基于遗传算法和模拟退火算法优化神经网络的铁路营业里程预测

提出应用遗传算法(GA)和模拟退火(SA)优化神经网络预测铁路营业里程.采用3层前馈神经网络实现铁路营业里程的时间序列预测,输入节点数为5,隐层节点数为8,输出节点数为1.对神经网络的连接权重和节点阈值的确定,采用GA和SA算法相结合的混合优化学习策略.两种算法结合时,SA算法处于外层,GA处于内层.GA采用实数编码,把要确定的神经网络节点连接权重和节点阈值作为基因串.数值计算结果表明混合优化的神经网络的学习速度和精度都比单纯BP算法得出的结果好.因此,用GA-SA混合优化的神经网络预测铁路营业里程是可行的.

Prediction of Length Sequence of Railways in Operation Based on Genetic Algorithm and Simulated Annealing Algorithm Optimized Neural Networks

A method is proposed for the prediction of length sequence of railways in operation based on genetic algorithm(GA) and simulated annealing(SA) optimized neural networks. A three-layer feedforward neural network (5 input neurons, 8 hidden neurons, 1 output neuron) is applied to the length sequence prediction of railways in operation. To obtain optimal weights, GA and SA algorithms are integrated to train the neural network. To combine these two algorithms, GA is put into each step of the SA algorithm. The weights that are all real numbers are coded as chromosomes of the GA. Compared with that of BP neural network the numerical results show that this model has the advantages of high prediction accuracy and high operational speed, indicating that the method is feasible.