Research On BP Neural Network Algorithm And Application Based On FPSO Optimization

As it is difficult to exactly identify the local optimal value based on traditional BP neural network algorithm in practices,and as the convergence speed is slow and the accuracy cannot reach the set standards,there's great gap between the output value and real value.Fractional calculus features flexible parameter,and high interference.This paper optimizes BP neural network algorithm by introducing FPSO which shows good performance in memory,convergence speed,global optimization,stability and precision and brings out the G-L defined FPSO-BP.The convergence speed,global optimization,stability and precision of the algorithm all have improvements after the upgradation.By optimizing the initial weight of BP neural network based on FPSO,an optimal training network is formed to minimize the gap between the output value and the expected value as much as possible.The paper comes as follows:1.Make a study on the principle of BP neural network algorithm and formula derivation process,and explain the reasons for the defects of this algorithm.2.Make a study on the theoretical basis and mathematical formula of FPSO;Bring out the process FPSO-BP and the deficiency of the current optimization algorithm.3.Make a study on the definition of the theory and formula of fractional calculus and explain how to optimize neural network algorithm by FPSO.Apply the new algorithm to environmental data prediction and improve the performance and accuracy of the new algorithm in data prediction.4.Make a study on the theoretical basis of image restoration,apply the new algorithm to image restoration,analyze the performance of the new algorithm in image restoration,and make evaluation of the new algorithm.The proposed algorithm is applied to environmental monitoring and image processing.Compared with BP neural network and FPSO,the new algorithm FPSO-BP boasts higher training speed,accuracy and efficiency,thus producing more accurate output of pollutant data prediction and giving better performance in image restoration.