202207727 1 Description A system and method for early detection of a leak in a fluid containing structure The invention relates to detection of a leak in a fluid containing structure, e.g. a water, oil, or gas distribution system, especially with focus on early prediction of leakage. Leakage in a fluid containing structure, e.g. a pipeline, a container, or a system of interconnected pipelines, containers, and/or other vessels, might result in a loss of the fluid. In case the fluid is a harmless substance and the quantity of lost fluid is low, the damage resulting from the loss might remain moderate and essentially limited to the financial value of the lost fluid. However, in case the fluid is hazardous and/or the quantity of lost fluid is substantial, severe damages of the environment of the fluid containing structure are expectable besides the pure financial damage. In any case, an early detection of leakage is of highest interest. For example, the fluid containing structure might be a fluid guiding pipeline system guiding a fluid from one or more points A to one or more points B, wherein the fluid might be water, oil, or some other liquid as well as a gas. In one of several imaginable concrete embodiments, the fluid containing structure is a water transmission and distribution network. Such a pipeline system typically comprises a plurality of interconnected vessels, e.g. tubes and/or containers. In a less complex embodiment, the fluid containing structure can be a simple tube or duct connecting two points A and B. In general terms, the fluid containing structure is a structure which contains a fluid or which guides a fluid. Leakage in the fluid containing structure might arise due to multiple reasons, for example wear and tear, mechanical interference, pressure spikes, faulty elements, etc. of the 202207727 2 vessels of the structure. As indicated above, an early leakage detection is of high interest, but fluid containing structures are often buried underground and/or are extremely extensive, i.e. large with lengths of several kilometers and complex, so that detection and localization of the leak becomes a challenging task. One solution to observe a fluid containing structure with regard to occurrence of leaks applies the “acoustic leak detection” (ALD) approach, which shall be the method of choice to detect a leak in the solution proposed herein as well. Just for example, an ALD method is disclosed in WO2014014378A1 and in US2022082467. Based on that and on further prior art it can assumed that the ALD method the its way of working is sufficiently well known. The detection via ALD might include the pure determination whether a leak is present or not and optionally the localization of such a leak. ALD utilizes one or more acoustic loggers, e.g. microphones, installed all over the fluid containing structure at suitable locations to record sound in the structure. In principle, ALD is based on differentiating leak sound, caused by the interaction of the fluid with the leak, from the regular sound or “noise” arising during normal operation of the fluid containing structure, e.g. caused by operation of devices of the fluid containing structure and/or caused by regular fluid behavior in the vessel, e.g. the fluid flowing through the vessel without abnormal disturbances. Therein, the leak sound is typically based on abnormal disturbances in the fluid in connection with the leak. For example, such abnormal disturbances can be caused by a part of the fluid streaming through the leak. Also, sound generating abnormal disturbances in the fluid can arise when the fluid flows along a section of a surface with irregularities due to the leak. However, the determination of leak sound and identification of location of leaks in the fluid containing structure by means of acoustic leak detection ALD is a challenging problem 202207727 3 because the measurable sound is an overlay of the regular sound and potential leak sound so that the ALD approach is typically disturbed by regular noise produced by various network elements of the fluid containing structure (in this context the regular sound is considered as noise). On the one hand, such noise might be interpreted by an evaluation system to be caused by an alleged leak such that false alarms are initiated. On the other hand, the noise might exceed the acoustic leak sound so that the leak cannot be detected. Both scenarios disturb reliable leakage detection. This is especially true in case an early leak detection is intended because in an early stage the leak might still be small so that sound and noise caused by the leak not yet dominate the overall sound in the structure and is, therefore, not detectable at the early stage. Therefore, a solution is required which enables a reliable, early detection of a leak while, at the same time, the risk of false alarms is minimized. This is solved by the method suggested in claim 1, by the system as per claim 10, and by the fluid containing structure of claim 12. A leakage determination method for early detection of a leak LK in a fluid containing structure with an acoustic leak detection (ALD) method foresees to monitor noise SND(t) in the fluid containing structure over time t by measuring the noise with one or more sensors. Preferably, the sensors are acoustic loggers or microphones as regularly used by typical acoustic leak detection (ALD) approaches. Based on the monitors noise SND(t), a trend measure TREND of the measured noise SND(t) over time t is determined and a presence of a leak LK is assumed depending on the determined trend measure TREND, i.e. certain properties of the trend TREND(t), e.g. a slope, are evaluated to conclude whether a leak LK exists or not. This solution enables predictive or early leak identification with low risk of false alarms. 202207727 4 The monitoring of the noise SND(t) can include a creation of a time series ti with i=1,2,…,I and I≥2 of noise measures NOI(ti), e.g. volumes of the respective noises SND(ti), by a respective series of I measurements of the noise SND(t), which shall include a continuous noise measurement with one or more suitable sensors over time t. Therein, each noise measure NOI(ti) of the time series ti represents noise SND(ti) in the structure at a point ti in time at a position of the respective sensor. The monitoring of the noise SND(t) can also be performed with a plurality of sensors. In that case, the noise measures NOI(ti) are combined noise measures, wherein each sensor provides individual noise measures NOI _201’ (ti), NOI _201” (ti), each individual noise measure NOI _201’ (ti), NOI _201” (ti) representing noise SND _201’ (ti), SND _201” (ti) in the structure (100) at a point ti in time at the positions of the respective sensors 210’, 201”. The individual noise measures NOI _201’ (ti), NOI _201” (ti) of different sensors (201’, 201”) representing the same point ti in time are combined to form the combined noise measures NOI(ti). The combination might be a summing up or an averaging of individual noise measures NOI _201’ (ti), NOI _201” (ti) to for the respective combined noise measure NOI(ti). With this, the accuracy of the approach is further improved. The trend measure TREND(t) can be determined by an interpolation of a certain subset SUBNOI(tj) of the time series ti of noise measures NOI(ti), wherein the subset SUBNOI(tj) with j=1,2,…,J and 2≤J≤I comprises at least two and at most all noise measures NOI(tj) from the entirety of noise measures NOI(t) of the time series ti. Therein, the noise measures NOI(tj) for the subset SUBNOI(tj) are selected from a basic set BASE of noise measures NOI(tj) such that a correlation coefficient r _201’ of the interpolation is optimized. 202207727 5 As an alternative, the subset SUBNOI(tj) can be composed of the J=NMAX most recent noise measures NOI(tj) selected from a basic set BASE of noise measures NOI(tj), i.e. the NMAX youngest noise measures NOI(tj) from the basic set BASE are selected to compose the subset SUBNOI(tj). Correspondingly, noise measures acquired earlier might not be considered any more. Therein, the basic set BASE of noise measures NOI(tk) can comprise the entirety of noise measures NOI(ti) of the time series or only those noise measures NOI(tk) from the entirety of noise measures NOI(ti) which represent noises SND(tk) which have been measured by the sensor at points tk in time at which intrinsic background noise SND(100) in the fluid containing structure at the location of the sensor was below a predefined threshold value THRES_NOI. Therein, the “intrinsic” background noise is noise caused by and in the fluid containing structure itself, e.g. by operation of devices and/or by the fluid flowing through the structure and/or by external influences. The volume or other properties of such intrinsic noise can be more or less disturbing and such influence can be excluded by the proposed measure. The trend measure TREND can be a time dependent function TREND(t), wherein the presence of the leak LK is assumed in case a slope M of the trend measure TREND(t) in its development over time t is positive, preferably in case the slope M is greater than a predefined minimum value Mmin, i.e. M>Mmin. This reduces the risk of false alarms, especially in a section of the trend measure TREND(t) curve representing the most recent time and at least over a minimum time span covering a suitable number NMP of noise measurement points, e.g. NMP=NMAX. It is well known that the “suitable” number NMP depends on the aspired and required reliability and exactness of the trend measure. 202207727 6 In case a slope M of the trend measure TREND(t) in its development over time t is positive, preferably in case the slope M is greater than a predefined minimum value Mmin, and, therewith, indicates an increase of the noise measures NOI(t) of the time series ti over time t, and the presence of the leak LK is assumed at a present point ta in time, a future point tx in time is estimated based on the trend measure TREND(t), e.g. utilizing the trend measure TREND(t) to extrapolate into the future, which represents the moment tx in time at which the noise measure NOI(tx) is expected to exceed a predefined tolerance value TOL_NOI. Regarding ta and tx, for example, the “present” point ta in time can be the point in time of the most recent noise measurement or it can be the point in time of the most recent determination of the trend measure TREND. Essentially, a temporal distance dT=tx- ta shall provide a measure for assessing how much time remains until the noise measure NOI(t) is expected to exceed the predefined tolerance value TOL_NOI. An alarm status for an operator of the fluid containing structure is only generated in case the temporal distance dT=tx-ta is less than a predefined threshold THRES_T, i.e. in case dT<THRES_T. Therein, THRES_T might be set to be, for example, a few days or hours, depending on the nature of the fluid containing structure and/or depending on the slope M. In case of a greater slope M, the respective alarm signal might be generated at an earlier stage, i.e. THRES_T is relatively small. while a lower slope M might allow further observation of the development to avoid false alarms, i.e. THRES_T might be set at a higher value. A corresponding leakage determination system for early detection of a leak LK in a fluid containing structure comprises at least one sensor for measuring a time series ti of noise SND(100)(t) in the fluid containing structure, resulting in noise measures NOI _201’ (ti), and an evaluation system for receiving and subsequent processing of the time series ti of noise measures NOI _201’ (ti) provided by the sensor 202207727 7 (201’), wherein the evaluation system (220) is configured to perform the method described above. Preferably, a fluid containing structure comprises such a leakage determination system. For example, the fluid in the structure is water or oil or gas and the structure is a network for providing the fluid from one or more sources to one or more consumers. The proposed approach is based on monitoring recorded noise as a sparse time series wherein the analysis and evaluation of the time series dos not only consider a current measurement value, but several measurements from prior dates. It will be possible to initiate an alarm status irrespective of whether the noise is above a certain threshold. The method allows to detect leaks at an early stage and, therewith, realizes a predictive approach, e.g. for detection of slow developing leaks. In summary, the proposed approach allows to run predictive non-invasive leak detection and to identify the small leaks, i.e. leaks causing noise below predefined thresholds as well as leaks at an early stage of development, both reducing overall fluid losses. It is to be understood that the elements and features recited in the appended claims may be combined in different ways to produce new claims that likewise fall within the scope of the present invention. Thus, whereas the dependent claims appended below depend from specific independent or dependent claims, it is to be understood that these dependent claims can, alternatively, be made to depend in the alternative from any preceding or following claim, whether independent or dependent, and that such new combinations are to be ^{understood as forming a part of the present specification.} 202207727 8 DESCRIPTION OF THE FIGURES In the following, possible embodiments of the different aspects of the present invention are described in more detail with reference to the enclosed figures. The objects as well as further advantages of the present embodiments will become more apparent and readily appreciated from the following description of the preferred embodiments, taken in conjunction with the accompanying figure in which: FIG 1 shows a fluid containing structure; FIG 2 shows a time development of noise measures; FIG 3 shows a flow chart of the leakage determination method. DETAILED DESCRIPTION FIG 1 is a simplified and reduced visualization of a fluid containing structure 100 with one or more vessels 101 for containing a fluid 110 and/or for guiding the fluid 110 from one or more points A, e.g. sources of the fluid, to one or more points B, e.g. consumers. In reality, the structure 100 can be much more complex. The arrows in the vessels indicated the direction of flow of the fluid from an upstream direction to a downstream direction. The vessels 101 can be, for example, pipes, tubes, or passages 101p, containers or reservoirs 101c, and/or further typical and/or imaginable structures 101x which might be required to fulfill the purpose of the fluid containing structure 100. However, such pipes, containers, and structures etc. will be subsumed in the following under the term “vessels” 101. Some of those vessels 101 can be interconnected at respective crossings. Moreover, the fluid containing structure 100 comprises one or more types of active devices 102, e.g. valves 102v, pumps 102p, and/or further typical and/or imaginable elements 102x 202207727 9 which might be required to fulfill the purpose of the fluid containing structure 100. However, such valves 102v, pumps 102p, and/or elements 102x will be subsumed in the following under the term “devices” 102. The devices 102 can be used to manipulate a flow of the fluid 110. Control of the devices 102 is performed by a control system 120 of the fluid containing structure 100. The control system 120, which can be a computer implemented system with a processor executing a respective software, is configured to control the devices 102 based on input data. The input data processed by the control system 120 can be data from an operator of the fluid containing structure 100 and/or data from sensors 103 positioned in the fluid containing structure 100 at positions of interest or relevance, e.g. along the vessels 101 and devices 102, e.g. positioned at critical crossings, important devices 102, and/or further positions of interest or relevance. Typically, but not necessarily, each device 102 would be equipped with a respective sensor 103 which provides relevant operational data of the assigned device 102. Therewith, the typical fluid containing structure 100 comprises a control system 120, one or more vessels 101, one or more devices 102, and/or sensors 103. In practice, the concrete setup and architecture of a particular fluid containing structure 100 depends on the purpose and field of use of the structure 100. Thus, in an exemplary minimal realization of the fluid containing structure 100 still underlying the present invention, the structure 100 merely has to contain and hold the fluid 110, i.e. it indeed just operates as a container. In that case, the structure 100 might only comprise one container vessel 101, but no further vessels. A slightly enhanced realization of this minimal realization might additionally comprise a valve 102 for releasing the fluid from the container vessel 101 and/or for letting additional 202207727 10 fluid 110 into the container vessel 101. In that case, a control system 120 for controlling the valve 102 might be foreseen or the valve 102 might be operated manually. In a more complex realization of the fluid containing structure 100 still underlying the present invention, the structure 100 comprises a plurality of vessels 101, a plurality of devices 102, a plurality of sensors 103, and a control system 120. Exemplarily, but not limiting the invention, such a complex fluid containing structure 100 can be embodied as a water guiding pipeline system (WGPS) 100. The WGPS 100 can be used to systematically guide the fluid 110, in this case water 110, from sources A to consumers B via the vessels 101, with the help of the devices 102, and controlled by the control system 120. Of course, a similar structure 100 can be used to guide other fluids like oil or gas etc. from A to B. In any realization of the fluid containing structure 100, occurrence of a leak LK might result in the damages and risks indicated above. Therefore, an early detection of leakage is of highest interest. Therein, “detection” might mean the pure determination whether a leak is present or not and optionally the localization of such a leak. For that reason, the fluid containing structure 100 is equipped with a leakage determination system 200 which is embodied and configured to apply the acoustic leak detection approach ALD as known in prior art. The ALD approach typically foresees to acoustically measure sound SND in the fluid containing structure 100 with the help of one or more acoustic loggers 201, e.g. embodied as corresponding microphones, and to analyze the measured sound SND, e.g. its volume, i.e. the loudness, to identify potential contributions to the sound SND caused by a presumable leak LK. Alternatively or additionally, the spectrum of the measured sound SND might be a basis for further analysis to detect leakage. However, an approach only based on volume is 202207727 11 a simplification, but it yields acceptable results for many cases. In general terms, a representation r(SND) or noise measure NOI, respectively, of the sound SND is measured, wherein the noise measure NOI might be, for example, the volume and/or the spectrum of the noise SND. In the scenario described herein, it is not of highest relevance which one or more of the noise properties are analyzed and it can be assumed, for the sake of simplicity, that the analysis described below is based on noise volume. The leakage determination system 200 comprises one or more acoustic loggers 201 installed throughout the fluid containing structure 100 at suitable locations as well as an evaluation system 220. The evaluation system 220, which can be a computer implemented system with a processor executing a respective software, can be an independent system or it can be implemented in the control system 120, if available. Each installed acoustic logger 201 acoustically measures current noise SND ₂₀₁ (ti) at the location of the acoustic logger 201 in the structure 100 at points ti in time t, e.g. at regular or irregular intervals and/or initiated by the evaluation system 220, and provides a corresponding noise measure NOI ₂₀₁ (ti) to the evaluation system 220 for further processing. Thus, a noise measure NOI _201’ (ti) provided by a particular acoustic logger 201’ to the evaluation system 220 is a measure of the real noise SND(ti) at the location of the acoustic logger 201’ at the point ti in time. The logger 201’, e.g. a microphone, only detects one or more parameters of the real noise SND _201’ (ti), but not the noise SND(ti) itself and the noise measure NOI _201’ (ti) can represent, for example, the volume, the spectrum, or some other property of the noise SND _201’ (ti) or it might be composed of these and/or other parameters which allow a reliable conclusion on the real noise SND _201’ (ti) at the location of the logger 201’ at the point ti in time. 202207727 12 The noise measurements can be sensitive for noise SND occurring both in the structure 100 itself, i.e. in the walls of the vessels 101 etc., and occurring in the fluid 110. For example, some acoustic loggers 201 might be arranged and configured such that they essentially measure noise in the fluid 110 while other acoustic loggers 201 are arranged and configured such that they essentially measure noise in the structure 100 itself. Alternatively or additionally, some or all acoustic loggers 201 can be arranged and configured such that they measure noise both in the structure 100 itself and in the fluid 110. In any case, the loggers 201 act as acoustic sensors for noise SND in the fluid containing structure 100 and the respective sensor measurement data, i.e. the noise measure NOI of the noise SND, are transferred to the evaluation system 220 for further processing for leakage determination. The processing and analysis of the noise measure NOI of the acoustic loggers 201 in the evaluation system 220 might result in the insight that the fluid containing structure 100 comprises a leak LK, e.g. in case the sound SND measured by one or more loggers 201’ shows suspicious volume or spectrum etc. in the noise measure NOI. In that case, the leakage determination system 200 is configured to initiate an alarm status which might, for example, notify an operator of the fluid containing structure 100 of the potentially critical situation and/or which might result in initialization of automatic measures like closing valves, stopping pumps, and/or shutting down the overall system. As night is usually a sufficiently quiet period with lowest regular noise SND(100) in the structure 100, typical acoustic measurements with acquisition of noise measures NOI take place at night in order to avoid unwanted disturbances, i.e. noises not related to leaks, so that potential leak sound SND(LK) can be distinguished from regular noise SND(100) more easily. Nevertheless, early and unambiguous identification of the presence of a leak LK with low risk of false alarms 202207727 13 during the processing and analysis of the noise measures NOI of the acoustic loggers 201 in the evaluation system 220 is challenging and/or might even involve relatively high manual efforts because, as indicated above, the overall noise SND and noise measure NOI measurable by the acoustic loggers 201 is an overlay of potential leak sound SND(LK) caused by a leak LK with the regular noise SND(100) of the fluid containing structure 100. Thus, the leakage determination system 200 has to be able to identify leak sound SND(LK) despite the presence of regular, unavoidable noise SND(100), respectively. The noise SND(100) can be caused by different sources. For example, the devices 102 of the structure 100 can be sources of acoustic noise SND(102) contributing to the noise SND(100), i.e. the noise SND(100) is essentially an overlay of the regular noises SND(102) generated by those devices 102 during operation of the structure 100. One concrete example for a regular acoustic noise SND(102) causing device 102 which is typically present in a fluid containing structure 100 might be a pump which operates at a certain frequency and which generates corresponding noise SND(102). Another concrete example for a regular noise SND (102) causing device 102 might be a pressure reducing valve (PRV) which is often used to reduce an operating pressure at suitable locations in the fluid containing structure 100. In certain operating regimes, the flow of fluid 110 through a PRV might produce an acoustic noise SND(PRV) with properties similar to the noise SND(LK) caused by a leak LK. Moreover, the vessels 101 themselves and especially elements of the fluid containing structure 100 which have an impact on the flow properties of the fluid 110 like the crossings or curvatures of vessels 101 give rise to disturbances and, therewith, to acoustic noise contributing to the regular sound. In general terms, the fluid containing structure 100 comprises a plurality of sources SRC of regular acoustic noise, especially including vessels 101 and/or devices 102, but also external sources of acoustic interference the structure 100 like external noises 202207727 14 from nearby traffic, e.g. cars or trains, or any activity by industries and people. However, in case the leakage determination system 200 has initiated the alarm status, a subsequent field inspection might be performed. During the field inspection, an expert assesses essentially manually whether a regular noise causing device 102, e.g. a PRV, is topologically in the vicinity of that acoustic logger 201’ that caused the alarm status. In case such PRV is near the logger 201’ and a noise measure NOI _201’ provided by that logger 201’ is proven to be dominated by noise SND(PRV) caused by the PRV, the alarm is classified as a false alarm and the alarm status can be reset. In specific cases additional investigations might be required to identify whether the alarm status is based on a false alarm and suitable network operations have to be executed. Such network operations typically require closing of the PRV under investigation to stop the flow through the PRV, which might result in isolating some consumers downstream of the PRV. In case the PRV is the source of the noise, such an experiment allows to stop the cavitation noise and to confirm that the noise was coming from the PRV. An alternative approach is to perform the sound measurements in close proximity to the location of the PRV to be able to unambiguously distinguish whether sound is generated by the PRV or not. Thus, the unambiguous, reliable identification of the presence of a leak LK with low risk of false alarms during the processing and analysis of the measurement data NOI of the acoustic loggers 201 in the evaluation system 220 is therefore challenging and/or effortful due to the presence of regular noise SND(100) generated during normal operation of the fluid containing structure 100 which can hinder the identification of sound SND(LK) caused by a leak LK. Moreover and in addition to the requirement of reliable leak detection despite the presence of noise SND(100), a reliable, early indication that a leak LK seems to be present is of high interest in order to be able to initiate counter measures at 202207727 15 an early stage to avoid damages before the leak gets even larger. The solution for reliable, early leakage detection proposed herein is based on the approach to observe and monitor noise SND(t) in the fluid containing structure 100 over time t. For that purpose, one or typically more acoustic loggers 201 are installed at suitable positions in the structure 100. Each logger 201 measures the noise SND ₂₀₁ (t) continuously or at least at a plurality of points ti in time t and provides the respective plurality of noise measures NOI ₂₀₁ (ti) to the evaluation system 220. For example, consecutive points t1, t2, t3,… in time t might be, but don’t have to be, equally distanced from each other. In any case, the point ti in time t for which a particular acoustic logger 201’ has measured noise SND _201’ (ti) would be provided to the evaluation system 220 as well, so that the evaluation system 220 is enabled to process the provided noise measures NOI ₂₀₁ (ti) with the respective points ti in time. In the following, a particular acoustic logger 201’ will be considered, but the explanations can be applied for each one if the acoustic loggers 201 installed in the fluid containing structure 100. The acoustic logger 201’ monitors the fluid containing structure 100 by measuring the noise SND _201’ (t) in the structure 100 at the location of the logger 201’ in a time series, i.e. over time t. As indicated, the respective noise measures NOI _201’ (ti) for consecutive points ti in time t with i=1,2,…,I as well as the measurement times ti themselves are provided to the evaluation system 220. The entirety of noise measures NOI _201’ (ti) provided to the evaluation system 220 forms a basic set BASE _201’ of noise measures NOI _201’ (ti), which is growing over time. The diagram depicted in FIG 2 shows exemplary noise measures NOI _201’ (ti) provided by the particular logger 201’ over time 202207727 16 t. The first noise measure NOI _201’ (t1) has been measured at a point t1 in time, while the most recent measurement to provide noise measure NOI _201’ (t9) has been executed at a point t9 in time, e.g. “now”, i.e. I=9 at the moment. FIG 2 assumes that an expanding leak LK is present in the structure 100. Correspondingly, the noise SND _201’ (t) at the location of the particular acoustic logger 201’ and the respective noise measure NOI ₂₀₁ ’(t) increase over time t. In more detail, an increase of the noise measures NOI ₂₀₁ ’(ti) is visible for points t5-t9 in time, while noise measures NOI ₂₀₁ ’(ti) for points t1-t4 in time are essentially stable and do not yet indicate a presence of the leak LK. The evaluation system 220 might be configured such that an alarm status is not initiated before the noise measure NOI _201’ (t) has exceeded a predefined tolerance threshold TOL_NOI, for example in order to avoid false alarms. Since none of the noise measures NOI ₂₀₁ ’(t) until and including point t9 in time t exceeds the threshold TOL_NOI, the alarm status is not yet initiated by the evaluation system 220 although an increase of noise measures is visible. Therefore, a risk arises that the expanding leak LK will be detected at a point in time at which severe damages have become unavoidable. The solution provided herein foresees that the evaluation system 220 processes certain noise measures NOI ₂₀₁ ’(ti) and the respective points ti in time received from the acoustic logger 201’ to determine a trend measure TREND _201’ (t) of the measured noise SND _201’ (t) represented by the noise measure NOI ₂₀₁ ’(t). The trend measure TREND _201’ (t) is a time dependent function, which is generated by an interpolation of a data set DATA _201’ consisting of a subset SUBNOI ₂₀₁ ’(tj) of the basic set BASE _201’ of noise measures NOI _201’ (ti) and the points tj in time corresponding to the noise measures NOI _201’ (tj) of the subset SUBNOI ₂₀₁ ’(tj), i.e. the points tj in time at which the measurements for determining the noise measures NOI _201’ (tj) have been performed. Preferably, a linear interpolation is applied. The presence of the leak LK can be assumed at an 202207727 17 early stage and before the noise measure NOI _201’ (t) exceeds the tolerance threshold TOL_NOI by evaluating the determined trend measure TREND _201’ (t), i.e. certain properties of the trend measure TREND _201’ (t) are evaluated in the evaluation system 220 to conclude whether a leak LK exists or not. The subset SUBNOI ₂₀₁ ’(tj) with j=1,2,…,J and 2≤J≤I comprises at least J=2 and at most all, i.e. J=I, noise measures NOI ₂₀₁ ’(ti) from the entirety of noise measures NOI ₂₀₁ ’(ti) of the time series. The selection of particular noise measures NOI ₂₀₁ ’(tj) from the basic set BASE _201’ of noise measures NOI _201’ (ti) to be included in the subset SUBNOI ₂₀₁ ’(tj) has the objective to create a subset SUBNOI ₂₀₁ ’(tj) the interpolation of which yields a trend measure TREND _201’ (t) with high reliability and accuracy, representing the development of NOI ₂₀₁ ’(ti) over time in the best possible way, for example expressed by the correlation coefficient of the linear interpolation. For example and as shown in FIG 2, the trend measure TREND _201’ (t) is a linear interpolation of a data set DATA _201’ consisting of the noise measures {NOI _201’ (t4), NOI _201’ (t5), NOI _201’ (t6), NOI _201’ (t7), NOI _201’ (t8), NOI _201’ (t9)} and corresponding points {t4, t5, t6, t7, t8, t9} in time. In that case, the respective subset SUBNOI ₂₀₁ ’(tj) only includes the noise measures NOI _201’ (t4), NOI _201’ (t5), NOI _201’ (t6), NOI _201’ (t7), NOI _201’ (t8), and NOI _201’ (t9), so that the remaining noise measures NOI _201’ (t1), NOI _201’ (t2), NOI _201’ (t3) are not considered in the interpolation. If those remaining noise measures NOI _201’ (t1), NOI _201’ (t2), NOI _201’ (t3) would be considered in the linear interpolation, the respective correlation coefficient r _201’ would indicate a worse accuracy of the trend measure TREND _201’ (t). In one exemplary embodiment, the selection of particular noise measures NOI ₂₀₁ ’(tj) from the basic set BASE _201’ of noise measures NOI _201’ (ti) to be included in the subset 202207727 18 SUBNOI ₂₀₁ ’(tj) can be based on optimizing the correlation coefficient r _201’ of the respective interpolation. In another exemplary embodiment for the selection of particular noise measures NOI ₂₀₁ ’(tj) from the basic set BASE _201’ of noise measures NOI _201’ (ti) to be included in the subset SUBNOI ₂₀₁ ’(tj), the subset SUBNOI ₂₀₁ ’(tj) is composed of the J=NMAX most recent noise measures NOI ₂₀₁ ’(tj) of the basic set BASE _201’ , i.e. the NMAX youngest noise measures NOI ₂₀₁ ’(tj) are selected from the basic set BASE _201’ to compose the subset SUBNOI ₂₀₁ ’(tj). Correspondingly, noise measures NOI ₂₀₁ ’(t) acquired earlier in time might not be considered any more. In the example shown in FIG 2, NMAX=6 is applicable, so that only the noise measures NOI _201’ (t4), NOI _201’ (t5), NOI _201’ (t6), NOI _201’ (t7), NOI _201’ (t8), and NOI _201’ (t9) are used to compose SUBNOI ₂₀₁ ’(tj). As soon as an additional noise measure NOI _201’ (t10) is available, the earlier noise measure NOI _201’ (t4) would drop out of SUBNOI ₂₀₁ ’(tj), being replaced by NOI _201’ (t10). In a further exemplary embodiment, all noise measures NOI ₂₀₁ ’(tj) from the basic set BASE _201’ of noise measures NOI _201’ (ti) are included in the subset SUBNOI ₂₀₁ ’(tj). Of course, other approaches to compose a subset SUBNOI ₂₀₁ ’(tj) might be imaginable. The basic set BASE _201’ of noise measures NOI _201’ (ti) might comprise the entirety of noise measures NOI _201’ (ti) provided by the particular acoustic logger 201’ so that each and every noise measure NOI ₂₀₁ ’(ti) provided by the logger 201’ is available and selectable, respectively, for building the subset SUBNOI ₂₀₁ ’(tj) in any one of the embodiments. However, the volume or other properties of intrinsic, typically fluctuating background noise in the fluid containing structure 100, for example caused by any kind of source SRC during regular operation as described above, can be more or less disturbing and overlaying the measurement of noise 202207727 19 SND _201’ (t) by the acoustic logger 201’ at certain times. Therefore, the basic set BASE _201’ out of which noise measures NOI ₂₀₁ ’(tk) can be selected to build the subset SUBNOI ₂₀₁ ’(tk) as foreseen in the exemplary embodiments offers only those noise measures NOI ₂₀₁ ’(tk) from the entirety of noise measures NOI ₂₀₁ ’(ti) for selection which represent noises SND _201’ (tk) which have been measured by the acoustic logger 201’ at points tk in time at which the intrinsic background noise SND(100)(tk) in the fluid containing structure (100) at the location of the acoustic logger 201’ was below a predefined threshold value THRES_NOI. Other noise measures noise measures NOI ₂₀₁ ’(ti) not fulfilling this condition might be deleted from the basic set BASE _201’ . The determination of intrinsic background noise SND(100)(ti) at points ti of time is not a core aspect of the solution provided herein. However, such insights might be achieved based on measurements with an additional sensor system 202 of the leakage determination system 200. Such additional sensor system 202 might include sensors 202a which are arranged and installed in the environment of the fluid containing structure 100 in order to determine background noise form peripheral structures. Other additional sensors 202b of the additional sensor system 202 might be arranged close to or integrated in potentially noise causing devices 102 so that they can determine background noise cause by such devices. Further additional sensors 202c of the additional sensor system 202 might be arranged throughout the fluid containing structure 100 at locations which are suspicious or known for generation of fluctuating background noise. The sensor data SENS(t) provided by the additional sensor system 202 can be provided to the evaluation system 220 which processes them to determine the intrinsic background noise SND(100)(t) over time. In case the intrinsic background noise SND(100)(ta) at a point ta in time exceeds the predefined threshold value THRES_NOI, the respective noise measures NOI _201’ (ta) would, not be included in the basic set BASE _201’ . 202207727 20 Thus, various approaches and consideration for composing the subset SUBNOI _201’ (tj) for determining the trend measure TREND _201’ (t) are imaginable. In any case, coming back to the detection of the leak LK, it has been mentioned earlier that the presence of the leak LK can be assumed at an early stage and even before the noise measure NOI _201’ (t) exceeds the tolerance threshold TOL_NOI by evaluating the determined trend measure TREND _201’ (t) which is generated by an interpolation of the data set DATA _201’ consisting of the subset SUBNOI ₂₀₁ ’(tj) of the basic set BASE _201’ of noise measures NOI _201’ (ti) and the respective points tj in time. For example, in case the trend measure TREND _201’ (t) is determined by a linear interpolation of the data set DATA _201’ , the respective slope M _201’ of the trend measure TREND _201’ (t) can be an indicator for the presence of a leak. Such presence of a leak LK might be assumed in case the slope M _201’ of the trend measure TREND(t) in its development over time t is positive because this indicates an increase of noise SND _201’ (100). Preferably, an additional precondition would be that the slope M _201’ is not only positive, but also exceeds a predefined minimum value Mmin, i.e. M _201’ >Mmin, to reduce the risk of false alarms. This consideration regarding the slope M _201’ is especially applicable in a section of the trend measure TREND(t) representing the most recent time and at least over a minimum time span, covering a suitable number NMP of noise measurement points, e.g. NMP=NMAX. It is well known that the “suitable” number NMP depends on the aspired and required reliability and exactness of the trend measure. In case the slope M _201’ fulfills the condition M _201’ >0, preferably M _201’ >Mmin, the evaluation system 220 might assume the presence of a leak LK and initiate an alarm status or similar, possibly including an alarm signal and/or other measures. As a further consideration, the evaluation system 220 might be configured to estimate a future point tx in time which represents the moment tx in time at which the noise measure NOI _201’ (t) is expected to exceed the predefined 202207727 21 tolerance value TOL_NOI. The estimation can apply the trend measure TREND _201’ (t) by utilizing the trend measure TREND _201’ (t) to extrapolate into the future. The precondition for initiating the alarm status does not only include M _201’ >0, preferably M _201’ >Mmin, as a necessary condition in this example, but it requires that a temporal distance dT=tx-ta between the estimated moment tx in time and a present point ta in time is less than a predefined threshold THRES_T, i.e. in case dT<THRES_T. For example, the “present” point ta in time can be the point in time of the most recent noise measurement, i.e. ta=t9 in the example of FIG 2, or it can be the point in time of the most recent determination of the trend measure TREND _201’ (t). Essentially, the temporal distance dT shall provide a measure for assessing how much time remains until the noise measure NOI _201’ (t) is expected to exceed the predefined tolerance value TOL_NOI. Therein, THRES_T might be set to be, for example, a few days or hours, depending on the nature of the fluid containing structure 100 and/or depending on the slope M _201’ . In case of a greater slope M _201’ , the respective alarm signal might be generated at an earlier stage, i.e. THRES_T is relatively small, while a lower slope M _201’ might allow further observation of the development to avoid false alarms, i.e. THRES_T might be set at a higher value. With that, the leakage determination system 200 and any one of the explained configuration options of the evaluation system 220 allows a reliable, early detection of a leak LK. Essentially and shown in FIG 3, the leakage determination method LDM comprises a step LDM1 of monitoring noise SND(t) in the fluid containing structure 100 over time t by measuring the noise with acoustic loggers, a step LDM2 of determining a trend measure TREND of the measured noise 202207727 22 SND(t) over time t, and a step LDM3 of assuming the presence of a leak LK depending on the determined trend measure TREND. It is possible to execute the above separately for different acoustic loggers 201’, 201” so that a corresponding number of independent conclusions whether a leak is present or not is created. However, it is imaginable to combine measurements of several acoustic loggers 201’, 201”, e.g. by adding noise measures NOI _201’ (ti), NOI _201” (ti) or by forming average values of individual noise measures NOI _201’ (ti), NOI _201” (ti) etc. In the latter case, each noise measure NOI(ti) in the basic set BASE is a combination COMB(NOI _201’ (ti), NOI _201” (ti),…) of the noise measures NOI _201’ (ti), NOI _201” (ti),… determined with acoustic loggers 201’, 201”, etc. Preferably, the acoustic loggers 201’, 201”, the noise measures of which are combined by summing or averaging etc. are located in the fluid containing structure 100 relatively close to each other to ensure that background noise SND(100) at their locations is similar. While the present invention has been described above by reference to various embodiments, it should be understood that many changes and modifications can be made to the described embodiments. It is therefore intended that the foregoing description be regarded as illustrative rather than limiting, and that it be understood that all equivalents and/or combinations of embodiments are intended to be included in this description. Thus, the invention is not restricted to the above illustrated embodiments but variations can be derived by a person skilled in the art without deviation from the scope of the invention.