Research For Voltage Optimization And Control Based On The Co-evolutionary Genetic Algorithm

Although overall voltage level of power system is operated very stably, butvoltage qualified rate is lower in local power grid, and the load terminal voltageexceeds the operating boundary. Voltage optimization and control is conducted toresearch, the scientific voltage regulator means is adopted to carry out the optimalallocation of reactive power, so that voltage qualification rate is increased, andvoltage quality is obviously improved, which is prevented to electrical accidents.Both in theory and practicality, it is conducive to operation stably of power systemunder different load conditions, and to improve the economics of grid operation.Electricity market is reformed and large regional power grid interconnection iscarried out orderly, so that the power system is gradually tend to be expanded largerand complicated, the number of control variables is became to increase much larger, itis more difficult to solve the higher dimension solution space. When the traditionaloptimization algorithms and artificial intelligence optimization methods is used tosolve the optimal voltage regulation problems, it is easily appeared prematureconvergence,slow iterating,insufficient memory and computational efficiencydecline, even the curse of dimensionality problem.Combining to the characteristics of voltage optimization and control problem inthe actual power system, the mathematical model is established to minimize theinvestment of the new voltage reactive power compensation device, which is order toincrease the voltage qualification rate, and the penalty function reflects that the nodevoltage is over limited,which is closer to the actual state grid; Newton-Raphson andfast decoupled power flow algorithm is discussed, the result is showed that thenumber of iterations is increased using fast decoupled power flow algorithm, but eachiteration of the calculation is much smaller, the calculation speed is increased, andmemory footprint is reduced; theoretical framework and superiority ofco-evolutionary genetic algorithm is deeply analyzed, divide and conquer the strategyis adopted, according to the degree of electrical distance between nodes, the entire grid is divided into multiple population partitions by a number of tight couplinginternal nodes, so that the large-scale complex grid optimization problem is dividedinto multiple sub-optimization problems, which is achieved to reduce the complexityof the optimize problem.Finally, the voltage optimization and control is studied in the IEEE-14standardnode system and an actual local power system, it is showed that the control variablesand state variables of the system were operated within the allowable range and overlimited is not happened, the voltage qualification rate is increased, the operationaldimension of complex optimization problems is reduced, computation time is speededup and convergence performance and optimization results is improved, the validityand practicality of the algorithm is verified.