Intelligent fault diagnosis method, system and equipment for chemical production equipment and medium

Abstract

The invention provides an intelligent fault diagnosis method and system for chemical production equipment, equipment and a medium, and the method comprises the steps: carrying out the multi-scale convolution of a thermal infrared image of the chemical production equipment under each operation condition, extracting the mutual information difference of image textures between different scales, and determining the hot spot interference characteristics of thermal fault detection under each operation condition; according to the thermal fault features of the chemical production equipment, thermal response element association is carried out on the thermal response areas of the chemical production equipment in different operation conditions, and thermal response association coefficients of the thermal fault features of the chemical production equipment in different operation conditions are obtained; and performing fuzzy class correlation evaluation on the structural dependence of the thermal response area of the equipment among different operation conditions through all the hot spot interference characteristics and the thermal response correlation coefficients to obtain a class correlation evaluation matrix, and positioning the thermal fault area of the chemical production equipment based on the class correlation evaluation matrix. By adopting the scheme of the invention, the structure dependence correlation positioning of the thermal fault area under multiple operation working conditions can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image data processing, more particularly, the present application relates to a chemical production equipment fault intelligent diagnosis method, system, device and medium. BACKGROUND

[0002] Chemical production equipment often faces complex working condition changes and environmental interference during operation, resulting in thermal abnormalities on the inside or surface of the equipment, and then reflecting potential fault states. Thermal infrared imaging technology is widely used in thermal fault detection of chemical equipment due to its non-contact and real-time monitoring advantages. By capturing infrared radiation information on the surface of the equipment, thermal infrared images can reflect the local temperature distribution characteristics of the equipment, providing intuitive and rich visual data for fault diagnosis. Image processing technology is a key means for analyzing complex texture information in thermal infrared images and identifying abnormal hot spots, providing important support for the accuracy and robustness of intelligent fault diagnosis.

[0003] Existing thermal fault diagnosis techniques generally rely on thermal infrared image features under a single working condition, which is difficult to fully consider the complex changes in thermal response of the equipment under multiple operating conditions, resulting in scattered fault feature expression and lack of cross-condition consistency analysis. In addition, existing technical methods usually ignore the structural dependency relationship between thermal response regions under different working conditions, and cannot effectively integrate thermal response information under multiple working conditions, which significantly affects the accuracy and stability of thermal fault positioning. Especially in the presence of hot spot interference and environmental noise, traditional methods are difficult to accurately distinguish between real fault hot spots and interference factors, resulting in a significant increase in misdiagnosis rate and a decrease in the reliability of the diagnosis results. Therefore, how to realize the structural dependency correlation positioning of the thermal fault region under multiple operating conditions has become a difficult problem in the industry. SUMMARY

[0004] The present application provides a chemical production equipment fault intelligent diagnosis method, system, device and medium, which can realize the structural dependency correlation positioning of the thermal fault region under multiple operating conditions.

[0005] In a first aspect, the present application provides a chemical production equipment fault intelligent diagnosis method, comprising the following steps: Responding to thermal fault detection of the chemical production equipment under different operating conditions, collecting thermal infrared images of the chemical production equipment under each operating condition by a thermal fault detection device; Performing multi-scale convolution on the thermal infrared images of the chemical production equipment under each operating condition to extract mutual information differences of image textures between different scales, and then determining hot spot interference features of the thermal fault detection under each operating condition according to all mutual information differences and structural information of the chemical production equipment. extracting thermal fault features of the chemical production equipment under each operating condition from all thermal infrared images, and correlating thermal response elements of the thermal response regions of the chemical production equipment between different operating conditions according to all thermal fault features to obtain thermal response correlation coefficients of the thermal fault features of the chemical production equipment between different operating conditions; performing fuzzy class correlation evaluation on structure dependency of the thermal response regions of the chemical production equipment between different operating conditions through the thermal spot interference features of the thermal fault detection under each operating condition and all thermal response correlation coefficients to obtain a class correlation evaluation matrix, and further correlating and positioning the thermal fault regions of the chemical production equipment based on the class correlation evaluation matrix.

[0006] Preferably, the thermal infrared images of the chemical production equipment under each operating condition are subjected to multi-scale convolution to extract mutual information differences of image textures between different scales, which specifically include: For each operating condition, the thermal infrared images are subjected to hierarchical convolution processing through convolution kernels of different scales to extract texture feature maps under corresponding convolution scales; texture contrast analysis is performed on the texture feature maps under any two scales to determine mutual information of textures between them; all mutual information of textures is determined to determine mutual information differences of image textures between different scales of the corresponding thermal infrared images under the operating condition, and further to obtain mutual information differences of image textures between different scales of the corresponding thermal infrared images under each operating condition.

[0007] Preferably, the thermal spot interference features of the thermal fault detection under each operating condition are determined according to all mutual information differences and structure information of the chemical production equipment, which specifically include: For each operating condition, texture change regions presenting significant information gain in the thermal infrared images are determined based on mutual information differences of image textures between different scales; non-fault thermal spot regions caused by equipment structure occlusion, structural gaps and material thermal capacity differences are located through the texture change regions combined with the structure information of the chemical production equipment; thermal spot interference features of the thermal fault detection under the operating condition are determined according to the non-fault thermal spot regions, and further to obtain the thermal spot interference features of the thermal fault detection under each operating condition.

[0008] Preferably, the thermal fault features of the chemical production equipment under each operating condition are extracted from all thermal infrared images, which specifically include: For each operating condition, temperature distribution analysis is performed on the thermal infrared images to identify abnormal temperature regions higher than the ambient thermal distribution pattern; geometric features, gray distribution features and boundary features of the abnormal temperature regions are extracted to further form a thermal anomaly feature set; The multiple features in the thermal anomaly feature set are integrated and classified by an image clustering method, and non-fault regions are removed, so as to obtain a thermal fault region feature with structural response anomaly; The thermal fault region feature is taken as a thermal fault feature of the chemical production equipment under the running condition, and then a thermal fault feature of the chemical production equipment under each running condition is obtained.

[0009] Preferably, the thermal response element correlation of the thermal response region of the chemical production equipment between different running conditions is determined according to all the thermal fault features, and the thermal response correlation coefficient of the thermal fault features of the chemical production equipment between different running conditions specifically includes: For each two running conditions, the thermal fault features extracted under the two running conditions are position-aligned according to the equipment spatial structure, and then the correlation relationship of the thermal fault features in each thermal response region is determined; The difference features of the thermal response elements of the same thermal response region under the two conditions are extracted; The thermal response correlation coefficient of the thermal fault features of the chemical production equipment between the two running conditions is determined according to the correlation relationship of the thermal fault features and the difference features of the thermal response elements, and then the thermal response correlation coefficient of the thermal fault features of the chemical production equipment between each two running conditions is obtained.

[0010] Preferably, the structural dependency of the thermal response region of the chemical production equipment between different running conditions is evaluated by fuzzy class correlation according to the thermal spot interference features detected by the thermal fault detection under each running condition and all the thermal response correlation coefficients, and the class correlation evaluation matrix specifically includes: The thermal fault information loss in the thermal fault feature extraction process of the chemical production equipment under each running condition is determined according to the thermal spot interference features detected by the thermal fault detection under each running condition; A fuzzy reasoning model is constructed based on all the thermal fault information losses and all the thermal response correlation coefficients; The structural dependency of the thermal response region of the chemical production equipment is calculated by fuzzy class membership through the fuzzy reasoning model, and the fuzzy dependency relationship of each thermal response region under different running conditions is obtained. The fuzzy dependency relationships of all the thermal response regions are aggregated, and the class correlation evaluation matrix is obtained.

[0011] Preferably, the thermal fault detection equipment specifically is a thermal infrared imager.

[0012] In a second aspect, the present application provides a chemical production equipment fault intelligent diagnosis system, which comprises: A collection module is configured to respond to thermal fault detection of the chemical production equipment under different running conditions, and collect thermal infrared images of the chemical production equipment under each running condition by a thermal fault detection equipment. The processing module is used to perform multi-scale convolution on the thermal infrared images of chemical production equipment under each operating condition, extract the mutual information differences of image texture between different scales, and then determine the hot spot interference features of thermal fault detection under each operating condition based on all mutual information differences and the structural information of chemical production equipment. The processing module is also used to extract thermal fault features of chemical production equipment under various operating conditions from all thermal infrared images, and to correlate thermal response elements of the thermal response regions of chemical production equipment under different operating conditions based on all thermal fault features, so as to obtain thermal response correlation coefficients of thermal fault features of chemical production equipment under different operating conditions. The execution module is used to perform fuzzy class correlation assessment on the structural dependence of the thermal response region of chemical production equipment under different operating conditions by using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients, to obtain a class correlation assessment matrix, and then to locate the thermal fault region of chemical production equipment based on the class correlation assessment matrix.

[0013] Thirdly, this application provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to acquire the code and execute the above-described intelligent fault diagnosis method for chemical production equipment.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-mentioned intelligent fault diagnosis method for chemical production equipment.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: In this embodiment, thermal fault detection of chemical production equipment under different operating conditions is performed by acquiring thermal infrared images of the equipment under each operating condition using a thermal fault detection device. Multi-scale convolution is applied to the thermal infrared images of the equipment under each operating condition to extract the mutual information differences in image texture between different scales. Then, based on all mutual information differences and the structural information of the chemical production equipment, hot spot interference features for thermal fault detection under each operating condition are determined. Thermal fault features for each operating condition are extracted from all thermal infrared images. Based on all thermal fault features, thermal response elements are associated with the thermal response regions of the equipment under different operating conditions to obtain thermal response correlation coefficients for thermal fault features under different operating conditions. Fuzzy class correlation evaluation is performed on the structural dependence of the thermal response regions of the equipment under different operating conditions using the hot spot interference features and all thermal response correlation coefficients for thermal fault detection under each operating condition to obtain a class correlation evaluation matrix. Then, the thermal fault regions of the chemical production equipment are associated and located based on the class correlation evaluation matrix.

[0016] Therefore, this application uses hot spot interference features and thermal response correlation coefficients to perform fuzzy class correlation assessment on the structural dependence of the thermal response region of chemical production equipment under different operating conditions, obtaining a class correlation assessment matrix. Then, based on the class correlation assessment matrix, the thermal fault region of the chemical production equipment is correlated and located. First, for each operating condition's thermal infrared image, a multi-scale convolution method is used to meticulously capture subtle changes in image texture at different scales, extracting the mutual information differences between scales. This reveals the statistical correlation between hot spots and background texture. Combined with equipment structural information, it achieves accurate differentiation between fault thermal response and non-fault hot spot interference, significantly improving the recognition accuracy of hot spot interference features. This step strengthens the robustness of feature extraction, significantly reduces the interference of environmental noise and false hot spots on diagnostic results, and provides stable and high-quality input features for subsequent correlation analysis of multi-condition thermal response regions, thus improving the overall accuracy and reliability of thermal fault detection. Then, based on all thermal fault features, the thermal response elements of the thermal response region of the chemical production equipment under different operating conditions are correlated to obtain the thermal response characteristics of the thermal fault features of the chemical production equipment under different operating conditions. By calculating the correlation coefficients between thermal fault features under different operating conditions, the similarity and dependence of thermal response features are quantified. This allows for a deeper understanding of the inherent consistency and structural evolution of the equipment's thermal response under multiple operating conditions, effectively addressing the shortcomings of isolated fault feature processing in traditional single-condition diagnosis. This multi-condition correlation analysis lays a solid data foundation for structural dependence assessment, enhances the spatiotemporal continuity and stability of thermal fault features, and significantly improves the accuracy of fault location and the reliability of diagnosis. Finally, fuzzy class correlation assessment is conducted on the structural dependence of the thermal response region of chemical production equipment under different operating conditions using hot spot interference features and thermal response correlation coefficients. This yields a class correlation assessment matrix. Fuzzy class correlation assessment can integrate the uncertainty and fuzziness in multi-source information, overcoming the limitations of traditional hard threshold judgment and improving the accuracy of structural dependence expression. The generated class correlation assessment matrix comprehensively reflects the correlation strength and structural dependence characteristics of the thermal response region under multiple operating conditions, providing a scientific basis for accurate correlation location of fault regions and significantly enhancing the diagnostic model's adaptability to complex operating condition changes and the stability of fault identification. In summary, the proposed solution can achieve structurally dependent correlation localization of thermal fault regions under multiple operating conditions. Attached Figure Description

[0017] Figure 1 This is an exemplary flowchart of an intelligent fault diagnosis method for chemical production equipment according to some embodiments of this application; Figure 2 This is a flowchart illustrating the determination of hot spot interference characteristics according to some embodiments of this application; Figure 3This is a flowchart illustrating the determination of thermal response correlation coefficients according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of an intelligent fault diagnosis system for chemical production equipment according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing an intelligent fault diagnosis method for chemical production equipment, according to some embodiments of this application. Detailed Implementation

[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 The figure is an exemplary flowchart of an intelligent fault diagnosis method for chemical production equipment according to some embodiments of this application. The intelligent fault diagnosis method 100 for chemical production equipment mainly includes the following steps: In step 101, thermal fault detection of chemical production equipment is performed under different operating conditions, and thermal infrared images of chemical production equipment under each operating condition are collected by thermal fault detection equipment.

[0020] It should be noted that the main purpose of the thermal fault detection of chemical production equipment under different operating conditions in this application is to comprehensively acquire the thermal response characteristics of the equipment under various actual operating conditions, so as to accurately identify the thermal fault manifestations that may exist under different operating conditions. By covering a variety of typical operating conditions, the law of the change of equipment thermal characteristics with operating conditions can be revealed, and the normal thermal changes driven by operating conditions can be distinguished from potential abnormal thermal behaviors, thus avoiding the one-sidedness and misjudgment of the diagnostic results under a single operating condition.

[0021] In practice, the thermal infrared images of chemical production equipment under various operating conditions can be acquired by the thermal fault detection equipment in the following way: key equipment nodes are selected at the chemical production site, and multiple typical operating states are set according to operating parameters (such as pressure and load). In each state, the thermal fault detection equipment is used to perform fixed-point thermal infrared imaging acquisition, and the timestamp and operating condition information are recorded simultaneously. It should be noted that the thermal fault detection equipment in this application is specifically a thermal infrared imager. The thermal infrared imager can capture the surface temperature distribution image of the chemical production equipment to reflect possible thermal anomalies inside the chemical production equipment. Its core function is to monitor changes in thermal state without contact.

[0022] In step 102, multi-scale convolution is performed on the thermal infrared images of chemical production equipment under each operating condition to extract the mutual information differences of image textures between different scales. Then, based on all the mutual information differences and the structural information of the chemical production equipment, the hot spot interference features for thermal fault detection under each operating condition are determined.

[0023] In some embodiments, the following steps can be used to perform multi-scale convolution on the thermal infrared images of chemical production equipment under each operating condition and extract the mutual information differences in image texture between different scales: For each operating condition, the thermal infrared image is processed by layered convolution using convolution kernels of different scales to extract the texture feature map at the corresponding convolution scale. Perform texture comparison analysis on texture feature maps at any two scales to determine the texture mutual information between them; Based on all the texture mutual information, the mutual information difference of the image texture of the corresponding thermal infrared image at different scales under the operating conditions is determined, and then the mutual information difference of the image texture of the corresponding thermal infrared image at different scales under each operating condition is obtained.

[0024] It should be noted that the hierarchical convolution processing in this application refers to using convolution kernels of multiple scales to perform multiple convolution operations on the same thermal infrared image to extract local texture features from different spatial perception ranges, so as to reveal the multi-level structural changes of the thermal infrared image from detail to global level; the texture mutual information in this application is an indicator to measure the similarity of texture information in different texture feature maps; the mutual information difference in this application is an indicator used to quantify the degree of difference of texture information at different spatial scales.

[0025] In specific implementation, for each operating condition, firstly, the thermal infrared image is processed by layered convolutional processing using convolutional kernels of different scales to extract the texture feature map at the corresponding convolutional scale. This can be achieved as follows: Common image processing libraries (such as OpenCV) can be used to preprocess the thermal infrared image, including grayscale conversion, size normalization, and noise reduction, to ensure the consistency and accuracy of subsequent processing. Multiple convolutional kernels of different sizes (e.g., 3×3, 5×5, 7×7, etc.) are selected, and without using a deep learning model, layered convolutional processing is performed on the thermal infrared image using a sliding window approach to extract the local texture feature map at each scale. The convolutional output at each scale represents the texture response features of the image within the corresponding spatial perception range. Secondly, texture comparison analysis is performed on the texture feature maps at any two scales to determine the texture mutual information between them. This can be achieved as follows: The texture features at any two scales are compared and analyzed. The feature maps are compared pairwise, and the gray-level distribution of each texture feature map is obtained using histogram matching. Then, the joint gray-level distribution of each pair of feature maps is statistically analyzed to determine the degree of information overlap between them. The information entropy comparison method in information theory is used to determine the similarity of the two in texture structure, and this similarity is taken as the texture mutual information between them. Then, the texture mutual information difference of the corresponding thermal infrared image at different scales under the operating condition can be determined by the following method: collect the texture difference results (i.e., texture mutual information) under all convolution scale combinations to obtain a set of difference indexes reflecting the changes in texture information at different scales, and take this set of difference indexes as the texture mutual information difference of the corresponding thermal infrared image at different scales under the operating condition. Finally, by repeating the above method, the texture mutual information difference of the corresponding thermal infrared image at different scales under each operating condition can be obtained.

[0026] In some embodiments, reference Figure 2 As shown in the figure, this is a flowchart illustrating the process of determining hot spot interference characteristics in some embodiments of this application. In this embodiment, the determination of hot spot interference characteristics for thermal fault detection under each operating condition based on all mutual information differences and the structural information of the chemical production equipment can be achieved through the following steps: In step 1021, for each operating condition, the texture change regions in the thermal infrared image that exhibit significant information gain are determined based on the mutual information difference of image textures at different scales; In step 1022, the non-faulty hot spot areas caused by equipment structure obstruction, structural gaps and material heat capacity differences are located by combining the texture change area with the structural information of the chemical production equipment. In step 1023, hot spot interference characteristics for thermal fault detection under operating conditions are determined based on the non-faulty hot spot region, thereby obtaining hot spot interference characteristics for thermal fault detection under each operating condition.

[0027] It should be noted that the non-faulty hot spot area in this application refers to a local temperature anomaly area caused by equipment structure obstruction, gaps or material thermal characteristics differences, which is unrelated to the actual fault; the hot spot interference feature in this application is a feature that measures the degree of interference of the non-faulty hot spot area with thermal fault detection.

[0028] In specific implementation, for each operating condition, firstly, determining the texture change regions exhibiting significant information gain in the thermal infrared image based on the mutual information difference of image textures at different scales can be achieved in the following way: extracting texture features at different scales of the thermal infrared image using multi-scale convolution operations, and identifying regions with significant texture abrupt changes and high information gain in the thermal infrared image by calculating the mutual information difference between any two scales, and taking these regions as texture change regions; secondly, locating non-faulty hot spot regions caused by equipment structural obstruction, structural gaps, and material heat capacity differences by combining the texture change regions with the structural information of the chemical production equipment can be achieved in the following way: the aforementioned texture change regions can be... Spatially comparing and registering the chemical regions with the structural information of the chemical equipment (such as computer-aided design structural diagrams or 3D point cloud models), an image-structure alignment algorithm (such as a rigid registration method based on feature point matching) is used to determine whether these regions fall on structural gaps, obstructions, or material interfaces with significant differences in heat capacity. This locates non-faulty hotspot regions induced by the structure. It should be further explained that the image-structure alignment algorithm (such as a rigid registration method based on feature point matching) used in this application refers to matching key texture feature points in the thermal infrared image with corresponding structural points in the known structural model of the chemical equipment, establishing a one-to-one spatial correspondence between image information and 3D structural information. Through this alignment, it can be determined whether the location of abnormal texture changes in the image corresponds precisely to areas such as structural gaps, obstructions, or material interfaces of the equipment. Since these structural factors often cause non-faulty hotspots (such as localized temperature anomalies caused by uneven heat conduction, different heat capacity, or blocked heat dissipation paths), locating them helps to accurately identify and eliminate such misleading thermal anomaly areas, ensuring that thermal fault analysis focuses on the actual equipment fault response. Finally, the hotspot interference characteristics for thermal fault detection under operating conditions can be determined based on the non-faulty hotspot areas in the following way: multi-dimensional thermal response characteristics of temperature distribution, texture structure, and boundary stability can be extracted from the located non-faulty hotspot areas. These characteristics reflect the morphology, intensity, and variation law of the hotspots. Combined with the spatial location of the area and equipment structural information, these characteristics are statistically analyzed and pattern recognized to clarify their difference from the actual fault thermal response. These characteristics are then summarized as hotspot interference characteristics to identify and eliminate the interference of non-faulty hotspots on the thermal fault detection process, thereby improving the accuracy and robustness of diagnosis.

[0029] In step 103, thermal fault features of chemical production equipment under various operating conditions are extracted from all thermal infrared images. Based on all thermal fault features, thermal response elements are correlated between the thermal response regions of chemical production equipment under different operating conditions to obtain thermal response correlation coefficients of thermal fault features of chemical production equipment under different operating conditions.

[0030] In some embodiments, extracting thermal fault features of chemical production equipment under various operating conditions from all thermal infrared images can be achieved using the following steps: For each operating condition, temperature distribution analysis is performed on thermal infrared images to identify abnormal temperature areas that are higher than the ambient thermal distribution pattern. The geometric features, grayscale distribution features, and boundary features of the abnormal temperature region are extracted to form a thermal anomaly feature set; The thermal anomaly feature set is integrated and classified by image clustering method, and non-fault areas are eliminated to obtain thermal fault area features with abnormal structural response. The thermal fault region characteristics are used as the thermal fault characteristics of chemical production equipment under operating conditions, thereby obtaining the thermal fault characteristics of chemical production equipment under each operating condition.

[0031] It should be noted that the thermal fault characteristics in this application refer to the thermal response characteristics that reflect local temperature anomalies and structural changes during abnormal operation of chemical equipment, and are used to identify and locate possible fault areas.

[0032] In specific implementation, firstly, temperature distribution analysis is performed on thermal infrared images to identify abnormal temperature regions that are higher than the ambient temperature distribution pattern. This can be achieved as follows: For thermal infrared images acquired under each operating condition, temperature threshold segmentation techniques (such as difference detection based on background modeling) are used to identify abnormal temperature regions where the temperature is significantly higher than the ambient temperature distribution. This step analyzes the temperature value distribution of image pixels, eliminates normal temperature fluctuations, and highlights potential fault areas. Secondly, the geometric features, grayscale distribution features, and boundary features of the abnormal temperature regions are extracted to form a thermal anomaly feature set. This can be achieved as follows: Multidimensional features are extracted from the identified abnormal temperature regions, including geometric features (such as area and shape indices), grayscale distribution features (statistical grayscale histogram, mean, and variance), and boundary features (edge ​​detection algorithms such as the Canny operator are used to detect boundary continuity and robustness). All extracted features are then combined into a set as the thermal anomaly feature set. Finally, image clustering methods are used to integrate and classify multiple features in the thermal anomaly feature set, eliminating non-faulty regions and obtaining thermal fault regions with abnormal structural responses. The features can be implemented as follows: Multidimensional features extracted from each abnormal temperature region (including region area, boundary perimeter, gray-level mean, gray-level variance, boundary gradient magnitude, etc.) can be uniformly organized into feature vectors. All feature vectors are then standardized (e.g., Z-score standardization) to eliminate the influence of dimensions and improve the stability of the clustering results. A K value (i.e., the number of clusters) is then set, which can be optimized using experience or by using the silhouette coefficient as an evaluation metric. All feature vectors are then used as input, and the K-means algorithm is used to divide them into K categories based on Euclidean distance. The algorithm aims to minimize the distance from each point to its cluster center. After clustering, the feature distribution characteristics of each type of center are statistically analyzed. Clusters with low center gray-level, small region area, or unstable boundary gradients are identified as non-fault regions and removed. The retained clusters represent regions with concentrated high temperatures, clear boundaries, and a certain degree of geometric stability. These clustering results are then used as the features of thermal fault regions with abnormal structural responses. Finally, these thermal fault region features are used as the thermal fault features of chemical production equipment under operating conditions.

[0033] In some embodiments, reference Figure 3 As shown in the figure, this is a flowchart illustrating the determination of thermal response correlation coefficients in some embodiments of this application. In this embodiment, thermal response elements are correlated between the thermal response regions of chemical production equipment under different operating conditions based on all thermal fault characteristics. The thermal response correlation coefficients of thermal fault characteristics of chemical production equipment under different operating conditions can be obtained by the following steps: In step 1031, for each of the two operating conditions, the thermal fault features extracted under the two operating conditions are aligned according to the spatial structure of the equipment, thereby determining the correlation between the thermal fault features in each thermal response region. In step 1032, the differences in thermal response elements of the same thermal response region under the two operating conditions are extracted; In step 1033, the thermal response correlation coefficient of the thermal fault characteristics of the chemical production equipment between the two operating conditions is determined based on the correlation relationship of the thermal fault characteristics and the difference characteristics of the thermal response elements, thereby obtaining the thermal response correlation coefficient of the thermal fault characteristics of the chemical production equipment between each pair of operating conditions.

[0034] It should be noted that the thermal response region in this application refers to a local area in chemical production equipment that exhibits thermal characteristic responses under thermal infrared images as operating conditions change; the thermal response correlation coefficient in this application is an indicator that measures the degree of correlation between changes in thermal fault characteristics in the same thermal response region under different operating conditions.

[0035] In specific implementation, for each pair of operating conditions, firstly, the extracted thermal fault features under the two operating conditions are aligned according to the equipment spatial structure. The correlation of thermal fault features in each thermal response region can be determined in the following way: a rigid registration method based on spatial constraints can be used to align the extracted thermal fault features under the two operating conditions according to the equipment spatial structure. This involves setting a geometric mapping relationship between thermal infrared image feature points and key points of the equipment structure to achieve spatial registration of thermal fault features, and statistically analyzing the information difference values ​​of thermal fault features in each thermal response region, using the average of all information difference values ​​as the quantification value of the correlation. Secondly, the difference features of thermal response elements in the same thermal response region under the two operating conditions can be extracted in the following way: after completing the position alignment, thermal response elements such as temperature gradient, texture distribution, and boundary morphology are extracted for each pair of corresponding thermal response regions, and the differences in these elements under the two operating conditions are calculated. This can be quantified using difference analysis or a statistical structural similarity evaluation method, and the differences in these differences are used as the correlation value for the same thermal response region under the two operating conditions. The thermal response characteristics of the thermal response elements under the two operating conditions are determined by the following method: Based on the thermal fault characteristics that have been aligned, the multi-dimensional thermal response elements such as temperature gradient, texture direction consistency, and boundary clarity in the same thermal response area under the two operating conditions are normalized to eliminate the influence of different scales and dimensions, and to ensure the comparability of each feature dimension in the evaluation. Then, combined with the stability of the performance of each thermal response element in thermal faults by the equipment structure, weight coefficients are set based on experience or training based on historical data. For example, the weight of temperature gradient change is higher than that of texture change. A weighted coefficient matrix is ​​constructed. The normalized difference value of each type of feature is multiplied by the corresponding weight and summed to obtain the thermal response correlation coefficient representing the degree of feature variation of the thermal response area under the two operating conditions. The principle is based on the linear weighted fusion idea, which emphasizes that under the premise of maintaining structural consistency, the smaller the change of different thermal response elements, the stronger the regional correlation, and vice versa, thereby realizing the differential correlation modeling between thermal fault responses under multiple operating conditions.

[0036] It should be noted that the proposed solution in this application realizes a multi-condition thermal fault feature mapping mechanism based on structural alignment and thermal response element difference analysis by constructing thermal response correlation coefficients between thermal fault features of chemical production equipment under different operating conditions. In the implementation process, the location registration of thermal fault features extracted under different operating conditions is first performed with spatial structure as a constraint to ensure semantic consistency between regions. Then, the quantitative differences of thermal response elements such as temperature gradient and thermal boundary morphology are extracted within the same thermal response region to form a discriminative feature difference vector. Furthermore, by combining the spatial correlation between thermal fault features, a weighted rule is used to construct thermal response correlation coefficients to express the correlation strength of thermal response changes in regions under different operating conditions. This correlation coefficient achieves a normalized representation of thermal feature changes while maintaining structural consistency, providing a quantitative basis for cross-condition thermal anomaly identification and enhancing the adaptability of the diagnostic model to complex operating environments and the accuracy of fault identification.

[0037] In step 104, the structural dependence of the thermal response region of the chemical production equipment under different operating conditions is evaluated by using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients to obtain a class correlation evaluation matrix. Then, the thermal fault region of the chemical production equipment is associated and located based on the class correlation evaluation matrix.

[0038] In some embodiments, the structural dependence of the thermal response region of chemical production equipment under different operating conditions is evaluated by fuzzy class correlation using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients. The resulting class correlation evaluation matrix can be achieved through the following steps: The thermal fault information loss during the extraction process of thermal fault features for chemical production equipment under each operating condition is determined based on the hot spot interference characteristics of thermal fault detection under each operating condition. A fuzzy inference model is constructed based on all thermal fault information loss and all thermal response correlation coefficients. The fuzzy inference model is used to calculate the fuzzy class membership degree of the structural dependence of the thermal response region of the chemical production equipment, so as to obtain the fuzzy dependence relationship of each thermal response region under different operating conditions. The fuzzy dependencies of all thermal response regions are aggregated to obtain the class correlation evaluation matrix.

[0039] It should be noted that the thermal fault information loss in this application is an indicator of the degree of distortion of the expression of real thermal fault features caused by interference factors during the feature extraction process of thermal infrared images; the fuzzy dependency relationship in this application is a fuzzy correlation index that measures the consistency of structural response in thermal response regions; and the class correlation evaluation matrix in this application is a fuzzy correlation matrix that characterizes the strength of structural dependence of each thermal response region under cross-operating conditions.

[0040] It should also be noted that the fuzzy inference model in this application is a knowledge-driven model that realizes nonlinear mapping between input and output based on fuzzy sets and rule bases. In this embodiment, the fuzzy inference model takes the thermal fault information loss and thermal response correlation coefficient as input variables, sets the corresponding fuzzy membership function to convert them into fuzzy quantities, and constructs an "if-then" form rule base in combination with the equipment structure dependency characteristics. The fuzzy inference model realizes the calculation of fuzzy class membership degree of the structural dependency of the thermal response region through four steps: fuzzification, rule matching, inference and defuzzification. The model can adopt a Mamdani type structure, and the output is a continuous value representing the strength of structural dependency, which has good expressive power and interpretability.

[0041] In specific implementation, firstly, the thermal fault information loss in the extraction process of thermal fault features for chemical production equipment under each operating condition can be determined based on the hot spot interference characteristics of thermal fault detection under each operating condition. This can be achieved in the following way: For the thermal infrared image acquired under each operating condition, after separating the thermal fault region using image segmentation technology (such as U-Net deep convolutional neural network), the texture feature difference analysis is performed by combining the hot spot interference characteristics output by the convolutional layer with the original image features to extract the degree of interference of the target thermal fault feature by the hot spot interference characteristics under that operating condition, and this degree of interference is used as the thermal fault information loss in the extraction process of thermal fault features for chemical production equipment under that operating condition. Secondly, the fuzzy inference model can be constructed based on all thermal fault information losses and all thermal response correlation coefficients in the following way: When constructing the fuzzy inference model, the thermal fault information loss and thermal response correlation coefficient can be used as input variables. A Mamdani-type fuzzy inference system is adopted, and the fuzzy membership function of the input variables (such as a Gaussian function to smoothly describe "low temperature") is set. The system identifies three fuzzy levels: "high loss," "medium loss," and "high loss." A rule base is constructed based on thermal conduction characteristics (e.g., "If information loss is high and the correlation coefficient is low, the dependency is weak," and "If information loss is low and the correlation coefficient is high, the dependency is strong"). Then, the fuzzy inference model calculates the fuzzy membership degree of the structural dependency of the thermal response region of the chemical production equipment. The fuzzy dependency relationship of each thermal response region under different operating conditions can be achieved as follows: After inputting specific values ​​(i.e., all thermal fault information loss and thermal response correlation coefficients), the fuzzy inference model performs fuzzification, rule matching, and fuzzy inference on the input variables of each thermal response region. Finally, the results are defuzzified using the central average method to obtain the fuzzy structural dependency degree (i.e., membership degree) of the thermal response region. Finally, the fuzzy dependency relationships of all thermal response regions are aggregated to obtain the class correlation evaluation matrix. This can be achieved as follows: The fuzzy dependency relationships (i.e., membership degrees) of all thermal response regions can be combined into a matrix according to image positional relationships, and this matrix is ​​used as the class correlation evaluation matrix.

[0042] It should be noted that, compared with existing thermal fault diagnosis technologies, this application's solution addresses the shortcomings of existing technologies in dealing with changes in multiple operating conditions and interference effects. It solves the technical problems of unstable thermal fault feature expression, weak correlation of structural response information, and low positioning accuracy under cross-operating conditions. Firstly, by introducing hot spot interference features, it can identify the impact of interference factors on thermal fault feature extraction under different operating conditions from the image texture level, quantitatively characterizing information loss and effectively enhancing the robustness of the feature extraction stage. Secondly, it uses thermal response correlation coefficients to quantitatively describe the stability of thermal response regions under different operating conditions, compensating for the deficiency of traditional methods in identifying the degree of structural thermal response coupling between operating conditions. Thirdly, it constructs a fuzzy inference model, using information loss and response correlation coefficients as input variables, and implements fuzzy class membership inference of structural dependence based on a fuzzy rule system, improving the adaptability to complex nonlinear structural response relationships. Fourthly, it relies on the aggregation of fuzzy dependencies to generate a class-related evaluation matrix, comprehensively reflecting the strength of structural dependence of each thermal response region under multiple operating conditions, ultimately achieving high-precision, interference-resistant correlation positioning of thermal fault regions, effectively improving the intelligence, stability, and accuracy of the diagnostic system in complex operating environments.

[0043] In some embodiments, the correlation and localization of thermal fault regions in chemical production equipment based on the class-related evaluation matrix can be achieved using the following steps: Initialize a thermal fault detection model suitable for thermal fault diagnosis of chemical production equipment; The class-related evaluation matrix is ​​used as the association constraint term of the thermal fault detection model; The thermal infrared images of the chemical production equipment under various operating conditions are input into the thermal fault detection model as input parameters of the thermal fault detection model. The thermal fault detection model is used to locate the thermal fault area of ​​the production equipment.

[0044] It should be noted that the association constraint in this application refers to the auxiliary regularization condition used to guide the model to consider the structural dependence between thermal response regions under different operating conditions during the training process.

[0045] It should also be noted that the thermal fault detection model in this application is a deep learning-based intelligent diagnostic model designed to automatically identify and locate thermal fault regions from thermal infrared images of chemical production equipment. Its technical principles include multi-layer convolutional neural networks (such as ResNet and DenseNet) used to extract multi-scale spatial features and texture information from the images. Through convolutional and pooling layers, the morphology, temperature distribution, and texture features of hot spots are gradually abstracted to form rich feature representations. The model typically integrates attention mechanisms (such as channel attention or spatial attention modules) to enhance attention to key thermal response regions and improve the ability to discriminate fault features. The class correlation evaluation matrix, as prior knowledge of the structural dependencies between thermal response regions, is embedded in the model through the feature fusion layer or added to the loss function as an additional regularization constraint. This prompts the model to fully consider the correlation between different thermal response regions and cross-condition stability during training and inference. This fusion structure not only enhances the model's adaptability to multi-condition interference but also effectively improves the accuracy and robustness of thermal fault localization, achieving accurate identification and localization of equipment thermal faults under complex operating conditions.

[0046] In specific implementation, firstly, initializing a thermal fault detection model suitable for thermal fault diagnosis of chemical production equipment can be achieved in the following way: when initializing the thermal fault detection model, a typical deep learning framework can be selected, such as a heat map classification and localization model with ResNet or DenseNet as the backbone network, and a spatial attention mechanism can be integrated at the output to enhance the attention to key areas; secondly, using the class-related evaluation matrix as an association constraint term of the thermal fault detection model can be achieved in the following way: the class-related evaluation matrix is ​​embedded in the thermal fault detection model as a structural dependency association constraint. Specifically, the class-related evaluation matrix is ​​introduced as an auxiliary input into the intermediate layer of the model (such as a feature fusion layer or attention layer) through a fusion layer, and participates in the loss function calculation as a regularization term during training, thereby prompting the thermal fault detection model to improve its performance. In the feature mapping stage, priority is given to structurally related regions. Then, the thermal infrared images of the chemical production equipment under various operating conditions are input into the thermal fault detection model. The input parameters of the thermal fault detection model can be implemented in the following way: the thermal infrared images under different operating conditions are used as the main input of the model, and after unifying the resolution and image format, they are input into the front end of the neural network to extract multi-scale spatial texture features. Finally, the thermal fault region of the chemical production equipment can be located by the thermal fault detection model in the following way: the trained model combines the class correlation evaluation matrix and image features in the feature space to output the thermal fault probability map of each thermal response region. The thermal fault region is accurately located by post-processing methods (such as threshold segmentation or connectivity region analysis) to achieve intelligent diagnosis and localization across operating conditions and structural dependencies.

[0047] It should be noted that this application addresses the technical bottlenecks of existing thermal fault diagnosis technologies, such as dispersed thermal response features under multiple operating conditions, insufficient utilization of structural dependency information, and low fault location accuracy. It proposes an association-based localization method based on a class-related evaluation matrix: First, by embedding the class-related evaluation matrix as an association constraint term into the thermal fault detection model, explicit modeling of structural dependencies between thermal response regions under different operating conditions is achieved, solving the problem that traditional methods cannot effectively integrate cross-condition structural information, leading to isolated and inconsistent diagnostic results. Second, the model uses multi-condition thermal infrared images as input and combines multi-scale thermal features extracted by a deep convolutional network, improving the richness and robustness of thermal fault feature representation. Finally, the association constraint guides the model to prioritize the identification of fault regions with strong structural correlations during the inference phase, achieving precise localization of thermal fault regions and consistent cross-condition discrimination. This effectively enhances the diagnostic system's adaptability to changes in operating conditions and disturbances, significantly improves the accuracy and stability of fault detection, and ensures the operational safety and maintenance efficiency of chemical production equipment.

[0048] On the other hand, in some embodiments, this application provides an intelligent fault diagnosis system for chemical production equipment, referencing... Figure 4 The figure is a schematic diagram of the structure of an intelligent fault diagnosis system for chemical production equipment according to some embodiments of this application. The intelligent fault diagnosis system 400 for chemical production equipment includes: a data acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to respond to thermal fault detection of chemical production equipment under different operating conditions, and to acquire thermal infrared images of chemical production equipment under various operating conditions through thermal fault detection equipment. Processing module 402 in this application is used to perform multi-scale convolution on the thermal infrared images of chemical production equipment under each operating condition, extract the mutual information difference of image texture between different scales, and then determine the hot spot interference features of thermal fault detection under each operating condition based on all mutual information differences and the structural information of chemical production equipment. In this application, the processing module 402 is also used to extract the thermal fault features of chemical production equipment under various operating conditions from all thermal infrared images, and to correlate the thermal response elements of the thermal response regions of chemical production equipment under different operating conditions based on all thermal fault features, so as to obtain the thermal response correlation coefficient of the thermal fault features of chemical production equipment under different operating conditions. The execution module 403 in this application is mainly used to perform fuzzy class correlation assessment on the structural dependence of the thermal response region of chemical production equipment under different operating conditions by using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients, to obtain a class correlation assessment matrix, and then to locate the thermal fault region of chemical production equipment based on the class correlation assessment matrix.

[0049] In addition, this application also provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to acquire the code and execute the above-described intelligent fault diagnosis method for chemical production equipment.

[0050] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device for implementing an intelligent fault diagnosis method for chemical production equipment according to some embodiments of this application. The intelligent fault diagnosis method for chemical production equipment in the above embodiments can... Figure 5 The computer device shown is used to implement this, and the computer device 500 includes at least one processor 501, a communication bus 502, a memory 503, and at least one communication interface 504.

[0051] Processor 501 can be a general-purpose central processing unit (CPU) or an application-specific integrated circuit (ASIC).

[0052] The communication bus 502 can be used to transmit information between the aforementioned components.

[0053] Memory 503 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 503 may exist independently and be connected to processor 501 via communication bus 502. Memory 503 may also be integrated with processor 501.

[0054] The memory 503 stores program code for executing the scheme of this application, and its execution is controlled by the processor 501. The processor 501 executes the program code stored in the memory 503. The program code may include one or more software modules. The intelligent fault diagnosis method for chemical production equipment in the above embodiments can be implemented by the processor 501 and one or more software modules in the program code in the memory 503.

[0055] Communication interface 504 uses any transceiver-like device to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0056] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0057] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0058] In addition, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned intelligent fault diagnosis method for chemical production equipment.

[0059] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0060] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for intelligent fault diagnosis of chemical production equipment, characterized in that, Includes the following steps: The system responds to thermal fault detection of chemical production equipment under different operating conditions by acquiring thermal infrared images of the chemical production equipment under various operating conditions. Multi-scale convolution is performed on the thermal infrared images of chemical production equipment under each operating condition to extract the mutual information difference of image texture between different scales. Then, based on all the mutual information differences and the structural information of the chemical production equipment, the hot spot interference features of thermal fault detection under each operating condition are determined. Thermal fault features of chemical production equipment under various operating conditions are extracted from all thermal infrared images. Based on all thermal fault features, thermal response elements are correlated between the thermal response regions of chemical production equipment under different operating conditions to obtain thermal response correlation coefficients of thermal fault features of chemical production equipment under different operating conditions. By using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients, a fuzzy class correlation assessment is performed on the structural dependence of the thermal response region of chemical production equipment under different operating conditions to obtain a class correlation assessment matrix. Then, based on the class correlation assessment matrix, the thermal fault region of chemical production equipment is correlated and located.

2. The method as described in claim 1, characterized in that, Multi-scale convolution is performed on the thermal infrared images of chemical production equipment under each operating condition to extract the mutual information differences in image texture between different scales. Specifically, this includes: For each operating condition, the thermal infrared image is processed by layered convolution using convolution kernels of different scales to extract the texture feature map at the corresponding convolution scale. Perform texture comparison analysis on texture feature maps at any two scales to determine the texture mutual information between them; Based on all the texture mutual information, the mutual information difference of the image texture of the corresponding thermal infrared image at different scales under the operating conditions is determined, and then the mutual information difference of the image texture of the corresponding thermal infrared image at different scales under each operating condition is obtained.

3. The method as described in claim 1, characterized in that, Based on all mutual information differences and structural information of chemical production equipment, the hot spot interference characteristics for thermal fault detection under each operating condition are specifically determined as follows: For each operating condition, the texture change regions in the thermal infrared image that exhibit significant information gain are determined based on the mutual information difference of image textures at different scales; By combining the texture variation regions with the structural information of the chemical production equipment, non-faulty hot spot regions caused by equipment structure obstruction, structural gaps, and material heat capacity differences can be located. Based on the non-faulty hot spot region, the hot spot interference characteristics of thermal fault detection under operating conditions are determined, and then the hot spot interference characteristics of thermal fault detection under each operating condition are obtained.

4. The method as described in claim 1, characterized in that, Extracting thermal fault characteristics of chemical production equipment under various operating conditions from all thermal infrared images specifically includes: For each operating condition, temperature distribution analysis is performed on thermal infrared images to identify abnormal temperature areas that are higher than the ambient thermal distribution pattern. The geometric features, grayscale distribution features, and boundary features of the abnormal temperature region are extracted to form a thermal anomaly feature set; The thermal anomaly feature set is integrated and classified by image clustering method, and non-fault areas are eliminated to obtain thermal fault area features with abnormal structural response. The thermal fault region characteristics are used as the thermal fault characteristics of chemical production equipment under operating conditions, thereby obtaining the thermal fault characteristics of chemical production equipment under each operating condition.

5. The method as described in claim 1, characterized in that, Based on all thermal fault characteristics, the thermal response elements of chemical production equipment under different operating conditions are correlated to obtain the thermal response correlation coefficients of thermal fault characteristics of chemical production equipment under different operating conditions. Specifically, these coefficients include: For each pair of operating conditions, the thermal fault features extracted under the two operating conditions are aligned according to the spatial structure of the equipment to determine the correlation between the thermal fault features in each thermal response region. Extract the differences in thermal response elements of the same thermal response region under the two operating conditions; Based on the correlation of the thermal fault characteristics and the difference characteristics of the thermal response elements, the thermal response correlation coefficient of the thermal fault characteristics of the chemical production equipment between the two operating conditions is determined, and then the thermal response correlation coefficient of the thermal fault characteristics of the chemical production equipment between each pair of operating conditions is obtained.

6. The method as described in claim 1, characterized in that, By using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients, a fuzzy class correlation assessment is performed on the structural dependence of the thermal response region of chemical production equipment under different operating conditions. The resulting class correlation assessment matrix includes: The thermal fault information loss during the extraction process of thermal fault features for chemical production equipment under each operating condition is determined based on the hot spot interference characteristics of thermal fault detection under each operating condition. A fuzzy inference model is constructed based on all thermal fault information loss and all thermal response correlation coefficients. The fuzzy inference model is used to calculate the fuzzy class membership degree of the structural dependence of the thermal response region of the chemical production equipment, so as to obtain the fuzzy dependence relationship of each thermal response region under different operating conditions. The fuzzy dependencies of all thermal response regions are aggregated to obtain the class correlation evaluation matrix.

7. The method as described in claim 1, characterized in that, The thermal fault detection device is specifically a thermal infrared imager.

8. A fault diagnosis system for chemical production equipment, characterized in that, include: The acquisition module is used to respond to thermal fault detection of chemical production equipment under different operating conditions. It acquires thermal infrared images of chemical production equipment under various operating conditions through thermal fault detection equipment. The processing module is used to perform multi-scale convolution on the thermal infrared images of chemical production equipment under each operating condition, extract the mutual information differences of image texture between different scales, and then determine the hot spot interference features of thermal fault detection under each operating condition based on all mutual information differences and the structural information of chemical production equipment. The processing module is also used to extract thermal fault features of chemical production equipment under various operating conditions from all thermal infrared images, and to correlate thermal response elements of the thermal response regions of chemical production equipment under different operating conditions based on all thermal fault features, so as to obtain thermal response correlation coefficients of thermal fault features of chemical production equipment under different operating conditions. The execution module is used to perform fuzzy class correlation assessment on the structural dependence of the thermal response region of chemical production equipment under different operating conditions by using the hot spot interference characteristics of thermal fault detection under each operating condition and all thermal response correlation coefficients, to obtain a class correlation assessment matrix, and then to locate the thermal fault region of chemical production equipment based on the class correlation assessment matrix.

9. A computer device comprising a memory and a processor, the memory storing code, characterized in that, The processor is configured to acquire the code and execute the intelligent fault diagnosis method for chemical production equipment as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent fault diagnosis method for chemical production equipment as described in any one of claims 1 to 7.

Images