Speed ​​Sensorless Control Of Induction Motor Based On Kalman Filter

In recent years, because of low cost and high reliability.induction motor has been widely used in industrial,agricultural, household,etc. In order to achieve the closed-loop speed control of induction motor, it commonly used the method that the mechanical sensor is mounted on the shaft of the induction motor.Usually,the mechanical speed sensor is installed in induction motor drive shaf to achieve the closed-loop speed control.But it can cause the problem such as volume increase, cost increase and reduce the system reliability by istalling the mechanical speed sensor.For this problem,speed sensorless control is a good solution to the problem above. So far, many kinds of speed sensorless algorithms have been proposed.In these algorithms, unscented kalman filter(UKF) has a good speed prediction performance and get the majority of scholars study.But in the application,many scholars find that the elapsed time of UKF is long.Also,UKF is easy to filter divergence and not suitable for high dimensional nonlinear systems above 3 dimension. To slove this problem, there is a further study in this paper. The specific research work are as follows:Fisrtly,introduce in detail the mathematical model of induction motor, inverter and the principle of voltage space vector,direct torque control system, which lay the theoretical foundation for the the establishment of induction motor speed sensorless direct torque control simulation model based on Kalman filter.Secondly, introduce unscented kalman filter(UKF). To the problem of UKF that its run time is too long and easy to filter divergence, improve it and propose a kind of algorithm called spherica simplex square root UKF to overcome these problems. On this basis, establish induction motor speed sensorless Matlab simulation model about UKF and spherica simplex square root UKF.It proves that the spherica simplex square root UKF algorithm can effectively reduce the running time and guarantee system not to filter divergence through the simulation. But found the spherica simplex square root UKF algorithm and UKF algorithm still can not be overcome the problem of decline of the filtering precision in high-dimensional nonlinear system. To this end, this paper adopts anonther kind of kalman filter algorithm called cubature kalman filter(CKF) which can be used for high dimensional nonlinear systems.Then, study the estimated performance of UKF and CKF in induction motor speed by Matlab simulation.As the result,it shows that CKF is shorter than UKF in running time,higher than UKF int filter precisions and has good prediction performance in the high-dimensional nonlinear system. Although CKF is shorter than UKF in running time,but just a little short,it is still too long in motor control system which require the high real-time performance.So, reduction dimension CKF algorithm is studyed in this paper,which can further reduce the computation time. At the same time, put forward a kind of algorithm called square root cubature kalman filter(SRCKF) that can solve the problem of filtering divergence.Finally, the result of research by Matlab shows that SRCKF is more practical in engineering which have the same filter precision as the UKF in speed prediction performance and effectively overcome the problem of filter divergence.Also it shows that reduction dimension CKF is superior than CKF because it have shorter running time in addition to the same filter precision as the CKF.