蒙特卡洛在Brightway2

问题描述

我在运行蒙特卡洛时遇到问题，我不确定问题出在哪里或什么地方。正常的LCA计算运行良好，我检查了交易所是否增加了不确定性。我也尝试用两种不同的方式编写代码，但是问题是一样的，结果是：[nan,nan,nan]

enter image description here

下面是我正在使用的代码

 mymethod1 = ('ReCiPe Midpoint (H) V1.13 no LT','climate change','GWP100')
 mymethod1

 el = t_db.get("chemical1 production")   
 functional_unit = {el: 1}
 lca = bw.LCA(functional_unit,mymethod1)
 lca.lci()
 lca.lcia()
 print(lca.score)

 mc = bw.MonteCarloLCA({el: 1},mymethod1)  
 mc_results = [next(mc) for x in range(5)] 
 mc_results

 #also tried doing like this
 mc = bw.MonteCarloLCA({bw.Database('exldb').get('chemical1 production'):1},mymethod1) 
 next(mc)
 mc_results = [next(mc) for x in range(5)] 
 mc_results

解决方法

我怀疑问题出在库存中，而不是影响评估方法中。我将确保您具有定义正确定义的概率分布所需的所有参数。在表here中，您可以看到需要根据所使用的概率分布类型定义哪些参数。

您可以写：

import math as math

for Input,Output,Row,Col,Type,Uncertainty_Type,Amount,Loc,Scale,Shape,Min,Max,Negative in lca.tech_params:

if Uncertainty_Type in [0,1]:
    assert math.isnan(Loc) is False
    
    assert math.isnan(Scale) or (Scale is None)
    assert math.isnan(Shape) or (Shape is None)
    assert math.isnan(Min) or (Min is None)
    assert math.isnan(Max) or (Max is None)
    
elif Uncertainty_Type in [2,3]:
    # lognormal and normal
    assert math.isnan(Loc) is False
    assert math.isnan(Scale) is False
    
    assert math.isnan(Shape) or (Shape is None)

elif Uncertainty_Type == 4:
    # uniform
    assert math.isnan(Max) is False
    
    assert math.isnan(Loc) or (Scale is None),f" loc {Loc}"
    assert math.isnan(Scale) or (Scale is None)
    assert math.isnan(Shape) or (Shape is None)

elif Uncertainty_Type in [5,7]:
    # triangular and discrete uniform
    assert math.isnan(Max) is False

elif Uncertainty_Type in [8,9]:
    # weibull and Gamma
    assert math.isnan(Scale) is False
    assert math.isnan(Shape) is False
    
elif Uncertainty_Type in [10,11]:
    # beta,generalized extreme value
    assert math.isnan(Loc) is False
    assert math.isnan(Shape) is False
    
elif Uncertainty_Type == 12:
    # student
    assert math.isnan(Shape) is False

检查是否已定义所有不确定性参数。

更新：stat-arrays存储库针对值不同于np.nan的可选列发出警告。我添加了一些断言来检查这一点..并意识到对于ecoinvent（3.6结果）不是这种情况

brightway montecarlo