Improved Distributed Estimation Algorithm Applied To Nonlinear Multi-constraint Optimization

Owing to the rapid development of science and technology, China's economy is developing at a high speed,meanwhile industry is also gradually rising. But due to the development of the machinery manufacturing, industry in our country is relatively late compared with other western countries, therefore,its development level is still a big gap compared with the developed countries in the West, especially on the problem of mechanical optimization design, the gap is more prominent. So, it is urgent to look for an optimal design scheme applied to the mechanical design optimization, which reduces the consumption of mechanical materials and manufacturing costs, and improve equipment performance and output. At present,many domestic scholars have carried on the thorough research to the mechanical design,and have obtained a lot of effective research results,put forward a number of new or improved algorithms to solve the problem of mechanical optimization design.However,for nonlinear constrained optimization problems,it is not mature enough.Traditional optimization methods such as Newton method and grid method can only find the optimal solution of the problem, and still have some limits such as difficult operation,long running and low search efficiency,it has been unable to meet the requirements of high efficiency of mechanical equipment.Therefore, how to find a more ideal new algorithm to solve the problem of nonlinear multi-constrained mechanical optimization design has become a hot topic in the field of optimizationMIMIC algorithm,as one of the bivariate correlation in distributed estimation algorithm,when studying the problem of probability model in the course of evolution,is easy to produce the phenomenon that it is too rely on the problem of the space distribution,and makes the algorithm to construct the probability model can not accurate representation of the solution space of the information.Thus after several iterations of the algorithm,decrease the diversity of population,and the speed of evolution is becoming more and more slowly.Theresult of changes in each generation are small,the case suggests that its local search ability are relatively weak.In view of this,based on the distributed estimation algorithm,the Particle swarm optimization algorithm and Trust region algorithm with strong local search capability were added respectively,we put forward the combination of particle swarm and trust region algorithm hybrid algorithm with MIMIC algorithm.Numerical experiments are carried out by standard functions.Through numerical experiments are carried out by standard functions,they can have a good result.And then applied to pressure vessels and optimal design of the box girder, the use of programming to solve the optimal solution,then compared with the original algorithm,improvement algorithms improve the optimization efficiency at the same time,also provide a new train of thought and method for the problems of optimal design of nonlinear constraint optimization and lays the foundation for follow-up study.