Prediction and Early Warning Model of Substation Project Cost Based on Data Mining

Abstract

Under the dual pressure of the reform of the power system and the increasing demand for electricity in society, power grid companies urgently need to improve the level of lean management of power engineering costs. Substation project is an important part of power engineering construction. Accurate cost forecasting can effectively control project budgets, improve corporate efficiency, and standardize capital management and control. Traditional cost prediction methods are difficult to solve the high-dimensional and non-linear problems of cost data. This paper builds a machine learning model based on data mining techniques to predict the cost of substation projects and improve the cost prediction ability. First, the importance of random forest is used to extract key influencing factors, and the support vector machine regression (SVR) cost prediction model based on Grid Search (GS) optimization is constructed. Then, it is compared with the prediction results of the Random Forest (RF) model to verify the effectiveness of the model proposed in this paper for the cost prediction of substation projects. Finally, the probability density function is constructed according to the distribution of historical data, and the threshold of cost early warning is determined based on the “3 sigma principle” to realize the early warning control of the cost of substation projects and provide better guidance for cost management decisions.