能否给出一个三目标的例子代码

当然可以！以下是一个使用MOEA/D算法解决三目标优化问题的简单示例代码：

import numpy as np

# 定义目标函数
def objective_function(x):
    f1 = x[0]**2
    f2 = (x[0]-2)**2
    f3 = (x[0]-4)**2
    return [f1, f2, f3]

# 定义MOEA/D算法
def moead(population_size, num_generations):
    num_objectives = 3
    num_variables = 1
    num_neighbors = 5
    
    # 初始化权重向量
    weights = np.random.uniform(size=(population_size, num_objectives))
    weights = weights / np.sum(weights, axis=1)[:, np.newaxis]
    
    # 初始化种群
    population = np.random.uniform(size=(population_size, num_variables))
    
    # 迭代进化
    for generation in range(num_generations):
        for i in range(population_size):
            # 选择邻居个体
            neighbors_indices = np.random.choice(population_size, num_neighbors, replace=False)
            neighbors = population[neighbors_indices]
            
            # 计算子问题适应度值
            subproblem_fitness = []
            for j in range(num_neighbors):
                fitness = objective_function(neighbors[j])
                subproblem_fitness.append(fitness)
                
            # 更新个体解集
            min_index = np.argmin(subproblem_fitness)
            population[i] = neighbors[min_index]
        
    return population

# 调用MOEA/D算法进行优化
population_size = 100
num_generations = 100

final_population = moead(population_size, num_generations)

# 输出最终的 Pareto 最优解集
for i in range(population_size):
    fitness = objective_function(final_population[i])
    print("Solution ", i+1, ": ", final_population[i], " Fitness: ", fitness)

请注意，这只是一个简单的示例，目标函数是手动定义的。在实际应用中，您需要根据具体的问题来定义自己的目标函数，并根据问题的特点进行适当的调整和改进。