目前在网络上没有找到现成的办法解决双坐标分组柱状图。为了画这个图，花了3小时。
实现效果：
源代码如下：
import matplotlib.pyplot as plt
a = [4,5,0]
b = [3,2,0]
c = [4,3,0]
d = [3,2,0]
a_2 = [0,0,0.13]
b_2 = [0,0,0.18]
c_2 = [0,0,0.18]
d_2 = [0,0,0.18]
labels = ['A', 'B', 'C']
plt.rcParams['axes.labelsize'] = 16 # xy轴label的size
plt.rcParams['xtick.labelsize'] = 12 # x轴ticks的size
plt.rcParams['ytick.labelsize'] = 14 # y轴ticks的size
# 设置柱形的间隔
width = 0.2 # 柱形的宽度
x1_list = []
x2_list = []
x3_list= []
x4_list= []
for i in range(len(a)):
x1_list.append(i)
x2_list.append(i+width)
x3_list.append(i+2*width)
x4_list.append(i+3*width)
# 创建图层
fig, ax1 = plt.subplots()
ax1.set_ylabel('X')
ax1.set_ylim(0, 6)
ax1.bar(x1_list,a, width=width, label="a",color='mediumspringgreen', align='edge')
plt.bar(x2_list,b, width=width, label="b",color='cornflowerblue', align='edge', tick_label=labels)
plt.bar(x3_list,c, width=width, label="c",color='aqua', align='edge', tick_label=labels)
plt.bar(x4_list,d, width=width, label="d",color='coral', align='edge', tick_label=labels)
ax6 = ax1.twinx()
ax6.set_ylim(0, 0.3)
ax6.set_ylabel('Y')
b1 = plt.bar(x1_list,a_2, width=width, label="a",color='mediumspringgreen', align='edge')
b2 = plt.bar(x2_list,b_2, width=width, label="b",color='cornflowerblue', align='edge', tick_label=labels)
b3 = plt.bar(x3_list,c_2, width=width, label="c",color='aqua', align='edge', tick_label=labels)
b4 = plt.bar(x4_list,d_2, width=width, label="d",color='coral', align='edge', tick_label=labels)
plt.legend(handles=[b1,b2,b3,b4],loc="upper center",ncol=4)
plt.show()
