ABSTRACT
The availability modeling and optimization of rice finishing and grading system (RFGS) of a Rice Milling Plant using two evolutionary techniques i.e. Genetic algorithm (GA) and Particle Swarm Optimization (PSO) has been done. The present case study addresses the industrial system consisting of six subsystems namely Abrasive Whitener, Rotary Shifter, Sizer, Polisher, Sortex, and Grader which are subjected to failures. A mathematical model of the Industrial system chosen for the study has been developed using the Markovian Approach (MA) to develop various differential-difference equations which are obtained from the state transition diagram as shown in Figure 10.2. The recursive method is applied to solve these equations and reduced to the Steady State condition required for RFGS. Further, the optimal value of steady state availability (SSA) of systems has been evaluated using MATLAB7.0.4 G.A. toolbox and PSO technique by developing a computer program coded in C language and simulation run on Pentium Core2Duo machine. The SSA obtained from both the evolutionary techniques (GA and PSO) is also compared. The optimization is carried out in two steps firstly to find out the optimum number of generations, thereafter the optimal number of population size/particle size in both the approaches. The Failure and Repair Rates (FRR) of the various subsystem of the system concerned are the governing parameters for optimization which are exponentially distributed. The latter technique shows better results as compared to the first approach in the present case study. The PSO gives 2.85% more availability at the 30th generation and 25 particle size as compared to GA results at 100th generation and 70 population size.
