Working condition anomaly detection method and device, equipment and storage medium
By classifying the data to be detected by the equipment and building a data reconstruction model, calculating the differential data to judge abnormal working conditions, the problem of lack of abnormal data in the prior art is solved, and the safety and reliability of the equipment are improved.
Patent Information
- Application Number
- CN202411336938.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-24
- Publication Date
- 2025-01-14
AI Technical Summary
The prior art is difficult to detect equipment abnormal operating conditions based on only health data, and the lack of abnormal data leads to difficulty in detection.
By obtaining the data to be detected and classifying it to obtain the data to be detected for each working condition, the data reconstruction model constructed based on historical health data is used to reconstruct the data, calculate the difference between the reconstructed data and the original data, and based on the differential data is different. Determine whether the working condition is abnormal.
It realizes abnormal working condition detection of equipment with only health data, improving the safety and reliability of equipment.
Smart Images
Figure CN119312129A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of data processing technology, and in particular to a method, device, equipment and storage medium for detecting abnormal working conditions. Background Art
[0002] In modern industry and manufacturing, equipment health status monitoring and accident prevention are key to ensuring production safety and efficiency. However, many devices can only provide data on healthy status, but lack data on abnormal status, which makes anomaly detection based on traditional methods difficult. This is because traditional methods usually rely on a large amount of abnormal data for training to achieve accurate equipment abnormal condition detection. Therefore, how to detect equipment abnormal conditions based on only healthy data is a technical problem that technical personnel need to solve urgently. Summary of the invention
[0003] In view of this, the present disclosure proposes an abnormal operating condition detection method, apparatus, device and storage medium, which can realize abnormal operating condition detection of equipment based on only health data.
[0004] According to a first aspect of the present disclosure, a method for detecting abnormal operating conditions is provided, comprising:
[0005] Acquire the data to be detected, and classify the data to be detected to obtain the data to be detected corresponding to each type of working condition;
[0006] The data reconstruction models corresponding to various working conditions are used to reconstruct the data to be detected corresponding to various working conditions to obtain the reconstructed data corresponding to various working conditions;
[0007] Calculate the difference data between the data to be tested and the reconstructed data corresponding to various working conditions;
[0008] Determine whether various operating conditions are abnormal based on the difference data;
[0009] Among them, the data reconstruction models corresponding to various working conditions are constructed based on the historical health data under various working conditions.
[0010] In a possible implementation, when the data to be detected is classified to obtain the data to be detected corresponding to each type of working condition, it includes:
[0011] The kmeans algorithm is used to classify the data to be detected to obtain clusters corresponding to various working conditions;
[0012] Based on the clusters corresponding to various working conditions, the data to be detected corresponding to various working conditions are determined.
[0013] In one possible implementation, when constructing data reconstruction models corresponding to various working conditions based on historical health data under various working conditions, it includes:
[0014] Obtain historical health data and model training parameters under various working conditions;
[0015] Based on the historical health data and model training parameters under various working conditions, the basic data reconstruction model is trained to obtain the data reconstruction models corresponding to various working conditions.
[0016] In a possible implementation, when determining whether various operating conditions are abnormal based on the difference data, it includes:
[0017] Obtain the residual thresholds for various working conditions;
[0018] Determine whether the difference data is greater than the residual threshold of the corresponding operating condition: when it is greater than the residual threshold of the corresponding operating condition, determine that the corresponding operating condition is abnormal; when it is less than or equal to the residual threshold of the corresponding operating condition, determine that the corresponding operating condition is normal.
[0019] In a possible implementation, the residual thresholds for each operating condition are determined based on the clusters corresponding to each operating condition.
[0020] In a possible implementation, when determining the residual threshold of each operating condition based on the clusters corresponding to each operating condition, it includes:
[0021] Based on the clusters corresponding to each working condition, the maximum distance within each class corresponding to each working condition is calculated;
[0022] The maximum distance within each class corresponding to each working condition is normalized to obtain the normalized maximum distance within each class corresponding to each working condition;
[0023] Based on the normalized maximum distance within the class corresponding to each working condition and the preset adjustment parameters, the residual threshold corresponding to each working condition is determined.
[0024] According to a second aspect of the present disclosure, a device for detecting abnormal working conditions is provided, comprising:
[0025] A data acquisition module is used to acquire the data to be detected, and classify the data to be detected to obtain the data to be detected corresponding to various working conditions;
[0026] A data reconstruction module is used to reconstruct the data to be detected corresponding to various working conditions by using data reconstruction models corresponding to various working conditions to obtain reconstructed data corresponding to various working conditions;
[0027] A difference calculation module, used to calculate the difference data between the to-be-detected data and the reconstructed data corresponding to various working conditions;
[0028] A data detection module, used to determine whether various operating conditions are abnormal based on the difference data;
[0029] Among them, the data reconstruction models corresponding to various working conditions are constructed based on the historical health data under various working conditions.
[0030] In one possible implementation, the data detection module includes:
[0031] Threshold acquisition module, used to obtain residual thresholds for various working conditions;
[0032] The operating condition detection module is used to determine whether the difference data is greater than the residual threshold of the corresponding operating condition: when it is greater than the residual threshold of the corresponding operating condition, the corresponding operating condition is determined to be abnormal; when it is less than or equal to the residual threshold of the corresponding operating condition, the corresponding operating condition is determined to be normal.
[0033] According to a third aspect of the present disclosure, an abnormal operating condition detection device is provided, comprising: a processor; and a memory for storing processor executable instructions; wherein the processor is configured to execute the method described in the first aspect of the present disclosure.
[0034] According to a fourth aspect of the present disclosure, a non-volatile computer-readable storage medium is provided, on which computer program instructions are stored, wherein the computer program instructions, when executed by a processor, implement the method described in the first aspect of the present disclosure.
[0035] The present disclosure provides a method, device, equipment and storage medium for detecting abnormal working conditions, the method comprising obtaining data to be detected, and classifying the data to be detected to obtain data to be detected corresponding to various working conditions; reconstructing the data to be detected corresponding to various working conditions using data reconstruction models corresponding to various working conditions to obtain reconstructed data corresponding to various working conditions; calculating the difference data between the data to be detected and the reconstructed data corresponding to various working conditions; determining whether various working conditions are abnormal based on the difference data; wherein the data reconstruction models corresponding to various working conditions are constructed based on historical health data under various working conditions. Through the above detection method, abnormal working condition detection of equipment can be realized when only health data is available, thereby improving the safety and reliability of the equipment.
[0036] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.
[0038] Figure 1 A flow chart showing a method for detecting abnormal operating conditions according to an embodiment of the present disclosure is shown;
[0039] Figure 2A schematic block diagram of an abnormal operating condition detection device according to an embodiment of the present disclosure is shown;
[0040] Figure 3 A schematic block diagram of an abnormal operating condition detection device according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION
[0041] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.
[0042] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.
[0043] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.
[0044] <Method Example>
[0045] Figure 1 FIG. 1 is a flow chart showing a method for detecting abnormal operating conditions according to an embodiment of the present disclosure. Figure 1 As shown, the method includes steps S1100-S1400:
[0046] S1100, obtain the data to be tested, and classify the data to be tested to obtain the data to be tested corresponding to each working condition. The data to be tested refers to a data set consisting of the health data of the equipment operation, which is composed of the parameters of different components collected by sensors, and may include temperature, pressure, flow, etc. For example, the equipment includes 4 components a, b, c, d, each of which contains 10, 15, 20, and 25 characteristic parameters, and their characteristics can be expressed as A = {a 1 ,a 2 ,...,a 10},B={b 1 ,b 2 ,...,b 15}, C={c 1 ,c 2 ,...,c 20}, D = {d 1 ,d 2 ,...,d 25}, the final data to be tested is M = {A, B, C, D}. It should be pointed out that the data to be tested is a data set consisting of healthy operating data without abnormal data.
[0047] When classifying the data to be detected, it can be implemented based on clustering algorithms such as self-organizing map (SOM), DBCAN density clustering, and kmeans algorithm, which is not specifically limited here.
[0048] It should be noted here that since the distance from each cluster center to each point in the cluster after classification is needed in the subsequent detection steps, and from the algorithm principle, the kmeans algorithm can directly obtain the cluster center without the need for clustering and then calculating, so when classifying the data to be detected, the kmeans clustering algorithm can be used as a priority.
[0049] When the data to be detected are classified based on the kmeans clustering algorithm to obtain the data to be detected corresponding to each type of working condition, the following steps may be included:
[0050] First, the kmeans clustering algorithm is used to classify the data to be detected and obtain clusters corresponding to various working conditions. Specifically, the following steps may be included:
[0051] Firstly, the kmeans clustering algorithm is used to classify the data to be detected and obtain multiple initial clusters, each of which corresponds to a working condition.
[0052] Secondly, determine whether the working condition corresponding to each cluster cluster is a normal working condition or a variable working condition. Specifically, traverse each cluster cluster, and for the current cluster cluster traversed, determine whether the number of data to be detected included in the current cluster cluster is greater than or equal to the preset number threshold. If it is greater than or equal to the number threshold, determine that the working condition corresponding to the current cluster cluster is a normal working condition, otherwise, determine that the working condition corresponding to the current cluster cluster is a variable working condition. Among them, the normal working condition refers to a stable working condition under normal operating conditions. For example, in the normal operating scenario of a ship, the normal working condition may include acceleration conditions, steering conditions, deceleration conditions, etc. Variable working conditions refer to transition conditions between any two normal working conditions. For example, when the normal working condition includes acceleration conditions, steering conditions, and deceleration conditions, the conditions of transition from acceleration conditions to steering conditions, the conditions of transition from acceleration conditions to deceleration conditions, and the conditions of transition from steering conditions to deceleration conditions are all variable conditions. The data quantity threshold is set according to the ranking of the number of clusters, and according to the statistical running time of normal working conditions, it is ensured that the sum of the sample numbers of clusters within the threshold accounts for more than 85% of the total sample number. Preferably, the quantity threshold can be set to 150,000. Finally, the clusters corresponding to each variable working condition are merged to form a cluster corresponding to a variable working condition, and the clusters corresponding to normal working conditions are not merged. Finally, the clusters corresponding to multiple normal working conditions and one cluster corresponding to a variable working condition are collectively referred to as clusters corresponding to various working conditions.
[0053] It should be noted here that as the number of normal operating conditions increases, the number of variable operating conditions will increase significantly. For example, when there are 15 normal operating conditions, the corresponding variable operating conditions will include 105. If a data reconstruction model is constructed for each variable operating condition in the later stage, the task of building the data reconstruction model will be large. However, the data between the variable operating conditions are similar. Therefore, multiple variable operating conditions can be merged into one variable operating condition. In this way, only one data reconstruction model corresponding to the variable operating condition needs to be constructed later, which greatly reduces the task of building the data reconstruction model.
[0054] Second, based on the clusters corresponding to each operating condition, the data to be detected corresponding to each operating condition is determined. Specifically, for each operating condition, all the data to be detected included in the clusters corresponding to each operating condition are used as the data to be detected corresponding to each operating condition.
[0055] S1200, reconstructing the to-be-detected data corresponding to each operating condition by using a data reconstruction model corresponding to each operating condition, and obtaining reconstructed data corresponding to each operating condition.
[0056] It should be noted here that before executing step S1200, it is necessary to first construct a data reconstruction model corresponding to each type of working condition.
[0057] In a possible implementation, the data reconstruction model corresponding to each working condition is constructed based on the historical health data under each working condition, which may specifically include the following steps:
[0058] First, obtain historical health data and model training parameters under various working conditions.
[0059] Specifically, when obtaining historical health data under various working conditions, the following steps may be included:
[0060] First, a set number of historical health data of the equipment is obtained. The historical health data refers to the historical operation data collected when the equipment is operating normally and without failure. For example, for a new steam turbine with a service life of more than 20 years but no more than three years in use, it has not had any failures within a three-year period. At this time, its historical operation data within three years of operation is the historical health data of the new steam turbine.
[0061] Secondly, the set number of historical health data are classified to obtain the historical health data under various working conditions. The specific classification process is shown in S1100 and will not be described in detail here.
[0062] The model training parameters refer to important factors that affect the model training effect. Among them, the model training parameters may include: at least one of batch size, number of training rounds (Epoch), learning rate (LearningRate) and optimizer (Optimizer), and multiple optional values are pre-configured for each model training parameter, wherein the optional value of each parameter can be set according to the hardware resources and the pre-experimental results of the sub-dataset generated using downsampling. For example, four optional values of 64, 128, 256, and 512 can be configured for batch size (Batch Size). Four optional values of 50, 100, 150, and 200 can be configured for the number of training rounds (Epoch). Four optional values of 0.01, 0.005, 0.001, and 0.0001 can be configured for the learning rate (LearningRate). Three optional values of stochastic gradient descent (SGD), Adam, and RMSprop can be configured for the optimizer (Optimizer).
[0063] Second, based on the historical health data and model training parameters under various working conditions, the basic data reconstruction model is trained to obtain the data reconstruction models corresponding to various working conditions. Among them, the basic data reconstruction model can adopt an abnormality detection algorithm model such as LSTM autoencoder model, isolation forest algorithm model or principal component analysis algorithm model, which is not specifically limited here. Since the LSTM autoencoder model has better performance in most scenarios, the LSTM autoencoder model is preferred as the basic data reconstruction model.
[0064] Before training the basic data reconstruction model, it is necessary to first construct multiple model training parameter groups based on the obtained model training parameters. Specifically, the optional values of each model training parameter are traversed and combined to obtain multiple groups of model training parameter groups. Continuing with the above embodiment, the batch size is 64, the number of training rounds is 50, the learning rate is 0.01, and the optimizer takes stochastic gradient descent (SGD) to obtain a group of model training parameter groups. Next, the batch size is 128, the number of training rounds is 50, the learning rate is 0.01, and the optimizer takes stochastic gradient descent (SGD) to obtain another group of model training parameter groups. Next, the batch size is 256, the number of training rounds is 50, the learning rate is 0.01, and the optimizer takes stochastic gradient descent (SGD) to obtain another group of model training parameter groups. By analogy, 192 groups of model training parameter groups will be obtained in the end.
[0065] After constructing multiple model training parameter groups, the basic data reconstruction model is trained based on the historical health data under various working conditions and multiple model training parameter groups to obtain the data reconstruction models corresponding to various working conditions. Since the process of the data reconstruction model corresponding to various working conditions is the same, the following takes one working condition (hereinafter referred to as the current working condition) as an example to illustrate the training process of the corresponding data reconstruction model. The specific training steps are as follows:
[0066] Traverse each model training parameter group.
[0067] For the traversed current model training parameter group, the basic data reconstruction model is trained based on the historical health data under the current working conditions and the current model training parameter group to obtain the data reconstruction model corresponding to the current model training parameter group. Each historical health data under the current working conditions is input into the data reconstruction model corresponding to the current model training parameter group, so as to reconstruct the reconstructed historical health data corresponding to each historical health data through the data reconstruction model corresponding to the current model training parameter group. Based on each historical health data and the corresponding reconstructed historical health data, the average absolute error of the data reconstruction model corresponding to the current model training parameter group is calculated. The average absolute error can be expressed by the formula:
[0068]
[0069] Among them, y i is the i-th historical health data, y i ′ is the reconstructed historical health data corresponding to the i-th historical health data, n is the number of historical health data under the current working condition, and MAE is the mean absolute error.
[0070] After the traversal is completed, the data reconstruction model corresponding to each model training parameter group and the mean absolute error of the corresponding data reconstruction model can be obtained.
[0071] The data reconstruction model with the smallest mean absolute error is selected as the data reconstruction model corresponding to the current working condition.
[0072] By referring to the above steps, the data reconstruction model corresponding to various working conditions can be obtained.
[0073] After completing the construction of the data reconstruction model corresponding to each working condition, the data reconstruction model corresponding to each working condition can be used to reconstruct the data to be detected corresponding to each working condition to obtain the reconstructed data corresponding to each working condition. Among them, the reconstruction process of the data reconstruction model on the data to be detected includes: the first step is to input the data to be detected into the encoder of the data reconstruction model, and convert the data to be detected into a low-dimensional latent space representation. The second step is to input the low-dimensional latent space representation into the decoder, and the decoder reconstructs the reconstructed data corresponding to the data to be detected based on the low-dimensional latent space representation.
[0074] S1300, calculating the difference data between the data to be detected and the reconstructed data corresponding to each working condition. Specifically, calculating the residuals between the data to be detected and the corresponding reconstructed data for each working condition, and using the residuals as the difference data between the data to be detected and the reconstructed data corresponding to each working condition. The residuals are represented by the mean absolute error, which can be expressed as:
[0075]
[0076] Among them, y i is the i-th data to be detected, y i ′ is the reconstructed data corresponding to the i-th data to be detected, n is the number of corresponding data to be detected under each working condition, and MAE is the residual value between the data to be detected and the corresponding reconstructed data.
[0077] S1400, determining whether various operating conditions are abnormal based on the difference data. Specifically, the following steps may be included:
[0078] First, obtain the residual threshold of each working condition. The residual threshold is determined according to the clustering of each working condition. Specifically, the steps can be as follows:
[0079] First, based on the clusters corresponding to each working condition, the maximum distance within the cluster corresponding to each working condition is calculated. Specifically, based on the clusters corresponding to each working condition, the distance from each data in the cluster to the cluster center is calculated, and the maximum distance is selected based on the distance from each data to the cluster center as the maximum distance within the cluster corresponding to each working condition.
[0080] Secondly, the maximum distance within each class corresponding to each working condition is normalized to obtain the normalized maximum distance within each class corresponding to each working condition. Specifically, in order to enable the maximum distance within different clusters to be compared, they are normalized. Generally, the maximum distance within all clusters can be normalized to a range of 0 to 1.
[0081] Next, based on the normalized maximum distance within each class and the preset adjustment parameters corresponding to each working condition, the residual threshold corresponding to each working condition is determined. Among them, for each working condition, the calculation formula of the corresponding residual threshold is as follows:
[0082] Residual threshold = α*Dnorm
[0083] In the formula, α is the preset adjustment parameter, and Dnorm is the normalized maximum distance within the class corresponding to each working condition.
[0084] Under normal circumstances, the data to be tested for each working condition and the historical health data during training have similar data distributions. At this time, α = 1, and the residual threshold is Dnorm. If there is a deviation between the data to be tested and the historical health data during training, the residual threshold can be adjusted by adjusting α. Finally, the residual thresholds for various working conditions are obtained.
[0085] Second, determine whether the difference data is greater than the residual threshold of the corresponding working condition: when it is greater than the residual threshold of the corresponding working condition, determine that the corresponding working condition is abnormal; when it is less than or equal to the residual threshold of the corresponding working condition, determine that the corresponding working condition is normal.
[0086] Therefore, the present disclosure provides a method for detecting abnormal working conditions, which includes: obtaining data to be detected, and classifying the data to be detected to obtain data to be detected corresponding to various working conditions; reconstructing the data to be detected corresponding to various working conditions using data reconstruction models corresponding to various working condition types, and obtaining reconstructed data of the data to be detected corresponding to various working conditions; calculating the difference data between the data to be detected and the reconstructed data corresponding to various working conditions and the corresponding working condition data; determining whether various working conditions are abnormal based on the difference data, and whether the working condition data is healthy data or abnormal data; wherein the data reconstruction models corresponding to various working conditions are constructed based on historical health data under various working condition categories. The above abnormality detection method can realize the detection of abnormal working conditions based on only healthy data, which effectively improves the reliability and safety of the equipment.
[0087] <Device Example>
[0088] Figure 2 FIG. 1 is a schematic block diagram of an abnormal operating condition detection device according to an embodiment of the present disclosure. Figure 2As shown, the device 100 includes: a data acquisition module 110, a data reconstruction module 120, a difference calculation module 130, and a data detection module 140. The data acquisition module 110 is used to acquire the data to be detected, and classify the data to be detected to obtain the data to be detected corresponding to each type of working condition; the data reconstruction module 120 is used to reconstruct the data to be detected corresponding to each type of working condition using a data reconstruction model corresponding to each type of working condition, and obtain the reconstructed data of the data to be detected corresponding to each type of working condition; the difference calculation module 130 is used to calculate the difference data between the data to be detected and the reconstructed data corresponding to each type of working condition and the corresponding working condition class data; the data detection module 140 is used to determine whether each type of working condition is abnormal based on the difference data, and whether the working condition class data is healthy data or abnormal data; wherein, the data reconstruction model corresponding to each type of working condition is constructed based on the historical health data under each type of working condition category.
[0089] In one possible implementation, the data detection module 140 includes: a threshold acquisition module, used to obtain the residual threshold of each type of operating condition; an operating condition detection module, used to determine whether the difference data is greater than the residual threshold of the corresponding operating condition: when it is greater than the residual threshold of the corresponding operating condition, the corresponding operating condition is determined to be abnormal; when it is less than or equal to the residual threshold of the corresponding operating condition, the corresponding operating condition is determined to be normal.
[0090] <Equipment Embodiment>
[0091] Figure 3 FIG. 1 is a schematic block diagram of an abnormal operating condition detection device according to an embodiment of the present disclosure. Figure 3 As shown, the device A 200 includes: a processor 210 and a memory 220 for storing executable instructions of the processor 210. The processor 210 is configured to implement any of the above-mentioned abnormal working condition detection methods when executing the executable instructions.
[0092] Here, it should be noted that the number of processors 210 may be one or more. At the same time, in the abnormal working condition detection device 200 of the embodiment of the present disclosure, an input device 230 and an output device 240 may also be included. Among them, the processor 210, the memory 220, the input device 230 and the output device 240 may be connected through a bus or in other ways, which are not specifically limited here.
[0093] The memory 220 is a computer-readable storage medium that can be used to store software programs, computer executable programs, and various modules, such as the program or module corresponding to the abnormal working condition detection method of the embodiment of the present disclosure. The processor 210 executes various functional applications and data processing of the A device 200 by running the software programs or modules stored in the memory 220.
[0094] The input device 230 may be used to receive input numbers or signals. The signals may be key signals related to user settings and function control of the device / terminal / server. The output device 240 may include display devices such as display screens.
[0095] <Storage Medium Embodiment>
[0096] According to a fourth aspect of the present disclosure, a non-volatile computer-readable storage medium is also provided, on which computer program instructions are stored. When the computer program instructions are executed by the processor 210, any of the above-mentioned abnormal operating condition detection methods is implemented.
[0097] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.
Claims
1. A method for detecting abnormal working conditions, characterized in that: include: Acquire the data to be detected, and classify the data to be detected to obtain the data to be detected corresponding to each type of working condition; The data reconstruction models corresponding to various working conditions are used to reconstruct the data to be detected corresponding to various working conditions to obtain the reconstructed data corresponding to various working conditions; Calculate the difference data between the data to be tested and the reconstructed data corresponding to various working conditions; Determine whether various operating conditions are abnormal based on the difference data; Among them, the data reconstruction models corresponding to various working conditions are constructed based on the historical health data under various working conditions.
2. The method according to claim 1, characterized in that When the data to be detected are classified to obtain the data to be detected corresponding to various working conditions, it includes: The kmeans algorithm is used to classify the data to be detected to obtain clusters corresponding to various working conditions; Based on the clusters corresponding to various working conditions, the data to be detected corresponding to various working conditions are determined.
3. The method according to claim 1, characterized in that: When constructing data reconstruction models corresponding to various working conditions based on historical health data under various working conditions, it includes: Obtain historical health data and model training parameters under various working conditions; Based on the historical health data and model training parameters under various working conditions, the basic data reconstruction model is trained to obtain the data reconstruction models corresponding to various working conditions.
4. The method according to claim 1, characterized in that: When determining whether various operating conditions are abnormal based on the difference data, it includes: Obtain the residual thresholds for various working conditions; Determine whether the difference data is greater than the residual threshold of the corresponding operating condition: when it is greater than the residual threshold of the corresponding operating condition, determine that the corresponding operating condition is abnormal; when it is less than or equal to the residual threshold of the corresponding operating condition, determine that the corresponding operating condition is normal.
5. The method according to claim 4, characterized in that The residual thresholds for each operating condition are determined based on the clusters corresponding to each operating condition.
6. The method according to claim 5, characterized in that When determining the residual thresholds for various working conditions based on the clusters corresponding to various working conditions, it includes: Based on the clusters corresponding to each working condition, the maximum distance within each class corresponding to each working condition is calculated; The maximum distance within each class corresponding to each working condition is normalized to obtain the normalized maximum distance within each class corresponding to each working condition; Based on the normalized maximum distance within the class corresponding to each working condition and the preset adjustment parameters, the residual threshold corresponding to each working condition is determined.
7. A device for detecting abnormal working conditions, characterized in that: include: A data acquisition module is used to acquire the data to be detected, and classify the data to be detected to obtain the data to be detected corresponding to various working conditions; A data reconstruction module is used to reconstruct the data to be detected corresponding to various working conditions by using data reconstruction models corresponding to various working conditions to obtain reconstructed data corresponding to various working conditions; A difference calculation module, used to calculate the difference data between the to-be-detected data and the reconstructed data corresponding to various working conditions; A data detection module, used to determine whether various operating conditions are abnormal based on the difference data; Among them, the data reconstruction models corresponding to various working conditions are constructed based on the historical health data under various working conditions.
8. The device according to claim 7, characterized in that The data detection module comprises: Threshold acquisition module, used to obtain residual thresholds for various working conditions; The operating condition detection module is used to determine whether the difference data is greater than the residual threshold of the corresponding operating condition: when it is greater than the residual threshold of the corresponding operating condition, the corresponding operating condition is determined to be abnormal; when it is less than or equal to the residual threshold of the corresponding operating condition, the corresponding operating condition is determined to be normal.
9. An abnormal working condition detection device, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the method described in any one of claims 1 to 6 when executing the executable instructions.
10. A non-volatile computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.