scheduling_algorithm_didi/didi_RL/train.py

import numpy as np
import pandas as pd
from entity import *
from didi_RL.RL_learning import RL

n_actions = 2
n_features = 3
train_step = 10

max_x = 30
max_y = 30
max_time = 144
gamma = 0.9

# max_x = 10
# max_y = 10
# max_time = 20
# gamma = 0.9

def train():
    # data = load('../train_data/train_data03.pkl')
    data = load('../train_data111.pkl')
    print("数据量：",sum(len(i) for i in data))
    step = 0
    learn_num = 0
    rl = RL(time_step=max_time)
    for i in range(len(data)-1,-1,-1):
        for match in data[i]:
            travel_time = match.travel_time
            rl.learn(s = (match.driver.x,match.driver.y),
                     s_ = (match.order.to_x,match.order.to_y),
                     t = match.order.order_time,
                     t_ = match.arrive_time,
                     r = (match.money / travel_time) * sum([gamma ** i for i in range(travel_time)]),
                     detal_t = travel_time
                     )
    print("ok")
    # rl.save_label('RL_q_label02.pkl')
    rl.save_label('RL_q_label03.pkl')

q_label_path = "RL_q_label.pkl"
def predict(s,s_,t,t_,r,detal_t):
    q_label = load(q_label_path)
    q_now = q_label[t-1][s][1]
    q_next = q_label[t_-1][s_][1]
    V = gamma ** detal_t * q_next - q_now + r
    return V


if __name__ == '__main__':
    # train()
    a = load('RL_q_label03.pkl')
    xx = []
    yy = []
    Z = []

    import matplotlib.pyplot as plt

    plt.xlabel('X')
    plt.ylabel('Y')
    plt.xlim(xmax=30, xmin=0)
    plt.ylim(ymax=30, ymin=0)

    colors1 = '#00CED1'  # 点的颜色
    colors2 = '#DC143C'
    area = np.pi * 8  # 点面积
    # 画散点图
    for k,v in a[108].items():
        # if 16<=k[0]<=24 and 16<=k[1]<=24:
        print(k,v)
        xx.append(k[0])
        yy.append(k[1])
        Z.append(v[1])
        colors = colors2 if v[1]>60 else colors1
        plt.scatter(k[0], k[1], s=area, c=colors, alpha=0.4)
    # plt.plot([0, 9.5], [9.5, 0], linewidth='0.5', color='#000000')
    plt.legend()
    # 三维图
    # from mpl_toolkits.mplot3d import Axes3D
    # fig = plt.figure()  # 定义新的三维坐标轴
    # ax3 = plt.axes(projection='3d')
    # ax3.plot_trisurf(xx, yy, Z,cmap='rainbow')

    # # 作图
    # ax3.plot_surface(X, Y, np.array(Z), cmap='rainbow')
    # ax3.contour(X,Y,Z, zdim='z',offset=-2，cmap='rainbow)   #等高线图，要设置offset，为Z的最小值
    plt.show()
    pass