提出了一种利用混合pi-sigma模糊神经推理方法建立瓦斯涌出量的预测模型.该模型采用高斯基函数作为模糊子集的隶属度函数,可在线动态调整隶属度函数和结论参数.与神经网络预测模型比较,该模型具有物理意义明确、原理清晰、收敛速度快、预测精度高等特点,在对某矿瓦斯涌出量数据的仿真结果表明,该方法预测准确度高、速度快,并且结果具有可重复性,证明该方法是有效的.为便于工程实际应用,在Matlab环境中开发了基于图形用户界面(GUI)的仿真应用界面,给出了使用方法和预测结果.实验同时表明,对所采用的数据,模型的训练精度设置为0.001时网络的泛化能力最好,网络训练精度和预测精度之间不具有正比关系.%A gas emission prediction model established by using reasoning method of hybrid pi-sigma fuzzy neural networks was proposed. The model adopted Gaussian function as a fuzzy membership function, and the membership functions and conclusions parameters of the model could be adjusted online dynamically. Compared with the neural network prediction model, the method has characteristics of clear physical meaning, clear principle, fast convergence, high prediction accuracy and so on. The gas emission data of a coal mine simulation results show that the prediction has a high accuracy, fast convergence and the prediction results can be repeated, it is proved that the method is effective. In order to facilitate the practical application, the authors developed a Graphical User Interface (GUI) application interface in the Matlab environment, and gave the method and prediction results. The experiments also show that, for the data, the generalization ability of the model is best when the training accuracy is set 0. 001, and the training accuracy and the prediction accuracy of the model do not have positive relationship.
展开▼
