Multi Agent-based Data Validity Analysis Of Internet Of Vehicles And Application Research

The quality of data in the Internet of Vehicles(IoV)plays a vital role in the management of intelligent transportation.Unreliable data will cause very serious consequences.The data validity of safe driving in the IOV is the basis of improving the safety of vehicles.Unlike traditional information systems,the data validity analysis of vehicle safety driving is faced with the diversity of abnormal data,the subjectivity and randomness of drivers’ driving behavior.Thus,combining the characteristics of IoV data and the drivers’ driving style,it is of great significance to develop a data validity analysis with faster speed,better performance,and more accurate detection.In this paper,Multi-agent is used to model and describe vehicle collaboration in the Internet of Vehicles.In addition,based on the Multi-agent model,data validity analysis of the safety-critical data in the IoV is analyzed from one-dimensional and multi-dimensional perspective.The specific research content is as follows:(1)In terms of the validity analysis of one-dimensional time series data in the Internet of Vehicles,this paper improves the traditional Long Short Term Memory(LSTM)network model.Based on the improved LSTM_BD variant model,this paper proposes a unit IOV data validity analysis algorithm LSTM_BD,and then uses Numenta Anomaly Benchmark(NAB)data set to compare with Hierarchical Temporal Memory(HTM)algorithm to verify the performance of LSTM_BD algorithm.It is proved that LSTM_BD algorithm has certain generality and can be applied to other fields.(2)In terms of validity analysis of multidimensional time series data in the Internet of Vehicles,this paper adds driving style parameters,and proposes a multivariate IOV data validity analysis algorithm ADD based on the Traffic Cellular Automata(TCA)model and the Gaussian mixed model(GMM).Considering that real-time anomaly detection of all data in the vehicle interaction data packet will cause a large computational overhead,this paper designs a TCA model based on the hybrid system theory and cellular automata theory.The TCA model is used to preprocess the data packets,so as to achieve the ideal data validity analysis results in limited computing resources.(3)Considering the driver’s driving style will have a certain degree of influence on the vehicle data in the real scene,this paper designs a new quantitative model of driving style.The model is used to quantify the driver’s driving style through data and convert it into the corresponding driving style parameter e,together with the pre-processed vehicle speed v,acceleration a,and distance d.The final data anomaly list is obtained through GMM calculations to complete the validity analysis of the IOV data.