BACKGROUND

[0001] Some printing systems include components that can, over time, become worn or defective, leading to a reduced performance of the printing system and/or print defects.

[0002] An example of one component that may become worn over time, is a belt used to drive (e.g. cause rotation or movement) of other components within the printing system. Some belts include teeth to engage with complementary recesses or spaces between teeth of other components of the printing system. Teeth of such a belt may become worn over time and, in some examples, may wear down completely or break away from the belt, leading to a reduction in performance of the belt.

BRIEF DESCRIPTION OF DRAWINGS

[0003] Examples will now be described, by way of non-limiting example, with reference to the accompanying drawings, in which:

[0004] Figure 1 is a schematic illustration of an example of apparatus for determining a state of a belt in a print agent application assembly;

[0005] Figure 2 is a schematic illustration of an example of apparatus for determining a state of a belt in a substrate conveying system;

[0006] Figure 3 is a flowchart of an example of a method of training a predictive model to determine a state of a belt of a print apparatus;

[0007] Figure 4 is a series of graphs showing data used to train a predictive model;

[0008] Figure 5 is a representation of various layers of a predictive model;

[0009] Figure 6 is a series of graphs showing outputs of a trained predictive model;

[0010] Figure 7 is a flowchart of an example of a method of determining a state of a belt of a print apparatus;

[0011] Figure 8 is a flowchart of a further example of a method of determining a state of a belt of a print apparatus;

[0012] Figure 9 is a simplified schematic illustration of an example of apparatus to determine an operating state of a belt in a print component; [0013] Figure 10 is a simplified schematic illustration of a further example of apparatus to determine an operating state of a belt in a print component;

[0014] Figure 11 is a simplified schematic illustration of an example of a machine- readable medium in communication with a processor; and

[0015] Figure 12 is a simplified schematic illustration of a further example of a machine-readable medium in communication with a processor.

DETAILED DESCRIPTION

[0016] A print apparatus includes many moving parts, some of which may suffer mechanical wear over time. In some cases, a component that suffers wear may start to function in a less-than-optimal manner, such that the component and/or the print apparatus as a whole suffers a drop-off in performance. The reduced performance may manifest itself in the form of reduced printing efficiency or reduced image quality in a resulting print product.

[0017] Wear or defective parts of some components may be visible to an operator or a user and, therefore, it may be clear when a component is to be repaired or replaced. However, some components within a print apparatus are positioned in places which not easily accessible, or which cannot be seen without first removing a cover or another component. One example of a component that may be worn through use is a belt. Belts may be used to drive components such as wheels, cogs and rollers in a print apparatus. Some belts may include teeth (referred to as a toothed belt or timing belt), and teeth of the belt may become worn over time. Belts with a relatively small amount of wear (e.g. where one or two teeth are worn) may continue to function, but less effectively than a belt with no defective teeth. Such wear may be referred to as gradual wear or gradual failure. A more significant defect occurs when the amount of wear (e.g. the number of damaged teeth) is such that the belt no longer functions (e.g. the belt is no longer able to drive the components to which it is connected). This may be referred to as catastrophic wear or catastrophic failure.

[0018] The present disclosure provides a mechanism by which various states of a component (e.g. states of defectiveness of a belt) may be determined by measuring parameters of a print apparatus. For example, a state of a first component of a print apparatus may be determined based on the measurement of a parameter of a second component of the print apparatus. In particular, examples disclosed herein make use of a trained predictive model or classifier to determine, on the basis of the measured parameters, the state of a belt. Part of the present disclosure relates to obtaining measurements of parameters of a print component in a particular state, and using it to train a predictive model, for example using machine learning techniques. Other parts of the disclosure relate to using a trained predictive model to determine the state of a component based on measurements of parameters.

[0019] According to one example, elements of the present disclosure may be used to determine a state of a belt in a print agent application assembly of a print apparatus. A print agent application assembly may be used to apply print agent, such as ink, to a surface, such as a surface of a printable medium or the surface of a roller used during a printing operation. One example of a print agent application assembly as a binary ink developer (BID). BIDs may be used in liquid electrophotography (LEP) printing, which involves transferring charged liquid ink via a series of rollers to a substrate or printable medium. In an LEP print apparatus, print agent may pass through a BID. Each BID handles print agent of a particular colour, so an LEP printing system may include, for example, seven BIDs. Print agent from a BID is selectively transferred from a print agent transfer roller - also referred to as a developer roller - of the BID in a layer of substantially uniform thickness to a photoconductive surface, such as a photo imaging plate (PIP). The selective transfer of print agent is achieved through the use of an electrically-charged print agent, also referred to as a“liquid electrophotographic ink”. During the printing process, the entire PIP of the print apparatus is charged, then areas representing an image to be printed are discharged. Print agent (e.g. LEP ink) is transferred to those portions of the PIP that have been discharged. The PIP transfers the print agent to a printing blanket, which subsequently transfers the print agent onto a printable substrate, such as paper. The discharged portions of the PIP represent the portion or portions of a pattern or image in which print agent from the BID is to be applied to the substrate.

[0020] Print agent that is not transferred from the developer roller to the PIP (i.e. in those areas where the PIP remains charged) remains on the developer roller of the BID, and is removed from the developer roller by components within the BID, such as a cleaner roller. Print agent may be removed from the cleaner roller using a sponge roller which includes an absorbent material to absorb print agent from the surface of the cleaner roller. A Bl D may also include an idler gear which may be connected to the sponge roller and the cleaner roller using a belt, such as a toothed belt or timing belt. The position of the idler gear relative to the cleaner roller and/or the sponge roller may be adjusted in order to increase and/or decrease the tension of the belt. In other examples, a belt (e.g. a timing belt) may be used elsewhere in a BID, to drive other components.

[0021] Figure 1 is a schematic illustration of an example of an apparatus 100 for determining a state of a belt in a print agent application assembly, such as a BID. The apparatus 100 includes a print agent application assembly (e.g. a BID) 102. Figure 1 shows just a portion of the BID 102 that is relevant to the present disclosure; it will be clear to those skilled in the relevant field that such a BID will include other components not shown in Figure 1. The BID 102 shown in Figure 1 includes a cleaner roller 104, the sponge roller 106 and an idler gear 108, which are connected to one another by a timing belt 110. The timing belt 110 includes a plurality of teeth (e.g. prism-shaped protrusions extending across a width of the belt) to engage with complimentary recesses or notches (e.g. between adjacent teeth) of a gear or cog. One or more of the rollers or gears (e.g. the cleaner roller 104, the sponge roller 106 and/or the idler gear 108) may be driven by a motor 112 (also referred to as a BID motor) which, in some examples, may comprise a servomotor. The motor 112 (shown connected to the cleaner roller 104 in Figure 1) drives

(i.e. causes rotation of) one roller connected to the belt 110, and its engagement with the belt leads to each of the other rollers being caused to rotate.

[0022] The motor 112 may be in communication with a processor or processing apparatus 114, for example via a wired connection or a wireless connection. The processor 114 may be located in the print apparatus in which the BID 102 is to be installed. In other examples, the processor 114 may be remote from the print apparatus.

[0023] Figure 2 is a schematic illustration of an example of an apparatus 200 for determining a state of a belt a substrate conveying system 202 of a print apparatus. The substrate conveying system 202 may be used to transport a substrate (e.g. a printable medium) along part of a print apparatus, for example after the substrate is has been printed. The substrate conveying system 202 may, in some examples, be referred to as an exit pickup conveyor (EPC). The substrate conveying system 202 shown in Figure 2 includes a first idler wheel 204, a second idler wheel 206, a pulley 208 and a motor 210, all of which are connected via a belt 212. In other examples, more or fewer wheels or rollers may be engaged with and driven by the belt 212. As with the example shown in Figure 1 , the belt 212 used in the substrate conveying system 202 comprises a toothed timing belt. In the example shown in Figure 2, the position of the motor 210 may be adjusted using a tensioning mechanism 214 in order to increase and/or decrease a tension of the belt 212. The motor 210 (e.g. a servomotor) drives the belt 212 which, in turn, drives the first idler wheel 204, the second idler wheel 206 and the pulley 208. [0024] The motor 210 may be in communication with the processor or processing apparatus 114 via a wired connection or a wireless connection. In other words, a single processor may, in some examples, control or communicate with multiple motors (e.g. the motors 112 and 210). In other examples, multiple processors may be provided, such that a separate processor may communicate with each motor 112, 210.

[0025] The processor 114 may obtain data measured in respect of a motor 112, 210. In some examples, measurements may be acquired in respect of a motor using components within the motor, and transmitted to the processor 114. In other examples, a separate component may be used to obtain measurements in respect of a motor, and transmit those measurements to the processor 114. In examples described herein, a velocity of the motor 112, 210 may be measured (e.g. from the motor encoder) to provide a first parameter, and a torque of the motor may be measured (e.g. based on an electrical current associate with the motor) to provide a second parameter. In other examples, other parameters may be measured, such as an electric current associated with the motor.

[0026] It has been recognised that, by measuring various parameters in respect of the motor 112, 210, it is possible to determine a state of the belt 110, 212 being driven by the motor. According to the present disclosure, parameters of motors may be measured when operating with belts having various identified defects (i.e. in different states), and used to train a predictive model, such that the trained model may be used to determine a state (e.g. a defect) of a belt when parameters measured in respect of the belt are provided as an input to the trained model. Figure 3 is a flowchart of an example of a method 300 of training a predictive model to determine a state of a belt of a print apparatus. The method 300 comprises, at block 302, obtaining a training data set. The training data set may be obtained by measuring operating parameters of a plurality of belts operating in print apparatus, each belt having been classified according one of a plurality of operating states. At block 304, the method 300 comprises training the predictive model, using the obtained training data, to classify a belt according to one of the plurality of operating states.

[0027] In some examples, the measured operating parameters may include one or more of: a velocity of a motor used to drive the belt; and a torque associated with a motor used to drive the belt. The plurality of operating states may, in some examples, include one or more of a suitable operating state; a low tension state wherein a tension of the belt is below a defined threshold; a partly defective state wherein the belt is partly defective and still capable of operating; and a defective state wherein the belt is defective. In a suitable operating state, the belt may function to an intended level, for example with little or no reduction in operating ability. A belt may be considered to be in a suitable operating state if its tension is above a defined threshold. A low tension state may be identified if the tension of the belt is below a defined threshold. Such a low tension may lead to a reduced operating efficiency. If low tension of a belt is identified, then the belt tension may be increased, for example by an operator. The belt may be considered to be in a partly defective state in a small number of its teeth (e.g. one or two teeth) are damaged or worn. In this state, the belt may be capable of driving the rollers to which it is connected, but there may be a slight reduction in efficiency. For example, a belt with a small number of teeth worn or missing may slip as it passes over teeth of a cog, wheel or roller. A belt may be considered to be a defective state when a larger number of its teeth are damaged or worn. For example, a belt having more than two adjacent worn or missing teeth may not operate at an intended level and, therefore, such a belt may be considered defective and in need of replacement.

[0028] In some examples, training the predictive model at block 304 may comprise training multiple predictive models, each to perform a separate classification task. For example, a first predictive model may be trained to classify a belt either as being in a suitable operating state (e.g. in which the tension of the belt is equal to or above a defined threshold) or as being in a low-tension state (e.g. in which the tension of the belt is below a defined threshold). The second predictive model may, for example, be trained to classify a belt either as being partially defective (e.g. in which the belt has just two adjacent worn or missing teeth) or as being fully defective (e.g. in which the belt has more than two adjacent worn or missing teeth). While a single classifier may be trained to classify a belt according to one of the four possible classifications, the use of multiple classifiers may lead to more accurate classification predictions.

[0029] Figure 4 is a series of graphs showing data used to train a predictive model. The data shown in Figure 4 was obtained using belts in various states operating in two different BIDs. For each BID, measurements were taken for: i) belts operating in an intended way (e.g. new belts), with an intended tension; ii) belts operating with a low (i.e. lower than intended) tension; iii) belts operating with two adjacent teeth removed; and iv) belts with more than two adjacent teeth. In one example, for a defective belt (i.e. a belt with more than two adjacent teeth missing), belts were used in which a sufficient number of adjacent teeth were missing such that teeth of the cleaner roller 104 (functioning as a drive gear) were unable to engage with teeth of the belt. For each belt, the velocity of the motor and the torque of the motor were measured with a 0.5 millisecond sampling rate, for a duration of 1.0235 seconds, giving 2040 examples. [0030] In Figure 4, graphs 402 to 408 show motor velocity data obtained in the BID and graphs 410 to 416 show motor torque obtained in the BID. The data in graphs 402 and 410 was obtained when fully functioning belts were used, the data in graphs 404 and 412 was obtained when low-tension belts were used, the data in graphs 406 and 414 is obtained when belts having to missing teeth were used, and the data in graphs 408 and 416 was obtained when defective belts (e.g. belts having more than two missing teeth) were used.

[0031] Data acquired (e.g. the data shown in Figure 4) in respect of the various belts may be used to train to one or more predictive models. In some examples, a single predictive model may be trained to identify a state of a belt based on a parameter (e.g. motor velocity or motor torque) measured in respect motor driving the belt. In other examples, multiple predictive models may be trained. For example, a first predictive model may be trained to distinguish between a fully-functioning belt and a belt having low-tension, and a second predictive model may be trained to distinguish between a belt having just two adjacent teeth missing and a belt having more than two adjacent teeth missing. By using two separate predictive models, a more accurate prediction may be made because some features of the data that are relevant to low-tension may also be relevant to belts with teeth missing.

[0032] Various different types of models may be used in a machine learning environment to be used as predictive models or classifiers identifying a state of a belt. In some examples, a model may be used which are selected from a group comprising artificial neural networks, support vector machines (SVMs), Bayesian networks and genetic algorithms. Other types of model may be used

[0033] Artificial neural networks or, simply, neural networks, will be familiar to those skilled in the art, but in brief, a neural network is a type of model that can be used to classify data (for example, classify, or identify the contents of image data). The structure of a neural network is inspired by the human brain. Neural networks are comprised of layers, each layer comprising a plurality of neurons. Each neuron comprises a mathematical operation. In the process of classifying a portion of data, the mathematical operation of each neuron is performed on the portion of data to produce a numerical output, and the outputs of each layer in the neural network are fed into the next layer sequentially. Generally, the mathematical operations associated with each neuron comprise one or more weights that are tuned during the training process (e.g. the values of the weights are updated during the training process to tune the model to produce more accurate classifications). [0034] For example, in a neural network model for classifying the contents of images, each neuron in the neural network may comprise a mathematical operation comprising a weighted linear sum of the pixel (or in three dimensions, voxel) values in the image followed by a non-linear transformation. Examples of non-linear transformations used in neural networks include sigmoid functions, the hyperbolic tangent function and the rectified linear function. The neurons in each layer of the neural network generally comprise a different weighted combination of a single type of transformation (e.g. the same type of transformation, sigmoid etc. but with different weightings). In some layers, the same weights may be applied by each neuron in the linear sum; this applies, for example, in the case of a convolution layer. The weights associated with each neuron may make certain features more prominent (or conversely less prominent) in the classification process than other features and thus adjusting the weights of neurons in the training process trains the neural network to place increased significance on specific features when classifying an image. Generally, neural networks may have weights associated with neurons and/or weights between neurons (e.g. that modify data values passing between neurons).

[0035] As briefly noted above, in some neural networks, such as convolutional neural networks, lower layers such as input or hidden layers in the neural network (i.e. layers towards the beginning of the series of layers in the neural network) are activated by (i.e. their output depends on) small features or patterns in the portion of data being classified, while higher layers (i.e. layers towards the end of the series of layers in the neural network) are activated by increasingly larger features in the portion of data being classified. As an example, where the data comprises an image, lower layers in the neural network are activated by small features (e.g. such as edge patterns in the image), mid level layers are activated by features in the image, such as, for example, larger shapes and forms, whilst the layers closest to the output (e.g. the upper layers) are activated by entire objects in the image.

[0036] In general, the weights of the final layers of a neural network model (known as the output layers) are most strongly dependent on the particular classification task being solved by the neural network. For example, the weights of outer layers may heavily depend on whether the classification task is a localisation task or a detection task. The weights of lower layers (e.g. input and/or hidden layers) tend to depend on the contents (e.g. features) of the data being classified.

[0037] Generally, neural network models may comprise feed forward models (such as convolutional neural networks, autoencoder neural network models, probabilistic neural network models and time delay neural network models), radial basis function network models, recurrent neural network models (such as fully recurrent models, Hopfield models, or Boltzmann machine models), or any other type of neural network model comprising weights.

[0038] The model used to identify a state of the belt, according to the present disclosure, may be considered to constitute a classifier as it is able to provide an output in the form of a classification of the belt (e.g. operating as intended, low tension, two teeth missing, or fully defective).

[0039] According to one example, a separable convolution layers-based model may be used as a predictive model. Figure 5 is a representation of various layers of a separable convolution layers-based model may be used in examples disclosed herein. In the example shown in Figure 5, the model is constructed from 4 separable convolution layers having 128 filters, 4 separable convolution layers having 64 filters, a sequence of layers in which functions are performed, including batch normalisation, max pooling and dropout, 4 separable convolution layers having 64 filters, four separable convolution layers having 32 filters, a global average pooling layer and, finally, a dense layer which includes a Sigmoid activation function to generate the final classification. Models having other structures may alternatively be used.

[0040] Figure 6 is a series of graphs showing the accuracy of predictions made by a prediction model trained using techniques described herein, using training and testing data. The data shown in Figure 6 was obtained using a separable convolution layer-based model as discussed above. Graph 602 shows training accuracy and validation accuracy as a function of the number of epochs, graph 604 shows training loss and validation loss as a function of the number of epochs, graph 606 shows predictions made using testing data, and graph 608 shows predictions made using training data. The results show that the trained model is able to correctly predict or determine the state of a belt with an accuracy exceeding 80%.

[0041] Once a predictive model has been trained, it may be used to predict a state of a belt based on parameters measured in respect of a motor connected to the belt. Figure 7 is a flowchart of an example of a method 700 of determining a state of a belt of a print apparatus. The method 700 comprises, at block 702, receiving, using a processor, a measurement of a parameter associated with the belt. At block 704, the method 700 comprises providing, using a processor, the measured parameter as an input to a trained classifier. The method 700 comprises, at block 706, obtaining, using a processor, as an output of the trained classifier an indication of a state of the belt. While various classifiers may be trained and used in the method 700, in one example, the classifier may comprise a convolutional neural network model.

[0042] As discussed previously, the belt may, in some examples, comprise a belt to cause rotation of one or more components of a print agent application assembly. For example, the belt may comprise a belt of a BID, such as a belt used to drive the cleaner roller 104 and the sponge roller 106 (see Figure 1). In other examples, the belt may comprise a belt to cause rotation of one or more components of the substrate conveying system. For example, the belt may comprise a belt and exit pickup conveyor of a print apparatus or a belt of another system used to convey or transport a substrate within a print apparatus. The belt may, for example, comprise a timing belt.

[0043] A BID may be installed within a print apparatus in such a way that the BID, or particularly the belt within the BID, cannot be seen easily by an operator without first removing the BID or one or more other components. Furthermore, a BID may include a housing or cover which is to be removed before the belt can be inspected to determine its state manually. Therefore, by applying the method 700 in relation to a belt in the BID, it is possible to determine a state of the belt indirectly, without inspecting the belt visually. If it is determined that the belt is of a state that would benefit from repair, replacement, adjustment or further inspection, then such further tasks may be performed. In some examples, therefore, measurements of a parameter associated with a belt may be obtained at intervals (e.g. hourly) and those measurements may be provided to a trained classifier to determine a state of the belt. In this way, an identification of a state of a belt that might warrant further inspection can be made at an early stage, before any significant deterioration in print quality or efficiency is experienced.

[0044] The parameter associated with the belt may be a parameter of a motor (e.g. a servomotor) associated with or used to drive the belt. The parameter may be measured using a separate component, such as a sensor or detector, and transmitted to the processor. The processor may, in some examples, comprise the processor 114 (see Figure 1) associated with the motor 112 in the BID 102 or the processor 216 (see Figure 2) associated with the motor 210 in the substrate conveying system 202. In other examples, a different processor may be used to perform the functions described in blocks 702 to 706. For example, measured parameters may be transmitted to a remote computer or server for processing. In such examples, the trained classifier may be stored in a memory associated with and accessible by the processor. [0045] The measured parameter may, in some examples, comprise a velocity of a motor used to drive the belt. Alternatively, the measured parameter may comprise a velocity of the belt itself. In other examples, the measured parameter may comprise a torque associated with a motor used to drive the belt.

[0046] Figure 8 is a flowchart of an example of a method 800 of determining a state of a belt of a print apparatus. The method 800 may comprise a block or blocks of the method 700 discussed above. The method 800 may, in some examples, comprise, at block 802, measuring a parameter associated with the belt using a servomotor used to drive the belt. The parameter may be measured prior to it being received by the processor at block 702.

[0047] The output of the trained classifier may be one of a plurality of defined outputs which are indicative of the state of the belt. The state of the belt (i.e. the defined outputs of the classifier) may, in some examples, comprise: i) a suitable operating state; ii) a low tension state wherein a tension of the belt is below a defined threshold; ii) a partly defective state wherein the belt is partly defective and still capable of operating; and iv) a defective state wherein the belt is defective. The suitable operating state may comprise a state in which the belt is fully functional, and is operating correctly, for example, with a tension above a defined threshold. Low tension is may comprise a state in which the belt has a tension lower than or equal to the defined threshold. The partly defective state may comprise a state in which the belt has two adjacent teeth that are worn or missing. The defective state may comprise a state in which the belt has more than two adjacent teeth that are worn or missing.

[0048] Once an indication of the state of the belt has been obtained (block 706), the method 800 may comprise, at block 804, taking an action. For example, the action may be taken responsive to the obtained indication of the state of the belt. The action to be taken at block 804 may depend on the determined or obtained indication of the state of the belt. In one example, at block 806, the method 800 may comprise, using a processor, providing for presentation to a user the obtained indication of the state of the belt. For example, the processor may cause the indication of the state to the belt to be presented to a user (e.g. by displaying the indication of the state of the belt on a display screen) so that the user can decide whether or not any further action should be taken.

[0049] As an alternative to, or in addition to, block 806, the method 800 may in some examples comprise, at block 808, generating, using a processor, a belt adjustment instruction signal based on the determined indication of the state of the belt. The belt adjustment instruction signal may comprise an instruction to a component (e.g. of the print apparatus) to perform a particular action. For example, a signal may be generated which contains instructions which, when executed by a processor or by some other component, cause a belt tensioning adjustment to be made. In one example, an instruction signal may be generated at block 808 and sent or delivered to a component which can increase or decrease the tension of the belt, such as the tensioning mechanism 214 of Figure 2. Thus, if, for example, the method 700, 800 determines that the belt is in a low-tension state, then an automatic adjustment may be made to increase the tension of the belt. In this way, if a belt is determined to be in a particular state that might lead to sub-optimal performance of the print apparatus, then an automatic adjustment may be made to put the belt into different state that may lead to an improved performance.

[0050] According to examples of the present disclosure, an apparatus is also provided. Figure 9 is a schematic illustration of an example of apparatus 900 to determine an operating state of a belt in a print component. The apparatus 900 comprises processing apparatus 902 which may, for example, comprise a processor or processing circuitry. The processing apparatus 902 is to receive (block 904) a measurement obtained in respect of the belt; input (block 906) the received measurement to a classification model, the classification model having been trained to determine an operating state of the belt; and determine (block 908) as an output of the classification model, an indication of an operating state of the belt.

[0051] The measurement received in respect of the belt may, for example, comprise a measurement received from a component associated with the belt, such as a motor used to drive the belt. In some examples, the measurement received may comprise a velocity of the motor and/or a torque of the motor. The classification model to which the obtained measurement is provided may comprise a prediction model or classifier as disclosed herein which may, for example, be trained using machine learning techniques. In one example, the classification model may comprise a convolutional neural network model. The indication of an operating state of the belt that is output by the classification model may comprise, for example, an indication that the belt is operating in a satisfactory manner (e.g. with the tension equal to or exceeding a defined threshold); an indication that the belt is operating and is under low-tension (e.g. the tension is below a defined threshold); an indication that the belt is operating with signs of a defect (e.g. the belt has two teeth that are worn or missing) and an indication belt is effective (e.g. the belt has more than two teeth that are more or missing). [0052] In some examples, as discussed herein, multiple classification models may be used. For example, a first classification model may distinguish between belts operating at a normal (e.g. sufficient) tension and a low tension, and a second classification model may distinguish between belts having two adjacent teeth missing and belts having more than two adjacent teeth missing.

[0053] Figure 10 is a schematic illustration of an example of apparatus 1000 to determine an operating state of a belt in a print component. The apparatus 1000 comprises the processor 902 discussed above, which may perform the functions set out in blocks 904, 906 and 908. The apparatus 1000 may further comprise a print component 1002 having a belt in respect of which an operating state is to be determined. The apparatus 1000 may further comprise a sensor 1004 to obtain a measurement in respect of the belt. The sensor 1004 may, in some examples, obtain the measurement from a component associated with the belt, such as a motor (e.g. a servomotor) used to drive the belt. Measurement obtained by the sensor 1004 may be transmitted to the processing apparatus 902 for processing.

[0054] In some examples, the apparatus 1000 may further comprise an adjustment unit 1006 to effect an adjustment to the belt or to a component associated with the belt, in response to the determined indication of the operating state of the belt. For example, the adjustment unit 1006 may adjust the tension of the belt. In some examples, the tension of the belt may be adjusted by operating a tensioning mechanism (e.g. the tensioning mechanism 214 of Figure 2) to increase or decrease the tension of the belt.

[0055] The apparatus 900, 1000 may comprise, or form part of, a print apparatus.

[0056] According to examples of the present disclosure, a machine-readable medium is also provided. Figure 11 is a schematic illustration of an example of a machine- readable medium 1104 in communication with the processor 1102. Such a machine- readable medium 1104 may comprise instructions which, when executed by the processor 1102, cause the processor to perform various functions described herein. For example, the machine-readable medium 1104 may comprise instructions which, when executed by the processor 1102, cause the processor to perform one or more blocks of the methods 300, 700, 800 described herein. In one example, the machine-readable medium 1104 may comprise: instructions (e.g. measurement receiving instructions 1106) which, when executed by the processor 1102, cause the processor to receive a measurement of a parameter associated with the belt; instructions (e.g. parameter providing instructions 1108) which, when executed by the processor, the processor to provide the measured parameter as an input to a trained classifier; and instructions (output obtaining instructions 1110) which, when executed by the processor, cause the processor to obtain as an output of the trained classifier an indication of a state of the belt.

[0057] Figure 12 is a schematic illustration of a further example of the machine- readable medium 1104 in communication with the processor 1102. In this example, the machine-readable medium 104 may comprise: instructions (e.g. training dataset obtaining instructions 1202) which, when executed by the processor, cause the processor to obtain a training dataset, for example by measuring operating parameters of a plurality of belts operating in print apparatus, each belt having been classified according one of a plurality of operating states; and instructions (e.g. predictive model training instructions 1204) which, when executed by the processor, cause the processor to train the predictive model, using the obtained training data, to classify a belt according to one of the plurality of operating states. The processor 1102 used to execute instructions may comprise a processor 114, 216, 902 described herein, or a separate processor, such as a processing apparatus or processing circuitry in the remote computing system, or in a server as part of a distributed computing environment.

[0058] Examples in the present disclosure can be provided as methods, systems or machine readable instructions, such as any combination of software, hardware, firmware or the like. Such machine readable instructions may be included on a computer readable storage medium (including but is not limited to disc storage, CD-ROM, optical storage, etc.) having computer readable program codes therein or thereon.

[0059] The present disclosure is described with reference to flow charts and/or block diagrams of the method, devices and systems according to examples of the present disclosure. Although the flow diagrams described above show a specific order of execution, the order of execution may differ from that which is depicted. Blocks described in relation to one flow chart may be combined with those of another flow chart. It shall be understood that each flow and/or block in the flow charts and/or block diagrams, as well as combinations of the flows and/or diagrams in the flow charts and/or block diagrams can be realized by machine readable instructions.

[0060] The machine readable instructions may, for example, be executed by a general purpose computer, a special purpose computer, an embedded processor or processors of other programmable data processing devices to realize the functions described in the description and diagrams. In particular, a processor or processing apparatus may execute the machine readable instructions. Thus functional modules of the apparatus and devices may be implemented by a processor executing machine readable instructions stored in a memory, or a processor operating in accordance with instructions embedded in logic circuitry. The term‘processor’ is to be interpreted broadly to include a CPU, processing unit, ASIC, logic unit, or programmable gate array etc. The methods and functional modules may all be performed by a single processor or divided amongst several processors.

[0061] Such machine readable instructions may also be stored in a computer readable storage that can guide the computer or other programmable data processing devices to operate in a specific mode.

[0062] Such machine readable instructions may also be loaded onto a computer or other programmable data processing devices, so that the computer or other programmable data processing devices perform a series of operations to produce computer-implemented processing, thus the instructions executed on the computer or other programmable devices realize functions specified by flow(s) in the flow charts and/or block(s) in the block diagrams.

[0063] Further, the teachings herein may be implemented in the form of a computer software product, the computer software product being stored in a storage medium and comprising a plurality of instructions for making a computer device implement the methods recited in the examples of the present disclosure.

[0064] While the method, apparatus and related aspects have been described with reference to certain examples, various modifications, changes, omissions, and substitutions can be made without departing from the spirit of the present disclosure. It is intended, therefore, that the method, apparatus and related aspects be limited only by the scope of the following claims and their equivalents. It should be noted that the above- mentioned examples illustrate rather than limit what is described herein, and that those skilled in the art will be able to design many alternative implementations without departing from the scope of the appended claims. Features described in relation to one example may be combined with features of another example.

[0065] The word“comprising” does not exclude the presence of elements other than those listed in a claim,“a” or“an” does not exclude a plurality, and a single processor or other unit may fulfil the functions of several units recited in the claims.

[0066] The features of any dependent claim may be combined with the features of any of the independent claims or other dependent claims.