Research On Aerodynamic Design Optimization Of Centrifugal Compressor Blade Configuration Based On Meta Heuristic Algorithm

Centrifugal pressurizer is an important power equipment to guarantee national defense security and promote national economic development,and has been widely used in strategic demand fields such as aviation,shipping,chemical industry and new energy.It is difficult to further improve the aerodynamic performance because of the large inverse pressure gradient,strong three-dimensional effect and nonconstant characteristics.Advanced geometric configuration aerodynamic design optimization method is an important technical means to achieve high load and high performance of the compressor,but the complex curved blade of the centrifugal compressor will face HEB problems,namely High-Dimension,ExpensiveComputationally and Black-box.Box problem.Therefore,in this paper,based on three main aspects of aerodynamic design optimization methods:parameterization methods,meta-heuristic optimization algorithms(multi-island genetic algorithm,particle swarm algorithm,multi-agent model-assisted evolutionary algorithm and fast non-dominated ranking genetic algorithm)and optimization strategies,we propose an improvement scheme for the optimization of complex curved blade aerodynamic configuration design parameters and apply it to centrifugal compressor blade aerodynamic design.The optimization is applied to the aerodynamic design of centrifugal compressor blades.The main researches of the paper contain:(1)Based on type surface perturbation and parametric mapping dimensionality reduction techniques,a dual Bezier surface parameterization method(D-Bezier)is proposed.Through the blade section arc normalization and Bezier surface unitization,the mapping model of blade suction surface geometry configuration and two Bezier surfaces is established,the influence of Bernstein basis function on blade geometry configuration is studied,and the type surface parameterization of complex surface blade geometry configuration is realized.Compared with the traditional profile parameterization method,the D-Bezier mapping dimensionality reduction parameterization method has the advantages of low-dimensional variables,smooth surface,controllable global configuration and blade thickness,etc.(2)Based on the three-dimensional spatial free deformation technique and the geometric properties of the spline basis function,a mathematical model for the FFD parameterization of the complex curved blade of a centrifugal compressor is established,the mapping equations between the spatial mesh body and the complex surface are derived,the solution method of the Monte Carlo iterative algorithm to calculate the local parameters of the nonlinear system of equations is discussed,the influence of the B spline basis function on the geometric configuration of the blade is studied,and the three-dimensional spatial mesh parameterization of the complex curved blade is realized based on the control of the vertex deformation and the original geometric configuration.The B-sample basis FFD parameterization method has the advantages of universality,flexibility,applicability to arbitrary geometrical shapes and local support,which makes it easy to refine the geometry of complex surfaces.(3)Based on the addition principle and model management strategy of multiagent model-assisted evolutionary algorithm,an improved single-objective element heuristic CALS-PSO evolutionary algorithm for fast solution of constrained problems is proposed,a dynamic penalty model based on the activation function is established,the iteration direction change law of the global optimal solution is studied,and the intrinsic mechanism for the improvement of the quality-seeking quantity is elucidated.The comprehensive performance of multiple single-objective metaheuristic optimization algorithms is compared and analyzed,and it is verified that the proposed multi-agent model-assisted particle swarm algorithm CALS-PSO has significant advantages over traditional optimization algorithms in terms of iteration speed when solving larger-scale expensive computational real-world problems.(4)Based on the non-dominated sorting division and distance operator screening criteria,we proposed a multi-objective evolutionary algorithm DNSGA for improving population diversity,established the non-dominated sorting genetic algorithm Pareto rank classification and spatial density operator screening model,derived the equations for assigning weights to the decision variables of blade geometric configuration,studied the effects of sorting operators on progeny diversity,and elucidated the intrinsic mechanism of population diversity enhancement.mechanism.It is shown through a large number of independent reproducibility experiments that the improved multi-objective DNSGA evolutionary algorithm improves the exploration ability of global search for superiority and improves the solution quality of the black box problem.The correctness of the theoretical analysis model was verified by comparing the experimental data of centrifugal impeller with the numerical calculation results.(5)Based on the spatial complex surface blade parameterization method,improved meta-heuristic optimization algorithm and human-in-the-loop idea,an efficient dimensionality reduction optimization and multi-case segmentation fine optimization search model is established.The objective function and constraints of the blade aerodynamic design optimization problem in various application scenarios are discussed,the optimal configuration of the parameterization method and optimization algorithm is studied,and the optimal geometric configuration of the complex curved blade is efficiently solved with the isentropic efficiency maximization as the optimization objective,and the optimal control of the shape of the centrifugal compressor blade aerodynamic performance is realized.The changes of blade geometry and flow field structure before and after optimization are studied,and the intrinsic mechanism of aerodynamic performance improvement is elucidated.The superiority of the proposed blade parameterization method and the improved metaheuristic optimization algorithm in terms of convergence speed and amount of seeking quality is analyzed,which helps to solve the HEB technical problems in the optimization of complex curved blade aerodynamic design of centrifugal compressor.