Overheat prediction system, overheat prediction method, and non-transitory computer-readable storage medium

Abstract

A system and method for predicting overheating and a non-transitory computer readable storage medium are disclosed. The method for predicting overheating is used to predict whether a subject has a tendency of overheating, and includes the following steps. A statistical model is established using a reference temperature distribution map to obtain a threshold temperature distribution map. The threshold temperature distribution map is used to filter a subject temperature distribution map to screen out potential abnormal temperature distribution maps. Cluster analysis is used to determine whether the potential abnormal temperature distribution maps belong to the same subset as the reference temperature distribution map. The potential abnormal temperature distribution maps that do not belong to the same subset as the reference temperature distribution map are marked as abnormal. When all the potential abnormal temperature distribution maps are marked as abnormal, an overheating warning information is generated.

Description

Technical Field

[0001] This disclosure relates to a prediction system and method, and in particular to an overheating prediction system and method. Background Technology

[0002] Most electronic components have operating temperature limits. When the temperature is too high, they will not function properly, will be damaged, or may even cause a fire. Therefore, for factories with many large pieces of equipment, ensuring that their instruments and equipment are within the operating temperature range is an important task for maintaining industrial safety.

[0003] However, current technology relies on manual handheld infrared guns to monitor equipment temperature at regular intervals, which is not only time-consuming but also consumes a lot of manpower. On the other hand, the gaps between each monitoring session create a window of opportunity, making it impossible to prevent situations where the temperature rises rapidly in a short period of time due to a malfunction. Furthermore, the amount of temperature change data obtained is difficult to apply to data analysis to understand the temperature change trend of the equipment.

[0004] Taking transformers as an example, there are also partial discharge sensors that detect the discharge phenomenon caused by high voltage damaging the insulation, in order to provide users with fault information. However, this method cannot provide early warning and prevent safety hazards caused by abnormal power supply and electrical room fires. Summary of the Invention

[0005] Therefore, this disclosure provides an overheating prediction method, the steps of which include: obtaining multiple reference temperature distribution maps and multiple temperature distribution maps to be measured; establishing a statistical model based on the reference temperature distribution maps to obtain a threshold temperature distribution map; filtering the temperature distribution maps to be measured using the threshold temperature distribution map to select at least one potential abnormal temperature distribution map; performing cluster analysis on the reference temperature distribution maps and the potential abnormal temperature distribution maps to determine whether the potential abnormal temperature distribution maps do not belong to the same subset as the reference temperature distribution maps; when the potential abnormal temperature distribution maps do not belong to the same subset as the reference temperature distribution maps, marking the potential abnormal temperature distribution maps as abnormal; and generating an overheating warning message.

[0006] Another disclosure provides an overheating prediction system comprising a thermal image acquisition device and a processing unit. The thermal image acquisition device is configured to capture images of an object under test to obtain multiple temperature distribution maps. The processing unit is communicatively connected to the thermal image acquisition device and configured to perform the following steps: acquiring multiple reference temperature distribution maps and multiple temperature distribution maps to be measured from the thermal image acquisition device; establishing a statistical model using the reference temperature distribution maps to obtain a threshold temperature distribution map; filtering the temperature distribution maps to be measured using the threshold temperature distribution map to identify at least one potential abnormal temperature distribution map; performing cluster analysis on the reference temperature distribution maps and the potential abnormal temperature distribution maps to determine whether the potential abnormal temperature distribution maps do not belong to the same subset as the reference temperature distribution maps; when the potential abnormal temperature distribution maps do not belong to the same subset as the reference temperature distribution maps, marking the potential abnormal temperature distribution maps as abnormal; and generating an overheating warning message.

[0007] This disclosure also provides a non-transitory computer-readable storage medium having at least one set of instructions stored thereon, which, when executed by a processing unit, perform the above-described overheating prediction method.

[0008] It should be understood that the foregoing general description and the following specific description are merely exemplary and explanatory, and are intended to provide further explanation of the requested disclosure. Attached Figure Description

[0009] To make the above and other objects, features, advantages and embodiments of this disclosure more apparent and understandable, the accompanying drawings are described below:

[0010] Figure 1 This is a schematic diagram of an overheat prediction system according to an embodiment of this disclosure.

[0011] Figure 2 This is a flowchart of an overheating prediction method according to an embodiment of this disclosure.

[0012] Figure 3a This is a schematic diagram of a temperature distribution map according to an embodiment of this disclosure.

[0013] Figure 3b This is a schematic diagram of an abnormal temperature distribution in an image capture according to an embodiment of this disclosure.

[0014] Figure 4 This is a schematic diagram of a thermal image capturing device according to an embodiment of this disclosure, capturing a reference temperature distribution map and a temperature distribution map to be measured over a period of time.

[0015] Figure 5 This is a schematic diagram of the threshold temperature distribution map and the temperature distribution map to be measured, according to an embodiment of this disclosure.

[0016] Figure 6 A schematic diagram illustrating cluster analysis of a reference temperature distribution map and a potential anomalous temperature distribution map according to an embodiment of this disclosure; and

[0017] Figure 7 This is a schematic diagram of an overheat prediction system according to another embodiment of this disclosure.

[0018] In the attached figures, the following labels are used:

[0019] 100: Overheat Prediction System

[0020] 110: Thermal image capturing device

[0021] 111: Displacement Detection Unit

[0022] 120: Processing Unit

[0023] 130: Output device

[0024] 200: Overheating Prediction Methods

[0025] S201~S212: Steps

[0026] OBJ: Test Item

[0027] SVR: Server

[0028] MD: Mobile Device

[0029] DP: Display device

[0030] IMGa: Temperature Distribution Map

[0031] IMGe1, IMGe2, IMGe3: Imaging anomaly temperature distribution map

[0032] IMGm1~IMGm30: Reference Temperature Distribution Map

[0033] IMGn1~IMGn10: Temperature distribution map to be measured

[0034] Pt: Threshold Temperature Distribution Map

[0035] P1, P2: Temperature distribution diagrams to be measured

[0036] S: Subset

[0037] P: Data point Detailed Implementation

[0038] To make the description in this disclosure more detailed and complete, reference can be made to the accompanying drawings and the various embodiments described below, in which the same numbers represent the same or similar elements.

[0039] Please refer to Figure 1This is a schematic diagram of the overheat prediction system 100 according to the first embodiment of this disclosure. Figure 1 As shown, the overheating prediction system 100 includes a thermal image acquisition device 110, a processing unit 120, and an output device 130. The thermal image acquisition device 110 is configured to capture images of a test object (OBJ) over a period of time, obtaining a series of temperature distribution maps of the OBJ. The processing unit 120 is communicatively connected to the thermal image acquisition device 110, receives these temperature distribution maps from the thermal image acquisition device 110, and is configured to establish a statistical model and perform cluster analysis based on these temperature distribution maps as reference data for overheating prediction of the OBJ. The output device 130 is communicatively connected to the processing unit 120, receives information from the processing unit 120, and issues a corresponding warning. The server SVR is communicatively connected to the processing unit 120 and is configured to receive a warning message from the processing unit 120. After receiving the warning message, the server SVR can send the warning message via SMS, Email, or other means to provide users with information about the warning message through mobile devices (MD) and / or display devices (DP). The specific technical means for building models, analyzing, and predicting will be described in the following paragraphs.

[0040] In some embodiments disclosed herein, the thermal image capturing device 110 may be an infrared thermal imager, which can capture images such as... Figure 3a The temperature distribution map shown is IMGa, which is an image or array composed of multiple pixels, each containing temperature information corresponding to its shooting location.

[0041] In some embodiments disclosed herein, the processing unit 120 may be a processor, a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), or any other type of processing element.

[0042] Please refer to further information. Figure 2 This is a flowchart of an overheating prediction method 200 according to an embodiment of this disclosure. The overheating prediction method 200 can be... Figure 1 The overheating prediction system 100 shown is implemented.

[0043] As shown in Figures 1 and 2, in step S201, the processing unit 120 acquires multiple reference temperature distribution maps from the thermal image capturing device 110. In some embodiments of this disclosure, these reference temperature distribution maps are a series of temperature distribution maps acquired by the thermal image capturing device 110 at a fixed frequency over a period of time from the object under test OBJ.

[0044] In step S202, the processing unit 120 removes reference temperature distribution maps with imaging abnormalities from these reference temperature distribution maps. Specifically, it determines whether each of the reference temperature distribution maps has imaging abnormalities based on the pixel percentage distribution of each map, and then further removes the reference temperature distribution maps determined to have imaging abnormalities. For example... Figure 3b The image capture anomaly temperature distribution maps IMGe1 ​​and IMGe2 are shown in the image. In these maps, a large area of ​​adjacent pixels contains the same temperature information (at this time, the pixel percentage distribution of the image capture anomaly temperature distribution maps IMGe1 ​​and IMGe2 will be obviously concentrated in a specific gray level. For example, if more than 50% of the pixels are concentrated in the same gray level, it can be judged as an anomaly). Or, if the image capture anomaly temperature distribution map IMGe3 is affected by noise, the temperature information is randomly distributed in the map (at this time, the pixel percentage distribution of the image capture anomaly temperature distribution map IMGe3 will be dispersed to all gray levels, without obvious distribution changes, which can also be judged as an anomaly). In fact, there are many image processing algorithms for judging image anomalies. This disclosure document is not limited to the two examples mentioned above.

[0045] After the processing unit 120 identifies reference temperature distribution maps with imaging anomalies, it removes them from the reference temperature distribution maps to avoid the biased information they contain reducing the accuracy of overheating prediction. In some embodiments of this disclosure, the processing unit 120 further obtains a temperature distribution map from the thermal image capturing device 110 that was captured at a similar time to the reference temperature distribution map with imaging anomalies, or captured at the same frequency after the time period of the reference temperature distribution maps to which it belongs, and replaces the reference temperature distribution map with it. For example, if two out of ten consecutively captured reference temperature distribution maps are determined to be abnormal, then the subsequent two reference temperature distribution maps are used to replace them.

[0046] In step S203, the processing unit 120 establishes a statistical model based on the reference temperature distribution maps to obtain a threshold temperature distribution map. The purpose is to establish a model using statistical methods based on the temperature change trends of the reference temperature distribution maps, serving as a reference benchmark for overheat prediction. The threshold temperature distribution map contains multiple threshold temperatures corresponding to each pixel, which are threshold values ​​calculated based on the quartiles in the statistical model. Specifically, the processing unit 120 statistically analyzes the co-location temperature information groups contained in co-location pixel groups at the same position in the reference temperature distribution maps. It first sets one or more temperature intervals based on a certain multiple of the interquartile range and / or standard deviation of the co-location temperature information groups. For example, in some embodiments of this disclosure, these temperature intervals can be represented by the following formulas:

[0047] [Formula 1]

[0048] [Q1(T n )-αIQR(Tn ),Q3(T n )+αIQR(T n )]

[0049] [Formula 2]

[0050] [mean(T n )-βstd(T n ),mean(T n )+βstd(T n )]

[0051] Where T n For this isotopic temperature information group, Q1 is a first quartile, Q3 is a third quartile, IQR is an interquartile range, α is a first fold, mean is a mean, std is a standard deviation, and β is a second fold.

[0052] Subsequently, temperature information exceeding these temperature ranges is removed from the isotopic temperature information group to obtain a corrected isotopic temperature information group, avoiding the influence of extreme values ​​on subsequent threshold temperature calculations. Then, the threshold temperature is calculated using a statistical model of the corrected isotopic temperature information group. For example, in some embodiments of this disclosure, the threshold temperature can be expressed by the following Equation 3:

[0053] [Formula 3]

[0054]

[0055] Where UCL is the threshold temperature. This is the corrected isotopic temperature information group, where ρ is a third multiplier.

[0056] The processing unit 120 traverses all pixels in the reference temperature distribution maps in the same way, calculates the corresponding threshold temperatures, and generates a threshold temperature distribution map.

[0057] In step S204, the processing unit 120 acquires multiple temperature distribution maps of the object under test from the thermal image capturing device 110. In some embodiments of this disclosure, these temperature distribution maps are a series of temperature distribution maps of the object under test (OBJ) captured within a shorter imaging time than the reference temperature distribution maps when the user wants to know whether the object under test (OBJ) has a tendency to overheat.

[0058] For the time points when these reference temperature distribution maps and these temperature distribution maps to be measured were taken, please refer to... Figure 4This is a schematic diagram of a thermal image capturing device 110 according to an embodiment of this disclosure, capturing reference temperature distribution maps IMGm1-30 and test temperature distribution maps IMGn1-10 over a period of time. The thermal image capturing device 110 captures one temperature distribution map of the test object OBJ every 60 seconds until 30 maps are accumulated, which are then used as the reference temperature distribution maps IMGm1-30. On the other hand, after capturing these reference temperature distribution maps, the thermal image capturing device 110 captures one temperature distribution map of the test object OBJ every 0.2 seconds until 10 maps are accumulated, which are then used as the test temperature distribution maps IMGn1-10.

[0059] Similar to step S202, in step S205, the processing unit 120 removes temperature distribution maps with imaging abnormalities from the temperature distribution maps to be measured. Specifically, it determines whether each of the temperature distribution maps to be measured has imaging abnormalities based on the pixel percentage distribution of each map, using the same method as the reference temperature distribution map used in step S202 to determine imaging abnormalities. After the processing unit 120 identifies the temperature distribution maps with imaging abnormalities, it removes them from the temperature distribution maps to avoid the biased information they contain reducing the accuracy of overheat prediction. Subsequently, the processing unit 120 obtains a temperature distribution map from the thermal image capturing device 110 that was captured at a similar time to the temperature distribution map with imaging abnormalities, or captured at the same frequency after the time period of the temperature distribution maps to be measured, and replaces the temperature distribution map with the one that has imaging abnormalities.

[0060] In step S206, the processing unit 120 filters the temperature distribution maps to be tested using the threshold temperature distribution map, and selects at least one potential abnormal temperature distribution map. Specifically, the processing unit 120 compares whether the temperature information of each pixel in each of the temperature distribution maps to be tested exceeds one of the corresponding threshold temperatures in the threshold temperature distribution map. If the number of pixels in the temperature distribution maps to be tested whose temperature information exceeds the threshold temperature is greater than a preset threshold, then it is marked as a potential abnormal temperature distribution map.

[0061] like Figure 5The diagram illustrates a threshold temperature distribution map Pt and test temperature distribution maps P1 and P2 according to an embodiment of this disclosure. For ease of explanation, the temperature information of each pixel in the threshold temperature distribution map Pt and the test temperature distribution maps P1 and P2 is represented in numerical form. During step S206, the processing unit 120 compares the threshold temperature of each pixel in the threshold temperature distribution map Pt with the temperature information of pixels at the same position in the test temperature distribution maps P1 and P2. Pixels marked with halftone dots in the test temperature distribution maps P1 and P2 indicate that their temperature information is higher than the corresponding threshold temperature. When the threshold is set to 5, the test temperature distribution map P1, because only one pixel's temperature information exceeds the corresponding threshold temperature, will not be marked as a potentially abnormal temperature distribution map. On the other hand, the test temperature distribution map P2 has 13 pixels' temperature information exceeding the corresponding threshold temperature; therefore, the test temperature distribution map P2 will be marked as a potentially abnormal temperature distribution map by the processing unit 120.

[0062] Next, in step S207, the processing unit 120 determines whether all the temperature distribution maps to be measured are marked as potentially abnormal temperature distribution maps. If not, it is determined that the temperature change of the test object OBJ is normal and there is no overheating trend for the time being. The processing unit 120 then executes step S213 to add the temperature distribution maps to be measured to the reference temperature distribution maps as reference data for the next determination of whether the test object OBJ has an overheating trend. When the next prediction is to be made, the process returns to step S203, and the processing unit 120 establishes a statistical model with a new set of reference temperature distribution maps.

[0063] On the other hand, if all the temperature distribution maps to be measured are marked as potential abnormal temperature distribution maps, the processing unit 120 further executes step S208 to perform cluster analysis on the reference temperature distribution maps and the potential abnormal temperature distribution maps to determine whether the potential abnormal temperature distribution maps belong to the same subset as the reference temperature distribution maps. When performing cluster analysis, the processing unit 120 can use different clustering algorithms as needed, such as DBSCAN (density-based spatial clustering of applications with noise), K-nearest neighbors algorithm, K-means clustering algorithm, and Linde–Buzo–Gray algorithm. In some embodiments of this disclosure, each of the reference temperature distribution maps and the potential abnormal temperature distribution maps is projected onto a coordinate plane to form its own data points. In one embodiment, the reference temperature distribution maps and the potential abnormal temperature distribution maps can be projected onto a two-dimensional coordinate plane with time on the X-axis and temperature on the Y-axis to form multiple data points. After the above projection, data points on the two-dimensional coordinate plane that are close in Euclidean distance are grouped into the same subset. That is, for example, data points with similar temperatures (within 5%) will be grouped into the same subset. Then, in step S209, the processing unit 120 determines whether the data points corresponding to these potential abnormal temperature distribution maps and the data points corresponding to these reference temperature distribution maps belong to the same subset.

[0064] Generally, the aforementioned reference temperature distribution maps are images taken under normal operating conditions (no overheating occurs), and the projected data points will be close to each other, forming a subset of normal data points. The purpose of step S209 is to determine whether the data points corresponding to the potential abnormal temperature distribution map fall into the subset of normal data points.

[0065] If one of the potential abnormal temperature distribution maps and the reference temperature distribution maps belong to the same subset, it is determined that the temperature change of the test object OBJ is normal and there is no overheating trend at present. The processing unit 120 then executes step S213, adding the test temperature distribution maps to the reference temperature distribution maps as reference data for the next judgment on whether the test object OBJ has an overheating trend. When the next prediction is to be performed, it returns to step S203, and the processing unit 120 establishes a statistical model with a new set of reference temperature distribution maps. On the other hand, if the processing unit 120 determines that neither the potential abnormal temperature distribution maps nor the reference temperature distribution maps belong to the same subset, it is determined that the temperature change of the test object OBJ has an overheating trend, and the processing unit 120 further executes step S210.

[0066] For information on how to use cluster analysis to determine whether each of the potential anomalous temperature distribution maps and the reference temperature distribution maps belongs to the same subset, please refer to [link to relevant documentation]. Figure 6 This is a schematic diagram illustrating cluster analysis of one of the reference temperature distribution maps and one of the potential abnormal temperature distribution maps according to an embodiment of this disclosure. The processing unit 120 first projects the reference temperature distribution maps onto a two-dimensional coordinate plane with time as one axis and temperature as the other, resulting in multiple data points. For example, the processing unit 120 uses a clustering algorithm to calculate a subset S corresponding to the data points of the reference temperature distribution maps. On the other hand, one of the potential abnormal temperature distribution maps is also projected onto the two-dimensional coordinate plane, becoming a data point P. If data point P falls within the range of subset S, it is determined that one of the potential abnormal temperature distribution maps and the reference temperature distribution maps belong to the same subset; otherwise, if... Figure 6 If the data point P shown in the figure does not fall within the range of subset S, then it is determined that one of the potential abnormal temperature distribution maps and the reference temperature distribution maps do not belong to the same subset.

[0067] Next, in step S210, the processing unit 120 marks these potential abnormal temperature distribution maps as abnormal. Since in step S209, these potential abnormal temperature distribution maps are not in the same subset as these reference temperature distribution maps, it means that the temperature change of the test object OBJ has deviated from the normal operating temperature represented by these reference temperature distribution maps. Therefore, the overheating prediction system 100 or the overheating prediction method 200 determines that the test object OBJ has an overheating trend, and thus marks these potential abnormal temperature distribution maps as abnormal.

[0068] Further, the processing unit 120 executes step 211 to generate an overheating warning message. In some embodiments of this disclosure, the overheating warning message includes identification information of the OBJ under test, potential abnormal temperature distribution maps, and information such as the time when the potential abnormal temperature distribution maps were captured.

[0069] Subsequently, in step S212, the processing unit 120 issues the overheat warning message. In some embodiments of this disclosure, the output device 130 issues the overheat warning message after receiving it from the processing unit 120. The overheat warning message can be issued through various technical means such as flashing lights, sound, or displaying text or graphics on the screen. Correspondingly, the output device 130 can be a device that includes a light-emitting unit, a sound-emitting unit, a display screen, and any other components with output functions, which can be arbitrarily combined according to user needs.

[0070] On the other hand, in other embodiments of this disclosure, the processing unit 120 may also transmit the overheating information to the server SVR. After receiving the overheating warning information, the server SVR may send the overheating warning information via SMS, email or other means to provide users with information about the overheating warning information through the mobile device MD and / or the display device DP.

[0071] like Figure 7 The diagram shown illustrates an overheating prediction system according to another embodiment of this disclosure. The thermal image capturing device 110 further includes a displacement detection unit 111, which can be implemented using gyroscopes, accelerometers, or other motion sensors to detect whether the thermal image capturing device 110 is colliding or moving. In some embodiments of this disclosure, when the thermal image capturing device 110 captures an image of an object under test, the displacement detection unit 111 continuously monitors the movement of the thermal image capturing device 110. If displacement is detected, a thermal image capturing device displacement signal is sent to the processing unit 120. Conversely, when the processing unit 120 receives the thermal image capturing device displacement signal, it issues a displacement warning message corresponding to the signal.

[0072] As described above regarding the technical means for issuing the overheating warning information, in some embodiments of this disclosure, the displacement warning information can be received by the output device 130 from the processing unit 120, and the output device 130 can issue the displacement warning information. The displacement warning information can be issued through various technical means such as flashing, sound, or displaying text or graphics on the screen. Alternatively, in other embodiments of this disclosure, the processing unit 120 can transmit the overheating information to the server SVR. After receiving the overheating warning information, the server SVR can issue the overheating warning information via SMS, email, or other means, providing users with information about the overheating warning information through the mobile device MD and / or the display device DP.

[0073] Although several embodiments have been described in detail above as examples, the overheating prediction system, overheating prediction method, and non-transitory computer-readable storage medium proposed in this disclosure can also be implemented in other hardware, software, storage media, or combinations thereof. Therefore, the scope of protection of this disclosure should not be limited to the specific implementations described in the embodiments of this disclosure, but should be determined by the appended claims.

[0074] It will be apparent to those skilled in the art to which this disclosure pertains that various modifications and variations can be made to the structure of this disclosure without departing from its scope or spirit. In view of the foregoing, the scope of protection of this disclosure also covers modifications and variations made within the scope of the appended patent applications.

Claims

1. A method for predicting overheating, characterized in that, The steps include: Multiple reference temperature distribution maps and multiple temperature distribution maps to be measured were obtained; A statistical model is established based on these reference temperature distribution maps to obtain a threshold temperature distribution map. The threshold temperature distribution map is used to filter these temperature distribution maps to be tested, and multiple potential abnormal temperature distribution maps are selected. Cluster analysis was performed on these reference temperature distribution maps and these potential abnormal temperature distribution maps to determine whether these potential abnormal temperature distribution maps belong to the same subset as these reference temperature distribution maps; When these potential abnormal temperature distribution maps and these reference temperature distribution maps do not belong to the same subset, these potential abnormal temperature distribution maps are marked as abnormal; as well as An overheating warning message is generated based on the temperature distribution map of these potential abnormalities marked as abnormal.

2. The overheating prediction method as described in claim 1, characterized in that, It further includes the following steps: Based on the pixel percentage distribution of each of these reference temperature distribution maps, determine whether there are any imaging abnormalities in each of these reference temperature distribution maps, and remove the reference temperature distribution maps that are determined to have imaging abnormalities. as well as Based on the pixel percentage distribution of each of these temperature distribution maps, determine whether there are any imaging abnormalities in each of these temperature distribution maps, and remove the temperature distribution maps that are determined to have imaging abnormalities.

3. The overheating prediction method as described in claim 1, characterized in that, It further includes the following steps: The overheat warning message may be transmitted to a mobile device or displayed on a display device.

4. The overheating prediction method as described in claim 3, characterized in that, The temperature distribution maps to be measured are captured by a thermal image capturing device, which is communicatively connected to a processing unit and includes a shift detection unit. When the thermal image capturing device shifts, the shift detection unit generates a shift signal. The overheating prediction method further includes the following steps: The shift signal is received by the processing unit; When the shift signal is received, the processing unit transmits a shift warning message to a mobile device or displays the shift warning message on a display device.

5. The overheating prediction method as described in claim 1, characterized in that, The threshold temperature distribution map contains multiple threshold temperatures corresponding to each pixel, which are calculated based on a threshold value of the quartiles in the statistical model. The step of screening these potential abnormal temperature distribution maps further includes: Compare the temperature information of each pixel in each of the temperature distribution maps to be tested to see if it exceeds a threshold temperature. If the number of pixels in one of the temperature distribution maps to be tested whose temperature information exceeds the threshold temperature is greater than a threshold, then it is marked as one of the potential abnormal temperature distribution maps.

6. An overheating prediction system, characterized in that, It includes: A thermal image capturing device configured to capture images of an object under test to obtain multiple temperature distribution maps; and A processing unit, communicatively connected to the thermal image capturing device, wherein the processing unit is configured to: The thermal image acquisition device acquires multiple reference temperature distribution maps and multiple temperature distribution maps to be measured. A statistical model is established based on these reference temperature distribution maps to obtain a threshold temperature distribution map. The threshold temperature distribution map is used to filter the temperature distribution maps to be tested, and at least one potential abnormal temperature distribution map is selected. If all the temperature distribution maps to be measured are marked as potential abnormal temperature distribution maps, then cluster analysis is performed on the reference temperature distribution maps and the potential abnormal temperature distribution maps to determine whether the potential abnormal temperature distribution maps and the reference temperature distribution maps belong to the same subset. When these potential abnormal temperature distribution maps and these reference temperature distribution maps do not belong to the same subset, these potential abnormal temperature distribution maps are marked as abnormal; as well as An overheating warning message is generated.

7. The overheating prediction system as described in claim 6, characterized in that, The processing unit is further configured as follows: Based on the pixel percentage distribution of each of these reference temperature distribution maps, determine whether there are any imaging abnormalities in each of these reference temperature distribution maps, and remove the reference temperature distribution maps that are determined to have imaging abnormalities. as well as Based on the pixel percentage distribution of each of these temperature distribution maps, determine whether there are any imaging abnormalities in each of these temperature distribution maps, and remove the temperature distribution maps that are determined to have imaging abnormalities.

8. The overheating prediction system as described in claim 6, characterized in that, The processing unit is further configured as follows: The overheat warning message may be transmitted to a mobile device or displayed on a display device.

9. The overheating prediction system as described in claim 8, characterized in that, The thermal image capturing device further includes a displacement detection unit, which generates a displacement signal when the thermal image capturing device is displaced. The overheating prediction system includes the following technical features: The shift signal is received by the processing unit; When the shift signal is received, the processing unit transmits a shift warning message to a mobile device or displays the shift warning message on a display device.

10. The overheating prediction system as described in claim 6, characterized in that, The threshold temperature distribution map contains multiple threshold temperatures corresponding to each pixel, which are calculated based on the quartiles in the statistical model. Furthermore, the processing unit is configured to: Compare the temperature of each pixel in each of the temperature distribution maps to be tested to see if it exceeds the corresponding threshold temperature. If the number of pixels in one of the temperature distribution maps to be tested that exceed the threshold temperature is greater than a threshold, then it is marked as the potential abnormal temperature distribution map.

11. A non-transitory computer-readable storage medium, characterized in that, It has at least one instruction stored thereon, which, when executed by a processing unit, performs an overheating prediction method, the overheating prediction method comprising the following steps: Multiple reference temperature distribution maps and multiple temperature distribution maps to be measured were obtained; A statistical model is established based on these reference temperature distribution maps to obtain a threshold temperature distribution map. The threshold temperature distribution map is used to filter the temperature distribution maps to be tested, and at least one potential abnormal temperature distribution map is selected. If all the temperature distribution maps to be measured are marked as potential abnormal temperature distribution maps, then cluster analysis is performed on the reference temperature distribution maps and the potential abnormal temperature distribution maps to determine whether the potential abnormal temperature distribution maps and the reference temperature distribution maps belong to the same subset. When these potential abnormal temperature distribution maps and these reference temperature distribution maps do not belong to the same subset, these potential abnormal temperature distribution maps are marked as abnormal; and An overheating warning message is generated.

Images