Application Study Of The Particle Swarm Optimization (PSO) Algorithm In The Optimization Of Parametric Building Information Modeling (BIM)

As the increasing application of various optimization algorithms in AEC community,especially in the field of building performance optimization,intelligent algorithms gradually assert the dominance over traditional algorithms due to being not strict with the conditions of decision variables,objective functions and constraints,and able to deal with large-scale and complex optimization problems).However,existing studies have limitations,for example,the integration of intelligent algorithms with BIM models needs the assistance of third party software or platforms,which leads to the application’s inconvenience for the designers;the optimization algorithms may be too random;and the verification of the calculation requires the calibration of operational parameters: targeting on specific problems,different algorithms should use different settings of the parameters.Considering the above problems,this thesis based on Optimo(a multi-objective optimization plug-in on Dynamo,built-in with the NSGA-II algorithm),implement the adding,testing and applying study of the Particle Swarm Optimization(PSO)algorithm on Revit + Dynamo parametric design BIM platform.The Multi-Objective Particle Swarm Optimization(MOPSO)algorithm is realized by hybrid programming of MATLAB and C# codes,employing the powerful scientific calculation capacity of MATLAB to create the PSO algorithm core and its testing.The solving of three commonly used test functions for multi-objective optimization,proves the validity of the coded PSO solver.Additionally,on the Microsoft Visual Studio 2017 IDE,this paper accomplishes the encapsulation,transformation and transmission of the variables,enabling the solver core to read the input data from Dynamo,and exporting the calculation results and major data into an Excel file.Finally,a case study of indoor artificial illumination design optimization is conducted,taking the illumination energy consumption and the degree of illumination uniformity as optimization objectives.This case verifies the validity and feasibility of the presented optimizing tool with MOPSO algorithm as its core and showcases the workflow of the optimizing tool application.The innovations of this thesis mainly twofolded:1.Algorithm Coding for architectural design optimization:Introducing nonlinear constraints for applications in broader conditions;introducing panelty functions to improve convergence speed;and introducing the coefficient of variation to maintain population diversity.2.Input and Output on the Dynamo Interface:Employing function handle in input nodes to improve comprehensibility of function inputs;exporting calculation results into an Excel file automatically to improve the convenience for later data processing and analyzing.