基于KELM的连续搅拌反应釜模型辨识

发布时间：2018-05-17 04:33

本文选题：核极限学习机 + 单隐层前馈神经网络　；参考：《控制工程》2017年10期

【摘要】：极限学习机(Extreme learning machine,ELM)是一种单隐层前馈神经网络(SLFNs),它随机选择网络的隐含层节点及其参数,训练时仅需调节输出层权值,因此ELM以极快的学习速度获得良好的推广性。考虑到ELM的特征映射函数未知时,可以将核矩阵引入到ELM中。针对模型未知的强非线性连续搅拌反应釜(Continuous Stirred Tank Reactor,CSTR),提出一种基于核极限学习机(Extreme Learning Machine with Kernels,KELM)的NARX模型辨识方法。以仿真的CSTR过程实例进行辨识实验,建立基于NARX-KELM的辨识模型。实验结果表明,在相同条件下,与带动量因子的BP神经网络、模糊神经网络(FNN)、GAP-RBF、MGAP-RBF神经网络、回声状态网络(ESN)、ELM等方法相比,KELM能够有效地改进辨识精度,而且性能更好,这表明了所提方法的有效性和应用潜力。
[Abstract]:Extreme learning machine (ELM) is a kind of single hidden layer feedforward neural network (SLFNs). It randomly selects hidden layer nodes and parameters of the network, and only adjusts the weight of the output layer during training. Therefore, ELM obtains good generalization with very fast learning speed. Considering that the feature mapping function of ELM is unknown, the kernel matrix can be introduced into ELM. For continuous Stirred Tank reactor with unknown model, a NARX model identification method based on extreme Learning Machine with Kernel Kernel KELM is proposed. The identification model based on NARX-KELM is established by using the simulation CSTR process example. The experimental results show that under the same conditions, compared with BP neural network with driving factor, fuzzy neural network with GAP-RBFN MGAP-RBF neural network and echo state network with ESNE / ELM, KELM can effectively improve the identification accuracy, and its performance is better. This shows the effectiveness and application potential of the proposed method.
【作者单位】：兰州交通大学自动化与电气工程学院;
【基金】：国家自然科学基金项目(51467008) 光电技术与智能控制教育部重点实验室(兰州交通大学)开放课题(KFKT2016-3)
【分类号】：TP183;TQ052

【相似文献】