公交车最优发车时间

问题描述

假设$[0,2]$时间内一共有两辆公交车发车，发车时间分别为$T_1,2$。而在这段时间内，乘客到达公交车站的人数服从Poisson(10t)，若要最小化乘客的总等待时间，求第一班公交车的发车时间$T_1$，即
$$\begin{array}{l}\min T=E\left({\sum }_{k=1}^{N(t_1)}(t_1-s_k)+{\sum }_{k=1}^{N(t_2)}(t_2-s_k)\right)\\ s.t.t_1+t_2=2\\ N(t_1)\sim Poisson(10\cdot t_1)\\ N(t_2)\sim Poisson(10\cdot t_2)\end{array}$$
直接利用过滤泊松模型，得到理论解：
$T=10\cdot ({\int }_0^{t_1}(t_1-s)ds+{\int }_0^{t_2}(t_2-s)ds)=5(t_1^2+t_2^2)\ge 10$

仿真验证

乘客到达时间仿真

import numpy as np
import matplotlib.pyplot as plt

def simulate_bus_station(arrival_rate, max_time):
    arrival_times = []
    last_arrival = 0

    # Continuously generate passenger arrival times until the last arrival is at 2 seconds
    while True:
        # Generate the next arrival time
        next_arrival = last_arrival + np.random.exponential(1 / arrival_rate)
        if next_arrival > max_time:
            # If the next arrival time is greater than 2 seconds, stop the simulation
            break
        arrival_times.append(next_arrival)
        last_arrival = next_arrival

    return arrival_times
    
np.random.seed(0)
# Set the arrival rate (lambda) for Poisson distribution
arrival_rate = 10  # For example, 10 passenger per second

# Simulate passenger arrivals
arrival_times = simulate_bus_station(arrival_rate,2)

# Plot the cumulative count of passengers vs. arrival time
plt.step(arrival_times, range(1, len(arrival_times) + 1), where='post')
plt.xlabel('Time (seconds)')
plt.ylabel('Cumulative number of passengers')
plt.title('Passenger Arrivals at a Bus Station')
plt.show()

平均等待总时间计算

def calculate_total_waiting_time(T1,T2,arrival_times):
    # 计算每个乘客的等车时间
    wait_times = []

    for arrival in arrival_times:
        if arrival <= T1:
            # 如果乘客在第一班车发车前到达，他们的等车时间是T1减去他们的到达时间
            wait_times.append(T1 - arrival)
        elif T1 < arrival <= T2:
            # 如果乘客在第一班车发车后，且第二班车发车前到达，他们的等车时间是T2减去他们的到达时间
            wait_times.append(T2 - arrival)

    # 计算总的等待时间
    total_wait_time = sum(wait_times)
    return total_wait_time
    
def calculate_average_total_waiting_time(arrival_rate, max_time, T1, sim_num):
    total_wait_times = np.zeros(shape=sim_num)
    for i in range(sim_num):
        arrival_times = simulate_bus_station(arrival_rate, max_time)
        total_wait_time = calculate_total_waiting_time(T1,max_time,arrival_times)
        total_wait_times[i] = total_wait_time
    average_total_wait_time = np.average(total_wait_times)
    return average_total_wait_time

# 设置仿真参数
lam = 10  # 平均每秒到达一个乘客
max_time = 2  # 模拟总时间为2秒
sim_num = 100 # 每种情况模拟100次
T1_values = np.linspace(0.1, 1.9, 50)  # 在区间(0.1, 1.9)内取50个不同的T1值
np.random.seed(0)

# 对不同的T1计算平均等待时间
average_wait_times = []
avg_time_in_theory = []
for T1 in T1_values:
    avg_wait = calculate_average_total_waiting_time(lam, max_time, T1, sim_num)
    average_wait_times.append(avg_wait)
    avg_time_in_theory.append(0.5*lam*(T1**2+(max_time-T1)**2))

# 使用Matplotlib绘制T1与平均等待时间的关系
plt.plot(T1_values, average_wait_times, label='Average Total Wait Time')
plt.plot(T1_values, avg_time_in_theory, label='Theoretically Average Total Wait Time')
plt.xlabel('Bus Departure Time T1 (s)')
plt.ylabel('Wait Time (s)')
plt.title('Average Total Wait Time vs. Bus Departure Time T1')
plt.legend()
plt.grid(True)
plt.show()