多重力辅助低推力（MGALT）版本的航天器轨道优化套件（STOpS）附matlab代码.zip

function [f_best_return,x_star_best_return,is_viable,neval] = ... MGALT_MBH_function(BOD,CONST,OPT,OPT_algo,VAR,x0,num_mig,count_MBH,loop) % FORM: [f_best_return,x_star_best_return,is_viable,neval] = ... % MGALT_MBH_function(BOD,CONST,OPT,OPT_algo,VAR,x0,num_mig,count_MBH,loop) % % |----------------------------------------------------------------------- % | % | NOTES: % | -This function is the actual implimentation of the MBH algorithm. % | The algorithm had to be broken into a function because it would % | not be feasible to fit into "MGALT_MBH" due to organization % | % | -A user who would like solutions where ALL of the transfer % | trajectories are feasible needs to chenge the comment under % | section 3.0 and 4.3 from "any(feasible)" to "all(feasible)". The % | rational for having any was explained in the accompanying thesis, % | as the randomness could potentially alows MBH to find a solution % | where both transfers are considered feasible. As on now, it will % | accept a solution where any transfer is feasible % | % |----------------------------------------------------------------------- % | % | INPUTS: % | -BOD (1,1) [struct] [unitless] % | A struct containing information pertaining to the planetary % | bodies. Contains list of bodies, launch windows and ToF, and % | planetary R/V/JD vectors. This struct has dynamic fields and % | will adapt to contain only the necesary information % | -CONST (1,1) [struct] [unitless] % | A struct containing constants used in the calcs. Contains % | values for AU, TU, Sun (rad/mu/rp) and (rad/mu/rp/SOI/per) % | for any bodies used in the optimization scheme. This is a % | dynamic struct and will adapt to contain only the necesary % | information % | -OPT (1,1) [struct] [unitless] % | A struct containing constants user options. Contains the save % | folder, ToF values, and more structs containing informaiton % | for the island model, cost parameters, weighting parameters, % | and all of the islands used in the optimization process % | -OPT_algo (1,1) [struct] [unitless] % | MBH option parameters. For a full explination of these % | parameters, see % | "Algorithms/Algorithm_Parameters/parametersMBH.m" % | -VAR (1,1) [struct] [unitless] % | A struct containing the variable limits % | -x0 (1,Nvar) [float] [unitless] % | The initial population which will be evaluated by the solver % | -num_mig (1,1) [int] [unitless] % | The current migration number % | -count_MBH (1,1) [int] [unitless] % | The current MBH island number % | -loop (1,1) [int] [unitless] % | The current loop number % | This is if the function is solving an initial set of members % | or a secondary set of members generated from MGALT_MBH_cluster % | % |----------------------------------------------------------------------- % | % | OUTPUTS: % | -f_best (1,Nvar) [float] [unitless] % | The respective cost of each member in 'x0' % | -member_final (1,Nvar) [float] [unitless] % | The new perturbed member consisting of the original member and % | the perturbations added to it % | -is_viable (1,1) [boolean] [unitless] % | The new perturbed member consisting of the original member and % | the perturbations added to it % | -nfeval (1,1) [int] [unitless] % | The current number of iterations for this MBH object % | % |----------------------------------------------------------------------- % | % | MISC: % | % |----------------------------------------------------------------------- %% Initialize % Number of iterations num_iter_global = size(x0,1); % Number of iterations for the outer loop % Preallocate Cost Parameters f = zeros(num_iter_global,1); f_best_return = zeros(num_iter_global,1); x_star_best_return = zeros(num_iter_global,size(x0,2)); is_viable = zeros(num_iter_global,1); % Perallocate the plot_vars struct. Need to run 1 execution to get field % names [~,vars] = feval(OPT.solver,x0(1,:),BOD,CONST,OPT,VAR); name = fieldnames(vars); for i0 = 1:length(name) plot_vars(num_iter_global).(name{i0}) = []; end %% feval if OPT.parallel % Function Handle fh = str2func(string(OPT.solver)); % https://www.mathworks.com/help/matlab/ref/str2func.html % Display info ll_par = fprintf('\n Parallel Processing \n'); ll_RAM = fprintf(2,' Watch CPU/RAM usage \n\n'); % Chunks to break parallel into to prevent a lot of RAM usage at once chunk_nums = 1000; % Set by Malloy, uses ~2.0GB of RAM per pass chunk_whole = floor(num_iter_global/chunk_nums); chunk_rems = mod(num_iter_global,chunk_nums); % Run through all searches minus mod for chunk_iter = 1:chunk_whole % Preallocate temp vars f_temp = zeros(chunk_nums,1); for i0 = 1:length(name) plot_vars_temp(chunk_nums).(name{i0}) = []; end data_temp(1:chunk_nums) = parallel.FevalFuture; % Display info ll_loop1 = fprintf(' Search: %1.0f-%1.0f / %1.0f\n',((chunk_iter-1)*chunk_nums+1),(chunk_iter*chunk_nums),num_iter_global); ll_loop2 = fprintf(' Search: %1.0f\n',(chunk_iter-1)*chunk_nums+1); % Perform parfeval on chunk n for chunk_num = 1:chunk_nums % Search Display fprintf(repmat('\b',1,ll_loop2)) ll_loop2 = fprintf(' Search: %1.0f\n',((chunk_iter-1)*chunk_nums)+chunk_num); data_temp(chunk_num) = parfeval(OPT.ppool,fh,2,x0(((chunk_iter-1)*chunk_nums)+chunk_num,:),BOD,CONST,OPT,VAR); end fprintf(repmat('\b',1,ll_loop2)) % Collect results for chunk n ll_collect = fprintf('\n ---Collecting results from parpool---\n'); for chunk_num = 1:chunk_nums [index,cost,plt_vars] = fetchNext(data_temp); f_temp(index) = cost; plot_vars_temp(index) = plt_vars; end % Save the collected results into the total struct f( ((chunk_iter-1)*chunk_nums+1):(chunk_iter*chunk_nums) ) = f_temp; plot_vars( ((chunk_iter-1)*chunk_nums+1):(chunk_iter*chunk_nums) ) = plot_vars_temp; % Clear display fprintf(repmat('\b',1,ll_collect)) fprintf(repmat('\b',1,ll_loop1)) clear f_temp plot_vars_temp data_temp end % Run through all remaining values if logical(chunk_rems) % Preallocate temp vars clear f_temp plot_vars_temp data_temp f_temp = zeros(chunk_rems,1); for i0 = 1:length(name) plot_vars_temp(chunk_rems).(name{i0}) = []; end data_temp(1:chunk_rems) = parallel.FevalFuture; % Display info ll_loop1 = fprintf(' Search: %1.0f-%1.0f / %1.0f\n',(chunk_whole*chunk_nums+1),(chunk_whole*chunk_nums+chunk_rems),num_iter_global); ll_loop2 = fprintf(' Search: %1.0f\n',(chunk_whole*chunk_nums+1)); % Perform parfeval on remainder for chunk_rem = 1:chunk_rems % Search Display fprintf(repmat('\b',1,ll_loop2)) ll_loop2 = fprintf(' Search: %1.0f\n',(chunk_whole*chunk_nums+chunk_rem)); data_temp(chunk_rem) = parfeval(OPT.ppool,fh,2,x0((chunk_whole*chunk_nums+chunk_rem),:),BOD,CONST,OPT,VAR); end fprintf(repmat('\b',1,ll_loop2)) % Collect