Research On Power System Economic Dispatch Solution Based On A Novel Constraint Handling Mechanism

Under the premise of ensuring power quality,economic dispatch of power system needs to save power generation cost,reduce energy consumption and meet power supply demand,which is of great significance to the country and power enterprises.For this reason,a large number of solving methods are applied to solve the problem of economic dispatch,in which the intelligent optimization algorithm has its unique features and has been widely used.However,with the expansion of power system scale,many factors need to be considered in actual production.When these factors are considered at the same time,the economic dispatch problem of power system becomes a nonlinear,nonconvex,discontinuous and high-dimensional problem,which leads to the limitation of intelligent optimization algorithm.Therefore,this paper proposes two solutions to the problem of constraint processing and population initialization in intelligent optimization algorithm.The main research contents are as follows:Firstly,the static economic dispatch model of power system is studied,which takes the minimum fuel cost as the objective function and considers the valve point effect,network transmission loss,active power balance,prohibited operation zones,unit output limit and unit climbing limit in actual load dispatch.This paper analyzes the main process of differential evolution algorithm,and expounds the basic theory of self-organizing map neural network.Secondly,a novel constraint handling mechanism is proposed.In order to solve the problem of penalty factor selection of traditional penalty function,the virtual generator unit method selects a set of reasonable penalty factor parameters based on the actual unit characteristics to solve the equality constraint problem.For the units that violate inequality constraints,the power distribution scheme in the new constraint handling mechanism adjusts the output power of each unit according to the characteristics of valve points,which effectively solves the inequality constraint problem and makes the unit output more reasonable.The new constraint handling mechanism is applied to the differential evolution algorithm to solve the economic scheduling problem,and the simulation tests are carried out on 13,40 and 80 units.Finally,a population initialization method is proposed to solve the economic dispatch problem of power system.In this method,a mature model is obtained by learning self-organizing map(SOM)neural network for many times,and the data needed by K-means is obtained by using the model.On the basis of SOM network learning,K-means clustering method is used to get more accurate range of initial values of each unit.The initial values of the population obtained by this method are tested in different scale power systems.The results show that the population initialization method can get ideal results in a short time,which proves that the method can accelerate the convergence and the effectiveness is verified.