基于PSO的RBF神经网络在热工系统辨识中的应用

在神经网络辨识大迟延对象时,模型类中迟延时间多是根据经验估测的,而不同的值对神经网络辨识的精度和效率就会不同。针对上述问题,将基于正交最小二乘(OLS)算法的径向基(RBF)神经网络和粒子群优化(PSO)算法相结合对热工系统的复杂对象进行辨识仿真。通过对电厂一次风量数据和平均床温数据的仿真实验结果表明,在RBF神经网络对大迟延对象进行辨识时,通过PSO算法进一步确定其最佳迟延时间,从而得到更精确的模型并提高辨识效率,可以取得良好的效果。

PSO algorithm based RBF neural network and its application in thermal system identification

The delay time of model category is always estimated based on the experience when neural network is used to identify the object with large delay time. The accuracy and efficiency of neural network identification will be different when the delay time is given different values.Against the above problem,we proposed a method which combined Radial Basis Function(RBF) neural network based on orthogonal least squares algorithm and Particle Swarm Optimization(PSO) algorithm when identified the complex objects in thermal process.Through the simulation to the first wind data and the average bed temperature data,the experimental result shows that it will achieve good results,get more accurate model and enhance efficiency of identification when the best time delay is further defined through PSO algorithm when RBF neural network is used to identify the object with large delay time.