Developed a method for solving the problem of multiobjective optimisation of anisotropic regulators of multimass electromechanical systems, allowing to satisfy the diverse requirements that apply to the work multimass electromechanical systems in different modes. Selection matrices by formed the stochastic robust control goal vector is carried out by solving a nonlinear programming problem, in witch the objective function and constraints that are generated by nonlinear scheme of compromise, which uses a combination of penalty function method with an internal point for local criteria and restrictions are permissible, and the penalty function method with an external point for the local criteria and restrictions that are not valid. Formulated mathematical programming problem is multiextremal, the calculation of the objective function is algorithmic in nature, including multiple solution of algebraic Riccati equations, Lyapunov equations and of special form equations for calculating the level of system anisotropy, which required for the synthesis of stochastic robust controller, which requires significant computing resources.
To solve the formulated multiextremal mathematical programming problem used bionic multiagent stochastic particle swarm optimization algorithms does not require the calculation of derivatives of optimized function reliably and allow to find the global optimum multiextremal ravine objective functions and objective functions with areas such as "plateau", significantly reducing the amount of computation of the objective function and significantly reduces the cost of computer time.
The results of comparisons of the dynamic characteristics of electromechanical multimass systems with synthesized anisotropic regulators and types regulators. It is shown, that the use of synthesized anisotropic regulators allowed to reduce the random error compensation of external disturbances, reduce regulation and reduce the system sensitivity to changes of parameters plant compared to a system with standard controllers.