부산대학교 도서관

Reasonable prediction of the progress of substation infrastructure projects can help to scientifically plan the construction period, reduce the waste of resources and time cost. However, substation infrastructure project is characterized by multiple complexities such as large scale, complex construction methods, long construction period and many scenarios involved, which leads to insufficient prediction scenarios for substation infrastructure projects, unthorough consideration of the complexity of the influencing factors, and insufficient use of the historical progress data, and other shortcomings of the existing research. To this end, this study adopts the random forest model to screen and quantify the key factors affecting construction progress based on expert consultation and historical progress data, and constructs a construction progress prediction model for substation infrastructure projects. Subsequently, the progress prediction models for three typical scenarios of civil construction project, electrical project and overhead line project of a substation infrastructure project of the State Grid Company were constructed based on the model respectively. The experimental results show that this paper's model improves the prediction accuracy by 5.8% and 18.3% compared with the modelling results of Extra-tree model and decision tree model, respectively. Finally, based on the modelling results and for different scenarios, relevant suggestions are provided to improve the shortcomings of the traditional schedule management approach.