numpy.bool_'对象没有属性'parameters'

Question

我遇到了一个奇怪的错误，报AttributeError: 'numpy.bool_' object has no attribute 'parameters'

import numpy as np
import math
import cvxpy as cp

def max_peak(power_signal,overall_ld):

    max_value = overall_ld[0] + power_signal[0]
    for i in range(1,time_slot):
        max_value= cp.maximum(max_value,\
                              overall_ld[i]+power_signal[i])
    return max_value

def offline_opt(overall_load):

    power_signal = cp.Variable(time_slot)
    
    obj_constraints = []

    for time in range(time_slot):
        SoC = 0
        for i in range(time):
            SoC += ((1-agg[3])**(i))*power_signal[time-i] 
        print("agg[0]=",agg[0])
        obj_constraints += [-agg[0]<=SoC, SoC<= agg[0]]
    
    prob = cp.Problem(cp.Minimize(max_peak(power_signal, overall_load)), \
                      constraints=obj_constraints)

    prob.solve(solver=cp.SCS)

    return prob.value

if __name__ == "__main__":

    time_slot = 3

    agg = np.array([10.0,2.0,2.0,0])
    
    offline_opt(np.array([10,5,3]))

问题出在行(obj_constraints += [-agg[0]<=SoC, SoC<= agg[0]])。但是，如果我将 agg[0] 替换为 10，-agg[0] 将替换为 -10。然后它通过并给了我一个结果。所以我检查了 agg[0],

的类型和值

IPdb [2]: agg[0]
10.0

IPdb [4]: type(agg[0])
<class 'numpy.float64'>

所以这是怎么回事？

谢谢！

完整堆栈：

  File "", line 1187, in <module>
    off_peak = offline_opt(overall_ld)

  File "", line 261, in offline_opt
    prob.solve(solver=cp.MOSEK)

  File "C:\Users\anaconda3\lib\site-packages\cvxpy\problems\problem.py", line 462, in solve
    return solve_func(self, *args, **kwargs)

  File "C:\Users\anaconda3\lib\site-packages\cvxpy\problems\problem.py", line 878, in _solve
    for parameter in self.parameters():

  File "C:\Users\anaconda3\lib\site-packages\cvxpy\utilities\performance_utils.py", line 70, in _compute_once
    result = func(self, *args, **kwargs)


  File "C:\Users\anaconda3\lib\site-packages\cvxpy\problems\problem.py", line 336, in parameters
    params += constr.parameters()

AttributeError: 'numpy.bool_' object has no attribute 'parameters'

更新：现在可以将代码复制并粘贴运行到您的机器上。它是独立的。

Answer 1

在 offline_opt 内循环的第一次迭代中，time 被设置为零。内部循环 (for i in range(time)) 不会发生（因为 range(0)），所以 SoC 仍然为零。

第一次迭代后的约束是： [-agg[0]<=0, 0<= agg[0]]，它们被评估为布尔值（因为没有涉及变量的表达式）。如果在第一次迭代后打印 obj_constraints 的值，您将看到它是 [True, True]。 CVXPY 不知道如何处理布尔值，因为它期望约束包含表达式。

“快速但愚蠢”的修复方法是在将 SoC 添加为约束之前检查它是否属于 cvxpy.Expression 类型：

if isinstance(SoC, cp.Expression):
    obj_constraints += [-agg[0]<=SoC, SoC<= agg[0]]

P.S - 而不是 [-agg[0]<=SoC, SoC<= agg[0]] 你可以简单地写 [SoC <= cp.abs(agg[0])]

numpy.bool_'对象没有属性'parameters'

numpy.bool_' object has no attribute 'parameters'

python

cvxpy