Optimization in the Operation of Process Plant through Genetic Programming
G P Basal, Alok Kumar Jain, A K Tiwari, Pradeep K. Chande · IETE Journal of Research · 2000
In the operation of process plant, it is experienced during abnormal conditions that the attendant has to face problem of confusing state because of many alarms. The unwanted alarms have to be suppressed, and thus involving the experience of the operator and in the case of automatic system, computer time is required to bring the system into normal condition. In this paper we propose a novel methodology called Genetic Programming (GP). This technique is a variant of Genetic Algorithms (GAs) that evolves on a dynamic length tree representations on the basis of fitness function. Such tree representations are interpreted as parse trees for a computer program. The initial population consisting of superior individuals and inferior individuals used to create two parent parse trees. These parent parse trees are then combined to generate off-spring population. Here the methodology is to provide assistance to the attendant indicating that plant's behaviour is drifting away from the normal operation and presence of the factors leading to malfunction. The strategy is to minimize error and optimize the operation. The implementation of Genetic Programming approach is shown in tabular form and validity of the approach is verified by a graph providing the state of art during three cases: worse case of generation, best case of generation and average fitness of the population. Further the results thus obtained are compared with the Genetic Algorithms-based fuzzy expert system for the optimization problem. Genetic programming demonstrates the usefulness over other approach in regard to simplicity and faster in response in solving the problem but lacks in flexibility which GA-based fuzzy expert system possesses.