Tuning of PID Parameters Using Artificial Neural Network
Nisha1, Avneesh Mittal 2, O.P.Sharma 3, Nitu Dhyani, Vijay Sharma, Avinashi Kapoor1 and T.K. Saxena National Physical Laboratory, Dr. K.S. Krishnan Road, New Delhi – 110 012 1 Department of Electronics, University of Delhi South Campus, New Delhi 110 021 2 Bhaskaracharya College of Applied Sciences, University of Delhi, Dwarka, ND-110 075 3 Kirori Mal College, University of Delhi, Delhi-110 007 e-mail Corresponding Author: firstname.lastname@example.org Abstract: Tuning of the PID controller in a varying environment is extremely difficult. For this purpose one has to use the adaptive PID controller. In the present paper a novel method for fast tuning of the PID controller has been presented and implemented on designed and developed hardware around the 89C51 microcontroller. Varying environment in the very old existing MLW-MK70, former East German bath has been created with the help of two microcontrollers. The artificial neural network (ANN) has been used to tune the PID parameters. The software has been written in Visual BASIC5.0 language.
A well known continuous PID controller is described using
⎛ 1 u = KP⎜e + ⎜ Ki ⎝
∫ e dt + K
de ⎞ ⎟ dt ⎟ ⎠
Keywords: PID, Artificial Neural Network (ANN), Temperature Controller, Adaptive Control I. INTRODUCTION Adaptive nature of Artificial Neural Network controllers[1-6] have made them a major area of interest among researchers in widespread fields[7-13], mainly because ANN controllers can efficiently learn the unknown or continuously varying environment and act accordingly. Industrial automation applications prefer PID (proportional Integral Derivative) controllers because of its simple structure and robustness etc. In this paper an ANN has been employed for tuning of PID parameters. The hardware has been designed and fabricated around Atmel’s 89C51 microcontroller to control the temperature in a 30 year old, former East German, water bath MLW MK 70. A specially designed varying environment has been created in the water bath using two microcontrollers. In this system, fresh water is allowed to flow continuously at a rate according to outflow of the hot water. The level of the water is kept constant inside the bath. II. DESIGN APPROACH For training/ learning the environment ANN has been utilized. By varying the environment the neural network was allowed to learn the system and accordingly the PID Parameters.
where u is the controller output, KP is the proportional gain, KI is the integral time, KD is the derivative time, and e is the error between the set point and the process output. For a digital control of ts sampling periods, we can write
The figure (1) shows the block diagram of the approach followed in the present work Set Point (Ts) + e NNPID Controller Water Bath
⎛ 1 u = K P ⎜ en + ⎜ Ki ⎝
en − en−1 ⎞ ⎟ ts ⎟ ⎠
Observed Temperature (To)
e = Ts- To
Fig.1: Block Diagram of the approach followed A Neural Network tuned PID (NNPID) which has two inputs, one output and three layers which are input layer, hidden layer and output layer. The input layer has two neurons and the output layer has one and their neurons are P-neurons. The hidden layer has three neurons and they are P-neuron (H1), I-neuron (H2) and D-neuron (H3) respectively. The NNPID is shown in Fig.2 In NNPID when suitable connective weights are chosen, a NNPID becomes a conventional PID controller. In the present case weights for the first layer to hidden layer are takes as
978-1-4244-3604-0/09/$25.00 ©2009 IEEE
w1hj= +1, w2hj= -1, w1ho =KP, w2ho =KI, w3ho =KD,
better control performances. A. Back-propagation algorithms
H1i=w1h1I1+ w2h1I2=Ts - Tos = Terr H2i=w1h2I1+ w2h2I2=Ts - Tos = Terr H3i=w1h3I1+ w2h3I2=Ts - Tos = Terr Where Ts is the set point and temperature Tos is observed temperature and Terr is the error. H1i, H2i and H3i...