论文部分内容阅读
针对非线性系统逆模型的学习问题,提出一种基于贝叶斯-高斯神经网络(BGNN)的设计方法.BGNN模型的训练分为两个步骤,首先利用群智能优化算法进行BGNN的离线结构训练;然后用训练好的BGNN模型在线整合历史数据,进行非线性系统逆模型的获取.对水轮发电机组非线性系统进行了BGNN逆模型的仿真,结果表明了BGNN逆模型设计方法具有结构简单、在线辨识效果好等优点,适于非线性离散系统的逆模型设计.
In order to solve the learning problem of nonlinear system inverse model, a design method based on Bayesian-Gaussian neural network (BGNN) is proposed. The training of BGNN model is divided into two steps. Firstly, the BGNN offline structure training Then the historical data is integrated online with the trained BGNN model to obtain the nonlinear system inverse model.The simulation of the BGNN inverse model of the hydrogenerator nonlinear system is carried out.The results show that the BGNN inverse model design method has the advantages of simple structure, Online identification effect is good, suitable for non-linear discrete system inverse model design.