A Machine Learning-Based Prediction Method and System for Low Oil Pressure Faults
By training a fault prediction model using machine learning models and using oil pressure data features to predict future faults, the problem of hysteresis detection in existing technologies is solved, enabling early warning and safety prevention.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- DONGFENG COMML VEHICLE CO LTD
- Filing Date
- 2023-08-14
- Publication Date
- 2026-05-26
AI Technical Summary
Existing technology can only detect faults when the oil pressure is low, which is delayed and cannot predict or warn in advance, posing a safety hazard.
By using machine learning models and leveraging vehicle oil pressure data features such as vehicle speed, engine speed, and coolant temperature, a fault prediction model is trained to predict potential low oil pressure faults in the near future and to send out real-time warnings.
It enables early prediction of low oil pressure faults, promptly reminding users to change the oil and avoiding driving safety issues caused by the fault.
Smart Images
Figure CN117131957B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of engine oil (lubricating oil) maintenance, and specifically to a method and system for predicting low oil pressure faults based on machine learning. Background Technology
[0002] Engine oil pressure testing is a crucial technology that helps drivers detect engine-related faults early and prevent unexpected situations while driving. Currently, the methods for testing engine oil pressure generally include:
[0003] (1) The oil pressure is detected in real time by the value of the sensor to determine whether the oil needs to be changed.
[0004] (2) The oil pressure is judged by the service life of the oil, and then the conclusion is drawn as to whether the oil needs to be changed.
[0005] (3) The condition of the engine oil (including oil pressure) is determined by analyzing the chemical composition inside the engine oil, so as to conclude whether the engine oil needs to be changed.
[0006] However, the above methods can only detect low oil pressure, meaning that the low oil pressure fault has already occurred before the detection. This means that the above methods can only work when the vehicle has already malfunctioned, which has a certain delay and poses certain safety risks. For example, the driver may cause a safety problem because he cannot change the oil in time. Summary of the Invention
[0007] In view of the deficiencies in the existing technology, the technical problem solved by the present invention is: how to predict in advance that a low oil pressure failure will occur in the future, and thus give users a timely warning to avoid safety problems caused by the low oil pressure failure.
[0008] To achieve the above objectives, the present invention provides a machine learning-based method for predicting low oil pressure faults, comprising the following steps: obtaining oil pressure samples within the prediction period, with each oil pressure sample corresponding to a vehicle; obtaining a fault prediction model by learning and training all oil pressure samples through a machine learning model; and obtaining a fault prediction result by calculating the oil pressure data of the vehicle under test using the fault prediction model.
[0009] Based on the above technical solution, the oil pressure sample includes: oil pressure data within the prediction period when the current oil pressure is low; oil pressure data within the prediction period when the current oil pressure is not low; each oil pressure data includes several aggregated features, and the aggregated feature values are obtained by aggregating all feature values of the feature within the aggregation period.
[0010] Based on the above technical solution, the aggregation features and feature values include:
[0011] Vehicle speed, engine speed, oil pressure, water temperature, and ambient temperature are all characteristic values: the maximum, minimum, and average values within the polymerization cycle.
[0012] Mileage, characterized by mileage within the aggregation period;
[0013] Travel time, with the characteristic value being: the travel time within the aggregation period;
[0014] Fuel consumption per 100 kilometers, characterized by: fuel consumption per 100 kilometers for each trip within the aggregation period.
[0015] Based on the above technical solution, the process of obtaining a fault prediction model by learning and training all oil pressure samples through a machine learning model includes: dividing all oil pressure samples into a training sample set and a test sample set; training the training sample set with an XGBoost model to obtain a fault prediction training model; and when the fault prediction result obtained by the fault prediction training model after calculating the test sample set meets the requirements, the fault prediction training model is used as the fault prediction model.
[0016] Based on the above technical solution, the process of obtaining the fault prediction result by calculating the oil pressure data of the vehicle under test through the fault prediction model includes: after receiving the fault prediction request of the vehicle, obtaining the oil pressure data of the vehicle, the composition of the oil pressure data is the same as that of the oil pressure sample; and obtaining the fault prediction result by calculating the oil pressure data of the vehicle under test through the fault prediction model.
[0017] The present invention provides a machine learning-based prediction system for low oil pressure faults, comprising an oil pressure data acquisition module, a fault prediction model training module, and a fault prediction module.
[0018] The oil pressure data acquisition module is used to: acquire oil pressure data and transmit it to the fault prediction model training module;
[0019] The fault prediction model training module is used to: learn and train a fault prediction model on all oil pressure samples within the prediction period through a machine learning model; and calculate the fault prediction result by using the oil pressure data of the vehicle under test through the fault prediction model.
[0020] The fault prediction module is used to receive and display the fault prediction results returned by the fault prediction model training module.
[0021] Based on the above technical solution, the oil pressure sample includes: oil pressure data within the prediction period when the current oil pressure is low, and oil pressure data within the prediction period when the current oil pressure is not low; each oil pressure data includes several aggregated features, and the aggregated feature values are obtained by aggregating all feature values of the feature within the aggregation period.
[0022] Based on the above technical solution, the aggregation features and feature values include:
[0023] Vehicle speed, engine speed, oil pressure, water temperature, and ambient temperature are all characteristic values: the maximum, minimum, and average values within the polymerization cycle.
[0024] Mileage, characterized by mileage within the aggregation period;
[0025] Travel time, with the characteristic value being: the travel time within the aggregation period;
[0026] Fuel consumption per 100 kilometers, characterized by: fuel consumption per 100 kilometers for each trip within the aggregation period.
[0027] Based on the above technical solution, the process of the fault prediction model training module to obtain a fault prediction model by learning and training all oil pressure samples through a machine learning model includes: dividing all oil pressure samples into a training sample set and a test sample set; training the training sample set with an XGBoost model to obtain a fault prediction training model; and when the fault prediction result obtained by the fault prediction training model after calculating the test sample set meets the requirements, the fault prediction training model is used as the fault prediction model.
[0028] Based on the above technical solution, the process of the fault prediction model training module to obtain the fault prediction result by calculating the oil pressure data of the vehicle under test through the fault prediction model includes: after receiving the fault prediction request of the vehicle, obtaining the oil pressure data of the vehicle, the composition of the oil pressure data is the same as that of the oil pressure sample; and obtaining the fault prediction result by calculating the oil pressure data of the vehicle under test through the fault prediction model.
[0029] Compared with the prior art, the advantages of the present invention are as follows:
[0030] This invention can use machine learning to train a fault prediction model based on relevant oil pressure data. The fault prediction model can calculate and push the probability of a vehicle's oil pressure failure in the future in real time, thereby providing timely warnings to users, reminding them of potential faults, and guiding customers to change the oil in advance to avoid driving safety problems caused by oil pressure-related faults. Attached Figure Description
[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0032] Figure 1 This is a flowchart illustrating the learning and training process of the fault prediction model in this embodiment of the invention.
[0033] Figure 2 This is a schematic diagram of the workflow of the machine learning-based prediction system for low oil pressure faults in an embodiment of the present invention. Detailed Implementation
[0034] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0035] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.
[0036] The method for predicting low oil pressure faults based on machine learning in this embodiment of the invention includes the following steps: taking a car as a unit, obtaining oil pressure samples (i.e., data related to oil pressure, with one sample corresponding to each car) within a prediction period (e.g., predicting a fault will occur in 3 days). After learning and training all oil pressure samples through a machine learning model, a fault prediction model is obtained. After calculating the oil pressure data of the vehicle under test using the fault prediction model, a fault prediction result is obtained. The fault prediction result includes whether a low oil pressure fault will occur after the prediction period and whether a low oil pressure fault will not occur after the prediction period.
[0037] Therefore, this invention can obtain a fault prediction model by performing machine learning training based on relevant data of engine oil pressure. Through the fault prediction model, the probability of engine oil pressure failure in the future can be calculated and pushed in real time, thereby enabling timely warning to users, reminding them of potential fault problems, and guiding customers to change the engine oil in advance to avoid driving safety problems caused by engine oil pressure-related faults.
[0038] Preferably, in the above method, one oil pressure sample includes:
[0039] Predict oil pressure data for the next period when the current oil pressure is low;
[0040] Predicts oil pressure data for the current period when the current oil pressure is not low.
[0041] As shown in Table 1, this embodiment uses sample labels to distinguish whether the oil pressure is low. A positive sample label indicates that the oil pressure is not low, represented by 0, and a negative sample label indicates that the oil pressure is low, represented by 1. The oil pressure data includes data related to and possibly related to the oil pressure, which is obtained by filtering from a large number of vehicle data.
[0042] Table 1. Composition of Sample Data
[0043] Is the current oil pressure low? Sample screening Positive / Negative Samples Tags (1 / 0) yes Vehicle data from the 3 days prior to the malfunction negative samples 1 no Vehicle data from the 3 days prior to the malfunction Positive samples 0
[0044] Each oil pressure data point includes several aggregated features. The aggregated feature values are obtained by aggregating all feature values within an aggregation period (e.g., 1 day). For example, the aggregated features corresponding to the oil pressure data might be vehicle speed, engine speed, and coolant temperature. In this case, vehicle speed represents the maximum, minimum, and average vehicle speed within the aggregation period.
[0045] Preferably, as shown in Table 2, the polymerization cycle in this embodiment is 1 day, and the polymerization characteristics and characteristic values mainly include:
[0046] Table 2. Aggregation characteristics and eigenvalues
[0047] Feature name Eigenvalues Speed Maximum, minimum, and average values for the day. Engine speed Maximum, minimum, and average values for the day. Oil pressure value Maximum, minimum, and average values for the day. water temperature Maximum, minimum, and average values for the day. Ambient temperature Maximum, minimum, and average values for the day. mileage Mileage driven on the day Travel time Total driving time on the day Operating conditions Statistical values of different working conditions fuel consumption per 100 kilometers fuel consumption per 100 kilometers …… ……
[0048] Preferably, the process of obtaining the fault prediction model by learning and training a machine learning model on all oil pressure samples in the above method includes:
[0049] Sample classification process: All oil pressure samples are divided into training sample set and test sample set. In this embodiment, the training sample set accounts for 80% of the oil pressure samples and the test sample set accounts for 20% of the oil pressure samples. The proportion of the training sample set and the test sample set is obtained from the experiment. If the proportion of the training sample set exceeds 80%, there will be insufficient test cases. If the proportion of the training sample set is lower, there will be fewer training samples, which will lead to insufficient training accuracy.
[0050] Learning and training process: Select the XGBoost model to train the training sample set to obtain the fault prediction training model. When the fault prediction training model obtains the fault prediction result after calculating the test sample set and meets the requirements, the fault prediction training model is used as the fault prediction model. That is, after each round of training on the training sample set using the XGBoost model, it is validated using the test sample set. The validation metrics include multiple metrics such as accuracy, recall, and AUC. This process is continued until the training is completed and the final model is saved.
[0051] The reason for choosing the XGBoost model is:
[0052] (1) High accuracy: XGBoost has high accuracy and predictive ability when processing structured data, especially performing well on large-scale datasets.
[0053] (2) Robustness: XGBoost can handle missing values and outliers, and is also very robust to noisy data.
[0054] (3) Scalability: XGBoost supports parallel computing and can utilize multi-core CPUs and distributed computing to process large-scale datasets quickly.
[0055] (4) Interpretability: XGBoost can output feature importance to help users understand the model’s prediction results.
[0056] (5) Prevent overfitting: XGBoost uses regularization techniques, such as L1 and L2 regularization, to prevent overfitting.
[0057] (6) Flexibility: XGBoost supports a variety of loss functions and evaluation metrics, which can be adjusted according to different tasks.
[0058] See Figure 1 As shown, the above learning and training process specifically includes:
[0059] S101: Input the data of the training sample set (referred to as the training set in the figure) into the XGBoost model to train and obtain the fault prediction training model, then proceed to S102.
[0060] S102: Input the data of the test sample set (referred to as the test set in the figure) into the fault prediction training model to calculate the fault prediction result, and then determine whether the fault prediction result meets the requirements (i.e. whether it meets the standards in the figure). If it does, use the fault prediction training model as the fault prediction model and output it; otherwise, adjust the data of the training sample set according to the reasons for not meeting the requirements and rewrite it to S101.
[0061] Preferably, the process of obtaining the fault prediction result by calculating the oil pressure data of the vehicle under test using the fault prediction model in the above method includes: after receiving the fault prediction request from the vehicle, obtaining the oil pressure data of the vehicle, the composition of which is the same as that of the oil pressure sample; and obtaining and displaying the fault prediction result by calculating the oil pressure data of the vehicle under test using the fault prediction model.
[0062] Specifically, the vehicle's oil pressure data is stored in the database in advance according to the format of the oil pressure sample, and can be retrieved directly from the database by vehicle ID. When displaying fault prediction results, all results can be pushed to the user client (i.e., pushes are made for both when a fault will occur and when it will not occur within the prediction period). In this embodiment, pushes are only made when a fault will occur within the prediction period.
[0063] Preferred, see Figure 2 As shown, the machine learning-based low oil pressure fault prediction system in this embodiment of the invention includes an on-board controller, an on-board TBOX, an oil pressure data acquisition module running on the on-board TBOX, a model training platform, a fault prediction model training module running on the model training platform, and a fault prediction module (in APP form) running on a user client.
[0064] The oil pressure data acquisition module is used to: acquire oil pressure data through the vehicle controller and transmit it to the fault prediction model training module through the vehicle TBOX.
[0065] The fault prediction model training module is used to: learn and train all oil pressure samples within the prediction period through a machine learning model to obtain a fault prediction model; and calculate the fault prediction results by using the oil pressure data of the vehicle under test through the fault prediction model. The fault prediction results include faults that will occur with low oil pressure after the prediction period and faults that will not occur with low oil pressure after the prediction period.
[0066] See Figure 2 As shown, the model training platform includes a cloud platform and an online service platform. The cloud platform includes a data platform and a machine learning platform. Accordingly, the fault prediction model training module will be divided into the following components in actual operation:
[0067] The sample acquisition and transformation module running on the data platform is used to generate oil pressure samples required for the fault prediction model after the vehicle oil data has undergone data cleaning, data aggregation and other operations.
[0068] The model training module running on the machine learning platform is used to: learn and train the machine learning model on all oil pressure samples within the prediction period to obtain a fault prediction model.
[0069] The fault prediction calculation module running on the online service platform is used to: calculate the fault prediction result by using the oil pressure data of the vehicle under test through the fault prediction model.
[0070] The fault prediction module is used to receive and display the fault prediction results returned by the fault prediction model training module.
[0071] In practical use, the workflow of the above prediction system includes:
[0072] (1) The data platform collects and forms oil pressure samples in real time. The fault prediction model training module learns and trains all oil pressure samples in the prediction period through machine learning model to obtain the fault prediction model.
[0073] (2) When a user applies for fault prediction through the fault prediction module, the vehicle controller collects the vehicle's oil pressure data (oil pressure value, vehicle speed, engine speed, elevation, temperature, etc.) in real time and sends it to the vehicle TBOX. The vehicle TBOX collects and uploads the vehicle's driving status data to the data platform. After receiving the uploaded data, the data platform performs data cleaning, data aggregation and other operations to generate the samples required by the low oil pressure detection model. Then, the model is trained through the machine learning platform, and the online service platform loads the model to provide real-time result calculation services. The result information is pushed to the driver through the mobile APP to remind the driver to check the oil and engine in time to avoid safety problems during vehicle operation due to untimely oil changes.
[0074] Preferably, the oil pressure sample includes: oil pressure data within the prediction period when the current oil pressure is low, and oil pressure data within the prediction period when the current oil pressure is not low.
[0075] In this embodiment, sample labels are used to distinguish whether the oil pressure is low. A positive sample label represents that the oil pressure is not low, which is represented by 0. A negative sample label represents that the oil pressure is low, which is represented by 1. The oil pressure data includes data that is related to and may be related to the oil pressure, and is obtained by filtering from a large number of vehicle data.
[0076] Each oil pressure data point includes several aggregated features. The aggregated feature values are obtained by aggregating all feature values within an aggregation period (e.g., 1 day). For example, the aggregated features corresponding to the oil pressure data might be vehicle speed, engine speed, and coolant temperature. In this case, vehicle speed represents the maximum, minimum, and average vehicle speed within the aggregation period.
[0077] Preferably, in this embodiment, the polymerization cycle is 1 day, and the polymerization characteristics and characteristic values mainly include:
[0078] Vehicle speed, with characteristic values including the maximum, minimum, and average values within the aggregation period;
[0079] Engine speed, with characteristic values including the maximum, minimum, and average values within the aggregation cycle;
[0080] Oil pressure values, with characteristic values including the maximum, minimum, and average values within the polymerization cycle;
[0081] Water temperature, with characteristic values including the maximum, minimum, and average values within the polymerization cycle;
[0082] Ambient temperature, with characteristic values including the maximum, minimum, and average values during the polymerization cycle;
[0083] Mileage, characterized by mileage within the aggregation period;
[0084] Travel time, with the characteristic value being: the travel time within the aggregation period;
[0085] Operating conditions, characterized by statistical values of operating conditions in different dimensions within the aggregation period;
[0086] Fuel consumption per 100 kilometers, characterized by: fuel consumption per 100 kilometers for each trip within the aggregation period.
[0087] Preferably, the process by which the fault prediction model training module obtains the fault prediction model after learning and training all oil pressure samples using a machine learning model includes:
[0088] Sample classification process: All oil pressure samples are divided into training sample set and test sample set. In this embodiment, the training sample set accounts for 80% of the oil pressure samples and the test sample set accounts for 20% of the oil pressure samples. The proportion of the training sample set and the test sample set is obtained from the experiment. If the proportion of the training sample set exceeds 80%, there will be insufficient test cases. If the proportion of the training sample set is lower, there will be fewer training samples, which will lead to insufficient training accuracy.
[0089] Learning and training process: Select the XGBoost model to train the training sample set to obtain the fault prediction training model. When the fault prediction training model obtains the fault prediction result after calculating the test sample set and meets the requirements, the fault prediction training model is used as the fault prediction model. That is, after each round of training on the training sample set using the XGBoost model, it is validated using the test sample set. The validation metrics include multiple metrics such as accuracy, recall, and AUC. This process is continued until the training is completed and the final model is saved.
[0090] The reason for choosing the XGBoost model is:
[0091] (1) High accuracy: XGBoost has high accuracy and predictive ability when processing structured data, especially performing well on large-scale datasets.
[0092] (2) Robustness: XGBoost can handle missing values and outliers, and is also very robust to noisy data.
[0093] (3) Scalability: XGBoost supports parallel computing and can utilize multi-core CPUs and distributed computing to process large-scale datasets quickly.
[0094] (4) Interpretability: XGBoost can output feature importance to help users understand the model’s prediction results.
[0095] (5) Prevent overfitting: XGBoost uses regularization techniques, such as L1 and L2 regularization, to prevent overfitting.
[0096] (6) Flexibility: XGBoost supports a variety of loss functions and evaluation metrics, which can be adjusted according to different tasks.
[0097] See Figure 1 As shown, the above learning and training process specifically includes:
[0098] S101: Input the data of the training sample set (referred to as the training set in the figure) into the XGBoost model to train and obtain the fault prediction training model, then proceed to S102.
[0099] S102: Input the data of the test sample set (referred to as the test set in the figure) into the fault prediction training model to calculate the fault prediction result, and then determine whether the fault prediction result meets the requirements (i.e. whether it meets the standards in the figure). If it does, use the fault prediction training model as the fault prediction model and output it; otherwise, adjust the data of the training sample set according to the reasons for not meeting the requirements and rewrite it to S101.
[0100] Preferably, the process by which the fault prediction model training module calculates the fault prediction result using the oil pressure data of the vehicle under test through the fault prediction model includes: upon receiving a fault prediction request from the vehicle, acquiring the vehicle's oil pressure data, the composition of which is the same as the oil pressure sample; and then calculating and displaying the fault prediction result using the oil pressure data of the vehicle under test through the fault prediction model.
[0101] Specifically, the vehicle's oil pressure data is stored in the database in advance according to the format of the oil pressure sample, and can be retrieved directly from the database by vehicle ID. When displaying fault prediction results, all results can be pushed to the user client (i.e., pushes are made for both when a fault will occur and when it will not occur within the prediction period). In this embodiment, pushes are only made when a fault will occur within the prediction period.
[0102] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer-readable storage media (or non-transitory media) and communication media (or transient media).
[0103] As is known to those skilled in the art, the term computer-readable storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer-readable storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.
[0104] For example, the computer-readable storage medium may be an internal storage unit of the electronic device described in the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., provided on the electronic device.
[0105] The above are merely specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the scope of the claims.
Claims
1. A method for predicting engine oil low pressure failure based on machine learning, the method comprising: The method includes the following steps: obtaining oil pressure samples within the prediction period, with each oil pressure sample corresponding to a car; learning and training all oil pressure samples using a machine learning model to obtain a fault prediction model; and calculating the fault prediction result using the oil pressure data of the vehicle under test through the fault prediction model. The oil pressure sample includes: oil pressure data within the prediction period when the current oil pressure is low; oil pressure data within the prediction period when the current oil pressure is not low; each oil pressure data includes several aggregated features, and the aggregated feature values are obtained by aggregating all feature values of the feature within the aggregation period. The aggregation features and feature values include: Vehicle speed, engine speed, oil pressure, water temperature, and ambient temperature are all characteristic values: the maximum, minimum, and average values within the polymerization cycle. Mileage, characterized by mileage within the aggregation period; Travel time, with the characteristic value being: the travel time within the aggregation period; Fuel consumption per 100 kilometers, characterized by: fuel consumption per 100 kilometers for each trip within the aggregation period.
2. The method of claim 1, wherein the machine learning-based prediction of engine oil low pressure failure is based on a plurality of engine oil low pressure failure events. The process of obtaining a fault prediction model by learning and training all oil pressure samples through a machine learning model includes: dividing all oil pressure samples into a training sample set and a test sample set; training the training sample set with an XGBoost model to obtain a fault prediction training model; and using the fault prediction training model to calculate the fault prediction result of the test sample set and obtaining the result meets the requirements. 3.The method of claim 1, wherein, The process of obtaining a fault prediction result by calculating the oil pressure data of the vehicle under test using a fault prediction model includes: after receiving a fault prediction request from the vehicle, obtaining the oil pressure data of the vehicle, the composition of which is the same as that of the oil pressure sample; and obtaining a fault prediction result by calculating the oil pressure data of the vehicle under test using a fault prediction model.
4. A machine learning based prediction system for oil low pressure failure characterized by: The system includes an oil pressure data acquisition module, a fault prediction model training module, and a fault prediction module; The oil pressure data acquisition module is used to: acquire oil pressure data and transmit it to the fault prediction model training module; The fault prediction model training module is used to: learn and train a fault prediction model on all oil pressure samples within the prediction period through a machine learning model; and calculate the fault prediction result by using the oil pressure data of the vehicle under test through the fault prediction model. The fault prediction module is used to: receive and display the fault prediction results returned by the fault prediction model training module; The oil pressure sample includes: oil pressure data within the prediction period when the current oil pressure is low, and oil pressure data within the prediction period when the current oil pressure is not low; each oil pressure data includes several aggregated features, and the aggregated feature values are obtained by aggregating all feature values of the feature within the aggregation period; The aggregation features and feature values include: Vehicle speed, engine speed, oil pressure, water temperature, and ambient temperature are all characteristic values: the maximum, minimum, and average values within the polymerization cycle. Mileage, characterized by mileage within the aggregation period; Travel time, with the characteristic value being: the travel time within the aggregation period; Fuel consumption per 100 kilometers, characterized by: fuel consumption per 100 kilometers for each trip within the aggregation period.
5. The machine learning based prediction system of oil low pressure failure as claimed in claim 4 wherein: The process of obtaining the fault prediction model training module by learning and training all oil pressure samples through a machine learning model includes: dividing all oil pressure samples into a training sample set and a test sample set; training the training sample set with an XGBoost model to obtain a fault prediction training model; and when the fault prediction result obtained by the fault prediction training model after calculating the test sample set meets the requirements, the fault prediction training model is used as the fault prediction model.
6. The machine learning based prediction system of oil low pressure failure as claimed in claim 4 wherein: The process of the fault prediction model training module to obtain fault prediction results by calculating the oil pressure data of the vehicle under test through the fault prediction model includes: after receiving the fault prediction request of the vehicle, obtaining the oil pressure data of the vehicle, the composition of the oil pressure data is the same as that of the oil pressure sample; and obtaining the fault prediction results by calculating the oil pressure data of the vehicle under test through the fault prediction model.