Oil quality anomaly prediction method, device and equipment

Abstract

The invention provides an oil quality anomaly prediction method, device and equipment, and the method comprises the steps: collecting oil quality data and feature data of different equipment, and constructing a feature index data set; training a GRU network model based on an attention mechanism by using the feature index data set; acquiring personalized feature data of a single user, and processing the personalized feature data to obtain a personalized data set with the same structure as the feature index data set; performing personalized training adjustment on the GRU network model by using the personalized data set to obtain a personalized oil quality anomaly prediction model; the to-be-detected oil quality is input to the personalized oil quality anomaly prediction model, the prediction result of the to-be-detected oil quality is output, high-quality anomaly prediction is performed through the personalized oil quality anomaly prediction model, only the to-be-detected oil quality needs to be input in real time, and compared with a manual sampling mode, the method has the advantages that timeliness is better, the oil quality prediction result is faster, and the accuracy is higher. And the problem of lagging is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of oil quality prediction, and in particular to a method, device and equipment for predicting abnormal oil quality. Background Art

[0002] Oil supervision in power plants refers to the chemical supervision and management of various oils used in power plants, such as turbine oil, insulating oil and fire-resistant oil. The main purpose is to ensure that the quality of these oils meets the relevant standards, thereby ensuring the safe and economical operation of equipment. Traditionally, the work of oil supervision in power plants is mainly to conduct index testing through regular manual sampling.

[0003] However, the manual sampling method has poor timeliness, resulting in delayed maintenance of oil quality. Summary of the invention

[0004] The present invention provides an oil quality abnormality prediction method, device and equipment, which are used to solve the defects of poor timeliness and time lag in manual oil quality detection in the prior art.

[0005] In a first aspect, the present invention provides a method for predicting abnormal oil quality, comprising:

[0006] Collect oil quality data and characteristic data of different equipment and build characteristic index data set;

[0007] Using the feature index dataset, training a GRU network model based on the attention mechanism;

[0008] Collecting personalized feature data of a single user, processing the personalized feature data, and obtaining a personalized data set with the same structure as the feature indicator data set;

[0009] Using the personalized data set, the GRU network model is trained and adjusted in a personalized manner to obtain a personalized oil quality abnormality prediction model;

[0010] The oil quality to be tested is input into the personalized oil quality abnormality prediction model, and the prediction result of the oil quality to be tested is output.

[0011] According to an oil quality abnormality prediction method provided by the present invention, the high-quality data includes normal oil quality data and abnormal oil quality data, and the characteristic data includes individual characteristics, environmental information and behavioral information:

[0012] The constructing of the characteristic indicator data set includes:

[0013] By using multiple interpolation, the normal oil quality data, the abnormal oil quality data, the individual characteristics, the environmental information and the behavioral information are analyzed and merged, and the observation rows of the incomplete data set are estimated and filled n times to generate n complete data sets;

[0014] Analyze the n complete data sets respectively, integrate the results from each filled data set, and obtain statistically inferred data;

[0015] The data are normalized and smoothed to obtain a characteristic index data set.

[0016] According to an oil quality abnormality prediction method provided by the present invention, the characteristic index data set is used to train a GRU network model based on an attention mechanism, including:

[0017] Sending the feature index data set into the GRU network model with attention mechanism through the update gate and the reset gate;

[0018] Divide the feature index data set into a training set and a validation set in a ratio of 7:3, and set a maximum number of training times;

[0019] The GRU network model with the attention mechanism is trained using the training set and the validation set, and if the number of training times is greater than the maximum number of training times, the GRU network model is output.

[0020] A method for predicting oil quality anomaly according to the present invention further includes:

[0021] If the training times are less than the maximum training times, obtaining the LOSS value;

[0022] Based on the LOSS value, the Adam optimizer is used to adjust the network parameters, and training is performed again until the number of training times is greater than the maximum number of training times.

[0023] A method for predicting oil quality anomaly according to the present invention further includes:

[0024] Using the test set to verify the output GRU network model, comparing the true value with the predicted value;

[0025] Based on the comparison result, the accuracy of the GRU network model is determined.

[0026] According to an oil quality abnormality prediction method provided by the present invention, after collecting personalized feature data of a single user and processing the personalized feature data to obtain a personalized data set with the same structure as the feature index data set, the method further includes:

[0027] The personalized data set is stored in the cloud to build a personalized user profile.

[0028] According to a method for predicting oil quality anomaly provided by the present invention, before constructing the characteristic index data set, the method further includes:

[0029] Performing time stamp synchronization on the oil quality data and characteristic data collected from different devices to obtain a consistent data source for different devices in the time dimension;

[0030] Data aggregation is performed on the data sources at different time points to reduce the amount of data.

[0031] A method for predicting oil quality anomaly according to the present invention further includes:

[0032] Based on the prediction results, determining the type of oil quality anomaly;

[0033] When the type is hydraulic oil abnormality or oil quality deterioration, the control automation program processes the abnormal oil quality;

[0034] When the type is oil sludge carbon deposition or fire-resistant oil quality deterioration, an instruction for manual intervention adjustment is issued to deal with the abnormal oil quality.

[0035] In a second aspect, the present invention further provides an oil quality abnormality prediction device, comprising:

[0036] The acquisition module is used to collect oil quality data and characteristic data of different equipment and build a characteristic indicator data set;

[0037] A training module, used to train a GRU network model based on an attention mechanism using the feature indicator dataset;

[0038] A synchronization module, used to collect personalized feature data of a single user, process the personalized feature data, and obtain a personalized data set with the same structure as the feature indicator data set;

[0039] A personalization module, used to use the personalized data set to perform personalized training and adjustment on the GRU network model to obtain a personalized oil quality abnormality prediction model;

[0040] The prediction module is used to input the oil quality to be detected into the personalized oil quality abnormality prediction model and output the prediction result of the oil quality to be detected.

[0041] In a third aspect, the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned methods for predicting abnormal oil quality is implemented.

[0042] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned oil quality abnormality prediction methods.

[0043] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned oil quality abnormality prediction methods.

[0044] The present invention provides an oil quality anomaly prediction method, device and equipment, which include collecting oil quality data and feature data of different equipment to construct a feature index data set; using the feature index data set to train a GRU network model based on an attention mechanism; collecting personalized feature data of a single user, processing the personalized feature data to obtain a personalized data set with the same structure as the feature index data set; using the personalized data set to perform personalized training and adjustment on the GRU network model to obtain a personalized oil quality anomaly prediction model; inputting the oil quality to be detected into the personalized oil quality anomaly prediction model, and outputting a prediction result of the oil quality to be detected. Since high-quality anomaly prediction is performed by using the personalized oil quality anomaly prediction model, it is only necessary to input the oil quality to be detected in real time. Compared with the manual sampling method, it is more timely and the oil quality prediction result is faster, which effectively solves the problem of lag. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 is a schematic flow chart of the oil quality abnormality prediction method provided in this embodiment;

[0047] Figure 2 is a schematic diagram of the structure of the oil quality abnormality prediction device provided in this embodiment;

[0048] Figure 3 It is a schematic diagram of the structure of the electronic device provided in this embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Figure 1 Schematic diagram of the process of predicting abnormal oil quality provided in this embodiment.

[0051] like Figure 1 As shown, the oil quality abnormality prediction method provided by the embodiment of the present invention mainly includes the following steps:

[0052] 101. Collect oil quality data and characteristic data of different equipment and build a characteristic indicator data set.

[0053] In a specific implementation process, firstly, we collect training samples, and according to the complexity of oil quality changes of different equipment, we refine the oil quality change related parameters, environmental background interference, and individual characteristic elements to form a variable set, and quantify the variables according to the characteristics of each variable. Therefore, it is necessary to collect normal oil quality data and abnormal oil quality data from different equipment, and collect individual characteristics, environmental information, and behavioral information. Among them, individual characteristics include: the role of the equipment, operating conditions, and index adjustment; environmental information includes: ambient temperature, humidity, and dust during measurement; behavioral information includes: whether oil purification technology is used, whether oil regeneration technology is used.

[0054] In the process of collecting all normal oil quality data, abnormal oil quality data and characteristic data, some data are not recorded, omitted or lost due to human factors, resulting in missing values, so the data needs to be preprocessed. For missing values, the multiple imputation method is used, including interpolation, analysis, and merging. The observation rows of the incomplete data set are estimated and filled n times to generate n complete data sets, and the n complete data sets are analyzed separately. The results from each filled data set are integrated to produce the final statistical inference. After obtaining the complete data set, each characteristic data is normalized and smoothed. Finally, a complete characteristic indicator data set can be obtained.

[0055] 102. Use the feature indicator dataset to train the GRU network model based on the attention mechanism.

[0056] The attention mechanism is used to analyze the relationship and importance level between equipment oil quality changes and various variables. The attention mechanism is introduced into the GRU model. The attention mechanism selects important information from a large amount of information and focuses on these important information, ignoring unimportant information. The weight represents the importance of information.

[0057] Therefore, the feature index dataset is first sent to the GRU network model with attention mechanism through the update gate and reset gate, where the update gate and reset gate are used for information transmission. Then, the feature index dataset is divided into training set and validation set in a ratio of 7:3, and the maximum number of training times is set; the GRU network model with attention mechanism is trained using the training set and validation set, and if the number of training times is greater than the maximum number of training times, the GRU network model is output. If the number of training times is less than the maximum number of training times, the LOSS value is obtained; based on the LOSS value, the network parameters are adjusted using the Adam optimizer, and training is performed again until the number of training times is greater than the maximum number of training times.

[0058] The final trained network is stored as a weight matrix, which is the pre-trained GRU network model. The output GRU network model is then verified using the test set to compare the actual value with the predicted value. Based on the comparison results, the accuracy of the GRU network model is determined.

[0059] 103. Collect personalized feature data of a single user, process the personalized feature data, and obtain a personalized data set having the same structure as the feature indicator data set.

[0060] Different users have different usage patterns, which will result in different personalized feature data. Therefore, the personalized feature data of a single user is collected in a targeted manner and then processed independently, including multiple interpolation, analysis, and merging. The observation rows are estimated and filled n times to obtain complete personalized feature data. Finally, normalization and smoothing are performed to obtain a personalized data set with the same structure as the feature indicator data set.

[0061] Personalized data sets are used to characterize the results of different users’ characteristic habits.

[0062] 104. Using personalized data sets, the GRU network model is trained and adjusted to obtain a personalized oil quality abnormality prediction model.

[0063] The personalized data set is used to perform personalized training and adjustment on the pre-trained GRU network model, so that the personalized oil quality abnormality prediction model obtained after personalized training can better meet the oil quality prediction needs of different users. Compared with a single network model, the different models trained with personalized data sets have more accurate and reliable results.

[0064] 105. Input the oil quality to be tested into the personalized oil quality abnormality prediction model, and output the prediction result of the oil quality to be tested.

[0065] After obtaining the personalized oil quality abnormality prediction model, you only need to input the oil quality to be tested into it to predict the oil quality, so as to quickly obtain the oil quality analysis results. When the oil quality is abnormal or is about to be abnormal, corresponding processing can be carried out in time, reducing the abnormal impact of oil quality on the equipment.

[0066] This also includes collecting current oil quality data as the oil quality to be detected through sensors, then determining the user device identifier corresponding to the oil quality to be detected, filtering the corresponding personalized oil quality abnormality prediction model through the user device identifier, and finally inputting the oil quality to be detected into the corresponding personalized oil quality abnormality prediction model to ensure the accuracy of the oil quality result prediction.

[0067] Furthermore, based on the above embodiment, this embodiment also includes: storing personalized data sets in the cloud to build a personalized user profile.

[0068] By uploading and archiving the user's personalized data, it is possible to analyze different users, flexibly adjust the equipment corresponding to different users, determine different maintenance times, and specifically ensure the safety and scientificity of the equipment suitable for different users.

[0069] Furthermore, based on the above embodiment, this embodiment also includes: synchronizing timestamps of oil quality data and feature data collected from different devices to obtain consistent data sources for different devices in the time dimension; and aggregating data sources at different time points to reduce the amount of data.

[0070] Specifically, through timestamp synchronization, we ensure that data from different data sources remain consistent in time dimension, which is convenient for subsequent analysis. Then, through data aggregation, data from multiple data sources or data from different time points of the same data source are aggregated according to certain rules to reduce the amount of data and improve analysis efficiency.

[0071] Furthermore, based on the above-mentioned embodiment, the present embodiment also includes: determining the type of oil quality abnormality based on the prediction result; when the type is hydraulic oil abnormality or oil quality deterioration, controlling the automatic program to process the abnormal oil quality; when the type is sludge carbon deposition or fire-resistant oil quality deterioration, issuing a manual intervention adjustment instruction to process the abnormal oil quality.

[0072] Specifically, different oil quality prediction results correspond to different corresponding means. Therefore, according to different results, the type of oil quality abnormality is first determined, and then a multi-level response mechanism is executed based on the abnormality type. Different response levels and processing procedures are set according to the severity and type of the abnormality. When the type is hydraulic oil abnormality or oil quality deterioration, it indicates that the abnormal problem is relatively simple and can be automatically adjusted through automatic control to save labor and ensure processing efficiency. When the type is sludge carbon deposition or fire-resistant oil quality deterioration, manual intervention is required at this time. Through manual inspection and maintenance, a comprehensive inspection of the equipment or oil quality is guaranteed to ensure normal oil quality and efficient operation of the equipment.

[0073] Based on the same general inventive concept, the present invention also protects an oil quality anomaly prediction device. The oil quality anomaly prediction device provided by the present invention is described below. The oil quality anomaly prediction device described below and the oil quality anomaly prediction method described above can be referenced to each other.

[0074] Figure 2 Schematic diagram of the structure of the oil quality abnormality prediction device provided in this embodiment.

[0075] like Figure 2 As shown, this embodiment provides an oil quality abnormality prediction device, comprising:

[0076] The collection module 201 is used to collect oil quality data and characteristic data of different equipment and construct a characteristic index data set;

[0077] A training module 202 is used to train a GRU network model based on an attention mechanism using a feature index data set;

[0078] Synchronization module 203, used to collect personalized feature data of a single user, process the personalized feature data, and obtain a personalized data set with the same structure as the feature indicator data set;

[0079] A personalization module 204 is used to perform personalization training and adjustment on the GRU network model using a personalization data set to obtain a personalization oil quality abnormality prediction model;

[0080] The prediction module 205 is used to input the oil quality to be detected into the personalized oil quality abnormality prediction model, and output the prediction result of the oil quality to be detected.

[0081] Furthermore, the high-quality data in this embodiment includes normal oil quality data and abnormal oil quality data, and the characteristic data includes individual characteristics, environmental information and behavioral information:

[0082] The acquisition module 201 is specifically used for:

[0083] By using multiple interpolation, the normal oil quality data, the abnormal oil quality data, the individual characteristics, the environmental information and the behavioral information are analyzed and merged, and the observation rows of the incomplete data set are estimated and filled n times to generate n complete data sets;

[0084] Analyze the n complete data sets respectively, integrate the results from each filled data set, and obtain statistically inferred data;

[0085] The data are normalized and smoothed to obtain a characteristic index data set.

[0086] Furthermore, the training module 202 in this embodiment is specifically used for:

[0087] Sending the feature index data set into the GRU network model with attention mechanism through the update gate and the reset gate;

[0088] Divide the feature index data set into a training set and a validation set in a ratio of 7:3, and set a maximum number of training times;

[0089] Using the training set and the validation set, the GRU network model with the attention mechanism is trained, and if the number of training times is greater than the maximum number of training times, the GRU network model is output;

[0090] If the training times are less than the maximum training times, obtaining the LOSS value;

[0091] Based on the LOSS value, the Adam optimizer is used to adjust the network parameters, and training is performed again until the number of training times is greater than the maximum number of training times.

[0092] Furthermore, the training module 202 in this embodiment is also specifically used for:

[0093] Using the test set to verify the output GRU network model, comparing the true value with the predicted value;

[0094] Based on the comparison result, the accuracy of the GRU network model is determined.

[0095] Furthermore, the present embodiment also includes a storage module, which is used to:

[0096] The personalized data set is stored in the cloud to build a personalized user profile.

[0097] Furthermore, this embodiment also includes a synchronization module, which is used to:

[0098] Performing time stamp synchronization on the oil quality data and characteristic data collected from different devices to obtain a consistent data source for different devices in the time dimension;

[0099] Data aggregation is performed on the data sources at different time points to reduce the amount of data.

[0100] Furthermore, this embodiment also includes a prompt module, which is used to:

[0101] Based on the prediction results, determining the type of oil quality anomaly;

[0102] When the type is hydraulic oil abnormality or oil quality deterioration, the control automation program processes the abnormal oil quality;

[0103] When the type is oil sludge carbon deposition or fire-resistant oil quality deterioration, an instruction for manual intervention adjustment is issued to deal with the abnormal oil quality.

[0104] Figure 3 It is a schematic diagram of the structure of the electronic device provided in this embodiment.

[0105] like Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the oil quality abnormality prediction method, which includes: collecting oil quality data and feature data of different devices to construct a feature index data set; using the feature index data set to train a GRU network model based on an attention mechanism; collecting personalized feature data of a single user, processing the personalized feature data to obtain a personalized data set with the same structure as the feature index data set; using the personalized data set to perform personalized training and adjustment on the GRU network model to obtain a personalized oil quality abnormality prediction model; inputting the oil quality to be detected into the personalized oil quality abnormality prediction model, and outputting the prediction result of the oil quality to be detected.

[0106] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0107] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the oil quality abnormality prediction method provided by the above-mentioned methods, and the method includes: collecting oil quality data and feature data of different equipment to construct a feature indicator data set; using the feature indicator data set to train a GRU network model based on an attention mechanism; collecting personalized feature data of a single user, processing the personalized feature data to obtain a personalized data set with the same structure as the feature indicator data set; using the personalized data set to perform personalized training and adjustment on the GRU network model to obtain a personalized oil quality abnormality prediction model; inputting the oil quality to be detected into the personalized oil quality abnormality prediction model, and outputting the prediction result of the oil quality to be detected.

[0108] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the oil quality abnormality prediction method provided by the above-mentioned methods, the method comprising: collecting oil quality data and feature data of different equipment to construct a feature indicator data set; using the feature indicator data set to train a GRU network model based on an attention mechanism; collecting personalized feature data of a single user, processing the personalized feature data to obtain a personalized data set with the same structure as the feature indicator data set; using the personalized data set to perform personalized training and adjustment on the GRU network model to obtain a personalized oil quality abnormality prediction model; inputting the oil quality to be detected into the personalized oil quality abnormality prediction model, and outputting a prediction result of the oil quality to be detected.

[0109] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0110] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting oil quality anomaly, characterized in that: include: Collect oil quality data and characteristic data of different equipment and build characteristic index data set; Using the feature index dataset, training a GRU network model based on the attention mechanism; Collecting personalized feature data of a single user, processing the personalized feature data, and obtaining a personalized data set with the same structure as the feature indicator data set; Using the personalized data set, the GRU network model is trained and adjusted in a personalized manner to obtain a personalized oil quality abnormality prediction model; The oil quality to be tested is input into the personalized oil quality abnormality prediction model, and the prediction result of the oil quality to be tested is output.

2. The oil quality abnormality prediction method according to claim 1, characterized in that: The high-quality data includes normal oil quality data and abnormal oil quality data, and the characteristic data includes individual characteristics, environmental information and behavioral information: The constructing of the characteristic indicator data set includes: By using multiple interpolation, the normal oil quality data, the abnormal oil quality data, the individual characteristics, the environmental information and the behavioral information are analyzed and merged, and the observation rows of the incomplete data set are estimated and filled n times to generate n complete data sets; Analyze the n complete data sets respectively, integrate the results from each filled data set, and obtain statistically inferred data; The data are normalized and smoothed to obtain a characteristic index data set.

3. The oil quality abnormality prediction method according to claim 1, characterized in that: The method of using the feature index data set to train a GRU network model based on an attention mechanism includes: Sending the feature index data set into the GRU network model with attention mechanism through the update gate and the reset gate; Divide the feature index data set into a training set and a validation set in a ratio of 7:3, and set a maximum number of training times; The GRU network model with the attention mechanism is trained using the training set and the validation set, and if the number of training times is greater than the maximum number of training times, the GRU network model is output.

4. The oil quality abnormality prediction method according to claim 3, characterized in that: Also includes: If the training times are less than the maximum training times, obtaining the LOSS value; Based on the LOSS value, the Adam optimizer is used to adjust the network parameters, and training is performed again until the number of training times is greater than the maximum number of training times.

5. The oil quality abnormality prediction method according to claim 3, characterized in that: Also includes: Using the test set to verify the output GRU network model, comparing the true value with the predicted value; Based on the comparison result, the accuracy of the GRU network model is determined.

6. The oil quality abnormality prediction method according to claim 1, characterized in that: After the personalized characteristic data of a single user is collected and the personalized characteristic data is processed to obtain a personalized data set with the same structure as the characteristic indicator data set, the method further includes: The personalized data set is stored in the cloud to build a personalized user profile.

7. The oil quality abnormality prediction method according to any one of claims 1 to 6, characterized in that: Before constructing the feature indicator data set, the method further includes: Performing time stamp synchronization on the oil quality data and characteristic data collected from different devices to obtain a consistent data source for different devices in the time dimension; Data aggregation is performed on the data sources at different time points to reduce the amount of data.

8. The method for predicting abnormal oil quality according to any one of claims 1 to 6, characterized in that: Also includes: Based on the prediction results, determining the type of oil quality anomaly; When the type is hydraulic oil abnormality or oil quality deterioration, the control automation program processes the abnormal oil quality; When the type is oil sludge carbon deposition or fire-resistant oil quality deterioration, an instruction for manual intervention adjustment is issued to deal with the abnormal oil quality.

9. An oil quality abnormality prediction device, characterized in that: include: The acquisition module is used to collect oil quality data and characteristic data of different equipment and build a characteristic indicator data set; A training module, used to train a GRU network model based on an attention mechanism using the feature indicator dataset; A synchronization module, used to collect personalized feature data of a single user, process the personalized feature data, and obtain a personalized data set with the same structure as the feature indicator data set; A personalization module, used to use the personalized data set to perform personalized training and adjustment on the GRU network model to obtain a personalized oil quality abnormality prediction model; The prediction module is used to input the oil quality to be detected into the personalized oil quality abnormality prediction model and output the prediction result of the oil quality to be detected.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the oil quality abnormality prediction method according to any one of claims 1 to 8 is implemented.

Images