Methods and systems for remote monitoring of equipment

Abstract

According to one aspect, a method for remote monitoring of electrical equipment includes receiving a scene of an infrared image of equipment to be monitored, registering the scene, and partitioning the scene into a plurality of fixed, predefined, and non-overlapping areas, each area displaying a portion of the equipment to be monitored. Each area is assigned to one or more groups, each group containing at least two areas. A data value is derived from the intensity values of the pixels in each respective area. An alarm condition, indicative of potential equipment failure based on analysis of the data values of the areas assigned to that group, is defined for each group. The method includes detecting that an alarm condition for one of the groups has been met, and providing an operator alert that the alarm condition for that group has been met.

Description

Atty Docket No.: 8070.00015WO PATENTMETHODS AND SYSTEMS FOR REMOTE MONITORING OF EQUIPMENTCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims priority to U.S. Provisional Patent Application No.63 / 748,328, filed January 22, 2025, entitled “METHODS AND SYSTEMS FOR REMOTE MONITORING OF ELECTRICAL EQUIPMENT,” which is assigned to the assignee hereof and is expressly incorporated herein by reference in its entirety.TECHNICAL FIELD

[0002] This disclosure relates to monitoring the health of electrical equipment, such as high-voltage electrical transformers. More specifically, it relates methods and systems for remote monitoring of electrical equipment.BACKGROUND

[0003] Electrical power grid substations often have many large, high-voltage transformers, which age over time and can fail due to short-circuits or internal high- voltage arcing — both of which may cause an explosion that destroys not only the transformer but adjacent equipment. Power companies are therefore highly motivated to detect potential failures of these large transformers and other types of electrical equipment and high-voltage electrical equipment before failure occurs. This requires continual monitoring of this equipment.

[0004] Conventional monitoring methods require the use of voltage, current, or temperature sensors attached to the transformer, which is expensive and intrusive and which increases the number of equipment that must be maintained, or require periodic analysis of samples of the oil taken from the transformer core, which is labor-intensive and time-consuming and which cannot provide real-time notification of impending failure.

[0005] Because an increase in operating temperature of a transformer is strongly correlated to failure, there have been recent attempts to remotely monitor the temperature of a transformer using infrared cameras, but these approaches suffer the disadvantage that they rely on a determination of an exact temperature reading, which is difficult to do even when using expensive IR cameras, because determination of an absolute temperatureAtty. Docket No.: 8070.0015WOreading requires corrections to the IR sensor output to compensate for ambient temperature, corrections to compensate for whether the equipment is currently in direct sunlight or in shade, etc., as well as continual calibration of the IR sensor itself.

[0006] Accordingly, in light of the disadvantages associated with conventional approaches to remote monitoring of electrical equipment, there is a need for improved methods and systems for remote monitoring of electrical equipment.BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Embodiments of the subject matter described herein will now be explained with reference to the accompanying drawings, wherein like reference numerals represent like parts, of which:

[0008] Figure 1A is a block diagram illustrating an exemplary system for remote monitoring of electrical equipment according to an embodiment of the subject matter described herein;

[0009] Figure IB is a detailed view of an infrared image of a scene in which portions of the image are defined as virtual probes that represent a portion of the equipment being monitored according to an embodiment of the subject matter described herein;

[0010] Figures 2A and 2B are example visible light and infrared images, respectively, acquired by an exemplary system for remote monitoring of electrical equipment according to an embodiment of the subject matter described herein;

[0011] Figures 3 A and 3B show example views of a transformer, provided by an infrared camera but shown as a black and white line drawing for clarity;

[0012] Figure 4 is a flow chart illustrating an exemplary process for remote monitoring of electrical equipment according to an embodiment of the subject matter described herein;

[0013] Figure 5 illustrates an infrared image where a range of data values are mapped to a single color or limited number of colors according to an embodiment of the subject matter described herein;

[0014] Figure 6 illustrates an example screen shot from a user display of an exemplary system for remote monitoring of electrical equipment according to embodiments of the subject matter described herein; andAtty. Docket No.: 8070.0015WO

[0015] Figure 7 is a block diagram of an exemplary hardware module for performing a function described herein.DETAILED DESCRIPTION

[0016] The subject matter disclosed herein includes methods, systems, and non-transitory computer readable media for remote monitoring of electrical equipment. Data can be derived from data different sources, e.g., IR images, RFID sensors, gas detectors, etc., and the data so provided is referred to as a data point. A data point is a numerical value provided by a data source, and may represent a temperature, a pressure, a gas concentration, a voltage, a current, or other measurable characteristic. Each data source provides data points periodically, continuously, or continually. Thus, each data source provides a stream of data points over time.

[0017] In some aspects, all data points are provided by a single data source type (e.g., all from IR cameras, all from RFID sensors, all from wired sensors, etc.). In some aspects, data points are provided by multiple data source types (e.g., from both IR cameras and physical sensors, from physical sensors and gas detectors, etc.) For example, recent improvements in RFID technology have produced RFID temperature sensors that can be accurately and reliably read from several meters away and tens or hundreds of times per second.

[0018] Two or more data sources, and thus the data points provided by those data sources, may be assigned to an analysis group. Data points within each analysis group may be compared with each other, tracked over time, subjected to mathematical or statistical analysis, or combinations thereof. A data source may be assigned to more than one analysis group.

[0019] At least one alarm metric is associated with each analysis group, where an alarm metric is a predefined condition or threshold that indicates a change in operating condition of the electrical equipment being monitored. Changes in operating condition include equipment failure, indications of a potential equipment failure, changes in operating parameters, changes in work output, detections of gas or fluid leakage, changes in operation temperature, and so on. When an alarm metric is satisfied, an alarm isAtty. Docket No.: 8070.0015WOgenerated to bring the change of operating condition to the attention of the operator of the electrical equipment.

[0020] For example, infrared (IR) cameras and RFID probes may report temperature data points of primary terminals of a large electrical transformer. These data points may be part of an analysis group that defines an alarm metric that triggers an alarm if one of the terminals has a temperature that is different from the other terminals by a threshold amount. In some aspects, an IR camera is mounted to provide an unchanging view of equipment to be monitored, where portions of the IR image generated by the IR camera are views of specific components of the equipment.

[0021] For example, an IR image of a transformer may include a first contiguous group of pixels that show primary terminal A of the transformer, a second contiguous group of pixels that show primary terminal B of the transformer, a third contiguous group of pixels that show primary terminal C of the transformer, a fourth contiguous group of pixels that show secondary terminal A of the transformer, a fifth contiguous group of pixels that show secondary terminal B of the transformer, and a sixth contiguous group of pixels that show secondary terminal C of the transformer. Such contiguous groups of pixels may be referred to herein as “virtual probes.” In this example, the IR intensity values of the pixels contained within the first contiguous group of pixels may be averaged to provide a representative intensity value for that contiguous group of pixels, and that representative intensity is a data point for that contiguous group of pixels. By the same process, a data point can be derived for each of the contiguous group of pixels.

[0022] It is noted that a virtual probe may comprise just one contiguous group of pixel, or it may comprise more than one contiguous group of pixels. For example, the operator may define a virtual probe to include all of the pixels of an IR image that display the side of a heat exchanger. If the view of the heat exchanger is unobstructed, the operator can capture the side of the heat exchanger using just one contiguous group of pixels. In this example, the virtual probe comprises just one contiguous group of pixels. However, if the view of the heat exchanger is obstructed by a vertical pole located between the heat exchanger and the IR camera, for example, the operator may need to define two contiguous groups of pixels in order cover the side of the heat exchanger without including the pixels that show the vertical pole - e.g., a first contiguous group ofAtty. Docket No.: 8070.0015WOpixels showing the portions of the heat exchanger visible to the left of the vertical pole and a second contiguous group of pixels showing the portions of the heat exchanger to the right of the vertical pole. In this example, the virtual probe will comprise the combination of the first and second contiguous groups of pixels just described.

[0023] In this example, the data points for the primary terminals of the transformer are part of a first analysis group, the data points for the secondary terminals of the transformer are part of a second analysis group, and the data points for both the primary and secondary terminals of the transformer are part of a third analysis group. In this example, each analysis group has its own alarm metric that triggers an alarm if one of the terminals has a temperature that is different from the other terminals by a threshold percentage, e.g., by 25%. It is noted that each data point may be the average intensity values of the pixels in that contiguous group of pixels - no conversion to a temperature value is needed (although there could also be a conversion-to-temperature step).

[0024] Using the IR image example above, it is noted that the specific set of pixels that shows the portion of the equipment being measured (primary terminal A, for example) is presumed to be the same from frame to frame of the IR image feed; that is, the scene is static and does not jitter - e.g., the IR camera does not shake. This allows the IR image to be partitioned into one or more non-overlapping regions of pixels, each area representing a component of the equipment being monitored (e.g., a first region being the group of pixels that shows primary terminal A, a second region being the group of pixels that shows primary terminal B, and so on.) This operation may also be referred to as mapping, e.g., mapping groups of pixels to distinct components of the equipment being monitored. In some aspects, this mapping may be done by the operator at system configuration, e.g., using a mouse to select and label groups of pixels within a reference image. This is referred to herein as a “static mapping.” In this example, each virtual probe is a group of pixels that display the image of a predefined portion of the equipment being monitored, e.g., a terminal, a bus bar, a transmission line, a valve, pump, etc.

[0025] It is noted, however, that in real-world implementations, the IR camera may shake, e.g., due to wind, earth tremors, or its position may change slightly, e.g., due to expansion and contraction of metal brackets, poles, rails, or mounting hardware. This causes the incoming IR image to differ slightly from the reference image that was used toAtty. Docket No.: 8070.0015WOmap regions of pixels to components of the equipment to be measured. As a result, a region of pixels may not be properly aligned with the component of the equipment to be measured, which may cause the intensity reading from group of pixels to not accurately reflect the temperature of the component that was mapped to that region of pixels.10026] Thus, in some aspects, for data points that are derived from IR images, the system includes an image registration or image stabilization step to register the incoming IR image to a reference IR image so that the static mapping is correct and the data points derived from the incoming IR image accurately reflect the portions of the equipment as intended. In some aspects, the image stabilization process may be performed within the camera. In some aspects, the image stabilization process may be performed by the server or processing system that performs the group analysis and generates the alarms, prior to the mapping step.10027] In some aspects, a camera may provide both an IR image and a visible light image; in some aspects, the IR image registration may be based on registration of the visible light image. In some aspects, multiple scenes may be acquired from multiple cameras, or acquired from a single camera that automatically changes position, orientation, focal length, zoom value, or other image characteristic between scene captures and automatically sends the captured scene to the system for processing. Each scene so captured may have its own independent mapping. In some aspects, a single component of electrical equipment to be monitored may appear in multiple scenes, e.g., in both a wide angle scene and a zoomed scene; in some aspects, the data points for that one component from different scenes may be fused.

[0028] In some aspects, this partitioning may be done by an artificial intelligence / machine learning (AI / 'ML) model at system configuration time and / or run time, e.g., to detect terminals, plugs, bus bars, bushings, switch jaws or hinges, disconnects, circuit breakers, or other identifiable components in each frame. This is referred to herein as “dynamic mapping”. In some aspects, AI / 'ML may be used to automatically identify anomalies in data from disparate data source types. For example, AI / 'ML may be trained to identify anomalies between data points derived from an IR image and data points provided by RFID devices, gas monitors, physical probes, etc. Data may be derived fromAtty. Docket No.: 8070.0015WOmultiple data sources, e.g., from multiple IR cameras, from different types of cameras, from cameras and wired or wireless probes, etc.

[0029] In another example, another analysis group may include temperature data points of primary terminals of an adjacent large electrical transformer, with a similar alarm metric. In yet another example, a third analysis group may include temperature data points of all of the primary terminals of both of the transformers, with an alarm metric that triggers an alarm if any of the terminals exceeds a maximum temperature. In yet another example, a fourth analysis group may include data provided by a gas detector and data provided by an RFID temperature monitor on a transformer seal, with an alarm metric that triggers an alarm if an unacceptable level of gas is detected in solution of transformer oil and a temperature of the transformer seal increases at higher than a threshold rate, for example. These analysis groups and alarm metrics are illustrative and not limiting. The use of multiple analysis groups with independent alarm metrics provides a real-time, multi-viewed analysis of equipment health.

[0030] One advantage to the methods and systems described herein is that they do not require accurate data readings. For example, temperature data derived from IR images does not have be calibrated to an accurate temperature for the alarm metrics to perform correctly. In fact, the alarm metrics perform correctly even using raw sensor data without conversion to a temperature value at all, which obviates the need for expensive sensors and / or computationally demanding conversion, compensation, and calibration routines.

[0031] The subject matter described herein may be implemented in hardware, software, firmware, or any combination thereof. As such, the terms “function” or “module” as used herein refer to hardware, software, and / or firmware for implementing the feature being described.

[0032] In one exemplary implementation, the subject matter described herein may be implemented using a computer readable medium having stored thereon executable instructions that when executed by the processor of a computer control the computer to perform steps. Exemplary' computer readable media suitable for implementing the subject matter described herein include disk memory devices, chip memory devices, programmable logic devices, application specific integrated circuits, and other non-transitory storage media. In one implementation, the computer readable medium mayAtty. Docket No.: 8070.0015WOinclude a memory accessible by a processor of a computer or other like device. The memory may include instructions executable by the processor for implementing any of the methods described herein. In addition, a computer readable medium that implements the subject matter described herein may be located on a single device or computing platform or may be distributed across multiple physical devices and / or computing platforms.

[0033] Methods and systems for remote monitoring and failure prediction of remotely monitored electrical equipment are provided herein. This methods and systems disclosed herein avoid the problems associated with determining an exact temperature reading and instead take advantage of the knowledge that similar equipment in similar conditions — e.g., multiple transformers at a single site — should exhibit approximately the same temperature characteristics, and that likelihood of failure of a particular transformer can be accurately predicted based on a relative comparison of similar equipment rather than an absolute temperature reading. For example, if one transformer is measurably hotter than its neighbors — regardless of what the actual temperature readings are — that transformer is more likely to fail. The same is true for a transformer that is measurably hotter than other transformers operating under the same or similar environment and operating conditions, regardless of whether the other transformers are physically close to or geographically diverse from each other.

[0034] Thus, unlike conventional systems that take great pains to take an exact temperature reading of a transformer and use that absolute temperature value to calculate likelihood of failure, the subject matter disclosed herein instead seeks to identify transformers that are relatively hotter than their neighbors by a threshold amount. As will be described in more detail below, the methods and systems disclosed herein allow such comparisons to be made across many collections of relative temperature measurements simultaneously and allow each comparative measurement to be associated with an independent set of comparison and alarm metrics, which provides a rich set of views by which equipment health may be monitored. Because the methods and systems described herein may operate on relative, rather than absolute, values, it is not necessary to perform computationally expensive and error-prone calibration of sensor data to an accurate absolute temperature.Atty. Docket No.: 8070.0015WO

[0035] Furthermore, because the mathematical analysis techniques employed herein are “data-agnostic”, i.e., they work regardless of what the data values represent, the methods and systems described herein not only do not require calibration of data values to an accurate temperature, but also do not require that the data be converted to a temperature at all. That is, they methods and systems described herein can accept as input the raw data sensor in any form, and in any unit of measurement, whether it be volts, amps, lumens, charge, etc. In other words, the systems and methods of the subject matter described herein correctly detect increased likelihood of equipment failure using raw data without the need to first convert that raw data into a temperature value. As a result, systems and programs operating according to the concepts described herein can be greatly simplified and streamlined while producing equally valuable and reliable results that are comparable to results produced by much more complicated and expensive systems that require calibrated temperature values as inputs. It should be noted that while data values may be converted to temperature values to be displayed for the equipment user’s convenience and ease of understanding, the conversion to temperature is not required for the analysis and detection of potential equipment failure.

[0036] Figure 1A is a block diagram illustrating an exemplary system for remote monitoring of electrical equipment according to an embodiment of the subject matter described herein. According to one aspect, system 100 includes a data acquisition module 102 for acquiring a set of data points. In some aspects, each data point representing a temperature associated with a piece of electrical equipment or a component thereof. The data points passed to a monitoring module 104, where the data points may be assigned to groups. Each data point may belong to multiple groups simultaneously, and each group contains two or more data points. An alarm metric is defined for each group, and one group’s metric may be different from another group’s metric. The defined alarm metrics are used to monitor and determine the health of the electrical equipment. Upon detection of an alarm condition according to one or more of the alarm metrics, the system may notify the user of the detected alarm condition.

[0037] For example, if one terminal of a multi-terminal transformer is hotter than the other terminals by a threshold amount, this may be indicative of a potential failure of that transformer. Thus, an example alarm metric might be to generate an alarm if any dataAtty. Docket No.: 8070.0015WOpoint within a group has a data value that exceeds the other data values by a threshold amount. If the data values of the data points are converted to temperatures, then the example alarm metric might be to generate an alarm if any temperature reading within the group exceeds the other temperature readings by a threshold amount. Other examples will be described below.

[0038] Data acquisition module 102 may receive data points via a variety of types of sensors. In the embodiment illustrated in Figure 1A, for example, data acquisition module 102 may use an infrared image sensor 106 for capturing an image 108 of a scene 110 that includes a view of the electrical equipment to be monitored 112. As used herein, the term “scene” refers to a collection of real-world objects as observed by an image sensor. Thus, “a scene that contains transformers”, for example, is a view, as seen by an image sensor, of a volume of space that includes transformers. Since the output of such an image sensor is a 2D (or 3D) image, the image of a scene is a representation of the actual scene. For this reason, the terms “image” and “scene” may be used interchangeably herein where the distinction can be determined by context or where the distinction is irrelevant in the context of the description.

[0039] In one embodiment, image sensor 106 may also include a visible light camera, in which case image 108 may include both a visible light view of the scene (left pane) and an infrared light view of the scene (right pane.) Data acquisition module may use the false color or grayscale image produced by infrared image sensor 106 to determine the temperature of the equipment being observed. For reasons that will be explained in more detail below, system 100 can use a technique that does not require an exact temperature reading or even conversion of the data to a temperature value. For this reason, the systems and methods described herein have a speed and cost advantage over conventional systems that rely on highly accurate temperature readings. Even if sensors used by the systems and method described herein provide numbers that purport to be actual temperature values, these values are also treated as mere numbers that are input into the calculations. No conversion or calibration to “true” temperature values is needed or performed.10040] According to another aspect, data acquisition module 102 may receive data points via other means. In the embodiment illustrated in Figure 1A, for example, closeAtty. Docket No.: 8070.0015WOup view 114 shows the use of a sensor 116 that is physically attached (or proximate) to the equipment being monitored and that wirelessly transmits a temperature or other sensor reading to data acquisition module 102. In this embodiment, data acquisition module 102 may include a wireless receiver for wireless communication with one or more sensors 116.

[0041] In embodiment, sensor 116 may be a radio frequency identifier (RFID) device that includes or that can be coupled with a temperature sensor. RFID devices may be active or passive. For passive RFID devices, system 100 may include RFID reader functionality to interrogate the passive RFID device and read the data value being transmitted from the RFID device in response to that interrogation. Active RFID devices may transmit the data unilaterally (e.g., the RFID device is a “beacon”), in response to a query or other trigger, or other operation. Other types of wireless sensors may be used for sensor 116.

[0042] Figure 1B is a detailed view of an infrared image of a scene in which portions of the image are defined as virtual probes (which may also be referred to as “visual probes”) that represent a portion of the equipment being monitored according to an embodiment of the subject matter described herein. In the example image 118, portions of the infrared image of a piece of electrical equipment - in this case a high voltage transformer - are identified as areas of the image that represent portions of the transformer that are to be monitored. These areas, which are referred to herein as “virtual probes”, are identified by white outlines. In one embodiment, the operator or user of the system manually selects the portions of the image, e.g., using a mouse, stylus, or other means to draw lines that enclose a collection of pixels within the image, as seen in Figure 1B. In another embodiment, an image analysis algorithm may be used to identify target or potential portions of the image, which a user may or may not then manually confirm. Examples of such image analysis algorithms include, but are not limited to, edge detection, color gradient detection / contour mapping, etc.

[0043] Each virtual probe may include one or more areas that enclose a contiguous group of pixels. That is, a single virtual probe may have just one contiguous group of pixels (e.g., the probe includes just one area of the image), or it may have multiple, non¬ contiguous areas, each area containing a contiguous group of pixels (e.g., the probeAtty. Docket No.: 8070.0015WOincludes multiple areas that do not touch each other). In this manner, a collection of separate areas of an image may be treated as a single unit. This is useful, for example, when it is desired to monitor the average temperature of a collection of objects within the virtual field. The same concept may be applied to group a collection of RFID devices (or the collection of data that such a group of devices would produce) as a single probe.

[0044] Although the term “virtual probe” is used throughout this document for simplicity, it will be understood that the information that is provided by a virtual probe may also be provided by a non- virtual or actual probe, e g., by the RFID devices described above. Just as a single probe may include multiple non-contiguous areas of an image, and a single probe may include a collection of data from multiple RFID devices, a single probe may also include combinations of the above. For example, a single probe may incorporate both data from an area of an infrared image and a temperature reading from an RFID. Thus, the use of the term “virtual probe” is not intended to refer exclusively to “areas of an image” but also contemplates non-virtual probes or combinations of virtual and non-virtual probes.

[0045] In the embodiment illustrated in Figure 1B, thirteen different portions of image 118 have been defined as virtual probes. In this example, probes Pl A, P1B, and PIC represent input terminals of a first three-phase transformer and probes P2A, P2B, and P2C represent output terminals of that transformer. Probes P3A, P3B, and P3C represent input terminals of a second three-phase transformer and probes P4A, P4B, and P4C represent output terminals of the second three-phase transformer. Virtual probe P5 represents the body of the transformer(s). A color scale on the right-side of image 118 shows the correlation between the false-color infrared image and temperature values, with white / yellow representing higher temperatures and blue / black representing lower temperatures. In this example image, the body of the transformer (P5) is bright, indicating that is hotter than the terminals (P

[0012] [ABC]), which are darker. It is noted again that data values may be converted to temperature values to be displayed for the equipment user’s convenience and ease of understanding, but the conversion to temperature is not required for the analysis and detection of potential equipment failure.

[0046] Once virtual probes have been defined, they can be assigned to one or more groups of probes. Using image 118 as an example, probes Pl A, P1B, and PIC may beAtty. Docket No.: 8070.0015WOmembers of a group called “T1 input”, while probes P2A, P2B, and P2C may be members of another group called “Tl output”. Probes may be assigned to multiple groups. For example, probes P1A, P1B, PIC, P3A, P3B, and P3C may form a group called “All_inputs”; probes Pl A, P1B, PIC, P2A, P2B, P2C, P3A, P3B, P3C, P4A, P4B, and P4C may form a group called “All terminals”, and so on.

[0047] Alternatively, the non-contiguous areas of image 118 labeled P l A, P1B, and P1B could have been grouped into a single “probe”, called “Pl”, for example. Where it is not necessary or advantageous to know the temperature of each individual terminal of a transformer, all of the terminals may instead be treated as a single probe. As stated above, any area of an image may belong to multiple probes, which means that the system may be configured to monitor each terminal separately and also monitor the group of terminals as a set, potentially with different rules, metrics, alarms, and so on, for each. Defining a single virtual probe as containing multiple non-contiguous areas, for example, also allows an object that is too large to fit on just one image - e.g,, some parts of it show up in one image and other parts of it show up in another image - to be monitored as a whole unit, by defining a single virtual probe as containing multiple areas that show up in different displays or that span multiple images or displays. This is useful to measure the health of very large pieces of equipment, for example.

[0048] Each group may be assigned its own set of rules for determining the health and operation status of the equipment represented by the probes in that group. The set of rules for a group, which may define include mathematical algorithms or other tools for analysis of the probe data, is collectively referred to as the “alarm metric” for the group, since one purpose of the system is to identify potential equipment failure and alert the operator, e.g., sound an alarm, when certain conditions are detected, but the same rules may be used for other purposes as well, such as to provide a picture of general health or operation status, etc., without necessarily generating an alarm.

[0049] An alarm metric can include any type of analysis of the data points that make up the group. The following list of examples is for illustration purposes and is not intended to be limiting:

[0050] One example alarm metric compares the value of each data point in the group to the values of the other data points in the group and determines whether any one valueAtty. Docket No.: 8070.0015WOis significantly different from the rest of the values in the group. For example, the alarm metric may look for outliers, e.g., it may calculate an average value for the group and look for data points that deviate from the average value by a threshold amount. In another example, a variance of the data points may be calculated, and an alarm raised if the variance exceeds a threshold value.

[0051] Another example alarm metric may plot a trend of data values over time and look for changes in data value that may indicate a problem. For example, an alarm metric may look for one or more data values that have a rate of data change over time higher than the rate measured for other probes in the group, which may indicate the impending failure of one phase or terminal of a transformer. Likewise, an alarm metric may set an alarm if it detects that all probes in the group show a rate of data change higher than expected, which may indicate the impending failure of the transformer as a whole.10052] Yet another example alarm metric compares the data values of one transformer to the data values of another transformer that is subject to the same environmental conditions, e.g., two transformers on the same site, subject to the same amount of sun, shade, cloudy or rainy weather, ambient temperature, etc., to see if one transformer is hotter than its similarly-situated neighbor. The advantage of such comparative measurements is that it is not necessary to determine an exact temperature: the system need only determine that there is a threshold difference in data values. For example, a system need only detect that transformer A is at least 20 degrees hotter, say, than transformer B: it is not necessary to determine whether their temperatures are 35 °C and 55°C, or 47°C and 67°C, or 39.1°C and 59.1°C, for example. The accuracy of the temperature reading is much less important than the difference in data values.

[0053] Yet another example alarm metric compares the data values of one transformer to data values of other transformers, regardless of where the other transformers are located, e.g., the other transformers could be geographically co-located or they could be geographically diverse.

[0054] Yet another example alarm metric may consider not only data values that represent measured equipment temperature but also data values that represent other aspects, such as current operating conditions of the equipment (e.g., whether it is operating at full capacity or not, whether it is operating under a full load or not, whetherAtty. Docket No.: 8070.0015WOit is a week day, weekend, or holiday, and so on), environment (e.g., whether it is raining or not, whether there is standing water or not, and so on), and even site occupancy (e.g., whether there are people present or not, and whether intruders and / or unauthorized personnel have been detected or not). Any type of data may be considered and / or factored into an alarm metric. Furthermore, data from any source, location, and / or type may be considered, including data from locations other than the location of the equipment being monitored.

[0055] This has several distinct benefits. One benefit is that it is not necessary to use expensive, high-precision infrared cameras or other imaging devices; cheaper, less accurate infrared image sensors can be used. Another benefit is that it is not necessary to perform a conversion of the data value to a temperature at all, much less performing the difficult and / or computationally expensive calibration of the image sensor to register the image color or intensity to a precise temperature value. Likewise, algorithms and routines to adjust temperature readings based on distance from the sensor may be obviated. In Figure 1B, for example, probes P4A, P4B, and P4C (herein referred to as “group P4”) appear to be brighter than probes Pl A, PIB, and PIC (herein referred to as “group Pl”), but this may be an artifact due to the fact that the probes in group Pl are physically closer to the camera than are the probes in group P4 If each group includes an alarm metric that looks for variation in data value within each group, an overheated terminal will be detected correctly even without having to calculate an exact temperature value compensated for distance from the camera. This is illustrated in Figures 2A and 2B.

[0056] Figures 2A and 2B are example visible light and infrared images, respectively, acquired by an exemplary system for temperature-agnostic remote monitoring of electrical equipment according to an embodiment of the subject matter described herein. Figure 2A shows the scene in visible light 200, and Figure 2B shows the same or similar scene in infrared 202. The relative or absolute data value of the scene is determined based on a color (for false color images) or intensity (for black and white images) of the image.

[0057] In Figure 2B, infrared image 202 is a view of a transformer in which three virtual probes, TA, TB, and TC, have been defined as representing the three-phase outputAtty. Docket No.: 8070.0015WOterminals of the transformer. This image shows that one of the terminals, represented by virtual probe TA, is markedly hotter than the other two terminals represented by virtual probes TB and TC, despite the fact that all three terminals are subject to the same environmental conditions. An alarm metric that looks for deviations from the average would detect TA as being of concern, even without knowing precisely the exact temperature of TA.10058] In addition, the ability to include terminal TA in other groups, each group having its own alarm metric, means that each virtual probe may be subject to multiple alarm metric simultaneously, which can increase the likelihood that impending failure may be detected. For example, in the scenario where the transformer in Figure 2B is overheating on all terminals ~ e g., TA, TB, and TC are all hotter than they should be, if TA, TB, and TC are members of another group that includes terminals from other, similarly situated transformers, that alarm metric may identify all three terminals TA, TB, and TC as being outliers when compared to terminals from other transformers, and raise an alarm appropriately. The many-to-many relationship between virtual probes and groups allows enormous flexibility to provide very robust coverage and real-time or offline analysis based on multiple data points and multiple algorithms.

[0059] The locations of the virtual probes presume that the locations of the portions of the equipment being monitored within the infrared image do not change. This presumption may not always be true, however. For example, if the infrared camera that is providing the infrared image shown in Figure 2A is mounted on a pole, tower, or other mounting point that is susceptible to wind or vibration, the view of the equipment being monitored may shift from frame to frame relatively rapidly. Likewise, if the infrared camera is mounted to a building or other object that may settle over time, the view of the equipment being monitored may also shift over time relatively slowly. In either scenario, the infrared image that is provided may be shifted so much that one or more of the virtual probes may no longer contain an image of the portion of equipment that the virtual probe was intended to monitor. An example of this is shown in FIGS. 3 A and 3b.

[0060] Figure 3 A shows an example view’ of a transformer 300, provided by an infrared camera but shown as a black and white line drawing for clarity. As shown in Figure 3A, six virtual probes have been defined, each virtual probe being a fixed andAtty. Docket No.: 8070.0015WOpredefined set of pixels within the scene corresponding to a fixed and predefined portion of the equipment to be monitored, which in this example is a transformer 300. Virtual probes A, B, and C monitor the primary’ terminals of the transformer 300 (referred to respectively as terminal TA, TB, and TC), and virtual probes D, E, and F monitor the secondary terminals of the transformer 300 (referred to respectively as terminal TD, TE, and TF).

[0061] The locations of the virtual probes A-F within the image are defined based on a presumption that the transformer 300 is always in the same location within the image, referred to herein as the “reference location.” Using the simplified example shown in Figure 3A, one corner of the transformer 300 is presumed to be located at the same location in the image, labeled XR and YR, where “R” stands for “reference.” So long as that corner of the transformer 300 is located at (XR, YR) in the image, the virtual probes A-F will accurately represent the terminals that the virtual probes are intended to monitor,

[0062] Figure 3B shows an example view of the transformer 300, provided by an infrared camera that has shifted slightly (e.g., due to wind, vibration, settling, etc.) and therefore provides an image of a view that is different from the expected, or reference, view. In the example shown in Figure 3B, the camera has moved or rotated slightly to the left and down, which causes the image of the transformer 300 to be slightly to the right and slightly above its reference location. As a result, the corner of the transformer 300 is not located at location (XR, YR) in the image, but at location (XA, YA) in the image, where “A” stands for “actual.” In this example, the term ΔX refers to the difference between XR and XA, and the term ΔY refers to the difference between YR and YA.

[0063] Because of the movement of the camera that provided the image shown in Figure 3B, the data captured by the virtual probes A-F no longer represent the intended portion of the equipment being monitored. That is, that the portions of the image that are contained within the fixed and predefined sets of pixels within the scene - i.e., the virtual probes - fail to display the portions of the equipment that the virtual probe is intended to monitor. For example, in Figure 3B, virtual probe A does not capture a terminal at all, virtual probe B captures data for terminal TA instead of terminal TB, virtual probe C captures data for TB instead of terminal TC, data for terminal TC is not captured by anyAtty. Docket No.: 8070.0015WOvirtual probe, and so on. This can lead to erroneous results during analysis of the shifted image.

[0064] Figure 3B illustrates a scenario in which the infrared image shifts so much that the portion of the image defined by a virtual probe do not contain any of the portion of the equipment that the particular virtual probe was intended to monitor, but in other scenarios, the movement of the camera - and thus the shift of the corresponding image -is small enough that the virtual probe contains an image of at least some of the portion of the equipment that the virtual probe is intended to monitor. This can also lead to erroneous results during analysis of the shifted image. If the image shifts such that the virtual probe also captures an image of a portion of equipment that is cool, this can cause the value of temperature calculated from that virtual probe to be lower than the actual temperature of the portion of the equipment that the virtual probe was intended to monitor. Likewise, if the image shifts such that the virtual probe also captures an image of a portion of equipment that is hot, this can cause the value of temperature calculated from that virtual probe to be higher than the actual temperature of the portion of the equipment that the virtual probe was intended to monitor.

[0065] Thus, the displacement and / or rotation of an actual image from a reference image can lead to inaccurate or erroneous results during analysis of the shifted image. In order to avoid this problem, improved techniques for remote monitoring of equipment are herein disclosed. In an aspect, infrared image data is subjected to an image registration technique prior to being portioned into virtual probes. Doing so ensures that the portions of the image that occupy the virtual probes display the portions of the equipment that the virtual probe is intended to monitor.

[0066] Image registration is the process of transforming different sets of data into one coordinate system to achieve spatial correspondence. The sets of data may be, for example, images from different sources or images taken from the same source at different times. One type of image registration is spatial alignment, which aligns multiple images of the same scene to a common coordinate system for accurate comparison and analysis. In some aspects, image registration aligns a series of captured images (e.g., such as a series of frames in a video stream or a series of time-lapse images) to each other to remove visible jitter. In other aspects, image registration aligns a series of capturedAtty. Docket No.: 8070.0015WOimages to a reference image - or aligns visual elements with the captured images to a reference location - so that the scene does not shift over time.

[0067] There are a number of techniques that can be used to ensure that the portions of the image that are contained within the fixed and predefined sets of pixels within the scene (i.e., the virtual probes) display the portions of the equipment that the virtual probe is intended to monitor, i.e., to ensure that there is spatial correspondence between each virtual probe and the portion or portions of the equipment being monitored that the respective virtual probe is intended to display.10068] In some aspects, the coordinate locations of the virtual probes are unchanged, but the incoming infrared image is adjusted (translated, rotated, scaled, etc.) so that virtual probes display the intended portions of the equipment being monitored. This approach may be referred to herein as “adjusting the image to the virtual probes.” Referring to the example in Figure 3B, the incoming image may be shifted by (ΔX,ΔY) –in this example, down and to the left - so that bottom corner of the transformer 300 is located at (XR, YR) rather than at (XA, YA). Once the image is properly registered, the virtual probes A-F display their respective intended portion or portions of the equipment being monitored.

[0069] In some aspects, the incoming infrared image is unmodified, but the locations of the virtual probes are adjusted (translated, rotated, scaled, etc.) so that they display their respective intended portions of the equipment being monitored. This approach may be referred to herein as “adjusting the virtual probes to the image.” Referring again to Figure 3B, in this example, the location of each of the virtual probes A-F is shifted by (-ΔX,-ΔY) – in this example, up and to the right - so that each virtual probe displays the respective intended portion or portions of the equipment being monitored.

[0070] In some aspects, both the image and the virtual probes may be adjusted m order to ensure spatial correspondence between the virtual probes and the portion or portions of the equipment being monitored that the respective virtual probe is intended to display. In some aspects, the decision whether to adjust the image to match the virtual probes or to adjust the virtual probes to match the image may be made according to an operator setting and / or may be made according to which approach is more cost-, time-, or resource-effective.Atty. Docket No.: 8070.0015WO

[0071] Image registration or image alignment algorithms can be classified into intensity-based and feature-based. One of the images is referred to as the moving or source and the others are referred to as the target, fixed or sensed images. Image registration involves spatially transforming the source / moving image(s) to align with the target image. The reference frame in the target image is stationary, while the other datasets are transformed to match to the target.

[0072] In some aspects, the infrared image data is subjected to an intensity-based image registration method. Intensity-based methods compare intensity patterns in images via correlation metrics, while feature-based methods find correspondence between image features such as points, lines, and contours. Intensity-based methods register entire images or sub-images. If sub-images are registered, centers of corresponding sub images are treated as corresponding feature points. Intensity-based methods include, but are not limited to:• Mutual Information: Measures the statistical dependency between the intensities of corresponding pixels m the images. It is particularly useful for multimodal image registration where images come from different modalities.• Mean Squared Error (MSE): quantifies the difference between pixel intensities in aligned images. It is a straightforward metric used to measure alignment quality.• B-splines: A flexible method for modeling smooth deformations. They represent the transformation as a combination of basis functions, allowing for smooth and continuous transformations.• Thin-Plate Splines: Used for interpolating non-rigid transformations by minimizing bending energy. They provide a smooth transformation that can handle complex deformations,

[0073] In some aspects, the infrared image data is subjected to a feature-based registration method. Feature-based methods establish a correspondence between a number of especially distinct points in images. Knowing the correspondence between a number of points in images, a geometrical transformation is then determined to map the target image to the reference images, thereby establishing point-by-point correspondence between the reference and target images. Methods combining intensity-based and feature-Atty. Docket No.: 8070.0015WObased information have also been developed. Feature-based methods include, but are not limited to:• SIFT (Scale-Invariant Feature Transform): A robust method for detecting and describing local features in images. It identifies keypoints that are invariant to scale and rotation, making it effective in matching features across images with varying sizes and orientations.• SURF (Speeded-Up Robust Features): An accelerated alternative to SIFT, designed to be faster while maintaining similar robustness. It uses integral images to speed up the computation of the detector and descriptor.• ORB (Oriented FAST and Rotated BRIEF): Combines the FAST keypoint detector with the BRIEF descriptor, providing a fast and efficient feature extraction method. It includes orientation information to improve robustness to rotation.• RANSAC (Random Sample Consensus): A robust algorithm used for estimating transformation models while handling outliers. It is often used in conjunction with feature-based methods to estimate the best transformation between images.

[0074] Image registration algorithms can also be classified according to the transformation models they use to relate the target image space to the reference image space. In some aspects, the infrared image data is subjected to linear transformations, which include rotation, scaling, translation, and other affine transforms. Linear transformations are global in nature, thus, they cannot model local geometric differences between images.

[0075] In some aspects, the infrared image data is subjected to 'elastic' or 'nonrigid' transformations. These transformations are capable of locally warping the target image to align with the reference image, Nonrigid transformations include radial basis functions (thin-plate or surface splines, multiquadrics, and compactly-supported transformations), physical continuum models (viscous fluids), and large deformation models (diffeomorphisms).

[0076] Transformations are commonly described by a parametrization, where the model dictates the number of parameters. For instance, the translation of a full image canAtty. Docket No.: 8070.0015WObe described by a single parameter, a translation vector. These models are called parametric models. Non-parametric models on the other hand, do not follow any parameterization, allowing each image element to be displaced arbitrarily.

[0077] Alternatively, many advanced methods for spatial normalization are building on structure preserving transformations homeomorphisms and diffeomorphisms since they carry smooth submanifolds smoothly during transformation. Diffeomorphisms are generated in the modern field of Computational Anatomy based on flow’s since diffeomorphisms are not additive although they form a group, but a group under the law of function composition. For this reason, flows which generalize the ideas of additive groups allow for generating large deformations that preserve topology, providing 1-1 and onto transformations. Computational methods for generating such transformation are often called LDDMM which provide flows of diffeomorphisms as the main computational tool for connecting coordinate systems corresponding to the geodesic flows of Computational Anatomy.

[0078] In some aspects, the infrared image data is subjected to spatial methods of image registration. Spatial methods operate in the image domain, matching intensity patterns or features in images. Some of the feature matching algorithms are outgrowths of traditional techniques for performing manual image registration, in which an operator chooses corresponding control points (CP) in images. When the number of control points exceeds the minimum required to define the appropriate transformation model, iterative algorithms like RANSAC can be used to robustly estimate the parameters of a particular transformation type (e.g. affine) for registration of the images.

[0079] In some aspects, the infrared image data is subjected to frequency- domain methods of image registration. Frequency-domain methods find the transformation parameters for registration of the images while working in the transform domain. Such methods work for simple transformations, such as translation, rotation, and scaling. Applying the phase correlation method to a pair of images produces a third image which contains a single peak. The location of this peak corresponds to the relative translation between the images. Unlike many spatial-domain algorithms, the phase correlation method is resilient to noise, occlusions, and other defects typical of medical or satellite images. Additionally, the phase correlation uses the fast Fourier transform to compute theAtty. Docket No.: 8070.0015WOcross-correlation between the two images, generally resulting in large performance gains. The method can be extended to determine rotation and scaling differences between two images by first converting the images to log-polar coordinates. Due to properties of the Fourier transform, the rotation and scaling parameters can be determined in a manner invariant to translation.

[0080] In some aspects, the infrared image data is subjected to single-modality or multi-modality image registration methods. Single-modality methods tend to register images in the same modality acquired by the same scanner / sensor type, while multi -modality registration methods tend to register images acquired by different scanner / sensor types. Multi -modality registration methods are often used in medical imaging as images of a subject are frequently obtained from different scanners. Examples include registration of brain CT / MRI images or whole body PET / CT images for tumor localization, registration of contrast- enhanced CT images against non-contrast-enhanced CT images for segmentation of specific parts of the anatomy, and registration of ultrasound and CT images for prostate localization in radiotherapy.

[0081] In some aspects, the infrared image data is subjected to registration methods having different levels of automation. Manual, interactive, semi-automatic, and automatic methods have been developed. Manual methods provide tools to align the images manually. Interactive methods reduce user bias by performing certain key operations automatically while still relying on the user to guide the registration. Semi-automatic methods perform more of the registration steps automatically but depend on the user to verify the correctness of a registration. Automatic methods do not allow any user interaction and perform all registration steps automatically.

[0082] In some aspects, an image similarity measure quantifies the degree of similarity between intensity patterns in two images. The choice of an image similarity measure depends on the modality of the images to be registered. Common examples of image similarity measures include cross-correlation, mutual information, sum of squared intensity differences, and ratio image uniformity. Mutual information and normalized mutual information are the most popular image similarity measures for registration of multimodality images. Cross-correlation, sum of squared intensity differences and ratio image uniformity are commonly used for registration of images in the same modality.Atty. Docket No.: 8070.0015WO

[0083] In some aspects, the infrared image data is subjected to a transformation model for image registration. Transformation models are mathematical models used to align images by mapping the coordinates of one image to another. They can be broadly categorized into:• Affine Transformation: A linear transformation that preserves points, straight lines, and planes. It includes rotation, scaling, translation, and shearing. Suitable for aligning images with similar perspectives but differing in scale or orientation.• Projective Transformation (Homograph y): A more complex transformation that can model changes in perspective, such as those caused by camera angles. Effective for aligning images captured from different viewpoints where perspective distortion is significant.• Non-Rigid Transformation: Models that account for deformations that are not purely linear, such as warping and stretching. Types include thin-plate splines, B- splines. Useful for aligning images with complex distortions or biological structures.

[0084] In some aspects, the infrared image data is subjected to a deep learning based method for image registration. Deep learning based methods include, but are not limited to:• U-Net: A convolutional neural network (CNN) architecture designed for image segmentation and registration. It features an encoder-decoder structure with skip connections, allowing it to capture detailed spatial information.• Voxelmorph: A deep learning model specifically designed for image registration.It predicts dense deformation fields that align images by learning from training data.• DeepReg: An end-to-end deep learning framework for medical image registration.It integrates both feature extraction and registration into a single model, trained to optimize registration directly.

[0085] Figure 4 is a flow chart illustrating an exemplary method 400 for remote monitoring of electrical equipment according to an embodiment of the subject matterAtty. Docket No.: 8070.0015WOdescribed herein. In some aspects, method 400 may be performed by an apparatus (e.g., any apparatus described herein). The method 400 includes the following steps:

[0086] At operation 402, the apparatus may receive a scene of an infrared image of equipment to be monitored. In some aspects, the apparatus may receive the scene from an infrared camera. In some aspects, the apparatus may retrieve the scene that has been stored in memory or data storage device, which may be local to or remote from the apparatus. In some aspects, the scene may be selected from a stream of images or extracted from a video stream.10087] At operation 404, the apparatus may register the scene. In some aspects, registering the scene comprises registering the scene to a reference scene. In some aspects, registering the scene may include registering the scene using an image registration algorithm. Example image registration algorithms include, but are not limited to, an intensity-based image registration method, a feature-based image registration method, a linear transformation method, an elastic transformation method, a spatial method image registration method, a frequency-domain method image registration method, a single-modality image registration method, a multi-modality image registration method, a deep-learning, machine-learning, or artificial intelligence based method, or a combination thereof.

[0088] At operation 406, the apparatus may partition the scene into a plurality of fixed, non-overlapping areas, each area having a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored.

[0089] At operation 408, the apparatus may assign each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas.

[0090] At operation 410, the apparatus may acquire a plurality of data values, one for each of the fixed, non-overlapping areas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area. In some aspects, the data values are used as-is, e.g., without converting the data value into a temperature value prior to analyzing the data to determine an alarm condition. In some aspects, the data values may be converted to temperature values prior to analyzing the data to determine an alarm condition, for purposes of providing information to an operator (e.g., in a monitor screen, dashboard, etc.), or for both.Atty. Docket No.: 8070.0015WO

[0091] Health of equipment often may be determined by looking at trends over time. In some aspects, the change of standard deviation over time (which may be referred to herein as “SIGMA*DELTA / T”) of a group of data points may provide valuable information about the equipment’s health. For example, if the standard deviation of the data from a group of probes changes over time such that the rate of change over a defined period changes more than a threshold amount (e g., the data value for one or more probes begins to deviate more and more from the other probes over time) the equipment corresponding to that group of probes may be flagged as a potential candidate for failure,

[0092] At operation 412, the apparatus may define an alarm condition for each of the one or more groups, where each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group. Example alarm conditions include, but are not limited to: detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold rate; detecting that a variance or standard deviation of data values from the fixed, non¬ overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-overlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold rate; or detecting that data values from the fixed, non-overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, where the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and where the first equipment and second equipment are subject to a same set of environmental conditions

[0093] At operation 414, the apparatus may detect that the alarm condition for a first group of the one or more groups has been met.

[0094] At operation 416, the apparatus may in response to detecting that the alarm condition for the first group has been met, provide an operator alert that the alarm condition for the first group has been met.

[0095] In some aspects, acquiring the plurality of data values further may include acquiring at least one additional data value from a sensor that is attached to theAtty. Docket No.: 8070.0015WOequipment to be monitored and assigning the at least one additional data value to at least one of the one or more groups. In some aspects, the sensor transmits the data value via a wireless communications links, or a combination thereof.

[0096] In some aspects, the apparatus includes a display for showing, to an operator, an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas. In some aspects, a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image. In some aspects, the apparatus is configured to adjust an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator. In some aspects, the apparatus is configured to map a sub-range of data values to a single color or intensity value. In some aspects, the apparatus is configured to display data values in a first subrange in color and to display data values in a second sub-range in grayscale or black and white.

[0097] To provide an easily-understood visual indicator of temperature of the equipment being monitored, a range of colors or intensity values to be displayed on a monitor or other display device may be mapped to a range of temperatures. It could also be said that the range of temperatures may be mapped to a range of colors or intensity values. For brevity of description, the term “color” will hereinafter be understood to mean “color or intensity value” unless it is clear from context that color exclusively is intended.

[0098] In one embodiment, a contiguous range of colors within the visible spectrum may be mapped across a contiguous range of temperatures. The mapping of color to temperature may be according to a linear function, a logarithmic function, an exponential function, or any other mathematical or non-mathematical function. For ease of operation, in one embodiment, the color map or intensity map may be adjusted automatically or manually by the user. For example, the range of temperatures signified by the available colors or intensity values may be expanded at the cost of resolution of data values. Likewise, the available colors or intensity values may be mapped to a smaller range of data values so that the user can distinguish temperatures at a higher resolution, but at the expense of limited range. An example of this can be seen in Figure IB, which displaysAtty. Docket No.: 8070.0015WOtemperatures in a range from 34.8°C to 101 °C using a continuous range of colors from black through blue, purple, read, orange, and yellow, ending with white, as shown in the color legend that appears within the image near the right edge of the image.

[0099] In one embodiment, the user may map a range of data values to a single color or limited number of colors, w’here each discrete color represents a temperature or range of temperatures. This is illustrated graphically in Figure 5.

[0100] Figure 5 illustrates an infrared image where a range of data values are mapped to a single color or limited number of colors according to an embodiment of the subject matter described herein. In the embodiment illustrated in Figure 5, infrared image 500 is subject to a color mapping algorithm that maps data values below a threshold value into a grayscale value and maps data values above the threshold value to yellow. Image 500 shows a portion 502 that is colored yellow to indicate that the data value of that portion of the equipment in the scene is above a threshold value. This technique makes it much easier for an operator to notice a high data value reading on the infrared image when compared to viewing a full color infrared image, such as is shown in Figure 2B. In other examples, data values may be mapped to yellow, orange, and red, corresponding to temperatures that are slightly higher than normal, higher than normal, and much higher than normal, for example. The ranges of temperatures assigned to the colors need not be contiguous. For example, one color may be used to display temperatures that are below a first threshold and another color may be used to display temperatures that are above a second threshold that is higher than the first threshold, and temperatures between the first and second thresholds are displayed in greyscale (or perhaps not displayed at all).

[0101] Figure 6 is an example screen shot from a user display of an exemplary system for remote monitoring of electrical equipment according to embodiments of the subject matter described herein. In Figure 6, an example graphical user interface 600 includes several panels, such as a graph of historical temperature data (top panel 602), a selection of different views to choose from (bottom panel 604), and information about the currently selected view (middle panel 606). In the embodiments illustrated in Figure 6, the middle panel 606 includes a visible light image 608, a thermal image 610, a site map 612 showing the available images and their relative locations to each other, and a probe list 614 showing the virtual probes defined in the particular view. During operation, aAtty. Docket No.: 8070.0015WOuser selects a view, either from the views in the bottom panel 604 or from the site map 612. This causes the middle panel 606 to change to show the visible light image 608 and thermal image 610, respectively, of the selected view. The user can then select probes in the probe list 614; data from the selected probes is displayed in the top panel 602.

[0102] The methods and systems disclosed herein may be implemented using hardware or hardware in combination with software and / or firmware. The functions described herein may be performed by one or more hardware modules.

[0103] Figure 7 is a block diagram of an exemplary system for performing a function described herein. In the embodiment illustrated in Figure 7, system 700 may include a processor 702 that executes instructions that may be stored locally and that may be fetched from main memory 704. Main memory may be volatile, non-volatile, or a combination of the two. System 700 may include non-volatile memory 706, such as ROM, EPROM, EEPROM, FLASH, and the like. System 700 may include a network interface device 708 for communicating over a network 710. System 700 may include a video display 712, which may be used to provide a graphic user interface (GUI) to a user, an alphanumeric input device 714, such as a keyboard, and a cursor control device 716, such as a mouse, pointer, stylus, touch screen, etc. System 700 may include mass storage, such as a hard disk drive or solid state drive 718, which may be used to store instruction code. System 700 may include a signal generation device 720 or other peripheral for communicating with or controlling external devices. The modules of system 700 may communicate with each other via one or more busses 722.

[0104] Calculating a temperature differential is much easier and less rigorous than calculating an exact temperature. For example, it is relatively easy to calculate that one transformer is 20 degrees hotter than another, but relatively hard to determine whether the temperatures of the two transformers are 80 degrees and 100 degrees, or 85 degrees and 105 degrees, or 77 degrees and 97 degrees, etc. By using temperature differentials, calculation of exact temperature values (and all of the associated adjustments, compensations, and calibrations that this calculation entails) is obviated. Moreover, by using data differentials - e.g., using raw sensor data before it is converted to temperatures at all - the detection of the relative health of the equipment may beAtty. Docket No.: 8070.0015WOperformed more simply and quickly but with equal reliability as prior art techniques which take great pains to ensure accurate temperatures from sensor data.

[0105] The additional benefit of this method is that it is now possible to take a single infrared image, or "scene" — of multiple transformers in the same facility, for example — and partition it into multiple areas of interest, or "probes", and detect potential failures based on differences of temperature between one set of pixels (one probe) and another set of pixels (another probe) using relatively inexpensive IR cameras, such as those used for surveillance instead of expensive IR cameras used for temperature measurements. No physical contact with the equipment being monitored is necessary — no sensors, no physical probes, no extraction of transformer oil for analysis needed,10106] Finally, because relatively inexpensive, surveillance-type IR cameras may be used, it is possible to use the same IR camera for both temperature monitoring and intrusion detection,

[0107] Sensor Fusion. Sensor Fusion is a term used by Sensei Solutions, LLC (hereinafter referred to as “Sensei Solutions”), to refer to the process of combining signal data from two or more sensors or systems operating in different spectra so as to discriminate between individual sensor-induced noise, and the confirmation of actual occurrence across a heterogeneous population of collocated sensors within a specific period of time. Sensor Fusion is also a term used by Sensei Solutions to refer to a line of products that provide, use, or support this process.

[0108] The importance of Sensor Fusion. Regardless of type or manufacturer, even the most sensitive, exotic and expensive sensors produce false signals (or" noise") under certain circumstances. Better sensors produce fewer false signals and may claim superior SNR (signal-noise ratio), but no sensor is immune to false stimulation under the right circumstances. Thus, any system which relies solely on one class of sensor is inherently incapable of exceeding the reliability of the weakest or noisiest sensor to which it is connected, and will be prone to false detection at a rate defined by the sensor's SNR.

[0109] The process of Sensor Fusion can significantly improve the certainty of valid detections and suppression of noise, by enforcing a logical protocol which requires detection of signals from two or more collocated sensors across two or more spectra within a certain period of time. Coincidental signals from collocated sensors indicate aAtty. Docket No.: 8070.0015WOmuch greater likelihood of valid detection - approximately an order of magnitude greater reliability for each spectra in which the phenomena is detected.

[0110] Sensor Fusion is important as a tool to reduce noise-induced false alarms, and increase reliability in any sensor-based automation system.

[0111] Application to Real-World Problems. Such a dramatic improvement is SNR directly affects the feasibility and economics of any sensor-based automation system, from CBM (condition-based Maintenance) to IDS (Intrusion Detection) and ESI (Electronic Signals Intelligence),

[0112] Two examples of the benefits which may be realized from deployment of Sensor Fusion are Physical Security and Condition-based Monitoring of critical infrastructure such as electrical substation equipment (Power Transformers, Breakers, Bushings etc,), generation assets (Turbines, GSUs) and transmission assets (Towers and transmission lines).

[0113] Sensor Fusion for Physical Security, In the Physical Security realm, Sensor Fusion may be applied to processing of the signal data from conventional IDS (Intrusion Detection Systems) such as PIRs, BLS (Buried-line and fence sensors) Video Analytics, GSR (Ground Surveillance Radar) and GDS (Gunshot Detection Systems). Each of these sensors operate in different spectra - PIRs and Thermography in the infra-red spectrum, BLS in the seismic, video analytics in the visible spectrum, GDS in the audible and radar in the GHz spectrum. Accordingly, properly configured sensors of complementary classes may be deployed to virtually eliminate the propagation of false signals produced by either individual sensor. This strategy has proven extremely effective when combining inexpensive PIR sensors with video analytics, and all the more so with the addition of GSR.

[0114] Sensor Fusion for CBM. The second example is the use of sensor fusion to verify equipment condition of critical assets like high-voltage power transformers. These devices are typically cooled by circulation of dielectric mineral oil around the cellulose- insulated windings inside the main tank of the transformer. In a healthy transformer the mineral oil will contain only minuscule levels of gases in solution. However, if the transformer has suffered any sort of electrical or mechanical insult (overloading, voltage-spike / impulse or physical damage from collision) the internal windings and theirAtty. Docket No.: 8070.0015WOinsulators may be damaged. Such damage typically takes the form of a breakdown in the cellulose insulation, and / or the shorting of some portion of the windings to the transformer tank / ground. In the first case, cellulose which has been stressed by overheating will release carbon dioxide and carbon monoxide. In the second case even the smallest short can result in arcing through the mineral oil - an event referred to as " Partial Discharge". This partial discharge results in a phase-change of the dielectric mineral oil to gas. This phase change produces a number of gases, many of which are highly explosive, even in minute quantities such as hydrogen, methane, ethane and acetylene. Sensors to measure the concentration of these gases in the dielectric solution called DGA (Dissolved Gas Analysis) producing readings based on samples drawn every four hours or so. Much can be determined about the health of the transformer based on the quantity and proportion of these gases; however it is an inexact science due to perfectly normal variations in what is normal from one transformer type and model to the next. Because of these wide variations only broad guidelines exist for safe vs. dangerous concentration of these gases, and except in the most extreme cases the DGA data alone is not sufficiently reliable to automatically make a decision to take a power transformer out of service, with the decision ultimately coming down to the experience of the engineer interpreting the data.

[0115] Sensor Fusion may be applied to this problem by fusing data from other sensors such as partial-discharge acoustic or RF sensors with the DGA, or by comparison across a number of transformers of the same make and model operating under similar conditions (environment and load). In the case of high-voltage transformers the economic consequences of decommissioning a working asset is only secondary to an actual failure of the transformer - neither is desirable or acceptable, and systems which "cry wolf are short-lived.

[0116] Multiple Sensors, Multiple Algorithms, Single View. Sensei Solution, LLC owns and produces a Transformetrics™ application module that provides single¬ dashboard access to all of the instrumentation data collected by Sensei Solutions’ MasterMind™ platform from any number of sensors and lED’s deployed on a transformer, including DGA, Partial Discharge, Load, Temperatures and BushingAtty. Docket No.: 8070.0015WOMonitors. A wide variety of sensors and IEDs are supported including devices from companies such as Kelman / GE, Serveron, Morgan- Schaeffer, Mistras, and Doble.

[0117] Any number or combination of user-provided and industry-standard algorithms may be applied to the sensor data, with the results graphically annotated and superimposed on the relevant data series. Out-of-the-box support for Duval Triangle and Rogers Ratios is provided, and may be augmented by any number of algorithms available from the Institute of Electrical and Electronics Engineers (IEEE) and other professional engineering organizations.

[0118] High Confidence, High Reliability. The only thing worse than an unexpected outage is an unnecessary outage. Because different algorithms respond to different circumstances with varying degrees of accuracy, any one algorithm can indicate fault conditions where none actually exists. Sensei Solutions’ proprietary analytic fusion engine reduces ambiguity and increases confidence by continuously cross-checking analytic results with multiple algorithms and methods which together significantly reduce the likelihood of an unnecessary service interruption, while increasing the likelihood of a valid detection in time to take preventative measures.

[0119] Intuitive Graphical Interface. All of the information pertinent to transformer health can be viewed and compared at-a-glance. Key summary information for each individual data point is presented in either tabular or graphical form, with 5, 15, and 30-day trends automatically calculated and stored for each data element.

[0120] Optimized For Large Data Sets. Sensei Solutions’ MasterMind™ User Interface is designed specifically to handle massive data sets while rendering responsive and informative virtualizations for any number of series or elements. Instantly access years’ worth of sampling data, and effortlessly perform retrospective analysis with eyepopping graphics and state-of-the-art algorithms.

[0121] Scalable, Enterprise-Ready Architecture. No other automation information system provides the level of scalability and flexibility achieved by Sensei Solutions’ MasterMind™ Surveillance Automation Platform. MasterMind™ processors scale from the field to the command center, and everywhere in-between, with software-adapter interfaces for many IEDs and Enterprise-class systems.Atty. Docket No.: 8070.0015WO

[0122] Rugged Package, Flexible Form-Factor. MasterMind™ host processors may be deployed as solid-state embedded devices at the edge of the network collecting and analyzing real-time data on-site, and they may also be deployed on the desktop for historical analysis, or in the data center for access to corporate databases, and in the command center for real-time depiction of site / device status.

[0123] In the detailed description above it can be seen that different features are grouped together in examples. This manner of disclosure should not be understood as an intention that the example clauses have more features than are explicitly mentioned in each clause. Rather, the various aspects of the disclosure may include fewer than all features of an individual example clause disclosed. Therefore, the following clauses should hereby be deemed to be incorporated in the description, wherein each clause by itself can stand as a separate example. Although each dependent clause can refer in the clauses to a specific combination with one of the other clauses, the aspect(s) of that dependent clause are not limited to the specific combination. It will be appreciated that other example clauses can also include a combination of the dependent clause aspect(s) with the subject matter of any other dependent clause or independent clause or a combination of any feature with other dependent and independent clauses. The various aspects disclosed herein expressly include these combinations, unless it is explicitly expressed or can be readily inferred that a specific combination is not intended (e.g., contradictory aspects, such as defining an element as both an insulator and a conductor). Furthermore, it is also intended that aspects of a clause can be included in any other independent clause, even if the clause is not directly dependent on the independent clause.

[0124] Aspect examples are described in the following numbered clauses:

[0125] Clause 1: A method, performed by processing circuitry, for remote monitoring of equipment, the method comprising: receiving a scene of an infrared image of equipment to be monitored; registering the scene; partitioning the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored; assigning each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas; acquiring a plurality of data values, one for each of the fixed, non- overlappingAtty. Docket No.: 8070.0015WOareas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area; defining an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group; detecting that the alarm condition for a first group of the one or more groups has been met; and in response to detecting that the alarm condition for the first group has been met, providing an operator alert that the alarm condition for the first group has been met.

[0126] Clause 2: The method of clause 1, wherein registering the scene comprises registering the scene to a reference scene.

[0127] Clause 3: The method of any of clauses 1 to 2, wherein registering the scene comprises registering the scene using an image registration algorithm.

[0128] Clause 4: The method of any of clauses 1 to 3, wherein using the image registration algorithm comprises using at least one of: an intensity-based image registration method; a feature-based image registration method; a linear transformation method; an elastic transformation method; a spatial method image registration method; a frequency-domain method image registration method; a single-modality image registration method; a multi-modality image registration method; a deep-learning, machine-learning, or artificial intelligence based method; or a combination thereof.

[0129] Clause 5: The method of any of clauses 1 to 4, wherein detecting that the alarm condition for one of the one or more groups has been met comprises at least one of: detecting that a variance or standard deviation of data values from the fixed, non¬ overlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold rate; detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-overlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold rate; or detecting that data values from the fixed, non-overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and theAtty. Docket No.: 8070.0015WOsecond group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

[0130] Clause 6: The method of any of clauses 1 to 5, wherein a value of at least one of the plurality of data values is converted to a temperature value.

[0131] Clause 7: The method of any of clauses 1 to 6 including providing, to the operator, a display that shows an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas.

[0132] Clause 8: The method of any of clauses 1 to 7, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image,

[0133] Clause 9: The method of any of clauses 1 to 8 including adjusting an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

[0134] Clause 10: The method of any of clauses 1 to 9 including mapping a sub-range of data values to a single color or intensity value.

[0135] Clause 11: The method of any of clauses 1 to 10 wherein data values in a first sub-range are displayed in color and wherein data values in a second sub-range are displayed in grayscale or black and white.

[0136] Clause 12: The method of any of clauses 1 to 11 wherein acquiring the plurality of data values further comprises acquiring at least one additional data value from a sensor that is attached to the equipment to be monitored and assigning the at least one additional data value to at least one of the one or more groups.

[0137] Clause 13: The method of any of clauses 1 to 12 wherein the sensor transmits the data value via a wired communications link, via a wireless communications links, or a combination thereof.

[0138] Clause 14: An apparatus, comprising: a memory; and at least one processor communicatively coupled to the memory, the at least one processor configured to: receive a scene of an infrared image of equipment to be monitored; register the scene; partition the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that displays a fixed portion of the equipment to be monitored; assign each of the plurality of areas to one or more groups, each group containing at least two ofAtty. Docket No.: 8070.0015WOthe plurality of areas; acquire a plurality of data values, one for each of the fixed, non¬ overlapping areas, derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area; define an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group; detect that the alarm condition for one of the one or more groups has been met; and in response to detecting that the alarm condition for one of the one or more groups has been met, provide an operator alert that the alarm condition for the one of the one or more groups has been met.

[0139] Clause 15: The apparatus of clause 14, wherein the at least one processor is configured to register the scene to a reference scene.

[0140] Clause 16: The apparatus of any of clauses 14 to 15, wherein the at least one processor is configured to register the scene using an image registration algorithm.

[0141] Clause 17: The apparatus of any of clauses 14 to 16, wherein the image registration algorithm comprises at least one of: an intensity-based image registration method; a feature-based image registration method; a linear transformation method; an elastic transformation method; a spatial method image registration method; a frequency¬ domain method image registration method; a single-modality image registration method; a multi-modality image registration method; a deep-learning, machine- learning, or artificial intelligence based method; or a combination thereof.

[0142] Clause 18: The apparatus of any of clauses 14 to 17, wherein the alarm condition for each of the one or more groups comprises at least one of: the condition that a variance or standard deviation of data values from areas in the group exceeds a threshold amount or has a change of rate that exceeds a threshold rate; the condition that a variance or standard deviation of data values from areas in the group differs from a variance or standard deviation of data values from areas in another group by a threshold amount or has a change of rate that exceeds a threshold rate; or the condition that data values from areas in a first group differ from data values from areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein theAtty. Docket No.: 8070.0015WOfirst equipment and second equipment are subject to a same set of environmental conditions.

[0143] Clause 19: The apparatus of any of clauses 14 to 18, wherein the at least one processor is configured to convert at least one of the plurality of data values to a temperature value.

[0144] Clause 20: The apparatus of any of clauses 14 to 19, further comprising a display that shows an image of the scene and a visual indication of the plurality’ of fixed, non-overlapping areas.

[0145] Clause 21: The apparatus of any of clauses 14 to 20, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image.

[0146] Clause 22: The apparatus of any of clauses 14 to 21, wherein the at least one processor is further configured to adjust an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

[0147] Clause 23: The apparatus of any of clauses 14 to 22, wherein the at least one processor is further configured to map a sub-range of data values to a single color or intensity value.

[0148] Clause 24: The apparatus of any of clauses 14 to 23, wherein the at least one processor is further configured to display data values in a first sub-range in color and to display data values in a second sub-range in grayscale or black and white.

[0149] Clause 25: The apparatus of any' of clauses 14 to 24, wherein the at least one processor is further configured to acquire at least one additional data value from a sensor that is attached to the equipment to be monitored and assign the at least one additional data value to at least one of the one or more groups.

[0150] Clause 26: The apparatus of any of clauses 14 to 25 wherein the sensor transmits the data value via a wired communications link, via a wireless communications links, or a combination thereof.

[0151] Clause 27: A device, comprising: means for receiving a scene of an infrared image of equipment to be monitored; means for registering the scene; means for partitioning the scene into a plurality of fixed, non-overlapping areas, each areaAtty. Docket No.: 8070.0015WOcomprising a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored; means for assigning each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas; means for acquiring a plurality of data values, one for each of the fixed, non-overlapping areas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area; means for defining an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group; means for detecting that the alarm condition for a first group of the one or more groups has been met; and means for, in response to detecting that the alarm condition for the first group has been met, providing an operator alert that the alarm condition for the first group has been met,

[0152] Clause 28: The device of clause 27, wherein means for registering the scene comprises means for registering the scene to a reference scene.

[0153] Clause 29: The device of any of clauses 27 to 28, wherein means for registering the scene comprises means for registering the scene using an image registration algorithm.

[0154] Clause 30: The device of any of clauses 27 to 29, wherein means for using the image registration algorithm comprises means for using at least one of: an intensity- based image registration method a feature- based image registration method a linear transformation method an elastic transformation method a spatial method image registration method a frequency-domain method image registration method a singlemodality image registration method a multi-modality image registration method a deeplearning, machine-learning, or artificial intelligence based method a deep- learning, machine- learning, or artificial intelligence based method a combination thereof.

[0155] Clause 31: The device of any of clauses 27 to 30, wherein means for detecting that the alarm condition for one of the one or more groups has been met comprises at least one of: means for detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold rate means for detecting that a variance orAtty. Docket No.: 8070.0015WOstandard deviation of data values from the fixed, non-overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, nonoverlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold rate means for detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-over lapping areas m another group by a threshold amount or has a rate of change over time that exceeds a threshold rate means for detecting that data values from the fixed, non-overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

[0156] Clause 32: The device of any of clauses 27 to 31, further comprising means for converting at least one of the plurality of data values is to a temperature value.

[0157] Clause 33: The device of any of clauses 27 to 32, including means for providing, to the operator, a display that shows an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas.

[0158] Clause 34: The device of any of clauses 27 to 33, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image.

[0159] Clause 35: The device of any of clauses 27 to 34, including means for adjusting an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

[0160] Clause 36: The device of any of clauses 27 to 35, including means for mapping a sub-range of data values to a single color or intensity value.

[0161] Clause 37: The device of any of clauses 27 to 36, wherein data values in a first sub-range are displayed in color and wherein data values in a second sub-range are displayed in grayscale or black and white.Atty. Docket No.: 8070.0015WO

[0162] Clause 38: The device of any of clauses 27 to 37, wherein means for acquiring the plurality of data values further comprises means for acquiring at least one additional data value from a sensor that is attached to the equipment to be monitored and means for assigning the at least one additional data value to at least one of the one or more groups.

[0163] Clause 39: The device of any of clauses 27 to 38, wherein the sensor transmits the data value via a wired communications link, via a wireless communications links, or a combination thereof

[0164] Clause 40: A non-transitory computer-readable medium storing computerexecutable instructions that, when executed by a device, cause the device to: receive a scene of an infrared image of equipment to be monitored; register the scene; partition the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored; assign each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas; acquire a plurality of data values, one for each of the fixed, nonoverlapping areas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area; define an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group; detect that the alarm condition for a first group of the one or more groups has been met; and in response to detecting that the alarm condition for the first group has been met, provide an operator alert that the alarm condition for the first group has been met.

[0165] Clause 41: The non-transitory computer-readable medium of clause 40, wherein the computer-executable instructions that, when executed by the device, cause the device to register the scene comprise computer-executable instructions that, when executed by the device, cause the device to register the scene to a reference scene.

[0166] Clause 42: The non-transitory computer-readable medium of any of clauses 40 to 41, wherein the computer-executable instructions that, when executed by the device, cause the device to register the scene comprise computer-executable instructions that, when executed by the device, cause the device to register the scene using an image registration algorithm.Atty. Docket No.: 8070.0015WO

[0167] Clause 43: The non-transitory computer-readable medium of any of clauses 40 to 42, wherein the computer-executable instructions that, when executed by the device, cause the device to using the image registration algorithm comprise computer-executable instructions that, when executed by the device, cause the device to use at least one of: an intensity-based image registration method; a feature-based image registration method; a linear transformation method; an elastic transformation method; a spatial method image registration method; a frequency-domain method image registration method; a single¬ modality image registration method; a multi-modality image registration method; a deeplearning, machine-learn, or artificial intelligence based method; or a combination thereof.

[0168] Clause 44: The non-transitory computer-readable medium of any of clauses 40 to 43, wherein the computer-executable instructions that, when executed by the device, cause the device to detecting that the alarm condition for one of the one or more groups has been met comprise computer-executable instructions that, when executed by the device, cause the device to at least one of: detect that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold rate; detect that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non¬ overlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold rate; or detect that data values from the fixed, non¬ overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

[0169] Clause 45: The non-transitory computer-readable medium of any of clauses 40 to 44, further comprising computer-executable instructions that, when executed by the device, cause the device to convert at least one of the plurality of data values to a temperature value.

[0170] Clause 46: The non-transitory computer-readable medium of any of clauses 40 to 45, further comprising computer-executable instructions that, when executed by theAtty. Docket No.: 8070.0015WOdevice, cause the device to provide, to the operator, a display that shows an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas.

[0171] Clause 47: The non-transitory computer-readable medium of any of clauses 40 to 46, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image.

[0172] Clause 48: The non-transitory computer-readable medium of any of clauses 40 to 47, further comprising computer-executable instructions that, when executed by the device, cause the device to adjust an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

[0173] Clause 49: The non-transitory computer-readable medium of any of clauses 40 to 48, further comprising computer-executable instructions that, when executed by the device, cause the device to map a sub-range of data values to a single color or intensity value.

[0174] Clause 50: The non-transitory computer-readable medium of any of clauses 40 to 49, wherein data values in a first sub-range are displayed in color and data values in a second sub-range are displayed in grayscale or black and white.

[0175] Clause 51: The non-transitory computer-readable medium of any of clauses 40 to 50, wherein the computer-executable instructions that, when executed by the device, cause the device to acquire the plurality of data values comprise computer-executable instructions that, when executed by the device, cause the device to acquire at least one additional data value from a sensor that is attached to the equipment to be monitored and assigning the at least one additional data value to at least one of the one or more groups.

[0176] Clause 52: The non-transitory computer-readable medium of any of clauses 40 to 51, wherein the computer-executable instructions that, when executed by the device, cause the device to acquire at least one additional data value from a sensor that is attached to the equipment to be monitored comprise computer-executable instructions that, when executed by the device, cause the device to receive the at least one additional data value via a wired communications link, via a wireless communications links, or a combination thereof.

Claims

Atty. Docket No.: 8070.0015WOCLAIMSWhat is claimed is:

1. A method, performed by processing circuitry, for remote monitoring of equipment, the method comprising:receiving a scene of an infrared image of equipment to be monitored; registering the scene;partitioning the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored;assigning each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas;acquiring a plurality of data values, one for each of the fixed, non-overlapping areas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area;defining an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, nonoverlapping areas assigned to that group;detecting that the alarm condition for a first group of the one or more groups has been met; andin response to detecting that the alarm condition for the first group has been met, providing an operator alert that the alarm condition for the first group has been met.

2. The method of claim 1, wherein registering the scene comprises registering the scene to a reference scene.

3. The method of claim 1, wherein registering the scene comprises registering the scene using an image registration algorithm.Atty. Docket No.: 8070.0015WO4. The method of claim 3, wherein using the image registration algorithm comprises using at least one of:an intensity-based image registration method;a feature-based image registration method;a linear transformation method;an elastic transformation method;a spatial method image registration method;a frequency- domain method image registration method;a single-modality image registration method;a multi-modality image registration method;a deep-learning, machine-learning, or artificial intelligence based method; or a combination thereof.

5. The method of claim 1, wherein detecting that the alarm condition for one of the one or more groups has been met comprises at least one of:detecting that a variance or standard deviation of data values from the fixed, non¬ overlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold rate;detecting that a variance or standard deviation of data values from the fixed, nonoverlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-overlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold rate; ordetecting that data values from the fixed, non-overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

6. The method of claim 1, wherein a value of at least one of the plurality of data values is converted to a temperature value.Atty. Docket No.: 8070.0015WO7. The method of claim 1 including providing, to the operator, a display that shows an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas.

8. The method of claim 7, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image.

9. The method of claim 8 including adjusting an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

10. The method of claim 8 including mapping a sub-range of data values to a single color or intensity value.

11. The method of claim 8 wherein data values in a first sub-range are displayed in color and wherein data values in a second sub-range are displayed in grayscale or black and white.

12. The method of claim 1 wherein acquiring the plurality of data values further comprises acquiring at least one additional data value from a sensor that is attached to the equipment to be monitored and assigning the at least one additional data value to at least one of the one or more groups.

13. The method of claim 12 wherein the sensor transmits the data value via a wired communications link, via a wireless communications links, or a combination thereof.

14. An apparatus, comprising:a memory; andAtty. Docket No.: 8070.0015WOat least one processor communicatively coupled to the memory, the at least one processor configured to:receive a scene of an infrared image of equipment to be monitored; register the scene;partition the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that displays a fixed portion of the equipment to be monitored;assign each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas;acquire a plurality of data values, one for each of the fixed, nonoverlapping areas, derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area;define an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group;detect that the alarm condition for one of the one or more groups has been met; andin response to detecting that the alarm condition for one of the one or more groups has been met, provide an operator alert that the alarm condition for the one of the one or more groups has been met.

15. The apparatus of claim 14, wherein the at least one processor is configured to register the scene to a reference scene.

16. The apparatus of claim 14, wherein the at least one processor is configured to register the scene using an image registration algorithm.

17. The apparatus of claim 16, wherein the image registration algorithm comprises at least one of:an intensity-based image registration method;Atty. Docket No.: 8070.0015WOa feature-based image registration method;a linear transformation method;an elastic transformation method;a spatial method image registration method;a frequency- domain method image registration method;a single-modality image registration method;a multi-modality image registration method;a deep-learning, machine-learning, or artificial intelligence based method; or a combination thereof.

18. The apparatus of claim 14, wherein the alarm condition for each of the one or more groups comprises at least one of:the condition that a variance or standard deviation of data values from areas in the group exceeds a threshold amount or has a change of rate that exceeds a threshold rate;the condition that a variance or standard deviation of data values from areas in the group differs from a variance or standard deviation of data values from areas in another group by a threshold amount or has a change of rate that exceeds a threshold rate; or the condition that data values from areas in a first group differ from data values from areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

19. The apparatus of claim 14, wherein the at least one processor is configured to convert at least one of the plurality of data values to a temperature value.

20. The apparatus of claim 14, further comprising a display that shows an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas.Atty. Docket No.: 8070.0015WO21. The apparatus of claim 20, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image.

22. The apparatus of claim 21, wherein the at least one processor is further configured to adjust an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

23. The apparatus of claim 21, wherein the at least one processor is further configured to map a sub-range of data values to a single color or intensity value.

24. The apparatus of claim 21, wherein the at least one processor is further configured to display data values in a first sub-range in color and to display data values in a second sub-range in grayscale or black and white.

25. The apparatus of claim 14, wherein the at least one processor is further configured to acquire at least one additional data value from a sensor that is attached to the equipment to be monitored and assign the at least one additional data value to at least one of the one or more groups.

26. The apparatus of claim 25 wherein the sensor transmits the data value via a wired communications link, via a wireless communications links, or a combination thereof.

27. A device, comprising:means for receiving a scene of an infrared image of equipment to be monitored; means for registering the scene;means for partitioning the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored;Atty. Docket No.: 8070.0015WOmeans for assigning each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas;means for acquiring a plurality of data values, one for each of the fixed, non¬ overlapping areas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area;means for defining an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, non-overlapping areas assigned to that group;means for detecting that the alarm condition for a first group of the one or more groups has been met; andmeans for, in response to detecting that the alarm condition for the first group has been met, providing an operator alert that the alarm condition for the first group has been met.

28. The device of claim 27, wherein means for registering the scene comprises means for registering the scene to a reference scene.

29. The device of claim 27, wherein means for registering the scene comprises means for registering the scene using an image registration algorithm.

30. The device of claim 29, wherein means for using the image registration algorithm comprises means for using at least one of:an intensity-based image registration methoda feature- based image registration methoda linear transformation methodan elastic transformation methoda spatial method image registration methoda frequency- domain method image registration methoda single-modality image registration methoda multi-modality image registration methodAtty. Docket No.: 8070.0015WOa deep-learning, machine- learning, or artificial intelligence based method a deep-learning, machine- learning, or artificial intelligence based method a combination thereof.

31. The device of claim 27, wherein means for detecting that the alarm condition for one of the one or more groups has been met comprises at least one of:means for detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold ratemeans for detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-overlapping areas m another group by a threshold amount or has a rate of change over time that exceeds a threshold ratemeans for detecting that a variance or standard deviation of data values from the fixed, non-overlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-overlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold ratemeans for detecting that data values from the fixed, non-overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

32. The device of claim 27, further comprising means for converting at least one of the plurality of data values is to a temperature value.

33. The device of claim 27, including means for providing, to the operator, a display that shows an image of the scene and a visual indication of the plurality of fixed, nonoverlapping areas.Atty. Docket No.: 8070.0015WO34. The device of claim 33, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image.

35. The device of claim 34, including means for adjusting an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

36. The device of claim 34, including means for mapping a sub-range of data values to a single color or intensity value.

37. The device of claim 34, wherein data values in a first sub-range are displayed in color and wherein data values in a second sub-range are displayed in grayscale or black and white.

38. The device of claim 27, wherein means for acquiring the plurality of data values further comprises means for acquiring at least one additional data value from a sensor that is attached to the equipment to be monitored and means for assigning the at least one additional data value to at least one of the one or more groups.

39. The device of claim 38, wherein the sensor transmits the data value via a wired communications link, via a wireless communications links, or a combination thereof.

40. A non-transitory computer-readable medium storing computer-executable instructions that, when executed by a device, cause the device to:receive a scene of an infrared image of equipment to be monitored;register the scene;partition the scene into a plurality of fixed, non-overlapping areas, each area comprising a fixed set of pixels that represents a shape of a fixed portion of the equipment to be monitored;Atty. Docket No.: 8070.0015WOassign each of the plurality of areas to one or more groups, each group containing at least two of the plurality of areas;acquire a plurality of data values, one for each of the fixed, non-overlapping areas, each data value being derived from intensity values of all of the pixels contained within the respective fixed, non-overlapping area;define an alarm condition for each of the one or more groups, wherein each alarm condition defines a condition indicative of potential equipment failure of which an operator should be alerted, based on an analysis of the data values of the fixed, nonoverlapping areas assigned to that group;detect that the alarm condition for a first group of the one or more groups has been met; andin response to detecting that the alarm condition for the first group has been met, provide an operator alert that the alarm condition for the first group has been met,41. The non-transitory computer-readable medium of claim 40, wherein the computer-executable instructions that, when executed by the device, cause the device to register the scene comprise computer-executable instructions that, when executed by the device, cause the device to register the scene to a reference scene.

42. The non-transitory computer-readable medium of claim 40, wherein the computer-executable instructions that, when executed by the device, cause the device to register the scene comprise computer-executable instructions that, when executed by the device, cause the device to register the scene using an image registration algorithm.

43. The non-transitory computer-readable medium of claim 42, wherein the computer-executable instructions that, when executed by the device, cause the device to using the image registration algorithm comprise computer-executable instructions that, when executed by the device, cause the device to use at least one of:an intensity-based image registration method;a feature- based image registration method;a linear transformation method;Atty. Docket No.: 8070.0015WOan elastic transformation method;a spatial method image registration method;a frequency- domain method image registration method;a single-modality image registration method;a multi-modality image registration method;a deep-learning, machine-learning, or artificial intelligence based method; or a combination thereof.

44. The non-transitory computer-readable medium of claim 40, wherein the computer-executable instructions that, when executed by the device, cause the device to detecting that the alarm condition for one of the one or more groups has been met comprise computer-executable instructions that, when executed by the device, cause the device to at least one of:detect that a variance or standard deviation of data values from the fixed, nonoverlapping areas in the group exceeds a threshold amount or has a rate of change over time that exceeds a threshold rate;detect that a variance or standard deviation of data values from the fixed, nonoverlapping areas in the group differs from a variance or standard deviation of data values from the fixed, non-overlapping areas in another group by a threshold amount or has a rate of change over time that exceeds a threshold rate; ordetect that data values from the fixed, non-overlapping areas in a first group differ from data values from the fixed, non-overlapping areas in a second group by a threshold amount, wherein the first group corresponds to portions of a first equipment and the second group corresponds to portions of a second equipment, and wherein the first equipment and second equipment are subject to a same set of environmental conditions.

45. The non-transitory computer-readable medium of claim 40, further comprising computer-executable instructions that, when executed by the device, cause the device to convert at least one of the plurality of data values to a temperature value.Atty. Docket No.: 8070.0015WO46. The non-transitory computer-readable medium of claim 40, further comprising computer-executable instructions that, when executed by the device, cause the device to provide, to the operator, a display that shows an image of the scene and a visual indication of the plurality of fixed, non-overlapping areas.

47. The non-transitory computer-readable medium of claim 46, wherein a hue or intensity of a pixel of the image represents a static data value or a static range of values corresponding to a temperature or a range of temperatures of the equipment at that location in the image,48. The non-transitory computer-readable medium of claim 47, further comprising computer-executable instructions that, when executed by the device, cause the device to adjust an intensity map or color map of the image to increase a range or sensitivity of the data values being represented to the operator.

49. The non-transitory computer-readable medium of claim 47, further comprising computer-executable instructions that, when executed by the device, cause the device to map a sub-range of data values to a single color or intensity value.

50. The non-transitory computer-readable medium of claim 47, wherein data values in a first sub-range are displayed in color and data values in a second sub-range are displayed in grayscale or black and white.

51. The non-transitory computer-readable medium of claim 40, wherein the computer-executable instructions that, when executed by the device, cause the device to acquire the plurality of data values comprise computer-executable instructions that, when executed by the device, cause the device to acquire at least one additional data value from a sensor that is attached to the equipment to be monitored and assigning the at least one additional data value to at least one of the one or more groups.Atty. Docket No.: 8070.0015WO52. The non-transitory computer-readable medium of claim 51, wherein the computer-executable instructions that, when executed by the device, cause the device to acquire at least one additional data value from a sensor that is attached to the equipment to be monitored comprise computer-executable instructions that, when executed by the device, cause the device to receive the at least one additional data value via a wired communications link, via a wireless communications links, or a combination thereof.

Images