"Lighting control system"

Technical Field

The invention relates to a control unit for a lighting control system according to the preamble of claim 1. Furthermore, the invention relates to a lighting control system, a lighting system, a method for controlling a lighting system, a computer program product and an HDR vision sensor according to claims, 3, 7, 8, 9 and 10, respectively. The invention furthermore relates to a control method according to the preamble of claim 11. The invention furthermore relates to a method for controlling a technical system, a computer program product and a control unit according to claims 18, 19 and 20, respectively.

Background Art

Electrical lighting is typically responsible for a large fraction of an office building's electricity consumption. Therefore, engineers and architects have over the last years more and more tried to find ways to reduce electric lighting in buildings and to find ways to use more daylight.

However/ this is not at all a trivial task. For example, it is not possible to light an office building exclusively with daylight: at times when it is dark outside, for example in the mornings, the evenings or during times of heavy rain, there is simply not enough daylight available, and any office building therefore always needs an electric lighting system, at least as a backup option. The fact that such an electric lighting system is there bears the risk that office occupants close window blinds and always work with the electric lighting switched on, even at times when enough daylight could be provided, for example by means of daylighting fagade elements or the like.

On the other hand, in office buildings which dispose of daylighting fagade elements or the like (or simply large windows), discomfort glare, i.e. so much light in an office occupant's view field that it makes the occupant feel uncomfortable, can occur. Ideally, an office lighting system therefore not only comprises an electric lighting system but also devices for avoiding glare, for example window blinds.

In addition to that, an office lighting system has to assure that there is always enough light on an office occupant's workspace, because otherwise the occupant cannot properly see his work, e.g. his paperwork or his computer keyboard.

An ideal office lighting system would therefore be able to always create a lighting scenario that offers an optimal trade-off between discomfort glare minimization, workplane illuminance maximization and electrical power minimization. Over the last few years, lighting control systems have therefore been developed, typically comprising a lighting control unit, different sensors and different actuators. Based on input values received from the sensors, e.g. illuminance sensors, the lighting control unit typically sends output values to the actuators which act on luminaires, window blinds and the like. Such lighting control systems are typically able to automatically guarantee a constant illuminance on an office occupant's workplane at every moment in time.

However, no fully automated control strategies exist which automatically avoid discomfort glare.

A common way to avoid glare sensation for the occupant as well as guaranteeing enough light includes using a rudimentary ceiling-mounted luminance meter that measures the horizontal illuminance (typically on a workplane level) right below the luminance meter. The principle of a common controller which integrates this measured value is quite straightforward: a control loop starts and the system performs a data acquisition task to sample the current indoor lighting condition. In the next step, the horizontal illuminance is compared with a lower boundary limit, for example 200 lux. According to the output, the controller either executes new commands to modify the settings of a lighting system (e.g. sending commands to the lighting facility of the building, sun shadings and electric lighting system), or the control loop continues for further analysis. During this further analysis, similar comparison/action tasks are carried out for other middle and upper illuminance limits and the corresponding commands are generated based on the outcome of the comparisons. The limits are chosen based on the indoor lighting norms and definition of the indices such as Useful Daylight illuminance (UDI). In the case where the outcomes of all the comparisons are negative, no action is taken and the data is logged. As soon as a control cycle ends (i.e. the control loop has come to an end), a new cycle is initiated. This solution does however not provide enough information to the building automation system (BAS) so as to deliver reliable commands for a dynamic automated lighting system. Moreover, the controllers do not have any systematic approach to filter out redundant, useless commands, which typically occur quite frequently. This deficiency leads to unnecessary and frequent lighting system status amendments (e.g. switching lights on and off, dimming, moving of blinds) which can quickly become annoying to the office occupants and which can therefore cause high automatic system refusal by building occupants.

Problem to be Solved

It is the object of the invention to solve or to at least diminish the above- mentioned disadvantages. In particular, it is the objective of the invention to find a way to automatically avoid discomfort glare in an office building, while still assuring an appropriate visual comfort and while minimizing electricity consumption.

Furthermore, it is an objective of the invention to find a way to automatically avoid discomfort glare in an office building, while still assuring an appropriate visual comfort, while minimizing electricity consumption, and while not being too annoying to the building occupants.

Solution to the Problem

This problem is solved by a control unit for a lighting control system, wherein the control unit is configured to calculate output values based on input values, wherein the output values include control commands for a lighting installation, wherein the input values include information about a light distribution, preferably a luminance distribution, in a view field of an office occupant, and the control unit is configured to calculate said output values based on said input values in such a way that discomfort glare for the office occupant is avoided, while sufficient workplane illuminance is guaranteed and electricity consumption is minimized.

Especially the fact that the control unit receives as input values information about a light distribution, preferably a luminance distribution, in the view filed of an office occupant enables the control unit to determine whether there is a risk of discomfort glare in the office occupant's view filed. The control unit can then efficiently and reliably avoid glare, for example by closing a window blind until the light distribution in the office occupant's view field is such that no discomfort glare occurs. At the same time, the control unit can act on a complementary electric lighting system in order to guarantee an appropriate illuminance on the office occupant's workplane, for example between 300 lux and 500 lux, while avoiding to supply too much electric lighting in order to keep an electricity consumption as low as possible.

In a preferred embodiment, the control unit is configured to carry out said avoiding of discomfort glare continuously. In this context, the term "continuously" means that the control unit is able to react immediately when discomfort glare occurs. This has the advantage of making the work environment very comfortable for the occupant because as soon as discomfort glare is detected, the control unit can intervene and for example move the window blinds such that the discomfort glare is overcome. However, it is also possible to have the control unit only periodically check whether discomfort glare occurs, for example every 5 minutes. In a particularly preferred embodiment, the control unit is configured such that it acts on the window blinds when a certain calculated glare value passes a certain threshold. In a preferred embodiment, the input values to the control unit comprise a sun profile angle, a current sun shading position and/or a current sun shading tilt angle and/or a current electric lighting power.

A lighting control system according to the invention comprises a control unit according to the invention and furthermore comprises a first light sensor, which is an HDR vision sensor, configured to at least partly supply said input values to the control unit. It is particularly preferred that the HDR vision sensor is configured such that it supplies said information about a light distribution in the view field of an office occupant to the control unit. In this context, the term "HDR vision sensor" is to be understood as follows: this photometric device allows for real-time capturing and analyzing of luminance maps of visual scenes with considerable accuracy and speed. It offers a 132dB intra-scene dynamic range encoded logarithmically with 149 steps per decade. Each HDR image therefore provides a complete record of the magnitude and spatial variation of the luminance in the field-of-view. "HDR" stands for "high dynamic range". Consequently, such an HDR vision sensor has the advantage of being able to supply a very detailed and very precise luminance distribution of the view field of the office occupant in a quick and reliable manner. However, it is not absolutely necessary to use an HDR vision sensor. It would in theory also be possible to use another sensor for creating the input values for the control unit.

In a preferred embodiment, the first light sensor is configured to calculate a glare value based on light captured by the first light sensor, wherein the glare value preferably comprises a daylight glare probability, and the first light sensor is configured to send said glare value as input value to the control unit. In a particularly preferred embodiment, the first light sensor is configured to calculate the glare value on-the-fly, meaning that the almost "real time" processed data, in particular the glare value, is provided to the control unit automatically, without any further offline or manual post-processing by means of any computer program. In a particularly preferred embodiment, the first light sensor is configured to send said glare value to the control unit at least every 20 seconds, preferably at least every 15 seconds, more preferably at least every 13 seconds, most preferably every 5 seconds.

The above-mentioned daylight glare probability is calculated according to the following equation:

DGP = 5.87χ1(Γ ⁵ £ _ν + 9.18xl0 ^_2 x + + °- ¹⁶

Where DGP is acronym for Daylight Glare Probability, E _v [lux] is the vertical eye illuminance, L _{s i} [cd/m ² ] is the average luminance of glare source i, ω _{5 ί} [sr] is the solid angle subtended by the glare source i and Ρ _; [-] is the Guth's position index for the glare source i, taking in to account the importance of the location of glare index with respect to the center of the image.

Calculating the glare value directly in the HDR vision sensor has the advantage of making it possible to only send the glare value to the control unit when the glare value passes a certain threshold and/or is such that discomfort glare occurs. Like this, unnecessary data transfer between the HDR vision sensor and the control unit, which could slow down the lighting control system, can be avoided. However, it is also possible for the control unit to calculate the glare value and the HDR vision sensor to only send raw luminance data to the control unit. Using the daylight glare probability for assessing a risk of discomfort glare is advantageous because it is considered to be a very reliable assessment of risk of discomfort glare which can furthermore be comparably easily and quickly calculated. However, it is also possible to use other values as glare values, for example a daylight glare index and/or a unified glare rating and/or a CIE glare index. In a preferred embodiment, the first light sensor comprises a digital signal processor configured to carry out image processing and the calculation of said glare value and/or in that the first light sensor is configured to be mounted at a height between 100 cm and 140 cm above a floor of an office room, preferably at a height between 110 cm and 130 cm above said floor, most preferably at a height of approximately 120 cm above said floor. In a particularly preferred embodiment, the first light sensor is configured to be mounted parallel to the line of view of the sitting office occupant, preferably next to the office occupant, or on a computer screen of the office occupant, or yet on a wall behind the occupant. Mounting the first light sensor in such a way has the advantage of obtaining realistic assessments of glare risk. However, it would in theory also be possible to locate the first light sensor differently, for example above the head of the office occupant.

In a preferred embodiment, the lighting control system comprises a second light sensor, wherein the second light sensor is preferably an HDR vision sensor, wherein the second light sensor is configured to measure an illuminance of one or more workplanes of an office room in which the lighting control system is installed. In a particularly preferred embodiment, the second light sensor is configured to be mounted on a ceiling of an office room and provides workplane illuminance values as input values to the control unit. Such a second light sensor has the advantage of making it extremely easy for the control unit to guarantee sufficient illuminances on the workplanes, because the control unit will always know about the actual illuminances on the workplanes and will be able to supply just as much additional electric light as needed, thus also minimizing the electricity consumption. Using an HDR vision sensor as second vision sensor has the advantage of making the illuminance measurement extremely precise. A lighting system according to the invention, which is preferably part of a building automation system, comprises a lighting control system according to the invention and further comprises a lighting installation, wherein the lighting installation comprises at least one lighting device and at least one window blind. The window blind is typically a Venetian blind or a roller blind. The lighting device is typically a ceiling-mounted luminaire, a freestanding luminaire or any other kind of luminaire or lamp. The control unit is configured to send the output values to the lighting installation such that, preferably at essentially every moment in time, discomfort glare is avoided, and preferably a best possible trade-off between discomfort glare minimization, workplane illuminance maximization and electrical power minimization is guaranteed. Preferably, in order to achieve this, an algorithm is typically continuously running inside the control unit which continuously determines optimal set points for the window blinds' altitude and/or tilt, and the optimal lighting power.

A method for controlling a lighting system in an office room according to the invention comprises the steps:

- capturing images of a view field of an office occupant by means of an HDR vision sensor,

- calculating, based on said images, a glare value, wherein the glare value preferably comprises a daylight glare probability and/or a daylight glare index and/or a unified glare rating and/or a CIE glare index, wherein the calculating is preferably carried out by means of a digital signal processor of the HDR vision sensor,

- supplying said glare value as input value to a control unit, preferably on the fly and/or continuously and/or event-based,

- measuring a workplane illuminance, preferably on the fly and supplying this workplane illuminance as input value to the control unit,

- determining, by means of the control unit and based on said input values, optimal settings for a lighting and/or sun shading installation, preferably based on an outcome of optimizing a cost function, including an electric lighting energy demand, a discomfort glare rating and/or a number of sun shading movements per day, and

- sending said optimal settings to the lighting installation by means of the control unit.

In a preferred embodiment, a data transfer from the HDR vision sensor to the control unit is carried out via a telemetry channel, wherein in this context the expression "telemetry channel" is to be understood as wired or wireless communication protocol to transfer data over pre-established channels to a remote agent such as a computer program or a repeater.

A computer program product, typically recorded on a computer-readable medium, comprises code for carrying out a method according to the invention.

An HDR vision sensor according to the invention comprises a fish eye lens, a CMOS light capturing sensor, and a digital signal processor, wherein the HDR vision sensor is configured to calculate a glare value, based on light captured by the light capturing sensor, wherein said calculation is preferably carried out on the fly, wherein the glare value preferably comprises a daylight glare probability and/or a daylight glare index and/or a unified glare rating and/or a CIE glare index, and the HDR vision sensor is configured to supply the glare value as input value to a control unit. In a preferred embodiment, the HDR vision sensor is spectrally calibrated and/or geometrically calibrated and/or photometrically calibrated, meaning that firstly, the spectral sensitivity of the sensor is modified through application of one or several glass- or gelatin-based optical filters so as to match the spectral sensitivity of the human eye {V( )). Secondly, the sensor is calibrated in order to eliminate the Vignetting phenomenon, a pure geometrical phenomenon, caused by the fisheye lens; reduction of an image's brightness or saturation at the periphery compared to the image center. Finally, the sensor is calibrated photometrically so that the value of each pixel is equal to the average luminance (e.g. in cd/m ² ) of corresponding patch of the area in the field of view of the sensor.

The HDR vision sensor works as follows: The light from its field of view arrives firstly on the fisheye lens. The fisheye lens collects the light and redirects the part of the light rays that are not parallel to its principal axes. In the next step, all the light rays going through its principal axis pass through one or several optical filters. The optical filters in this case change the spectral composition of the light so as to pass only visible part of the light. In the next step, the light rays arrive on the photodetector of the vision sensor where the light photons are transformed to voltage differences proportionally to the light intensity. The output of the photodetector is a matrix of 320 X 240 pixels in greyscale unit; ranging from 0 to 0123. It is fed in the next step to the digital signal processing unit to perform the image analysis and extract the daylight glare value, such as the daylight glare probability. As soon as this variable is ready, it is sent over the telemetry channel to the remote agent as an input for the control unit.

The problem is furthermore solved by a method for controlling a lighting system wherein the method comprises a first sub-method, wherein the first sub-method comprises a first data acquisition step during which a set of current lighting parameters is acquired, a fuzzy logic step during which a fuzzy logic sequence is run in order to determine a preferred sun-shading position based on the set of current lighting parameters, a checking step during which a check sequence is run to decide whether a sun shading system forming part of the lighting system is to be actuated such as to reach the preferred sun-shading position, and an actuating step during which a command is sent to the sun shading system by which the sun shading system is forced into the preferred sun-shading position, if during the checking step it was decided that the sun shading system is to be actuated such as to reach the preferred sun-shading position, wherein the set of current lighting parameters comprises a visual comfort parameter.

The combination of a fuzzy logic step for finding a preferred sun-shading position (which typically would seem ideal from an energy-efficiency and visual comfort point of view) with a checking step during which it is assessed whether it is really a good idea to carry out the necessary actions for achieving the preferred sun-shading position (in order to reduce annoyances) and the use of a visual comfort parameter (for reducing the glare risk) solves the problem in a synergistic and yet simple way.

In a preferred embodiment, the visual comfort parameter is a delta discomfort glare probability (DDGP). Preferably, a calibrated High Dynamic Range (HDR) vision sensor provides a discomfort glare index named Daylight Glare Probability (DGP), which, similar to any other variable that reports the occurrence probability of any phenomenon, ranges from 0% to 100%. This range is typically divided into four spans associated with subjective linguistic terms: imperceptible [0, 35%]; perceptible [35%, 40%]; disturbing [40%, 45%]; intolerable [45%, 100%]. In the standard setup, the threshold for imperceptible glare sensation is set to 35%. This threshold is typically referred to as DGP_ref. Since this value might vary from person to person, DGP_ref might not be equal to 35% all the time. To take into account these interpersonal differences, the new variable DDGP (delta daylight glare probablity) is introduced. It Is calculated as follows: DDGP = DGP_measured - DGP_ref. The fact that the visual comfort parameter is a DDGP therefore has the advantage that interpersonal differences can be accounted for by adapting the DGP_ref to each person by considering this person's preferences. However, it would also be possible to use another visual comfort parameter, like for example a unified glare rating (UGR) value or the like. In a preferred embodiment, the set of current lighting parameters comprises a horizontal workplane illuminance and/or a solar height and/or an azimuth angle. In this context, the horizontal workplane illuminance is an illuminance value at a certain point of a predefined horizontal workplane, for example at 80 cm or 100 cm above floor level, the solar height is typically expressed in degrees, and so is the azimuth angle, which corresponds to an azimuth of the sun with regard to a building fagade.

In a preferred embodiment, the preferred sun shading position comprises a first position value for a first sun shading and a second position value for a second sun shading. The use of two separate sun shading position values for two different sun shadings allows for a maximum amount of flexibility because the two sun shadings, which could typically be an upper sun shading and a lower sun shading, can then be actuated separately. However, it would also be possible to send one single position value to several sun shadings or to only foresee a single sun shading.

In a preferred embodiment, the checking step comprises a time check and/or a threshold check and/or a seriousness check. During the time check, it is typically checked how much time has elapsed since a last sun shading position change, and only if a minimum amount of time has elapsed since this last position change, another sun shading position change is considered further. The minimum amount of time is typically at least 5 minutes, preferably at least 10 minutes, more preferably at least 15 minutes. Such a time check has the advantage to guarantee that amendments to the sun shading positioning are not occurring too often. During the threshold check, it is typically checked whether a modification of a sun shading position will lead to a sufficiently important change in window coverage percentage. Only if a pre-defined threshold of window coverage percentage change is exceeded, the sun position change is considered further. In a preferred embodiment, a first threshold exists for increasing the window coverage percentage and a second threshold exists for decreasing it. In a preferred embodiment, the first threshold equals at least 5%, preferably at least 10%, more preferably at least 15% and/or the second threshold equals at least 10%, preferably at least 20%, more preferably at least 30%. Such a threshold check has the advantage to guarantee that too unimportant and thus annoying movements are avoided. During the seriousness check, it is typically tested whether a sun shading position change that is being instructed by the fuzzy logic step has been obtained during the fuzzy logic step such that the rule set of the fuzzy logic step is sufficiently applied. In other words: during the seriousness check, it is checked whether the fuzzy logic rules have been sufficiently applied. The objective of the seriousness check is to find out whether the rule set of the fuzzy logic step is sufficiently utilized for deriving the sun shading position change. In preferred embodiments, the seriousness of a whole rule base used in the fuzzy logic step is the sum of all fuzzy rules' weights. If the sum of the weights is close to 0, this means that the output value of the fuzzy logic step is not seriously determined. Preferably, the output weights correspond to IRR parameters and/or outputs of an evalfis command in Matlab.

In a preferred embodiment, the checking step comprises a visual comfort parameter check, wherein the visual comfort parameter check is preferably configured to override the time check and/or the threshold check and/or the seriousness check. This has the advantage that a main focus is put on avoiding glare. This aspect of the invention is based on the surprising finding by the inventors that it is especially discomfort glare which can cause building occupants to reject a building automation system. In a typical embodiment, a command is sent to the sun shading system if the DDGP value is higher than 5%. In a preferred embodiment, the method comprises a second sub-method, wherein the second sub-method comprises a second acquiring step during which a second set of current lighting parameters is acquired, an illuminance assessment step during which an illuminance level is assessed, and an electric lighting commanding step during which a command is sent to an electric lighting system forming part of the lighting system in order to achieve a desired illuminance level, if during the illuminance assessment step it was found that the illuminance is insufficient. Preferably, as a function of illuminance level insufficiency, the electric lighting system is tuned either to "full power" mode or "dimming" mode. The full power mode is preferably activated in the case where the difference between the current measured illuminance level and the reference illuminance level (e.g. 300 lux) is larger than the illuminance level that can be maximally provided by the electric lighting system. For example, in the case where the current illuminance level is 100 lux, the reference illuminance level is 300 lux and the maximum illumination by electric lighting is 150 lux, the electric lighting system is tuned to "full power" mode ((300-100) lux > 150 lux). Otherwise, the lighting system is preferably set to dimming mode. By applying this strategy, the electric lighting energy demand is minimized. Preferably, the electric lighting system is deactivated as the measured current illuminance level is larger than the reference illuminance value. Such a second sub-method meant to control an electric lighting system has the advantage that electric lighting can also be controlled. The separation into two distinct sub-methods allows for dealing with daylighting and electric lighting independently and flexibly. However, it would also be possible to control daylighting and electric lighting at the same time.

In a preferred embodiment, the first sub-method is carried out before the second sub-method. This has the advantage to also try everything possible to use available daylight in the first place, and then to only add artificial lighting when it is really necessary. This order for the sub-methods can therefore be considered to help to cut down electricity demands. However, it would theoretically also be possible to carry out the second sub-method before the first one.

In a method for controlling a technical system according to the invention, this method comprises a calculation step during which a preferred system setting is determined based on at least one input variable, wherein the method further comprises a filtration step, during which it is assessed whether the technical system should actually really reach the preferred system setting, wherein the filtration step preferably comprises a seriousness check. This method cannot only be used in lighting but for example also for assuring that appropriate instructions are, for example, given to robots that interact with humans. In a preferred embodiment, the method for controlling a technical system comprises a fuzzy logic step as described above, in particular a fuzzy logic step during which a seriousness of a whole rule base used in the fuzzy logic step is the sum of all fuzzy rules' weights, and during the seriousness check it is checked whether this seriousness is above a certain threshold, and an action command to reach the preferred system setting is only issued if the seriousness is above the threshold. In a preferred embodiment, the method for controlling a technical system comprises a time check and/or a threshold check and or a first sub- method and/or a second sub-method as described above. In a particular embodiment, the method for controlling a technical system comprises an override step configured to override the seriousness check and/or the time check and/or the threshold check.

A computer program product according to the invention comprises code for carrying out any of the methods according to the invention. A control unit according to the invention is configured to carry out any of the methods according to the invention.

FIGURES

In the following, the invention is described in detail by means of a drawing, wherein shows:

Figure 1 : A schematic view of an office room in which a lighting system

according to the invention is installed.

Figure 2: Flow diagram of a control method according to one embodiment of the invention,

Figure 3: Fuzzy logic control step with four inputs and two outputs,

Figure 4: Membership functions for input variable DDGP,

Figure 5: Schematic overview of rule aggregation during the fuzzy logic

control step according to the invention, and

Figure 6: Zoom-in on output section of Figure 5.

Description of Preferred Embodiments

Figure 1 shows schematic view of an office room 1 in which a lighting system according to the invention is installed. In office room 1 , an office occupant 2 is sitting on a chair at a workplace in front of a computer screen. A surface on the table at the workspace of the office occupant 2 is referred to as workplane. This is typically an area of the table where the office occupant's 2 computer keyboard would be located.

The office room 1 comprises a window 3. The window 3 is equipped with a window blind 4. The window blind 4 is made from fabric of beige color and is a roller blind. The office room furthermore comprises a daylighting system 5, which comprises a rounded surfaces coated with aluminum. This daylighting system 5 is configured to reflect daylight D that hits the aluminum coating onto a ceiling of the office room 1 , from where it is reflected downwards. The daylighting system 5 is equipped with a daylighting system blind 6, which is a roller blind.

The office room 1 is furthermore equipped with two ceiling-mounted luminaires 7.1 , 7.2. These ceiling-mounted luminaires 7.1 , 7.2 are equipped with electric light sources such as LEDs or fluorescent tubes and are configured to supply artificial light to the office room 1 and in particular the office occupant's 2 workplane at times where not enough daylight D reaches the office room 1 through the window 3 and the daylighting system 5.

The office room 1 is furthermore equipped with a control unit 8. The control unit 8 is linked to the window blind 4, the daylighting system blind 6 and to the luminaires 7.1 , 7.2. The control unit 8 is configured to create output values and to send them as instructions to the window blind 4, the daylighting system blind 6 and to the luminaires 7.1 , 7.2. These instructions can for example be dimming instructions for the luminaires 7.1 , 7.2 and/or up/down commands for the blinds 4, 6.

Furthermore, the office room 1 comprises a first light sensor 9. The first light sensor 9 is an HDR vision sensor. It is mounted on a movable rod 10 at approximately 120 cm above floor level, i.e. approximately at an eye level of the office occupant 2. The first light sensor 9 is configured to measure a luminance distribution in a view field of the office occupant 2 and to calculate a daylight glare probability from this luminance distribution. The first light sensor 9 is furthermore configured to supply this daylight glare probability as input value to the control unit 8 via a telemetry network, which is typically a wireless network (not visualized in Figure 1 ).

The office room 1 furthermore comprises a ceiling-mounted second light sensor 11 , which is also an HDR vision sensor. The second light sensor 11 is configured to measure an illuminance on the workplane of the office occupant 2.

The control unit 8 and the light sensors 9, 11 form a lighting control system. This lighting control system, together with the blinds 4, 6 and the luminaires 7.1 , 7.2 form a lighting system of the office room 1.

This lighting system operates as follows:

The first light sensor 9 measures a visual discomfort index, for example the daylight glare probability, for the view field of the office occupant 2, and the second light sensor 11 measures the workplane illuminance for the office occupant 2 and detects the presence of the office occupant 2. The first light sensor 9 and the second light sensor 11 make their data available as input values to the control unit 8. The control unit then regulates the lighting installation of the office room 1 , namely the luminaires 7.1 , 7.2 and/or the window blind 4 and/or the daylighting system blind 6 based on these input values.

Even though the first light sensor 9 is mounted on the rod 10 next to the office occupant 2 in Figure 1 , it is also possible to mount it on the computer screen of the office occupant 2 or yet on a wall or partition behind the office occupant 2. Furthermore, it is possible to foresee multiple first light sensors 9. This has the advantage to make it possible to more reliable detect discomfort glare in the office room 1.

Image processing and calculation of the visual discomfort index is carried-out by the digital signal processor embedded in the first light sensor 9. However, it would also be possible to transfer raw images to a remote digital signal processor for further image processing and data handling for control strategies.

In any case, the data transfer within the lighting system can be carried out by means of wireless or wired communication networks, or a combination of both. It is for example possible that the first light sensor 9 communicates with the control unit 8 through a wireless communication channel, whereas the blinds 4, 6, the luminaires 7.1 , 7.2 and the second light sensor 11 are connected to the control unit 8 by wire. This has the advantage that the first light sensor 9 is freely movable within the office room 1 , whereas, at the same time, the highest possible reliability and safety of the lighting system is guaranteed by the conventional wired connections of the other components which is typically less prone to failures and/or interferences.

The first light sensor 9 continuously evaluates the daylight glare probability based on the luminance distribution detected by it and provides it as input value to the control unit 8. Other visual comfort indices, such as Daylight Glare Index (DGI), Unified Glare Rating (UGR) and/or CIE Glare Index (CGI) are also measurable by the first light sensor 9.

The control unit 8 comprises is a closed loop controller that is configured to apply different control strategies such as predictive algorithms and/or user- adaptive algorithms and/or fuzzy logic algorithms. The control unit 8 can furthermore be configured to not only control the lighting installation, but also to interact with control units for heating, ventilation and air conditioning, so-called HVAC control units, or to even act as a combined lighting and HVAC control unit.

The ceiling-mounted luminaires 7.1 , 7.2 can be replaced by or combined with other ceiling-mounted luminaires and/or pendant luminaires and/or freestanding luminaires and/or recessed ceiling luminaires and/or recessed wall luminaires and/or spotlights and/or downlights.

The blinds 4, 6 can have different embodiments such as internal or external roller blinds, or yet internal or external Venetian blinds.

The second light sensor 11 can be installed directly above the workplane of the office occupant 2 (as shown in Figure 1 ), or at any other place on the ceiling of the office room 1. The workplane illuminances of several work stations and/or workplanes are measurable with one and the same second light sensor 11 , but it is also possible to foresee multiple second light sensors 11.

Figure 2 shows a block diagram of a method for controlling a lighting system according to the invention. The method starts with a start step SO and ends with an end step S12. This procedure is called one cycle. As soon as one cycle is finished, the method is typically restarted.

The method comprises two sub-methods. The first sub-method is responsible for defining and commanding the positions of a sun shading system forming part of the lighting system based on the captured data. The steps S1 to S7 belong to this first sub-method. The second sub-method is responsible for defining a state and commanding an electric lighting system forming part of the lighting system. The steps S8 to S11 belong to this second sub-method.

At the beginning of each sub-method, a data acquisition step S1 and S8, respectively, is performed in order to provide the control method with the actual lighting conditions of the office room. The reason for dividing the control method into two sub-methods is to privilege the daylighting over the electric lighting by improving first the lighting condition of the office by adjusting the sun shading system.

During the method, a calibrated High Dynamic Range (HDR) vision sensor provides a discomfort glare index named Daylight Glare Probability (DGP) (as explained in [1], see Section "Cited References" below) as input variable to be dealt with during the method. The DGP, similar to any other variable that reports the occurrence probability of any phenomenon, ranges from 0% to 100%. This range is divided into four spans associated with subjective linguistic terms: "imperceptible" (0% to 35%), "perceptible" (35% to 40%), "disturbing" (40% to 45%) and intolerable (45% to 100%). For the purpose of glare avoidance, the threshold for imperceptible glare sensation is typically set to 35%. This threshold is referred to as DGP_ref. However, the threshold for imperceptible glare sensation typically varies from person to person and DGP_ref might therefore not be equal to 35% for all persons. To take into account these interpersonal differences, the method in the present embodiment uses a new variable, namely a delta daylight glare probability (DDGP), which is calculated as follows:

DDGP = DGP_measured - DGP_ref, wherein the variable DGP_measured is a measured DGP at a particular moment in time and wherein DGP_ref is a customized DGP for a particular office occupant, in particular an office occupant who will be using the lighting system which is controlled by the control method according to the invention.

The DDGP is used as an input to the first sub-method according to the following procedure:

During a data acquisition step S1 , the current indoor lighting condition in an office room is sampled. In particular, four current lighting parameters are determined during this step, namely the DDGP 12, a horizontal workplane illuminance 13, a solar height 14 and an azimuth angle 15. These four current lighting parameters are used as inputs for the fuzzy logic step S2, as can be seen in Figure 3. The outputs 16 and 17 of the fuzzy logic step S2 of the present embodiment are new sun shading positions 16 and 17 for two sun shadings, which correspond to the preferred sun-shading position at a given moment in time.

During the fuzzy logic step S2, a fuzzy logic sequence is run in order to determine the preferred sun-shading position based on the set of current lighting parameters. In this particular embodiment, new commands for the positions of the sun shading system and the seriousness of the fuzzy logic step about these commands are created as outputs of the fuzzy logic step S2. More details regarding the fuzzy logic step S2 will be explained later on.

The outputs of the fuzzy logic step S2 are used as input variables for a checking step. This checking step comprises a time check S3, a threshold check S4 and a seriousness check S6. The reasons for the presence of the checking step is that the inventors have found out that the commands created by the fuzzy logic step S2 cannot always be applied as such to the sun shading system for the following two reasons:

1 ) The number of the sun shading and electric lighting status amendments should be minimized so as to avoid bothering the building occupants; 2) The system should remain as agile as possible when there is a high risk of glare sensation.

In the time check S3, a minimum time interval, e.g. 15 minutes, is introduced in order to avoid frequent amendments of the shading position. If there is any command to the sun shading system generated by fuzzy logic step S2 before this minimum time interval (which typically starts once a sun shading positioning has been effectuated) has elapsed, the command is ignored and the method continues directly with a data logging step S11 (see below). However, in case this minimum time interval has elapsed, the command is passed to the threshold check S4.

In the threshold check S4, a threshold is introduced in order to avoid jerky, and occasionally useless and annoying movements of sun shading system. For a command that would lead to an increasing of a window coverage percentage (in other words: a closing of sun shading) a lower threshold, namely 15%, is chosen. A higher threshold is chosen for a command that would lead to a decreasing of the window coverage percentage (in other words: an opening of sun shading), for which a threshold of 30% is chosen. The reason for these different thresholds is the fact that it is preferable to encourage more agile glare protection capacity and therefore to accept lower thresholds for cases where sun shadings are to be closed. For example, the threshold of 15% means that only such commands are carried out which lead to at least 15% more window area being covered by the sun shading than before. This means that if the difference between the outcome of the fuzzy logic step S2 (new position) for the sun shading and its actual position is higher than the threshold, the commands are passed to the seriousness check S6. Otherwise, the command is ignored and the method continues directly with the data logging step S11 (see below).

In the seriousness check S6, the seriousness of the command generated by the fuzzy logic step S2 is taken into account. In fact, the inventors have found out by means of experiments that not all of the commands are serious and strong enough to be taken into consideration. The reason for this is that the structure and outcome of the fuzzy logic step S2 depends basically on the experience of the controller designer who designs the fuzzy logic step S2 and in particular the rules. The controller designer defines each rule based on some subjective observations and assumptions regarding the behavior of the system states in different situations. Most of the time, these rules might not be sufficient enough to cover all the possible conditions that a system might experience. Hence, the outcomes might not be "serious" and "strong" all the time. To address this challenge and avoid applying weak, or in other words: not sufficiently serious, commands, a threshold for seriousness of the output of the fuzzy logic step S2 is defined. This threshold is chosen to be 15% of seriousness. If any command has a seriousness of more than 15%, it is sent to the sun shading system to execute it. If not, the command is ignored and the method continues directly with the data logging step S11 (see below).

The first sub-method furthermore comprises a visual comfort parameter check S5. This visual comfort parameter check S5 is assigned to keep the system agile enough to react to sudden strong risks of glare sensation. If the DDGP is higher than a certain threshold - for example 5% -, regardless of any possible outcomes of the checks S3, S4 and S6, the command of the fuzzy logic step S2 is passed on to the actuators.

During an actuating step S7, the commands having been filtered during the checking step are executed. This actuating step S7 sends the commands to the motors of the sun shading system. This is the end of the first sub-method of the control cycle.

The second sub-method of the method for controlling a lighting system begins with a second data acquisition step S8 during which the new indoor lighting conditions - i.e. the lighting conditions which have been reached after the first sub-method has been carried out - are re-evaluated.

During an illuminance check step S9, it is determined if a horizontal illuminance on a predefined workplane is high enough. If this is the case, there is no need for applying electric lighting. In this case, the method continues directly with the data logging step S11 (see below). Otherwise, the necessary complementary illumination intensity needed by the electric lighting is determined. This can include determining a dimming level of the electric lighting system if a dimmable system is in place.

During a lighting step S10, lighting commands in line with the outcome of the illuminance check step S9 are applied to the electric lighting system.

In the data logging step S11 , the complete record of the inputs, decisions and applied commands are registered for post-analysis purposes.

The functioning of the fuzzy logic step S2 shall now be explained in more detail.

In general, fuzzy inference is the process of formulating the mapping from a given input to an output using fuzzy logic. It is composed of five steps: i) fuzzify inputs, ii) apply fuzzy operators, iii) apply implication method, iv) aggregate all outputs and finally v) deffuzify. In the fourth step of the process, all the outputs of the rules are aggregated. Because decisions are based on the testing of all of the rules in a Fuzzy Inference System (FIS), the rules must be combined in some manner in order to make a decision. Aggregation is the process by which the fuzzy sets that represent the outputs of each rule are combined into a single fuzzy set. Aggregation only occurs once for each output variable, just prior to the fifth and final step, namely defuzzification. The input of the aggregation process is the list of "truncated output" functions returned by the implication process for each rule. The output of the aggregation process is one fuzzy set for each output variable. In general, the output of each rule of a fuzzy process is combined, or aggregated, into a single fuzzy set whose membership function assigns a weighting for every output (tip) value. The explained principles are valid for the crisp output sets; these sets are used in this embodiment for generating commands for the position of the two sun shadings.

In the final step of the fuzzy process, the above-mentioned step v) or defuzzification step, the centroid method is applied to return the center of an area under a curve that represents the result of aggregation. The mathematical operation behind the centroid method is depicted in Equation 1 , where the value y ₀ is the value of the fuzzy inference output, μΒ' (yj) is the value of the aggregated result at yj and j is the sweeping parameter over the output range and F is the upper maximum of the output range.

(Equation 1 )

The "seriousness" value of the present embodiment is the denominator of Equation 1 , which is the weight of fuzzy or crisp output sets.

seriousness =

(Equation 2) In the case where the sum of the aggregated results is too small, it implies that none of the rules are seriously contributed to the aggregated results. In this case, although y ₀ , the final output, is mathematically calculable and it is theoretically valid, it is not derived based on considerable aggregated values. This case normally happens when the rules does not have much to "say" based on the inputs.

During the fuzzy logic step S2 of the embodiment shown in Figure 2, the input variables are fuzzified: according to predefined membership functions, the numerical variables are translated to fuzzified variables. The membership functions of the DDGP are defined as shown in Figure 4. They are defined based on the boundaries defined for the DGP in document [1] by Wienold and Christoffersen (see details in Section "Cited References" below). Reference to the definitions of the boundaries for the DGP in document [1] is hereby made.

In Figure 4, four membership functions for the input variable DDGP are shown, namely (from top to bottom) a first membership function for the linguistic term "imperceptible", a second membership function for the linguistic term "perceptible", a third membership function for the linguistic term "disturbing" and a fourth membership function for the linguistic term "intolerable".

Figure 5 shows a rule-base according to the invention predefined by a control method designer. Based on this rule base, the control method provides the proper positioning outputs for two sun shading devices as follows:

First of all, the inputs are inserted to each rule and based on the internal mechanism of the rule, and then the decisions for the position of sun shadings are determined. As depicted in the two right columns of Figure 5, the decision of the rule for each shading position is created by a) the position of the sun shading device, for example 30% window coverage; and b) the "seriousness" of the rule about this decision. The respective position of the sun shading device is shown by the horizontal positions of the striped vertical bars 19 in the white output boxes and the seriousness is shown by the ratio of the length of the thick black bars 18 with respect to the total height of the striped vertical bars 19. For example, for rule 5, as the first output, the striped vertical bar 8 is located in the far right side of the white box (e.g. 100%) and it shows that the decision of this rule for this output is 100% of window coverage. In addition, we observe that the thick black bar 18 fills up around 30% of the total height of the striped vertical bar 19. The rest of the bar remains striped and thinner than the thick black bar 18. This shows that the rule 5 is 30% sure about the value of its output (e.g. 100%). This step is repeated for all the rules in the rule base. A zoom on the output part of Figure 5 is shown in Figure 6.

The weighted sum of the output of each rule determines the final value and the seriousness of the fuzzy logic controller for this set of inputs. The final outputs are shown in the bottom right corner of Figure 5 and at the bottom of Figure 6, respectively. In this example, it is shown that the sun shading window coverage for sun shading #1 is about 100% as depicted in Output 1 (horizontal position of the checkerboard pattern vertical line v1 ) and for sun shading #2 it is about 60% as depicted in Output 2 (horizontal position of the checkerboard pattern vertical line v2). Once again, we notice here that there are some vertical thick black bars 18 in the final outputs of the fuzzy logic step S2. These thick black bars 18 obviously can be interpreted as the degree of the seriousness of the controller about the value of its outputs. The fact that there are two thick black bars 18 in the box Output 2 (which corresponds to the second sun shading) is due to the contributions of the two rules (in this example, rule 4 and 5) to the final output of the fuzzy logic set. The overall seriousness will be the sum of the two thick black bars 18 following to the definition of the seriousness in Equation 2 above. In one embodiment, a method for controlling a lighting system in an office room according to the invention comprises the steps:

- capturing images of a view field of an office occupant by means of an HDR vision sensor,

- calculating, based on said images, a glare value, wherein the glare value preferably comprises a daylight glare probability and/or a daylight glare index and/or a unified glare rating and/or a CIE glare index, wherein the calculating is preferably carried out by means of a digital signal processor of the HDR vision sensor,

- supplying said glare value as input value to a control unit, preferably on the fly and/or continuously and/or event-based,

- measuring a workplane illuminance, preferably on the fly and supplying this workplane illuminance as input value to the control unit,

- determining, by means of the control unit and based on said input values, optimal settings for a lighting and/or sun shading installation, preferably based on an outcome of optimizing a cost function, including an electric lighting energy demand, a discomfort glare rating and/or a number of sun shading movements per day, and

- sending said optimal settings to the lighting installation by means of the control unit,

wherein the method further comprises a first sub-method, wherein the first sub- method comprises:

- a first data acquisition step (S1 ) during which a set of current lighting parameters (12, 13, 14, 15) is acquired,

- a fuzzy logic step (S2) during which a fuzzy logic sequence is run in order to determine a preferred sun-shading position based on the set of current lighting parameters (12, 13, 14, 15)

- a checking step (S3, S4, S5, S6) during which a check sequence is run to decide whether a sun shading system forming part of the lighting system is to be actuated such as to reach the preferred sun-shading position, and - an actuating step (S7) during which a command is sent to the sun shading system by which the sun shading system is forced into the preferred sun- shading position, if during the checking step it was decided that the sun shading system is to be actuated such as to reach the preferred sun-shading position, wherein the set of current lighting parameters (12, 13, 14, 15) comprises a visual comfort parameter (12).

The invention is not limited to the preferred embodiments described here. The scope of protection is defined by the claims.

Furthermore, the following claims are hereby incorporated into the Description of Preferred Embodiments, where each claim may stand on its own as a separate embodiment. While each claim may stand on its own as a separate embodiment, it is to be noted that - although a dependent claim may refer in the claims to a specific combination with one or more other claims - other embodiments may also include a combination of the dependent claim with the subject matter of each other dependent or independent claim. Such combinations are proposed herein unless it is stated that a specific combination is not intended. Furthermore, it is intended to include also features of a claim to any other independent claim even if this claim is not directly made dependent to the independent claim.

It is further to be noted that methods disclosed in the specification or in the claims may be implemented by a device having means for performing each of the respective acts of these methods. Cited References

[1] J. Wienold and J. Christoffersen, "Evaluation methods and development of a new glare prediction model for daylight environments with the use of CCD cameras," Energy Build., vol. 38, no. 7, pp. 743-757, Jul. 2006.

Reference list

1 Office room

2 Office occupant

3 Window

4 Window blind

5 Daylighting system

6 Daylighting system blind

7.1 , 7.2 Luminaire

8 Control unit

9 First light sensor

10 Rod

11 Second light sensor

12 Delta Daylight Glare Probability (DDGP)

13 Horizontal workplane illuminance

14 Solar height

15 Azimuth angle

16 First position value

17 Second position value

18 Thick black bars (representing seriousness)

19 Striped vertical bars

SO Start step

S1 First data acquisition step

S2 Fuzzy logic step

S3 Time check

S4 Threshold check

S5 Visual comfort parameter check

S6 Seriousness check

S7 Actuating step 58 Second data acquisition step

59 Illuminance check step

510 Lighting step

511 Data logging step

512 End step

D Daylight

v1 , v2 vertical lines in respective output boxes for window 1 and window