Asset monitoring system and method, control system and electric power system

Abstract

To monitor an asset (12, 14, 16, 17) of an electric power system (10; 170; 180) are provided, an asset monitoring system (40, 51) is operative to convert asset images to grayscale images, to perform an asset state assessment based on the grayscale images, and generate output based on the asset state assessment.

Description

[0001] 1

[0002] ASSET MONITORING SYSTEM AND METHOD, CONTROL SYSTEM AND ELECTRIC POWER SYSTEM

[0003] TECHNICAL FIELD

[0004] Embodiments relate to asset monitoring systems and asset monitoring methods operative for monitoring an asset of an electric power system. Embodiments relate in particular to asset monitoring systems and asset monitoring methods that facilitate monitoring of assets having an asset tank.

[0005] BACKGROUND

[0006] Electric power systems are important infrastructure components. Examples of such electric power systems comprise electric power generation, transmission, and / or a distribution systems. Such electric power systems comprise a plurality of assets. Many of these assets are designed to remain in field operation for longer periods, such as in excess of several years or even several decades. The failure of an asset of an electric power system can have potentially catastrophic consequences. Thus, it is important that the assets be monitored. For illustration, it is desirable to detect potentially adverse effects of aging and / or operating conditions to which some of the assets have been subjected to take corrective action prior to asset failure. Important examples of such electric power system assets comprise transformers, transformer accessories, reactors, high voltage direct current valves, or other electric power system assets, without being limited thereto.

[0007] Conventionally, asset monitoring involves deployment of maintenance personnel for on-site inspection and maintenance. This process may not always be adequate for triggering a mitigating or corrective action in a timely manner.

[0008] D. Kumar and M. A. Ansari, "Condition monitoring of electrical assets using digital IRT and Al technique," Journal of Electrical Systems and Information Technology, Volume 5, Issue 3, December 2018, Pages 623-634. Elsevier (2018) disclose a technique useful for monitoring an asset of an electric power system by means of infra-red thermo-gram (IRT) and artificial intelligence (Al) techniques. The IRT sensor system and processing suggested in this scientific publication involves complex dedicated hardware for thermal imaging, which can be sensitive to ambient conditions, and complex processing techniques, which may be challenging to implement in some scenarios.

[0009] Thus, there is still a need for improved techniques of performing asset monitoring. There is, in particular, a need for techniques that facilitate asset monitoring using image sensors that are robust and reliable. SUMMARY

[0010] It is an objective of the invention to provide methods and processing systems that provide enhanced techniques of monitoring an asset of an electric power system. There is, in particular, a need for techniques that can be implemented using sensors that are versatile and readily available in association with processing techniques operative to detect at least some types of asset degradation scenarios, such as contaminations, in a reliable and robust manner, for at least some possible root causes for a degradation of asset state. It is an optional objective of the invention to provide enhanced techniques of monitoring an asset that can identify whether the asset has a state in which it can cause pollution of its environment and / or whether ambient contaminants are located so as to have the potential of affecting the asset safety and operation.

[0011] According to embodiments, an asset monitoring system and an asset monitoring method as recited in the claims are provided. The dependent claims define preferred or advantageous embodiments.

[0012] Embodiments of the invention an asset monitoring system and an asset monitoring method that convert asset images having several image channels into grayscale images having a single image channel. The conversion may comprise a clustering (such as k means clustering) or other image recognition technique. The asset monitoring system and asset monitoring method are operative to perform an asset state assessment (such as detection of a contaminant on or in proximity to an asset tank of the asset) based on whether an area covered by pixels having one or several given grayscale value, i.e., a proportion of the pixels having the one or several given grayscale values, changes. Thereby, the asset state assessment can be performed efficiently, even when operating on asset images captured using a conventional image sensor with several channels in the visible and / or nearinfrared spectral range.

[0013] According to an aspect of the invention, there is provided an asset monitoring system operative to monitor an asset of an electric power system. The asset monitoring system comprises at least one processing circuit operative to obtain a first asset image and a second asset image, the first asset image and the second asset image representing at least a portion of the asset captured from a same acquisition position relative to the asset at different acquisition times. The at least one processing circuit is operative to perform an image conversion to generate a first grayscale image from the first asset image and a second grayscale image from the second asset image. The at least one processing circuit is operative to determine a first numerical value indicative of a proportion of pixels in the first grayscale image fulfilling a grayscale value-based criterion and a second numerical value indicative of a proportion of pixels in the second grayscale image fulfilling the grayscale value-based criterion. The at least one processing circuit is operative to provide output via at least one interface based on an asset state assessment that is based on a comparison of the first numerical value and the second numerical value. Various effects and advantages are associated with the asset monitoring system. The asset monitoring system is operative to perform the asset state assessment based on a quantity that can be efficiently determined, namely a proportion of pixels in a grayscale image that fulfils a grayscale valuebased criterion (such as having a pixel value that corresponds to a given one of the possible grayscale values or any one of a subset of the possible grayscale values). The asset monitoring system is operative to perform the asset state assessment, based on change in a proportion of pixels in the grayscale images having one or several given grayscale value(s) (such as gray; or either gray or black) to identify various possible root causes of asset degradation, such as for example a highly undesirable mineral oil leak or any other contaminant at or in proximity to an asset tank, which has the potential of causing pollution in the environment of the asset. The processing performed by the asset monitoring system can operate on color images having several image channels in the visible and / or infrared spectral range, while allowing a possible degradation to be detected efficiently using the change in the proportion of pixels in the transformed grayscale images having one or several given grayscale value(s) as indicator for a degradation root cause (such as insulation liquid leakage from an asset tank; corrosion on the asset tank; etc.).

[0014] The term "grayscale image" refers to an image having a single image channel only. Thus, by conversion to grayscale images, an area-based detection of changes in asset state may be implemented in a robust and efficient manner, based on the proportion of pixels in the grayscale images that have a given grayscale value (such as gray) or that have grayscale value in a given subset of the possible grayscale values (such as either gray or black).

[0015] The at least one processing circuit may be operative to perform the image conversion such that it comprises a clustering technique or the application of an alternate image recognition technique, such as for example but not limited to Artificial Neural Networks (ANNs) Convolutional Neural Networks (CNNs), and / or Deep Learning ANN(s).

[0016] Thereby, the grayscale images can be generated using a technique capable of taking into account pixel values and pixel proximity when generating the grayscale images.

[0017] The at least one processing circuit may be operative to identify, based on the comparison, a change in state of an asset tank of the asset to perform the asset state assessment and to provide the output based on the change in state of the asset tank.

[0018] Thereby, an image-based state assessment of the asset tank can be implemented using an efficient and robust processing, which is capable of processing asset images having several image channels as obtainable using conventional camera chips that are converted into grayscale images to detect changes of the asset tank over time in an efficient and robust manner. The at least one processing circuit may be operative to identify, based on the comparison, a change of an area covered by at least one contaminant to perform the asset state assessment and to generate the output based on the change of the area.

[0019] Thereby, an image-based state assessment that detects contaminants can be implemented using an efficient and robust processing, which is capable of processing asset images having several image channels as obtainable using conventional camera chips that are converted into grayscale images to detect an increase in contaminant-covered area over time in an efficient and robust manner. The output may be generated responsive to and based on the detected change in area, thereby providing an image processing-based technique of asset monitoring capable of triggering an action responsive to the increase in contaminant-covered area.

[0020] A contaminant may be any material resulting from egress of material from the asset to an environment external to the asset tank (such as insulation liquid leaking from the asset tank) or resulting from ingress of material to the environment of the asset tank (such as flood water approaching the asset tank floor).

[0021] The at least one contaminant may comprise an insulation liquid leaking from the asset . The at least one processing circuit may be operative to identify, based on the comparison, the change of the area covered by the insulation liquid. The output may be dependent on the detected change of the area covered by the insulation liquid.

[0022] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing. The output may be generated responsive to and based on the detected change in area covered by the insulation liquid, thereby providing an image processing-based technique capable of triggering an action responsive to the increase in insulation liquid-covered area. Timely detection of insulation liquid leaks is of importance to ensure safety and reliability of asset operation and to keep pollution effects low.

[0023] The at least one processing circuit may be operative to discriminate the insulation liquid from liquids other than the insulation liquid and / or from transient light condition-dependent effects to perform the asset state assessment. The output may be dependent on the change in area covered by the insulation liquid, i.e., taking into account areas that may be covered by liquids other than the insulation liquid and / or changes in grayscale images caused by transient light condition-dependent effects.

[0024] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing. Timely detection of insulation liquid leaks is of particular importance to ensure safety and reliability of asset operation and to keep pollution effects low. The at least one processing circuit may be operative to perform a data-driven processing technique to discriminate the insulation liquid from the liquids other than the insulation liquid and / or from the transient light condition-dependent effects.

[0025] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing. Timely detection of insulation liquid leaks is of importance to ensure safety and reliability of asset operation and to keep pollution effects low or eliminate their risk completely.

[0026] The asset state assessment may be operative to obtain, responsive to detecting a change in the proportion of pixels in the second grayscale image that suggests insulation liquid leakage, a third asset image, perform the image conversion to generate a third grayscale image, and verify, based on the third grayscale image, that the change in the proportion of pixels in the second grayscale image that suggests insulation liquid leakage is not caused by a transient light condition-dependent effect (such as a shadow cast by a cloud or other transient effect).

[0027] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing.

[0028] The at least one contaminant may comprise flood water. The at least one processing circuit may be operative to generate the output responsive to detection of the flood water based on the comparison.

[0029] Thereby, an image-based state assessment that detects potential risks associated with flooding can be implemented using an efficient and robust processing. Timely detection of flooding is of importance to ensure safety and reliability of asset operation.

[0030] The at least one contaminant may comprise a corroded metal portion. The at least one processing circuit may be operative to generate the output responsive to detection of the corroded metal portion based on the comparison.

[0031] Thereby, an image-based state assessment that detects potential risks associated with asset tank corrosion can be implemented using an efficient and robust processing. Timely detection of asset tank corrosion is of importance to ensure safety and reliability of asset operation and mitigate the risk of environmental pollution.

[0032] The at least one processing circuit may be operative to perform the image conversion such that all pixels of the first grayscale image and all pixels of the second grayscale image have grayscale values selected from a group consisting of ten or less than ten discrete values.

[0033] Thereby, the image conversion can be implemented in such a way that the number of possible different grayscale values are less than in conventional grayscale images. A segmentation, using a clustering or other image processing technique capable of performing a segmentation (such as the application of an artificial intelligence (Al) model), is efficiently performed during the image conversion. Interpretation of the first and second grayscale images by the asset monitoring system is facilitated.

[0034] The at least one processing circuit may be operative to perform the image conversion such that all pixels of the first grayscale image and all pixels of the second grayscale image have grayscale values selected from a group consisting of three discrete values (such as a first grayscale value representing black, a second grayscale value representing gray, and a third grayscale value representing white) or four discrete values (such as a first grayscale value representing black, a second grayscale value representing dark gray, a third grayscale value representing light gray, and a fourth grayscale value representing white).

[0035] Thereby, the image conversion can be implemented in such a way that the number of possible different grayscale values are less than in conventional grayscale images. A segmentation, using the clustering, or another image recognition technique, such as for example Artificial Neural Networks (ANN) in their multiple configurations, such as convolutional neural nets and / or deep learning, among others, is efficiently performed during the image conversion. Interpretation of the first and second grayscale images by the asset monitoring system is facilitated.

[0036] The at least one processing circuit may be operative to perform the image conversion such that it comprises a k-means clustering.

[0037] Thereby, the image conversion can include an inherent segmentation of the asset images.

[0038] Each of the first asset image and the second asset image may have three image channels or more than three image channels.

[0039] Thereby, the asset monitoring system can operate on asset images that can be captured using conventional, robust image sensors having several image channels, such as charge coupled device (CCD) or CMOS image sensors having several image channels corresponding to different portions of the visible and / or near-infrared spectral range.

[0040] The three image channels or the more than three image channels may comprise one or several image channels in the visible and / or infrared spectral range.

[0041] Thereby, the asset monitoring system can operate on asset images that can be captured using conventional, robust image sensors having several image channels.

[0042] The three image channels or the more than three image channels may comprise one or several image channels in the visible spectral range (e.g., from 380 nm to 700 nm or from 380 nm to 780 nm).

[0043] Thereby, the asset monitoring system can perform image-based asset state assessment based on asset images that can be captured using conventional, robust image sensors having several image channels. The asset monitoring system is operative to detect a change in asset state (such as an increase in amount of insulation liquid leaked from an asset tank) that causes changes in the visible spectral range. The three image channels or the more than three image channels may comprise one or several image channels in the near infrared spectral range (e.g., from 780 nm to 2500 nm).

[0044] Thereby, the asset monitoring system can perform image-based asset state assessment taking into account changes in the near infrared spectral range, such as changes indicative of local heat generation.

[0045] The asset monitoring system may comprise an image acquisition device comprising an image sensor operative to capture the first asset image and the second asset image. The at least one processing circuit may be operative to communicatively interface with the image acquisition device to receive the first asset image and the second asset image. The at least one processing circuit may be operative to communicatively interface with the image acquisition device to trigger acquisition of the first asset image and of the second asset image by the image acquisition device.

[0046] Thereby, the asset monitoring system is operative to acquire the first asset image and the second asset image at a given elapsed time for comparison.

[0047] The image acquisition device may comprise an image sensor having three channels or more than three channels for different spectral ranges. The image acquisition device may comprise a CCD or CMOS image sensor.

[0048] Thereby, the asset monitoring system can capture images that can be captured using conventional, robust image sensors operative to capture images having several image channels.

[0049] The image sensor may be operative such that the channels or the more than three channels may comprise one or several image channels in the visible and / or infrared spectral range.

[0050] Thereby, the asset monitoring system can operate on asset images that can be captured using conventional, robust image sensors having several image channels.

[0051] The image sensor may be operative such that the channels or the more than three channels may comprise one or several image channels in the visible spectral range (e.g., from 380 nm to 700 nm or from 380 nm to 780 nm).

[0052] Thereby, the asset monitoring system can perform image-based asset state assessment based on asset images that can be captured using conventional, robust image sensors having several image channels. The asset monitoring system is operative to detect a change in asset state (such as an increase in amount of insulation liquid leaked from an asset tank) that causes changes in the visible spectral range.

[0053] The image sensor may be operative such that the channels or the more than three channels comprise one or several image channels in the near infrared spectral range (e.g., from 780 nm to 2500 nm). Thereby, the asset monitoring system can perform image-based asset state assessment taking into account changes in the near infrared spectral range, such as changes indicative of local heat generation.

[0054] The asset monitoring system may comprise a positioning device operative to position the image sensor at the same image acquisition position relative to the asset for acquisition of the first asset image and the second asset image.

[0055] Thereby, a consistent position of the image acquisition position relative to the asset can be ensured.

[0056] The positioning device may be operative to position the image sensor at the same translatory position and rotatory orientation relative to the asset for acquisition of the first asset image and the second asset image.

[0057] Thereby, a consistent translatory and rotatory position of the image acquisition position relative to the asset can be ensured, facilitating the comparison of the first and second grayscale images.

[0058] The at least one processing circuit may be operative to communicatively interface with the positioning device to cause at least one actuator of the positioning device to position the image sensor at the same translatory position and rotatory orientation relative to the asset for acquisition of the first asset image and the second asset image.

[0059] Thereby, a consistent translatory and rotatory position of the image acquisition position relative to the asset can be ensured, while allowing the image sensor to be repositioned (e.g., for capturing images of the asset from different viewing directions).

[0060] The at least one interface may comprise a human machine interface. The at least one processing circuit may be operative to generate the output comprising an alarm, warning, or other status information for outputting via the human machine interface responsive to the asset state assessment.

[0061] Thereby, the asset monitoring system is operative to provide alarms or warnings. The human machine interface may be a human machine interface of a control center, e.g., of a power grid control center.

[0062] Alternatively or additionally, the at least one interface may comprise a data interface. The at least one processing circuit may be operative to generate the output comprising control data for outputting via the data interface to automatically trigger a control action responsive to the asset state assessment. The control action may cause operation of primary system equipment of an electric power system. The control action may be a mitigating control action that mitigates effects of a detected degradation in asset state.

[0063] Thereby, the asset monitoring system is operative to automatically trigger the control action, providing an enhanced degree of automation. The asset monitoring system may be or may comprise a transformer monitoring system, a reactor monitoring system, and / or a high voltage direct current (HVDC) valve monitoring system.

[0064] Thereby, the technical effects of the asset monitoring system are attained in association with an asset which, due to its complexity and its challenging operating conditions, may be particularly prone to exhibiting changes in asset state that require action (e.g., maintenance) to be taken.

[0065] According to another aspect of the invention, there is provided a control system for an electric power system. The control system may comprise the asset monitoring system of an aspect or embodiment. The control system may be operative to control the electric power system responsive to the asset state assessment.

[0066] Thereby, a control action can be triggered (e.g., automatically, semi-automatically, or using operator input in association with the result of the asset state assessment) based on the asset state assessment.

[0067] According to another aspect of the invention, there is provided an electric power system. The electric power system may comprise an asset. The electric power system may comprise the asset monitoring system according to an aspect or embodiment operative to perform the asset state assessment for the asset and / or the control system according to an aspect or embodiment operative to perform an electric power system control operation responsive to the asset state assessment.

[0068] Thereby, the effects described in association with the asset monitoring system are obtained in the electric power system.

[0069] The electric power system may comprise any one or any combination of: a substation; a transmission grid; a distribution grid; a power generation system; a distributed energy resource (DER); a converter / inverter of an HVDC system comprising a plurality of valves.

[0070] The asset may comprise an asset tank. The asset monitoring system may be operative to perform the asset state assessment to identify a contamination on or in proximity to the asset tank.

[0071] Thereby, an image-based state assessment of the asset tank can be implemented using an efficient and robust processing, which is capable of processing asset images having several image channels as obtainable using conventional camera chips that are converted into grayscale images to detect changes of the asset tank over time in an efficient and robust manner.

[0072] The asset may comprise a transformer, a reactor, or a HVDC valve.

[0073] Thereby, the technical effects of the asset monitoring system are attained in association with an asset which, due to its complexity and its challenging operating conditions, may be particularly prone to exhibiting changes in asset state that require action (e.g., maintenance) to be taken.

[0074] According to another aspect of the invention, there is provided an asset monitoring method for monitoring an asset of an electric power system. The asset monitoring method comprises obtaining, by an asset monitoring system, a first asset image and a second asset image, the first asset image and the second asset image representing at least a portion of the asset captured from a same acquisition position relative to the asset at different acquisition times. The asset monitoring method comprises performing an image conversion to generate a first grayscale image from the first asset image and a second grayscale image from the second asset image. The asset monitoring method comprises determining a first numerical value indicative of a proportion of pixels in the first grayscale image fulfilling a grayscale value-based criterion and a second numerical value indicative of a proportion of pixels in the second grayscale image fulfilling the grayscale value-based criterion. The asset monitoring method comprises providing output via at least one interface based on an asset state assessment that is based on a comparison of the first numerical value and the second numerical value.

[0075] Various effects and advantages are associated with the asset monitoring method. The asset monitoring method is operative to perform the asset state assessment based on a quantity that can be efficiently determined, namely a proportion of pixels in a grayscale image that fulfils a grayscale value-based criterion (such as having a pixel value that corresponds to a given one of the possible grayscale values or any one of a subset of the possible grayscale values). The asset monitoring method is operative to perform the asset state assessment, based on change in a proportion of pixels in the grayscale images having one or several given grayscale value(s) (such as gray; or either gray or black) to identify various possible root causes of asset degradation, such as a contaminant at or in proximity to an asset tank, which has the potential of causing pollution in the environment of the asset. The processing performed in the asset monitoring method can operate on color images having several image channels in the visible and / or infrared spectral range, while allowing a possible degradation to be detected efficiently using the change in the proportion of pixels in the grayscale images having one or several given grayscale value(s) as indicator for a degradation root cause (such as insulation liquid leakage from an asset tank; corrosion on the asset tank; etc.).

[0076] Additional features of the asset monitoring system and the effects attained thereby correspond to the features and effects disclosed in association with the asset monitoring system.

[0077] The method may be performed automatically by the asset monitoring system, the control system, or the electric power system according to an aspect or embodiment.

[0078] The asset monitoring method may comprise identifying, by the asset monitoring system and based on the comparison, a change in state of an asset tank of the asset to perform the asset state assessment and providing the output based on the change in state of the asset tank.

[0079] Thereby, an image-based state assessment of the asset tank can be implemented using an efficient and robust processing, which is capable of processing asset images having several image channels as obtainable using conventional camera chips that are converted into grayscale images to detect changes of the asset tank over time in an efficient and robust manner. The asset monitoring method may comprise identifying, by the asset monitoring system and based on the comparison, a change of an area covered by at least one contaminant to perform the asset state assessment and to generate the output based on the change of the area.

[0080] Thereby, an image-based state assessment that detects contaminants can be implemented using an efficient and robust processing, which is capable of processing asset images having several image channels as obtainable using conventional camera chips that are converted into grayscale images to detect an increase in contaminant-covered area over time in an efficient and robust manner. The output may be generated responsive to and based on the detected change in area, thereby providing an image processing-based technique of asset monitoring capable of triggering an action responsive to the increase in contaminant-covered area.

[0081] The at least one contaminant may comprise an insulation liquid leaking from the asset. The asset monitoring method may comprise identifying, by the asset monitoring system and based on the comparison, the change of the area covered by the insulation liquid. The output may be dependent on the detected change of the area covered by the insulation liquid.

[0082] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing. Timely detection of insulation liquid leaks is of importance to ensure safety and reliability of asset operation and to keep pollution effects low.

[0083] The asset monitoring method may comprise discriminating the insulation liquid from liquids other than the insulation liquid and / or from transient light condition-dependent effects to perform the asset state assessment. The output may be dependent on the change in area covered by the insulation liquid, i.e., taking into account areas that may be covered by liquids other than the insulation liquid and / or changes in grayscale images caused by transient light condition-dependent effects.

[0084] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing. Timely detection of insulation liquid leaks is of particular importance to ensure safety and reliability of asset operation and to keep pollution effects low.

[0085] The asset monitoring method may comprise performing a data-driven processing technique to discriminate the insulation liquid from the liquids other than the insulation liquid and / or from the transient light condition-dependent effects.

[0086] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing. Timely detection of insulation liquid leaks is of importance to ensure safety and reliability of asset operation and to keep pollution effects low.

[0087] The asset monitoring method may comprise obtaining, responsive to detecting a change in the proportion of pixels in the second grayscale image that suggests insulation liquid leakage, a third asset image, performing the image conversion to generate a third grayscale image, and verifying, based on the third grayscale image, that the change in the proportion of pixels in the second grayscale image that suggests insulation liquid leakage is not caused by a transient light condition-dependent effect (such as a shadow cast by a cloud or other transient effect).

[0088] Thereby, an image-based state assessment that detects insulation liquid leakage can be implemented using an efficient and robust processing.

[0089] The at least one contaminant may comprise flood water. The asset monitoring method may comprise generating the output responsive to detection of the flood water based on the comparison.

[0090] Thereby, an image-based state assessment that detects potential risks associated with flooding can be implemented using an efficient and robust processing. Timely detection of flooding is of importance to ensure safety and reliability of asset operation.

[0091] The at least one contaminant may comprise a corroded metal portion. The asset monitoring method may comprise generating the output responsive to detection of the corroded metal portion based on the comparison.

[0092] Thereby, an image-based state assessment that detects potential risks associated with asset tank corrosion can be implemented using an efficient and robust processing. Timely detection of asset tank corrosion is of importance to ensure safety and reliability of asset operation and mitigate the risk of environmental pollution.

[0093] In the asset monitoring method, the asset monitoring system may perform the image conversion such that all pixels of the first grayscale image and all pixels of the second grayscale image have grayscale values selected from a group consisting of ten or less than ten discrete values.

[0094] Thereby, the image conversion can be implemented in such a way that the number of possible different grayscale values are less than in conventional grayscale images. A segmentation, using the clustering or other image recognition technique, may be efficiently performed during the image conversion. Interpretation of the first and second grayscale images by the asset monitoring system is facilitated.

[0095] In the asset monitoring method, the asset monitoring system may perform the image conversion such that all pixels of the first grayscale image and all pixels of the second grayscale image have grayscale values selected from a group consisting of three discrete values (such as a first grayscale value representing black, a second grayscale value representing gray, and a third grayscale value representing white) or four discrete values (such as a first grayscale value representing black, a second grayscale value representing dark gray, a third grayscale value representing light gray, and a fourth grayscale value representing white).

[0096] Thereby, the image conversion can be implemented in such a way that the number of possible different grayscale values are less than in conventional grayscale images. A segmentation, using the clustering or other image processing technique, may be efficiently performed during the image conversion. Interpretation of the first and second grayscale images by the asset monitoring system is facilitated.

[0097] In the asset monitoring method, the asset monitoring system may perform the image conversion such that it comprises a k-means clustering.

[0098] Thereby, the image conversion can include an inherent segmentation of the asset images.

[0099] In the asset monitoring method, each of the first asset image and the second asset image may have three image channels or more than three image channels.

[0100] Thereby, the asset monitoring system can operate on asset images that can be captured using conventional, robust image sensors having several image channels, such as charge coupled device (CCD) or CMOS image sensors having several image channels corresponding to different portions of the visible and / or near-infrared spectral range.

[0101] The three image channels or the more than three image channels may comprise one or several image channels in the visible and / or infrared spectral range.

[0102] Thereby, the asset monitoring system can operate on asset images that can be captured using conventional, robust image sensors having several image channels.

[0103] The three image channels or the more than three image channels may comprise one or several image channels in the visible spectral range (e.g., from 380 nm to 700 nm or from 380 nm to 780 nm).

[0104] Thereby, the asset monitoring system can perform image-based asset state assessment based on asset images that can be captured using conventional, robust image sensors having several image channels. The asset monitoring system is operative to detect a change in asset state (such as an increase in amount of insulation liquid leaked from an asset tank) that causes changes in the visible spectral range.

[0105] The three image channels or the more than three image channels may comprise one or several image channels in the near infrared spectral range (e.g., from 780 nm to 2500 nm).

[0106] Thereby, the asset monitoring system can perform image-based asset state assessment taking into account changes in the near infrared spectral range, such as changes indicative of local heat generation.

[0107] The asset monitoring method may comprise capturing, by an image acquisition device comprising an image sensor, the first asset image and the second asset image. The method may comprise communicatively interfacing, by at least one processing circuit of the asset monitoring system, with the image acquisition device to receive the first asset image and the second asset image. The method may comprise communicatively interfacing, by the at least one processing circuit, with the image acquisition device to trigger acquisition of the first asset image and of the second asset image by the image acquisition device. Thereby, the asset monitoring system is operative to acquire the first asset image and the second asset image.

[0108] The image acquisition device may comprise an image sensor having three channels or more than three channels for different spectral ranges. The image acquisition device may comprise a CCD or CMOS image sensor.

[0109] Thereby, the asset monitoring system can capture images that can be captured using conventional, robust image sensors operative to capture images having several image channels.

[0110] The image sensor may be operative such that the channels or the more than three channels may comprise one or several image channels in the visible and / or infrared spectral range.

[0111] Thereby, the asset monitoring system can operate on asset images that can be captured using conventional, robust image sensors having several image channels.

[0112] The image sensor may be operative such that the channels or the more than three channels may comprise one or several image channels in the visible spectral range (e.g., from 380 nm to 700 nm or from 380 nm to 780 nm).

[0113] Thereby, the asset monitoring system can perform image-based asset state assessment based on asset images that can be captured using conventional, robust image sensors having several image channels. The asset monitoring system is operative to detect a change in asset state (such as an increase in amount of insulation liquid leaked from an asset tank) that causes changes in the visible spectral range.

[0114] The image sensor may be operative such that the channels or the more than three channels comprise one or several image channels in the near infrared spectral range (e.g., from 780 nm to 2500 nm).

[0115] Thereby, the asset monitoring system can perform image-based asset state assessment taking into account changes in the near infrared spectral range, such as changes indicative of local heat generation.

[0116] The method may comprise positioning, by a positioning device, the image sensor at the same image acquisition position relative to the asset for acquisition of the first asset image and the second asset image.

[0117] Thereby, a consistent position of the image acquisition position relative to the asset can be ensured.

[0118] The method may comprise positioning, by the positioning device, the image sensor at the same translatory position and rotatory orientation relative to the asset for acquisition of the first asset image and the second asset image.

[0119] Thereby, a consistent translatory and rotatory position of the image acquisition position relative to the asset can be ensured, facilitating the comparison of the first and second grayscale images. The method may comprise communicatively interfacing, by at least one processing circuit of the asset monitoring system, with the positioning device to cause at least one actuator of the positioning device to position the image sensor at the same translatory position and rotatory orientation relative to the asset for acquisition of the first asset image and the second asset image.

[0120] Thereby, a consistent translatory and rotatory position of the image acquisition position relative to the asset can be ensured, while allowing the image sensor to be repositioned (e.g., for capturing images of the asset from different viewing directions).

[0121] The at least one interface may comprise a human machine interface. The asset monitoring method may comprise generating the output comprising an alarm, warning, or other status information for outputting via the human machine interface responsive to the asset state assessment.

[0122] Thereby, the asset monitoring method is operative to provide alarms or warnings. The human machine interface may be a human machine interface of a control center, e.g., of a power grid control center.

[0123] Alternatively or additionally, the at least one interface may comprise a data interface. The asset monitoring method may comprise generating the output comprising control data for outputting via the data interface to automatically trigger a control action responsive to the asset state assessment. The control action may cause operation of primary system equipment of an electric power system. The control action may be a mitigating control action that mitigates effects of a detected degradation in asset state.

[0124] Thereby, the asset monitoring system is operative to automatically trigger the control action, providing an enhanced degree of automation.

[0125] The asset monitoring method may be or may comprise a transformer monitoring method, a reactor monitoring method, and / or a high voltage direct current (HVDC) valve monitoring method.

[0126] Thereby, the technical effects of the asset monitoring method are attained in association with an asset which, due to its complexity and its challenging operating conditions, may be particularly prone to exhibiting changes in asset state that require action (e.g., maintenance) to be taken.

[0127] According to another aspect of the invention, there is provided a control method of controlling an electric power system, comprising performing, by a control system, at least one control action based on the output provided by the asset monitoring system.

[0128] Thereby, the risk of human-induced errors is mitigated in performing electric power system control operations.

[0129] According to another aspect of the invention, there is provided machine-readable instruction code comprising machine-readable instructions which, when executed by at least one processing circuit, cause the at least one processing circuit to perform the method according to an aspect or embodiment of the invention. The effects attained by the machine-readable instruction code correspond to the effects disclosed in association with the methods and systems according to various embodiments.

[0130] According to another aspect of the invention, there is provided non-transitory storage medium having stored thereon machine-readable instruction code comprising machine-readable instructions which, when executed by at least one processing circuit, cause the at least one processing circuit to perform the method according to an aspect or embodiment of the invention.

[0131] The effects attained by the non-transitory storage medium correspond to the effects disclosed in association with the methods and systems according to various embodiments.

[0132] The asset monitoring systems and methods can be used in association with an electric power grid or sub-systems thereof, such as a power system substation, without being limited thereto.

[0133] BRIEF DESCRIPTION OF THE DRAWINGS

[0134] Embodiments of the invention will be described with reference to the drawings in which similar or identical reference signs designate elements with similar or identical configuration and / or function.

[0135] Figure 1 is a schematic diagram of an asset monitoring system.

[0136] Figure 2 is a block diagram of at least one processing circuit of the asset monitoring system.

[0137] Figure 3 is a schematic representation of an electric power system comprising the asset monitoring system.

[0138] Figure 4 is a flow chart.

[0139] Figure 5 is a schematic representation of an electric power system comprising the asset monitoring system and an asset.

[0140] Figure 6 shows a first grayscale image.

[0141] Figure 7 shows a second grayscale image.

[0142] Figure 8 shows a first grayscale image.

[0143] Figure 9 shows a second grayscale image.

[0144] Figure 10 shows a first grayscale image.

[0145] Figure 11 shows a second grayscale image.

[0146] Figure 12 is a is a schematic representation of an asset and a contaminant including flood water.

[0147] Figure 13 is a flow chart.

[0148] Figure 14 is a schematic representation of operation of the asset monitoring system.

[0149] Figure 15 is a flow chart.

[0150] Figure 16 is a graph showing a variation of a proportion of pixels having a given grayscale value for various image acquisition times.

[0151] Figure 17 is a flow chart.

[0152] Figure 18 is a flow chart. Figure 19 is a schematic representation of an electric power system comprising the asset monitoring system.

[0153] Figure 20 is a schematic representation of an electric power system comprising the asset monitoring system.

[0154] Figure 21 is a schematic representation of the asset monitoring system.

[0155] Figure 22 is a flow chart.

[0156] Figure 23 is a schematic diagram of an asset monitoring system.

[0157] DETAILED DESCRIPTION OF EMBODIMENTS

[0158] Embodiments of the invention will be described with reference to the drawings. In the drawings, similar or identical reference signs designate elements with similar or identical configuration and / or function.

[0159] Embodiments relate to methods and systems useful in association with asset monitoring of an asset of an electric power system. More particularly, the invention provides processing systems and methods and systems operative to perform an image-based asset monitoring. The image-based asset monitoring comprises converting asset images, each respectively having several image channels associated with a different one of several spectral ranges (such as red, green, and blue spectral ranges of the visible spectrum and / or a near-infrared spectral range), into grayscale images, each respectively having a single image channel. A segmentation is performed, which may be performed using, e.g., a clustering (such as k-means clustering) or another image processing technique such as application of an artificial intelligence (Al) model. The clustering or other image processing that groups pixels may be performed in association with (e.g., during or upon) the conversion into the grayscale images. Based on a change in the proportion of pixels having a given grayscale value (such as a grayscale value corresponding to black or gray) or in the proportion of pixels having any one of a subset of possible grayscale values (such as any one of two out of three or four possible grayscale values), a change in asset state is detected. The system or method is operative to automatically perform an action (such as a control action) responsive to detecting, based on the change in proportion of pixels fulfilling a grayscale value-based criterion, a change in asset state.

[0160] As used herein, "asset monitoring" refers to at least monitoring the asset with regard to one or several characteristics, such as presence of one or several contaminants (optionally including at least one contaminant indicative of a degradation of the asset), presence of local hotspots, or other asset characteristics.

[0161] As used herein, "grayscale image" refers to an image having a single image channel only. Each pixel of the grayscale image may have a value selected from a set of supported grayscale values. The set of supported grayscale values may consist of less than ten, e.g., three or four different grayscale values. The techniques disclosed herein are also applicable when the grayscale image has grayscale values selected from a greater set of supported grayscale values.

[0162] As used herein, "asset image" refers to an image representing at least a portion of the asset, wherein the asset image may have more than one (e.g., three or four) image channels. At least one of several asset images (such as the first asset image) may represent a non-degraded asset state to facilitate the identification of asset degradation.

[0163] As used herein, "numerical value indicative of a proportion of pixels fulfilling a grayscale valuebased criterion" refers to any quantifier that is dependent on the proportion of pixels fulfilling the grayscale value-based criterion. Examples for such quantifiers include the number of pixels fulfilling the grayscale value-based criterion; the number of pixels divided by the total number of pixels in the grayscale image; an area covered by the number of pixels fulfilling the grayscale value-based criterion; an area covered by the number of pixels fulfilling the grayscale value-based criterion divided by the area imaged in the asset image.

[0164] As used herein, "clustering" refers to a process of grouping pixels. The clustering may comprise a k-means clustering, without being limited thereto. Image recognition techniques other than, e.g., k- means clustering may be used, such as image recognition techniques that comprise processing by at least one Al model, e.g., at least one ML model, or other data-driven processing technique.

[0165] As used herein, "asset" refers to an asset of an electric power system. The asset may comprise an asset having an asset tank. The asset may comprise an insulation liquid (such as an insulation oil) within the asset tank. The asset may comprise a transformer (e.g., a power transformer or instrumentation transformer), a reactor, and / or a HVDC valve.

[0166] Thus, embodiments of the invention provide techniques of performing asset monitoring using a processing technique that is simple, robust, and can operate on asset images having several image channels (such as asset images captured using a CCD or CMOS sensor having three or four image channels in the visible and / or near-infrared spectral range).

[0167] Figure 1 shows a schematic block diagram of an asset monitoring system 30 for performing APM for assets of an electric power system.

[0168] The asset monitoring system 30 comprises one or several interfaces 31, 32, a storage system 33, and at least one processing circuit 40. The asset monitoring system 30 comprises at least one interface 31 operative to obtain image data 61 comprising several asset images from an image acquisition device. The asset monitoring system 30 may be operative to store at least one of the asset images and / or a grayscale image generated therefrom in the storage system 33 for use in comparing grayscale images generated from the asset images. The asset monitoring system 30 is operative to perform an asset state assessment, based on a comparison of several grayscale images generated from the several asset images. The asset state assessment may be operative to detect a contaminant at or in proximity to an asset tank, to identify a leak in the asset tank, and / or to identify presence of flood water in proximity to the asset tank. The asset monitoring system 30 is operative to generate output based on the asset state assessment. The asset monitoring system 30 may be operative such that the output causes a control action to be performed and / or causes an alarm, warning, or other status information relating to the asset to be output via a human machine interface (HMI). The asset monitoring system 30 may be operative to generate and provide output comprising control data 62, such as control commands, based on the asset state assessment. The asset monitoring system 30 may be operative to generate and provide the control data 62 such that the control data 62 triggers an action, such as a protective, mitigating, or other action dependent on the asset state assessment. The asset monitoring system 30 may be operative to generate and provide the control data 62 to trigger acquisition of additional asset images, e.g., for discriminating transient effects caused by varying shadow conditions and / or lighting conditions from changes of the asset state (such as an increase in an amount of leaked insulation liquid). Alternatively or additionally, the asset monitoring system 30 may be operative to generate and provide an HMI control to control the provision of an alarm, warning, or other status information 63 via an HMI.

[0169] To perform the asset monitoring, the at least one processing circuit 40 is operative to perform an image conversion 40 to generate a first grayscale image from a first asset image included in the image data 61 and a second grayscale image from a second asset image included in the image data 62. The first asset image and the second asset image may respectively represent at least part of the asset and / or its environment. The first asset image and the second asset image may respectively be taken from a same image acquisition position (e.g., from a same translatory position and rotatory orientation) relative to the asset. The first asset image and the second asset image are captured at different image acquisition times. The image conversion 41 comprises a clustering 42. The clustering

[0170] 42 may comprise a k-means clustering. A number of clusters may be pre-defined (e.g., equal to three or four). Alternatively, the number of clusters may be an inherent result of the clustering, with a constraint being imposed on the number of clusters. Each of the grayscale images has a single image channel. Pixel values of the grayscale images may be selected from a comparatively small set of supported grayscale image values, such as three or four different grayscale image values.

[0171] The at least one processing circuit 40 is operative to perform a determination 43 of a first numerical value indicative of a proportion of pixels in the first grayscale image fulfilling a grayscale value-based criterion. The at least one processing circuit 40 is operative to perform the determination

[0172] 43 of a second numerical value indicative of a proportion of pixels in the second grayscale image fulfilling the grayscale value-based criterion. The grayscale value-based criterion may be whether the respective pixel of the first / second grayscale image has a given one of the possible grayscale values that are supported (such as three or four distinct grayscale values). The grayscale value-based criterion may be whether the respective pixel of the first / second grayscale image has is included in a subset of the possible grayscale values that are supported (such as a subset of two out of three or four distinct grayscale values). The proportion of such pixels is indicative of an area proportion in the first / second grayscale image in which the pixels of the first / second grayscale image fulfill the grayscale value-based criterion.

[0173] The at least one processing circuit 40 is operative to perform an asset state assessment 44, based on the proportions determined by the pixel proportion determination 43. The at least one processing circuit 40 may be operative to determine whether there is a change in the proportion and / or whether the change or rate of change in the proportion of pixels in the grayscale images having a given one of the possible grayscale values fulfill a threshold criterion (such as the change or rate of change being greater than a threshold).

[0174] The at least one processing circuit 40 is operative to perform an output generation 45 to provide the output via at least one interface 31, 32 based on an asset state assessment 44. The output generation 45 may be operative to provide output comprising control data 62 and / or HMI output 63 (such as an alarm, warning, or other status information) based on a result of the asset state assessment 44.

[0175] The at least one processing circuit 40 may comprise any one or any combination of integrated circuits, integrated semiconductor circuits, processors, controllers, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), circuit(s) including quantum bits (qubits) and / or quantum gates, without being limited thereto, to perform the processing of the image data 61. The at least one processing circuit 40 may be operative to execute instruction code that may be stored in the storage system 33 to perform the processing of the image data 61.

[0176] Figure 2 is a schematic representation of the at least one processing circuit 40.

[0177] The at least one processing circuit 40 is operative to perform the image conversion 41 comprising the clustering 42. The clustering 42 may be a k-means clustering. The clustering 42 groups individual pixels within the first grayscale image and individual pixels within the second grayscale image based on their similarity and / or proximity. Such pixel clustering enables efficient analysis of the grayscale images, facilitating tasks such as segmentation, classification, and object recognition.

[0178] In one implementation, the clustering 42 of the grayscale images is performed with a predetermined number of clusters. This approach involves predefining the desired cluster number (such as three or four clusters), providing for greater control over the resulting segmentation. The procedure can be performed, e.g., as follows, with the processing being performed by the at least one processing circuit 40:

[0179] Preprocessing: The preprocessing is optional. If performed, the preprocessing may comprise a normalization to a specified range (e.g., 0 to 1) to ensure consistency across different grayscale images. The preprocessing may comprise noise reduction and / or image enhancement.

[0180] Feature Extraction: Relevant features may be determined from each pixel which are indicative of its intensity or texture attributes. Feature extraction may be performed based on pixel intensities, local gradients, texture descriptors, or a combination of these.

[0181] Cluster Initialization: Based on the predetermined cluster number, centroids of clusters are defined. Examples include randomly selecting initial cluster centroids or utilizing techniques such as k-means++.

[0182] Cluster Assignment: Each pixel is assigned to the nearest cluster centroid based on a distance metric, such as Euclidean distance, in the feature space. The cluster assignment is updated until convergence is reached, ensuring a balanced distribution of pixels across clusters.

[0183] Cluster Update: The centroids of the clusters are updated, e.g., by computing the mean or centroid of the assigned pixels.

[0184] The cluster assignment and update operations are repeated until convergence or a predetermined stopping criterion is met.

[0185] The at least one processing circuit 40 may be operative to assess the quality of the clustering results using evaluation metrics such as within-cluster sum of squares or silhouette score. The at least one processing circuit 40 may be operative to adjust the clustering parameters or repeat the process with different initializations if the results do not meet the desired criteria.

[0186] In a further implementation, the clustering 42 of the grayscale images is performed with the number of clusters being determined as an inherent outcome of the clustering process. This approach allows for the automatic identification of meaningful clusters, without requiring a priori information on the number of clusters. The procedure can be performed, e.g., as follows, with the processing being performed by the at least one processing circuit 40:

[0187] Preprocessing: The preprocessing is optional. If performed, the preprocessing may comprise a normalization to a specified range (e.g., 0 to 1) to ensure consistency across different grayscale images. The preprocessing may comprise noise reduction and / or image enhancement.

[0188] Feature Extraction: Relevant features may be determined from each pixel which are indicative of its intensity or texture attributes. Feature extraction may be performed based on pixel intensities, local gradients, texture descriptors, or a combination of these.

[0189] Cluster Initialization: The at least one processing circuit 40 may be operative to apply an adaptive clustering algorithm. Examples include Density-Based Spatial Clustering of Applications with Noise (DBSCAN) or Mean Shift. Thereby, the initial number of clusters can be determined based on the density or mode of the feature space.

[0190] Cluster Assignment and Update: The at least one processing circuit 40 may be operative to assign pixels to clusters based on density or mode connectivity, iteratively expanding clusters and updating their centroids.

[0191] The cluster assignment and update operations are repeated until convergence or a predetermined stopping criterion is met.

[0192] The at least one processing circuit 40 may be operative to assess the quality of the clustering results using a quality metrics and adjust the clustering parameters based on the assessment. The at least one processing circuit 40 may be operative to adjust the clustering parameters or repeat the process with different initializations if the results do not meet the desired criteria.

[0193] The determination 43 of numerical values indicative of a proportion of pixels in the grayscale image fulfilling a grayscale value-based criterion may comprise a determination 46 of the number of pixels in the grayscale image that fulfills the grayscale value-based criterion. For illustration, a number of pixels in the first / second grayscale image having a given grayscale value (such as "black" or "gray") may be counted and divided by the total number of pixels in the respective grayscale image. Alternatively or additionally, area-based techniques may be used. For illustration, (smooth) boundaries may be determined around pixels belonging to a desired cluster (such as cluster associated with "black" or "gray") and the area thereof may be determined and divided by the image area. The determination of the proportion, in particular when implemented in an area-based manner, may optionally include a compensation of perspective distortions (accounting for the fact that locations on the asset more remote from the image acquisition device occupy a comparatively smaller area in the asset image than locations closer to the image acquisition device).

[0194] The asset state assessment 44 may comprise a comparison 47 performed based on at least two grayscale images generated from the asset images. The comparison 47 may comprise (e.g., may be) a comparison of the proportion of pixels determined to fulfill the pixel value based criterion in the first grayscale image and the proportion of pixels determined to fulfill the pixel value based criterion in the second grayscale image. The asset state assessment 44 may comprise identifying, based on the change in the proportion of pixels, a change in asset state, such as a change in an amount of insulation liquid leaked from an asset tank, a change in corrosion of the asset tank, presence of contaminants external to the asset (such as flood water), or other changes that affect the asset state.

[0195] The asset state assessment 44 may optionally comprise a root cause check 48. The root cause check 48 may comprise discriminating transient light condition effects (such as transient shadows) from changes in asset state. The root cause check 48 may be performed by considering more than two grayscale images, e.g., by triggering acquisition of a third (optionally fourth etc.) asset image, conversion of the third (optionally fourth etc.) asset image into a third ((optionally fourth etc.) grayscale image, and verifying that there is a consistent trend in the change in proportion of a certain grayscale value within the grayscale images over time. Alternatively or additionally, the root cause check 48 may include an object recognition using a classifier that assigns various shapes to different types of contaminants and / or that indicates, depending on the shape, the likelihood of the respective shape in the grayscale image representing a contaminant or a light condition dependent effect. The classifier may be based on a data-driven processing model, such as a deep learning or shallow learning model trained using labeled asset images.

[0196] The output generation 45 may comprise an interface control 49. The interface control 49 may comprise controlling at least one data interface 31 to output control data 62 to perform a control action responsive to the asset state assessment. Alternatively or additionally, the interface control 49 may comprise controlling at least one HMI to output an alarm, warning, or other state information related to the asset based on the asset state assessment.

[0197] Figure 3 is a schematic representation of an electric power electric power system 10 comprising the asset monitoring system 30. The electric power system 10 comprises a primary system 11 that may comprise assets of an electric power generation, transmission, and / or distribution system. The primary system 11 may comprise at least part of an electric power grid. The primary system 11 comprises assets that may comprise power transformers 12, 14, switchgear 13, or other assets that may be associated with a power transmission or power distribution line 15.

[0198] A secondary system of the electric power system 11 comprises measurement instrumentation, such as current and voltage transformers 16, 17 (which are collectively referred to as measurement transformers), dissolved gas analysis (DGA) sensors, and / or other sensors.

[0199] The electric power system 10 comprises an automation control system 20. The automation control system 20 comprises a plurality of devices 21, 22, 23. At least some of the devices 21, 22, 23 of the automation control system 20 may be operative to perform protection functions or other functions that involve a control of components of the primary system 11, such as switchgear 13 or a power transformer 12, 14. At least some of the devices of the automation control system 20 may be operative to execute a decision logic based on measurements, such as measurements received from measurement instrumentation, which may include the current transformer 16, the voltage transformer 17, and / or phasor measurement units.

[0200] The electric power system 10 comprises a communication system 24. The communication system 24 may be or may comprise a communication network. The communication system 24 may comprise a plurality of communication links by which devices of the automation control system 20 communicate with each other and / or with central systems, such as a SCADA system or other control system and the asset monitoring system 30. Communication may be performed using communication devices such as one or several gateway devices 25. The communication system 24 may comprise a substation automation system (SAS) communication system, such as an inter-substation and / or intra-substation communication system.

[0201] The at least one processing circuit 40 is communicatively interfaced with an image acquisition device 51 of the asset monitoring system 30. The asset monitoring system 30 may comprise a positioning device 52 operative to maintain and / or actively position the image acquisition device 51 at a same translatory position and rotatory orientation relative to an asset (such as the transformer 12) to acquire the asset images.

[0202] The at least one processing circuit 40 is communicatively interfaced with at least one HMI device 34, 35 of the asset monitoring system 30 to provide output based on the asset state assessment.

[0203] The electric power system 10 comprises a control system 26, such as a control center (e.g., a national and / or regional grid control center), a distribution grid control system, a distributed energy resource (DER) control system, an HVDC control system, without being limited thereto. The asset monitoring system 30 may be operative to communicatively interface with the control system 26 to provide control data thereto. The control data may be operative to generate an alarm, warning, or other status information relating to the asset for provision by the control system 26. The control data may be operative to enable the control system 26 to perform an action acting on primary system equipment of the primary system 10, based on the asset state assessment. The control action acting on the primary system equipment may be performed automatically, semiautomatically, and / or based on both an operator input and the result of the asset state assessment.

[0204] Figure 4 is a flow chart of a method 70. The method 70 may be performed automatically by the asset monitoring system 30. The method 70 is a method of monitoring an asset. The method 70 may be a transformer monitoring method, a reactor monitoring method, a HVDC valve monitoring method, and / or an electric power system control method that performs at least one control operation based at least on the asset state assessment.

[0205] At process block 71, the asset monitoring system 30 receives image data comprising several asset images. The several asset images do not need to be obtained concurrently, but may be received one by one as they are acquired. At least one of the several asset images may represent a non-degraded asset state.

[0206] At process block 72, the asset monitoring system 30 processes the asset images to generate several grayscale images. The image conversion may comprise a clustering, such as a k-means clustering.

[0207] At process block 73, the asset monitoring system 30 determines numerical values for each of several grayscale images, which numerical values quantify a proportion of pixels fulfilling a grayscale value-based criterion in the respective grayscale image. As previously explained, the determination of the numerical values may comprise determining the number of pixels fulfilling the grayscale valuebased criterion, and / or determining areas within the grayscale images.

[0208] At process block 74, the asset monitoring system 30 performs a comparison of at least two of the numerical values determined at process block 73. The asset monitoring system 30 may compare the proportion for more than two (e.g., for three, four, or more than four) grayscale images obtained from associated asset images, to confirm that there is a consistent trend in the change in proportion (as opposed to a non-monotonous, transient effect that may be caused by transient phenomena such as shadow cast). The asset state assessment is based on the comparison. The asset state assessment may comprise assessing whether there is a leak of insulation liquid from an asset tank of the asset, whether there is corrosion at the asset tank, and / or whether there is

[0209] At process block 75, the asset monitoring system 30 generates and provides output based on the asset state assessment. Provision of the output may comprise controlling an HMI and / or generating control data for effecting control actions in the electric power system.

[0210] Figure 5 is a schematic representation of an electric power system 10. The electric power system 10 comprises the asset 12 and the asset monitoring system 30. The electric power system 10 may optionally comprise a control system 26 operative to perform control actions, responsive to an asset state assessment obtained by the asset monitoring system 30.

[0211] The asset 12 may comprise an asset tank 80. The asset tank 80 may have insulation liquid (such as insulation oil) provided therein, with the insulation liquid surrounding conductors and / or other insulation material (such as paper insulation). In the illustrated electric power system 10, the asset 12 may comprise a transformer (such as a power transformer or measurement transformer) having the asset tank 80. The asset 12 may comprise accessory components such as bushings 81, a breather 82, or other accessory components, depending on a type of the respective asset 80.

[0212] The asset monitoring system 30 comprises an image acquisition device 51 positioned relative to the asset by a positioning device 52. The asset monitoring system 30 comprises a communication link 53 to communicatively interface the at least one processing circuit 40 with the image acquisition device 51. The communication link 53 may comprise a communication link of an intra-substation communication system and / or of an inter-substation communication system. The at least one processing circuit 40 may be operative to control image acquisition and to process the acquired asset images, as already discussed herein.

[0213] The asset monitoring system 30 may be operative to detect, based on the processing of the asset images, insulation liquid 84 leaking from the asset tank. The asset monitoring system 30 may be operative to detect, based on the processing of the asset images, insulation liquid 84 leaking onto a base 89 (such as a concrete base plate 89) on which the asset 12 is positioned and / or insulation liquid 85 leaking from the asset tank at a location spaced from the base 89. The asset monitoring system 30 may be operative to detect, based on the processing of the asset images, corrosion 86 on the asset tank 80.

[0214] The detection of contaminants (such as insulation liquid and / or corroded metal) on or in proximity to the asset tank 80 is based on a comparison of proportions of pixels fulfilling a grayscale value-based criterion, optionally in combination with additional checks (such as a classifier-based technique to classify the type of contaminant).

[0215] Operation of the asset monitoring system and the asset monitoring method will be illustrated further with reference to Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, and Figure 11.

[0216] Figure 6 shows a first grayscale image 91 obtained from a first asset image (e.g., in the nondegraded state of the asset). Figure 6 shows a second grayscale image 92 obtained from a second asset image at a later time (e.g., after a usage period which may be in excess of several years for a power system asset). In the illustrated implementation, the asset monitoring system 30 is operative to generate the first grayscale image 91 and the second grayscale image 92 such that all pixels of the first grayscale image 91 and of the second grayscale image 92 have grayscale values selected from a set of only three different values (black, gray, white). An increase in a number of pixels 93 having a given one of the grayscale values (e.g., black) is detected, causing the proportion of pixels with grayscale value being equal to black to have increased from the first grayscale image 91 to the second grayscale image 92. This change is detected by the asset monitoring system 30. The asset monitoring system 30 is operative to generate output based on the comparison to flag that there is a potential of an asset degradation.

[0217] Figure 8 shows a first grayscale image 94 obtained from a first asset image (e.g., in the nondegraded state of the asset). Figure 9 shows a second grayscale image 95 obtained from a second asset image at a later time (e.g., after a usage period which may be in excess of several years for a power system asset). In the illustrated implementation, the asset monitoring system 30 is operative to generate the first grayscale image 94 and the second grayscale image 95 such that all pixels of the first grayscale image 94 and of the second grayscale image 95 have grayscale values selected from a set of only three different values (black, gray, white). An increase in a number of pixels 96 having a given one of the grayscale values (e.g., gray) is detected, causing the proportion of pixels with grayscale value being equal to gray to have increased from the first grayscale image 94 to the second grayscale image 95. This change is detected by the asset monitoring system 30. The asset monitoring system 30 is operative to generate output based on the comparison to flag that there is a potential of an asset degradation. The asset monitoring system 30 may be operative to perform additional checks to confirm that the change is attributable to a change in asset state. The additional checks may comprise a classifier or other data-driven processing that takes into account the location of the change (e.g., at the base 89 that supports the asset tank) and / or the shape of the pixels having a different grayscale 1 value (e.g., characteristic shape of gravity-driven insulation oil flow) to verify that there is a leak, causing insulation liquid to flow from the asset tank onto the base 89.

[0218] Figure 10 shows a first grayscale image 97 obtained from a first asset image (e.g., in the nondegraded state of the asset). Figure 11 shows a second grayscale image 98 obtained from a second asset image at a later time (e.g., after a usage period which may be in excess of several years for a power system asset). In the illustrated implementation, the asset monitoring system 30 is operative to generate the first grayscale image 97 and the second grayscale image 98 such that all pixels of the first grayscale image 97 and of the second grayscale image 98 have grayscale values selected from a set of only three different values (black, gray, white). An increase in a number of pixels 99 having a given one of the grayscale values (e.g., gray) is detected, causing the proportion of pixels with grayscale value being equal to gray to have increased from the first grayscale image 97 to the second grayscale image 98. This change is detected by the asset monitoring system 30. The asset monitoring system 30 is operative to generate output based on the comparison to flag that there is a potential of an asset degradation. The asset monitoring system 30 may be operative to perform additional checks to confirm that the change is attributable to a change in asset state. The additional checks may comprise a classifier or other data-driven processing that takes into account the location of the change (e.g., on the asset tank) and / or the shape of the pixels having a different grayscale value (e.g., characteristic shape of gravity-driven insulation oil flow) to verify that there is a leak, causing insulation liquid to flow from an interior of the asset tank onto an outer wall of the asset tank.

[0219] The asset monitoring system 30 and the asset monitoring method may additionally or alternatively be operative to detect contaminants external to the asset, which have the potential of affecting safety and / or reliability of the asset. For illustration, the asset monitoring system 30 and the asset monitoring method may be operative to identify flood water, using the processing disclosed herein.

[0220] Figure 12 is a schematic representation of the asset 12 having an asset wall 80 supported on a base 89. The asset monitoring system 30 and the asset monitoring method may be operative to identify flood water 100, using the processing disclosed herein. The asset monitoring system 30 and the asset monitoring method may be operative to trigger an alarm or warning prior to the flood water reaching the asset 12, to thereby mitigate the risk of asset failure and / or to thereby enable protective measures to be performed.

[0221] The asset monitoring system 30 and the asset monitoring method provide the technical effect that they can use images having several image channels, such as RGB images captured using a CCD or CMOS camera chip, as input of the processing. Figure 13 and Figure 14 further illustrate operation of the asset monitoring system 30 and the asset monitoring method.

[0222] Figure 13 is a flow chart of a process 110. The process 110 may be performed automatically by the asset monitoring system 30 to generate the grayscale image, including the clustering. Conversion into a grayscale image, including the clustering, provides a robust and efficient detection of possible asset degradation while operating on images having several image channels (such as RGB images captured using a CCD or CMOS camera chip) as input.

[0223] At process block 111, the asset monitoring system 30 combines several image channels of an asset image to the single channel of the grayscale image. The combination may comprise determining, for each pixel in the asset image, a weighted sum of the values in the different image channels. Weights used in the weighted summation may be dependent on spectral characteristics of the camera sensor used to captured the asset image (e.g., the spectral sensitivity of the camera sensor for the various image channels). The result of the combination may have intensities, at each of the pixels, which may be selected from a greater range of values (such as 256 different possible intensity values).

[0224] At process block 112, the asset monitoring system 30 performs clustering, such as k-means clustering. Thereby, image segmentation can be efficiently performed for asset state assessment. The k-means clustering may be performed using a predetermined fixed number of clusters (such as three or four different clusters) or allowing the number of clusters to vary within certain limits (e.g., not more than ten clusters). The result of the clustering, thus, is a grayscale image in which each pixel has a grayscale value selected from a set that corresponds to the number of different clusters (e.g., three or four different grayscale values). The clustering at process block 112 may be based at least on the intensities at process block 111 and pixel location.

[0225] Figure 14 is a schematic representation to further illustrate the processing performed by the asset monitoring system 30 and in the asset monitoring method. The image acquisition device 50 comprises an image sensor 54. The image sensor 54 may comprise a CCD or CMOS sensor or other camera chip. The image sensor 54 is operative to capture several image channels of an image and provide, in the image data 61, the several image channels 121, 122, 123. In the illustrated example, the image channel 121 corresponds to red, the image channel 122 corresponds to green, and the image channel 123 corresponds to blue. While the intensities look similar in the various image channels, this is due to the fact that the colors in the captured scene do not exhibit pronounced variations in spectral intensity between red, green, and blue (although some variations are discernible when comparing, e.g., the base under the asset tank with the ground on which it is arranged). The at least one processing circuit 40 is operative to communicatively interface with the image sensor 54 to obtain the image data 61 and generate the first and second grayscale image(s) 124, as previously explained. The at least one processing circuit 40 is operative to perform the asset state assessment based on the change in proportion of pixels having certain grayscale values (e.g., gray or black) and generate output based on the result of the comparison.

[0226] Figure 15 is a flow chart of a method 130. The method 130 may be performed automatically by the asset monitoring system 30. The method 130 is a method of monitoring an asset. The method 130 may be a transformer monitoring method, a reactor monitoring method, a HVDC valve monitoring method, and / or an electric power system control method that performs at least one control operation based at least on the asset state assessment.

[0227] Process blocks 71, 72, and 73 may be implemented as described in association with Figure 4.

[0228] At process block 131, the asset monitoring system 30 determines whether there is a change in the proportion of pixels fulfilling the grayscale value-based criterion in the second grayscale image as compared to the first grayscale image. Process block 131 may also include a threshold comparison to compare the magnitude of the change in the pixel proportion to a threshold, to ensure that nonsignificant changes do not trigger a corrective or mitigating action or the issuance of an alarm or warning. If there is no change, or if the change does not fulfill the threshold criterion, the method may return to process block 71. Otherwise, the method 130 may continue at process block 132.

[0229] At process block 132, the asset monitoring system 30 may verify that one additional trigger criterion or several additional trigger criteria are fulfilled. The additional trigger criterion or trigger criteria may comprise: a verification that there is a consistent change that is not attributable to transient changes in light conditions. The asset monitoring system 30 may perform the verification based on, e.g., image acquisition of at least one third asset image, conversion of the at least one third asset image into at least one third grayscale image, and verifying that there is a change in the proportion of pixels from the first to the second to the at least one third grayscale image that is monotonous as a function of time. Alternatively or additionally, the asset monitoring system 30 may perform the verification based on data-driven processing techniques, such as by using a classifier having an input operative to receive the second asset image and / or the second grayscale image and having an output indicating the type of contaminant and / or the probability for various types of contaminants. a determination of a type of contaminant: The asset monitoring system 30 may perform the verification based on data-driven processing techniques, such as by using a classifier having an input operative to receive the second asset image and / or the second grayscale image and having an output indicating the type of contaminant and / or the probability for various types of contaminants.

[0230] If the additional trigger criterion or the additional trigger criteria are not fulfilled, the method 130 returns to process block 71. Otherwise, the method 130 proceeds to process block 75.

[0231] At process block 75, the asset monitoring system 30 generates output based on the asset state assessment. Process block 75 may be implemented as explained with reference to Figure 4. Figure 16 is a graph illustrating a dependency of the proportion of pixels fulfilling the grayscale value-based criterion, as a function of image acquisition time of the asset image from which the respective grayscale image is generated.

[0232] Responsive to detecting a change in the proportion of pixels having a certain grayscale value (such as black or gray), the asset monitoring system 30 may be operative to obtain a third asset image. The second asset image may be captured with a first delay 141 after acquisition of the first asset image. The first delay 141 may be greater than a day, greater than a week, greater than a month, or even greater than a year. The third asset image may be captured with a second delay 142 after acquisition of the second asset image. The second delay 142 may be shorter than the first delay 141. For illustration, the second delay 142 may be less than a day, less than an hour, or less than 10 minutes. The asset monitoring system 30 may actively trigger acquisition of the third asset image to discriminate transient effects from changes in the grayscale image(s) attributable to contaminants or other degradation effects. The asset monitoring system 30 may trigger acquisition of the third asset image with the second delay 142 that is shorter than the first delay 141.

[0233] The time-dependent evolution of the proportion of pixels fulfilling the grayscale value-based criterion may be evaluated by the asset monitoring system 30 to discriminate transient effects that can be caused by varying light conditions from more persistent effects indicative of a change in asset state. For illustration, the asset monitoring system 30 may be operative to determine that a non- monotonous variation of the proportion of pixels in which there is only a temporary increase in the proportion at the image acquisition time t2of the second asset image, with the proportion at the image acquisition time t3of the third asset image being comparable to that at the image acquisition time ti of the first asset image, is attributable to a transient effect rather than a degradation of the asset, such that no alarm is to be raised and / or no corrective action is to be automatically initiated.

[0234] Figure 17 is a flow chart of a method 150. The method 150 may be performed automatically by or using the asset monitoring system 30, a control system comprising the asset monitoring system 30, and / or the electric power system 10 comprising the asset monitoring system 30.

[0235] At process block 151, the asset monitoring system 30 processes several asset images captured from a same image acquisition position relative to the asset at different times, to detect a contaminant at or in proximity to the asset. The contaminant detection may comprise a detection of insulation liquid leaking from an asset tank 80. Alternatively or additionally, the contaminant detection may comprise a detection of a corroded metal portion on the asset tank 80. Alternatively or additionally, the contaminant detection may comprise detecting presence of foreign material external to the asset, such as detection of flood water 100.

[0236] At process block 152, the asset monitoring system 30 generates output based on the contaminant detection. Generating the output may comprise controlling an HMI to provide an alarm, warning, or other status information. Alternatively or additionally, generating the output may comprise providing control data for use in performance of a control action in the electric power system 10.

[0237] Figure 18 is a flow chart of a method 160. The method 160 may be performed automatically by or using the asset monitoring system 30, a control system comprising the asset monitoring system 30, and / or the electric power system 10 comprising the asset monitoring system 30.

[0238] At process block 161, the asset monitoring system 30 processes several asset images captured from a same image acquisition position relative to the asset at different times, to detect a change in state of the asset tank. The detection of a change in state of the asset tank may comprise a detection of insulation liquid in proximity of the asset tank 80, which is indicative of an aperture in the asset tank 80. Alternatively or additionally, the contaminant detection may comprise a detection of a corroded metal portion on the asset tank 80. Alternatively or additionally, the detection of a change in state of the asset tank may comprise detecting a corroded metal portion on the asset tank 80.

[0239] At process block 162, the asset monitoring system 30 generates output based on the detected change in state of the asset tank 80. Generating the output may comprise controlling an H Ml to provide an alarm, warning, or other status information. Alternatively or additionally, generating the output may comprise providing control data for use in performance of a control action in the electric power system 10.

[0240] The asset monitoring system 30 disclosed herein may be used to suggest or automatically perform control operations acting on the primary system 11 and / or secondary system of the electric power system 10.

[0241] Figure 19 and Figure 20 shows systems 170, 180 according to embodiments which respectively comprise a control system 174, 184. The control system 174, 184 may comprise or may be control system of a substation automation system, a national or regional control center, a microgrid control center for a microgrid comprising distributed energy resources (DERs), a high voltage direct current (HVDC) control system, without being limited thereto. The control system 174, 184 comprises at least one control circuit 175, 185 operative to generate and provide control commands 176, 186, which may comprise control commands acting on primary system components and / or secondary system components (such as commands for changing tap changer position, commands relating to transformer cooling, commands acting on switchgear 13, or combinations thereof). The at least one control circuit 175, 185 may comprise any one or any combination of integrated circuits, integrated semiconductor circuits, processors, controllers, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), circuit(s) including quantum bits (qubits) and / or quantum gates, without being limited thereto, to generate and provide the control commands 176, 186. The control system 174, 184 may be operative to enable the control commands 176, 186 to be generated based on the asset state assessment performed by the asset monitoring system 30, optionally also taking into account operator input (such as confirmations that a suggested control action is to be performed).

[0242] The control system 174, 184 may be operative such that the at least one control circuit 175, 185 generates and outputs control commands 176, 186 based on the output 62 generated by the asset monitoring system 30. The asset monitoring system 30 may be separate from the control system 174 and operative to communicatively interface with the control system 174 (Figure 19) or may be integrated into the control system 184 (Figure 20). The asset monitoring system 30 may comprise an HMI 172 or the control system 184 may comprise an HMI 182 to provide output responsive to the asset state assessment performed by the asset monitoring system 30, such as alerts, warnings, and / or recommendations for actions to be taken.

[0243] As illustrated in Figure 19, the positioning device for positioning the image acquisition device 51 relative to the asset need not be passive but may be controllable to change the translatory position and / or rotatory orientation of the image acquisition device 51 relative to the asset. In this case, the asset monitoring system 30 may be operative to control the positioning device 55 such that the image acquisition device 51 is positioned at a same translatory position and a same rotatory orientation relative to the asset for acquisition of the several images, which are processed to perform the asset state assessment.

[0244] Figure 21 is a schematic representation of an asset monitoring system 50 that comprises the at least one processing circuit 40, the image acquisition device 51, and an HMI device 191. The HMI device 191 may be portable or wearable device. The at least one processing circuit 40 may be operative to communicatively interface with the HMI device 191 via a communication link 192, which may be implemented as a wireless or wired point-to-point communication link or which may be a communication link established over a local area network (LAN) or WAN. The at least one processing circuit 40 may be operative to generate the output, based on the asset state assessment, to cause an alarm, warning, or other status information relating to the asset to be output via the HMI device 191. The at least one processing circuit 40 may be operative to communicatively interface with the image acquisition device 51 via a further communication link 193, which may comprise a point-to-point communication link or a communication link established over a network (such as a local area network (LAN) or WAN). The further communication link 193 may comprise a communication link established over an intra-substation or inter-substation communication system 24.

[0245] The asset monitoring system 30 disclosed herein may be used to suggest or automatically perform control operations, such as by issuing an alert, warning, or other status information and / or triggering control actions acting on the primary system 11 and / or secondary system of the electric power system

[0246] 10. Figure 22 is a flow chart of a method 200. The method 200 may be performed automatically by or using the asset monitoring system 30, the control system 184, and / or the electric power system 10.

[0247] At process block 201, the asset monitoring system 30 performs asset state monitoring. The asset state monitoring may be performed using any one or any combination of the techniques disclosed herein.

[0248] At process block 202, the asset monitoring system 30 or a control system 26, 184 causes a mitigating and / or corrective action to be performed responsive to the asset state assessment of the asset monitoring system 30. The mitigating and / or corrective action may be performed automatically, semi-automatically, and / or based on operator input in combination with the asset state assessment.

[0249] Various effects and advantages are attained by the asset monitoring system and method according to embodiments. The asset monitoring system and method provide enhanced techniques of performing asset monitoring. The asset monitoring system and method reduce the risk for human- induced error in asset monitoring and / or control of an electric power system. The asset monitoring system and method are operative to perform asset monitoring based on images that can be acquired using robust image sensors, such as CCD or CMOS sensors.

[0250] While embodiments have been described in detail with reference to the drawings, various modifications may be implemented in other embodiments. For illustration rather than limitation:

[0251] • While embodiments have been described in which the image conversion comprises a clustering, the image conversion may comprise alternate image recognition techniques to group pixels when converting the image data to grayscale images. For illustration, artificial intelligence (Al) models may be used, such as one or several machine learning (ML) models trained to perform, e.g., a segmentation task that groups pixels based on their pixel values and their geometrical arrangement. The image conversion may comprise any one, several, or any combination of one or several ANN(s), one or several CNN(s), and / or one or several deep learning models, such as deep learning ANN(s).

[0252] Figure 23 illustrates an asset monitoring system operative to apply one or several Al model(s) 212 to image data to generate the first grayscale image and / or the second grayscale image. The one or several Al model(s) 212 may comprise any or any combination of one or several ANN(s) 213, one or several CNN(s), and / or one or several deep learning models, such as deep learning ANN(s). The Al model(s) 212 may be implemented as ML models trained to group pixels, e.g., by assigning them to different grayscale values. The Al model(s) 212 may have an input operative to receive pixels of the image data captured by the image sensor(s), and may have an output operative to provide data indicating to which group of pixels in the first or grayscale image the respective pixels are to be assigned. The asset monitoring system 30 may be operative to perform additional or alternative image processing techniques to perform the image conversion.

[0253] • While embodiments have been described in which the asset monitoring system and method are operative to monitor transformers, the techniques may be applied to other electric power system assets such as reactors, HVDC valves, and / or other assets.

[0254] • While embodiments have been described in which the asset monitoring system and method are operative to monitor power transformers of the primary system, the techniques may be applied to other transformers such as measurement transformers.

[0255] • While embodiments have been described in which the asset monitoring system and method are operative to obtain and process images having three or four color channels in the visible spectral range, the asset monitoring system and method may be operative to obtain images having at least one image channel in the near-infrared spectral range and / or having different numbers of image channels. For illustration, the techniques disclosed herein may be used in association with thermal images obtained using a camera operative to capture spatially resolved temperature measurements.

[0256] Embodiments may be used in association with a power grid having renewables penetration, such as power grid comprising renewable energy systems (such as DERs), and / or HVDC systems.

[0257] This description and the accompanying drawings that illustrate aspects and embodiments of the present invention should not be taken as limiting-the claims defining the protected invention. In other words, while the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative and not restrictive. Various mechanical, compositional, structural, electrical, and operational changes may be made without departing from the spirit and scope of this description and the claims. In some instances, well- known circuits, structures, and techniques have not been shown in detail in order not to obscure the invention. Thus, it will be understood that changes and modifications may be made by those of ordinary skill within the scope and spirit of the following claims. In particular, the present invention covers further embodiments with any combination of features from different embodiments described above and below.

[0258] The disclosure also covers all further features shown in the Figures individually although they may not have been described in the afore or following description. Also, single alternatives of the embodiments described in the Figures and the description and single alternatives of features thereof can be disclaimed from the subject matter of the invention or from disclosed subject matter. The disclosure comprises subject matter consisting of the features defined in the claims or the embodiments as well as subject matter comprising said features. The term "comprising" does not exclude other elements or process blocks, and the indefinite article "a" or "an" does not exclude a plurality. A single unit or process block may fulfil the functions of several features recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Components described as coupled or connected may be electrically or mechanically directly coupled, or they may be indirectly coupled via one or more intermediate components. Any reference signs in the claims should not be construed as limiting the scope.

[0259] A machine-readable instruction code may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via a wide area network or other wired or wireless telecommunication systems. Furthermore, a machine-readable instruction code can also be a data structure product or a signal for embodying a specific method such as the method according to embodiments.

Claims

CLAIMS1. An asset monitoring system (30) operative to monitor an asset (12, 14, 16, 17) of an electric power system (10; 170; 180), the asset monitoring system (30) comprising: at least one processing circuit (40) operative to obtain a first asset image and a second asset image, the first asset image and the second asset image representing at least a portion of the asset (12, 14, 16, 17) captured from a same acquisition position relative to the asset at different acquisition times, perform an image conversion to generate a first grayscale image (91; 94; 97) from the first asset image and a second grayscale image (92; 95; 98) from the second asset image, determine a first numerical value indicative of a proportion of pixels in the first grayscale image (91; 94; 97) fulfilling a grayscale value-based criterion and a second numerical value indicative of a proportion of pixels in the second grayscale image (92; 95; 98) fulfilling the grayscale value-based criterion, and provide output (62, 63) via at least one interface (31, 32; 34, 35; 172; 191) based on an asset state assessment that is based on a comparison of the first numerical value and the second numerical value.

2. The asset monitoring system (30) of claim 1, wherein the at least one processing circuit (40) is operative to identify, based on the comparison, a change in state of an asset tank (80) of the asset (12, 14, 16, 17) to perform the asset state assessment and to provide the output (62, 63) based on the change in state of the asset tank.

3. The asset monitoring system (30) of claim 1 or claim 2, wherein the at least one processing circuit (40) is operative to identify, based on the comparison, a change of an area covered by at least one contaminant (84, 85, 86; 100) to perform the asset state assessment and to generate the output (62, 63) based on the change of the area.

4. The asset monitoring system (30) of claim 3, wherein the at least one contaminant (84, 85, 86; 100) comprises an insulation liquid (84, 85) leaking from the asset (12, 14, 16, 17) and the at least one processing circuit (40) is operative to identify, based on the comparison, the change of the area covered by the insulation liquid (84, 85).

5. The asset monitoring system (30) of claim 4, wherein the at least one processing circuit (40) is operative to discriminate the insulation liquid (84, 85) from liquids (100) other than the insulation liquid (84, 85) and / or from transient light condition-dependent effects to perform the asset state assessment.

6. The asset monitoring system (30) of claim 5, wherein the at least one processing circuit (40) is operative to perform a data-driven processing technique to discriminate the insulation liquid (84, 85) from the liquids (100) other than the insulation liquid (84, 85) and / or from the transient light condition-dependent effects.

7. The asset monitoring system (30) of any one of claims 3 to 6, wherein the at least one contaminant (84, 85, 86; 100) comprises flood water (100), wherein the at least one processing circuit (40) is operative to generate the output (62, 63) responsive to detection of the flood water (100) based on the comparison.

8. The asset monitoring system (30) of any one of claims 3 to 7, wherein the at least one contaminant (84, 85, 86; 100) comprises a corroded metal portion (86), wherein the at least one processing circuit (40) is operative to generate the output (62, 63) responsive to detection of the corroded metal portion (86) based on the comparison.

9. The asset monitoring system (30) of any one of the preceding claims, wherein the at least one processing circuit (40) is operative to perform the image conversion such that all pixels of the first grayscale image (91; 94; 97) and all pixels of the second grayscale image (92; 95; 98) have grayscale values selected from a group consisting of three discrete values or four discrete values.

10. The asset monitoring system (30) of any one of the preceding claims, wherein each of the first asset image and the second asset image has three image channels (121-123) or more than three image channels, wherein the three image channels (121-123) or the more than three image channels comprise one or several image channels (121-123) in the visible and / or infrared spectral range.

11. The asset monitoring system (30) of claim 10, further comprising an image acquisition device (51) comprising an image sensor (54) operative to capture the first asset image and the secondasset image, wherein the at least one processing circuit (40) is operative to communicatively interface with the image acquisition device (51) to receive the first asset image and the second asset image.

12. The asset monitoring system (30) of claim 11, further comprising a positioning device (52; 55) operative to position the image sensor (54) at the same image acquisition position relative to the asset (12, 14, 16, 17) for acquisition of the first asset image and the second asset image13. The asset monitoring system (30) of any one of the preceding claims, wherein the at least one processing circuit (40) is operative such that the image conversion comprises a clustering technique.

14. The asset monitoring system (30) of any one of the preceding claims, wherein the at least one processing circuit (40) is operative such that the image conversion comprises an image recognition technique.

15. The asset monitoring system (30) of claim 14, wherein the image recognition technique comprises application of at least one of: an Artificial Neural Network, ANN, a Convolutional Neural Network, CNN, a Deep Learning ANN,16. The asset monitoring system (30) of any one of the preceding claims, wherein: the at least one interface (31, 32; 34, 35; 172; 191) comprises a human machine interface (34, 35; 172; 191) and the at least one processing circuit (40) is operative to generate the output (62, 63) comprising an alarm or warning for outputting via the human machine interface (34, 35; 172; 191) responsive to the asset state assessment; and / or the at least one interface (31, 32; 34, 35; 172; 191) comprises a data interface (31, 32) and the at least one processing circuit (40) is operative to generate the output (62, 63) comprising control data for outputting via the data interface (31, 32) to automatically trigger a control action responsive to the asset state assessment.

17. A control system (184) for an electric power system (10; 180), comprising the asset monitoring system (30) of any one of the preceding claims, the control system (184) being operative to control the electric power system (10; 180) responsive to the asset state assessment.

18. An electric power system (10; 180), comprising:an asset (12, 14, 16, 17); and the asset monitoring system (30) of any one of claims 1 to 16 operative to perform the asset state assessment for the asset or the control system (184) of claim 17 operative to perform an electric power system control operation responsive to the asset state assessment.

19. The electric power system (10; 170; 180) of claim 18, wherein the asset (12, 14, 16, 17) comprises an asset tank (80), wherein the asset monitoring system (30) is operative to perform the asset state assessment to identify a contamination (84, 85, 86; 100) on or in proximity to the asset tank (80).

20. The electric power system (10; 170; 180) of claim 18 or claim 19, wherein the asset comprises a transformer (12, 14, 16, 17), a reactor, or a high voltage direct current, HVDC, valve.

21. An asset monitoring method for monitoring an asset (12, 14, 16, 17) of an electric power system (10; 170; 180), the asset monitoring method comprising: obtaining, by an asset monitoring system (30), a first asset image and a second asset image, the first asset image and the second asset image representing at least a portion of the asset captured from a same acquisition position relative to the asset at different acquisition times; performing an image conversion to generate a first grayscale image (91; 94; 97) from the first asset image and a second grayscale image (92; 95; 98) from the second asset image; determining a first numerical value indicative of a proportion of pixels in the first grayscale image (91; 94; 97) fulfilling a grayscale value-based criterion and a second numerical value indicative of a proportion of pixels in the second grayscale image (92; 95; 98) fulfilling the grayscale value-based criterion; and providing output (62, 63) via at least one interface (31, 32; 34, 35; 172; 191) based on an asset state assessment that is based on a comparison of the first numerical value and the second numerical value.

22. The asset monitoring method of claim 21, wherein the asset monitoring method is performed by the asset monitoring system (30) of any one of claims 1 to 16, the control system (184) of claim 17, or the electric power system (10; 180) of any one of claims 18 to 20.

23. Machine-readable instruction code comprising machine-readable instructions which, when executed by at least one processing circuit (40), cause the at least one processing circuit (40) to perform the method of claim 21 or claim 22.

Images