Artificial Intelligence Based Non-linear PID Speed and Current Control of Brushless DC Motor for Electric Vehicle Drive. (The Case of Bajaj Qute)

Abstract

Fast depletion of fossil fuels, environmental pollution, and global atmospheric challenges are the major driving forces to explore alternative environmental friendly energy sources. In line with this, fossil fuel driven vehicles contribute a significant share in environmental pollution. Electric vehicles (EV) are alternative solutions. However, energy efficiency and utilization are the major issues in EV. According to the current research status, BLDC motor drives have better efficiency out of different electric drives even though it has relatively high torque ripple. Energy efficient and better dynamic performance features of BLDCM drive for EV application can be achieved by using appropriate control techniques. In this regard, a neural network (NN) based nonlinear PID (NPID) speed controller, genetic algorithm (GA) based PI current controller, and pulse-width-modulated (PWM) based inverter drive of BLDCM for the propulsion of EV Bajaj Qute are implemented for this work. To achieve this, vehicle dynamics, BLDCM, and inverter models are done mathematically. Also, for implementation purpose the corresponding models are done using MATLAB software tool. Different cascade control architectures (like ZN-PID speed and PI current, GA-PID speed and PI current, GA-NPID speed and PI current, and NN-NPID speed and GA-PI current) for the specified application are implemented and performance evaluations are done. Accordingly, NN-NPID speed controller has a settling time of 0.0275sec, 0% overshoot, and 0% steady-state error under a constant load of 20Nm condition and per 1500rpm rated speed. On the other hand, the GA-NPID speed controller has better performance than ZNPID and GA-PID controllers, and it has a settling time of 0.03sec and 0% overshoot. However, due to a sudden load change of 30Nm, it has a +0.7% steady-state error which is twice that of the NN-NPID controller result (0.3%). Furthermore, GA-PI current controller results settling time of 0.02sec and 0% overshoot. Besides, ZN-PI current controller is performed and it has a settling time of 0.09sec and +50% overshoot which is less performance than the GA-PI controller. In addition, the performance of PWM based inverter drive showed 6% torque ripple compared to relay based inverter drive with 20% torque ripple. In general, the performance evaluations of the implemented different cascade controllers revealed that NN-NPID speed and GA-PI current cascade controller along with PWM based inverter drive for the specified application outperformed other controllers.