NN tuned BLDC motor

Description

In this thesis the Brush less DC motor speed is controlled by the Neural Network. The Neural Network tuned Brush Less DC motor speed is controlled via PI gain parameters optimal selection. So the optimal solution obtained via Neural Network control PI gain value. The results are compared with the PI tuned speed control of Brush Less DC motor.

Dissertation statement

We developed a simulink model of Brush Less DC motor speed control by using the PID controller circuit. The gain parameters of PID module are control by Neural Network and self tuning. The speed control of Brush Less DC motor is control by Neural Network method more efficiently and provided stability to the different applications.

Brush Less DC motor

In case of DC motor category there are two types of motors available in the market. The names of the two motors are Permanent magnet synchronous motor  and Permanent Magnet Brush less DC motor. The EMF production in these motors are in backward manner [1].

A Brush Less DC motor is operated by the direct current (dc) which having the automatically controlled commutation system instead of mechanically controlled commutation present in the synchronous motor. It is also known as the trapezoidal Permanent Magnet motor because of their output waveform shapes. It has the trapezoidal back emf waveform and quasi square waveform.

Dynamic behavior of Brush Less DC motor

The Brush Less DC motor is similar to the construction of conventional DC motor with only difference of physical commutator. In Brush Less DC motor there is no physical commutator present in the inner part of the machine. The dynamic model of Brush Less DC motor shown in the figure 1. It is similar to the model of Permanent Magnet (PM) DC motor.

Figure 1 Basic dynamic model of Brush Less DC motor

Speed control methods of Brush Less DC motor

1. Proportional Integral control
2. Fuzzy logic control
3. Artificial Neural network control

Above three-speed control methods of Brush Less DC motor, we choose the Neural network control method.

PI controller

Figure 2 reflects the block diagram of proportional-integral control. It has one inner current loop and one outer speed loop. The difference of the reference and real speed provided to the PI controller as input.

Figure 2 Block diagram of PI controlled

Artificial Neural Network

The artificial neural network is inspired by a biological nervous system like human brains. It is a mathematical model and provides the non-linear relationship between the system input and output. The structure of the neural network consist of three main input, and output and hidden layers.  In this controller, the values of gain parameters of the PI tuned because of the weights and biases of the neural network. The optimal value of  kp and ki obtained after the training of Neural Network.

Figure 4 Architecture of Neural Network

A machine learning algorithm structure three-layer neural network shown in the figure 4 input layer followed by the hidden layer and output layer, and provided optimal selection of gain parameters of PI controller.

Implementation

We proposed machine learning  Neural Network for optimal selection of gain parameters of PI. Neural Network structure inspired by the human brain so it has high decision making ability.

The ANN network also applied to the different applications like Automatic Digital Modulation Detection by neural network available at free-thesis.

ANN has the two main components which are called linear components and nodes. The nodes are fabricated as the input and output node which further connected to the outside environment. The PID controller gain parameters Proportional gain (kp ), Integral gain (ki ) and Derivative gain (kd ) are tune by the Neural network.. Each layer has the different weights and biases which affect the output of the network.

Steps of proposed scheme

1. Developed Simulink model of Brush Less DC motor with the inverter and other control equipment
2. Load the data of Brush Less DC motor into the MATLAB command selected as the input and output
3. Set the initial value gain parameters of PID controller which further tuned by the Neural Network
4. Optimal value of Neural Network tuned PI ( gain parameters selected)
5. Smooth speed of Brush Less DC motor obtained and minimizes the overshoot and settling down rate
6. Compare the results of machine learning Neural Network tune PI controller with PI controller.

Conclusion

This thesis provided the speed control of Brush Less DC motor using the Neural Network tuned PI controller. A simulink model developed in MATLAB with Neural Network control. The Neural Network system tunes the gain parameters of PID controller and provided the effective speed with fast settling down point. From the simulation model it is clear that Neural Network-PID controller provides the best results than Proportional-Integrative-Derivative controlled Brush Less DC motor.  It provides fast control of speed to attain the steady state condition than the PID based method.

References

1. Xiong, G. Junguo, C. Jian and J. Biao, “Research on Speed Control System of Brushless DC Motor Based on Neural Network,” 2015 8th International Conference on Intelligent Computation Technology and Automation (ICICTA), Nanchang, 2015, pp. 761-764.
2. Sreekala. and V. Jose, “Application of neural network in speed control of brushless DC motor using soft switching inverter,” 2012 IEEE International Conference on Motorering Education: Innovative Practices and Future Trends (AICERA), Kottayam, 2012, pp. 1-5.
3. K. V Jha, V. K. Verma, P. Prince, B. Priyadarshini and R. K. Ranjan, “PSO Based Design of Current CCII-PID Controller for the Speed Control of BLDC Motor,” 2018 International Conference on Power, Energy, Control and Transmission Systems (ICPECTS), Chennai, 2018, pp. 245-248.
4. Muhammed A.Ibrahim1 , Ausama Kh. Mahmood2 , Nashwan Saleh Sultan3, ‘Optimal PID controller of a brushless DC motor using genetic algorithm’, International Journal of Power Electronics and Drive System (IJPEDS) Vol. 10, No. 2, June 2019, pp. 822~830.
5. Shamseldin, Mohamed; Mohamed Abdel Ghany; Abdel Ghany Mohamed.International Journal of Power Electronics and Drive Systems; Yogyakarta9, Iss. 2,  (Jun 2018): 536-545.