Evaluation of Fixed Length Gene Genetic Programming (FLGGP) Method in Reservoir System Optimal Operation

This paper employs nonlinear programming (NLP), genetic algorithm (GA), and fixed length gene genetic programming (FLGGP) for the real-time operation of a three-reservoir system (Karoon4, Khersan1, and Karoon3 reservoirs) in which dependent and independent approaches are used to forecast the hydroelectric energy generated by the system. Unlike the forecast-independent approach, in the forecast-dependent approach, the value of release in each period depends on the reservoir in flow of the same period. Moreover, nonlinear decision rule (NLDR) curves are considered, and the total deficiency function as well as efficiency criteria are used to investigate the results of each procedure used. Finally, the performances of real-time operation of single- and three-reservoir systems are investigated and compared. Results indicate that the FLGGP gives the most efficient function for the extraction of reservoir operation rules in both the approaches examined. Comparison of the forecast-dependent and independent approaches revealed no significant differences. Therefore, the forecast-independent approach may be recommended for application in the extraction of reservoir operation rules due to its simplicity and ease of application.