Design And Improvement Research Of Backtracking Search Optimization Algorithm For DED Problem

Dynamic economic dispatch problem(DED)occupies an important role in power system operation.Its objective is to determine the optimal combination of power outputs of the available units to fulfill the forecasted load demand in the best economical manner,while all physical and operational constraints are satisfied.This paper solves the DED problem with valve point effect which is a high-dimensional constrained optimization problem with non-smooth and non-convex characteristics.In the early stage,it is difficult to obtain the global optimal solution when using linear programming,nonlinear programming,quadratic programming or Lagrange relaxation method to solve the DED problem because of the non-linear and non-convex of generator.In recent years,with the rapid development of meta-heuristic algorithms and its advantages in solving non-linear and non-convex problems,more and more scholars have used meta-heuristic algorithms to solve the DED problem.Backtracking search optimization algorithm,as a meta-heuristic algorithm with strong global search ability,has been used to solve various optimization problems in the real world.In view of the difficulties of non-convex and multiple local optima in the DED problem,this paper designs two adaptive BSAs to solve the DED problem with valve point effect.In addition,it is the first application of BSA to the DED problem.The main research work is as follows:1.Designing an adaptive hybrid backtracking search optimization algorithm(AHBSA).First,AHBSA constructs a coupled structure through hybridizing the strong local search of the DE/best/1 on the basis of the strong global search of BSA,which helps to realize the trade-off between exploration and exploitation capabilities.Second,to further promote the coupling of the coupled structure,two coupled strategies,namely a new BSA mutation strategy and an adaptive parameter control mechanism,are also presented.To verify the effectiveness and feasibility of the proposed AHBSA,five well-known DED cases on two test systems are employed.Experimental results demonstrate that our proposed algorithm can obtain competitive solution compared with other approaches available in literature and the original BSA.2.Designing an adaptive dual-learning backtracking search optimization algorithm(DABSA).In DABSA,a dual-learning strategy(DL),which contains two mutation operators with different optimal information,is first presented to keep a balance between exploration and exploitation capabilities.Second,an adaptive parameter control mechanism(APC)based on the current iteration is developed to choose a suitable parameter value within the crossover operation.To evaluate the performance of the proposed DABSA,three test systems with a total of four instances are considered.Compared to the original BSA,DABSA is far superior to BSA in terms of optimal cost,robustness and convergence speed.Its performance and those of reported representative algorithms are also compared,experimental results demonstrate its competitiveness in yielding lower generation costs along with higher robustness.In order to further verify the performance of the proposed algorithm in solving non-convex DED problem,the experimental results obtained from AHBSA and DABSA are compared in terms of optimal value,standard deviation and convergence.The results show that AHBSA is superior to DABSA in terms of optimal value,and DABSA is superior to AHBSA in terms of algorithm robustness and convergence.In addition,compared with other reported algorithms,the two proposed algorithms are highly competitive in terms of solution quality and robustness.