Internet of vehicles vehicle state prediction method and system

Abstract

The invention provides an Internet of Vehicles vehicle state prediction method. The Internet of Vehicles vehicle state prediction method comprises the following steps that vehicle operation data, environment data and driving behavior data are collected through a vehicle-mounted sensor array, a communication sub-module, an infrared camera, a biosensor, an OBD interface and a CAN bus; de-noising processing is carried out on the collected original data by utilizing an edge side FPGA acceleration unit and cloud Spark distributed calculation, and vehicle speed fluctuation ratio characteristics, 15-dimensional vehicle dynamic characteristics, 8-dimensional environment characteristics and 12-dimensional behavior characteristics are extracted; and constructing a multivariate prediction model based on the extracted features, wherein the multivariate prediction model comprises but is not limited to a neural network model, a time sequence model and a machine learning model. According to the Internet of Vehicles vehicle state prediction method provided by the invention, multi-source data fusion acquisition is performed, data comprehensiveness is ensured, and a foundation is laid for accurate prediction; edge cloud end cooperative processing is achieved, real-time performance and computing power are both considered, and data processing efficiency is improved; the multi-element model is constructed and fused, the adaptability of the model to a complex scene is enhanced, and the prediction precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Vehicles, and in particular to a method and system for predicting vehicle states in an Internet of Vehicles. Background Art

[0002] The Internet of Vehicles is a new service model that uses the new generation of information and communication technology to achieve all-round network connections between vehicles, vehicles and roads, vehicles and people, and vehicles and the cloud. It is an important branch of the Internet of Things. It uses vehicles as mobile nodes and uses network communication technology to enable vehicles to have "perception, communication, computing, and decision-making" capabilities, thereby improving traffic efficiency, ensuring driving safety, and optimizing user experience.

[0003] In the Internet of Vehicles (IoV) environment, vehicle status prediction is crucial for traffic safety, intelligent driving, and fleet management. Existing vehicle status prediction methods have problems such as single data collection dimension, insufficient real-time processing, and poor model adaptability. They are unable to accurately respond to changes in vehicle status in complex traffic environments, resulting in low prediction accuracy and an inability to provide reliable support for intelligent decision-making.

[0004] Therefore, it is necessary to provide a method for predicting vehicle status in an Internet of Vehicles to solve the above technical problems. Summary of the Invention

[0005] The present invention provides a method for predicting vehicle status in an Internet of Vehicles (IoV), which solves the problems of low vehicle status prediction accuracy, poor real-time performance, and insufficient adaptability in the prior art.

[0006] To solve the above technical problems, the present invention provides a method for predicting vehicle status in an Internet of Vehicles (IoV), comprising the following steps:

[0007] S1, collects vehicle operation data, environmental data and driving behavior data through the vehicle sensor array, communication submodule, infrared camera and biosensor, OBD interface and CAN bus;

[0008] S2: Utilize edge-side FPGA acceleration units and cloud-based Spark distributed computing to denoise the collected raw data and extract vehicle speed fluctuation characteristics, 15-dimensional vehicle dynamic characteristics, 8-dimensional environmental characteristics, and 12-dimensional behavioral characteristics.

[0009] S3. Build a multivariate prediction model based on the extracted features, including but not limited to a neural network model, a time series model, and a machine learning model;

[0010] S4. Fusion and optimization of multivariate models to improve prediction accuracy through model performance evaluation and parameter tuning;

[0011] S5. Apply the optimized model to vehicle status prediction, output the prediction results through the real-time warning interface and intelligent decision-making interface, and present them through HUD visual, voice and tactile multimodal methods;

[0012] S6. Use the data closed-loop feedback unit to collect actual operating data, iteratively train the model, and achieve continuous optimization of prediction capabilities.

[0013] A vehicle network vehicle state prediction system, such as the vehicle network vehicle state prediction method, comprises: a data acquisition module, an edge-cloud fusion processing module, a multivariate model construction module, a model fusion and optimization module, and an application and feedback module;

[0014] Preferably, the input end of the edge cloud fusion processing module is wirelessly connected to the output end of the data acquisition module;

[0015] Preferably, the input end of the multivariate model building module is wirelessly connected to the output end of the edge cloud fusion processing module;

[0016] Preferably, the input end of the model fusion and optimization module is wirelessly connected to the output end of the multivariate model construction module;

[0017] Preferably, the input end of the application and feedback module is wirelessly connected to the output end of the model fusion and optimization module.

[0018] Preferably, the data acquisition module includes a vehicle-mounted sensor array, a communication submodule, an infrared camera and a biosensor, an OBD interface and a CAN bus.

[0019] Preferably, the edge-cloud fusion processing module includes an edge-side FPGA acceleration unit and a cloud-side Spark distributed computing unit, which are used to realize raw data denoising, vehicle speed fluctuation feature extraction, 15-dimensional vehicle dynamic feature extraction, and 8-dimensional environmental feature and 12-dimensional behavioral feature fusion.

[0020] Preferably, the multivariate model building module is used to build multivariate prediction models such as neural network models, time series models and machine learning models.

[0021] Preferably, the model fusion and optimization module is used to fuse multivariate models and achieve model optimization through performance evaluation and parameter tuning.

[0022] Preferably, the application and feedback module includes a real-time warning interface and an intelligent decision-making interface for outputting prediction results.

[0023] The application and feedback module includes a HUD visual, voice and tactile multimodal output unit for presenting prediction information in multiple ways.

[0024] The application and feedback module includes a data closed-loop feedback unit for collecting actual operation data.

[0025] The application and feedback module includes an AEB system linkage control unit, an autonomous driving following strategy optimization unit, and a fleet charging scheduling algorithm unit, which are used to realize the actual application of the prediction results.

[0026] Compared with the related art, the vehicle state prediction method provided by the present invention has the following advantages:

[0027] Beneficial effects:

[0028] The present invention provides a method for predicting vehicle status in an Internet of Vehicles (IoV) network, which integrates and collects multi-source data to ensure data comprehensiveness and lay the foundation for accurate prediction.

[0029] Edge and cloud collaborative processing takes into account both real-time performance and computing power, improving data processing efficiency;

[0030] Multi-model construction and fusion to enhance the model's adaptability to complex scenarios and improve prediction accuracy;

[0031] The data closed-loop feedback mechanism enables continuous optimization of the model and continuous improvement of prediction capabilities;

[0032] Multimodal output and multi-scenario applications expand the practical application value of prediction results and improve the intelligence level of the Internet of Vehicles system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic structural diagram of a preferred embodiment of a method for predicting vehicle status in an Internet of Vehicles (IoV) provided by the present invention;

[0034] Figure 2 for Figure 1 The structural diagram of the data acquisition module shown;

[0035] Figure 3 for Figure 1 The schematic diagram of the structure of the edge cloud fusion processing module shown;

[0036] Figure 4 for Figure 1 The schematic diagram of the structure of the multivariate model building module shown;

[0037] Figure 5 for Figure 1 The structural diagram of the model fusion and optimization module shown;

[0038] Figure 6 for Figure 1 The schematic diagram of the application and feedback module is shown in FIG. DETAILED DESCRIPTION

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0040] Please refer to Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 and Figure 6 ,in, Figure 1 A schematic structural diagram of a preferred embodiment of a method for predicting vehicle status in an Internet of Vehicles (IoV) provided by the present invention; Figure 2 for Figure 1 The structural diagram of the data acquisition module shown; Figure 3 for Figure 1 The schematic diagram of the structure of the edge cloud fusion processing module shown; Figure 4 for Figure 1 The schematic diagram of the structure of the multivariate model building module shown; Figure 5 for Figure 1 The structural diagram of the model fusion and optimization module shown; Figure 6 for Figure 1 A vehicle state prediction method for an Internet of Vehicles (IoV) includes the following steps:

[0041] S1, collects vehicle operation data, environmental data and driving behavior data through the vehicle sensor array, communication submodule, infrared camera and biosensor, OBD interface and CAN bus;

[0042] S2: Utilize edge-side FPGA acceleration units and cloud-based Spark distributed computing to denoise the collected raw data and extract vehicle speed fluctuation characteristics, 15-dimensional vehicle dynamic characteristics, 8-dimensional environmental characteristics, and 12-dimensional behavioral characteristics.

[0043] S3. Build a multivariate prediction model based on the extracted features, including but not limited to a neural network model, a time series model, and a machine learning model;

[0044] S4. Fusion and optimization of multivariate models to improve prediction accuracy through model performance evaluation and parameter tuning;

[0045] S5. Apply the optimized model to vehicle status prediction, output the prediction results through the real-time warning interface and intelligent decision-making interface, and present them through HUD visual, voice and tactile multimodal methods;

[0046] S6. Use the data closed-loop feedback unit to collect actual operating data, iteratively train the model, and achieve continuous optimization of prediction capabilities.

[0047] A vehicle network vehicle state prediction system, such as the vehicle network vehicle state prediction method, comprises: a data acquisition module, an edge-cloud fusion processing module, a multivariate model construction module, a model fusion and optimization module, and an application and feedback module;

[0048] The input end of the edge cloud fusion processing module is wirelessly connected to the output end of the data acquisition module;

[0049] The input end of the multivariate model building module is wirelessly connected to the output end of the edge cloud fusion processing module;

[0050] The input end of the model fusion and optimization module is wirelessly connected to the output end of the multivariate model construction module;

[0051] The input end of the application and feedback module is wirelessly connected to the output end of the model fusion and optimization module.

[0052] The data acquisition module includes a vehicle-mounted sensor array, a communication submodule, an infrared camera and a biosensor, an OBD interface and a CAN bus.

[0053] The edge-cloud fusion processing module includes an edge-side FPGA acceleration unit and a cloud-side Spark distributed computing unit, which are used to realize raw data denoising, vehicle speed fluctuation feature extraction, 15-dimensional vehicle dynamic feature extraction, and 8-dimensional environmental feature and 12-dimensional behavior feature fusion.

[0054] The multivariate model building module is used to build multivariate prediction models such as neural network models, time series models and machine learning models.

[0055] The model fusion and optimization module is used to fuse multivariate models and achieve model optimization through performance evaluation and parameter tuning.

[0056] The application and feedback module includes a real-time warning interface and an intelligent decision-making interface for outputting prediction results.

[0057] The application and feedback module includes a HUD visual, voice and tactile multimodal output unit for presenting prediction information in multiple ways.

[0058] The application and feedback module includes a data closed-loop feedback unit for collecting actual operation data.

[0059] The application and feedback module includes an AEB system linkage control unit, an autonomous driving following strategy optimization unit, and a fleet charging scheduling algorithm unit, which are used to realize the actual application of the prediction results.

[0060] The data acquisition module realizes the comprehensive collection of vehicle operation data, environmental data, and driving behavior data. The on-board sensor array includes accelerometers, gyroscopes, and tire pressure sensors to perceive the vehicle's motion status in real time; the communication sub-module obtains external data such as road environment and traffic signals through 4G / 5G and V2X communication technologies; infrared cameras and biosensors are used to monitor the driver's status; the OBD interface and CAN bus obtain the vehicle's internal electronic control system data.

[0061] Edge-cloud fusion processing module: includes an edge-side FPGA acceleration unit, a cloud-side Spark distributed computing unit, a raw data denoising unit, a vehicle speed fluctuation feature extraction unit, a 15-dimensional vehicle dynamic feature extraction unit, and an environment and behavior feature fusion unit. The edge-side FPGA acceleration unit performs real-time denoising preprocessing on the collected raw data to reduce data interference; the cloud-side Spark distributed computing unit further processes the data, extracts vehicle speed fluctuation features, 15-dimensional vehicle dynamic features (such as acceleration, steering angle, and brake pressure), and fuses 8-dimensional environmental features (such as weather, road conditions, and traffic flow) with 12-dimensional behavior features (such as driver operation frequency and reaction time) to form a multi-dimensional feature vector.

[0062] Multivariate model building module: includes neural network model building unit, time series model building unit, and machine learning model building unit, which respectively build LSTM neural network model, ARIMA time series model, random forest and other machine learning models to model vehicle status from different angles.

[0063] Model Fusion and Optimization Module: This module consists of a model fusion unit, a model evaluation unit, and a parameter tuning unit. The model fusion unit integrates multivariate models using methods such as weighted fusion and stacked fusion. The model evaluation unit evaluates model performance using metrics such as mean squared error and mean absolute error. The parameter tuning unit optimizes model parameters based on the evaluation results to improve prediction accuracy.

[0064] Application and feedback module: includes real-time warning interface, intelligent decision-making interface, data closed-loop feedback unit, HUD multimodal output unit, AEB system linkage control unit, autonomous driving following strategy optimization unit, fleet charging scheduling algorithm unit, driving behavior data collection unit, model iteration training unit. The real-time warning interface and intelligent decision-making interface transmit the prediction results to the vehicle control system and the Internet of Vehicles platform; the HUD multimodal output unit presents prediction information to the driver through visual images, voice prompts, tactile vibrations, etc.; the AEB system linkage control unit automatically triggers the automatic emergency braking system according to the predicted collision risk; the autonomous driving following strategy optimization unit optimizes the following distance and speed of the autonomous driving vehicle; the fleet charging scheduling algorithm unit provides a charging scheduling plan for the electric vehicle fleet; the driving behavior data collection unit collects actual driving data; the model iteration training unit uses feedback data to retrain the model to achieve continuous optimization of the model.

[0065] The working principle of the vehicle state prediction method provided by the present invention is as follows:

[0066] First, the data acquisition module collects all-round vehicle operation-related data and transmits it to the edge-cloud fusion processing module; the edge-side FPGA acceleration unit performs real-time denoising, and the cloud-side Spark distributed computing unit extracts multidimensional features; the multivariate model construction module builds different types of prediction models, and the model fusion and optimization module fuses and optimizes the models; the application and feedback module applies the optimized model to actual prediction, outputs the prediction results and collects feedback data for iterative training of the model, forming a closed-loop system with continuously improved prediction capabilities.

[0067] Compared with the related art, the vehicle state prediction method provided by the present invention has the following advantages:

[0068] Beneficial effects:

[0069] The present invention provides a method for predicting vehicle status in an Internet of Vehicles (IoV) network, which integrates and collects multi-source data to ensure data comprehensiveness and lay the foundation for accurate prediction.

[0070] Edge and cloud collaborative processing takes into account both real-time performance and computing power, improving data processing efficiency;

[0071] Multi-model construction and fusion to enhance the model's adaptability to complex scenarios and improve prediction accuracy;

[0072] The data closed-loop feedback mechanism enables continuous optimization of the model and continuous improvement of prediction capabilities;

[0073] Multimodal output and multi-scenario applications expand the practical application value of prediction results and improve the intelligence level of the Internet of Vehicles system.

[0074] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for predicting vehicle status in an Internet of Vehicles, characterized in that: The following steps are involved: S1, collects vehicle operation data, environmental data and driving behavior data through the vehicle sensor array, communication submodule, infrared camera and biosensor, OBD interface and CAN bus; S2: Utilize edge-side FPGA acceleration units and cloud-based Spark distributed computing to denoise the collected raw data and extract vehicle speed fluctuation characteristics, 15-dimensional vehicle dynamic characteristics, 8-dimensional environmental characteristics, and 12-dimensional behavioral characteristics. S3. Build a multivariate prediction model based on the extracted features, including but not limited to a neural network model, a time series model, and a machine learning model; S4. Fusion and optimization of multivariate models to improve prediction accuracy through model performance evaluation and parameter tuning; S5. Apply the optimized model to vehicle status prediction, output the prediction results through the real-time warning interface and intelligent decision-making interface, and present them through HUD visual, voice and tactile multimodal methods; S6. Use the data closed-loop feedback unit to collect actual operating data, iteratively train the model, and achieve continuous optimization of prediction capabilities.

2. A vehicle network vehicle state prediction system, according to the vehicle network vehicle state prediction method of claim 1, characterized in that: include: Data acquisition module, edge-cloud fusion processing module, multivariate model building module, model fusion and optimization module, and application and feedback module; The input end of the edge cloud fusion processing module is wirelessly connected to the output end of the data acquisition module; The input end of the multivariate model building module is wirelessly connected to the output end of the edge cloud fusion processing module; The input end of the model fusion and optimization module is wirelessly connected to the output end of the multivariate model construction module; The input end of the application and feedback module is wirelessly connected to the output end of the model fusion and optimization module.

3. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The data acquisition module includes a vehicle-mounted sensor array, a communication submodule, an infrared camera and a biosensor, an OBD interface and a CAN bus.

4. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The edge-cloud fusion processing module includes an edge-side FPGA acceleration unit and a cloud-side Spark distributed computing unit, which are used to realize raw data denoising, vehicle speed fluctuation feature extraction, 15-dimensional vehicle dynamic feature extraction, and 8-dimensional environmental feature and 12-dimensional behavior feature fusion.

5. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The multivariate model building module is used to build multivariate prediction models such as neural network models, time series models and machine learning models.

6. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The model fusion and optimization module is used to fuse multivariate models and achieve model optimization through performance evaluation and parameter tuning.

7. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The application and feedback module includes a real-time warning interface and an intelligent decision-making interface for outputting prediction results.

8. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The application and feedback module includes a HUD visual, voice and tactile multimodal output unit for presenting prediction information in multiple ways.

9. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The application and feedback module includes a data closed-loop feedback unit for collecting actual operation data.

10. The vehicle state prediction system of the Internet of Vehicles according to claim 2, characterized in that: The application and feedback module includes an AEB system linkage control unit, an autonomous driving following strategy optimization unit, and a fleet charging scheduling algorithm unit, which are used to implement practical applications of the prediction results.

Images