Performance Analysis of Single Neuron Adaptive PID, WNN and ANFIS Type Controller Based PMBLDC Motor Drive

Abstract

Permanent Magnet Brushless DC (PMBLDC) Motor Drive is familiar as an innovative research for their compact size, silent operation, reliability, high efficiency, simple construction, easy to control and less maintenance requirements. Design cost of PMBLDC motor drive has been reduced drastically due to the invention of rare earth magnetic materials and advancement of power semiconductor devices. The semiconductor power devices are turned on or off sequentially using triggering and commutation circuits. Based on the rotor position the controller generates requisite signals to control the operation of the power inverter. A high performance controller is required to obtain desired performance of the drive system. It is noticed that PMBLDC Motor has a complexity to handle multi variable and nonlinear system. The motor speed control is frequently needed for controlling various drives such as robotics, copters, electric vehicles and other drives of applications. The high performance drives require very fast response, high efficiency and parameter insensitive. It is not possible to achieve desired performance with conventional PI controller. Tuning of controller parameters is necessary to achieve desired performance. To overcome these problems this study proposes the following controllers (1) Single Neuron based Adaptive (SNA) Controller, (2) Single Neuron based Adaptive PID (SNAPID) Controller, (3) Adaptive Neuro Fuzzy Inference System (ANFIS) based Controller with Radial Basis Function (RBF), (4) ANFIS Controller based on Takagi- Sugeno Model, (5) ANFIS Controller based on Line Voltage Model and (6) Wavelet Neural Network (WNN) based Controller. A PI controller with constant parameter is also designed and it is tuned by Ziegler- Nichols method. Each controller was simulated by developing software in C++ environment and performance of each controller is compared with PI controller. To get better speed, torque and current responses field orientation control method and square wave reference current input to the machine are considered with the above controllers. In this process, direct axis current is considered to zero and quadrature axis current produces useful torque. The drive performance was tested under different operating conditions such as constant starting condition, sudden load torque changes, speed variation and parameter changes. All proposed controllers perform well under speed variation and constant starting condition except fixed PI and WNN based controller. It is also observed that their speed responses are fast. For suddenly increment of load torque, motor speed is fallen by WNN based Controller, fixed PI and SNAPID Controller but SNA Controller, ANFIS Controller based on Takagi-Sugeno Model, ANFIS controller based on Line Voltage Model and ANFIS Controller with RBF perform well with constant speed. Finally, it is seen that all controllers with PMBLDC Motor drive work effectively without any performance degradation in the increment of stator resistance hence, the proposed controllers perform as insensitive controller with the variation of stator resistance.