基于神经网络法燃油泵驱动电机转速匹配研究  

Characteristics Curve Fitting of Fuel Pump Based on Legendre Neural Network

作　　者：李实求[1] 郝帅[1] 马旭[1] 程咏梅[1] 

机构地区：[1]西北工业大学,自动化学院,陕西西安710072

出　　处：《电力电子技术》2012年第2期111-113,共3页Power Electronics

基　　金：航空科学基金(20100853010)~~

摘　　要：航空燃油泵驱动电机负责给燃油泵提供所需的动力。由于燃油驱动电机与燃油泵之间存在很强的非线性及耦合性,使得燃油泵无法正常运行在工况点,造成严重的能量损耗。针对该问题,提出了利用Legendre神经网络法来解决燃油泵与驱动电机转速匹配问题。首先,利用Legendre神经网络法拟合出燃油泵的特性曲线,然后再利用该方法建立以燃油泵的流量、扬程为输入,驱动电机转速为输出的神经网络模型。燃油泵与驱动电机转速匹配仿真结果表明,相比于传统的最小二乘法,Legendre神经网络法具有更高的精度。Aviation fuel pump driving motor is responsible for the delivery of power to each engine.As the presence of strong nonlinear and coupling between fuel pump driving system and fuel pump, the fuel pump does not run in the operating point and resulting in energy waste.The method of Legendre neural network is used to solve the problem that fuel pump and driving motor speed matching.Firstly,Legendre neural network is used to fit the fuel pump characteristics curve.Secondly,this method is used again to establish a neural network model in which fuel pump flow and head is input and drive motor speed is output.The simulation results of matching between fuel pump and drivingmotor show that Legendre neural network has higher precision compared to traditional method of least squares.

关 键 词：驱动电机 神经网络 航空燃油泵 曲线拟合 

分 类 号：TM301[电气工程—电机]