Research On Economic Dispatch Of Power System Based On Memory Kinetic-Molecular Theory Optimization Algorithm

Economic Dispatch(ED)is a classical optimization problem in electric power.A reasonable load distribution plan can not only improve the efficiency of generating electricity and save cost,but also reduce the amount of discharged pollutants.Moreover,it has great strategic significance for energy.With the increasing scale and intelligence of power grid,ED problem gradually becomes high-dimensional and complex.This paper's research mainly studies the improvement of optimization algorithm based on kinetic-molecular theory(KMTOA)and application in the ED problem.The related research works are as follows:(1)The importance of the research of the ED problem is expounded.Then,two mathematical model and constraints of ED problem are described.Aiming at the characteristics of ED problem,such as high-dimensional,multi-constrained and nonlinear,the classical optimization algorithm and intelligent optimization algorithm solving ED problem are introduced in detail.The reasons why the classical optimization algorithm is not competent for the modern power grid are analyzed.At the same time,the advantages and disadvantages of intelligent optimization algorithms which based on different mechanism are analyzed.Aiming at the shortcomings of intelligent algorithm,KMTOA which includes the search mechanism,algorithm steps and advantages is introduced.And,the part of KMTOA which to be improved is also analyzed.(2)Aiming at the low precision and falling easily into early maturing of KMTOA,Optimization algorithm based on kinetic-molecular theory with memory(MKMTOA)is proposed by introducing the mature memory principle.Inspired the memory process,the individual in the population is regarded as a particle with memory values.Then,according to the calculation of memory value and the memory threshold,the individual is divided into 4 memory population libraries:long-term memory,short-term memory,instantaneous memory and forgetting.Simulating the selection and forgetting of memory,the memory of the individual in the 4 population library are continuously iterated by the mathematical model.It can increase the diversity of the population.According to the guide operator of memory elite,selecting the excellent individuals from the long-term memory population library is used to guide the rest individuals.It can speed up the convergence speed of algorithm and avoid the premature.(3)Through the test experiments of 12 single objective benchmark functions,it shows that MKMTOA has higher precision and faster convergence speed in the lower dimension function than KMTOA.The feasibility and validity of high dimensionnal problems and hyper-high dimensional problems are verified by the representative benchmark functions.And then,through the ED experiments on the 6-unit,13-unit and 40-unit system,it shows the applicability and efficiency of MKMTOA from small-scale to large-scale power system scheduling.(4)Considering the multiplicity of ED objectives,the single objective optimization algorithm is difficult to satisfy.On the basis of MKMTOA,the idea of non-dominated sorting is introduced,and the memory library is divided into four level Pareto frontiers by non-dominated fast sorting based on memory classification.MKMTOA transforms multi-objective optimization algorithm is achieved successfully.And,multi-objective optimization algorithm based on kinetic-molecular theory with memory(MOMKMTOA)is proposed.The algorithm preserves the model of updating memory and forgetting memory to ensure the diversity of the algorithm.In order to avoid the local optimal,selecting randomly the leading elite from the first level memory frontier by selection strategy of the memory elite is used to guide the rest individuals.(5)The comprehensive evaluation index IGD of the multi-bjective optimization is selected by MOMKMTOA to test the performance of 10 multi target benchmark functions.It shows that the uniformity and convergence characteristics of the algorithm have the best effect through the data comparison of different multi-objective algorithms.The IEEE-30 with 6-unit is selected as the test system for the engineering application.The test system considering network loss and ignoring network loss are regard as the two cases.It fully proves the performance of MOMKMTOA to solve complex multi-objective ED problem.