building-agents/plott.py

import re
import pandas as pd
import matplotlib.pyplot as plt
from scipy.ndimage import gaussian_filter1d
from matplotlib.patches import ConnectionPatch


def extract(text):
    matches = re.findall(pattern, text)
    rewards = []
    unbalances = []

    for match in matches:
        reward, unbalance = match
        rewards.append(float(reward))
        unbalances.append(float(unbalance) / 1000)

    rewards_s = pd.Series(rewards).rolling(window=3).mean()
    unbalances_s = pd.Series(unbalances).rolling(window=3).mean()
    return rewards, rewards_s, unbalances, unbalances_s


pattern = r"reward: ([\-\d\.]+),\s*unbalance:\s*([\d\.]+)"

with open('res/ddpg.txt', 'r', encoding='utf-16') as f1:
    ddpg = f1.read()
with open('res/ppo_1016.txt', 'r', encoding='utf-8') as f2:
    ppo = f2.read()
with open('res/ppo_llm.txt', 'r', encoding='utf-8') as f3:
    ppo_llm = f3.read()
with open('res/sac.txt', 'r', encoding='utf-8') as f4:
    sac = f4.read()
with open('res/td3.txt', 'r', encoding='utf-8') as f5:
    td3 = f5.read()

ddpg_rewards, ddpg_rewards_s, ddpg_unbalances, ddpg_unbalances_s = extract(ddpg)
td3_rewards, td3_rewards_s, td3_unbalances, td3_unbalances_s = extract(td3)
sac_rewards, sac_rewards_s, sac_unbalances, sac_unbalances_s = extract(sac)
ppo_rewards, ppo_rewards_s, ppo_unbalances, ppo_unbalances_s = extract(ppo)
lmppo_rewards, lmppo_rewards_s, lmppo_unbalances, lmppo_unbalances_s = extract(ppo_llm)

fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(8, 10))

ax1.plot(ddpg_rewards, color='#E03C31', alpha=0.2)
ax1.plot(td3_rewards, color='#1D8C55', alpha=0.2)
ax1.plot(sac_rewards, color='#FFA500', alpha=0.2)
ax1.plot(ppo_rewards, color='#3498DB', alpha=0.2)
ax1.plot(lmppo_rewards, color='#A52A2A', alpha=0.2)
ax1.plot(ddpg_rewards_s, label='DDPG', color='#E03C31', alpha=0.8)
ax1.plot(td3_rewards_s, label='TD3', color='#1D8C55', alpha=0.8)
ax1.plot(sac_rewards_s, label='SAC', color='#FFA500', alpha=0.8)
ax1.plot(ppo_rewards_s, label='PPO', color='#3498DB', alpha=0.8)
ax1.plot(lmppo_rewards_s, label='LMPPO', color='#A52A2A', alpha=0.8)
ax1.set_xlabel('Time [h]\n(a)', fontsize=14, labelpad=0)
ax1.set_ylabel('Reward[-]', fontsize=14)

# 子图
inset_ax1 = fig.add_axes([0.45, 0.6, 0.25, 0.15])  # [x, y, 宽, 高]
inset_ax1.plot(ddpg_rewards, color='#E03C31', alpha=0.8)
inset_ax1.plot(td3_rewards, color='#1D8C55', alpha=0.8)
inset_ax1.plot(sac_rewards, color='#FFA500', alpha=0.8)
inset_ax1.plot(ppo_rewards, color='#3498DB', alpha=0.8)
inset_ax1.plot(lmppo_rewards, color='#A52A2A', alpha=0.8)
inset_ax1.set_xlim(900, 1000)
inset_ax1.set_ylim(-150, -20)
inset_ax1.grid(True, which='both', axis='both', linestyle='--', linewidth=0.5, color='gray')
inset_ax1.tick_params(axis='both', which='major', labelsize=10)

ax1.grid(True, which='both', axis='both', linestyle='--', linewidth=0.5, color='gray')
ax1.legend(fontsize=12, handletextpad=0.5, labelspacing=0.3)

# 使用ConnectionPatch连接主图和子图
xy_main = (900, 0)
xy_inset = (900, -20)
xy_main2 = (1000, 0)
xy_inset2 = (1000, -20)

con = ConnectionPatch(xyA=xy_inset, xyB=xy_main, coordsA="data", coordsB="data",
                      axesA=inset_ax1, axesB=ax1, linestyle='--', linewidth=1, color='gray')
con2 = ConnectionPatch(xyA=xy_inset2, xyB=xy_main2, coordsA="data", coordsB="data",
                       axesA=inset_ax1, axesB=ax1, linestyle='--', linewidth=1, color='gray')
fig.add_artist(con)
fig.add_artist(con2)

ax2.plot(ddpg_unbalances, color='#E03C31', alpha=0.2)
ax2.plot(td3_unbalances, color='#1D8C55', alpha=0.2)
ax2.plot(sac_unbalances, color='#FFA500', alpha=0.2)
ax2.plot(ppo_unbalances, color='#3498DB', alpha=0.2)
ax2.plot(lmppo_unbalances, color='#A52A2A', alpha=0.2)
ax2.plot(ddpg_unbalances_s, label='DDPG', color='#E03C31', alpha=0.8)
ax2.plot(td3_unbalances_s, label='TD3', color='#1D8C55', alpha=0.8)
ax2.plot(sac_unbalances_s, label='SAC', color='#FFA500', alpha=0.8)
ax2.plot(ppo_unbalances_s, label='PPO', color='#3498DB', alpha=0.8)
ax2.plot(lmppo_unbalances_s, label='LMPPO', color='#A52A2A', alpha=0.8)
ax2.set_xlabel('Time [h]\n(b)', fontsize=14, labelpad=0)
ax2.set_ylabel('Power imbalance[KW]', fontsize=14)

# 子图
inset_ax2 = fig.add_axes([0.45, 0.25, 0.25, 0.15])
inset_ax2.plot(ddpg_unbalances, color='#E03C31', alpha=0.8)
inset_ax2.plot(td3_unbalances, color='#1D8C55', alpha=0.8)
inset_ax2.plot(sac_unbalances, color='#FFA500', alpha=0.8)
inset_ax2.plot(ppo_unbalances, color='#3498DB', alpha=0.8)
inset_ax2.plot(lmppo_unbalances, color='#A52A2A', alpha=0.8)
inset_ax2.set_xlim(900, 1000)
inset_ax2.set_ylim(0, 2)
inset_ax2.grid(True, which='both', axis='both', linestyle='--', linewidth=0.5, color='gray')
inset_ax2.tick_params(axis='both', which='major', labelsize=10)

ax2.grid(True, which='both', axis='both', linestyle='--', linewidth=0.5, color='gray')
ax2.legend(fontsize=12, handletextpad=0.5, labelspacing=0.3)

xy_main3 = (900, 0)
xy_inset3 = (900, 0)
xy_main4 = (1000, 0)
xy_inset4 = (1000, 0)

con3 = ConnectionPatch(xyA=xy_inset3, xyB=xy_main3, coordsA="data", coordsB="data",
                       axesA=inset_ax2, axesB=ax2, linestyle='--', linewidth=1, color='gray')
con4 = ConnectionPatch(xyA=xy_inset4, xyB=xy_main4, coordsA="data", coordsB="data",
                       axesA=inset_ax2, axesB=ax2, linestyle='--', linewidth=1, color='gray')
fig.add_artist(con3)
fig.add_artist(con4)

plt.show()
fig.savefig("compare.png", format='png', dpi=600, bbox_inches='tight')