Lubricating oil aging prediction method based on environmental difference

Abstract

The invention discloses a lubricating oil aging prediction method based on environment difference, and relates to the related field of industrial maintenance, and the method comprises the steps: collecting initial performance parameters of lubricating oil under an initial environment condition, and constructing a lubricating oil initial data set; using environment conditions and using performance parameters are collected on a preset sliding time window node, and a lubricating oil using data set is constructed; taking the lubricating oil initial data set as static input, taking the lubricating oil use data set as dynamic input, and constructing a lubricating oil environment differential aging prediction model based on a long short-term memory network; and inputting the to-be-used environment conditions into the lubricating oil environment differential aging prediction model for lubricating oil aging prediction, and generating and outputting a lubricating oil aging prediction result. The technical problem that the prediction result of an existing lubricating oil aging prediction method is not accurate enough is solved, and the technical effect of improving the accuracy of lubricating oil aging prediction is achieved.

Description

Technical Field

[0001] The present application relates to the field of industrial maintenance, and in particular to a lubricant aging prediction method based on environmental differences. Background Art

[0002] Lubricating oil plays the role of lubrication, cooling, rust prevention and cleaning in mechanical equipment. However, as the use time increases, the lubricating oil will gradually age. After the lubricating oil ages, its performance parameters such as viscosity, flash point, kinematic viscosity and alkalinity will change, resulting in reduced lubrication effect and increased energy consumption. By predicting the aging of lubricating oil, performance problems of the lubricating oil can be discovered and dealt with in a timely manner, maintaining its good performance and thus reducing equipment wear. In the existing technology, the aging prediction of lubricating oil is usually based on simple physical and chemical tests, or statistical models based on historical data. These methods do not fully consider the changes in lubricating oil performance under different environmental conditions. In actual use, the performance of lubricating oil will be affected by various environmental factors such as temperature, humidity, and pressure. Since the existing technology lacks the accurate capture and processing of changes in environmental factors, its prediction results are often not accurate enough.

[0003] In the current related technologies, the lubricant aging prediction method has the technical problem of inaccurate prediction results. Summary of the Invention

[0004] The present application provides a lubricant aging prediction method based on environmental differentials, adopts technical means such as continuously collecting the lubricant's environmental conditions and corresponding performance parameters at preset sliding time window nodes, constructing a lubricant environmental differential aging prediction model, and performing lubricant aging prediction. The lubricant environmental differential aging prediction model learns the changing laws of lubricant performance under different environmental conditions, thereby accurately predicting the performance changes of lubricants under different environmental conditions, and achieving the technical effect of improving the accuracy of lubricant aging prediction.

[0005] This application provides a lubricant aging prediction method based on environmental differences, including: Interactively collect the initial performance parameters of the lubricant under initial environmental conditions and build an initial lubricant data set; Based on the initial environmental conditions, at a preset sliding time window node, the environmental conditions and corresponding performance parameters of the lubricant are collected and acquired to construct a lubricant usage dataset, wherein the environmental conditions and the initial environmental conditions contain the same types of environmental variables, and the values ​​of the environmental variables change over time, and the performance parameters and the initial performance parameters contain the same types of parameters; Using the lubricant initial dataset as static input and the lubricant usage dataset as dynamic input, a lubricant environmental differential aging prediction model is constructed based on a long short-term memory network; The environment conditions to be used are input into the lubricating oil environment differential aging prediction model to perform lubricating oil aging prediction, and a lubricating oil aging prediction result is generated and output.

[0006] In a possible implementation, the lubricant initial dataset is used as a static input, the lubricant usage dataset is used as a dynamic input, and a lubricant environmental differential aging prediction model is constructed based on a long short-term memory network, and the following processing is performed: The lubricant initial data set is used as an initial embedding vector for static input, the use environment conditions in the lubricant use data set are used as dynamic input, and the corresponding use performance parameters are used as output information, and a lubricant use performance parameter predictor is constructed based on long short-term memory network training; Using the usage environment conditions and usage performance parameters in the lubricant usage data set as input information and the remaining service life of the lubricant as output information, a lubricant remaining service life predictor is constructed based on long short-term memory network training; The lubricating oil performance parameter predictor and the lubricating oil remaining service life predictor are sequentially connected to construct a lubricating oil environmental differential aging prediction model.

[0007] In a possible implementation, the use environment conditions in the lubricating oil use data set are used as dynamic input to perform the following processing: The use environment conditions include at least three environmental variables, any two of the environment variables in the use environment conditions are used as target variables, and the environmental variables excluding the target variables in the use environment conditions and the use performance parameter are used as control variables, and a partial correlation analysis is performed to obtain a partial correlation coefficient between the two target variables; If the partial correlation coefficient is greater than a preset partial correlation threshold, the two target variables are interacted with each other as a new dynamic input.

[0008] In a possible implementation, the environmental conditions to be used are input into the lubricating oil environmental differential aging prediction model to perform aging prediction, and a lubricating oil aging prediction result is generated and output, and the following processing is performed: Inputting the environmental conditions to be used into the lubricating oil performance parameter predictor and outputting the predicted performance parameters; Traversing the predicted usage performance parameters to extract a first predicted usage performance parameter; It is determined whether the first predicted performance parameter exceeds a first predicted performance parameter threshold; if so, a lubricant aging warning signal is activated, and a lubricant aging prediction result is generated and output based on the lubricant aging warning signal.

[0009] In a possible implementation, the traversal of the prediction performance parameters performs the following processing: If any of the predicted usage performance parameters does not exceed the corresponding predicted usage performance parameter threshold, the environmental conditions to be used and the predicted usage performance parameters are input into the lubricant remaining service life predictor to predict the lubricant remaining service life, and a lubricant aging prediction result is generated based on the lubricant remaining service life prediction result and output.

[0010] In a possible implementation, the environment conditions to be used and the predicted performance parameters are input into the lubricant remaining service life predictor to predict the lubricant remaining service life, and the following processing is performed: Performing time serialization on the environment condition to be used and the predicted performance parameter to obtain time series data of the environment condition to be used and time series data of the predicted performance parameter; Based on the time series data of the environmental condition to be used, randomly extracting time series data of the first environmental variable to be used; Based on the predicted usage performance parameter time series data, randomly extracting first predicted usage performance parameter time series data; Interacting the first to-be-used environmental variable time series data with the first predicted performance parameter time series data to obtain random interaction time series data; Time trends are captured for the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data, and the remaining service life of the lubricant is predicted based on the trend.

[0011] In a possible implementation, the time trend of the environmental condition time series data to be used, the predicted performance parameter time series data, and the random interaction time series data is captured, and the remaining service life of the lubricant is predicted based on the trend, and the following processing is performed: Capturing trend components, seasonal components, and random components in the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data; Capturing the characteristics of the trend component, seasonal component and random component respectively to obtain trend component characteristics, seasonal component characteristics and random component characteristics; The remaining useful life of the lubricant is predicted based on the trend component characteristics, seasonal component characteristics and random component characteristics in combination with weight distribution.

[0012] In a possible implementation, the environmental conditions to be used are input into the lubricating oil environmental differential aging prediction model to perform lubricating oil aging prediction, and a lubricating oil aging prediction result is generated and outputted, and the following processing is performed: performing data enhancement on the lubricant usage dataset to obtain an enhanced lubricant usage dataset; testing the lubricant environmental differential aging prediction model using the enhanced lubricant usage dataset; The lubricating oil environmental differential aging prediction model is modified according to the test results.

[0013] The lubricant aging prediction method based on environmental differential proposed in this application first interactively collects the initial performance parameters of the lubricant under the initial environmental conditions to construct an initial lubricant data set. Then, based on the initial environmental conditions, at the preset sliding time window node, the lubricant usage environmental conditions and corresponding usage performance parameters are collected to construct a lubricant usage data set, wherein the usage environmental conditions and the initial environmental conditions contain the same type of environmental variables, and the values ​​of the environmental variables change with time, and the usage performance parameters and the initial performance parameters contain the same type of parameters. Then, the lubricant initial data set is used as static input, and the lubricant usage data set is used as dynamic input. A lubricant environmental differential aging prediction model is constructed based on a long short-term memory network. Finally, the environmental conditions to be used are input into the lubricant environmental differential aging prediction model to perform lubricant aging prediction, generate and output lubricant aging prediction results, and achieve the technical effect of improving the accuracy of lubricant aging prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0015] Figure 1 A schematic flow chart of a lubricating oil aging prediction method based on environmental differences provided in an embodiment of the present application; Figure 2 A schematic diagram of a flow chart for predicting the remaining useful life of lubricating oil using the lubricating oil aging prediction method based on environmental differences provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0017] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0018] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0019] The present application embodiment provides a lubricant aging prediction method based on environmental differences, such as Figure 1 As shown, the method includes: Step S100 interactively collects the initial performance parameters of the lubricant under initial environmental conditions to construct an initial lubricant dataset. Specifically, the initial environmental conditions are the environmental conditions of the lubricant before or immediately after use, and the initial performance parameters are the performance parameters exhibited by the lubricant under these initial environmental conditions, used to describe the initial state of the lubricant. The environmental parameters of the lubricant in its initial state, such as temperature, humidity, and pressure, are determined. Using specialized testing equipment or methods, the initial performance parameters of the lubricant under these initial environmental conditions, such as viscosity, acid value, and flash point, are collected. The collected initial environmental conditions and corresponding initial performance parameters are organized into a dataset to form the initial lubricant dataset.

[0020] Step S200, based on the initial environmental conditions, collects and obtains the environmental conditions for use of the lubricant and the corresponding performance parameters at the preset sliding time window nodes to construct a lubricant use data set, wherein the environmental conditions for use and the environmental variables included in the initial environmental conditions are the same type, and the values ​​of the environmental variables change over time, and the performance parameters for use and the parameters included in the initial performance parameters are the same type. Specifically, the preset sliding time window is a preset time period for regularly collecting lubricant usage data, and the preset sliding time window is set according to actual needs. At each preset sliding time window node, the environmental conditions for use of the lubricant (the environmental conditions in which the lubricant is used during actual use) are collected, and the types of environmental variables for the collected environmental conditions are the same as the types of environmental variables included in the initial environmental conditions. That is, if the initial environmental conditions include three environmental variables: temperature, humidity, and pressure, then the collected environmental conditions also include the values ​​of the three environmental variables: temperature, humidity, and pressure. At each time window, we collect the lubricant's performance parameters under the corresponding environmental conditions. These parameters describe the lubricant's actual performance during use and describe its state changes. These parameters cover the same types as the initial performance parameters. The collected environmental conditions and corresponding performance parameters are organized into a data set to form a lubricant usage dataset.

[0021] Step S300 constructs a lubricant environmental differential aging prediction model based on a long short-term memory (LSTM) network, using the initial lubricant dataset as static input and the lubricant usage dataset as dynamic input. Specifically, the initial lubricant dataset and the lubricant usage dataset undergo necessary preprocessing, such as data cleaning and normalization, to meet model training requirements. Leveraging the LSTM network's ability to process time series data, the lubricant environmental differential aging prediction model is constructed. Using the initial lubricant dataset as static input and the lubricant usage dataset as dynamic input, the LSTM network is trained to learn the aging patterns of lubricants under different environmental conditions.

[0022] In one possible implementation, step S300 further includes step S310, wherein a lubricant performance parameter predictor is constructed based on long-short-term memory (LSTM) network training, using the initial lubricant dataset as a static input of an initial embedding vector, the environmental conditions in the lubricant usage dataset as a dynamic input, and the corresponding performance parameters as output information. Specifically, the initial lubricant dataset is converted into an initial embedding vector using methods such as word embedding technology, and a LSTM network is designed, using the initial embedding vector as a static input, the environmental conditions as a dynamic input (changing over time), and the performance parameters corresponding to the environmental conditions as output information. The lubricant performance parameter predictor is trained using the lubricant usage dataset, enabling the LSTM network to learn the correlation between environmental conditions and lubricant performance parameters. Step S320, wherein the environmental conditions and performance parameters in the lubricant usage dataset are input information, and the remaining useful life of the lubricant is output information, is constructed based on the LSTM network training. Specifically, the environmental conditions and performance parameters from the lubricant usage dataset are used as input information. A long-short-term memory (LSTM) network is then designed. This network uses the environmental conditions and performance parameters as input and the remaining useful life of the lubricant (the time or mileage the lubricant can still be used under specific conditions) as output. The lubricant usage dataset is used to train a LSS predictor, enabling the LSTM network to predict the remaining useful life of the lubricant. Step S330 involves sequentially connecting the LSS predictor with the LSS predictor to construct a LSS environmental differential aging prediction model. Specifically, the LSS predictor trained in step S310 and the LSS predictor trained in step S320 are sequentially connected to form a complete LSS environmental differential aging prediction model. This implementation method breaks down lubricant aging prediction into two subtasks: first, predicting the lubricant's performance parameters under different operating conditions, and then predicting the lubricant's remaining service life based on these performance parameters. This divide-and-conquer strategy helps the lubricant environmental differential aging prediction model more accurately learn the complex process of lubricant aging, achieving the technical effect of more accurately predicting lubricant aging.

[0023] In one possible implementation, the environmental conditions in the lubricant usage dataset are used as dynamic inputs. Step S310 further includes step S311, wherein the environmental conditions include at least three environmental variables, and a partial correlation analysis is performed using any two of the environmental variables as target variables. The environmental variables excluding the target variables and the performance parameters are used as control variables to obtain a partial correlation coefficient between the two target variables. Specifically, environmental conditions data are extracted from the lubricant usage dataset. These environmental conditions include at least three environmental variables, such as temperature, humidity, and pressure. Two of the environmental variables are randomly selected from all of the environmental variables as target variables, for example, temperature and humidity are selected as a set of target variables. The remaining environmental variables, excluding the selected target variables, and the performance parameters are used as control variables in the partial correlation analysis to eliminate their potential impact on the relationship between the target variables. A statistical method (such as a partial correlation coefficient formula or statistical software) is used to calculate the partial correlation coefficient between the two target variables. The partial correlation coefficient measures the correlation between the two target variables while holding the other variables constant. In step S312, if the partial correlation coefficient is greater than a preset partial correlation threshold, the two target variables are interacted and used as a new dynamic input. Specifically, the calculated partial correlation coefficient is compared with a preset partial correlation threshold, which is determined based on practical application context and statistical theory to determine whether there is a significant correlation between the two target variables. If the partial correlation coefficient is greater than the preset partial correlation threshold, it indicates that after controlling for other variables, the two target variables still have a significant correlation. In this case, the two target variables are interacted (e.g., by multiplication, ratio, polynomial, etc.) to generate a new variable as a new dynamic input. This interaction further captures the nonlinear relationship between the two target variables. The generated new dynamic input is added to the existing dynamic input set for training the long-short-term memory network. This implementation method uses partial correlation analysis to identify which environmental variables have significant correlations under operating conditions. These correlations may have a significant impact on the performance parameters of the lubricant. The environmental variables with significant correlations are interacted and used as new dynamic inputs, helping the long-short-term memory network model more accurately capture the interactions between these environmental variables, thereby achieving the technical effect of improving the accuracy of lubricant performance parameter prediction.

[0024] In step S400, the environmental conditions to be used are input into the lubricant environmental differential aging prediction model to perform lubricant aging prediction, generating and outputting lubricant aging prediction results. Specifically, the environmental conditions for the lubricant for which lubricant aging prediction is to be performed are input into the trained lubricant environmental differential aging prediction model. The lubricant environmental differential aging prediction model performs lubricant aging prediction based on the input environmental conditions to be used, combined with the learned lubricant aging patterns. The prediction results (predicted lubricant aging conditions or performance indicators) are output in the form of numerical values, charts, etc. for user reference and decision-making. The embodiments of the present application utilize technical means such as continuously collecting lubricant environmental conditions and corresponding performance parameters at preset sliding time window nodes, constructing a lubricant environmental differential aging prediction model, and performing lubricant aging prediction. This enables the lubricant environmental differential aging prediction model to learn the patterns of lubricant performance changes under different environmental conditions, thereby accurately predicting lubricant performance changes under different environmental conditions, achieving the technical effect of improving the accuracy of lubricant aging prediction.

[0025] In one possible implementation, step S400 further includes step S410, inputting the environmental conditions to be used into the lubricant performance parameter predictor and outputting predicted performance parameters. Specifically, the actual environmental conditions of the lubricant to be predicted are input into the trained lubricant performance parameter predictor. Based on the input environmental conditions to be used, the lubricant performance parameter predictor predicts the lubricant performance parameters under those environmental conditions and outputs the predicted performance parameters as prediction results. Step S420, traversing the predicted performance parameters to extract a first predicted performance parameter. Specifically, the predicted performance parameters obtained in step S410 are traversed, and the first (or a specified sequence of) predicted performance parameters are selected as the first predicted performance parameter. Step S430, determining whether the first predicted performance parameter exceeds a first predicted performance parameter threshold. If so, a lubricant aging warning signal is activated, and a lubricant aging prediction result is generated and output based on the lubricant aging warning signal. Specifically, the first predicted usage performance parameter is compared with a preset first predicted usage performance parameter threshold value. The first predicted usage performance parameter threshold value is a preset numerical value used to determine whether the first predicted usage performance parameter exceeds a normal range. If the first predicted usage performance parameter exceeds the first predicted usage performance parameter threshold value, it indicates that the performance of the lubricant has exceeded its safe usage range, and a lubricant aging warning signal is activated. Based on the activated lubricant aging warning signal, a lubricant aging prediction result is generated and output. The lubricant aging prediction result includes the performance status of the lubricant (aging warning) and related performance indicator data. This implementation method achieves the technical effect of avoiding equipment failure or damage due to lubricant aging by predicting the usage performance parameters of the lubricant under different usage environment conditions, monitoring the working status of the lubricant in real time, and issuing timely warnings when performance problems are predicted.

[0026] In one possible implementation, step S400 of traversing the predicted performance parameters further includes step S440: if any of the predicted performance parameters do not exceed the corresponding predicted performance parameter threshold, the predicted performance parameters are input into the lubricant remaining service life predictor to predict the lubricant remaining service life. A lubricant aging prediction result is generated and output based on the lubricant remaining service life prediction result. Specifically, after traversing the predicted performance parameters, a check is performed to determine whether any of the predicted performance parameters exceed their corresponding predicted performance parameter threshold (a value set for each predicted performance parameter to determine whether the predicted performance parameter exceeds a normal range). If all predicted performance parameters do not exceed their corresponding predicted performance parameter threshold, it indicates that the current lubricant performance is within a normal range. At this point, the predicted performance parameters are input into the lubricant remaining service life predictor using the predicted performance parameters as input. The lubricant remaining service life predictor predicts the remaining service life of the lubricant under the inputted environmental conditions and predicted performance parameters. Based on the remaining service life prediction results output by the lubricant remaining service life predictor, a lubricant aging prediction result is generated. This lubricant aging prediction result includes information about the remaining service life of the lubricant and possible aging trends or states. Finally, the generated lubricant aging prediction result is output to the user or system. This implementation method further predicts the remaining service life of the lubricant when all predicted performance parameters are within the normal range, providing the user or system with more comprehensive and accurate lubricant aging prediction information. This achieves the technical effect of helping the user or system understand the remaining service life of the lubricant in advance and formulate appropriate maintenance plans or replacement strategies.

[0027] like Figure 2As shown, in one possible implementation, the inputting of the environmental conditions to be used and the predicted performance parameters into the lubricant remaining useful life predictor to predict the lubricant remaining useful life, step S440, further includes step S441: performing time series conversion on the environmental conditions to be used and the predicted performance parameters to obtain time series data of the environmental conditions to be used and time series data of the predicted performance parameters to be used. Specifically, the environmental conditions to be used (e.g., data that varies over time, such as temperature, humidity, and pressure) are time series converted to generate the time series data of the environmental conditions to be used. Similarly, the predicted performance parameters (e.g., data that varies over time, such as viscosity and acid value) are time series converted to generate the time series data of the predicted performance parameters to be used. Step S442: Based on the time series data of the environmental conditions to be used, randomly extracting time series data of a first environmental variable to be used. Specifically, time series data of an environmental variable (e.g., temperature or humidity) is randomly selected from the time series data of the environmental conditions to be used. Step S443: Based on the time series data of the predicted performance parameters to be used, randomly extracting time series data of the first predicted performance parameter to be used. Specifically, time series data of a performance parameter (such as viscosity or acid value) is randomly selected from the predicted performance parameter time series data. Step S444 involves interacting the first environmental variable time series data to be used with the first predicted performance parameter time series data to obtain randomly interacted time series data. Specifically, the randomly extracted first environmental variable time series data to be used and the first predicted performance parameter time series data are interacted. The interaction can be simple multiplication or addition, or more complex mathematical or statistical operations. Through this interaction, the potential relationship between the environmental variable and the performance parameter is explored. Step S445 involves capturing time trends in the environmental condition time series data to be used, the predicted performance parameter time series data, and the randomly interacted time series data, and predicting the remaining useful life of the lubricant based on these trends. Specifically, time trend analysis is performed on the environmental condition time series data to be used, the predicted performance parameter time series data, and the randomly interacted time series data. Time series analysis techniques (such as moving average and exponential smoothing) are used to identify the trends of these time series data. Based on these trends, a prediction model is used to predict the remaining useful life of the lubricant. This implementation method comprehensively and deeply analyzes the temporal changes in lubricant performance by performing time serialization, random extraction, interaction, and time trend capture on the usage environment conditions and predicted usage performance parameters. Random extraction and interactive analysis increase the diversity and robustness of the lubricant remaining service life predictor, enabling it to adapt to changes in different environments and conditions, achieving the technical effect of improving the accuracy and reliability of lubricant remaining service life prediction.

[0028] In one possible implementation, step S445 further includes step S4451, capturing the trend component, seasonal component, and random component from the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data. Specifically, time series decomposition techniques (such as STL decomposition and X-12-ARIMA seasonal adjustment) are used to decompose the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data into three main components: a trend component (long-term trend), a seasonal component (cyclical variation), and a random component (random fluctuation or error). The trend component represents the long-term trend of the data over time, such as an increase, decrease, or stability; the seasonal component represents cyclical variation in the data, which is related to seasons or periodic events; and the random component represents random fluctuation or error in the data, which is caused by unpredictable factors. Step S4452, the characteristics of the trend component, seasonal component, and random component are captured to obtain trend component characteristics, seasonal component characteristics, and random component characteristics. Specifically, the characteristics of each component (trend component, seasonal component, and random component) are captured. Trend component characteristics include slope, growth rate, or decline rate; seasonal component characteristics include seasonal amplitude and seasonal cycle; and random component characteristics include volatility, skewness, and kurtosis. In step S4453, the remaining useful life of the lubricant is predicted based on the characteristics of the trend, seasonal, and random components, combined with weights assigned. Specifically, a weighted combination is performed based on the captured trend, seasonal, and random component characteristics, combined with preset or learned weights. The remaining useful life of the lubricant is predicted based on the weighted combination results. This implementation method uses time series decomposition technology to decompose the raw data into trend, seasonal, and random components, and captures the characteristics of each component separately. This provides a more detailed understanding and analysis of the different variation patterns in the data, more accurately capturing and simulating the impact of these components on the remaining useful life of the lubricant, thereby achieving the technical effect of improving prediction accuracy.

[0029] In one possible implementation, step S400 further includes step S450, performing data enhancement on the lubricant usage dataset to obtain an enhanced lubricant usage dataset. Specifically, a portion of representative data is selected from the original lubricant usage dataset, covering different usage environment conditions, usage durations, lubricant types, and the like. The selected data is transformed, including rotation, translation, scaling, noise addition, label smoothing, and the like, to increase the diversity and richness of the dataset. New lubricant usage performance parameter data is simulated and generated using existing physical or statistical models. The transformed and simulated data are fused with the original lubricant usage dataset to form an enhanced lubricant usage dataset. Step S460, testing the lubricant environment differential aging prediction model using the enhanced lubricant usage dataset. Specifically, a trained lubricant environment differential aging prediction model is loaded and tested using the enhanced lubricant usage dataset to evaluate the performance of the lubricant environment differential aging prediction model on the enhanced lubricant usage dataset. Evaluation metrics may include accuracy, recall, F1 score, mean squared error, and the like. Step S470, correct the lubricant environment differential aging prediction model based on the test results. Specifically, analyze the performance of the lubricant environment differential aging prediction model on the enhanced lubricant usage dataset, find out the shortcomings of the lubricant environment differential aging prediction model, and adjust the lubricant environment differential aging prediction model based on the performance analysis results, including changing the model structure, adjusting the model parameters, introducing new features, etc. Use the adjusted model structure and parameters to retrain on the enhanced lubricant usage dataset, and repeat the above steps until the performance of the lubricant environment differential aging prediction model on the enhanced lubricant usage dataset reaches the expected level. This implementation method effectively improves the performance of the lubricant environment differential aging prediction model through steps such as data enhancement, model testing and correction, and achieves the technical effect of enabling the lubricant environment differential aging prediction model to more accurately and reliably predict the remaining service life of the lubricant and the aging trend under environmental conditions.

[0030] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A lubricating oil aging prediction method based on environmental difference, characterized in that: The method comprises: Interactively collect the initial performance parameters of the lubricant under initial environmental conditions and build an initial lubricant data set; Based on the initial environmental conditions, at a preset sliding time window node, the environmental conditions and corresponding performance parameters of the lubricant are collected and acquired to construct a lubricant usage dataset, wherein the environmental conditions and the initial environmental conditions contain the same types of environmental variables, and the values ​​of the environmental variables change over time, and the performance parameters and the initial performance parameters contain the same types of parameters; Using the lubricant initial dataset as static input and the lubricant usage dataset as dynamic input, a lubricant environmental differential aging prediction model is constructed based on a long short-term memory network; Inputting the environmental conditions to be used into the lubricating oil environmental differential aging prediction model to perform lubricating oil aging prediction, generating and outputting a lubricating oil aging prediction result; The method uses the lubricant initial data set as static input and the lubricant usage data set as dynamic input to construct a lubricant environmental differential aging prediction model based on a long short-term memory network, including: The lubricant initial data set is used as an initial embedding vector for static input, the use environment conditions in the lubricant use data set are used as dynamic input, and the corresponding use performance parameters are used as output information, and a lubricant use performance parameter predictor is constructed based on long short-term memory network training; Using the usage environment conditions and usage performance parameters in the lubricant usage data set as input information and the remaining service life of the lubricant as output information, a lubricant remaining service life predictor is constructed based on long short-term memory network training; The lubricant oil performance parameter predictor and the lubricant oil remaining service life predictor are sequentially connected to construct a lubricant oil environmental differential aging prediction model; The step of using the usage environment conditions in the lubricating oil usage dataset as dynamic input includes: The use environment conditions include at least three environmental variables, any two of the environment variables in the use environment conditions are used as target variables, and the environmental variables excluding the target variables in the use environment conditions and the use performance parameter are used as control variables, and a partial correlation analysis is performed to obtain a partial correlation coefficient between the two target variables; If the partial correlation coefficient is greater than a preset partial correlation threshold, the two target variables are interacted with each other as a new dynamic input.

2. The lubricating oil aging prediction method based on environmental difference according to claim 1, characterized in that: The method of inputting the environmental conditions to be used into the lubricating oil environmental differential aging prediction model to perform aging prediction, generating and outputting the lubricating oil aging prediction result, includes: Inputting the environmental conditions to be used into the lubricating oil performance parameter predictor and outputting the predicted performance parameters; Traversing the predicted usage performance parameters to extract a first predicted usage performance parameter; It is determined whether the first predicted performance parameter exceeds a first predicted performance parameter threshold; if so, a lubricant aging warning signal is activated, and a lubricant aging prediction result is generated and output based on the lubricant aging warning signal.

3. The lubricating oil aging prediction method based on environmental difference according to claim 2, characterized in that: The method of traversing the predicted usage performance parameters further includes: If any of the predicted usage performance parameters does not exceed the corresponding predicted usage performance parameter threshold, the environmental conditions to be used and the predicted usage performance parameters are input into the lubricant remaining service life predictor to predict the lubricant remaining service life, and a lubricant aging prediction result is generated based on the lubricant remaining service life prediction result and output.

4. The lubricating oil aging prediction method based on environmental difference according to claim 3, characterized in that: The method of inputting the environmental conditions to be used and the predicted performance parameters into the lubricating oil remaining service life predictor to predict the remaining service life of the lubricating oil includes: Performing time serialization on the environment condition to be used and the predicted performance parameter to obtain time series data of the environment condition to be used and time series data of the predicted performance parameter; Based on the time series data of the environmental condition to be used, randomly extracting time series data of the first environmental variable to be used; Based on the predicted usage performance parameter time series data, randomly extracting first predicted usage performance parameter time series data; Interacting the first to-be-used environmental variable time series data with the first predicted performance parameter time series data to obtain random interaction time series data; Time trends are captured for the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data, and the remaining service life of the lubricant is predicted based on the trend.

5. The lubricating oil aging prediction method based on environmental difference according to claim 4, characterized in that: The method of capturing time trends of the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data, and predicting the remaining service life of the lubricant based on the trend, includes: Capturing trend components, seasonal components, and random components in the time series data of the environmental conditions to be used, the time series data of the predicted performance parameters, and the random interaction time series data; Capturing the characteristics of the trend component, seasonal component and random component respectively to obtain trend component characteristics, seasonal component characteristics and random component characteristics; The remaining useful life of the lubricant is predicted based on the trend component characteristics, seasonal component characteristics and random component characteristics in combination with weight distribution.

6. The lubricating oil aging prediction method based on environmental difference according to claim 1, characterized in that: The method further comprises: inputting the environmental conditions to be used into the lubricating oil environmental differential aging prediction model to perform lubricating oil aging prediction, generating and outputting a lubricating oil aging prediction result; performing data enhancement on the lubricant usage dataset to obtain an enhanced lubricant usage dataset; testing the lubricant environmental differential aging prediction model using the enhanced lubricant usage dataset; The lubricating oil environmental differential aging prediction model is modified according to the test results.

Images