基于多层感知机的过程变量趋势诊断方法研究及应用

Research and Application of Process Variable Trend Diagnosis Method Based on Multi⁃layer Perceptron

针对基于统计特性的趋势诊断方法存在诊断滞后及复杂过程变量趋势诊断准确率不高的问题,提出一种基于多层感知机(MLP)的仪表过程变量趋势诊断方法。该方法学习过程历史数据特征间的非线性关系,优化时间窗口内不同时刻过程数据的权重因子,最终提高复杂过程变量趋势诊断及时性和准确率。在支持OPC UA协议的智能压力变送器上的验证结果表明:该方法具有计算量较小和过程变量趋势诊断准确率高的优点。

In order to solve the problems of the trend diagnosis method based on statistical characteristics,such as the lag of diagnosis and the low accuracy of trend diagnosis of complex process variables,this paper proposes a trend diagnosis method of process variables based on multi-layer perceptron in the transmitter.The method learns the nonlinear relationship among the characteristics of process historical data,and optimizes the weight factors of process data at different times in the time window.Finally,it can improve the timeliness and accuracy of trend diagnosis of complex process variables.The verification results on a new type of pressure transmitter supporting OPC UA protocol show that this method has the advantages of less calculation and high accuracy of process variable trend diagnosis.