基于径向基神经网络的螺杆泵转速设定方法

Setting method for progressing cavity pump speed based on radial basis function neural network

为了研究受多种因素影响的螺杆泵转速控制系统,提出一种基于径向基神经网络的螺杆泵转速设定方法.利用径向基函数(RBF)神经网络对螺杆泵转速进行分析及预测,通过对螺杆泵的历史数据分析处理,得到螺杆泵转速的时间序列.将时间序列视为一个从输入到输出的非线性映射,并引入RBF神经网络来进行非线性映射的逼近.通过对网络进行学习与训练仿真实验,并与BP神经网络预测结果对比,表明应用RBF神经网络对螺杆泵转速进行短期预测精度更高、效果更好.该神经网络结构简单,非线性逼近能力强,通过对非样本点数据的实验验证,证明了该系统的可行性,具有一定的实用价值.

In order to investigate the control system of progressing cavity pump （PCP） speed influenced by multi-factors, a setting method for PCP speed based on radial basis function （RBF） neural network was proposed. The PCP speed was analyzed and predicted with RBF neural network. Through analyzing and processing the historical data of PCP, the time series of PCP speed was obtained. The time series could be regarded as a nonlinear mapping from input to output, and the RBF neural network was introduced to approximate the nonlinear mapping. Through performing the simulation experiments in both learning and training for the network and comparing the experimental results with the prediction results of BP neural network, it is revealed that when the RBF neural network is used for the short-term prediction of PCP speed, the precision is higher and the effect is better. The proposed neural network has simple structure and strong nonlinear approximating capability. The experimental verification of non-sample point data proves that the system is feasible and has certain practical value.