Abstract:Existing evaluation and cost management systems for distribution network projects are limited by indicator coverage and model accuracy, making it difficult to meet the refined requirements of engineering analysis and cost prediction. An integrated evaluation and cost prediction framework for distribution network projects is constructed by combining an improved Analytic Hierarchy Process (AHP) -entropy weight method, multiple regression analysis, and K-means clustering. First, an entropy weight mechanism is introduced to modify the weight allocation and reduce the subjectivity of traditional AHP in indicator assignment. Second, the K-means algorithm is applied to cluster engineering samples, reducing prediction deviation caused by sample differences. Finally, multiple regression analysis is used to achieve quantitative evaluation and cost prediction based on the indicator system. Simulation results show that the integrated model can improve prediction accuracy and reduce errors, demonstrating good engineering applicability and economic benefits.