Program, diagnostic method, and diagnostic device

Abstract

In a program, a diagnostic method, and a diagnostic device, a computer executes: a first process in which a plurality of regions are set in a temperature image showing a temperature distribution of the device; a second process of determining a change in a first correlation relationship, which is a correlation relationship of first feature quantities related to temperature between regions different from each other; and a third process for diagnosing an abnormality of the device on the basis of a change in the first correlation.

Description

Technical Field

[0001] This disclosure relates to procedures, diagnostic methods, and diagnostic devices. Background Technology

[0002] Anomaly diagnosis is widely used to detect malfunctions in various devices. One type of anomaly diagnosis involves determining the presence or absence of anomalies based on the device's temperature. For example, as disclosed in Patent Document 1, a technique is proposed that uses an infrared camera to capture a temperature image showing the temperature distribution of the device and performs anomaly diagnosis based on the temperature image.

[0003] Existing technical documents Patent documents Patent document 1: Japanese Patent Application Publication No. 2021-181286. Summary of the Invention

[0004] The problem the invention aims to solve As mentioned above, several techniques related to anomaly diagnosis focusing on device temperature have been proposed, but new proposals regarding techniques for diagnosing device anomalies with higher accuracy are expected.

[0005] In view of the above-mentioned problems, the present disclosure aims to provide a procedure, diagnostic method and diagnostic device capable of diagnosing device malfunctions with high accuracy.

[0006] Solution for solving the problem To address the aforementioned problems, one approach of this disclosure involves a computer performing: a first process of defining multiple regions in a temperature image showing the temperature distribution of the device; a second process of determining changes in a first correlation, wherein the first correlation is a correlation between temperature-related first characteristic quantities between different regions; and a third process of performing anomaly diagnosis of the device based on the changes in the first correlation.

[0007] The first feature quantity can include multiple feature quantities. In the second process, in addition to the change of the first correlation, the change of the second correlation is further determined. The second correlation is the correlation between different types of first feature quantities in the same region, or the correlation between different types of first feature quantities corresponding to different regions. In the third process, in addition to the change of the first correlation, anomaly diagnosis is also performed based on the change of the second correlation.

[0008] In the second processing, in addition to the change in the first correlation, the change in the third correlation can be further determined. The third correlation is the correlation between the first feature and the second feature, which is independent of the region. In the third processing, in addition to the change in the first correlation, anomaly diagnosis is also performed based on the change in the third correlation.

[0009] The second characteristic quantity may include characteristic quantities related to the operating state of the device.

[0010] The second characteristic may include characteristic quantities related to the thermal input and output of the device.

[0011] In the first processing, multiple regions can be set in the temperature image based on the visible light image of the device.

[0012] To address the aforementioned problems, in one aspect of the diagnostic method disclosed herein, a computer performs: a first process of defining multiple regions in a temperature image showing the temperature distribution of the device; a second process of determining changes in a first correlation, wherein the first correlation is a correlation between temperature-related first feature quantities between different regions; and a third process of performing anomaly diagnosis of the device based on the changes in the first correlation.

[0013] To address the aforementioned problems, one aspect of the diagnostic device disclosed herein includes: an infrared camera that captures a temperature image showing the temperature distribution of the device; and a processing unit that performs: a first processing step of defining multiple regions in the temperature image; a second processing step of determining a change in a first correlation, wherein the first correlation is a correlation between temperature-related first feature quantities between different regions; and a third processing step of performing anomaly diagnosis of the device based on the change in the first correlation.

[0014] Invention Effects According to the procedures, diagnostic methods, and diagnostic apparatus disclosed herein, device malfunctions can be diagnosed with high precision. Attached Figure Description

[0015] Figure 1 This is a schematic diagram illustrating the general structure of a diagnostic device according to an embodiment of the present disclosure; Figure 2 This is a block diagram illustrating an example of the functional structure of a processing apparatus according to an embodiment of the present disclosure; Figure 3 This is a diagram illustrating an example of a visible light image of an embodiment of the present disclosure; Figure 4 This is a diagram illustrating an example of a temperature image of an embodiment of the present disclosure; Figure 5 This is a flowchart illustrating an example of the processing flow performed by the processing apparatus according to an embodiment of the present disclosure; Figure 6 This is a diagram illustrating an example of a benchmark dataset for an embodiment of this disclosure; Figure 7 This is a diagram illustrating an example of a diagnostic object dataset according to an embodiment of this disclosure; Figure 8 This is a diagram illustrating an example of a benchmark dataset for a first variation of this disclosure; Figure 9 This is a diagram illustrating an example of a benchmark dataset for a second variation of this disclosure. Detailed Implementation

[0016] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings. The dimensions, materials, and other specific values ​​shown in the embodiments are merely illustrative for ease of understanding and do not limit the scope of this disclosure, except where specifically requested otherwise. Furthermore, in this specification and the accompanying drawings, elements having substantially the same function or structure are omitted from repeated description by using the same reference numerals. Additionally, elements not directly related to this invention are omitted from the illustrations.

[0017] Figure 1 This is a schematic diagram showing the general structure of the diagnostic device 1 according to an embodiment of the present disclosure. The diagnostic device 1 performs malfunction diagnosis of the device. Figure 1 In the example, diagnostic device 1 performs an abnormality diagnosis on engine 2, which is equivalent to one of the devices. However, the device that is the subject of diagnosis by diagnostic device 1 is not limited to engine 2.

[0018] like Figure 1 As shown, the diagnostic device 1 includes an infrared camera 11, a visible light camera 12, and a processing device 13.

[0019] Infrared camera 11 captures a temperature image. Specifically, infrared camera 11 has multiple imaging elements that sense far-infrared radiation emitted by an object. By sensing far-infrared radiation through these imaging elements, a temperature image is captured. In the temperature image, the distribution of far-infrared radiation intensity is shown as a temperature distribution. In this embodiment, infrared camera 11 captures a temperature image IM1 showing the temperature distribution of engine 2. The obtained temperature image IM1 is output to processing device 13.

[0020] Visible light camera 12 captures visible light images. Specifically, visible light camera 12 has multiple imaging elements that sense visible light. By sensing visible light through these imaging elements, a visible light image is captured. In this embodiment, visible light camera 12 captures a visible light image IM2 that reflects the engine 2. The obtained visible light image IM2 is output to processing device 13.

[0021] The processing unit 13 performs various processes related to the diagnosis of abnormalities in the engine 2. The processing unit 13 includes, for example, a central processing unit (CPU), a ROM storing programs, and RAM serving as a working area.

[0022] Figure 2This is a block diagram illustrating an example of the functional structure of the processing apparatus 13 according to an embodiment of the present disclosure. Figure 2 As shown, the processing device 13 includes, for example, an acquisition unit 13a, a setting unit 13b, a determination unit 13c, a diagnostic unit 13d, and a storage unit 13e. The functions of the processing device 13 described below are implemented by a program executed by a central processing unit within the processing device 13. The functions of the processing device 13 described below can be implemented by a single device or shared by multiple devices.

[0023] The acquisition unit 13a acquires various types of information. For example, the acquisition unit 13a acquires information from the infrared camera 11, the visible light camera 12, and the storage unit 13e.

[0024] The setting unit 13b sets multiple regions in the temperature image IM1. As will be described later, during the anomaly diagnosis of the engine 2, feature quantities are calculated for each region in the temperature image IM1. Examples of feature quantities include, for instance, the average temperature. Details regarding these feature quantities will be described later.

[0025] Determination unit 13c determines the changes in the correlation between the feature quantities. For example, the correlation between feature quantities in different regions can be cited as an example. Details regarding the correlation between feature quantities will be described later. Determination unit 13c determines the changes in this correlation based on the correlation between the feature quantities during normal operation of engine 2.

[0026] The diagnostic unit 13d performs anomaly diagnosis on engine 2 based on the determination results of the changes in the correlation between the characteristic quantities by the determination unit 13c. For example, if the diagnostic unit 13d diagnoses an engine 2 anomaly when the change in the correlation between the characteristic quantities of engine 2 under normal conditions is large, based on the determination results of the changes in the correlation between the characteristic quantities of engine 2 under normal conditions is large. Details regarding anomaly diagnosis will be described later.

[0027] Storage unit 13e stores various types of information. The information stored in storage unit 13e is used for various processes performed by processing device 13.

[0028] Figure 3 This is a diagram illustrating an example of a visible light image IM2 according to an embodiment of the present disclosure. As described above, engine 2 is projected into the visible light image IM2. Although in Figure 3 Illustrations are omitted, but in the visible light image IM2, brightness is shown, for example, in shades. Figure 3 In the example, cylinders 21, 22, 23, and 24, pipes 25, 26, 27, and 28, and a turbocharger 29 are shown as components of engine 2. However, the structure of engine 2 is not limited to these components. Figure 3 Examples.

[0029] Figure 4 This is a diagram illustrating an example of a temperature image IM1 according to an embodiment of the present disclosure. As described above, temperature image IM1 shows the temperature distribution of engine 2. Although in Figure 4 The illustration is omitted, but in the temperature image IM1, the temperature is shown, for example, in shades. Figure 4 In the example, the temperature image IM1 is divided into regions R1, R2, R3, R4, R5, R6, R7, R8, and R9, corresponding to cylinders 21, 22, 23, 24, pipes 25, 26, 27, 28, and the booster 29, respectively. Hereinafter, without specifically distinguishing between regions R1 to R9, they will be referred to simply as region R. Details regarding the processing related to the setting of region R will be described later.

[0030] The following is for reference Figure 3 and Figure 4 The processing performed by the processing device 13 will be described using an example. However, the visible light image IM2 and the temperature image IM1 are not limited to... Figure 3 and Figure 4 For example, the parts reflected in different images in Engine 2 may differ from those in... Figure 3 and Figure 4 The examples show the parts that appear in each image. For example, the parts that can be omitted in... Figure 4 In the example, a portion of the regions R1 to R9 are defined, and other regions R can also be added.

[0031] Figure 5 This is a flowchart illustrating an example of the processing flow performed by the processing apparatus 13 according to an embodiment of the present disclosure. Figure 5 The processing flow shown is executed, for example, when a user performs a specified operation on the processing device 13.

[0032] when Figure 5 At the start of the processing flow shown, in step S101, the acquisition unit 13a acquires a temperature image IM1 showing the temperature distribution of the engine 2 from the infrared camera 11. After step S101, in step S102, the acquisition unit 13a acquires a visible light image IM2 showing the engine 2 from the visible light camera 12.

[0033] Following step S102, in step S103, the setting unit 13b sets multiple regions R in the temperature image IM1. Specifically, the setting unit 13b sets multiple regions R in the temperature image IM1 based on the visible light image IM2.

[0034] The positions of the infrared camera 11 and the visible light camera 12 are different. Therefore, the display positions of the same part in the engine 2 are different between the temperature image IM1 and the visible light image IM2. The setting unit 13b can estimate the positional relationship of the display positions of the same part in the engine 2 between the temperature image IM1 and the visible light image IM2 based on the positional relationship between the infrared camera 11 and the visible light camera 12. For example, information showing the positional relationship between the infrared camera 11 and the visible light camera 12 is pre-stored in the storage unit 13e.

[0035] The setting unit 13b can determine, for example, which part of the engine 2 exists in which display position in the visible light camera 12 by performing image processing such as edge detection on the visible light camera 12. Therefore, the setting unit 13b can determine which part of the engine 2 exists in which display position in the temperature image IM1 based on the determination result of which part of the engine 2 exists in which display position in the visible light camera 12, and the estimation result of the positional relationship of the display positions of the same part of the engine 2 between the temperature image IM1 and the visible light image IM2.

[0036] The setting unit 13b can set the region R in which each part of the engine 2 exists in the temperature image IM1 based on the determination result of which part of the engine 2 exists in which display position in the temperature image IM1. Figure 4 In the example, the setting unit 13b sets the regions R in which cylinders 21, 22, 23, 24, pipes 25, 26, 27, 28, and booster 29 exist as region R1, region R2, region R3, region R4, region R5, region R6, region R7, region R8, and region R9, respectively, in the temperature image IM1.

[0037] However, the setting unit 13b can also set multiple regions R in the temperature image IM1 without using the visible light image IM2. For example, if information showing which part of the engine 2 exists in which display position in the temperature image IM1 is pre-stored in the storage unit 13e, the setting unit 13b can use this information to set multiple regions R in the temperature image IM1 without using the visible light image IM2. Furthermore, the setting unit 13b can also obtain information showing which part of the engine 2 exists in which display position in the temperature image IM1 based on the detection results of a sensor (e.g., LIDAR) capable of detecting the position information of each part of the engine 2.

[0038] exist Figure 5Following step S103, in step S104, the determination unit 13c obtains the benchmark dataset. The benchmark dataset is a dataset that summarizes the calculation results of the feature quantities of each region R of the engine 2 under normal conditions. The benchmark dataset is pre-stored in the storage unit 13e.

[0039] Figure 6 This is a diagram illustrating an example of a benchmark dataset D1 used in embodiments of this disclosure. Figure 6 The benchmark dataset D1 summarizes the calculation results of the first feature quantity F11 for each region R at each time T. The first feature quantity F11 is a temperature-related feature quantity. Specifically, the first feature quantity F11 is the average temperature in each region R. The determination unit 13c calculates the first feature quantity F11 for each region R at each time T based on the temperature image IM1 obtained at each time T.

[0040] The determination unit 13c calculates the first characteristic quantity F11 of each region R at time T1. Next, the determination unit 13c calculates the first characteristic quantity F11 of each region R at time T2. Then, the determination unit 13c calculates the first characteristic quantity F11 of each region R at time T3. Thus, the determination unit 13c calculates the first characteristic quantity F11 of each region R for each of the n times T from time T1 to Tn. Furthermore, in Figure 6 The diagram showing the actual calculated value of the first characteristic quantity F11 is omitted. However, at the same time T, the value of the first characteristic quantity F11 can differ for each region R.

[0041] exist Figure 5 Following step S104, in step S105, the determination unit 13c generates a diagnostic object dataset. The diagnostic object dataset is obtained by adding the calculation results of the first feature quantity F11 of each region R at the current time to the baseline dataset D1.

[0042] Figure 7 This is a diagram illustrating an example of a diagnostic object dataset D2 according to an embodiment of this disclosure. Figure 7 In the diagnostic dataset D2, for Figure 6 The determination unit 13c calculates the first feature quantity F11 of each region R at the current time Tk based on the baseline dataset D1 and the calculation results of the first feature quantity F11 of each region R at the current time Tk.

[0043] exist Figure 5Following step S105, in step S106, the determining unit 13c determines the change in the correlation between the feature quantities. Specifically, the determining unit 13c determines the change in this correlation based on the correlation between the feature quantities when the engine 2 is functioning normally. More specifically, the determining unit 13c determines the change in the aforementioned correlation based on the benchmark dataset D1 and the diagnostic object dataset D2.

[0044] exist Figure 6 and Figure 7 In the example, the determination part 13c determines the correlation of the first feature quantity F11 between different regions R, that is, the change of the first correlation.

[0045] First, determine the probability density function of the first feature quantity F11 in each region R of the benchmark dataset D1 in part 13c.

[0046] For example, the determination unit 13c uses a data set G1 consisting of n first feature quantities F11 in region R1 to calculate the probability density function of the first feature quantity F11 in region R1. In this case, the probability density function of the first feature quantity F11 in region R1 shows the probability density of each value of the first feature quantity F11 in data set G1. Similarly, the determination unit 13c uses a data set G2 consisting of n first feature quantities F11 in region R2 to calculate the probability density function of the first feature quantity F11 in region R2. In this case, the probability density function of the first feature quantity F11 in region R2 shows the probability density of each value of the first feature quantity F11 in data set G2. The determination unit 13c also calculates the probability density function of the first feature quantity F11 in the reference dataset D1 for each of regions R3 to R9 in the same way.

[0047] Next, the determination unit 13c calculates the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the benchmark dataset D1 based on the probability density function of the first feature quantity F11 in region R1 and the probability density function of the first feature quantity F11 in region R2 in the benchmark dataset D1. This correlation coefficient is equivalent to an index showing the strength of the correlation between the two probability density functions. For all pairs of two regions R that can be selected from regions R1 to R9, except for the pair of regions R1 and R2, the determination unit 13c also calculates the correlation coefficient of the first feature quantity F11 between the two regions R in the benchmark dataset D1. The correlation coefficient calculated in this way is equivalent to the first correlation of the engine 2 under normal conditions (i.e., the correlation of the first feature quantity F11 between different regions R).

[0048] In addition, the determination unit 13c calculates the probability density function of the first feature quantity F11 in each region R of the diagnostic object dataset D2.

[0049] For example, the determination unit 13c uses a data set G1k consisting of n+1 first feature quantities F11 in region R1 to calculate the probability density function of the first feature quantity F11 in region R1. In this case, the probability density function of the first feature quantity F11 in region R1 shows the probability density of each value of the first feature quantity F11 in data set G1k. Similarly, the determination unit 13c uses a data set G2k consisting of n+1 first feature quantities F11 in region R2 to calculate the probability density function of the first feature quantity F11 in region R2. In this case, the probability density function of the first feature quantity F11 in region R2 shows the probability density of each value of the first feature quantity F11 in data set G2k. The determination unit 13c also calculates the probability density function of the first feature quantity F11 in the diagnostic target dataset D2 for each of regions R3 to R9 in the same way.

[0050] Next, the determination unit 13c calculates the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the diagnostic subject dataset D2 based on the probability density function of the first feature quantity F11 in region R1 and the probability density function of the first feature quantity F11 in region R2 in the diagnostic subject dataset D2. This correlation coefficient is equivalent to an index showing the strength of the correlation between the two probability density functions. For all pairs of two regions R that can be selected from regions R1 to R9, except for the pair of regions R1 and R2, the determination unit 13c also calculates the correlation coefficient of the first feature quantity F11 between the two regions R in the diagnostic subject dataset D2. The correlation coefficient calculated in this way is equivalent to the first correlation at the current time Tk (i.e., the correlation of the first feature quantity F11 between different regions R).

[0051] Then, for each pair of two regions R, the determination unit 13c calculates the difference between the correlation coefficient obtained using the benchmark dataset D1 as described above and the correlation coefficient obtained using the diagnostic target dataset D2 as described above. For example, for the pair of regions R1 and R2, the determination unit 13c calculates the difference between the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the benchmark dataset D1 and the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the diagnostic target dataset D2. The determination unit 13c also calculates the difference between the correlation coefficient obtained using the benchmark dataset D1 and the correlation coefficient obtained using the diagnostic target dataset D2 for all pairs of two regions R that can be selected from regions R1 to R9, except for the pair of regions R1 and R2. The difference calculated in this way corresponds to an index showing the change in the correlation of the first feature quantity F11 between the different regions R.

[0052] exist Figure 5 Following step S106, in step S107, the diagnostic unit 13d performs anomaly diagnosis on engine 2. Figure 5 The processing flow shown has ended.

[0053] In step S107, the diagnostic unit 13d performs anomaly diagnosis of engine 2 based on the change in the first correlation determined in step S106. For example, if, among two pairs of regions R that can be selected from regions R1 to R9, there exists a pair where the difference between the correlation coefficient obtained using the benchmark dataset D1 and the correlation coefficient obtained using the diagnostic object dataset D2 is greater than or equal to a benchmark value, the diagnostic unit 13d diagnoses that an anomaly has occurred in engine 2. In this case, the diagnostic unit 13d can also diagnose what kind of anomaly has occurred in engine 2 or where the anomaly has occurred in engine 2 based on which pair has a difference greater than or equal to the benchmark value.

[0054] The diagnostic unit 13d can also calculate the anomaly degree as the difference between the correlation coefficient obtained using the reference dataset D1 and the correlation coefficient obtained using the diagnostic object dataset D2 between two regions R. In this case, the diagnostic unit 13d can also have the storage unit 13e store information showing the anomaly degree of each pair of two regions R that can be selected from regions R1 to R9. Alternatively, information showing the anomaly degree of each pair of two regions R can be calculated and stored in the storage unit 13e when an anomaly in engine 2 is artificially generated. The information showing the anomaly degree of each pair of two regions R is stored in the storage unit 13e in association with information such as what kind of anomaly occurred in engine 2, or where the anomaly occurred in engine 2. Therefore, by comparing the information showing the anomaly degree of each pair of two regions R obtained at the current time Tk with the information stored in the storage unit 13e showing the anomaly degree of each pair of two regions R obtained in the past, the diagnostic unit 13d can appropriately diagnose what kind of anomaly occurred in engine 2, or where the anomaly occurred in engine 2.

[0055] As described above, in the program, diagnostic method, and diagnostic apparatus 1 of the embodiments of this disclosure, a computer (processing unit 13 in the above example) executes a first process, a second process, and a third process. The first process sets multiple regions R in a temperature image IM1 showing the temperature distribution of the device (engine 2 in the above example). The second process determines the correlation between different regions R and the temperature-related first characteristic quantity (first characteristic quantity F11 in the above example), i.e., the change in the first correlation relationship. The third process performs device anomaly diagnosis based on the change in the first correlation relationship. In the above example, the setting unit 13b executes the first process, the determining unit 13c executes the second process, and the diagnostic unit 13d executes the third process. Thus, by focusing on the degree to which the temperature relationship between various parts of the device changes relative to the normal state of the device, device anomalies can be diagnosed with high precision.

[0056] Furthermore, according to the procedure, diagnostic method, and diagnostic apparatus 1 of the embodiments of this disclosure, by visualizing the changes in the first correlation relationship (in the example above, the difference between the correlation coefficient obtained using the benchmark dataset D1 and the correlation coefficient obtained using the diagnostic object dataset D2) of the temperature-related first feature quantity F11 between different regions R, abnormalities of the apparatus can be more easily diagnosed.

[0057] Furthermore, the procedure, diagnostic method, and diagnostic device 1 according to the embodiments of this disclosure, for example, simplify the diagnostic device 1 compared to methods such as diagnosing device malfunctions by installing temperature sensors at various parts of the device. Additionally, the diagnostic device 1 is cheaper than the methods described above. Furthermore, the diagnostic device 1 can be easily transported compared to the methods described above. Finally, the flexibility in the placement of the diagnostic device 1 is increased compared to the methods described above.

[0058] Specifically, in the procedure, diagnostic method, and diagnostic apparatus 1 of the embodiments of this disclosure, in the first process (specifically, the process of setting multiple regions R in a temperature image IM1 showing the temperature distribution of the device), multiple regions R are set in the temperature image IM1 based on a visible light image IM2 reflecting the device. Therefore, it is possible to set regions R after appropriately determining which part of the device (engine 2 in the above example) exists in which display position in the temperature image IM1, thus enabling more appropriate setting of multiple regions R in the temperature image IM1.

[0059] In the above, refer to Figures 1 to 7 The structure and operation of diagnostic device 1 have been explained. However, the structure and operation of diagnostic device 1 are not limited to the examples described above.

[0060] For example, the example described above shows that the diagnostic device 1 has one pair of infrared camera 11 and visible light camera 12. However, the diagnostic device 1 can also have two or more pairs of infrared camera 11 and visible light camera 12. In this case, it is possible to perform abnormality diagnosis of engine 2 on a wider range.

[0061] For example, the example described above illustrates that the region R set in the temperature image IM1 corresponds to the range where each part of the engine 2 exists. However, the region R set in the temperature image IM1 may also include both the range where each part of the engine 2 exists and the range where no part of the engine 2 exists. Furthermore, the region R set in the temperature image IM1 may correspond to a single pixel of the temperature image IM1. In this case, a region R is set for each pixel of the temperature image IM1.

[0062] For example, the example described above uses the average temperature of each region R as the first characteristic quantity F11. However, other characteristic quantities besides the average value mentioned above (e.g., the variance of the temperature in each region R, as described later) can also be used as the first characteristic quantity F11.

[0063] For example, the above description illustrates an example where, in step S106, the determining unit 13c determines the difference between the correlation coefficient obtained using the benchmark dataset D1 and the correlation coefficient obtained using the diagnostic subject dataset D2 as the change in the correlation of the first feature quantity F11 between different regions R, i.e., the first correlation relationship. However, in step S106, the determining unit 13c may also determine the ratio of the correlation coefficient obtained using the benchmark dataset D1 to the correlation coefficient obtained using the diagnostic subject dataset D2 as the change in the correlation of the first feature quantity F11 between different regions R, i.e., the first correlation relationship.

[0064] For example, the above describes an example of calculating the correlation coefficient of the first feature quantity F11 between two regions R in the benchmark dataset D1 and the correlation coefficient of the first feature quantity F11 between two regions R in the diagnostic target dataset D2 in step S106. However, these two processes may not be performed. For example, the determination unit 13c may directly determine the change in the correlation relationship of the first feature quantity F11 between two regions R, i.e., the change in the first correlation relationship, for each pair of two regions R that can be selected from regions R1 to R9, based on the probability density function of the first feature quantity F11 in each region R in the benchmark dataset D1 and the probability density function of the first feature quantity F11 in each region R in each region R in the diagnostic target dataset D2. Specifically, such processing can be achieved using a known method called direct probability density ratio estimation (see "Abnormality Investigation and Transformation Investigation (Mechanical Learning Program Series)" by Tsuyoshi Ide and Masaru Sugiyama, Kodansha, August 8, 2015, etc.). By using direct probability density ratio estimation and omitting the two processes mentioned above, the computational load can be reduced.

[0065] For example, the above describes an example of calculating the correlation coefficient of the first feature quantity F11 between two regions R in the benchmark dataset D1 and the first feature quantity F11 between two regions R in the diagnostic target dataset D2 in step S106. However, in both of these processes, the determination unit 13c may calculate the partial correlation coefficient instead of the correlation coefficient, or calculate the partial correlation coefficient in addition to the correlation coefficient. In this case, the determination unit 13c can determine the change in the first correlation relationship, i.e., the correlation of the first feature quantity F11 between regions R that are different from each other, based on the difference or ratio between the partial correlation coefficient obtained using the benchmark dataset D1 and the partial correlation coefficient obtained using the diagnostic target dataset D2.

[0066] The above describes an example of using the first characteristic quantity F11 as a characteristic quantity in the anomaly diagnosis of engine 2. However, in the anomaly diagnosis of engine 2, other characteristic quantities besides the first characteristic quantity F11 can also be used. The following describes a first and a second variation as examples of using other characteristic quantities besides the first characteristic quantity F11.

[0067] Figure 8 This is a diagram illustrating an example of the benchmark dataset D1, which is a first variation of this disclosure. Figure 8In the baseline dataset D1, in addition to the calculation results of the first feature quantity F11 for each region R at each time T, the calculation results of the first feature quantity F12 for each region R at each time T are also summarized. The first feature quantity F12, like the first feature quantity F11, is a temperature-related feature quantity. Specifically, the first feature quantity F12 is the variance of the temperature in each region R. The determination unit 13c calculates the first feature quantity F12 for each region R at each time T based on the temperature image IM1 obtained at each time T. Furthermore, in Figure 8 The diagram showing the actual calculated value of the first characteristic quantity F12 is omitted. However, at the same time T, the value of the first characteristic quantity F12 can differ for each region R.

[0068] In the first variation, the determination part 13c generates a pair Figure 8 The dataset obtained by adding the calculation results of the first feature quantity F11 and the calculation results of the first feature quantity F12 of each region R at the current time Tk to the baseline dataset D1 is used as the diagnostic target dataset D2. Furthermore, in the first variant example, the determination unit 13c determines the changes in the correlation between the feature quantities based on the baseline dataset D1 and the diagnostic target dataset D2 obtained as described above.

[0069] First, the determination unit 13c calculates the probability density function of the first feature quantity F11 in each region R of the benchmark dataset D1, and the probability density function of the first feature quantity F12 in each region R of the benchmark dataset D1.

[0070] For example, the determination unit 13c uses a data set G3 consisting of n first feature quantities F12 in region R1 to calculate the probability density function of the first feature quantity F12 in region R1. In this case, the probability density function of the first feature quantity F12 in region R1 shows the probability density of each value of the first feature quantity F12 in data set G3. Additionally, the determination unit 13c calculates the probability density function of the first feature quantity F12 in region R2 of the reference dataset D1. Specifically, the determination unit 13c uses a data set G4 consisting of n first feature quantities F12 in region R2 to determine the probability density function of the first feature quantity F12 in region R2. In this case, the probability density function of the first feature quantity F12 in region R2 shows the probability density of each value of the first feature quantity F12 in data set G4. The determination unit 13c similarly calculates the probability density function of the first feature quantity F12 in the reference dataset D1 for each of regions R3 to R9.

[0071] Next, the determination unit 13c determines the correlation relationships for the reference dataset D1, considering all combinations of data groups corresponding to each column. That is, in the first variation, for the reference dataset D1, in addition to calculating the correlation coefficient of the first feature quantity F11 between different regions R, the determination unit 13c also calculates the correlation coefficient of the first feature quantity F12 between different regions R, the correlation coefficient of the first feature quantity F11 and the first feature quantity F12 within the same region R, and the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 corresponding to different regions R respectively. The method for calculating the correlation coefficient of the first feature quantity F12 between different regions R is the same as the method for calculating the correlation coefficient of the first feature quantity F11 between different regions R described above.

[0072] For example, the determination unit 13c calculates the correlation coefficient between the first feature quantity F11 and the first feature quantity F12 in the same region R1 of the benchmark dataset D1 based on the probability density function of the first feature quantity F11 in region R1 of the benchmark dataset D1 and the probability density function of the first feature quantity F12 in region R1 of the benchmark dataset D1. This correlation coefficient is equivalent to an index showing the strength of the correlation between the two probability density functions. The determination unit 13c also calculates the correlation coefficient between the first feature quantity F11 and the first feature quantity F12 in the same region R for each of regions R2 to R9.

[0073] For example, the determination unit 13c calculates the correlation coefficients of the first feature quantity F11 in region R1 of the benchmark dataset D1 and the first feature quantity F12 in region R2 of the benchmark dataset D1, respectively, corresponding to the first feature quantities F11 and F12 that are different from each other in regions R1 and R2 of the benchmark dataset D1. This correlation coefficient is equivalent to an index indicating the strength of the correlation between the two probability density functions. Furthermore, the determination unit 13c calculates the correlation coefficients of the first feature quantities F11 and F12 that are different from each other in regions R2 and R1 of the benchmark dataset D1, respectively, based on the probability density functions of the first feature quantity F11 in region R2 of the benchmark dataset D1 and the first feature quantity F12 in region R1 of the benchmark dataset D1. This correlation coefficient is equivalent to an index indicating the strength of the correlation between the two probability density functions. For all pairs of regions R that can be selected from regions R1 to R9, except for the pair of regions R1 and R2, the determination unit 13c also calculates the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 corresponding to the regions R that are different from each other.

[0074] In addition, for the diagnostic dataset D2, the determination unit 13c, in the same manner as the baseline dataset D1, determines the correlation between all combinations of data groups corresponding to each column.

[0075] Then, the determination unit 13c calculates the differences between the correlation coefficients of the same comparison object among the various correlation coefficients obtained as described above using the benchmark dataset D1 and the various correlation coefficients obtained as described above using the diagnostic object dataset D2.

[0076] For example, the determination unit 13c calculates the difference between the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the benchmark dataset D1 and the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the diagnostic subject dataset D2. Additionally, the determination unit 13c calculates the difference between the correlation coefficient of the first feature quantity F12 between regions R1 and R2 in the benchmark dataset D1 and the correlation coefficient of the first feature quantity F12 between regions R1 and R2 in the diagnostic subject dataset D2. As described above, the difference calculated in this way corresponds to an index showing the change in the correlation between the first feature quantities (specifically, the first feature quantity F11 or the first feature quantity F12) between different regions R, i.e., an indicator of the change in the first correlation relationship.

[0077] For example, the determination unit 13c calculates the difference between the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 in the same region R1 of the benchmark dataset D1 and the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 in the same region R1 of the diagnostic subject dataset D2. Furthermore, the determination unit 13c calculates the difference between the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 corresponding to different regions R1 and R2 in the benchmark dataset D1 and the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 corresponding to different regions R1 and R2 in the diagnostic subject dataset D2. These differences are equivalent to indicators showing the change in the second correlation relationship, which indicates the correlation between the first feature quantity F11 and the first feature quantity F12 in the same region R, or the correlation between the first feature quantity F11 and the first feature quantity F12 corresponding to different regions R.

[0078] In the first variation, in addition to the change in the first correlation described above, the diagnostic unit 13d also performs anomaly diagnosis of engine 2 based on the change in the second correlation described above. For example, in the first variation, if there are regions R where the difference between the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 in the same region R is greater than or equal to a baseline value, the diagnostic unit 13d diagnoses an anomaly of engine 2. Furthermore, in the first variation, if there are pairs of regions R where the difference between the correlation coefficients of the first feature quantity F11 and the first feature quantity F12 in regions R that are different from each other is greater than or equal to a baseline value, the diagnostic unit 13d diagnoses an anomaly of engine 2. However, the determination unit 13c may also determine the ratio of the correlation relationship as an indicator showing the change in the second correlation relationship, instead of the difference in the correlation relationship.

[0079] As described above, in the first variation, the first characteristic quantities F11 and F12 include multiple characteristic quantities. In the second process (specifically, the process of determining the correlation between temperature-related first characteristic quantities between different regions R, i.e., the change of the first correlation), in addition to the change of the first correlation, the process further determines the correlation between different types of first characteristic quantities F11 and F12 in the same region R, or the correlation between different types of first characteristic quantities F11 and F12 corresponding to different regions R, i.e., the change of the second correlation. In the third process (specifically, the process of performing device anomaly diagnosis based on the change of the first correlation), in addition to the change of the first correlation, the anomaly diagnosis is also performed based on the change of the second correlation. Thus, by focusing on the degree to which the temperature relationship between various parts of the device changes based on the normal state of the device (engine 2 in the above example), more information can be used for anomaly diagnosis, thereby enabling more accurate diagnosis of device anomalies.

[0080] In the above, examples of using the first characteristic quantity F11 as the average temperature in each region R and the first characteristic quantity F12 as the variance of the temperature in each region R were given as examples of various first characteristic quantities. However, the combination and number of first characteristic quantities used for anomaly diagnosis are not limited to the examples above. For example, the maximum temperature in each region R can also be used as a first characteristic quantity. For example, more than three first characteristic quantities can also be used.

[0081] The above describes an example of processing various correlation coefficients between characteristic quantities. However, in the first variation, the direct probability density ratio estimation can be used, omitting the process of calculating the correlation coefficient. Furthermore, in the first variation, the determination unit 13c can calculate the partial correlation coefficient instead of the correlation coefficient, or calculate the partial correlation coefficient in addition to the correlation coefficient.

[0082] Figure 9 This is a diagram illustrating an example of the benchmark dataset D1, which is a second variation of this disclosure. Figure 9 In the baseline dataset D1, in addition to the calculation results of the first feature quantities F11 and F12 for each region R at each time T, the calculation results of the second feature quantities F21 and F22 for each time T are also summarized. The second feature quantities F21 and F22 are feature quantities that do not depend on the region R. That is, the second feature quantities F21 and F22 do not depend on the position in engine 2, and are common feature quantities at all positions.

[0083] Specifically, the second characteristic quantity F21 is a characteristic quantity related to the operating state of the engine 2. Examples of the second characteristic quantity F21 include the output of the engine 2 and its rotational speed. The determination unit 13c can, for example, obtain the output of the engine 2 at each time T based on information obtained from the control device that controls the operation of the engine 2. In addition, the determination unit 13c can, for example, obtain the rotational speed of the engine 2 at each time T based on the detection result of the sensor that detects the rotational speed of the engine 2.

[0084] Furthermore, the second characteristic quantity F22 is a characteristic quantity related to the heat input and output of the engine 2. Examples of the second characteristic quantity F22 include, for example, the ambient temperature of the engine 2, the temperature of the coolant, the temperature of the lubricating oil, the temperature of the fuel, the temperature of the intake air, the airflow state of the surrounding air, and the temperature of the surrounding heat source. The determination unit 13c can, for example, obtain this information at each time T based on the detection results of the sensors that detect this information. However, the determination unit 13c can also calculate the aforementioned temperatures (e.g., the temperature around the engine 2) at each time T based on the temperature image IM1 obtained at each time T.

[0085] In the second variation, the determination part 13c generates a pair Figure 9 The dataset obtained by adding the calculation results of the first feature quantities F11 and F12 of each region R at the current time Tk to the baseline dataset D1, and the calculation results of the second feature quantities F21 and F22, is used as the diagnostic target dataset D2. Furthermore, in the second variation, the determination unit 13c determines the changes in the correlation between the feature quantities based on the baseline dataset D1 and the diagnostic target dataset D2 obtained as described above.

[0086] First, the determination unit 13c calculates the probability density functions of the first feature quantities F11 and F12 in each region R of the benchmark dataset D1, and the probability density functions of the second feature quantities F21 and F22 in each region R of the benchmark dataset D1.

[0087] For example, the determination unit 13c uses a data set G5 consisting of n second feature quantities F21 to calculate the probability density function of the second feature quantity F21. In this case, the probability density function of the second feature quantity F21 shows the probability density of each value of the second feature quantity F21 in the data set G5. Similarly, the determination unit 13c uses a data set G6 consisting of n second feature quantities F22 to calculate the probability density function of the second feature quantity F22. In this case, the probability density function of the second feature quantity F22 shows the probability density of each value of the second feature quantity F22 in the data set G6.

[0088] Next, the determination unit 13c determines the correlation between all combinations of data groups corresponding to each column for the reference dataset D1. That is, in the second variation, for the reference dataset D1, in addition to determining the correlation coefficients between the first feature quantities, the determination unit 13c also determines the correlation between the first feature quantity (specifically, the first feature quantity F11 or the first feature quantity F12) and the second feature quantity (specifically, the second feature quantity F21 or the second feature quantity F22).

[0089] For example, the determination unit 13c calculates the correlation coefficient between the first feature quantity F11 and the second feature quantity F21 in region R1 of the benchmark dataset D1 based on the probability density function of the first feature quantity F11 in region R1 of the benchmark dataset D1 and the probability density function of the second feature quantity F21 in the benchmark dataset D1. The determination unit 13c also calculates the correlation coefficient between the first feature quantity F11 and the second feature quantity F21 in each region R for each of regions R2 to R9.

[0090] For example, the determination unit 13c calculates the correlation coefficient between the first feature quantity F12 and the second feature quantity F21 in region R1 of the benchmark dataset D1 based on the probability density function of the first feature quantity F12 in region R1 of the benchmark dataset D1 and the probability density function of the second feature quantity F21 in the benchmark dataset D1. This correlation coefficient is equivalent to an index showing the strength of the correlation between the two probability density functions. The determination unit 13c also calculates the correlation coefficient between the first feature quantity F12 and the second feature quantity F21 in each region R for each of regions R2 to R9.

[0091] For example, the determination unit 13c calculates the correlation coefficient between the first feature quantity F11 and the second feature quantity F22 in region R1 of the benchmark dataset D1 based on the probability density function of the first feature quantity F11 in region R1 of the benchmark dataset D1 and the probability density function of the second feature quantity F22 in the benchmark dataset D1. This correlation coefficient is equivalent to an index showing the strength of the correlation between the two probability density functions. The determination unit 13c also calculates the correlation coefficient between the first feature quantity F11 and the second feature quantity F22 in each region R for each of regions R2 to R9.

[0092] For example, the determination unit 13c calculates the correlation coefficient between the first feature quantity F12 and the second feature quantity F22 in region R1 of the benchmark dataset D1 based on the probability density function of the first feature quantity F12 in region R1 of the benchmark dataset D1 and the probability density function of the second feature quantity F22 in region R1 of the benchmark dataset D1. This correlation coefficient is equivalent to an index showing the strength of the correlation between the two probability density functions. The determination unit 13c also calculates the correlation coefficient between the first feature quantity F12 and the second feature quantity F22 in each region R for each of regions R2 to R9.

[0093] In addition, for the diagnostic dataset D2, the determination unit 13c, in the same manner as the baseline dataset D1, determines the correlation between all combinations of data groups corresponding to each column.

[0094] Then, the determination unit 13c calculates the differences between the correlation coefficients of the same comparison object among the various correlation coefficients obtained as described above using the benchmark dataset D1 and the various correlation coefficients obtained as described above using the diagnostic object dataset D2.

[0095] For example, the determination unit 13c calculates the difference between the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the benchmark dataset D1 and the correlation coefficient of the first feature quantity F11 between regions R1 and R2 in the diagnostic subject dataset D2. Additionally, the determination unit 13c calculates the difference between the correlation coefficient of the first feature quantity F12 between regions R1 and R2 in the benchmark dataset D1 and the correlation coefficient of the first feature quantity F12 between regions R1 and R2 in the diagnostic subject dataset D2. As described above, the difference calculated in this way corresponds to an index showing the change in the correlation between the first feature quantities (specifically, the first feature quantity F11 or the first feature quantity F12) between different regions R, i.e., an indicator of the change in the first correlation relationship.

[0096] For example, the determination unit 13c calculates the difference between the correlation coefficients of the first feature quantities F11 and F12 in the same region R1 of the benchmark dataset D1 and the correlation coefficients of the first feature quantities F11 and F12 in the same region R1 of the diagnostic subject dataset D2. Furthermore, the determination unit 13c calculates the difference between the correlation coefficients of the first feature quantities F11 and F12 corresponding to different regions R1 and R2 in the benchmark dataset D1 and the correlation coefficients of the first feature quantities F11 and F12 corresponding to different regions R1 and R2 in the diagnostic subject dataset D2. As described above, the difference calculated in this way is equivalent to an index showing the change in the second correlation relationship, which indicates the correlation between the first feature quantities F11 and F12 in the same region R, or the correlation between the first feature quantities F11 and F12 corresponding to different regions R.

[0097] For example, the determination unit 13c calculates the correlation coefficients of the first feature quantity F11 and the second feature quantity F21 in region R1 of the benchmark dataset D1, and the difference between these correlation coefficients and those of the first feature quantity F11 and the second feature quantity F21 in region R1 of the diagnostic subject dataset D2. Additionally, the determination unit 13c calculates the correlation coefficients of the first feature quantity F12 and the second feature quantity F21 in region R1 of the benchmark dataset D1, and the difference between these correlation coefficients and those of the first feature quantity F12 and the second feature quantity F21 in region R1 of the diagnostic subject dataset D2. Furthermore, the determination unit 13c calculates the correlation coefficients of the first feature quantity F11 and the second feature quantity F22 in region R1 of the benchmark dataset D1, and the difference between these correlation coefficients and those of the first feature quantity F11 and the second feature quantity F22 in region R1 of the diagnostic subject dataset D2. Finally, the determination unit 13c calculates the correlation coefficients of the first feature quantity F12 and the second feature quantity F22 in region R1 of the benchmark dataset D1, and the difference between these correlation coefficients and those of the first feature quantity F12 and the second feature quantity F22 in region R1 of the diagnostic subject dataset D2. These differences are equivalent to indicators showing the changes in the correlation between the first and second characteristic quantities, i.e., the third correlation.

[0098] In the second variation, the diagnostic unit 13d performs anomaly diagnosis of engine 2 based not only on changes in the first correlation and the second correlation, but also on changes in the third correlation. For example, in the second variation, if there exists a region R in which the difference between the correlation coefficients of the first feature quantity (specifically, the first feature quantity F11 or the first feature quantity F12) and the second feature quantity (specifically, the second feature quantity F21 or the second feature quantity F22) in any region R is greater than or equal to a baseline value, the diagnostic unit 13d diagnoses an anomaly of engine 2. However, the determination unit 13c may also use the ratio of the correlation to indicate changes in the third correlation, instead of the difference in the correlation.

[0099] As described above, in the second variation, in the second process (specifically, the process of determining the correlation between temperature-related first characteristic quantities, i.e., the change of the first correlation, between different regions R), in addition to the change of the first correlation, the correlation between the first characteristic quantity (in the above example, the first characteristic quantity F11 or the first characteristic quantity F12) and a second characteristic quantity (in the above example, the second characteristic quantity F21 or the second characteristic quantity F22, which is independent of region R), i.e., the third correlation, is further determined. In the third process (specifically, the process of diagnosing device anomalies based on the change of the first correlation), in addition to the change of the first correlation, the anomaly diagnosis is also based on the change of the third correlation. Thus, by considering the extent to which the temperature relationship between various parts of the device (in the above example, engine 2) changes relative to the normal state of the device, the influence of the second characteristic quantity on the temperature of various parts of the device can be taken into account for anomaly diagnosis, thereby enabling more accurate diagnosis of device anomalies.

[0100] Specifically, in the second variation, the second characteristic quantity includes a characteristic quantity (the second characteristic quantity F21) related to the operating state of the device (engine 2 in the above example). Therefore, it is possible to consider the impact of the device's operating state on the temperature of various parts of the device for anomaly diagnosis, thus enabling more accurate diagnosis of device anomalies.

[0101] Specifically, in the second variation, the second characteristic quantity includes a characteristic quantity (the second characteristic quantity F22 in the above example) related to the thermal input and output of the device (engine 2 in the above example). Therefore, it is possible to consider the impact of the thermal input and output state of the device on the temperature of various parts of the device for anomaly diagnosis, thus enabling more accurate diagnosis of device anomalies.

[0102] The above illustrates an example where, in addition to changes in the first and second correlations, changes in the third correlation are also identified, and anomaly diagnosis is made based on these changes. However, in the above example, it is also possible to omit the determination of changes in the second correlation. For example, one could... Figure 9 In the benchmark dataset D1, one of the first feature quantity F11 and the first feature quantity F12 is omitted, and there is one type of first feature quantity.

[0103] The above describes an example of processing various correlation coefficients between characteristic quantities. However, in the second variation, the direct probability density ratio estimation can be used, omitting the process of calculating the correlation coefficient. Furthermore, in the second variation, the determination unit 13c can calculate the partial correlation coefficient instead of the correlation coefficient, or calculate the partial correlation coefficient in addition to the correlation coefficient.

[0104] The embodiments have been described above with reference to the accompanying drawings, but it is self-evident that this disclosure is not limited to the above embodiments. Those skilled in the art will understand that various modifications or alterations will be readily apparent within the scope of the claims, and these naturally fall within the technical scope of this disclosure.

[0105] Explanation of reference numerals in the attached figures 1: Diagnostic device, 2: Engine (device), 11: Infrared camera, 12: Visible light camera, 13: Processing device, 13a: Acquisition unit, 13b: Setting unit, 13c: Determination unit, 13d: Diagnostic unit, 13e: Storage unit, F11: First feature quantity, F12: First feature quantity, F21: Second feature quantity, F22: Second feature quantity, IM1: Temperature image, IM2: Visible light image, R1: Region, R2: Region, R3: Region, R4: Region, R5: Region, R6: Region, R7: Region, R8: Region, R9: Region.

Claims

1. A program in which, To make the computer perform: The first process involves defining multiple regions in a temperature image showing the temperature distribution of the device; The second process involves determining the change in the first correlation, which is the correlation of a temperature-related first feature between the different regions. as well as The third step involves diagnosing any abnormalities in the device based on changes in the first correlation.

2. The procedure according to claim 1, wherein, The first feature quantity includes multiple feature quantities. In the second process, in addition to the change in the first correlation, a change in the second correlation is further determined. The second correlation is the correlation between different types of the first feature quantities within the same region, or the correlation between different types of the first feature quantities corresponding to different regions. In the third process, in addition to the change in the first correlation, the anomaly diagnosis is also performed based on the change in the second correlation.

3. The procedure according to claim 1 or 2, wherein, In the second process, in addition to the change in the first correlation, a change in the third correlation is further determined. The third correlation is the correlation between the first feature and a second feature that is independent of the region. In the third process, in addition to the change in the first correlation, the anomaly diagnosis is also performed based on the change in the third correlation.

4. The procedure according to claim 3, wherein, The second characteristic quantity includes characteristic quantities related to the operating state of the device.

5. The procedure according to claim 3, wherein, The second characteristic quantity includes characteristic quantities related to the thermal input and output of the device.

6. The procedure according to claim 1 or 2, wherein, In the first process, a plurality of regions are defined in the temperature image based on the visible light image of the device.

7. A diagnostic method, wherein, Computer execution: The first process involves defining multiple regions in a temperature image showing the temperature distribution of the device; The second process involves determining the change in the first correlation, which is the correlation of a temperature-related first feature between the different regions. as well as The third step involves diagnosing any abnormalities in the device based on changes in the first correlation.

8. A diagnostic device, wherein, have: An infrared camera captures temperature images showing the temperature distribution of the device; and The processing device performs the following first processing step: defining multiple regions in the temperature image. The second process involves determining the change in the first correlation, which is the correlation of a temperature-related first feature between the different regions. And a third process, based on changes in the first correlation, to diagnose anomalies in the device.

Images