Application Of BP Neural Network Based On Improved Particle Swarm Optimization Algorithm In Stock Predictio

The stock market has always been an important channel of financing and investment for all parts of society,playing the role of optimizing resource allocation and serving as a barometer of national economic development,and accurate prediction of stock price is of great significance to promote the stability and prosperity of stock market.However,the factors affecting stock prices are complex,covering political,cultural,economic and other aspects,and stock prices often exhibit non-linear instability,which makes traditional stock price forecasting methods unsatisfactory in terms of forecasting effectiveness.With the popularity of the Internet,the application of artificial intelligence technology in stock prediction is gradually increasing,among which Back-propagation Artificial Neural Networks is highly sought after due to its self-learning,fault tolerance,fast prediction speed and high short-term prediction accuracy features,which show unique advantages in handling non-linear time series prediction problems.However,when constructing the BP neural network model,it is considered that the initial weights and thresholds of the network are randomly determined,easily causing the instability of the network,and the memory function of each particle in the particle swarm optimization algorithm can continuously iterate to find the optimal solution,so the particle swarm optimization algorithm is used to optimize its initial weights and thresholds.To further optimize the Particle Swarm Algorithm,enrich the population diversity and avoid falling into the local optimal solution,the introduction of Artificial Immune Algorithm and Genetic Algorithm as improved Particle Swarm Optimization Algorithm also allows the improved Particle Swarm Algorithm to optimize the BP neural network with higher accuracy in terms of weights and thresholds.In order to avoid uncontrollable influences and ensure the selected data is more representative.In this paper,the small data amount of Luanhuaneng stock price and the large data amount of Shanghai stock index are selected as the prediction objects respectively.The data of several groups of different time periods are selected,and the opening price,the highest price,the lowest price,the closing price,the amount of rise and fall,and the rise and fall amount are used to predict the opening price of the following days.The validity of this algorithm is verified by comparing the experimental results using the model built by the improved BP neural network and the model built by the BP neural network to predict the same data set.