Petrochemical equipment-based model fusion training method and system

Abstract

The invention belongs to the technical field of petrochemical industry deep fusion, and provides a petrochemical equipment-based model fusion training method and system, and the method comprises the steps: recognizing a diagnosis model which accords with the fault diagnosis of petrochemical equipment, and carrying out the clustering analysis of different operation parameters of the petrochemical equipment during the current fault, and carrying out the classification and integration of the same kind, obtaining a unit parameter set, and determining an optimal diagnosis model corresponding to the unit parameter set; unit parameter groups in different unit parameter sets are extracted and integrated in combination with training verification performance of the optimal diagnosis model on the different unit parameter sets, and a training database of the optimal diagnosis model is obtained; performing training verification on the optimal diagnosis models, determining training verification weights of the optimal diagnosis models, and performing parallel fusion on all the optimal diagnosis models to obtain a target diagnosis model; the operation parameters of the petrochemical equipment during the current fault are used as input, and the fault index coefficient of the petrochemical equipment is output, so that the accuracy of petrochemical equipment fault diagnosis is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of deep integration of the petrochemical industry, and specifically relates to a training method and system for model fusion based on petrochemical equipment. Background Art

[0002] With the rapid development of information technology, data-driven fault diagnosis methods have gradually become a research hotspot. Among them, using machine learning and deep learning algorithms to construct diagnostic models for fault diagnosis of petrochemical equipment shows great potential. However, in the actual application process, the existing diagnostic model construction and application methods still have many limitations. On the one hand, in the model training stage, most of them only simply use some operating parameters of petrochemical equipment during the historical operation cycle for training, and fail to fully explore and utilize the rich operating data of the equipment at different fault times, resulting in the model being unable to comprehensively and deeply learn the operating characteristics and fault characteristics of petrochemical equipment, making the model lack generalization ability and difficult to guarantee the diagnostic accuracy when facing complex and changeable actual fault situations. On the other hand, different degrees of faults often correspond to abnormal changes in different operating parameters of petrochemical equipment. The existing methods often do not conduct detailed classification and targeted analysis of operating parameters, but mix all parameters together for model training, which makes it difficult for the model to accurately capture the key parameter information highly related to a specific fault degree, further restricting the improvement of the model's diagnostic performance.

[0003] Therefore, the present invention provides a training method and system for model fusion based on petrochemical equipment. Summary of the Invention

[0004] In order to make up for the deficiencies of the prior art and solve at least one technical problem proposed in the background art.

[0005] The technical solution adopted by the present invention to solve its technical problems is: A training method for model fusion based on petrochemical equipment, comprising the following steps: Using the basic parameter groups of petrochemical equipment at different fault times during the historical operation cycle, training and validating the models in the existing model library to obtain a diagnostic model library; Through cluster analysis of different operating parameters of petrochemical equipment during the current fault, splitting the basic parameter groups to obtain unit parameter groups, and using the unit parameter sets obtained by classifying and integrating the unit parameter groups according to the type of operating parameters to train and validate each diagnostic model in the diagnostic model library to determine the optimal diagnostic model; Combining the training and validation performance of the optimal diagnostic model on different unit parameter sets, extracting and integrating the unit parameter groups within different unit parameter sets to obtain a training database for the optimal diagnostic model; Through the training and verification of the training database, determine the training and verification weights of each optimal diagnostic model, and fuse all the optimal diagnostic models in parallel to obtain the target diagnostic model; Use the operating parameters of the petrochemical equipment at the current failure as the input of the target diagnostic model, and output the failure index coefficient of the petrochemical equipment.

[0006] Furthermore, the acquisition method of the diagnostic model library is as follows: Use the basic parameter groups of the petrochemical equipment at different failure times during the historical operation cycle to train and verify the model; Through the error comparison between the failure index coefficient output by the model verification and the failure index coefficient included in the basic parameter group, mark the basic parameter group as the verification matching group and the non-verification matching group, and process and analyze to obtain the verification matching feature and the verification deviation feature, and conduct deviation analysis on the verification matching feature and the verification deviation feature to obtain the verification reflection value; Mark the model with the verification reflection value less than the verification reflection threshold as the diagnostic model, and summarize to obtain the diagnostic model library.

[0007] Furthermore, the acquisition method of the verification matching feature and the verification deviation feature is as follows: Count the quantity proportion of the verification matching group in the basic parameter group to obtain the verification matching feature; Mark the error between the failure index coefficient output by the model verification that is not within the preset allowable error range and the failure index coefficient included in the basic parameter group as the failure index error, calculate the absolute deviation ratio between the failure index error and the endpoint value of the adjacent nearest preset allowable error range, obtain the out-of-bounds error ratio of the non-verification matching group, and perform averaging processing on the out-of-bounds error ratios of all non-verification matching groups to obtain the verification deviation feature.

[0008] Furthermore, the acquisition method of the unit parameter set is as follows: Obtain different operating parameters of the petrochemical equipment at the current failure, perform clustering analysis using the hierarchical clustering algorithm, split the operating parameters that are in the same category after clustering analysis from the basic parameter group, combine them to form a unit parameter group, and integrate the unit parameter groups with the same type of operating parameters into a unit parameter set.

[0009] Furthermore, the determination method of the optimal diagnostic model is as follows: Use the unit parameter groups included in the unit parameter set to train, verify and test each diagnostic model in the diagnostic model library, calculate the error between the failure index coefficient output by the diagnostic model and the failure index coefficient included in the unit parameter group, mark the unit parameter group as the matching unit parameter group and the non-matching unit parameter group, and process and analyze to obtain the verification compliance feature and the verification non-compliance feature, and through the deviation analysis of the verification compliance feature and the verification non-compliance feature, obtain the verification compliance value; Based on any unit parameter set, obtain the diagnostic model corresponding to the maximum verification compliance value when the unit parameter set is used for training, verification, and testing of each diagnostic model, and use it as the optimal diagnostic model corresponding to the unit parameter set.

[0010] Further, the ways to obtain the verification compliance feature and the verification non-compliance feature are as follows: Count the quantity proportion of the matching unit parameter groups in the unit parameter set to obtain the verification compliance feature; Mark the error between the fault index coefficient output by the diagnostic model outside the preset error range and the fault index coefficient included in the unit parameter group as the non-matching index error, calculate the absolute deviation ratio between the non-matching index error and the endpoint value of the adjacent nearest preset error range to obtain the out-of-bounds error ratio of the non-matching unit parameter groups, and perform averaging processing on the out-of-bounds error ratios of all non-matching unit parameter groups to obtain the verification non-compliance feature.

[0011] Further, the process of obtaining the training database is as follows: Obtain the verification compliance features after the optimal diagnostic model is trained, verified, and tested with different unit parameter sets, sum them to obtain the total verification compliance value, calculate the ratio between the verification compliance feature and the total verification compliance value to obtain the extraction ratio of the unit parameter group corresponding to the unit parameter set, and extract and integrate the unit parameter groups in each unit parameter set according to the extraction ratio of the unit parameter group corresponding to the unit parameter set to obtain the training database of the optimal diagnostic model.

[0012] Further, the way to determine the training and verification weights of the optimal diagnostic model is as follows: Based on any optimal diagnostic model, obtain the fault index coefficient output by the optimal diagnostic model, and perform error comparison and analysis with the fault index coefficient included in the unit parameter group to identify the adaptable unit parameter groups in the unit parameter group; Count the quantity proportion of the adaptable unit parameter groups in the training database to obtain the verification adaptability feature; Calculate the ratio between the verification adaptability feature and the total verification adaptability value to obtain the training and verification weights of the optimal diagnostic model; Among them, the total verification adaptability value is the sum of all verification adaptability features; Combine each optimal diagnostic model with the corresponding training and verification weights and add them together to obtain the target diagnostic model.

[0013] Further, the way to output the fault index coefficient of the petrochemical equipment is as follows: Use the operating parameters of the petrochemical equipment during the current fault as the input of the target diagnostic model, output through each optimal diagnostic model in the target diagnostic model, and combine the output results of each optimal diagnostic model with the corresponding weights to output the fault index coefficient of the petrochemical equipment.

[0014] A training system based on model fusion of petrochemical equipment, comprising: Diagnostic model library acquisition module: Using the basic parameter groups of petrochemical equipment at different fault times during the historical operation cycle, training and validating the models in the existing model library to obtain a diagnostic model library; Optimal diagnostic model identification module: Through cluster analysis of different operating parameters of petrochemical equipment during the current fault, splitting the basic parameter groups to obtain unit parameter groups, and using the unit parameter sets obtained by classifying and integrating the unit parameter groups according to the type of operating parameters, training and validating each diagnostic model in the diagnostic model library to determine the optimal diagnostic model; Training database construction module: Combining the training and validation performance of the optimal diagnostic model on different unit parameter sets, extracting and integrating the unit parameter groups within different unit parameter sets to obtain the training database of the optimal diagnostic model; Target model acquisition module: Through the training and validation of the training database, determining the training and validation weights of each optimal diagnostic model, and fusing all the optimal diagnostic models in parallel to obtain a target diagnostic model; Fault diagnosis output module: Using the operating parameters of petrochemical equipment during the current fault as the input of the target diagnostic model, and outputting the fault index coefficient of the petrochemical equipment.

[0015] The beneficial effects of the present invention are as follows: Obtain the basic parameter sets of the petrochemical equipment at different fault times during the historical operation cycle, train and verify the models in the existing model library, identify the diagnostic models that conform to the fault diagnosis of the petrochemical equipment, and integrate them to obtain a diagnostic model library; Through the cluster analysis of different operating parameters of the petrochemical equipment at the current fault, split the basic parameter sets to obtain unit parameter sets, and classify and integrate the unit parameter sets according to the types of operating parameters of the petrochemical equipment to obtain unit parameter sets. Use the unit parameter sets to train, verify and test each diagnostic model in the diagnostic model library to determine the optimal diagnostic model corresponding to the unit parameter sets; Combine the training and verification performances of the optimal diagnostic models on different unit parameter sets, determine the extraction ratios of the unit parameter sets within different unit parameter sets, and extract and integrate the unit parameter sets within different unit parameter sets according to the extraction ratios to obtain the training database of the optimal diagnostic model; Use the training database to train and verify the optimal diagnostic model, determine the training and verification weights of each optimal diagnostic model, and fuse all the optimal diagnostic models in parallel according to the training and verification weights to obtain the target diagnostic model; Use the operating parameters of the petrochemical equipment at the current fault as the input of the target diagnostic model, and output the fault index coefficient of the petrochemical equipment. While the model of the present invention can comprehensively train and learn the operating data of the petrochemical equipment, it can more deeply train and learn the relatively matching and adaptable operating data of the petrochemical equipment, improve the accuracy of the diagnostic model, and improve the accuracy of the fault diagnosis of the petrochemical equipment by incorporating multiple models in parallel for fault diagnosis of the petrochemical equipment. Brief Description of the Drawings

[0016] The present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 is the flowchart of the steps of a training method for model fusion based on petrochemical equipment according to an embodiment of the present invention; Figure 2 is the module diagram of a training system for model fusion based on petrochemical equipment according to an embodiment of the present invention. Detailed Embodiments

[0018] In order to make the technical means, creative features, achieved purposes and effects of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0019] Embodiment 1: Please refer to Figure 1 As shown, a training method and system for model fusion based on petrochemical equipment according to an embodiment of the present invention includes the following steps: Step 1: Obtain the basic parameter sets of the petrochemical equipment at different fault times during the historical operation cycle, train and verify the models in the existing model library, identify the diagnostic models that conform to the fault diagnosis of the petrochemical equipment, and integrate them to obtain a diagnostic model library; In step one, the basic parameter group includes the operation data of the petrochemical equipment at different failure moments during the historical operation cycle and the failure index coefficients. Among them, the operation data includes, but is not limited to, the temperature, pressure, flow rate, rotation speed, etc. of the petrochemical equipment. Among them, the failure index coefficients can be the efficiency decline coefficient, product variation coefficient, fatigue damage coefficient, etc. of the petrochemical equipment; It should be noted that there is a basic parameter group at different failure moments of the petrochemical equipment during the historical operation cycle. Among them, each basic parameter group is composed of the operation parameters of the petrochemical equipment at the same failure moment and the failure index coefficients. It should also be noted that the basic parameter group is obtained through the operation record report of the petrochemical equipment; In step one, the identification method of the diagnostic model that conforms to the fault diagnosis of petrochemical equipment is as follows: Based on any one of the existing model libraries; Use the basic parameter groups of the petrochemical equipment at different failure moments during the historical operation cycle to train and verify the model, and compare the failure index coefficients output by the model verification with the failure index coefficients included in the basic parameter group; If the error between the failure index coefficient output by the model verification and the failure index coefficient included in the basic parameter group is within the preset allowable error range, then mark the basic parameter group as a verification matching group; If the error between the failure index coefficient output by the model verification and the failure index coefficient included in the basic parameter group is not within the preset allowable error range, then mark the basic parameter group as a non-verification matching group; After completing the training and verification of the model using the basic parameter groups of the petrochemical equipment at different failure moments during the historical operation cycle, count the quantity ratio of the verification matching groups in the basic parameter group to obtain the verification matching characteristics; If the error between the failure index coefficient output by the model verification and the failure index coefficient included in the basic parameter group is not within the preset allowable error range, then mark the error between the failure index coefficient output by the model verification and the failure index coefficient included in the basic parameter group as the failure index error, calculate the deviation ratio between the failure index error and the endpoint value of the adjacent nearest preset allowable error range, and take the absolute value to obtain the out-of-bounds error ratio of the non-verification matching group. Perform an averaging process on the out-of-bounds error ratios of all non-verification matching groups to obtain the verification deviation characteristics; Exemplarily, assume that the failure index error is xi, the preset allowable error range is [xz - xp], assume that the failure index error xi is not within the preset allowable error range [xz - xp], and xi > xp, then the out-of-bounds error ratio WB is: Perform a difference process on the verification matching characteristics and the verification deviation characteristics to obtain the verification reflection value; In some preferred embodiments, the verification reflection value is compared with the verification reflection threshold; If the verification reflection value is greater than or equal to the verification reflection threshold, no processing is performed; If the verification reflection value is less than the verification reflection threshold, the model is marked as a diagnostic model; All diagnostic models are integrated to obtain a diagnostic model library; It should be noted that the physical meaning represented by the verification reflection value is as follows: the verification reflection value is obtained by taking the difference between the verification matching feature and the verification deviation feature. Among them, the verification matching feature represents the proportion of the number of basic parameter groups that meet the model verification passing, and the larger the proportion, the higher the verification passing rate of the model. The verification deviation feature reflects the verification passing deviation of the model for the basic parameter groups that are not verified as a whole. The larger the verification passing deviation, the worse the verification passing adaptability of the model for the basic parameter groups. The verification reflection value comprehensively reflects the verification performance of the model; Step 2: Through the clustering analysis of different operating parameters of the petrochemical equipment during the current failure, the basic parameter group is split to obtain unit parameter groups, and the unit parameter groups are classified and integrated according to the types of petrochemical equipment operating parameters to obtain a unit parameter set. The unit parameter set is used to train, verify, and test each diagnostic model in the diagnostic model library to determine the optimal diagnostic model corresponding to the unit parameter set; In Step 2, the method for obtaining the unit parameter group is as follows: Obtain different operating parameters of the petrochemical equipment during the current failure, perform clustering analysis using the hierarchical clustering algorithm, and split the operating parameters that are in the same class after clustering analysis from the basic parameter group and combine them to form a unit parameter group; For example, the basic parameter group includes operating parameters such as temperature, pressure, flow rate, and rotational speed. If the clustering result shows that temperature and pressure are operating parameters in the same class, then temperature and pressure are split from the basic parameter group to form a unit parameter group. Among them, the unit parameter group also includes a fault index coefficient, and the fault index coefficient is the fault index coefficient in the split basic parameter group; Among them, the process of performing clustering analysis using the hierarchical clustering algorithm is as follows: Initially, the operating parameters are standardized. After the processing, each operating parameter is regarded as a separate cluster; Calculate the similarity between all clusters, and merge the two clusters with the highest similarity into a new cluster. Among them, the similarity calculation method can be the single-linkage algorithm (nearest neighbor algorithm), the complete-linkage algorithm (farthest neighbor algorithm), or the average-linkage algorithm; Repeat the above steps until all operating parameters are merged into one cluster, then the clustering is completed; In Step 2, the method for obtaining the unit parameter set is as follows: Integrate the unit parameter groups with the same type of operating parameters into a unit parameter set; For example, summarize and integrate the unit parameter groups formed by all temperature and pressure combinations into a unit parameter set; In step two, the process of determining the optimal diagnostic model corresponding to the unit parameter set is as follows: Use the unit parameter groups included in the unit parameter set to conduct training verification tests on each diagnostic model in the diagnostic model library. Based on any one diagnostic model, obtain the fault index coefficient output by the diagnostic model, and calculate the error with the fault index coefficient included in the unit parameter group. If the error is within the preset error range, mark the unit parameter group as a matching unit parameter group; otherwise, if the error is not within the preset error range, mark the unit parameter group as a non-matching unit parameter group; Count the proportion of the number of matching unit parameter groups in the unit parameter set to obtain the verification compliance feature; Mark the error between the fault index coefficient output by the verification of the diagnostic model not within the preset error range and the fault index coefficient included in the unit parameter group as the non-matching index error. Calculate the deviation ratio between the non-matching index error and the endpoint value of the adjacent nearest preset error range, and take the absolute value to obtain the over-limit error ratio of the non-matching unit parameter group. Perform an averaging process on the over-limit error ratios of all non-matching unit parameter groups to obtain the verification non-compliance feature; Obtain the deviation between the verification compliance feature and the verification non-compliance feature to obtain the verification compliance value; Based on any one unit parameter set, obtain the diagnostic model corresponding to the maximum verification compliance value when the unit parameter set conducts training verification tests on each diagnostic model, as the optimal diagnostic model corresponding to the unit parameter set; Exemplarily, assume that the existing diagnostic models include: Diagnostic Model 1, Diagnostic Model 2, Diagnostic Model 3... Diagnostic Model n, where n is the diagnostic model quantity number. Use the unit parameter set to conduct training verification tests on each diagnostic model respectively. Among them, there is a verification compliance value between the unit parameter set and each diagnostic model. Select the diagnostic model corresponding to the maximum verification compliance value as the optimal diagnostic model of the unit parameter set; It should be noted that the calculation and analysis principles of the verification compliance value and the verification reflection value are the same, both reflecting the verification training performance of the model; Step three: Combine the training verification performances of the optimal diagnostic model on different unit parameter sets, determine the extraction ratio of the unit parameter groups in different unit parameter sets, and extract and integrate the unit parameter groups in different unit parameter sets according to the extraction ratio to obtain the training database of the optimal diagnostic model; In step three, the process of obtaining the training database is as follows: Based on any one optimal diagnostic model; Obtain the verification compliance characteristics of the optimal diagnostic model after training, verification, and testing with different unit parameter sets, sum them to obtain the total verification compliance value, calculate the ratio between the verification compliance characteristics and the total verification compliance value to obtain the extraction ratio of the unit parameter group corresponding to the unit parameter set, and extract and integrate the unit parameter groups within each unit parameter set according to the extraction ratio of the unit parameter group corresponding to the unit parameter set to obtain the training database of the optimal diagnostic model; Exemplarily, assume that the verification compliance characteristics of the optimal diagnostic model after training, verification, and testing with different unit parameter sets are YZ1, YZ2, YZ3......YZb respectively, where YZb represents the verification compliance characteristic of the optimal diagnostic model after training, verification, and testing with the b-th unit parameter set, and b represents the numbering of the unit parameter sets. Then, the extraction ratio cqi of the unit parameter group corresponding to the i-th unit parameter set is: where YZi represents the verification compliance characteristic after training, verification, and testing with the i-th unit parameter set; Assume that there are 3 unit parameter sets, and the extraction ratios of the corresponding unit parameter groups are 70%, 20%, and 10% respectively. Then, extract and integrate the unit parameter groups within the 3 unit parameter sets. For the training database of the optimal diagnostic model, the proportions of the extracted unit parameter groups in the training database are 70%, 20%, and 10%; It can be understood that by combining the verification compliance characteristics of the optimal diagnostic model on different unit parameter sets, the extraction ratios of the unit parameter groups within different unit parameter sets are determined, and the unit parameter groups within different unit parameter sets are extracted and integrated according to the extraction ratios to obtain the training database of the optimal diagnostic model. The purpose is to enable the diagnostic model to comprehensively and thoroughly train and learn the operation data of petrochemical equipment, and at the same time, more deeply train and learn the relatively matching and adaptable operation data of petrochemical equipment, thereby improving the accuracy of the diagnostic model; Step Four: Use the training database to train and verify the optimal diagnostic model, determine the training and verification weights of each optimal diagnostic model, and fuse all the optimal diagnostic models in parallel according to the training and verification weights to obtain the target diagnostic model; In Step Four, the process of obtaining the target diagnostic model is as follows: Use the unit parameter groups included in the training database to train and verify the optimal diagnostic model; Based on any one of the optimal diagnostic models, obtain the fault index coefficient output by the optimal diagnostic model, and calculate the error with the fault index coefficient included in the unit parameter group. If the error is within the preset error range, mark the unit parameter group as an adaptable unit parameter group; otherwise, if the error is not within the preset error range, mark the unit parameter group as a non-adaptable unit parameter group; Statistically adapt the parameter group of the unit in the training database to obtain the verification adaptation features; Calculate the ratio of the verification adaptation features to the total verification adaptation value to obtain the training verification weights of the optimal diagnostic model; Among them, the total verification adaptation value is the sum of all verification adaptation features; Add up each optimal diagnostic model combined with its corresponding training verification weight to obtain the target diagnostic model; Exemplarily, assume that the optimal diagnostic models include: Optimal Diagnostic Model 1, Optimal Diagnostic Model 2, Optimal Diagnostic Model 3... Optimal Diagnostic Model z, where z represents the numbering of the optimal diagnostic models, and the training verification weights corresponding to the optimal diagnostic models are q1, q2, q3... qz respectively, where qz represents the training verification weight corresponding to the z-th optimal diagnostic model. Then the obtained target diagnostic model is: Among them, q1 + q2 +... + qz = 1; It can be understood that all the optimal diagnostic models are fused in parallel according to the training verification weights to obtain the target diagnostic model. Different optimal diagnostic models perform outstandingly and have strong adaptability on the unit parameter groups corresponding to different operating data. Through parallel fusion, the performance of each model in its respective advantageous fields can be integrated, comprehensively improving the accuracy of the target diagnostic model in the overall data distribution; Step Five: Use the operating parameters of the petrochemical equipment during the current failure as the input of the target diagnostic model, and output the fault index coefficient of the petrochemical equipment; In Step Five, use the operating parameters of the petrochemical equipment during the current failure as the input of the target diagnostic model, and output through each optimal diagnostic model in the target diagnostic model. The output results of each optimal diagnostic model are combined with the corresponding weights to output the fault index coefficient of the petrochemical equipment; Exemplarily, assume that the fault index coefficients output by Optimal Diagnostic Model 1, Optimal Diagnostic Model 2, Optimal Diagnostic Model 3... Optimal Diagnostic Model z are gc1, gc2, gc3... gcz respectively, where gcz is the fault index coefficient output by the z-th optimal diagnostic model. Then the fault index coefficient ZGX of the petrochemical equipment finally output by the target diagnostic model is: The technical solution of the embodiment of the present invention is as follows: Obtain the basic parameter groups of the petrochemical equipment at different failure times during the historical operation cycle, train and verify the models in the existing model library, identify the diagnostic models that conform to the fault diagnosis of the petrochemical equipment, and integrate them to obtain a diagnostic model library; Through the cluster analysis of different operating parameters of the petrochemical equipment at the current failure, split the basic parameter groups to obtain unit parameter groups, and classify and integrate the unit parameter groups according to the types of operating parameters of the petrochemical equipment to obtain unit parameter sets. Use the unit parameter sets to train, verify and test each diagnostic model in the diagnostic model library to determine the optimal diagnostic model corresponding to the unit parameter set; Combine the training and verification performances of the optimal diagnostic model on different unit parameter sets to determine the extraction ratio of the unit parameter groups in different unit parameter sets, and extract and integrate the unit parameter groups in different unit parameter sets according to the extraction ratio to obtain the training database of the optimal diagnostic model; Use the training database to train and verify the optimal diagnostic model to determine the training and verification weights of each optimal diagnostic model, and fuse all the optimal diagnostic models in parallel according to the training and verification weights to obtain the target diagnostic model; Use the operating parameters of the petrochemical equipment at the current failure as the input of the target diagnostic model, and output the fault index coefficient of the petrochemical equipment. While the model of the present invention can comprehensively train and learn the operating data of the petrochemical equipment, it can more deeply train and learn the relatively matching and adaptable operating data of the petrochemical equipment, improve the accuracy of the diagnostic model, and improve the accuracy of the fault diagnosis of the petrochemical equipment by fusing multiple models in parallel for fault diagnosis of the petrochemical equipment.

[0020] Embodiment 2: A training system based on model fusion of petrochemical equipment, including the following modules: Diagnostic model library acquisition module: Obtain the basic parameter groups of the petrochemical equipment at different failure times during the historical operation cycle, train and verify the models in the existing model library, identify the diagnostic models that conform to the fault diagnosis of the petrochemical equipment, and integrate them to obtain a diagnostic model library; Based on any one of the models in the existing model library; Use the basic parameter groups of the petrochemical equipment at different failure times during the historical operation cycle to train and verify the model, and compare the fault index coefficient output by the model verification with the fault index coefficient included in the basic parameter group; If the error between the fault index coefficient output by the model verification and the fault index coefficient included in the basic parameter group is within the preset allowable error range, mark the basic parameter group as a verification matching group; If the error between the fault index coefficient output by the model verification and the fault index coefficient included in the basic parameter group is not within the preset allowable error range, mark the basic parameter group as a non-verification matching group; After completing the training and verification of the model using the basic parameter sets at different fault times during the historical operation cycle of the petrochemical equipment, count the proportion of the verification matching groups in the basic parameter sets to obtain the verification matching features; If the error between the fault index coefficient output by the model verification and the fault index coefficient included in the basic parameter set is not within the preset allowable error range, mark the error between the fault index coefficient output by the model verification and the fault index coefficient included in the basic parameter set as the fault index error, calculate the deviation ratio between the fault index error and the endpoint value of the adjacent nearest preset allowable error range, and take the absolute value to obtain the out-of-bounds error ratio of the non-verification matching groups. Perform an averaging process on the out-of-bounds error ratios of all non-verification matching groups to obtain the verification deviation features; Perform a difference process on the verification matching features and the verification deviation features to obtain the verification reflection value; If the verification reflection value is less than the verification reflection threshold, mark the model as a diagnostic model; Optimal diagnostic model identification module: Through the clustering analysis of different operating parameters of the petrochemical equipment at the current fault, split the basic parameter set to obtain unit parameter sets, and perform homogeneous classification and integration on the unit parameter sets according to the types of operating parameters of the petrochemical equipment to obtain unit parameter sets. Use the unit parameter sets to perform training, verification, and testing on each diagnostic model in the diagnostic model library to determine the optimal diagnostic model corresponding to the unit parameter sets; Obtain different operating parameters of the petrochemical equipment at the current fault, and use the hierarchical clustering algorithm for clustering analysis. Split the operating parameters that are in the same category after clustering analysis from the basic parameter set and combine them to form unit parameter sets; Among them, the process of performing clustering analysis using the hierarchical clustering algorithm is as follows: Initially, perform standardization processing on the operating parameters. After the processing, regard each operating parameter as a separate cluster; Calculate the similarity between all clusters, and merge the two clusters with the highest similarity into a new cluster. Among them, the similarity calculation method can be the single-linkage algorithm (nearest neighbor algorithm), the complete-linkage algorithm (farthest neighbor algorithm), or the average-linkage algorithm; Repeat the above steps until all operating parameters are merged into one cluster, then the clustering is completed; Integrate the unit parameter sets with the same type of operating parameters into a unit parameter set; Use the unit parameter sets included in the unit parameter sets to perform training, verification, and testing on each diagnostic model in the diagnostic model library. Based on any diagnostic model, obtain the fault index coefficient output by the diagnostic model, and calculate the error with the fault index coefficient included in the unit parameter set. If the error is within the preset error range, mark the unit parameter set as a matching unit parameter set. Otherwise, if the error is not within the preset error range, mark the unit parameter set as a non-matching unit parameter set; Statistically match the number ratio of the unit parameter groups in the unit parameter set to obtain the verification compliance feature; Mark the error between the fault index coefficient output by the verification of the diagnostic model and the fault index coefficient included in the unit parameter group as the non-matching index error, calculate the deviation ratio between the non-matching index error and the adjacent nearest preset error range endpoint value, and take the absolute value to obtain the out-of-bounds error ratio of the non-matching unit parameter group. Perform averaging processing on the out-of-bounds error ratios of all non-matching unit parameter groups to obtain the verification non-compliance feature; Obtain the deviation between the verification compliance feature and the verification non-compliance feature to obtain the verification compliance value; Based on any unit parameter set, obtain the diagnostic model corresponding to the maximum verification compliance value when the unit parameter set is used for training and verification tests of each diagnostic model, and use it as the optimal diagnostic model corresponding to the unit parameter set; Training database construction module: Combine the training and verification performance of the optimal diagnostic model on different unit parameter sets, determine the extraction ratio of the unit parameter groups in different unit parameter sets, and extract and integrate the unit parameter groups in different unit parameter sets according to the extraction ratio to obtain the training database of the optimal diagnostic model; Based on any optimal diagnostic model; Obtain the verification compliance features after the optimal diagnostic model is trained and verified on different unit parameter sets, sum them to obtain the total verification compliance value, calculate the ratio between the verification compliance feature and the total verification compliance value to obtain the extraction ratio of the unit parameter groups corresponding to the unit parameter set, and extract and integrate the unit parameter groups in each unit parameter set according to the extraction ratio of the unit parameter groups corresponding to the unit parameter set to obtain the training database of the optimal diagnostic model; Target model acquisition module: Use the training database to train and verify the optimal diagnostic model, determine the training and verification weights of each optimal diagnostic model, and fuse all optimal diagnostic models in parallel according to the training and verification weights to obtain the target diagnostic model; Use the unit parameter groups included in the training database to train and verify the optimal diagnostic model; Based on any optimal diagnostic model, obtain the fault index coefficient output by the optimal diagnostic model, and calculate the error with the fault index coefficient included in the unit parameter group. If the error is within the preset error range, mark the unit parameter group as an adaptable unit parameter group. Otherwise, if the error is not within the preset error range, mark the unit parameter group as a non-adaptable unit parameter group; Statistically match the number ratio of the adaptable unit parameter groups in the training database to obtain the verification adaptation feature; Calculate the ratio between the verification adaptation feature and the total verification adaptation value to obtain the training and verification weight of the optimal diagnostic model; Among them, the total verification adaptation value is the sum of all verification adaptation features; Add the optimal diagnosis models and their corresponding training and validation weights to obtain the target diagnosis model; Fault diagnosis output module: Use the operating parameters of the petrochemical equipment during the current fault as the input of the target diagnosis model, and output the fault index coefficient of the petrochemical equipment; Use the operating parameters of the petrochemical equipment during the current fault as the input of the target diagnosis model, output through the optimal diagnosis models in the target diagnosis model, and combine the output results of each optimal diagnosis model with the corresponding weights to output the fault index coefficient of the petrochemical equipment.

[0021] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the description in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A training method for model fusion based on petrochemical equipment, characterized by: The following steps are involved: Using the basic parameter groups of petrochemical equipment at different fault moments in the historical operation cycle, the models in the existing model library are trained and verified to obtain a diagnostic model library; Through cluster analysis of different operating parameters of petrochemical equipment during current faults, the basic parameter group is split to obtain the unit parameter group. The unit parameter set obtained by classifying and integrating the unit parameter groups according to the operating parameter type is used to train and verify each diagnostic model in the diagnostic model library to determine the optimal diagnostic model. Combined with the training and verification performance of the optimal diagnostic model on different unit parameter sets, the unit parameter groups in different unit parameter sets are extracted and integrated to obtain the training database of the optimal diagnostic model; Through the training and verification of the training database, the training and verification weights of each optimal diagnostic model are determined, and all the optimal diagnostic models are integrated in parallel to obtain the target diagnostic model; The operating parameters of the petrochemical equipment at the time of the current fault are used as the input of the target diagnosis model, and the fault index coefficient of the petrochemical equipment is output.

2. A training method for model fusion based on petrochemical equipment according to claim 1, characterized in that: The diagnostic model library is obtained in the following manner: The model is trained and verified using the basic parameter groups of petrochemical equipment at different fault moments in the historical operation cycle; By comparing the error between the fault index coefficient output by the model verification and the fault index coefficient contained in the basic parameter group, the basic parameter group is marked as a verification matching group and a non-verification matching group, and the verification matching feature and the verification deviation feature are obtained by processing and analysis, and the deviation analysis of the verification matching feature and the verification deviation feature is performed to obtain the verification reflection value; Models with verification response values ​​less than the verification response threshold are marked as diagnostic models, and the models are summarized to obtain a diagnostic model library.

3. A training method for model fusion based on petrochemical equipment according to claim 2, characterized in that: The verification matching feature and the verification deviation feature are obtained in the following manner: Count the number ratio of the verification matching group in the basic parameter group to obtain the verification matching feature; The error between the fault index coefficient of the model verification output that is not within the preset allowable error range and the fault index coefficient contained in the basic parameter group is marked as the fault index error. The absolute deviation ratio between the fault index error and the nearest endpoint value of the preset allowable error range is calculated to obtain the out-of-bounds error ratio of the non-verification matching group. The out-of-bounds error ratios of all non-verification matching groups are averaged to obtain the verification deviation characteristics.

4. The training method for model fusion based on petrochemical equipment according to claim 1 is characterized in that: The unit parameter set is obtained in the following manner: The different operating parameters of the petrochemical equipment during the current fault are obtained, and cluster analysis is performed using the hierarchical clustering algorithm. The operating parameters that are classified into one category after cluster analysis are split from the basic parameter group and combined to form a unit parameter group. The unit parameter groups with the same operating parameter type are integrated into a unit parameter set.

5. The training method for model fusion based on petrochemical equipment according to claim 1 is characterized in that: The optimal diagnostic model is determined as follows: The unit parameter groups included in the unit parameter set are used to train and verify each diagnostic model in the diagnostic model library, and the fault index coefficients output by the diagnostic model are used for error calculation and analysis with the fault index coefficients included in the unit parameter group. The unit parameter groups are marked as matching unit parameter groups and non-matching unit parameter groups, and the verification compliance features and verification non-compliance features are obtained through processing and analysis. The verification compliance value is obtained through deviation analysis of the verification compliance features and verification non-compliance features. Based on any unit parameter set, the diagnostic model corresponding to the maximum verification compliance value of the unit parameter set when performing training verification tests on each diagnostic model is obtained as the optimal diagnostic model corresponding to the unit parameter set.

6. A training method for model fusion based on petrochemical equipment according to claim 5, characterized in that: The method for obtaining the verification compliance feature and the verification non-compliance feature is as follows: The number ratio of matching unit parameter groups in the unit parameter set is counted to obtain verification compliance characteristics; The error between the fault index coefficient of the diagnostic model verification output that is not within the preset error range and the fault index coefficient contained in the unit parameter group is marked as a non-matching index error. The absolute deviation ratio between the non-matching index error and the nearest adjacent preset error range endpoint value is calculated to obtain the out-of-bounds error ratio of the non-matching unit parameter group. The out-of-bounds error ratios of all non-matching unit parameter groups are averaged to obtain the verification non-conformity feature.

7. The training method for model fusion based on petrochemical equipment according to claim 1 is characterized in that: The process of acquiring the training database is as follows: The verification compliance features of the optimal diagnostic model after training, verification and testing of different unit parameter sets are obtained, and the sum is used to obtain the total verification compliance value. The ratio between the verification compliance features and the total verification compliance value is calculated to obtain the extraction ratio of the unit parameter group corresponding to the unit parameter set. The unit parameter groups in each unit parameter set are extracted and integrated according to the extraction ratio of the unit parameter group corresponding to the unit parameter set to obtain the training database of the optimal diagnostic model.

8. The training method for model fusion based on petrochemical equipment according to claim 1 is characterized in that: The method for determining the training and verification weights of the optimal diagnostic model is: Based on any optimal diagnosis model, the fault index coefficient output by the optimal diagnosis model is obtained, and the error comparison analysis is performed with the fault index coefficient included in the unit parameter group to identify the adaptive unit parameter group in the unit parameter group; Count the number ratio of the adaptive unit parameter groups in the training database to obtain the verified adaptive features; The ratio of the validation adaptation feature to the total validation adaptation value is calculated to obtain the training validation weight of the optimal diagnostic model; Among them, the total value of verified adaptation is the sum of all verified adaptation features; Each optimal diagnostic model is combined with the corresponding training and verification weights and then added to obtain the target diagnostic model.

9. The training method for model fusion based on petrochemical equipment according to claim 1, characterized in that: The method of outputting the fault index coefficient of the petrochemical equipment is: The operating parameters of the petrochemical equipment at the time of the current fault are used as the input of the target diagnosis model, and the output is obtained through each optimal diagnosis model in the target diagnosis model. The output results of each optimal diagnosis model are combined with the corresponding weights to output the fault index coefficient of the petrochemical equipment.

10. A training system based on model fusion of petrochemical equipment, characterized by: include: Diagnostic model library acquisition module: Use the basic parameter groups of petrochemical equipment at different fault moments in the historical operation cycle to train and verify the models in the existing model library to obtain the diagnostic model library; Optimal diagnostic model identification module: Through cluster analysis of different operating parameters of petrochemical equipment at the time of current fault, the basic parameter group is split to obtain the unit parameter group, and the unit parameter set obtained by classifying and integrating the unit parameter group according to the type of operating parameters is used to train and verify each diagnostic model in the diagnostic model library to determine the optimal diagnostic model; Training database construction module: Combine the training and verification performance of the optimal diagnostic model on different unit parameter sets, extract and integrate the unit parameter groups in different unit parameter sets, and obtain the training database of the optimal diagnostic model; Target model acquisition module: Determine the training and verification weights of each optimal diagnostic model through training and verification of the training database, and integrate all the optimal diagnostic models in parallel to obtain the target diagnostic model; Fault diagnosis output module: takes the operating parameters of the petrochemical equipment at the time of the current fault as the input of the target diagnosis model, and outputs the fault index coefficient of the petrochemical equipment.

Images