IGBT junction temperature measuring method and system based on image digital identification and thermal imaging
Through image digital recognition and thermal imaging technology, the IGBT surface temperature distribution is obtained in real time, solving the problem of measuring IGBT junction temperature fluctuations and temperature inhomogeneity, achieving efficient protection of IGBT devices and stable operation of power electronic systems.
Patent Information
- Application Number
- CN202510516686.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-23
- Publication Date
- 2025-07-18
AI Technical Summary
The prior art is difficult to accurately and efficiently obtain IGBT junction temperature fluctuations and chip surface temperature unevenness, resulting in accelerated aging of IGBT devices and instability of power electronic systems.
Using a method based on image digital recognition and thermal imaging, the thermal map and video of the IGBT surface are obtained through a thermal imager, the areas of interest of key points are segmented, image preprocessing and feature extraction are performed, and real-time temperature measurement and early warning are achieved through time series prediction.
Real-time and accurate measurement and prediction of IGBT surface temperature is realized, and the stability and reliability of power electronic systems are improved. It is suitable for IGBT devices of different models and operating conditions.
Smart Images
Figure CN120340015A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image recognition, and particularly relates to a method and system for measuring the IGBT junction temperature based on image digital recognition and thermal imaging. Background Art
[0002] Insulated Gate Bipolar Transistor (IGBT), as the leading device in medium and high-power power electronic equipment, is widely used in scenarios such as frequency converters and motor drive systems. For example, in a frequency converter, IGBT is the core power device, and in a motor drive system, IGBT is responsible for converting direct current into alternating current to drive the motor to operate. The reliability and safety of IGBT determine whether the entire power electronic system can operate efficiently.
[0003] Junction temperature refers to the temperature of the PN junction inside a semiconductor device, and it is one of the important indicators for measuring the operating state of IGBT. Excessive junction temperature will lead to a decline in IGBT performance, accelerate the aging process of the device, increase thermal stress, cause damage to the internal chip, and even trigger faults, thereby affecting the stability and reliability of the entire power electronic system. With the development of power electronic technology, while improving the power level and switching frequency of IGBT modules, their reliability is increasingly affected by the junction temperature, and the electromagnetic interference generated by themselves also affects other electronic systems. Therefore, through junction temperature detection, the operating state of IGBT can be monitored in real time to ensure its operation within a safe range, which is of great significance for preventing equipment failures, extending service life, and ensuring the efficient operation of power electronic systems.
[0004] The PN junction of IGBT is located inside the semiconductor chip, and it is very difficult to directly contact and measure its temperature. Even for an internal temperature sensor close to the junction, due to response time and thermal coupling problems, it cannot immediately reflect the actual temperature of the junction. Existing IGBT junction temperature detection methods mostly focus on the IGBT junction temperature prediction task based on machine learning neural networks, using existing datasets to train models and achieve temperature estimation and prediction. However, the junction temperature fluctuation of IGBT is extremely susceptible to the power loss generated during device operation and causes uneven temperature distribution on the internal chip surface. Existing methods measure the IGBT temperature through a temperature measuring instrument, and the obtained temperature needs to be monitored manually, with low efficiency and high labor costs. Summary of the Invention
[0005] The present invention aims to provide a method and system for measuring the IGBT junction temperature based on image digital recognition and thermal imaging, in order to solve the problem that existing data-driven machine learning prediction methods cannot accurately and efficiently obtain temperature data for the complexity of IGBT junction temperature fluctuations and the non-uniformity of chip surface temperature.
[0006] The basic solution provided by the present invention is: a thermal imaging IGBT junction temperature measurement system based on image digital recognition, and the system includes:
[0007] A heat map acquisition module, configured to set n key points representing the IGBT junction temperature, and to acquire in real time the heat map and heat map video of the IGBT surface, where the heat map shows the corresponding IGBT surface temperature distribution and the temperatures of the n key points;
[0008] A video frame splitting module, configured to read the heat map video and output a number of frames of images at a preset frame interval;
[0009] An interested region segmentation module, configured to preset and segment the interested region images of the n key points of the first frame image, and automatically segment and acquire the interested region images of the corresponding n key points of other frame images according to the interested region images of the n key points of the first frame image;
[0010] An IGBT temperature image preprocessing module, configured to preprocess all the interested region images corresponding to the frame images to obtain images recognizable by a preset device;
[0011] An IGBT temperature digital extraction module, configured to extract the digital features included in the preprocessed interested region images to obtain corresponding temperature digits, further configured to aggregate all the temperature digits to form a temperature digit set, and further configured to display the temperature digits and their set in a preset manner; the temperature digits represent the real-time temperature.
[0012] Based on the thermal imaging IGBT junction temperature measurement system based on image digital recognition, the present invention also provides a method for measuring the IGBT junction temperature based on image digital recognition and thermal imaging, and the method includes:
[0013] Set n key points representing the IGBT junction temperature, and acquire in real time the heat map and heat map video of the IGBT surface, where the heat map shows the corresponding IGBT surface temperature distribution and the temperatures of the n key points;
[0014] Read the heat map video and output a number of frames of images at a preset frame interval;
[0015] Preset and segment the interested region images of the n key points of the first frame image, and automatically segment and acquire the interested region images of the corresponding n key points of other frame images according to the interested region images of the n key points of the first frame image;
[0016] Preprocess all the interested region images corresponding to the frame images to obtain images recognizable by a preset device;
[0017] Extract the digital features contained in the preprocessed region of interest image to obtain the corresponding temperature digits, and also used to aggregate all temperature digits to form a temperature digit set, and also used to display the temperature digits and their set in a preset manner;
[0018] After preprocessing the real-time temperature, use the time series prediction method to predict the real-time temperature and output the predicted temperature;
[0019] Analyze and process the real-time temperature output by the IGBT temperature digit extraction module and the predicted temperature output by the temperature prediction module, and alarm for abnormal temperature.
[0020] The working principle and advantages of the present invention are as follows:
[0021] Aiming at the complexity of IGBT junction temperature fluctuation and the non-uniformity of chip surface temperature, this solution adopts a combination of hardware and software to complete the acquisition of non-uniform surface temperature and junction temperature data of IGBT, avoiding the disadvantages of data-driven machine learning prediction methods, and has the following advantages:
[0022] The obtained heat map and video can directly reflect the real-time temperature distribution on the IGBT surface at the corresponding moment and time period, can adapt to the complexity of the temperature change of the power electronic device itself, and there is no need to build a complex mathematical model.
[0023] Thermal images are conducive to realizing multi-point temperature measurement on the chip surface. The selection and quantity of n key points are reasonable, which can specifically reflect the important temperature points related to the IGBT junction temperature. By dividing the regions of interest of the n key points and realizing digital display, the uneven temperature on the chip surface can be obtained in real time and accurately. Specifically, due to the composition, structure, and processing technology of the material affecting the thermal conductivity of the material, for the material of the same IGBT device, its microscopic composition and structure may be uneven, resulting in differences in thermal conductivity in local parts of the device, further leading to uneven temperature distribution throughout the device. This solution summarizes and selects the highest temperature of the upper bridge arm of the IGBT, the highest temperature of the lower bridge arm of the IGBT, the average temperature of the module, and the estimated junction temperature of the module as key points for temperature extraction and digital display according to the operating conditions of the IGBT. The upper and lower bridge arms are the main conduction paths of the IGBT, bearing most of the load current. When the current flows through the upper and lower bridge arms, conduction losses and switching losses will occur, resulting in a significant increase in the IGBT temperature. At the same time, too large a temperature difference between the upper and lower bridge arms may lead to uneven current distribution, accelerating the aging of the high-temperature-side device and even causing device failure. Therefore, the upper and lower bridge arms are selected as one of the key points; the surface average temperature Ts is the temperature calculated by the thermal imager itself based on the heat map of the measured object surface, comprehensively reflecting the overall thermal load (overall thermal state) of the IGBT. By paying attention to the overall thermal load of the IGBT, misjudgment caused by only relying on local point temperatures can be avoided. Therefore, the average temperature display on the thermal imager is selected as one of the key (temperature extraction) points. The junction temperature is calculated based on the surface average temperature Ts of the IGBT and the thermal resistance R_th, and the estimated junction temperature T_j display of the thermal imager is selected as one of the key (temperature extraction) points, which can monitor the temperature safety of the IGBT.
[0024] In addition, this solution not only completes the measurement of the uneven temperature on the IGBT surface, but also makes predictions based on the measured real-time temperature, processes the missing values and outliers of the real-time temperature, and improves the accuracy of temperature measurement; at the same time, it performs abnormal alarms based on the real-time temperature and the predicted temperature, and uses the early warning results of the real-time temperature and the predicted temperature to verify each other, which can make the early warning more accurate and improve the protection ability of the system for the IGBT.
[0025] This solution is not limited by the training data of the IGBT junction temperature, and the method is complete. As the IGBT changes with time loss, load, and operating conditions, its performance is extremely vulnerable to specific conditions. This method directly obtains the IGBT junction temperature based on the thermal imager and reads the temperature numbers on the heat map based on the image digital recognition method, which is applicable to IGBTs under different conditions such as different models, operating conditions, and application scenarios. Description of the Drawings
[0026] Figure 1Schematic diagram of the structure of the IGBT junction temperature measurement system based on image digital recognition and thermal imaging provided by the embodiments of the present invention;
[0027] Figure 2 Schematic diagram of the digital display of the temperature of each region of interest in the heat image provided by the first embodiment of the present invention Figure 1 ;
[0028] Figure 3 Schematic diagram of the digital display of the temperature of each region of interest in the heat image provided by the first embodiment of the present invention Figure 2 . Specific embodiments
[0029] The following is a more detailed description through specific embodiments:
[0030] The embodiment is basically as shown in the appendix Figure 1 : An IGBT junction temperature measurement system based on image digital recognition and thermal imaging, the system includes:
[0031] A heat map acquisition module, configured to set n key points representing the IGBT junction temperature, and to acquire in real time the heat map and heat map video of the IGBT surface, where the heat map shows the temperature distribution of the corresponding IGBT surface and the temperatures of the n key points.
[0032] Specifically, the heat map acquisition module includes a thermal imager, and key settings and real-time acquisition of the heat map and heat map video are performed through the thermal imager. In this embodiment, the thermal imager uses a Fluke Ti401PRO model infrared thermal imager.
[0033] The n key points include the highest temperature of the IGBT upper bridge arm, the highest temperature of the IGBT lower bridge arm, the module average temperature, and the module estimated junction temperature. (1) The basis for selecting the upper and lower bridge arms as key points: The upper and lower bridge arms are the main conduction paths of the IGBT, carrying most of the load current. When the current flows through the upper and lower bridge arms, conduction losses and switching losses will occur, resulting in a significant increase in the IGBT temperature. At the same time, too large a temperature difference between the upper and lower bridge arms may lead to uneven current distribution, accelerating the aging of high-temperature measuring devices and even causing device failure. Therefore, the upper and lower bridge arms are selected as one of the key points. (2) The surface average temperature Ts is the temperature calculated by the thermal imager itself based on the heat map of the surface of the measured object, comprehensively reflecting the overall thermal load (overall thermal state) of the IGBT. By paying attention to the overall thermal load of the IGBT, misjudgment caused by only relying on local point temperatures is avoided. Therefore, the average temperature display on the thermal imager is selected as one of the key (temperature extraction) points. (3) Estimation of the thermal imager junction temperature Tj: Calculate the junction temperature according to the IGBT surface average temperature Ts and the thermal resistance R_th:
[0034] T_j = T_s + R_{th} * P_loss
[0035] Among them, P_loss is the total loss of the IGBT (conduction loss + switching loss), and R_{th} is the thermal resistance from the junction to the case, which is obtained from the device specification sheet.
[0036] The video frame segmentation module is used to read the heat map video and output a number of frame images at a preset frame interval.
[0037] Specifically, in this embodiment, in order to obtain the temperature in real time, the preset frame interval is 1 frame. Here, frame segmentation means converting the video into a series of frame images heatmap j , j = 1, 2,... n.
[0038] The region of interest segmentation module is used to preset and segment the region of interest images of n key points in the first frame image, and automatically segment and obtain the region of interest images of the corresponding n key points in other frame images according to the region of interest images of n key points in the first frame image;
[0039] Specifically, the positions where the temperature data of n key points are displayed on each heat map remain unchanged.
[0040] For the preset segmentation, the region of interest image of the i-th key point on the first frame image heatmap1 is manually segmented using opencv During segmentation, the region occupied by the temperature display of the key point is mainly used as the region of interest, that is, the displayed data is framed.
[0041] ROI tem is a rectangular area, i = 1, 2,... n, and the ROI tem coordinate values (x, y) of the upper left corner and the width and height (w, h) of the ROI tem are used to automatically segment and obtain the ROI j on heatmap tem images, denoted as
[0042] The IGBT temperature image preprocessing module includes a grayscale conversion module, a black and white inversion module, and a binary conversion module; specifically:
[0043] The grayscale conversion module is used to convert the region of interest image into a grayscale image; that is:
[0044] Convert the original into a grayscale image
[0045]
[0046] Among them, is the region of interest image corresponding to the i-th key point of the j-th heat image; is the corresponding grayscale image; gray_trans(·) means traversing all pixels in · and calculating the grayscale value according to the following formula:
[0047]
[0048] where R, G, and B represent the intensity values of the red, green, and blue components respectively.
[0049] The black and white inversion module is used to convert the black and white pixels in the grayscale image to obtain a digital image with a white background and a black target; that is:
[0050] Convert the black and white pixels in to obtain a digital image with a white background and a black target That is:
[0051]
[0052] where w2b_trans(·) means traversing all pixels in · and calculating the new pixel value Dst according to the following formula:
[0053] Dst = 255 - Gray
[0054] The binary conversion module is used to convert the black and white pixels in the digital image to obtain a binary image, that is:
[0055] Convert the black and white pixels in to obtain the binary image of That is:
[0056]
[0057] where bin_trans(·) means traversing all pixels in · and calculating the new pixel value P according to the following formula:
[0058] if Dst > T, P = 255
[0059] if Dst < T, P = 0
[0060] where T is the set threshold.
[0061] The IGBT temperature digital extraction module is used to extract the digital features contained in the preprocessed region of interest image to obtain the corresponding temperature digits, and is also used to aggregate all temperature digits to form a temperature digit set, and is also used to display the temperature digits and their set in a preset manner.
[0062] It includes a feature extraction module, a character recognition module, and a post-processing module, which are used to extract the digital features contained in
[0063] The feature extraction module is used to extract features such as the contour, shape, and texture of characters;
[0064] The character recognition module is used to send the extracted features into a pre-trained convolutional neural network for the recognition of characters in
[0065] The post-processing module is used to optimize character recognition, and the optimization methods include text splicing, correction, and language model.
[0066] This system also includes a temperature prediction module, which includes a data preprocessing module and a time series prediction module. The data preprocessing module in the temperature prediction module mainly processes missing values and outliers according to the possible missing values and missing decimal points of the real-time temperature extracted by the IGBT temperature digital extraction module. For the missing value at the missing moment t or the obviously inaccurate temperature value T i (such as the previous and next temperatures of a certain extracted temperature are 71.0 and 71.2 respectively, and the currently extracted temperature is 711), the following processing is carried out:
[0067]
[0068] where t prev represents the previous moment, and t next represents the next moment.
[0069] Construct time series features for the preprocessed data. For each temperature variable T, take the past p steps (p can be set manually), and the constructed lag features are:
[0070] T lag =[T(t - 1), T(t - 9)…T(t - p)]∈R p×5
[0071] The calculation process of the time series prediction module is as follows:
[0072] Update gate Z t : Z t =σ(W z [x t , h t-1 )
[0073] Reset gate r t : r t =σ(W r [x t , h t-1 )
[0074] Candidate hidden state
[0075] Hidden state h t : h t = (1 - Z t ) ⊙ h t-1 + Z t ⊙ h t
[0076] where, x t is the current input feature; h t-1 is the hidden state of the previous time step; W z is the learnable weight matrix of the update gate, mapping the input and the hidden state to the update gate; W r is the learnable weight matrix of the reset gate; W h is the learnable weight matrix of the candidate state; σ(·) represents the Sigmoid activation function; tanh(·) is the hyperbolic tangent activation function; ⊙ represents the Hadamard product.
[0077] The temperature anomaly alarm module is used to monitor the real-time or predicted temperature and alarm for the abnormal temperature. The output of the temperature anomaly alarm module includes two parts. Among them, the real-time temperature from the IGBT temperature digital extraction module outputs the real-time temperature anomaly alarm after passing through the temperature anomaly alarm module, and the predicted temperature from the temperature prediction module outputs the predicted temperature anomaly alarm after passing through the temperature anomaly alarm module.
[0078] Specifically, the input real-time temperature time series and predicted temperature time series T ts are judged as follows:
[0079] 0 ≤ t ≤ T max , it is judged that the temperature at the current time step t belongs to the safe temperature, and continue to judge If it is judged that the temperature at the current time step t is abnormal, an alarm is issued.
[0080] In actual use, the thermal imager can be fixed on the surface of the power electronic device. The heat map of the device can be obtained in real time through this system. A series of video images of the heat map are obtained through fixed-frame segmentation, and the real-time temperature of each key point included in all the images is obtained. The ROI of each key point is obtained, as Figure 2 shown. A single heat image shows the temperature corresponding to the key point and the estimated junction temperature.
[0081] Such as Figure 3As shown, all the data obtained from all the thermal images are aggregated to form a data set, and the data set is stored in an excel table for subsequent analysis.
[0082] Applying the above-mentioned IGBT junction temperature measurement system based on image digital recognition and thermal imaging, this solution also proposes an IGBT junction temperature measurement method based on image digital recognition and thermal imaging. The method includes:
[0083] Set n key points representing the IGBT junction temperature, and real-time obtain the thermal image and thermal image video of the IGBT surface. The thermal image shows the corresponding IGBT surface temperature distribution and the temperatures of the n key points.
[0084] Read the thermal image video and output several frames of images at a preset frame interval.
[0085] Preset the region of interest (ROI) image for segmenting the n key points of the first frame image, and automatically segment and obtain the ROI images of the corresponding n key points of other frame images according to the ROI image of the n key points of the first frame image.
[0086] Preprocess all the ROI images corresponding to the frame images to obtain images that can be recognized by the preset device.
[0087] Extract the digital features contained in the preprocessed ROI images to obtain the corresponding temperature numbers, also used to aggregate all the temperature numbers to form a temperature number set, and also used to display the temperature numbers and their set in a preset manner.
[0088] After preprocessing the real-time temperature, use the time series prediction method to predict the temperature and output the predicted temperature.
[0089] Analyze and process the real-time temperature output by the IGBT temperature digital extraction module and the predicted temperature output by the temperature prediction module, and alarm for abnormal temperature.
[0090] It can be understood that the above method can achieve the same process as the above system and achieve the same effect, which will not be elaborated here.
[0091] The IGBT junction temperature measurement method and system based on image digital recognition and thermal imaging provided in this embodiment, aiming at the complexity of IGBT junction temperature fluctuations and the non-uniformity of the chip surface temperature, this solution adopts a combination of hardware and software to complete the acquisition of non-uniform surface temperature and junction temperature data of the IGBT, avoiding the disadvantages of data-driven machine learning prediction methods. It realizes the real-time acquisition of the thermal image video of the IGBT surface with the temperatures of multiple key points based on the thermal imager; frames the video to obtain a series of frame images, and manually segment the rectangular area of the temperature key point i based on opencv to obtain the region of interest. For a series of Preprocess the image to obtain a computer-recognizable image; obtain the temperature displayed in the image through digital extraction methods.
[0092] The obtained heat map video can directly reflect the real-time temperature distribution on the surface of the IGBT at the corresponding moment, adapt to the complexity of the temperature change of the power electronic device itself, and there is no need to construct a complex mathematical model. The thermal image is conducive to realizing multi-point temperature measurement on the chip surface. The selection and quantity of n key points (the highest temperature of the upper bridge arm of the IGBT, the highest temperature of the lower bridge arm of the IGBT, the average temperature of the module, and the estimated junction temperature of the module) are reasonable, which can specifically reflect the important temperature points related to the IGBT junction temperature. By segmenting the regions of interest of the n key points and realizing digital display, the uneven temperature on the chip surface can be obtained in real time and accurately.
[0093] In addition, this solution not only completes the measurement of the uneven temperature on the surface of the IGBT, but also makes predictions based on the measured real-time temperature, processes the missing values and outliers of the real-time temperature, and improves the accuracy of temperature measurement; at the same time, it performs abnormal alarms based on the real-time temperature and the predicted temperature, and uses the warning results of the real-time temperature and the predicted temperature to verify each other, which can make the warning more accurate and improve the protection ability of the system for the IGBT.
[0094] This solution is not limited to the training data of the IGBT junction temperature, and the method is complete. As the IGBT is affected by time loss, load, and operating conditions, its performance is extremely vulnerable to specific conditions. This method directly obtains the IGBT junction temperature based on a thermal imager and reads the temperature numbers on the heat map based on the image digital recognition method, and is applicable to IGBTs under different conditions such as different models, operating conditions, and application scenarios.
[0095] The above are only the embodiments of the present invention. Common general knowledge such as specific structures and characteristics known in the solution is not described in detail here. Those of ordinary skill in the art know all the common general knowledge in the technical field to which the invention belongs before the application date or the priority date, can know all the existing technologies in this field, and have the ability to apply the conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, improve and implement this solution in combination with their own abilities. Some typical well-known structures or well-known methods should not become an obstacle for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent.
Claims
1. An image digital recognition and thermal imaging IGBT junction temperature measurement system, characterized in that, The system includes: A heat map acquisition module, configured to set n key points representing the IGBT junction temperature, and acquire in real time a heat map and a heat map video of the IGBT surface, where the heat map shows the corresponding IGBT surface temperature distribution and the temperatures of the n key points; A video frame division module, configured to read the heat map video and output a number of frame images at a preset frame interval; A region of interest segmentation module, configured to preset and segment the region of interest images of the n key points of the first frame image, and automatically segment and acquire the region of interest images of the corresponding n key points of other frame images according to the region of interest images of the n key points of the first frame image; An IGBT temperature image preprocessing module, configured to preprocess all the region of interest images corresponding to the frame images to obtain images recognizable by a preset device; An IGBT temperature digital extraction module, configured to extract the digital features included in the preprocessed region of interest images to obtain corresponding temperature digits, further aggregate all the temperature digits to form a temperature digit set, and further display the temperature digits and their set in a preset manner; the temperature digits represent the real-time temperature.
2. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, wherein It further includes a temperature prediction module, configured to receive the real-time temperature output by the IGBT temperature digital extraction module, perform preprocessing using a data preprocessing module, and then predict the real-time temperature using a time series prediction module to output a predicted temperature; A temperature anomaly alarm module, configured to receive the real-time temperature output by the IGBT temperature digital extraction module and the predicted temperature output by the temperature prediction module, analyze and process them, and alarm for abnormal temperatures.
3. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, wherein The n key points include the highest temperature of the IGBT upper bridge arm, the highest temperature of the IGBT lower bridge arm, the module average temperature, and the module estimated junction temperature.
4. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 3, wherein The module estimated junction temperature T_j is calculated using the following formula: T_j = T_s + R_{th} * P_loss where, T_s is the module average temperature; R_{th} is the thermal resistance from the junction to the case; P_loss is the total loss of the IGBT.
5. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, characterized in that, The heat map acquisition module includes a thermal imager.
6. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, characterized in that The preset segmentation is to manually segment using opencv, and take the area where the key point temperature is displayed as the region of interest.
7. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, characterized in that, The region of interest is a rectangular region. The coordinate values (x, y) of the upper left corner of each region of interest of the first frame image and the width and height (w, h) of the region of interest are extracted, and the corresponding region of interest images of other frame images are automatically segmented and acquired according to (x, y, w, h).
8. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, characterized in that, The IGBT temperature image preprocessing module includes a grayscale image conversion module, a black and white inversion module, and a binary conversion module, configured to convert the region of interest image into a grayscale image, and then perform black and white pixel conversion to obtain a binary image; The IGBT temperature digital extraction module includes a feature extraction module, a character recognition module, and a post-processing module, configured to extract the features of the characters in the preprocessed region of interest images; input the extracted features into a pre-trained convolutional neural network for character recognition and optimization.
9. The IGBT junction temperature measurement system based on image digital recognition and thermal imaging according to claim 1, characterized in that, Aggregate all the data obtained from all the frame images to form a data set and store it in an excel table.
10. A method for measuring the IGBT junction temperature based on image digital recognition and thermal imaging, characterized in that The method includes: Set n key points representing the IGBT junction temperature, and obtain the heat map and heat map video of the IGBT surface in real time, where the heat map shows the corresponding IGBT surface temperature distribution and the temperatures of the n key points; Read the heat map video and output several frames of images at a preset frame interval; Preset the region of interest images of the n key points of the first frame image, and automatically segment and obtain the region of interest images of the corresponding n key points of other frame images according to the region of interest images of the n key points of the first frame image; Preprocess all the region of interest images corresponding to the frame images to obtain images recognizable by a preset device; Extract the digital features contained in the preprocessed region of interest images to obtain the corresponding temperature numbers, which are also used to aggregate all the temperature numbers to form a temperature number set, and are also used to display the temperature numbers and their set in a preset manner; the temperature numbers represent the real-time temperature; After preprocessing the real-time temperature, use the time series prediction method to predict the real-time temperature and output the predicted temperature; Analyze and process the real-time temperature output by the IGBT temperature digital extraction module and the predicted temperature output by the temperature prediction module, and alarm for abnormal temperatures.