基于并联逆的超声波液位传感器故障误差补偿方法

针对过程控制系统(PCS)液位控制单元中超声波液位传感器故障引起的非线性误差,基于神经网络建模方法设计了一种并联逆补偿环节,有效地补偿了故障误差对系统的影响.通过实验分别获得故障传感器与正常传感器测得的液位值,并进行相应的预处理;基于上述离线数据,分别利用LM-BP、径向基函数(RBF)神经网络的非线性逼近特性,设计逆映射中的并联补偿环节;为验证传感器故障误差的补偿效果,基于OPC技术与Matlab搭建了PCS半实体实验平台,将设计的并联补偿环节置于搭建的PCS液位控制单元中进行闭环实验.结果表明,所建方法能有效补偿传感器故障产生的非线性误差,抑制了故障影响在系统中的传播,实体实验也显现了方法的工程可用性.

Ultrasonic Level Sensor Fault Error Compensation Based on Parallel Inverse Method

Considering the nonlinearity error caused by the ultrasonic level sensor fault in process control system (PCS) level control, a parallel inverse link of compensation was presented based on the method of neural network modeling, which could compensate the effect of fault error. Firstly, the measurement of fault error and actual level were obtained through experiments, and the corresponding pretreatment was done. Then based on the above offline data, the parallel compensation link of the inverse mapping was designed by using the nonlinear approximation properties of Levenberg-Marquardt back propagation (LM-BP) and radial basis function (RBF) neural network. In order to verify the validity and the actual availability of the compensation method, a PCS semi-entity experimental platform was established based on OPC and Matlab technology, etc. The parallel compensation link designed in this paper was placed in level control unit of PCS and closed-loop experiments were implemented. The results showed that the method was effective to compensate the failure nonlinear error for ultrasonic level sensor that inhibited the influence on the system propagation, and suitable to apply in practical engineering.