The Research Of Electric Vehicle IPMSM Parameter Robust Estimation Method

As the non-renewable energy production is decreasing, the advance of modern technology, electric cars become the research focus in the today.Built-in permanent magnet synchronous motor(IPMSM) with its small size, good performance, high efficiency, high reliability, etc, are widely used in electric vehicle drive system.Because electric cars, often in a state of frequent starting, braking deceleration, and magnetic circuit saturation caused by changes in temperature as the variable parameters of inductance changes rapidly and accurately identify the parameters of the motor are needed to establish mathematical model for real-time loss, on the basis of solving the optimization of the motor current, achieve precise vector control to ensure the optimal operation of the motor moment.In this paper the basic structure of permanent magnet synchronous motor, a mathematical model has carried on the simple introduction and analysis.Studied the motor stator winding resistance losses, eddy current and hysteresis loss, stray loss, wind resistance and friction loss, etc., respectively, to establish the mathematical model of various losses, on the basis of the obtained electric controlled loss model of permanent magnet synchronous motor, identified the need to estimate the parameters of the motor efficiency optimization.Further analysis the mathematical model of the motor parameters such as the motor inductance, resistance and loss model and estimated parameters of the mathematical relationship of motor parameter estimation can be reduced the loss in the model parameters.The paper introduced the theory of least absolute deviation method and the simple application, and through comparing the four kinds of simple system simulation accuracy and robustness of parameter estimation method.Analysis found that there is no parameter estimation method, kalman filtering method, least square method and other methods have a common shortcomings, lack of robustness.But least absolute deviation(LAD) is not sensitive to the individual abnormal measured value, at the same time of measuring noise distribution, without any requirements for stronger robustness.Through the example simulation found that LAD method though there is no requirement for measuring the noise distribution, the method for parameter estimation of the time still affected by colored noise is very large.In order to make up for the defect of traditional LAD method, puts forward the improved method of least absolute deviation(ELAD).To change traditional LAD method of objective function method, can significantly improve the method to estimate the robustness of motor parameters.Because solving ELAD objective function operation is very hard to achieve, so on the basis of the theory of optimization theory and the variational ELAD method to estimate parameters will be converted to optimization problem, the method can be further deduced to estimate parameters of the system of equations, to solve the equations for the parameters of the motor estimates.The method of numerical calculation is more complex, computing speed is slow, in the practical application can make use of the recursive neural network in the executable parallel computing hardware FPGA to achieve rapid solutionEstablished the built-in permanent magnet synchronous motor control system and ELAD method motor to estimate parameters of the simulation model.And through the output current and torque waveform control system model was verified the correctness of the simulation to estimate the parameter values of the motor, and analysis contrast to verify its accuracy and robustness.