Abstract Crack prevention is a significant issue in the construction process of concrete dams. The vast majority of concrete cracks are related to temperature variations, and hence, temperature control is a primary method used to prevent cracks. This paper presents a new integrated concept, named Feedback Design (FD) of temperature control measures, which incorporates real-time temperature monitoring, temperature field simulation and construction process simulation in a large system to optimize a temperature control scheme for concrete dams. We used a fiber-optic Distributed Temperature Sensing (DTS) system to monitor the temperature variation process of typical dam monoliths in real time. By incorporating a genetic algorithm, neural network algorithm, and finite element theory, we proposed an intelligent inversion method to obtain thermal parameters of concrete dams based on the temperature monitoring data. Meanwhile, we built a simulation model of a concrete dam’s construction process based on a cycling network technique to obtain accurate construction environmental parameters. Finally, we simulated the temperature field of a typical dam monolith based on initial temperature control measures, inversion thermal parameters, and accurate construction environmental parameters. Temperature variance lines of typical points in the model obtained from the finite element method (FEM) accord well with the measured values obtained from the DTS system. We applied the FD approach to the Xiluodu arch dam located in southwest China to forecast the temperature fields and optimize the temperature control scheme of the dam blocks planned for pouring. The FD approach has been proven reliable and efficacious.