Differential Evolution Based On Two-stage Mutation Strategy And Its Application In Crack Propagation Path Control

Cracks may penetrate or affect key structures in construction machinery,and greatly reducing material strength and service life.In recent decades,many scholars have studied the law of crack propagation under external load,which makes the crack propagation path control an effective way to reduce the effect of cracks.In order to solve the path control problem in crack propagation,many heuristic algorithms are used in optimization,such as genetic algorithm and particle swarm algorithm.But the general heuristic algorithm has low calculation efficiency,and it takes a lot of calculation to converge to the ideal result.In this paper,the improved differential evolution algorithm and expansion finite element method are applied to the optimization of crack propagation path,and the path of crack propagation is controlled by optimizing the position and size of the opening.Crack growth calculation is a timeconsuming process.In order to reduce the calculation cost,the extended finite element method based on reanalysis is used to solve the path of crack growth.The differential evolution algorithm(DE)has a simple structure,few parameters,and is easy to implement.It also has strong global search capabilities.However,the traditional differential evolution algorithm has the disadvantages of poor calculation efficiency and low precision of optimization results when dealing with complex practical problems.Therefore,this paper proposes a differential evolution based on two-stage mutation strategy(Ts DE).In the first phase of the Ts DE algorithm,a combination of two mutation strategies is used to strengthen the global search capability of the algorithm in the early stage.During the evolution process,the individual adaptively adjusts the mutation strategy used;in the second stage of the Ts DE algorithm,a new domain mutation strategy based on Euclidean distance is proposed to improve the convergence speed of the algorithm later.In addition,a combination of control parameter adjustment mechanisms is used to improve the performance of these mutation strategies,and a linear population reduction mechanism is used to dynamically exclude individuals with lower rankings in evolutionary effects,further improving the computational efficiency of the algorithm.Finally,the CEC2017 benchmark test is used to evaluate the performance differences between the Ts DE algorithm and other differential evolution algorithm variants.The results show that the performance of Ts DE algorithm is better than other comparison algorithms.In order to further verify the performance of the Ts DE algorithm in crack growth path control,this paper compares the Ts DE algorithm with other optimization algorithms in two crack growth examples with no initial holes and with initial holes.The results show that the Ts DE algorithm has higher accuracy in the crack propagation path control.