Abstract: Abstract: For the selection of temperature control measures for massive concrete, traditional methods are fully in accordance with the industry standard requirements and subject to repeated artificial amending by practical experience in engineering design and construction. Therefore, it is inefficient and limited by the designer's experience. In this paper, an improved particle swarm optimization (PSO) combined with concrete temperature field and stress field based on the finite element method (FEM) was tested to select the optimal concrete temperature control measures. In the simulation cases, two optimization objectives were defined. The first objective was that the tensile stress of multi feature points should satisfy a safety-cracking factor of at least 1.80. The second was that the whole temperature measures cost should be minimal. The optimization computation was implemented separately with only a single objective for safety factor and both objectives for safety and cost. From the results of 6 calculation cases for a fictitious small-scale concrete dam structure, the following conclusions were drawn. 1) The results show that the proposed method can achieve automatic finding of the temperature control measures optimization, and the optimization results are more scientific and more reasonable. If the ranges of the temperature control parameters can be defined reasonably, the dependence of optimization on humans can be decreased. It will increase the scientificity and persuasion of the temperature control scheme, especially under the complicated situation of more optimization objects, which is very difficult to draw a most reasonable quantitative measures composition. 2) The efficiency of the whole research is improved noticeably. According to experience, if the safety factor is taken as the only objective, the optimization will cost 5 to 7 days by a medium level researcher. In this paper, the time cost of the intelligent optimization is only 2.2 days. 3) After considering the two-objective optimization, the total costs of temperature control measures can be significantly reduced by at least 9% under the condition of ensuring crack-prevention safety. 4) The total calculation time will be influenced by the types, the number and changes of temperature control measures, the locations and number of feature points, and the number of optimization objectives. A high performance personal computer is tested in this paper. The optimization time cost of 5 feature points and 300 days of simulation is 2.7 times the one of 3 feature points and 80 days of simulation. The optimization time cost of the dual-objective is 1.6 times the single-objective. Therefore, a high performance parallel machine should be used to implement the proposed intelligent method for a large-scale engineering structure in a multi-objective, multi-measure, and multi-feature-point optimization task. 5) If the equivalent cooling pipe algorithm is adopted to replace the current explicit one, the optimization for pipe distances will become more feasible. 6) The cost of different temperature control measures in this paper may not be suitable for every construction site. In a factual application case, the checked prices and cost weight should be considered. For future research, how to define the reasonable weights for different feature points at different locations of different structures is the next challenge.