论文部分内容阅读
模糊神经网络应用于热力系统建模,虽能取得较好的效果,但当模糊规则较多时,网络学习速度较慢。针对这个问题,对传统的模糊神经网络进行了改进。利用Kohonen自组织网络对数据信息进行聚类。然后利用粗糙集规则约减的方法,获取模糊神经网络最小规则,以提高模糊神经网络的学习速度。经过锅炉汽压回路模型的仿真实验结果表明:粗糙模糊神经网络学习速度较传统模糊神经网络有较大提高,同时网络误差有所降低。
The application of fuzzy neural network to the modeling of thermodynamic system can achieve better results, but when the fuzzy rules are more, the network learning speed is slower. To solve this problem, the traditional fuzzy neural network is improved. Clustering data information using Kohonen ad hoc networks. Then, rough rules are used to reduce the rules to obtain the minimum rules of fuzzy neural network to improve the learning speed of fuzzy neural network. The simulation results of the boiler steam pressure circuit model show that the learning speed of the rough fuzzy neural network is greatly improved compared with the traditional fuzzy neural network, and the network error is reduced.