The phenomena of dust sticking, hardening and bridging on the inner wall of ash silo in thermal power plant seriously affect the safe operation of thermal power units, so it is necessary to clean the ash on the inner wall of ash silo regularly. In view of the limited working range of the existing ash silo cleaning robot and the poor cleaning effect caused by the insufficient telescopic ratio and low stiffness of its mechanical arms, reciprocating rope row drive mechanism and multi-level nested box-type arm body was introduced into the design of the overall structure of the mechanical arm with large stiffness and telescopic ratio of 1: 7. Therefore, an efficient working robot for the all-round cleaning of the inner wall of the ash silo was developed. Based on the depth analysis of the intrinsic structure of the cascade arm and the internal and external coupling force system of the mechanical arm combined with Castigliano's second theorem and the principle of linear elasticity, an overall stiffness model of the variable length mechanical arm under complex alternating loads was established. Considering the stress distribution characteristics of the mechanical arm under extreme working conditions, through the periodic topology optimization of hole shape of each discrete sub region in the low stress region, a global stiffness optimization method of the mechanical arm under severe mass constraint was proposed. The performance simulation and test results showed that the mass of the mechanical arm was greatly reduced after optimization, and the mechanical arm had ideal telescopic interval and stiffness characteristics. There was a certain deviation between the simulation value and test value of the end offset of the mechanical arm, which was caused by the machining and assembly errors of the mechanical arm. The research results provide an important reference for the structural design and stiffness optimization of ash silo cleaning robot.