ELECTRIC VEHICLE SLIPPAGE PREVENTION SYSTEM BASED ON NEURAL NETWORK CONTROLLER

Abstract

А functional diagram of the skidding prevention system was built, a mathematical model of an asynchronous electric drive of an electric vehicle was built, and a neuro-regulator was synthesized. The asynchronous motor is selected according to the equivalent power method, and the standard urban WLT cycle is taken as the basic cycle cycle. The mathematical model of the mechanical part is built taking into account the possibility of simulating the slippage of each of the driving wheels separately with different coupling coefficients. The model consists of a power converter, a battery, a speed regulator, a torque regulator, an asynchronous motor, a braking resistor, a unit for generating set signals, a mechanics unit and measuring units. The internal control system is built on the basis of DTC vector control using Matlab blocks. This work uses the NARMA-L2 control block, which is included in the Neural Network ToolboxTM. A simplified model (object model) was built, neural network parameters such as the number of hidden layers, discretization, number of samples, and number of epochs were selected for neural network training. The neural network was trained taking into account the linearized model of the object, which reflects the qualitative type of real processes in the system. Regardless of the linearization of the system, the output signal with a minimal error (about 1%) corresponds to the input. An analysis of the obtained results of network training was carried out. Simulation of system operation without skidding and skidding of one of the wheels was carried out. The simulation results are compared with the error obtained during neural network training, namely the discrepancy between the output and input signals. The model does not take into account side drift, so only the discrepancy in wheel speeds and the linear change in the speed of the electric vehicle can be observed. The possibility of using the traction electric drive using the method of intelligent neural networks in the safety system of the electric vehicle is shown. A conclusion was made about the operability and efficiency of the system using a neurocontroller to prevent one of the possible slippage modes.