Support Vector Regression Substation Cost Prediction Based On Improved Genetic Algorithm

With the increase in the demand for social power,the number and scale of power transmission and transformation project construction continue to expand.However,due to the impact of many factors such as the external environment of the project construction,design depth,policies and regulations,it is difficult for investment plan managers to make high-accuracy plans.This condition makes a large deviation between the final investment and the planned investment.How to reasonably estimate the project cost,control the cost of transmission and transformation construction projects,and improve the efficiency of investment capital utilization has become a hot topic in related fields.In view of the problem of power grid project cost prediction,the substation project is selected as the research object.It has characteristics of small data volume,complicated influencing factors and nonlinear correlation.The support vector regression based on Genetic Algorithm,multiple linear regression and neural network are realized and compared.It is concluded that support vector regression has the characteristics of simplicity,high accuracy and strong generalization ability in the prediction of substation engineering cost,which can meet the needs of auxiliary investment decision-making.It is the most suitable model in the this project.An improved genetic algorithm is proposed for the parameter selection of the support vector regression.The improvement includes the selection operation combining roulette wheel selection and elitism selection,and the adaptive adjustment of crossover and mutation probability.The comparative analysis shows that the improved genetic algorithm can accelerate the evolution speed of the population and inhibit the degradation of the population.It has the characteristics of stability and high efficiency when the parameters are optimized,and the improvement effect is obvious.