神经网络在交流异步电力测功机软测量中的应用

The soft- sensing modal of the alternating current asynchronous electrical dynamometer based on neural networks

测功机是对各种动力与传动机械进行性能测试的主要设备,广泛用于齿轮或液力变速器、汽车分动箱、风机和水泵、发动机等装置转矩和输入、输出机械功率测试.从对传统的模型参考自适应方法分析的基础上,结合ADLINE模型,构建了以测功机的定子电流、电子电压为输入量的交流异步电力测功机转速和机械转矩的软测量模型.该模型不需要定子电阻参数,同时将ADLINE模型应用到可调模型来解决PI参数的不确定性,另外为解决参考模型中转子时间常数对参考量输出的影响,建立了以温度为输入的ADLINE模型来在线辨识转子时间常数,并通过加权因子来调整模型的输出.实验证明,该方法具有可行性,为转矩和输入、输出机械功率的测试提供了一种新的方法.

The alternating current asynchronous electrical dynamometer is a major testing- equipment used largely in the power- mechanical field for the measurement of the input power,output power and the torque of gear wheel or torque converter,transfer case,wind turbine or pump,engine and so on. A soft- sensing modal of the alternating current asynchronous electrical dynamometer that input by the voltage and current of the stator used model reference adaptive control and ADLINE model was established based on the analysis of the tradition model reference adaptive control. The new soft- sensing modal didn＇t need the stator resistance and the ADLINE model was used in the adjustable model to solve the uncertainty of the PI parameters. To avoid the rotor＇s time constant causing the error to the output of the reference model,an ADLINE model input with the temperature was used to identification the time constant and the output of the model was adjusted by weight factor. The experiments prove that the proposed modal is effective,thus affords a new method for the measurement of the input power,output power and the torque in the power- mechanical field.