Shape optimization method of B-spline skin free-form thin-shell structures based on plant growth simulation algorithm
In order to find a structural form that not only satisfies the architectural aesthetics but also has good mechanical performance,according to the basic principle of plant growth simulation algorithm(PGSA),a variable-step parallel search mechanism is proposed on the basis of GSL&PS-PGSA,which generates a new simulated plant growth algorithm(GSL&VPS-PGSA).The proposed algorithm is to be introduced into the field of shape optimization of free-form thin-shell structures.This shape optimization method is based on the B-spline skinning method.Firstly,the curves are reconstructed through several type value points.Then,the number of type value points,orders,and knot vectors are adjusted to obtain a series of sectional control curves with similar properties,which are fitted into surfaces by skinning method.On this basis,taking the type value point coordinates of B-spline curve as the design variables,and taking the overall strain energy as the optimization goal,the GSL&VPS-PGSA is used to obtain a free-form thin-shell structure with coordinated shape and mechanical performance.The analysis results of a square arch shell and a complex curved structure show the B-spline skin method can create the free-form surfaces with arbitrary shapes and smoothness through parametric modeling,and its shape control parameters such as control points and orders provide the conditions for further shape optimization.The variable step size parallel search mechanism in GSL&VPS-PGSA can concisely control the solution accuracy and greatly improve the algorithm efficiency.Compared with genetic algorithm and inverse lifting method,the optimization effect of GSL&VPS-PGSA is more advantageous.The proposed shape optimization method can obtain a reasonable structural surface that mainly bears membrane stress,which is suitable for the shape optimization problems of large-scale free-form thin-shell structures with multiple variables,complex constraints and fine mesh division.
国家科技期刊平台
NSTL
万方数据
