A method and device for detecting abnormality of thermal imaging of uncooled infrared detector
By dividing the equipment surface into sub-regions and establishing a comprehensive evaluation index, the problems of accuracy and timeliness in equipment anomaly detection in existing technologies have been solved, and efficient detection and early warning of equipment anomalies have been achieved.
Patent Information
- Application Number
- CN202411231051.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-09-04
AI Technical Summary
Existing uncooled infrared detector thermal imaging technology lacks fine segmentation and data correlation analysis in equipment anomaly detection, resulting in low accuracy in detecting local anomalies and the inability to provide timely warnings.
By dividing the equipment surface into sub-regions and combining the grayscale similarity index and the temperature change function slope evaluation index, a comprehensive evaluation index is established to achieve accurate and timely judgment of equipment anomalies.
It improves the accuracy and sensitivity of anomaly detection, enabling timely discovery and location of potential anomalies, and providing support for equipment failure prevention and timely maintenance.
Smart Images

Figure CN118730600B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of equipment anomaly detection technology, and in particular to a method and apparatus for detecting anomalies in thermal imaging using an uncooled infrared detector. Background Technology
[0002] Uncooled infrared detector thermal imaging is a high-tech temperature measurement and imaging technology. It utilizes an uncooled infrared detector as its core component, converting the infrared radiation energy emitted from an object's surface into an electrical signal. This signal is then amplified and processed by a signal processing circuit, ultimately presenting the signal as an image. This technology eliminates the need for cooling the detector, resulting in low cost, high reliability, and fast response. Uncooled infrared detector thermal imaging technology is widely used in industry, medicine, environmental monitoring, and military fields, enabling rapid and accurate detection of temperature distribution on object surfaces, achieving non-contact temperature measurement and thermal state analysis.
[0003] In the field of modern industrial production and equipment operation and maintenance, uncooled infrared detector thermal imaging technology is widely used as an efficient non-destructive testing method in equipment condition monitoring and fault diagnosis. This technology captures the infrared radiation emitted by the surface of the equipment and converts it into a thermal image, displaying the temperature distribution of the equipment surface in grayscale form, thereby realizing the visual monitoring of the equipment's operating status.
[0004] However, existing technologies still have some shortcomings when applied to equipment anomaly detection. Firstly, they often lack fine-grained segmentation of the equipment surface. In practical applications, they typically treat the overall temperature of the equipment as a single monitoring object, ignoring the differences in temperature variations between different areas. This leads to reduced accuracy in detecting localized anomalies. Secondly, existing technologies often lack effective integration and analysis methods when processing data from different monitoring periods. They usually simply analyze the data from each monitoring period independently, ignoring the correlation and trends between the data, which also reduces the accuracy of anomaly detection. Furthermore, existing technologies often only detect anomalies after they occur, failing to provide early warnings and resulting in untimely anomaly detection. Summary of the Invention
[0005] (a) Technical problems to be solved
[0006] To address the technical problems in the background art, this invention proposes a method and device for detecting anomalies in thermal imaging of an uncooled infrared detector. By dividing the surface of the device into sub-regions, and based on the gray-scale similarity index between pixels in each monitoring cycle and the slope evaluation index of the fitted temperature change function, a comprehensive evaluation index for each monitoring cycle is obtained, thereby determining whether the device is abnormal; thus increasing the accuracy and timeliness of the judgment of device anomalies.
[0007] (II) Technical Solution
[0008] To achieve the above objectives, the present invention provides the following technical solution:
[0009] A method for detecting anomalies in thermal imaging using an uncooled infrared detector, comprising:
[0010] Real-time thermal images of the equipment during normal operation are obtained from historical data; grayscale similarity index of adjacent pixels at component and runtime is used. The surface of the equipment is divided into sub-regions; a monitoring cycle is set, and the surface thermal image of the equipment under test is acquired at the end of each monitoring cycle while the equipment is operating, and preprocessing operations are performed.
[0011] During normal equipment operation, thermal images of each sub-region are acquired at the end of each monitoring cycle. Combined with data from previous monitoring cycles, temperature change functions are fitted to the temperatures of all pixels in each sub-region on the surface of the normal equipment to obtain the standard slope range of the temperature change function for each sub-region at the end of each monitoring cycle. At the end of each monitoring cycle, the temperature change function of the initial pixel points in several sub-regions on the surface of the device to be tested is fitted based on the temperature of the previous monitoring cycles, to obtain the slope of the temperature change function of each sub-region at the end of each monitoring cycle. ;
[0012] At the end of each monitoring cycle, calculate the grayscale similarity index between each pixel in each sub-region of the thermal image of the device under test during the current monitoring cycle and the initial pixel in the corresponding sub-region. ;Calculate the grayscale similarity index between all pixels in each sub-region and the initial pixel. Similarity index threshold for grayscale Combined, the first evaluation index for each monitoring period is obtained. The slope of the temperature change function for each sub-region With respect to the standard slope range of the corresponding temperature change function Combined, a second evaluation index is obtained for each monitoring period. The first assessment index With the second assessment index By combining these factors, a comprehensive evaluation index for each monitoring period can be obtained. ;
[0013] The comprehensive evaluation index for each monitoring cycle is combined with the comprehensive evaluation index for all previous monitoring cycles to form the first anomaly determination index for the device under test in that monitoring cycle. The first anomaly detection index The device is compared with the preset first and second judgment thresholds to determine whether the device under test is abnormal.
[0014] Specifically, real-time thermal images of the equipment during normal operation are obtained from historical data. After the operation is completed, the grayscale value of each pixel in the real-time thermal image is obtained one by one through image processing software, and a grayscale value change function is constructed.
[0015] By randomly selecting any pixel in the thermal image As the initial pixel, the pixel Place adjacent pixels into pixels In the set of candidate pixels to be merged, calculate the pixel points. Pixels in its candidate pixel set Gray similarity index between The expression is:
[0016] ;
[0017] in, Indicates the total duration of normal device operation; sets the grayscale similarity index threshold. ,like Then the pixel points From pixels Extract and place the candidate merged pixels into the pixel set. Merge the pixel set and combine the pixels Place adjacent pixels into pixels From the set of candidate pixels to be merged, continue to judge the pixels. Its grayscale similarity index with the remaining pixels in the candidate merge pixel set; if , to pixels From pixels The candidate pixels to be merged are extracted from the set but placed into the pixel. Merge the set of pixels and continue to judge the pixels. The grayscale similarity index between the pixels and the remaining pixels in the candidate pixel set.
[0018] Furthermore, when pixels When the set of candidate pixels to be merged is empty, the pixels will be... With pixels Pixels are extracted from the merged pixel set and merged to obtain a sub-region; the initial pixel is marked in each sub-region, that is, the first pixel in each sub-region;
[0019] To obtain all sub-regions within each component region, use... Indicates the first The first component area Sub-regions, using Indicates the first The first component area The first sub-region 1 pixel;
[0020] The weighting coefficient of each component is calculated by analyzing the number of times anomalies occur at each component in historical data. ,in This indicates the total number of parts on the surface of the device, and .
[0021] Specifically, at the end of each monitoring cycle, the grayscale values of all pixels in each sub-region during normal equipment operation are acquired and converted into specific temperature values. The temperature values of corresponding pixels in the current monitoring cycle and the previous several monitoring cycles are placed in the same Cartesian coordinate system. The temperature values of corresponding pixels in the current monitoring cycle and the previous several monitoring cycles are labeled in the Cartesian coordinate system. A linear regression method is used to fit the temperature change function image within the current monitoring cycle. The fitted temperature change function image is a straight line, and the slope of the fitted function is obtained. The maximum and minimum values of the slopes of the fitted temperature change function for all pixels in each sub-region are combined to obtain the standard slope range of the temperature change function for the current sub-region. ,in, Indicates the first One monitoring cycle.
[0022] Specifically, at the end of each monitoring cycle, the grayscale values of all pixels in each sub-region during the operation of the device under test are acquired and converted into specific temperature values. Combined with the temperature values of the corresponding initial pixels in the previous several monitoring cycles, a temperature change function is fitted to each initial pixel to obtain the slope of the fitted function at the initial pixels of each sub-region at the end of each monitoring cycle. And the slope of the fitted function at that pixel point. As the slope of the temperature change function for the corresponding sub-region.
[0023] Specifically, within each monitoring cycle, the grayscale value of each pixel in each sub-region of the thermal image of the device under test is monitored in real time. At the end of each monitoring cycle, the grayscale similarity index between each pixel in each sub-region of the thermal image of the device under test in the current monitoring cycle and the initial pixel in the corresponding sub-region is calculated. ;
[0024] The grayscale similarity index between all pixels in each sub-region and the initial pixel. Similarity index threshold for grayscale Combined, the first evaluation index for each monitoring period is obtained. The expression is:
[0025] ;
[0026] in, Indicates the first The weighting coefficient of each component Indicates the first The total number of sub-regions in each component region Indicates the first The first component area The total number of pixels in each sub-region; It means that when When the expression is evaluated, the result is: Otherwise, the result of the expression is .
[0027] Furthermore, the slope of the temperature change function for each sub-region With respect to the standard slope range of the corresponding temperature change function Combined, the slope evaluation index of each sub-region is obtained. The expression is:
[0028] ;
[0029] in, Representing an interval The minimum value in the interval is the minimum value in the interval. The lower limit; Representing an interval The maximum value in the interval is the value of the interval. The upper limit;
[0030] Slope evaluation index for all sub-regions Combined, a second evaluation index is obtained for each monitoring period. The expression is: ;
[0031] The first evaluation index With the second assessment index By combining these factors, a comprehensive evaluation index for each monitoring period can be obtained. The expression is: ;in , The first evaluation index is respectively With the second assessment index The weighting coefficients.
[0032] Specifically, the first anomaly judgment index of the device under test is calculated sequentially within each monitoring cycle. If the calculation is based on the first anomaly judgment index of the device under test within the first monitoring period; Then the anomaly detection index at this time Comprehensive evaluation index of the first monitoring period Equal, that is If the calculation is the first... The first anomaly judgment index of the device under test within each monitoring cycle ,and Then the first Comprehensive evaluation index for each monitoring period Combined with the comprehensive evaluation index of all previous monitoring cycles, the first anomaly determination index of the equipment under test in this monitoring cycle is formed. The expression is:
[0033] ;
[0034] in, The comprehensive evaluation index for the first monitoring period The weighting coefficients, To monitor the periodic coefficient, and ; For the first Comprehensive evaluation index for each monitoring cycle Weighting coefficients; The value can be: .
[0035] Furthermore, a first judgment threshold is set. and the second judgment threshold ,and , At the end of each monitoring cycle, the first anomaly judgment index of the device under test within the current monitoring cycle will be calculated. Compared with the first and second judgment thresholds, if If so, it is determined that the device under test has not shown any abnormality during the current monitoring period;
[0036] like Then the comprehensive evaluation index calculated for the next monitoring cycle needs to be used. As the second anomaly detection index in the current monitoring cycle And the second anomaly detection index Compared with the first judgment threshold In comparison, if If so, it is determined that no abnormality has occurred in the device under test during the current monitoring period; if If so, it is determined that the device under test has an abnormality within the current monitoring period;
[0037] like If so, it is determined that the device under test has an abnormality within the current monitoring period.
[0038] An anomaly detection device for thermal imaging using an uncooled infrared detector, comprising:
[0039] The sub-region division and preprocessing module divides the equipment surface into several component regions according to different types of components; it acquires real-time thermal images of the equipment during normal operation from historical data, and merges pixels with similar gray value changes by calculating the gray-scale similarity index between adjacent pixels to obtain several sub-regions in each component region; at the end of each monitoring cycle, it preprocesses the acquired thermal images of the equipment surface to be inspected and adds corresponding annotations based on the sub-region division results.
[0040] The slope fitting module acquires the grayscale values of all pixels in each sub-region during the operation of both normal and under-test devices at the end of each monitoring cycle, and converts them into specific temperature values. Combining the temperature data from the previous monitoring cycles, it fits a temperature change function to each pixel to obtain the slope of the fitted function. For normal devices, it combines the maximum and minimum values of the slopes of the fitted temperature change functions for all pixels in each sub-region to obtain the standard slope range of the temperature change function for the current sub-region. For the under-test device, it obtains the slope of the temperature change function for each sub-region at the end of each monitoring cycle.
[0041] The comprehensive evaluation module calculates the grayscale similarity index between each pixel in each sub-region of the thermal image of the device under test and the initial pixel in the corresponding sub-region at the end of each monitoring cycle, and obtains the first evaluation index by combining the grayscale similarity index threshold; it combines the slope of the temperature change function of each sub-region with the standard slope range of the corresponding temperature change function to obtain the second evaluation index for each monitoring cycle; and it combines the first evaluation index with the second evaluation index to obtain the comprehensive evaluation index for each monitoring cycle.
[0042] The anomaly detection module combines the comprehensive evaluation index of each monitoring cycle with the comprehensive evaluation index of all previous monitoring cycles to form the first anomaly detection index of the device under test in that monitoring cycle; it sets a first detection threshold and a second detection threshold, and compares the first anomaly detection index with these two thresholds to determine whether the device under test has an anomaly.
[0043] (III) Beneficial Effects
[0044] This invention provides a method and apparatus for detecting anomalies in thermal imaging using an uncooled infrared detector, which has the following advantages:
[0045] 1. Through precise sub-region division and real-time monitoring, it can accurately capture minute temperature changes on the surface of the device under test, thereby timely detecting and locating potential thermal imaging anomalies; this not only improves the accuracy and reliability of anomaly detection, but also helps to quickly locate problematic components, providing strong support for equipment maintenance and repair.
[0046] 2. By real-time monitoring and fitting of the temperature of pixels in several sub-regions on the surface of normal equipment and equipment under test, a standard slope range of temperature change in each sub-region was established, and the temperature change of the equipment under test was detected as abnormal. This improved the sensitivity and accuracy of anomaly detection, and enabled timely detection of thermal imaging anomalies on the surface of equipment, providing an effective means for equipment fault prevention and early diagnosis.
[0047] 3. By combining the grayscale similarity index and the slope of the temperature change function for evaluation, and calculating the comprehensive evaluation index for each monitoring cycle, a comprehensive and accurate assessment of equipment thermal imaging anomalies is achieved. This not only improves the sensitivity and accuracy of anomaly detection, but also reflects the temperature distribution and change trend on the equipment surface, providing strong support for equipment fault prediction and timely maintenance.
[0048] 4. By calculating the comprehensive evaluation index for each monitoring cycle and combining it with the comprehensive evaluation index of all previous monitoring cycles to form the first anomaly judgment index, dynamic and continuous monitoring of equipment thermal imaging anomalies is achieved. This not only improves the accuracy and reliability of anomaly detection, but also enables real-time monitoring and timely early warning of equipment status through the setting of monitoring cycles. This helps to quickly discover and locate potential thermal imaging anomalies, providing strong support for equipment fault prevention and timely maintenance. Attached Figure Description
[0049] Figure 1 This is a flowchart of the steps of the uncooled infrared detector thermal imaging anomaly detection method provided by the present invention;
[0050] Figure 2 This is a schematic diagram of the structure of the uncooled infrared detector thermal imaging anomaly detection device provided by the present invention. Detailed Implementation
[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0052] refer to Figure 1 This invention provides a method for detecting anomalies in thermal imaging using an uncooled infrared detector, comprising:
[0053] Step 1: Obtain real-time thermal images of the equipment during normal operation from historical data; based on the grayscale similarity index of adjacent pixels at the component and during operation. The surface of the equipment is divided into sub-regions; a monitoring cycle is set, and the surface thermal image of the equipment under test is acquired at the end of each monitoring cycle while the equipment is operating, and preprocessing operations are performed.
[0054] Step one includes the following steps:
[0055] Step 101: Obtain the design drawing of the device to be tested. Divide the surface of the device to be tested into several component areas according to different types of components. The several types of components on the surface of the device to be tested represent components that use different materials and have different functions. For example, for a device containing multiple motors and heat sinks, each motor and heat sink is divided into a component area. Calculate the weight coefficient of each component by the number of times anomalies occur at each component in historical data. ,in This indicates the total number of parts on the surface of the device, and ;
[0056] Step 102: Divide each component area into sub-regions. Specifically, obtain real-time thermal images of the equipment during normal operation from historical data. After the operation is completed, use image processing software to obtain the grayscale value of each pixel in the real-time thermal image and construct a grayscale value change function. The horizontal axis represents time and the vertical axis represents grayscale value. The thermal image is an image captured by an uncooled infrared detector that captures infrared radiation emitted from the surface of an object and converts it into a grayscale image, which is displayed in grayscale form. Different grayscale values represent different temperatures.
[0057] By calculating the gray-level similarity index between adjacent pixels Pixels with similar grayscale value changes (i.e., temperature changes) during normal equipment operation are merged to form several sub-regions; specifically, this is done by randomly selecting any pixel from the thermal image. As the initial pixel, the pixel Place adjacent pixels into pixels In the set of candidate pixels to be merged, calculate the pixel points. Merge pixels with other candidate pixels Gray similarity index between pixel For pixels The expression for any pixel in the set of candidate pixels to be merged is:
[0058] ;
[0059] in, This indicates the total duration of normal device operation, and sets a grayscale similarity index threshold. Compare the relationship between the gray-level similarity index and the gray-level similarity index threshold between two pixels. This means that the grayscale values of two pixels change similarly during normal device operation, and the pixels are considered similar. From pixels Extract and place the candidate merged pixels into the pixel set. Merge the pixel set and combine the pixels Place adjacent pixels into pixels From the set of candidate pixels to be merged, continue to judge the pixels. The grayscale similarity index between it and the remaining pixels in its candidate merge pixel set; if This indicates that the grayscale values of two pixels do not change similarly during normal device operation, and the pixels are... From pixels Extract the candidate pixels from the set but do not merge the pixels. Insert pixels Merge the set of pixels and continue to judge the pixels. The grayscale similarity index between the grayscale of the remaining pixels in its candidate merge pixel set;
[0060] When pixel When the set of candidate pixels to be merged is empty, the pixels will be... With pixels Pixels from the merged pixel set are extracted and merged to obtain a sub-region; several sub-regions are obtained in each component area on the normal device surface using the above method, and the initial pixel points are marked in each sub-region, such as the points in the sub-regions mentioned above. Among them, the gray-scale similarity index threshold The value is obtained by continuously simulating the merging results and using the simulated merging results in the operation of another normal device to verify whether the merging results are accurate each time.
[0061] To obtain all sub-regions within each component region, use... Indicates the first The first component area Sub-regions, using Indicates the first The first component area The first sub-region 100 pixels; the first pixel in each sub-region is the initial pixel;
[0062] Step 103: Set the monitoring cycle. At the end of each monitoring cycle, acquire the surface thermal image of the device under test through an uncooled infrared detector. After preprocessing, divide the surface thermal image of the device under test acquired in each monitoring cycle into corresponding sub-regions according to the sub-region division results in Step 102, and mark the initial pixel points in each sub-region.
[0063] The preprocessing includes noise removal, image correction, and grayscale adjustment. Noise removal aims to eliminate random errors or interference in the image, which may be caused by various factors such as sensor noise, transmission errors, and quantization errors. Specifically, noise removal is performed using filter methods. Image correction aims to eliminate distortion and falsification in the image, making the image more accurately reflect the actual scene. Grayscale adjustment aims to improve the contrast and sharpness of the image, and grayscale adjustment of each sub-region is performed using histogram equalization.
[0064] When using this method, refer to steps 101 to 103:
[0065] Through precise sub-region division and real-time monitoring, minute temperature changes on the surface of the device under test can be accurately captured, thereby timely detection and location of potential thermal imaging anomalies. This not only improves the accuracy and reliability of anomaly detection but also helps to quickly locate problematic components, providing strong support for equipment maintenance and repair.
[0066] Step 2: During normal equipment operation, acquire thermal images of each sub-region at the end of each monitoring cycle. Combine this with data from previous monitoring cycles to fit the temperature change function at all pixels in each sub-region of the normal equipment surface, obtaining the standard slope range of the temperature change function for each sub-region at the end of each monitoring cycle. At the end of each monitoring cycle, the temperature change function of the initial pixel points in several sub-regions on the surface of the device to be tested is fitted based on the temperature of the previous monitoring cycles, to obtain the slope of the temperature change function of each sub-region at the end of each monitoring cycle. ;
[0067] Step 201: At the end of each monitoring cycle, acquire the grayscale values of all pixels in each sub-region during normal equipment operation, and obtain the specific temperature values of these pixels using a lookup table. Combine this with the temperature values of the corresponding pixels in the previous monitoring cycles, and fit a temperature change function for each pixel. Specifically, place the temperature values of the corresponding pixels in the current monitoring cycle and the previous monitoring cycles into the same Cartesian coordinate system. The horizontal axis of the Cartesian coordinate system represents the number of monitoring cycles, and the vertical axis represents the temperature value. Mark the temperature values of the corresponding pixels in the current monitoring cycle and the previous monitoring cycles in the Cartesian coordinate system, and use linear regression to fit the temperature change function image within the current monitoring cycle. The fitted temperature change function image is a straight line, and the slope of the fitted function is obtained. Combine the maximum and minimum values of the slopes of the fitted temperature change functions for all pixels in each sub-region to obtain the standard slope range of the temperature change function for the current sub-region. ,in, Indicates the first One monitoring cycle;
[0068] Step 202: At the end of each monitoring cycle, acquire the grayscale values of all pixels in each sub-region during the operation of the device under test, and obtain the specific temperature values of these pixels using a lookup table. The operating environment of the device under test is consistent with that of the normal device, and the initial temperature of the device under test is the same as that of the normal device. Combining the temperature values of the corresponding initial pixels in the previous monitoring cycles, fit the temperature change function for each initial pixel based on the method in Step 201, and obtain the slope of the fitted function for the initial pixels in each sub-region at the end of each monitoring cycle. And the slope of the fitted function at that pixel point. As the slope of the temperature change function of the corresponding sub-region;
[0069] When using it, refer to steps 201 to 202:
[0070] By real-time monitoring and fitting of the temperature of pixels in several sub-regions on the surface of normal equipment and equipment under test, a standard slope range for temperature change in each sub-region was established, and the temperature change of the equipment under test was detected as abnormal. This improved the sensitivity and accuracy of anomaly detection, enabling timely detection of thermal imaging anomalies on the surface of equipment, and providing an effective means for equipment fault prevention and early diagnosis.
[0071] Step 3: At the end of each monitoring cycle, calculate the grayscale similarity index between each pixel in each sub-region of the thermal image of the device under test and the initial pixel in the corresponding sub-region during the current monitoring cycle. ;Calculate the grayscale similarity index between all pixels in each sub-region and the initial pixel. Similarity index threshold for grayscale Combined, the first evaluation index for each monitoring period is obtained. The slope of the temperature change function for each sub-region With respect to the standard slope range of the corresponding temperature change function Combined, a second evaluation index is obtained for each monitoring period. The first assessment index With the second assessment index By combining these factors, a comprehensive evaluation index for each monitoring period can be obtained. ;
[0072] Step 301: Within each monitoring cycle, monitor the grayscale value of each pixel in each sub-region of the thermal image of the device under test in real time. At the end of each monitoring cycle, calculate the grayscale similarity index between each pixel in each sub-region of the thermal image of the device under test and the initial pixel in the corresponding sub-region within the current monitoring cycle. The expression is: ,in, Indicates the first The time at the end of each monitoring cycle, the above formula represents the integration of the gray values of each pixel within the current monitoring cycle;
[0073] Step 302: Calculate the grayscale similarity index between all pixels in each sub-region and the initial pixel. Similarity index threshold for grayscale Combined, the first evaluation index for each monitoring period is obtained. The expression is:
[0074] ;
[0075] in, Indicates the first The weighting coefficient of each component Indicates the first The total number of sub-regions in each component region Indicates the first The first component area The total number of pixels in each sub-region; It means that when When the expression is evaluated, the result is: Otherwise, the result of the expression is ;
[0076] Step 303: Calculate the slope of the temperature change function for each sub-region. With respect to the standard slope range of the corresponding temperature change function Combined, the slope evaluation index of each sub-region is obtained. The expression is:
[0077] ;
[0078] in, Representing an interval The minimum value in the interval is the minimum value in the interval. The lower limit; Representing an interval The maximum value in the interval is the value of the interval. The upper limit;
[0079] Step 304: Evaluate the slope index of all sub-regions. Combined, a second evaluation index is obtained for each monitoring period. The expression is: ;
[0080] Step 305: The first evaluation index... With the second assessment index By combining these factors, a comprehensive evaluation index for each monitoring period can be obtained. The expression is: ;in , The first evaluation index is respectively With the second assessment index The weighting coefficients are obtained by analyzing the historical abnormal data of the equipment to be tested using the entropy weighting method. The entropy weighting method measures the degree of variation or the amount of information of each indicator by calculating the information entropy of each indicator, and then determines the weight of each indicator.
[0081] When using this method, refer to steps 301 to 305:
[0082] By combining the grayscale similarity index and the slope of the temperature change function for evaluation, and calculating the comprehensive evaluation index for each monitoring cycle, a comprehensive and accurate assessment of equipment thermal imaging anomalies is achieved. This not only improves the sensitivity and accuracy of anomaly detection, but also reflects the temperature distribution and change trend of the equipment surface, providing strong support for equipment fault prediction and timely maintenance.
[0083] Step 4: Combine the comprehensive evaluation index of each monitoring cycle with the comprehensive evaluation index of all previous monitoring cycles to form the first anomaly judgment index of the equipment under test in that monitoring cycle. The first anomaly detection index The device under test is compared with the preset first and second judgment thresholds to determine whether there is an abnormality.
[0084] Step 401: Calculate the first anomaly judgment index of the device under test in each monitoring cycle. If the calculation is based on the first anomaly judgment index of the device under test within the first monitoring period; ,Right now Then the anomaly detection index at this time Comprehensive evaluation index of the first monitoring period Equal, that is If the calculation is the first... The first anomaly judgment index of the device under test within each monitoring cycle ,and Then the first Comprehensive evaluation index for each monitoring period The comprehensive evaluation index compared with all previous monitoring periods ( This is combined to form the first anomaly judgment index for the equipment under test during the monitoring period. The expression is:
[0085] ;
[0086] in, The comprehensive evaluation index for the first monitoring period The weighting coefficients, The monitoring cycle coefficient describes the phenomenon that the longer the target equipment operates, the greater the probability of anomalies occurring. ; For the first Comprehensive evaluation index for each monitoring cycle The weighting coefficients; it is necessary to ensure that the sum of the weighting coefficients of the comprehensive evaluation index for all monitoring periods is 1, then The value of should be: .
[0087] Step 402: Set the first judgment threshold and the second judgment threshold ,and , At the end of each monitoring cycle, the first anomaly judgment index of the device under test within the current monitoring cycle will be calculated. Compared with the first and second judgment thresholds, if If the result is negative, it means that the device under test has not shown any abnormality during the current monitoring cycle, and the abnormality determination will continue for the next monitoring cycle.
[0088] like If the result is negative, it indicates that the device under test is suspected of having an abnormality during the current monitoring period, and the comprehensive evaluation index calculated for the next monitoring period needs to be adjusted. As the second anomaly detection index in the current monitoring cycle And the second anomaly detection index Compared with the first judgment threshold In comparison, if This indicates that the device under test has not shown any abnormalities during the current monitoring period; if If the signal is 0, it indicates that the device under test has malfunctioned during the current monitoring period, and a device malfunction command will be sent.
[0089] like If the signal is 0, it indicates that the device under test has malfunctioned during the current monitoring period, and a device malfunction command will be sent.
[0090] Once an abnormality is detected in the equipment under test, stop its operation.
[0091] When using this method, refer to steps 401 to 402:
[0092] By calculating the comprehensive evaluation index for each monitoring cycle and combining it with the comprehensive evaluation index of all previous monitoring cycles to form the first anomaly judgment index, dynamic and continuous monitoring of equipment thermal imaging anomalies is achieved. This not only improves the accuracy and reliability of anomaly detection, but also enables real-time monitoring and timely early warning of equipment status through the setting of monitoring cycles. This helps to quickly discover and locate potential thermal imaging anomalies, providing strong support for equipment fault prevention and timely maintenance.
[0093] refer to Figure 2 The present invention also provides an anomaly detection device for thermal imaging using an uncooled infrared detector, comprising:
[0094] The sub-region division and preprocessing module divides the equipment surface into several component regions according to different types of components; it acquires real-time thermal images of the equipment during normal operation from historical data, and merges pixels with similar gray value changes by calculating the gray-scale similarity index between adjacent pixels to obtain several sub-regions in each component region; at the end of each monitoring cycle, it preprocesses the acquired thermal images of the equipment surface to be inspected and adds corresponding annotations based on the sub-region division results.
[0095] The slope fitting module acquires the grayscale values of all pixels in each sub-region during the operation of both normal and under-test devices at the end of each monitoring cycle, and converts them into specific temperature values. Combining the temperature data from the previous monitoring cycles, it fits a temperature change function to each pixel to obtain the slope of the fitted function. For normal devices, it combines the maximum and minimum values of the slopes of the fitted temperature change functions for all pixels in each sub-region to obtain the standard slope range of the temperature change function for the current sub-region. For the under-test device, it obtains the slope of the temperature change function for each sub-region at the end of each monitoring cycle.
[0096] The comprehensive evaluation module calculates the grayscale similarity index between each pixel in each sub-region of the thermal image of the device under test and the initial pixel in the corresponding sub-region at the end of each monitoring cycle, and obtains the first evaluation index by combining the grayscale similarity index threshold; it combines the slope of the temperature change function of each sub-region with the standard slope range of the corresponding temperature change function to obtain the second evaluation index for each monitoring cycle; and it combines the first evaluation index with the second evaluation index to obtain the comprehensive evaluation index for each monitoring cycle.
[0097] The anomaly detection module combines the comprehensive evaluation index of each monitoring cycle with the comprehensive evaluation index of all previous monitoring cycles to form the first anomaly detection index of the device under test in that monitoring cycle; it sets a first detection threshold and a second detection threshold, and compares the first anomaly detection index with these two thresholds to determine whether the device under test has an anomaly.
[0098] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer storage medium or transmitted through a computer storage medium.
[0099] Computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. Computer storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).
[0100] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this invention should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for detecting abnormalities in thermal imaging of an uncooled infrared detector, characterized in that: include: Get real-time thermal images of normal equipment operation from historical data; based on the gray similarity index Gsi of adjacent pixels at components and operation (x,y) Divide the surface of the equipment into sub-areas; set a monitoring cycle, obtain the surface thermal image of the equipment to be tested at the end of each monitoring cycle when the equipment to be tested is in operation, and perform preprocessing operations; When the equipment is operating normally, the thermal images of each sub-area are obtained at the end of each monitoring cycle, and the temperature change function is fitted for the temperature of all pixels in several sub-areas on the surface of the normal equipment in combination with the previous monitoring cycles, and the standard slope range of the temperature change function of each sub-area at the end of each monitoring cycle is obtained. Specifically: At the end of each monitoring cycle, the grayscale values of all pixels in each sub-area during normal equipment operation are obtained and converted into specific temperature values; the temperature values of the corresponding pixels in the current monitoring cycle and the previous monitoring cycles are placed in the same plane rectangular coordinate system, the temperature values of the corresponding pixels in the current monitoring cycle and the previous monitoring cycles are marked in the plane rectangular coordinate system, and the linear regression method is used to fit the temperature change function image in the current monitoring cycle to obtain the slope of the fitting function; the maximum and minimum values of the slopes of the temperature change function obtained by fitting all pixels in each sub-area are combined to obtain the standard slope range of the temperature change function of the current sub-area Wherein, b represents the bth monitoring period; At the end of each monitoring cycle, the temperature change function is fitted at the temperature of the initial pixel points in several sub-areas of the surface of the detection device in combination with the previous monitoring cycles, and the slope of the temperature change function of each sub-area at the end of each monitoring cycle is obtained. Specifically: At the end of each monitoring cycle, the grayscale values of all pixels in each sub-area when the device to be detected is running are obtained and converted into specific temperature values; combined with the temperature values at the corresponding initial pixels in the previous monitoring cycles, the temperature change function is fitted to each initial pixel, and the slope of the fitting function at the initial pixel of each sub-area at the end of each monitoring cycle is obtained. The slope of the fitting function at the pixel point The slope as a function of the temperature change in the corresponding sub-region; At the end of each monitoring cycle, the grayscale similarity index of each pixel point in each sub-region of the thermal image of the device to be detected in the current monitoring cycle and the initial pixel point in the corresponding sub-region is calculated. The grayscale similarity index of all pixels in each sub-region and the initial pixel is calculated. Combined with the grayscale similarity index threshold Gsi0, the first evaluation index Fei of each monitoring cycle is obtained b , specifically: In each monitoring cycle, the grayscale value of each pixel in each sub-region of the thermal image of the device to be detected is monitored in real time. At the end of each monitoring cycle, the grayscale similarity index of each pixel in each sub-region of the thermal image of the device to be detected in the current monitoring cycle and the initial pixel in the corresponding sub-region is calculated. The grayscale similarity index of all pixels in each sub-region and the initial pixel is calculated. Combined with the grayscale similarity index threshold Gsi0, the first evaluation index Fei of each monitoring cycle is obtained b , the expression is: Among them, α A Indicates the weight coefficient of the Ath component, n A represents the total number of sub-areas in the Ath component area, represents the total number of pixels in the a-th sub-region of the A-th component region; Indicates that when , the result of the expression is 1; otherwise, the result of the expression is The slope of the temperature change function of each sub-area The corresponding temperature change function standard slope range Combined to obtain the second evaluation index Sei for each monitoring period b , specifically: The slope of the temperature change function of each sub-area The corresponding temperature change function standard slope range Combined to get the slope evaluation index of each sub-area The expression is: in, Representation interval The minimum value in the interval The lower limit of Representation interval The maximum value in the interval The upper limit of The slope evaluation index of all sub-areas Combined, the second evaluation index Sei for each monitoring period is obtained b , the expression is: The first evaluation index Fei b With the second evaluation index Sei b Combined, we get the comprehensive evaluation index Zei for each monitoring period b , specifically: The first evaluation index Fei b With the second evaluation index Sei b Combined, we get the comprehensive evaluation index Zei for each monitoring period b , the expression is: Zei b =β1*Fei b +β2*Sei b ; β1 and β2 are the first evaluation index Fei b With the second evaluation index Sei b The weight coefficient of The comprehensive evaluation index of each monitoring cycle is combined with the comprehensive evaluation index of all previous monitoring cycles to form the first abnormality determination index Fdi of the equipment to be detected in the monitoring cycle b , the first abnormality determination index Fdi b The detected value is compared with the preset first and second determination thresholds to determine whether the device to be detected is abnormal.
2. The method for detecting abnormalities in thermal imaging of an uncooled infrared detector according to claim 1, characterized in that: Obtain real-time thermal images of normal equipment operation from historical data. After the operation is completed, use image processing software to obtain the grayscale value of each pixel in the real-time thermal image one by one, and construct a grayscale value change function. Randomly select any pixel point P in the thermal image as the initial pixel point, put the adjacent pixels of pixel point P into the candidate merged pixel point set of pixel point P, and calculate the gray similarity index Gsi between pixel point P and pixel point Q in its candidate merged pixel point set. (P,Q) , the expression is: Where T represents the total time of normal operation of the device; set the gray similarity index threshold Gsi0, if Gsi (P,Q) ≤Gsi0, then take the pixel Q from the candidate merged pixel set of the pixel P and put it into the merged pixel set of the pixel P, and put the adjacent pixels of the pixel Q into the candidate merged pixel set of the pixel P, and continue to judge the grayscale similarity index of the pixel P and the remaining pixels in its candidate merged pixel set; if Gsi (P,Q) >Gsi0, take pixel Q out from the candidate merged pixel set of pixel P but do not put it into the merged pixel set of pixel P, and continue to determine the grayscale similarity index between pixel P and the remaining pixels in its candidate merged pixel set.
3. The method for detecting abnormalities in thermal imaging of an uncooled infrared detector according to claim 2, characterized in that: When the candidate merged pixel set of pixel point P is an empty set, pixel point P and the pixel points in the merged pixel set of pixel point P are taken out and merged to obtain a sub-region; an initial pixel point is marked in each sub-region, that is, the first pixel point in each sub-region; Get all sub-areas in each widget area, using A a To represent the ath sub-area in the Ath component area, use represents the i-th pixel in the a-th sub-area of the A-th component area; The weight coefficients {α1, α2, …, α n }, where n represents the total number of components on the device surface, and α1+α2+…+α n =1.
4. The method for detecting abnormalities in thermal imaging of an uncooled infrared detector according to claim 1, characterized in that: Calculate the first abnormality determination index Fdi of the equipment to be detected in each monitoring cycle in turn b ; If the first abnormality determination index Fdi1 of the device to be detected in the first monitoring cycle is calculated, then the abnormality determination index Fdi1 is equal to the comprehensive evaluation index Zei1 of the first monitoring cycle, that is, Fdi1 = Zei1; if the first abnormality determination index Fdi1 of the device to be detected in the bth monitoring cycle is calculated b , and b>1, then the comprehensive evaluation index Zei b Combined with the comprehensive evaluation index of all previous monitoring cycles, the first abnormality determination index Fdi of the equipment to be detected in this monitoring cycle is formed b , the expression is: Where γ is the weight coefficient of the comprehensive evaluation index Zei1 of the first monitoring cycle, o is the monitoring cycle coefficient, and o = 0.1; [γ + (b-1) * o] is the comprehensive evaluation index Zei1 of the bth monitoring cycle b The weight coefficient of ; the value of γ is:
5. The method for detecting abnormalities in thermal imaging of an uncooled infrared detector according to claim 4, characterized in that: Set the first determination threshold Fdt and the second determination threshold Sdt, and Fdt = 0.9, Sdt = 0.7; at the end of each monitoring cycle, calculate the first abnormality determination index Fdi of the device to be detected in the current monitoring cycle b Compared with the first and second judgment thresholds, if Fdt≤Fdi b ≤1, it is judged that there is no abnormality in the equipment to be detected during the current monitoring period; If Sdt ≤ Fdi b <Fdt, the comprehensive evaluation index Zei calculated in the next monitoring period b+1 shall be used as the second abnormal determination index Sdi for the current monitoring period b , and the second abnormal determination index Sdi b shall be compared with the first determination threshold Fdt. If Fdt ≤ Sdi b ≤ 1, it is determined that the device to be detected in the current monitoring period has no abnormality; if Sdi b <Fdt, it is determined that the device to be detected in the current monitoring period has an abnormality; If Fid b <Sdt, then it is determined that the device to be detected has an abnormality during the current monitoring period.
6. An uncooled infrared detector thermal imaging anomaly detection device, used to implement any of the methods described in claims 1 to 5, characterized in that: include: The sub-region division and pre-processing module divides the equipment surface into several component regions according to different types of components; obtains real-time thermal images of normal equipment operation from historical data, and merges pixels with similar grayscale value changes by calculating the grayscale similarity index between adjacent pixels to obtain several sub-regions in each component region; at the end of each monitoring cycle, pre-processes the acquired thermal image of the surface of the equipment to be tested, and makes corresponding annotations based on the sub-region division results; The slope fitting module obtains the grayscale values of all pixels in each sub-area when the normal device and the device to be detected are in operation at the end of each monitoring cycle, and converts them into specific temperature values; combines the temperature data of the previous monitoring cycles, fits the temperature change function to each pixel, and obtains the slope of the fitting function; for normal devices, combines the maximum and minimum values of the slopes of the temperature change function obtained by fitting all pixels in each sub-area, and obtains the standard slope range of the temperature change function of the current sub-area; for the device to be detected, the slope of the temperature change function of each sub-area at the end of each monitoring cycle is obtained; The comprehensive evaluation module calculates the grayscale similarity index of each pixel point in each sub-region of the thermal image of the device to be detected and the initial pixel point in the corresponding sub-region at the end of each monitoring cycle, and obtains the first evaluation index by combining the grayscale similarity index threshold; combines the slope of the temperature change function of each sub-region with the corresponding standard slope range of the temperature change function to obtain the second evaluation index of each monitoring cycle; combines the first evaluation index with the second evaluation index to obtain the comprehensive evaluation index of each monitoring cycle; The abnormality judgment module combines the comprehensive evaluation index of each monitoring cycle with the comprehensive evaluation index of all previous monitoring cycles to form a first abnormality judgment index of the equipment to be detected within the monitoring cycle; sets a first judgment threshold and a second judgment threshold, and compares the first abnormality judgment index with the two thresholds to determine whether the equipment to be detected has an abnormality.
Citation Information
Patent Citations
Infrared image processing method and device, computer device and readable storage medium
CN109447935A
Phase modifier fault monitoring method and system based on infrared image recognition
CN110910383A
Ground surface temperature anomaly detection and positioning method, device and system based on unmanned aerial vehicle
CN111307291A
Substation anomaly detection method and system
CN118505606A
