Based On The Particle Swann Optimization_Neural Network Integration Algorithm In Internet of Vehicles Application

Internet of Vehicles, an emerging industry, is being gradually popularized and applied. This paper is to propose a new construction method to neural network ensembles that based on the key technology to multi-sensor data fusion of Internet of Vehicles system, which is an effective solution to the problems such as the information collection about vehicle traffic. Training a group of neural networks Independently, and then use discrete particle swarm optimization (PSO) algorithm and describe all possible neural network ensemble when the value of particles is 0 or 1 in multi-dimensional space. The estimated value to predict error of the network integration is expressed by correlation degree among individual networks that made up of integration, and it will be used as the fitness function during the optimization process. We will select the individual networks which has much differences among the parts that get involved in neural network ensemble.