粒子群遗传算法代码用于优化交通列车时刻表

以下是基于粒子群遗传算法的交通列车时刻表优化代码示例：

import random
import numpy as np

# 定义常量
POP_SIZE = 100  # 种群大小
DNA_SIZE = 24   # DNA长度
CROSS_RATE = 0.8  # 交叉率
MUTATION_RATE = 0.05  # 变异率
N_GENERATIONS = 1000  # 迭代次数
N_PARTICLES = 20 # 粒子数
N_ITERATIONS = 20 # 迭代次数

# 生成随机时刻表
def generate_schedule(dna_size):
    return [random.randint(0, 23) for _ in range(dna_size)]

# 计算适应度函数
def fitness(schedule):
    # 计算列车之间的平均时间间隔
    diffs = [schedule[i+1]-schedule[i] for i in range(len(schedule)-1)]
    avg_diff = np.mean(diffs)
    return 1 / avg_diff

# 基因重组
def crossover(parent, pop, cross_rate):
    if np.random.rand() < cross_rate:
        i_ = np.random.randint(POP_SIZE, size=1)
        cross_points = np.random.randint(0, 2, DNA_SIZE).astype(np.bool)
        parent[cross_points] = pop[i_, cross_points]
    return parent

# 基因突变
def mutate(child, mutation_rate):
    for point in range(DNA_SIZE):
        if np.random.rand() < mutation_rate:
            child[point] = np.random.randint(0, 24)
    return child

# 粒子初始化
def particle_initialize(dna_size):
    return np.array([generate_schedule(dna_size) for _ in range(N_PARTICLES)])

# 粒子适应度计算
def particle_fitness(particles):
    fitnesses = np.array([fitness(p) for p in particles])
    return fitnesses

# 粒子最优适应度及最优解更新
def particle_update(particles, fitnesses, best_particle, best_fitness):
    # 找到当前粒子群中最优的适应度和对应的粒子
    max_fitness_index = np.argmax(fitnesses)
    if fitnesses[max_fitness_index] > best_fitness:
        best_fitness = fitnesses[max_fitness_index]
        best_particle = particles[max_fitness_index]
    return best_particle, best_fitness

# 粒子群算法
def particle_swarm_optimization(dna_size):
    # 初始化粒子群及其适应度
    particles = particle_initialize(dna_size)
    fitnesses = particle_fitness(particles)

    # 初始化最优解及其适应度
    best_particle = particles[0]
    best_fitness = fitnesses[0]

    # 迭代寻找最优解
    for i in range(N_ITERATIONS):
        # 粒子更新
        for j in range(N_PARTICLES):
            particles[j] = crossover(particles[j], particles, CROSS_RATE)
            particles[j] = mutate(particles[j], MUTATION_RATE)
        # 计算适应度
        fitnesses = particle_fitness(particles)
        # 更新最优解
        best_particle, best_fitness = particle_update(particles, fitnesses, best_particle, best_fitness)
    return best_particle

# 测试
best_schedule = particle_swarm_optimization(DNA_SIZE)
print(f'最优时刻表: {best_schedule}')
print(f'最优适应度: {fitness(best_schedule)}')

该代码使用粒子群遗传算法来寻找最优的交通列车时刻表。在初始化时，首先生成一个包含多个粒子（即多个随机时刻表）的粒子群，并计算每个粒子的适应度（即该时刻表的列车之间的平均时间间隔）。随后，使用交叉和变异操作来更新每个粒子，并再次计算每个粒子的适应度。在多次迭代后，最终找到具有最优适应度的粒子，即最优的交通列车时刻表。