基于代数法和优化法的异步电动机动态参数辨识方法对比研究

Dynamic parameter identification of asynchronous motor based on algebraic analysis and optimization modelling

针对在线应用中的电动机动态参数辨识问题,描述了异步电动机数模混合建模框架,提出了一种基于数据驱动及显式解析的数模混合辨识方法。分别以电机负荷电压/幅值及有功/无功功率两类PMU量测数据作为输入输出,基于PSASP三阶异步机动态方程,利用代数分析建立了电机惯性系数、暂态电抗、稳态电抗、转子回路时间常数4种基本参数的解析表达式;介绍了一类基于状态空间模型的数模混合参数优化辨识方法,通过遗传算法求取了最优参数使负荷有功/无功功率模型模拟信号与PMU测量信号拟合误差最小;在改进的3机9节点系统中,对基于模拟生成的PMU时序数据进行了对比分析。研究结果表明:代数法在辨识准确性和计算效率上均优于优化辨识法。

Aiming at identification of key asynchronous motor load( AML) parameters for online application,the data-model hybrid modelling framework was firstly given and a data-model hybrid analytic method of AML dynamic parameter identification was developed. With the active/reactive power,terminal voltage and current measured by PMU as input,the fundamental four parameters,including the rotor inertia,rotor transient reactance,rotor steady reactance,rotor time constant were derived by algebra analysis based on the AML dynamic equations.Another class of data-driven and state-space model based optimization modelling method was introduced,in which a genetic algorithm( GA)was adopted to seek the minimum fitting error between the modelled active/reactive power and the PMU counterparts. Finally,a comparison study between these two methods was conducted on a modified three-machine nine-bus systemthe. The results indicate that the proposed algebra analytic method can outperform the optimization method in both identification accuracy and computation efficiency.