The Research Of Doubly-fed Motor Based On The Neural Network Inverse System Control Strategy And Simulation

For complex nonlinear systems, the inverse system approach can play a role in feedback linearization and multi-variable decoupling, and this idea is simple and easy to understand. But to get the analytical forms of inverse system, it depends on not only the analytical expression of the precise mathematical model, but also some specific system parameters. However, in the specific circumstances, lots of the controlled systems are highly coupled with strong nonlinear characteristics, so it is difficult to realize in precise analytical expression, we can only approximate the model instead. But to establish its precise mathematical model and solve analytical expressions of inverse system, always we have to use many skills of solving, sometimes it is still difficult to achieve, and many non-linear controlled system's internal state and the specific parameters are part of the measurable, sometimes it is even difficult to obtain. In addition, in the actual process, system parameters is also affected by the changes of the surrounding environment, which will cause great volatility to result in the inverse system not be established completely.As the neural network has characteristics of multiple-input and multiple-output, self-learning ability, and it is able to approximate complex non-linear function in any accuracy, therefore, neural network is a very effective method to create a non-analytic forms of the inverse system.Double-fed Motor(DFM) is a multivariable, nonlinear and strongly coupled system. In this paper, based on vector control of Double-fed Motor, the control strategy of artificial neural network(ANN) inverse system for double-fed motor control is proposed, the DFM mathematical model in the synchronous M-T reference frame in the stator flux oriented is established, and the state equations expressions of DFM are given, in the end, the reversibility of the system is confirmed by using the interactor Algorithm. Neural network inverse model of DFM is constructed with MATLAB, by adding integrators and connecting with original system, it comes into a pseudo-linear composite system. The decoupled waves of the speed and flux are shown through simulation, and the simulation result shows that the control strategy of ANN inverse system used in DFM is feasible.