In view of the complex operation mechanism of anti-missile equipment system, the unclear structure which makes it difficult to select a suitable efficiency evaluation model, so the effectiveness evaluation of anti-missile equipment system is studied by the method of "data-driven + deep learning". ased on the operational process of the anti-missile equipment system, the evaluation index of the effectiveness of the anti-missile system is constructed from four aspects: detection and tracking, command and control, firepower interception and integrated support. To solve the problems of PSO algorithm, such as local extremum and premature convergence, an improved particle swarm optimization algorithm is proposed to optimize the parameters of SVR, and an IPSO-SVR efficiency evaluation model is established. On the basis of extracting, processing and analyzing a large number of experimental data, the IPSO-SVR model is trained and studied to obtain nonlinear fitting of the effectiveness of the anti-missile equipment system. The experimental results show that the proposed method has a very small error between the expected output and the actual output and it has high fitting accuracy, which means this method has high reliability and feasibility