GA-BP神经网络在老人负性情绪预测中的应用

Application of Genetic Algorithm-back Propagation Neural Network in Negative Emotion Prediction for the Elderly

BP神经网络在当前社会上有着极为广泛的应用,但是传统的BP神经网络存在收敛速度较慢、容易陷入局部最优点的缺点.本文利用遗传算法(Genetic Algorithm,GA)自动调整搜索方向、利于全局择优的特点,构建基于遗传算法优化的BP神经网络模型,用于预测养老机构老年人的负性情绪,重点通过预测的准确性验证模型可行性.本文以北京大学开放研究数据平台的中国健康与养老追踪调查数据空间(CHARLS)作为主要研究数据空间;预测结果表明,粒子群算法(Partical Swarm Optimization,PSO)和遗传算法都能够提升BP神经网络的收敛速度,同时避免陷入局部最优;粒子群算法优化的PSO-BP神经网络在收敛速度上更快,遗传算法优化的GA-BP神经网络在准确度上更优.考虑到养老机构对于数据实时性要求不高,因此选取遗传算法作为BP神经网络在负性情绪预测上的优化方案是目前阶段较为良好的选择.

Back-Propagation Neural network is widely used by now,but the traditional back-propagation neural network has the disadvantages of slow convergence speed and easily fall into local optimum.In consideration of the automatically adjust search direction and good for global optimization by Genetic Algorithm,construction of Back-Propagation Neural Network Model Based on Genetic Algorithms,used on prediction the negative emotion of elderly people in the nursing house,key point of the feasibility in this model is the accuracy of prediction.In this paper,the data space of China Health and Retirement Longitudinal Study(CHARLS)based on Peking University Open Research Data Platform is taken as the main research data space.The prediction results show that both Partial Swarm Optimization(PSO)and Genetic Algorithm can improve the convergence speed of BP neural network,and avoid falling into local optimum.The PSO-BP neural network has faster convergence speed.The GA-BP neural network is better in accuracy.Considering that the nursing house do not have high requirements on the real-time data,the selection of Genetic Algorithm as the optimization scheme of BP neural network in negative emotion prediction is a better choice at the current stage.