论文部分内容阅读
在泵特性与泵模型研究的基础上,提出了一种基于k-均值聚类算法RBF(Radial Basis Function)神经网络的建模方法,采用k-均值聚类算法通过输入数据样本优化神经网络隐层中心向量与基宽参数,大大地优化了神经网络结构,提高了神经网络的性能;采用最小二乘法通过输入输出数据样本优化了隐层与输出层的连接权值;利用检测的数据分别对泵QH(流量与扬程)特性及泵综合模型进行了神经网络训练,并进行校验。试验结果表明:通过合理地选择隐层节点个数及重叠系数,训练后的神经网络模型可以代替传统的泵特性与泵综合模型的多项式方程,具有较高的精度。
Based on the research of pump characteristics and pump model, a modeling method based on k-means clustering algorithm RBF (Radial Basis Function) neural network is proposed. The k-means clustering algorithm is adopted to optimize the neural network by inputting data samples Layer center vectors and base width parameters, which greatly optimize the neural network structure and improve the performance of the neural network; using least squares method to optimize the connection weight of hidden layer and output layer by using input and output data samples; Pump QH (flow and lift) characteristics and pump integrated model of neural network training, and verification. The experimental results show that the trained neural network model can replace the traditional polynomial equations of pump characteristics and pump integrated model with high precision by reasonably selecting the number of hidden nodes and the overlapping coefficients.