您的位置：首页 > 论文页面
智能模糊边界模块化神经网络的瓦斯预测研究
发表时间：2012-02-28 浏览量：1426 下载量：689
| 全部作者： | 付华，杨義葵，李欣欣 |
| 作者单位： | 辽宁工程技术大学电气与控制工程学院 |
| 摘 要： | 提出一种基于智能模糊边界模块化的神经网络算法，并应用于瓦斯预测。采用分而治之的思想，利用遗传算法自动寻优机制对输入量自动模块化进行划分，然后根据自定义模糊隶属度函数在划分边界一侧按照一定的模糊隶属度设定模糊边界带，从而解决各模块划分点附近预测控制结果的跳跃问题。最后对划分好的各模块样本分别用对应的神经网络进行训练，通过预测合成模块，得出预测结果。该方法相对于单一神经网络模型，提高了学习效率和泛化能力，有效地改善了模型预测精度，实验证明：该算法对瓦斯预测建模具有优越性和有效性。 |
| 关 键 词： | 电气测量技术及其仪器仪表；模块化神经网络；遗传算法；瓦斯预测；模糊边界 |
| Title： | Research on intelligent fuzzy boundary modular neural networks in gas prediction |
| Author： | FU Hua, YANG Yikui, LI Xinxin |
| Organization： | Faculty of Electrical and Control Engineering, Liaoning Technical University |
| Abstract： | The paper proposes a neural network algorithm based on intelligent fuzzy boundary modular and applies it to gas forecast. According to the ‘divide and rule’ thought, the paper uses the genetic algorithm which adopts automatic optimization mechanism to divide the input of the automatic modular, and sets a fuzzy boundary zone based on certain fuzzy membership according to user-defined fuzzy membership functions on one side of the division boundary, which can solve the leap problem of predictive control results near module partition. Finally, the module divided samples with corresponding neural network are respectively trained, and the prediction results are concluded through the forecast synthesis module. Compared with the single neural network model, this method has improved the learning efficiency, generalization ability and accuracy of prediction model effectively. The experimental results show that the algorithm has advantages and effectiveness on gas forecast model. |
| Key words： | electric measurement technology and instrument; modular neural networks; genetic algorithm; gas forecasting; fuzzy boundaries |
| 发表期数： | 2012年2月第4期 |
| 引用格式： | 付华，杨義葵，李欣欣. 智能模糊边界模块化神经网络的瓦斯预测研究[J]. 中国科技论文在线精品论文，2012，5（4）：372-376. |
0
评论数 0请您登录
暂无评论
下载全文
多维论文