增加解的评价指标，最终解的重复过滤和排序，可选输出数量
This commit is contained in:
parent
6a482e6af1
commit
dfdb10d207
|
|
@ -71,7 +71,7 @@ class Config:
|
|||
"""算法参数配置类：存储NSGA-II的各类参数"""
|
||||
def __init__(self):
|
||||
# 种群参数
|
||||
self.pop_size = 200 # 种群大小
|
||||
self.pop_size = 300 # 种群大小
|
||||
self.N1_ratio = 0.2 # 优先成本的种群比例
|
||||
self.N2_ratio = 0.2 # 优先延期的种群比例
|
||||
self.N3_ratio = 0.3 # 强制风险企业的种群比例
|
||||
|
|
@ -79,10 +79,33 @@ class Config:
|
|||
# 遗传操作参数
|
||||
self.crossover_prob = 0.8 # 交叉概率
|
||||
self.mutation_prob = 0.3 # 变异概率
|
||||
self.max_generations = 100 # 最大进化代数
|
||||
self.max_generations = 500 # 最大进化代数
|
||||
# 惩罚系数
|
||||
self.delta = 1.3 # 变更惩罚系数
|
||||
# 早停参数
|
||||
self.early_stop_patience = 50 # 连续多少代无改进则早停
|
||||
# 目标函数数量
|
||||
self.objective_num = 2 # 双目标（成本+延期）
|
||||
self.duplicate_threshold = 0.1 # 重复解保留数量
|
||||
self.print_top_n = 5 # 打印前N个最优解
|
||||
|
||||
class DataStructures:
|
||||
"""数据结构工具类：提供评价指标计算等功能"""
|
||||
@staticmethod
|
||||
def calculate_evaluation_index(objectives, optimal_cost, optimal_tardiness):
|
||||
"""
|
||||
计算评价指标
|
||||
:param objectives: 解的目标值 (成本, 延期)
|
||||
:param optimal_cost: 最优成本值
|
||||
:param optimal_tardiness: 最优延期值
|
||||
:return: 评价指标值
|
||||
"""
|
||||
cost, tardiness = objectives
|
||||
# 避免除以零（成本最优值为0时的保护）
|
||||
if optimal_cost == 0:
|
||||
cost_ratio = 0 if cost == 0 else float('inf')
|
||||
else:
|
||||
cost_ratio = cost / optimal_cost
|
||||
# 延期处理（+1避免除以零）
|
||||
tardiness_ratio = (tardiness + 1) / (optimal_tardiness + 1)
|
||||
return cost_ratio + tardiness_ratio
|
||||
62
main.py
62
main.py
|
|
@ -7,6 +7,8 @@ from encoder import Encoder
|
|||
from genetic_operators import GeneticOperator
|
||||
from nsga2 import NSGA2
|
||||
from visualizer import ResultVisualizer
|
||||
from data_structures import DataStructures
|
||||
"""主函数：执行NSGA-II算法求解多目标优化问题（整数化版本）"""
|
||||
def main():
|
||||
"""主函数：执行NSGA-II算法求解多目标优化问题（整数化版本）"""
|
||||
try:
|
||||
|
|
@ -31,6 +33,7 @@ def main():
|
|||
print("初始化种群（整数化）...")
|
||||
population = encoder.initialize_population()
|
||||
print(f"初始化种群完成，（种群大小，染色体长度）: {population.shape if population.size > 0 else '空'}")
|
||||
|
||||
# 若种群初始化失败（为空），直接退出
|
||||
if population.size == 0:
|
||||
print("错误：种群初始化失败，无法继续进化")
|
||||
|
|
@ -53,6 +56,7 @@ def main():
|
|||
ranks, fronts = nsga2.fast_non_dominated_sort(objectives)
|
||||
current_front = fronts[0] if fronts else [] # 第0层为最优前沿
|
||||
current_front_objs = [objectives[i] for i in current_front] if current_front else []
|
||||
|
||||
best_front = population[current_front] if current_front else [] # 更新当前最优前沿解（整数）
|
||||
best_front_objs = current_front_objs # 更新当前最优前沿目标值（整数）
|
||||
# 记录收敛趋势（基于最优前沿的平均目标值，整数）
|
||||
|
|
@ -75,6 +79,7 @@ def main():
|
|||
# 校验选择后的种群大小
|
||||
assert len(selected) == config.pop_size, \
|
||||
f"选择后的种群大小({len(selected)})与目标大小({config.pop_size})不符"
|
||||
|
||||
# 交叉操作（两点交叉）- 修复索引越界问题
|
||||
offspring = [] # 子代种群（整数）
|
||||
selected_len = len(selected) # selected的长度（等于pop_size）
|
||||
|
|
@ -98,6 +103,7 @@ def main():
|
|||
if len(offspring) < config.pop_size:
|
||||
offspring.append(child2)
|
||||
i += 2 # 处理下一对个体
|
||||
|
||||
else:
|
||||
# 直接添加当前父代（避免i+1越界）
|
||||
offspring.append(selected[i])
|
||||
|
|
@ -119,6 +125,7 @@ def main():
|
|||
population, objectives = nsga2.environmental_selection(combined, combined_objs)
|
||||
# 校验环境选择后的种群大小和整数类型
|
||||
population = population.astype(int)
|
||||
|
||||
assert len(population) == config.pop_size, \
|
||||
f"环境选择后的种群大小({len(population)})与目标大小({config.pop_size})不符"
|
||||
# 早停检查（连续多代无改进则停止）
|
||||
|
|
@ -135,9 +142,62 @@ def main():
|
|||
# 7. 结果可视化与输出（整数化）
|
||||
print("进化完成，处理结果（整数化）...")
|
||||
if len(best_front_objs) > 0:
|
||||
# 1. 过滤重复解
|
||||
unique_front = []
|
||||
unique_front_objs = []
|
||||
seen = set()
|
||||
|
||||
for sol, obj in zip(best_front, best_front_objs):
|
||||
# 将解和目标值组合为元组用于查重
|
||||
sol_tuple = tuple(sol.tolist())
|
||||
obj_tuple = tuple(obj)
|
||||
if (sol_tuple, obj_tuple) not in seen:
|
||||
seen.add((sol_tuple, obj_tuple))
|
||||
unique_front.append(sol)
|
||||
unique_front_objs.append(obj)
|
||||
|
||||
# 2. 计算评价指标并排序
|
||||
if unique_front_objs:
|
||||
# 找出各目标的最优值
|
||||
optimal_cost = min(obj[0] for obj in unique_front_objs)
|
||||
optimal_tardiness = min(obj[1] for obj in unique_front_objs)
|
||||
|
||||
# 计算每个解的评价指标
|
||||
evaluated_solutions = []
|
||||
for sol, obj in zip(unique_front, unique_front_objs):
|
||||
index = DataStructures.calculate_evaluation_index(
|
||||
obj, optimal_cost, optimal_tardiness
|
||||
)
|
||||
evaluated_solutions.append((sol, obj, index))
|
||||
|
||||
# 按评价指标升序排序
|
||||
evaluated_solutions.sort(key=lambda x: x[2])
|
||||
|
||||
# 选择前n个解
|
||||
top_n = min(config.print_top_n, len(evaluated_solutions))
|
||||
top_solutions = evaluated_solutions[:top_n]
|
||||
top_population = [sol for sol, _, _ in top_solutions]
|
||||
top_objectives = [obj for _, obj, _ in top_solutions]
|
||||
|
||||
# 打印排序后的解的评价指标
|
||||
print("\n" + "=" * 100)
|
||||
print(f"排序后的前{top_n}个最优解（评价指标从小到大）")
|
||||
print("=" * 100)
|
||||
for i, (_, obj, idx) in enumerate(top_solutions):
|
||||
print(f"解 {i + 1}: 成本={obj[0]}, 延期={obj[1]}, 评价指标={idx:.4f}")
|
||||
else:
|
||||
top_population = []
|
||||
top_objectives = []
|
||||
|
||||
# 保持原有图表绘制逻辑不变
|
||||
visualizer.plot_pareto_front(all_objectives, best_front_objs) # 绘制帕累托前沿（整数）
|
||||
visualizer.plot_convergence(convergence_history) # 绘制收敛趋势（整数）
|
||||
visualizer.print_pareto_solutions(best_front, best_front_objs) # 打印最优解详情（整数）
|
||||
|
||||
# 打印处理后的最优解详情
|
||||
if top_population:
|
||||
visualizer.print_pareto_solutions(top_population, top_objectives)
|
||||
else:
|
||||
print("未找到有效帕累托前沿解")
|
||||
else:
|
||||
print("未找到有效帕累托前沿解")
|
||||
except Exception as e:
|
||||
|
|
|
|||
|
|
@ -1,6 +1,9 @@
|
|||
# 染色体工具类
|
||||
"""结果可视化工具类：绘制帕累托前沿、收敛曲线，打印最优解详情（适配整数化）"""
|
||||
import matplotlib.pyplot as plt
|
||||
import numpy as np
|
||||
from chromosome_utils import ChromosomeUtils # 染色体工具类
|
||||
from chromosome_utils import ChromosomeUtils
|
||||
|
||||
class ResultVisualizer:
|
||||
"""结果可视化工具类：绘制帕累托前沿、收敛曲线，打印最优解详情（适配整数化）"""
|
||||
def __init__(self, utils: ChromosomeUtils):
|
||||
|
|
@ -12,6 +15,7 @@ class ResultVisualizer:
|
|||
self.figsize = (12, 8) # 图表大小
|
||||
# 企业名称列表（风险企业+供应商）
|
||||
self.supplier_names = ["RiskEnterprise"] + self.utils.supplier.names
|
||||
|
||||
def plot_pareto_front(self, all_objectives: list[tuple[int, int]],
|
||||
non_dominated_objectives: list[tuple[int, int]]):
|
||||
"""
|
||||
|
|
@ -38,6 +42,7 @@ class ResultVisualizer:
|
|||
plt.grid(True, alpha=0.5)
|
||||
plt.savefig('pareto_front_integerized.png', dpi=300, bbox_inches='tight') # 保存图片
|
||||
plt.show()
|
||||
|
||||
def plot_convergence(self, convergence_history: list[tuple[int, int]]):
|
||||
"""
|
||||
绘制收敛趋势图（每代最优前沿的平均成本和延期，整数化）
|
||||
|
|
@ -64,6 +69,7 @@ class ResultVisualizer:
|
|||
plt.tight_layout() # 调整布局
|
||||
plt.savefig('convergence_history_integerized.png', dpi=300, bbox_inches='tight') # 保存图片
|
||||
plt.show()
|
||||
|
||||
def print_solution_details(self, solution: np.ndarray, objectives: tuple[int, int]):
|
||||
"""
|
||||
打印单个解的详细信息（企业选择、产能、数量等，整数化）
|
||||
|
|
@ -101,16 +107,17 @@ class ResultVisualizer:
|
|||
print("✅ 所有物料数量满足需求约束（整数匹配）")
|
||||
else:
|
||||
print("❌ 部分物料数量未满足需求约束")
|
||||
|
||||
def print_pareto_solutions(self, population: np.ndarray, objectives: list[tuple[int, int]]):
|
||||
"""
|
||||
打印帕累托前沿解的详细信息（最多打印前5个，整数化）
|
||||
:param population: 帕累托前沿解的种群（整数）
|
||||
:param objectives: 对应的目标值列表（整数）
|
||||
打印排序后的帕累托前沿解详细信息（按评价指标排序）
|
||||
:param population: 排序后的帕累托前沿解种群
|
||||
:param objectives: 对应的目标值列表
|
||||
"""
|
||||
print("\n" + "=" * 100)
|
||||
print("帕累托前沿解详细方案（整数化）")
|
||||
print("按评价指标排序的帕累托前沿解详细方案（整数化）")
|
||||
print("=" * 100)
|
||||
# 打印前5个解（或所有解，取较少者）
|
||||
for i in range(min(5, len(population))):
|
||||
print(f"\n===== 最优解 {i + 1} =====")
|
||||
# 打印所有选定的解
|
||||
for i in range(len(population)):
|
||||
print(f"\n===== 排序解 {i + 1} =====")
|
||||
self.print_solution_details(population[i], objectives[i])
|
||||
Loading…
Reference in New Issue