cpk_curve.py

#!/usr/bin/env python
# coding: utf-8

# # 计算cpk并绘图
# 
# ## 导入使用到的modules

# In[9]:


import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
import math

title = None
legendlist = []


# ## 定义计算cpk和绘图的函数

# In[10]:


def cpk_calc(df_data: pd.DataFrame, usl, lsl,serialTitle):
    """
    :param df_data: 数据dataframe
    :param usl: 数据指标上限
    :param lsl: 数据指标下限
    :return:
    """
    global title
    sigma = 3
    # 若下限为0, 则使用上限反转负值替代
    if int(lsl) == 0:
        lsl = 0 - usl
 
    # 数据平均值
    #print(type(df_data.mean()))
    #print(df_data.mean())


    u = df_data.mean()[serialTitle]

    #print('mean',u)

    # 数据标准差
    stdev = np.std(df_data.values, ddof=1)
    #print('stdev',stdev)

    # 生成横轴数据平均分布
    x1 = np.linspace(u - sigma * stdev, u + sigma * stdev, 1000)
    #print(type(x1))

    # 计算正态分布曲线
    y1 = np.exp(-(x1 - u) ** 2 / (2 * stdev ** 2)) / (math.sqrt(2 * math.pi) * stdev)
    #print(type(y1))

    cpu = (usl - u) / (sigma * stdev)
    cpl = (u - lsl) / (sigma * stdev)
    # 得出cpk
    cpk = min(cpu, cpl)

    # 使用matplotlib画图
    plt.xlim(x1[0] - 0.5, x1[-1] + 0.5)
    plt.plot(x1, y1)


    #plt.hist(df_data.values, 15, density=True)
    if title is None:
        title = "{0} :CPK={1},mean = {2},stdev = {3}\n".format(serialTitle,cpk,u,stdev)
    else:
        title = title + ("{0} :CPK={1},mean = {2},stdev = {3}\n".format(serialTitle,cpk,u,stdev))
    plt.title(title)
    legendlist.append(serialTitle)
    
    
    plt.vlines([x1[np.argmax(y1)]],ymin = 0,ymax = y1[np.argmax(y1)], color='gray', linestyle='--')  # 


# ## 导入待处理的数据

# In[11]:


xlsx_file = "cpkdata.xlsx"
"""
# read_excel方法读取文件
xlsx_feature = pd.read_excel(xlsx_file, usecols=[1, 2, 3, 4, 5, 6, 7]) 
#usecols属性读取excel文件中的指定列，列号从0起始；skiprows属性跳过excel文件中指定行，行号从0起始

"""
# read_excel方法返回类型为DataFrame，不需要再次转换
feature = pd.read_excel(xlsx_file, usecols=[1, 2, 3, 4, 5,6])
#print(type(feature))
print(feature)


# ## 处理数据并 生成cpk曲线

# In[12]:


usl = 969.5
lsl = 930.5
std = 950
sigma = 3

stdev = 0
cpk_calc(feature,usl,lsl,'ambient')
cpk_calc(feature,usl,lsl,'18c')
cpk_calc(feature,usl,lsl,'28c')
cpk_calc(feature,usl,lsl,'aftercycle')
cpk_calc(feature,usl,lsl,'storage20')
cpk_calc(feature,usl,lsl,'storage60')

plt.axvline(usl, color='r', linestyle='--', label='USL')
legendlist.append('USL')

plt.axvline(lsl, color='r', linestyle='--', label='LSL')
legendlist.append('LSL')

plt.axvline(std, color='r', linestyle='--', label='STD')   
legendlist.append('STD')

plt.axvline(std - sigma * stdev, color='blue', linestyle='--', label='-3 Sigma')
legendlist.append('-3 Sigma')

plt.axvline(std + sigma * stdev, color='blue', linestyle='--', label='3 Sigma')
legendlist.append('3 Sigma')

plt.legend(legendlist)
plt.xlabel('Readings')
plt.ylabel('Probability Density')
plt.savefig('cpk1.png',bbox_inches='tight')


plt.show()


# In[ ]:





# In[ ]: