Аннотация
Abstract The development of methods for the approximation synthesis of flat lever mechanisms in the Kazakh school of the theory of machines and mechanisms is considered. Particular attention is paid to the practical aspects of applying approximate kinematic geometry to problems of synthesizing guides and moving, transmission, and adjustable mechanisms. Improved numerical–analytical methods of synthesis are proposed taking multiple quality criteria of mechanisms into account. To improve the design efficiency, a multicriteria optimization algorithm was developed using modern machine learning methods, in particular DeepSurNet-NSGA and others. New designs of mechanisms are presented, including 6-link mechanisms for a walking robot leg, mechanisms for exoskeletons of the lower and upper limbs, as well as a hinge-lever mechanism, and a Cartesian manipulator with two degrees of freedom. The prospects for applying the obtained solutions in robotics, biomechanics, and mechanical engineering are substantiated. The role of intelligent systems in the automation of synthesis and analysis of the functional capabilities of mechanisms is emphasized.