Contract Management Strategies for Nuclear Power Projects Based on Fuzzy Evaluation and Machine Learning Algorithm

Nuclear power engineering projects have the characteristics of high engineering cost, complex contract structure, and long contract period. The contract management directly affects the cost of nuclear power projects. However, as a part of project management, contract management is different from project quality, safety, progress and other management with clear standard specifications. However, rare research works are conducted to deal with assessment of the success of contract management of nuclear projects using fuzzy methods and machine learning algorithms. This article explores the evaluation of contract management capability and efficiency in nuclear power engineering projects by developing an evaluation method for project management maturity. The index system for contract management evaluation is established using the full life cycle method and a fuzzy theory-based method. This article also uses machine learning algorithms such as XGBoost and Random Forest (RF) to verify the fuzzy methods, and rank top features of contract management strategies. Classification results show that XGBoost and Random Forest algorithms are cost effective and time saving. XGBoost algorithm achieves better classification accuracy of 95% compared to RF algorithm. Among the top five features affecting the success of a contract management strategy, program integrity achieved 0.122570 score followed by efficiency of dispute settlement with 0.113232. Models’ accuracy results and ranking score show that the optimized efficiency has potential to increase efficiency of contract management using large datasets on nuclear projects.