International Journal of Electronic Engineering Research ISSN 0975 - 6450 Volume 2 Number 1 (2010) pp.

87–100 © Research India Publications http://www.ripublication.com/ijeer.htm

Intelligent PI Controller for Speed Control of D.C. Motor
1

Brijesh Singh, 2Surya Prakash, 3Ajay Shekhar Pandey and 4S.K. Sinha
1

Research scholar in the Department of Electrical Engineering, IT BHU , Varanasi, UP, India E-mail: er.brijesh81@gmail.com 2 Senior Non Commissioned Officer in the Indian Air Force at Tejpur, Assam, India E-mail: sprakashgiri0571@yahoo.com 3 Asstt. Professor in the Department of Electrical Engineering, Kamla Nehru Institute of Technology, Sultanpur, UP, India Corresponding author E-mail: shekhar.ajay04@rediffmail.com, ajayshekharpandey@gmail.com 4 Asstt. Professor in the Department of Electrical Engineering, Kamla Nehru Institute of Technology, Sultanpur, UP, India E-mail: sinhask98@engineer.com

Abstract An Intelligent PI Controller has been developed for speed control of separately excited D.C. Motor by exploiting the features of Artificial Neural Networks (ANN). The aim of the proposed schemes is to improve tracking performance of separately excited D.C. motor as compared to the conventional (PI) control strategy. Performance of these new controllers has been verified through simulation using MATLAB/SIMULINK package. In the present work the ANN concept is applied for both control strategies i.e. current and speed for D.C. separately excited motor. Intelligent PI Controller enhances the performance and dynamics of the D.C. motor in comparison to conventional PI controllers. Excellent results added to the simplicity of the drive system, makes the ANN based control strategy suitable for a vast number of industries, paper mills etc. Index Terms: DC motor, ANN, PI Controller, NARMA-L2 Control

Introduction
The separately excited Direct current (DC) motors with conventional Proportional-

88

Brijesh Singh et al

Integral (PI) speed controller are generally used in industry. This can be easily implemented and are found to be highly effective if the load changes are small. However, in certain applications, like rolling mill drives or machine tools, where the system parameters vary substantially and conventional PI or PID controller is not preferable due to the fact that, the drive operates under a wide range of changing load characteristics. The introduction of variable-speed drives increases the automation and productivity and, in the process, efficiency. Decreasing the energy input or increasing the efficiency of the mechanical transmission and processes can reduce the energy consumption. The system efficiency can be increased from 15 to 27 % by the introduction of variable speed drive operation in place of constant speed operation. The energy saving aspect of variable speed drive operation has the benefits of conservation of valuable natural resources, reduction of atmospheric pollution through lower energy production and consumption, and competitiveness due to economy [1]. The artificial neural network (ANN), often called the neural network, is the most generic form of Artificial Intelligence for emulating the human thinking process compared to the rule-based Expert system and Fuzzy Logic [2]. Multilayer neural networks have been applied in the identification and control of dynamic systems. The three typical commonly used neural network controllers: model predictive control, NARMA-L2 control, and model reference control are representative of the variety of common ways in which multilayer networks are used in control systems. As with most neural controllers, they are based on standard linear control architectures [3]. There are a number of articles that use ANNs applications to identify the mathematical D.C. motor model and then this model is applied to control the motor speed [4]. They also use inverting forward ANN with input parameters for adaptive control of D.C. motor [5]. The work, presents the ANN application in speed estimating and controlling separately excited D.C. motor. The motor speed is controlled by forecasting method and forecasting task which ANNs undertake from the terminal voltage parameter, armature current and a reference speed. DC motors have staged a come back with the advent of the silicon controlled rectifiers used for power conversion, facilitating a wide range speed control of these motors. The emphasis has been given on two quadrant operation. With the help of transfer function models, analysis of the performance of the dc motor drives for different cases has been done. Broadly, two control strategies with ANN are proposed for separately excited D.C. motor, current control and speed control strategy. 3-Φ Converter Controlled DC motor Drives The control schematic of a two converter – controlled separately excited dc motor is shown in Fig.1. The thyristor bridge converter gets its ac supply through a three phase transformer and fast acting ac contactors. The field is separately excited, and the field supply can not be kept constant or regulated. The DC motor has a tacho-generator whose output is utilized for closing the speed loops. The motor is driving a load considered to be frictional for this treatment. The output of the tacho-generator is filtered to remove the ripples to provide the signal, ωmr. The speed command and ωr*

Intelligent PI Controller for Speed Control of D.C. Motor

89

is compared to the speed signal to produced a speed error signal. This signal is processed through a proportional plus integrator (PI) controlled to determine the torque command. The torque command is limited, to keep it within the safe current limits. The armature current command ia* is compared to the actual armature current ia to have a zero current error. In case, there is an error, a PI current controller process it to alter the control signal vc. The control signal accordingly modifies the triggering angle α to be sent to the converter for implementation [2]. The operation of closed speed controlled drive is explained from one or two particular instances of speed command. A speed from zero to rated value is commanded, and the motor is assumed to be at standstill, this will generate a large speed error and a torque command and in turn an armature current command. The armature current error will generate the triggering angle to supply a preset maximum dc voltage across the motor terminals.

Figure 1: Speed-controller two-quadrant dc motor drive.

The inner current loop will maintain the current at level permitted by its commanded value. When the rotor attains the commanded value, the torque command will settle down to a value equal to the sum of load torque and other motor losses to keep the motor in steady state. The design of the gain and time constant of the speed and current controllers is of paramount importance in meeting the dynamic specifications of the motor drives.

90

Brijesh Singh et al

Modeling of DC motor load connected system (A Transfer Function Approach)
During the starting of separately excited D.C. motors motor, its starting performance is affected by its nonlinear behavior. Also as per the requirement of application, the speed control of DC motors can be done by using different control strategy like current control and speed control [6]. For controlling the speed of DC motor, PI control strategy is applied with current controller and speed controller. The modeling of DC machine, the transfer function model of 3-phase converter controlled DC motor drives and conventional PI controllers for the speed control of a D.C. motor has been reported in literature [7]. ANN based controller is also useful for improving the performance of DC motor over PI controllers [3], [8]. A SIMULINK model has been developed to test the performance of the ANN controller approach and conventional PI controller mode on DC motor drive. The transfer function model of motor and load are shown in Fig.2. The results obtained by the proposed controllers with different control strategies (current controller, speed controller or both), for achieving the better response has been performed. For improving the overall performance of D.C. motor ANN controller is also used. The tachogenerator has the transfer function.
Gω ( s ) = 0 . 065 1 + 0 .002 s

(1)

The speed reference voltage has a maximum of 18 V. The maximum current permitted in the motor is 20 A. Transfer function of all subsystems of given plant model are taken as per [7]. Now the transfer functions of different sub-systems of speed controlled DC drives plant model are following:
motor = K1
load =

1 + sTm 1 + 0.7s = (1 + sT1 )(1 + sT2 ) (1 + 0.1077s)(1 + 0.0208s)

(2) (3) (4) (5) (6)

K b / Bt 1 . 26 / 0 . 0869 14 . 5 . = = 1 + sT m 1 + 0 .7 s 1 + 0 .7 s

G c(s) =
G s (s) =

K c (1 + sT c ) 2 . 33 (1 + 0 . 0208 s ) = . sT c 0 . 0208 s
K s (1 + sT s ) 28 . 73 (1 + 0 . 0188 s ) . = sT s 0 . 0188 s

(7) (8)

where Ks, Kc, K1, Kr, are the gain of speed controller, current controller, motor and converter. Tm,, Tc ,Ts and Tr are the time constant for motor, current controller, speed controller and converter plant and Gs, Gw, Gc, and Hc are the speed controller, speed

Intelligent PI Controller for Speed Control of D.C. Motor

91

controller feedback gain, current controller gain, current feedback gain respectively.

Figure 2: Block diagram of the motor-load coupled drives (A transfer function model).

Simulink Plant Models
(A) Conventional PI Control using current and speed control In this control strategy, current control and speed control are applying for improving the performance of DC motor drive. Fig.3 shows the simulation plant model for this case. The results are shown in Fig.4 (a) and Fig.4 (b). The response shows that the speed of motor can achieve the steady state value with in a small time. The settling time of motor drive is reducing by applying the both control strategy. (Shown in Table.1) Table 1 Specification With Current Control With Speed and Current Control Settling time ts (Sec.) 1.085 1.23 Steady state error (p.u.) 0.01 0.02

Figure 3 : Simulink plant model with speed and current Control strategy

92

Brijesh Singh et al

Figure 4 (a): Output Speed with Current Control Strategy.

Figure 4 (b): Output Speed with Current and Speed Control Strategy.

(B) ANN Based intelligent PI controller (Using NARMA-L2 control) Typically, there are two steps involved, when using neural networks for control system. In system identification stage, a neural network model of the plant to control has been developed where as, in the control design stage; the neural network plant model has been designed to train the controller. In this work, for making the intelligent PI controller, NARMA L-2 controller is used to enhance the performance of D.C. machine. The central idea of this type of control is to transform nonlinear system dynamics into linear dynamics by canceling the nonlinearities. In this case study, the ANN controller has been used to generate the control signal for converters to control the speed of motor according to the plant output. The control signal for converters according to plant output is generated by trained ANN on the basis of plant identification. The different types of cases are as follows:

Intelligent PI Controller for Speed Control of D.C. Motor • • •

93

Using ANN tool as current controller Using current and speed controller both with individual ANN tools Using current and speed controller both with single ANN tool as a controller.

(I) With Current Control The plant layout for the ANN based Intelligent PI Controller with current control is shown in Fig.5 (a) along with the reference model Fig.5 (b) for training purpose. The Neural network specifications are given in Table 2. The training pattern and response of the system for plant model with current control are shown in Fig.6 and 7 respectively. The result (Table 3) shows that the response of the system is better than the Conventional PI current controller. The settling time and steady state error is further reduced effectively.

Figure 5(a): Simulink plant model with only current control strategy with ANN based PI controller.

Figure 5(b): Simulink reference model for training of ANN with current control strategy.

Table 2

Table 3

94

Brijesh Singh et al

Figure 6: Plant input and output training data generation for only current control strategy with ANN based PI controller.

Figure 7: Plant output with current control strategy using ANN tool as current controller.

(II) With Speed and Current Control Strategy In this case individual Neural Networks are used as a current and speed controller and both control signals are compared and given to the plant. The plant layout and reference training of the model for ANN are shown in Fig.8(a) and Fig.8(a) respectively. The output speed response is shown in Fig.9. The ANN specification and results are given in Table.4 & 5. The result shows that the settling time is reduced by applying this control strategy.

Intelligent PI Controller for Speed Control of D.C. Motor

95

Figure 8(a): Simulink plant model with individual ANN based PI controller for current and speed.

Figure 8(b): Simulink reference model for training of ANN for speed and current control strategy.

Figure 9: Output speed with individual ANN based PI controller for current and speed controller.

96 Table 4 Results Settling Time ts (sec.) Steady state value (p.u.) Steady State error (p.u.)

Brijesh Singh et al

1.4 0.98 0.02

Table 5 ANN for ANN for Speed controller current controller Number of inputs 5 3 Number of outputs 1 2 Number of hidden layer 5 2 Number of Training samples 500 500 Number of Training Epochs 150 150 Plant Specification

(III) Single Intelligent PI Controller with Speed and Current Control In this case only single neural network is used. The plant, training and response of the system are shown in Figures 10, 11 & 12 respectively. Reference model used is same as that used in Simulink reference model for training of ANN for speed and current control strategy Fig. 8(b). The result shows that the response of the plant is better in comparison to pervious cases as well as conventional PI control methods. ANN specifications and results are shown in Table. 6 & 7 respectively.

Figure 10: Simulink Plant Model with current and speed control strategy using single ANN based PI controller. Table 6 Table 7

Intelligent PI Controller for Speed Control of D.C. Motor

97

Figure 11: Training data for using single ANN based PI controller for current and speed both.

98

Brijesh Singh et al

Figure 12: Plant output speed for using single ANN based PI controller for current and speed both.

Results and Discussion
In this work, the performance of a D.C. motor with a constant load using different control strategy, conventional (PI) and intelligent (ANN) controller is evaluated on the basis of settling time, maximum overshoot and steady state error.

Table 8 Speed Response Maximum Steady Settling state time overshoot (mp) (ts) sec. p.u. error (ess) p.u. Conventional (PI) controller using current 1.085 0.1 0.01 control Conventional (PI) controller using current 1.23 0.69 0.02 control and speed control strategy ANN approach using current control strategy 1.7 No overshoot 0.02 Cases ANN approach for current control and speed control strategy both by using individual ANN tool ANN approach for current control and speed control strategy both by using single ANN tool 1.4 No overshoot 0.02

0.8

No overshoot

0.03

Intelligent PI Controller for Speed Control of D.C. Motor

99

The simulation of the complete drive system is carried out based on training for different reference plant with different control strategy. The D.C. motor has been successfully controlled using an ANN. Firstly, one ANN controller is used with only current control strategy. The results with speed and current control strategy are better as compared to the results obtained with only current control. In next case, the two ANN are trained to emulate functions: One for estimating the speed of DC motor and other for controlling it. The simulation result (Fig.12) shows that the response of the plant with both current and speed controller is better as compared to conventional methods (Table- 8). The results prove that the complete D.C. drive system is robust to parameter variations. The speed control loop of the drive is simulated with a Conventional PI controller; in order to compare the performances to those obtained from the respective ANN based drive system. The dynamic and steady state performance of the ANN based controlled drive is much better than the PI controlled drives (Table-8). By using the ANN controller concept with D.C. motor the performance and dynamics of the D.C. motor is improved in comparison to conventional PI controllers. All the comparisons for the different cases are tabulated in Table -8.

Conclusion
By using ANN mode controller for the separately excited DC motor speed control, the following advantages have been realized. The speed response for constant load torque shows the ability of the drive to instantaneously reject the perturbation. The design of controller is highly simplified by using a cascade structure for independent control of flux and torque. Excellent results added to the simplicity of the drive system, makes the ANN based control strategy suitable for a vast number of industrial, paper mills etc. The sharpness of the speed output with minimum overshoot defines the precision of the proposed drive. Settling time has been reduced to a label of 0.8 sec.

References
[1] [2] Ion Boldea, S.A. Nasar, Electric Drives, Taylor & Francis, CRC Press, 2006. B.K. Bose, “Expert system, fuzzy logic, and Neural Networks applications in Power Electronics and motion control,” Proc. of the IEEE, vol.82 pp.13031323, Aug. 1994. K.S. Narendra and K. Parthasarathy, “ Identification and control of dynamical system using neural networks,” IEEE Trans., Neural Network, vol. 1. pp. 4-27, Mar. 1990. Phan Quoc Dzung and Le Minh Phuong, “ ANN- Control system dc motor,” IEEE Trans., Neural Network, 1998.

[3]

[4]

100 [5]

Brijesh Singh et al S. weerasoory and M.A. AI-Sharkawi, “identification and control of d.c. motor using back propagation neural networks” IEEE transactions on energy conversion, vol. 6, No. 4 pp 669, Dec. 1991. J.Santana, J.L.. Naredo, F. Sandoval, I. Grout, and O.J. Argueta, “Simulation and construction of a speed control for a DC series motor,” mechatronics, vol. 12, issues 9-10, Nov.-Dec.2002,pp. 1145-1156 P.C. Sen, Principle of Electric Machines and Power Electronics, John Wiley and Sons, !989. J.H. Mines, MATLAB Supplement to Fuzzy and Neural Approaches in engineering, John Wiley, NY, 1997. S.G. German-Gankin-The computing modeling for power electronics system in MATLAB, 2001. S. Hykin, Neural Networks, Macmillan, NY, 1994.

[6]

[7] [8] [9] [10]

Appendix
The following motor parameters and ratings are used for designing speed-controlled dc motor drives, maintaining the field flux constant. Power rating = 5HP, D.C. motor input voltage = 220V, Armature current rating = 8.3A, Rated speed = 1470 rpm, Armature resistance Ra = 4 Ω, Moment of inertia J = 0.0607 kg – m2, Armature Inductance La = 0.072 h, Viscous friction coefficient Bt = 0.0869 N – m/rad/ sec, Torque constant Kb = 1.26 V/rad/sec. Converter specifications: Supplied voltage = 230 V, 3 – Phase, A.C. Frequency = 50 Hz, Maximum control input voltage is ± 10 V.