Research On The Off-line Acquisition Method Of EPS PMSM Parameters

The electric power steering system (EPS) can not only improve the vehicle’s fueleconomy, reduce the difficulty of arranging the engine compartment’s space, but alsoimprove the feeling of driver’s steering, but the last advantage require a good control for themotor. It is necessary to identify the motor’s parameters, because it can improve the motor’sdynamic response and in favor of the online fault detect, at the same time, it can determinetheir variation with the operating status, in fact, and when the EPS use the permanent magnetsynchronous motor (PMSM), this thing will be more important. This paper uses the PMSMas the research object, at the playground of researching the variation with the operatingstatus and study the acquisition method of motor’s parameters.There are several methods to acquire motor’s parameters, including direct measurementmethod, the estimator method, and the identification method and so on. The directmeasurement method can’t use the running data, the accuracy is limited, but theimplementation is simple, the result don’t exist non-convergence problems; the estimatormethod can be used in online estimation, but there results have the situation ofnon-convergence; the identification method is usually used in off-line circumstance, it alsocan use the motor’s running data to acquire its parameters. With the research background,this paper uses the direct measurement method and the identification method.Based on the study of home and abroad literature, decided the technical plan of thedirect measurement method, using the proportionate relationships between the speed of rotorflux and the motor’s mechanical speed to acquire the pole pairs; motor’s internal resistanceis measured via voltammetry which is simple and ensure results accurate.The measurementof direct axis inductance and quadrature-axis inductance is achieved through the voltageintegration method;motor rotor flux amplitude is obtained using the experiment of reverse driving test. On this basis, we carry out various experiments to verify the feasibility of eachmeasurement plan.To cope with the simulation method of studying the off-line parameters identification,we found the simulation model of PMSM in the Matlab/simulink circumstance based on thestudy of the rotor field oriented, and use the rotor field oriented control to achieve themotor’s speed control.An important issue in motor’s parameters identification is choosing the optimizealgorithm, the genetic algorithm (GA) can solve the multi-convex problem, and only requirethe input-output relationship, no extra demands for the optimize object, at the same time, wecan’t proof monotony of the motor, so we chose the GA as the optimization algorithm. Thedesign variables include the stator resistance, the direct axis inductance, the quadrature axisinductance and the rotor flux amplitude.The range of design variables, population size, crossover rate, mutation rate, andgenerations are main part of the GA’s parameters, they will affect the result and the calculateconsumption. Usually, the crossover rate and mutation rate are using the usual choice ofmethod.In case of no pre-knowledge, setting the range of design variables is a “painful” process.Based on the direct measurement of motor parameters on the understanding of the generalvariation with the motor running, you can reduce the limit on the range of design variables inoptimization methods for parameter identification. It is another key technology thatdetermining the recognition result is a true or not. This article uses two constraints, and whenthey meet, that is true. These constraints include: the recognition results don’t change whenincreasing in population numbers and algebra; in the same input and load conditions, theresponse of the motor phase current is in accord with the experimental results when therecognition result is set into the motor simulation model. Based on this analysis, simulationand result analysis of the off-line identification of motor parameters, verify the feasibility ofthis method.In order to further research on the identification method of motor parameters estimation, we organize the motor experiment, and use these experimental datas to conduct motorparameter identification, and analysis results show that using this method to identify motorparameters can get good identification result.In this paper, the research works laid the foundation for the follow-up study ofelectromagnetic coefficients vary with the changes of the running state.