for循环的另一种方法是按顺序迭代数据帧

Question

我有一个非常大的数据帧，用于一年中每一秒的电池放电时间表。在

基本的系列事件是

• 事件发生
• 蓄电池放电
• 停止放电
• 放电停止后X秒，开始充电
• 电池充满电后停止充电

数据帧看起来像这样。。。（请原谅我糟糕的格式）

Index | Freq | Case | Battery OP | Power Required | Battery Energy | SOC | Response timer | Charge Power |

01/01/2016 | 49.5862 | C | Discharging | 300.512 | 1500 | 99.85 | 3 | 0 |

01/01/2016 | 49.5862 | C | Charging    | 0       | 1500 | 99.85 | 3 | 1500 |

我目前正在使用for循环和一些if/elseif语句遍历每一行，检查电池是否需要充电。在

不过，我觉得效率很低。我可能快没记性了，要不然就得几天才能完成。在

我让它运行了一个周末，它仍然没有完成

我确定有更好的方法来做这个，但我不知道。问题是它必须是连续的。每秒钟需要根据电池的电量和之前的SOC%/能量计算充电状态或电池能量。在

这里有可复制的代码（尽量减少）

^{pr2}$

Answer 1

由于内存不足，最好的方法是使用panda数据帧的apply方法。这种方法称为矢量化。在

一个例子如下df.apply(numpy.sqrt, axis=1)

您可以查看文档以了解更多详细信息：http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.apply.html#pandas.DataFrame.apply

Answer 2

这就是我可能重新编写代码的方法。我只是简单地将最初的7列简化为一个字典，然后使用pd.DataFrame()将它们转换为一个适当的数据帧。然后，我只需在遍历构造的数据帧时应用if...elif语句。在

import numpy as np
import pandas as pd


Battery_W = 1000
Battery_Wh = 1000/ 3
starting_SOC = 0.75
charge_delay = 5
charging = True

#initialize test Dictionary 
test = {}

#add your test elements as a tuple
data = (50.00,50.00,49.99,49.98,49.87,49.76,49.65,49.25,50.00,50.00,50.00,50.00,50.00,50.00,49.99,49.78,49.67,49.46,49.25,49.25,50.00,50.00,50.00,49.95,49.65,49.45,49.65,49.55,50.00,50.00,50.00,50.00,50.00,50.00,50.00,49.95,49.65,49.45,49.65,49.55,49.99,49.68,50.00,50.00,50.00,50.00,50.00,50.00,50.00,50.00)
index = 0
#"Battery Operation mode" is not calculated seperately now
#"Calculate power output required" is not calculated seperately now
for d in data:
    if d <= 49.75:
    test[index] = {'Freq': d,
                   'Case': 'C',
                   'Battery_OP':'Discharging',
                   'Power_Required':0,
                   'Battery_Energy':0.0,
                   'SOC':0,
                   'Charge_Power' :0.0
                   }
    elif d > 49.75:
        test[index] = {'Freq': d,
                   'Case': 'B',
                   'Battery_OP': 'Idle',
                   'Power_Required': 1000,
                   'Battery_Energy': 0.0,
                   'SOC': 0,
                   'Charge_Power': 0.0}
   index +=1
#This is how I convert the dictionary into a df for the first-time
year_test = pd.DataFrame(test.values())

year_test['Response_timer'] = year_test.groupby('Battery_OP').cumcount()
year_test['Response_timer'][year_test['Battery_OP'] == 'Idle'] = 0

year_test['ChargePower'] = 0.00
year_test['BatteryEnergy'] = 0.0
year_test['BatteryEnergy'].iloc[0:charge_delay] = Battery_Wh * starting_SOC

j = charge_delay
#instead of using the range(), try to manipulate it using `itertuples()`
#This is most probably where you are losing your time..
for row in year_test.itertuples():
    if row.Index <5:
        continue
    if year_test.iloc[j-charge_delay, 3]  > 0 and year_test.iloc[j - ((charge_delay) -1), 3] == 0 :
        "charge at max rate"
        year_test.iloc[j,7] = Battery_W
        year_test.iloc[j,2] = "Charging"
        charging = True

    elif charging == True and year_test.iloc[j-1,4] < starting_SOC * Battery_Wh:
        "check if battery charged"
        year_test.iloc[j,7] = Battery_W
        year_test.iloc[j,2] = "Charging"

    elif year_test.iloc[j-1,4] >= starting_SOC * Battery_Wh or charging == False:
        charging = False
        year_test.iloc[j,7] = 0.0

    "New Battery Energy"
    year_test.iloc[j,4] = year_test.iloc[(j-1),4] - ((year_test.iloc[j,3])/60/60) + ((year_test.iloc[j,7])/60/60)
    if year_test.iloc[j,4] > Battery_Wh :
        year_test.iloc[j,4] = Battery_Wh

"Calculate battery SOC% for empty"
year_test['SOC'] = year_test['BatteryEnergy'] / Battery_Wh * 100