社会情感算法优化神经网络的短时交通流预测

针对反向传播(BP)神经网络用于交通流预测易陷入局部最优且寻优速度慢的问题,采用了社会情感优化(SEO)BP神经网络的参数,以SEO中的个体为一个BP神经网络,以3种情绪为表现形式,通过个体间的合作竞争进行寻优.运用Levy、正态、柯西分布3种情绪随机选择策略,通过不同方式实现了以不同的概率选择不确定的情绪,使SEO中情绪更好地模拟人的正常心理变化.实验表明:该模型较其他模型更有利于搜寻全局最优解,能有效提高短时交通流的预测精度.

Short-term traffic flow forecasting based on SEO optimized neural network

Using back propagation(BP) neural network in traffic flow predicting easy to fall into local optimum and speed of optimizing is slow. Therefore,using social emotional optimization (SEO) algorithm to optimize the parameters of BP network. In SEO,each individual represents a BP network and optimizing through cooperation and competition between individuals that having three emotions. To choose uncertain emotion with different probabilities,use three emotional random selection strategys based on Levy,normal and Cauchy distribution in different ways and this solution can make the update mode of emotion better simulate people 's normal psychological change. Experimental results show that compared with other forecasting models,these models are more advantageous to search the global optimal solution and the predicting precision can be effectively improved.