Improved Particle Swarm Neural Network In The Entry And Exit Cargo System

With the rapid development of economy and trade, the substantial increasing trades bring the challenge to the cargo inspection.The original artificial average random inspection method waste a lot of human resource and cannot recognize the unqualified products effectively. Considering the inspection need to be more quickly and more accurate, the entry and exit cargo risk evaluation system was proposed. The paper mainly focus on the improvement of particle swarm optimization algorithm and using it to train the neural network and make the risk evaluation system get a better performance. The opposition-based learning reconfiguring elite particle swarm optimization algorithm(OREPSO) was proposed and train the neural network to realize and optimize the cargo risk evaluation and prediction in the entry and exit cargo system.The paper proposes three improvement strategy:individual and social learning coefficient’s dynamic adjustment by introducing the concept of elite swarm、the reconfiguration to jump out of the local optimum, the opposition-based learning method for initialization and reconfiguration to achieve better performance. Based on these strategies the paper proposes the opposition-based learning reconfiguring elite set particle swarm optimization algorithm(OREPSO);In order to research the performance of OREPSO, the paper chooses four different test functions to do the comparative research with other kinds of improved algorithms. It includes: convergence speed、robustness、the 2 versions which are local best and global best and represent two kinds of topological structure of particle swarm、The analysis of the stability region;In the final the paper uses OREPSO to train the neural network and do the comparative research with other three kinds of neural network in cargo risk evaluation system. The test results prove OREPSO is feasible for the system and has a better performance.