Designing Evaluation Indices for Joint Operation Systems Base on Random Forest with Generalized Linear Regression Algorithm

Joint operation system is a typical complex system with nonlinear, emergent properties and diversity. Traditional capability index selection methods cannot describe the capability of the joint operation system objectively and accurately because of the subjectivity and randomness in the process of selection. Here, the joint operation simulation experiment based on simulation deduction system is designed, and the assembled machine learning model is built, the Random Forest with Generalized Linear Regression algorithm, to farm the multidimensional and holographic operation data deeply. The random forest algorithm model is used to deeply mine the “multi angle and full dimension” confrontation data obtained from the simulation experiment to screen the significant basic parameter indexes of joint operation strike-capability, and on this basis, the quadratic linear regression is carried out. Finally, the important target threat index of joint operation system, which reflects the complex characteristics of joint operation system, such as operational uncertainty, adaptability and emergence, is constructed. It provides a feasible decision-making basis for the commander to adjust the operational plan in real time and evaluate and predict the plan.