Improved Particle Swarm Algorithm And Its Application In Power System

Electric power system is a complex system, increasing the size and social to the powersupply of "safe, reliable, affordable, high quality, low-carbon" and many other qualityindicators under the background of ever-higher demands, to ensure the implementation ofpower system operation control of the target, face a variety of complex optimization problems.Its complexity is mainly reflected in the objectives and constraints of complex, more extreme,high-dimensional, multi-objective, and there exist many uncertain factors, these aresignificant challenges on modeling and algorithm for innovation, how to build a suitableoptimization models and creating practical and efficient algorithm is the key to solving theproblem. Modern intelligent optimization algorithm with zero on the objective functionrequirements, implementing simple and advantages of parallel search, in the field of complexoptimization are a growing number of applications received significant attention. Thisresearch aimed at: from group intelligent of nature features starting, analysis particle swarmalgorithm learning mode of constructed and the key elements, tried to awareness PSO easy to" premature " of fundamental causes; made innovation of particle swarm learning mode, andbased on new mode design efficient PSO improved algorithm; application random analysistheory and difference method research particle swarm dynamic trajectory and stability domain,further reveals PSO of work mechanism, for algorithm of improved provides theoryFoundation; will improved algorithm and power system specific problem phase combinationand solution, More effective and practical programmes.Swarm intelligence and Bionics principle, stochastic analysis theory and finite differencemethods, matroid theory, optimization theory and algorithms design theory is a powerful toolto study this problem. This article uses a combination of theoretical study and empiricalanalysis method, the main research content and outcomes include:1. Swarm intelligence theory application research on the main reason of PSO is easy to"premature", pointed out that the original learning model nature of "passive" mechanism isone of the important reasons. Come up with a new sense of exploration with active learningmodel and Logistic chaos through new learning mode for combining technology, developed anew and improved Particle Swarm Optimization (AEPSO), and by experimental test analysisto verify the validity of the algorithm.2. Based on matroid theory, proposed a matroid property of General-suitable forcombinatorial optimization problems of a combined population of evolutionary learning mode:Modular evolutionary learning mode based on vector (m1,m2,m-m1-m2). In turn proposed an improved discrete Particle Swarm Optimization (MDPSO), using Backpack to experimentaltests improved algorithms in combinatorial optimization problems such as analysis, feasibilityand effectiveness of the verification algorithm.3. Based on the culture in the Organization’s core values,"belief space" constructingpersonalized dynamic adaptation of cultural information in parameters, fostering speciesdiversity. Come up with an improved multi-objective Particle Swarm Optimization(CBCMPSO), using multi-objective evaluation index system and algorithm testing function toimprove the quality and performance of the algorithm of the test and analysis of experiments,indicating a new algorithm for multiple quality indicators are better.4. Derivation of out stochastic analysis theory and finite difference methods for dynamicanalysis model of a new particle swarm (γ-GPSO model) or proved algorithm depends on theparameter settings for one or two-stage stability region expression draws out of the stable area,contour map of the spectral radius and mean, a map providing a theoretical basis for improvedPSO and its parameter settings and intuitive tools. Derivation of out stochastic analysis theoryand finite difference methods for dynamic analysis model of a new particle swarm (γ-GPSOmodel) or proved algorithm depends on the parameter settings for one or two-stage stabilityregion expression draws out of the stable area, contour map of the spectral radius and mean, amap providing a theoretical basis for improved PSO and its parameter settings and intuitivetools.5. Gives examples of improved algorithm in power system applications, furthervalidation of the feasibility and effectiveness of the algorithm:1) design and simulation ofAEPSO economic dispatching algorithm testing;2) MDPSO design of demand-side resourceoptimization algorithm and simulation testing.