Оптимизируйте типы данных системы
В этом примере показано, как оптимизировать типы данных, используемые системой на основе заданных допусков.
Чтобы начаться, откройте систему, для которой вы хотите оптимизировать типы данных.
model = 'ex_auto_gain_controller'; sud = 'ex_auto_gain_controller/sud'; open_system(model)
Создайте fxpOptimizationOptions объект задать ограничения и допуски, чтобы удовлетворить вашим целям проекта. Установите UseParallel свойство fxpOptimizationOptions возразите против true запускать итерации оптимизации параллельно. Можно также задать размеры слова, чтобы позволить в проекте через AllowableWordLengths свойство.
opt = fxpOptimizationOptions('AllowableWordLengths', 10:24, 'UseParallel', true)
opt =
fxpOptimizationOptions with properties:
MaxIterations: 50
MaxTime: 600
Patience: 10
Verbosity: High
AllowableWordLengths: [10 11 12 13 14 15 16 17 18 19 20 21 22 23 24]
Advanced Options
AdvancedOptions: [1×1 struct]
Используйте addTolerance метод, чтобы задать допуски к различиям между исходным поведением системы и поведением с помощью оптимизированных типов данных с фиксированной точкой.
tol = 10e-2; addTolerance(opt, [model '/output_signal'], 1, 'AbsTol', tol);
Используйте fxpopt функционируйте, чтобы запустить оптимизацию. Программное обеспечение анализирует области значений объектов в вашей системе в соответствии с проектом и ограничениями, заданными в fxpOptimizationOptions объект применить типы гетерогенных данных к вашей системе при минимизации общей битной ширины.
result = fxpopt(model, sud, opt);
+ Preprocessing + Modeling the optimization problem - Constructing decision variables - Binding constraints + Running the optimization solver - Evaluating new solution: cost 180, does not meet the tolerances. - Evaluating new solution: cost 198, does not meet the tolerances. - Evaluating new solution: cost 216, does not meet the tolerances. - Evaluating new solution: cost 234, does not meet the tolerances. - Evaluating new solution: cost 252, does not meet the tolerances. - Evaluating new solution: cost 270, does not meet the tolerances. - Evaluating new solution: cost 288, does not meet the tolerances. - Evaluating new solution: cost 306, meets the tolerances. - Updated best found solution, cost: 306 - Evaluating new solution: cost 304, meets the tolerances. - Updated best found solution, cost: 304 - Evaluating new solution: cost 300, meets the tolerances. - Updated best found solution, cost: 300 - Evaluating new solution: cost 299, meets the tolerances. - Updated best found solution, cost: 299 - Evaluating new solution: cost 298, meets the tolerances. - Updated best found solution, cost: 298 - Evaluating new solution: cost 297, meets the tolerances. - Updated best found solution, cost: 297 - Evaluating new solution: cost 296, meets the tolerances. - Updated best found solution, cost: 296 - Evaluating new solution: cost 295, meets the tolerances. - Updated best found solution, cost: 295 - Evaluating new solution: cost 294, does not meet the tolerances. - Evaluating new solution: cost 293, meets the tolerances. - Updated best found solution, cost: 293 - Evaluating new solution: cost 292, meets the tolerances. - Updated best found solution, cost: 292 - Evaluating new solution: cost 291, meets the tolerances. - Updated best found solution, cost: 291 - Evaluating new solution: cost 290, meets the tolerances. - Updated best found solution, cost: 290 - Evaluating new solution: cost 289, meets the tolerances. - Updated best found solution, cost: 289 - Evaluating new solution: cost 287, meets the tolerances. - Updated best found solution, cost: 287 - Evaluating new solution: cost 283, meets the tolerances. - Updated best found solution, cost: 283 - Evaluating new solution: cost 282, meets the tolerances. - Updated best found solution, cost: 282 - Evaluating new solution: cost 281, meets the tolerances. - Updated best found solution, cost: 281 - Evaluating new solution: cost 280, meets the tolerances. - Updated best found solution, cost: 280 - Evaluating new solution: cost 279, meets the tolerances. - Updated best found solution, cost: 279 - Evaluating new solution: cost 278, meets the tolerances. - Updated best found solution, cost: 278 - Evaluating new solution: cost 277, does not meet the tolerances. - Evaluating new solution: cost 276, meets the tolerances. - Updated best found solution, cost: 276 - Evaluating new solution: cost 275, meets the tolerances. - Updated best found solution, cost: 275 - Evaluating new solution: cost 274, meets the tolerances. - Updated best found solution, cost: 274 - Evaluating new solution: cost 273, meets the tolerances. - Updated best found solution, cost: 273 - Evaluating new solution: cost 272, meets the tolerances. - Updated best found solution, cost: 272 - Evaluating new solution: cost 270, meets the tolerances. - Updated best found solution, cost: 270 - Evaluating new solution: cost 266, meets the tolerances. - Updated best found solution, cost: 266 - Evaluating new solution: cost 265, meets the tolerances. - Updated best found solution, cost: 265 - Evaluating new solution: cost 264, meets the tolerances. - Updated best found solution, cost: 264 - Evaluating new solution: cost 263, meets the tolerances. - Updated best found solution, cost: 263 - Evaluating new solution: cost 262, meets the tolerances. - Updated best found solution, cost: 262 - Evaluating new solution: cost 261, meets the tolerances. - Updated best found solution, cost: 261 - Evaluating new solution: cost 260, does not meet the tolerances. - Evaluating new solution: cost 259, does not meet the tolerances. - Evaluating new solution: cost 260, meets the tolerances. - Updated best found solution, cost: 260 - Evaluating new solution: cost 259, meets the tolerances. - Updated best found solution, cost: 259 - Evaluating new solution: cost 258, meets the tolerances. - Updated best found solution, cost: 258 - Evaluating new solution: cost 257, meets the tolerances. - Updated best found solution, cost: 257 - Evaluating new solution: cost 255, meets the tolerances. - Updated best found solution, cost: 255 - Evaluating new solution: cost 251, meets the tolerances. - Updated best found solution, cost: 251 - Evaluating new solution: cost 250, meets the tolerances. - Updated best found solution, cost: 250 - Evaluating new solution: cost 249, meets the tolerances. - Updated best found solution, cost: 249 - Evaluating new solution: cost 248, meets the tolerances. - Updated best found solution, cost: 248 - Evaluating new solution: cost 247, meets the tolerances. - Updated best found solution, cost: 247 - Evaluating new solution: cost 246, meets the tolerances. - Updated best found solution, cost: 246 - Evaluating new solution: cost 245, does not meet the tolerances. - Evaluating new solution: cost 244, does not meet the tolerances. - Evaluating new solution: cost 245, meets the tolerances. - Updated best found solution, cost: 245 - Evaluating new solution: cost 244, meets the tolerances. - Updated best found solution, cost: 244 + Optimization has finished. - Neighborhood search complete. - Maximum number of iterations completed. + Fixed-point implementation that met the tolerances found. - Total cost: 244 - Maximum absolute difference: 0.077478 - Use the explore method of the result to explore the implementation.
Используйте explore метод OptimizationResult объект, result, запустить Инспектора Данных моделирования и исследовать проект, содержащий самое маленькое общее количество битов при поддержании числовых допусков, заданных в opt объект.
explore(result);