Study On Inteligent Optimization Algorithm And Its Applications For Reactive Power Optimization In Power System

Reactive power optimization plays an important role in optimal operation of power systems, which directly effects on the security and effectiveness of power system operation. With the development of intelligent optimization algorithms, exploring them with better convergence and robustness for solving reactive power optimization becomes an attractive research area and is of theoretical and engineering practical significance. On the basis of previous research achievements, the paper investigates single/multiple objective oriented search algorithms and single/multiple objective dynamic multi-group self-adaptive diferential evolution algorithms to deal with static single/multiple objective reactive power optimization, dynamic single/multiple objective reactive power optimization, and benchmark function optimization. The contents mainly include five parts listed as follows:(1) A single-objective oriented search algorithm (OSA) is proposed. In OSA, the search individual simulates human behavior, and the search-object (the optimal solution of the objective function) works like an intelligent agent that can transmit oriented information to search individuals, so that search-individuals and the search-object can communicate with each other. Through testing on typical benchmark function optimization and static single-objective reactive power optimization based on OSA, the efficiency of the proposed algorithm is verified and OSA is proved to be an efficient tool for dealing with static single-objective reactive power optimization in power system.(2) A multi-objective oriented search algorithm (MOOSA) is proposed. MOOSA is based on OSA with the evaluation and selection strategies of multi-objective optimization to search well distributed and converged at Pareto-optimal front. Through testing on static multi-objective reactive power optimization based on MOOSA, the simulations prove that the proposed algorithm has the capability of trading off different objective function values and searching a well distributed set of solutions as closed to the true Pareto-optimal front as possible. As a result, MOOSA is proved to be an efficient tool for dealing with multi-objective reactive power optimization considering static voltage stability in power system.(3) A single-objective dynamic multi-group self-adaptive differential evolution algorithm (DMSDE) is proposed. DMSDE is an improved algorithm based on self-adapting control parameters modified differential evolution (SACPMDE). Through testing on static single/multiple objective reactive power optimization based on DMSDE, the results demonstrate that DMSDE is an efficient method to dispatch static single/multiple objective reactive power flow. In order to further improve the performance of the proposed algorithm, local search method is employed, namely DMSDE with local search algorithm (DMSDELS). The results show that DMSDELS is better in the search precision, convergence property and has strong ability to escape from the local sub-optima through testing on typical benchmark function optimization.(4) DMSDE is employed to solve dynamic single-objective reactive power optimization. Subjected to the maximum daily allowable number of switching operations for control devices which includes transformer taps and capacitor banks, the objective of the dynamic single-objective reactive power optimization under the varying daily load of power system is to minimize active power losses while maintaining acceptable daily voltage profiles. The simulations on dynamic single-objective reactive power optimization illustrate that DMSDE is an efficient method.(5) A multi-objective dynamic multi-group self-adaptive differential evolution algorithm (MODMSDE) is proposed. MODMSDE is based on DMSDE with the non-dominated sorting method and crowding-distance method of multi-objective optimization to evaluate and select new population. Considering that the limitation to the maximum daily allowable number of switching operations for control devices couples different time-period optimization together, an implementation based on dynamic multi-objective optimization is employed to solve dynamic multi-objective reactive power optimization. Through testing on dynamic-based reactive power optimization and benchmark function optimization based on MODMSDE, the efficiency of the proposed algorithm is verified and the implementation based on dynamic multi-objective optimization for the dynamic reactive power model is effective.