Optimal management of hydropower production: Case of Memve’ele hydropower reservoir policy

Abstract

Optimizing the production of electrical energy through the efficient management of reservoirs is one of the major concerns in power supply. Then, optimization methods have been employed to solve the problem of service continuity in a generation. Many reservoir management issues are indeed addressed without achieving a desired solution in line with demand. The main motivation for state-of-the-art technology is to develop robust and intelligent methods that enable strict compliance with production restrictions while meeting energy demand. This paper follows this perspective by proposing the water supply and hydroelectric production through a hybrid method combining the differential evolution and two constraints processing treatments (TCPT). The interest of the constraint method using here, including its superiority of feasible solutions (SF) and auto-adaptative penalty (AP) is its ability to overcome premature convergence and to obtain very accuracy near optimal solutions. AP technique improve by its ability process through tedious trial and error methods to select the best penalty coefficient during implementation. SF, for its part, is a technique for comparing unrealistic solutions on the basis of the violation of global constraint aimed at pushing unrealistic solutions towards the feasible region, while comparing two workable solutions on objective value improves the overall solution. To compare the performance of the proposed hybrid method of optimization with the SF and AP techniques as a constraint method, a series of experiments was firstly carried out using 25 well know benchmark functions. The obtained results as opposed to the current hybrid method such as TCPT-PSO, TCPT-ABC and TCPT-GA have exhibited the method’s superiority. Based on these performances, we employed the proposed hybrid method to manage the water supply and hydroelectric production of the Memve’ele reservoir. The application on the management of Memve’ele Reservoir demonstrates its accuracy in providing a near-optimal solution on the order of 99.7% for 1945s, 1985s and 1988s of demand during the seven hydrological years of the Ntem Valley with real mean square error (RMSE) in order of 10−7for water supply and hundredth for hydropower reservoir operation.

Keywords

• Two constraint processing techniques
• Differential evolution
• Reservoir
• Constraint processing
• Management
• Optimization