烟花算法改进BP神经网络预测模型及其应用

Prediction Model Based on Improved BP Neural Network with Fireworks Algorithm and Its Application

针对传统BP神经网络预测模型泛化能力弱且预测精度低的问题,首先对现有算法优化神经网络预测模型的不足进行了比较分析;然后,将烟花算法引入到神经网络模型中,利用烟花爆炸算子同时爆炸扩散的机理,对神经网络权重和阈值的寻优过程进行了优化,提出了一种基于烟花算法改进BP神经网络的预测模型。最后,以某纺织企业的棉纺质量数据为例,对提出的基于烟花算法改进BP神经网络的预测方法进行了应用验证。通过与其他算法改进BP神经网络的预测模型进行对比分析,结果表明:该预测方法对纱线质量的预测精度达到97.88%,而且该预测方法与其他方法相比,预测误差率下降了49.52%,并在寻优速度和寻优精度方面表现出较高性能。

To solve the weak generalization ability and low accuracy of the prediction model based on the traditional BP neural network,a prediction model based on the improved BP neural network with fireworks algorithm was proposed.Firstly,the deficiencies of the existing algorithms for optimizing neural network prediction model were analyzed.Secondly,fireworks algorithm was introduced into neural network model,and the mechanism of simultaneous explosion and diffusion of the fireworks explosive operator was adopted to improve the searching process of the weights and threshold in neural network.And further,a prediction model based on the improved BP neural network with fireworks algorithm was proposed.Finally,using spinning quality data as an example to verify the spinning quality prediction,and a comparative study was also made with quality prediction model based on BP neural network optimized by other swarm algorithms.The results show that the proposed method is better than the neural network based on other swarm algorithm on searching speed and accuracy,and prediction rate is up to 97.88%,the false alarm rate is decreased by 49.52%.