The electric power sector occupies a key position in the national economy. To fulfill current cost-control requirements for power engineering projects, relevant personnel must adopt focused management strategies so that expenses can be controlled through more refined management. In this paper, the learning capability of BP neural networks is employed to refine the affiliation function of fuzzy rules and thereby strengthen the integration of the TS fuzzy system. An enhanced Bayesian classifier is introduced for evaluation, while a heuristic function is incorporated to reduce discretization problems in continuous values, thereby establishing a fuzzy neural network prediction model through K-Means clustering, with the training speed and effectiveness of the BP neural network further enhanced by the Levenberg-Marquardt optimization approach. The whole life-cycle cost of a power project is further classified by time, and model parameters are simultaneously applied in operational calculations to eliminate the price factor from the mathematical model. The evaluation scores of the three levels, general, good, and excellent, in relation to the cost management performance of electric power projects are 0.709, 0.731, and 0.69, respectively, and the performance evaluation in all models is close to 0.7, which indicates that artificial intelligence technology produces a meaningful effect on cost control. After these project cost management measures were carried out, the cost of an electric power project finally realized a balance of 4,395,400 yuan. By optimizing the construction plan, the expected cost-control target for the electric power project was ultimately achieved.
