基于逐层预测模型的感应电机效率优化滑模控制  

Sliding Mode Control of Efficiency Optimization of Induction Motors Based on Layer-to-Layer Prediction Model

作　　者：苗敬利[1] 黄远[1] 

机构地区：[1]河北工程大学信息与电气工程学院,邯郸056038

出　　处：《电工技术学报》2014年第3期206-212,共7页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金(11272112);河北省科技攻关(12213912D)资助项目

摘　　要：针对轻载时感应电机效率优化运行问题,减少转子磁链是一种有效的方法,但转子磁链受电动机参数变化的影响,为了切实保证感应电机运行效率最优,在不同的工况下得到合适的转子磁链给定值成了问题的关键。为了提高转子磁链的预测精度,本文提出了一种基于二叉树型分层神经网络的逐层预测模型。该模型通过逐层细化预测范围并选用多个相同结构的神经网络进行递推,在复杂的工况下得到合适的转子磁链给定值,实现系统效率最优。此外,为了提高效率优化过程中电机转速的响应速度,设计了一种基于新型趋近律的滑模变结构控制策略,从而保证系统轻载时运行效率高并且转速响应迅速。仿真和实验验证了方案的有效性。Lessening the given rotor flux can improve the efficiency of induction motors control system at light load. As the rotor flux is affected by variable parameters of induction motor, it is a critical problem to obtain the appropriate given rotor flux in different operating conditions for efficiency optimization of induction motor. In order to improve the predictive accuracy of rotor flux, a novel pattern based on binary tree hierarchical layer-to-layer prediction model is proposed and it reduces prediction range of each network and uses more networks for degree elevation. This model gives suitable rotor flux to fulfill optimal efficiency in complex operation condition. Furthermore, a novel reaching law of sliding mode variable structure control strategy is proposed to meet the requirement of rapidity during efficiency optimization. Simulation and experiment verify good effectiveness of the proposed approach.

关 键 词：感应电动机 效率 逐层预测 神经网络 滑模控制 

分 类 号：TM301.2[电气工程—电机]