The Concrete mix proportion of 502 groups was collected as training data, and the model constructed by the BP neural network which optimized by genetic algorithm combined with particle swarm optimization algorithm.The punishment function was append to the objective function of the particle swarm optimization in modelling. Considering the raw material cost of concrete and several key factors affecting the compressive strength of concrete, the fitness value of objective function to punish, solve the concrete mix proportion in the design of nonlinear constraints discrete variables and continuous variables, and thus achieve the goal of control concrete cost and optimize the mix proportion. The 27 sets of concrete mix ratios after cost reduction are obtained, and the compressive strength test was conducted by the established model. The test results show that the fit ratio of the ratio of the combined ratio cost to the target cost is close to 97%, When the concrete cost per cubic meter is reduced by 5 yuan, 10 yuan and 15 yuan, the performance of mix ratio exported by the model meets the strength requirements.
