Study on Tradeoffs Between Weapon System Cost and Performance Based on Support Vector Machine

In the weapon system development and design process, support vector machine was brought forward to implement the tradeoffs between system performance parameters and cost to effectively control system cost and enhance cost benefit. In current adopted methods, the analysis process of iterative methods is too complex, whereas small perturbation methods adopt excessive approximate means to assure the statistic linear relationship between performance parameters and cost, which is hard to be insured in the practical application. Support vector machine has a favorable adaptability to the modeling of nonlinear system and can disposal the small sample problem in weapon system cost analysis, which can solve some practical difficulties existing in cost analysis. According to the essence of tradeoffs between weapon system performance parameters and cost, this paper discussed the general process of tradeoff analyses and validated the practicability and effectivity of the adopted method through the application analysis on a case. Both theoretical analyses and practical applications show that the system identification model established by support vector machine has the satisfying accuracy and generalization capability, and the optimal performance parameters can be obtained by adopting Davidon least square methods (DLS). The results are satisfying.