191 lines
6.2 KiB
Python
191 lines
6.2 KiB
Python
import numpy as np
|
|
import pandas as pd
|
|
import json
|
|
from scipy.optimize import minimize
|
|
from pyswarm import pso
|
|
|
|
# 读取数据
|
|
file_path = './data.csv'
|
|
data = pd.read_csv(file_path)
|
|
|
|
# 初始化状态
|
|
initial_soc = 0.4 # 电池初始荷电状态
|
|
initial_sg1 = 0
|
|
initial_sg2 = 0
|
|
initial_sg3 = 0
|
|
battery_capacity = 500 * initial_soc
|
|
|
|
# 参数设置
|
|
battery_efficiency = 0.9
|
|
battery_degradation = 0.01
|
|
battery_holding = 0.1
|
|
solar_cofficient = 0.01
|
|
volatage_change_percent = 0.1
|
|
wind_cofficient = 0.01
|
|
|
|
# 发电机参数
|
|
DG_parameters = {
|
|
'gen_1': {'a': 0.0034, 'b': 3, 'c': 30, 'power_output_max': 150, 'power_output_min': 10, 'ramping_up': 100,
|
|
'ramping_down': 100},
|
|
'gen_2': {'a': 0.001, 'b': 10, 'c': 40, 'power_output_max': 375, 'power_output_min': 50, 'ramping_up': 100,
|
|
'ramping_down': 100},
|
|
'gen_3': {'a': 0.001, 'b': 15, 'c': 70, 'power_output_max': 500, 'power_output_min': 100, 'ramping_up': 200,
|
|
'ramping_down': 200}
|
|
}
|
|
|
|
NUM_GEN = len(DG_parameters.keys())
|
|
period = 100 # 处理前 100 个数据
|
|
|
|
|
|
def get_dg_info(parameters):
|
|
p_max = []
|
|
p_min = []
|
|
ramping_up = []
|
|
ramping_down = []
|
|
a_para = []
|
|
b_para = []
|
|
c_para = []
|
|
for name, gen_info in parameters.items():
|
|
p_max.append(gen_info['power_output_max'])
|
|
p_min.append(gen_info['power_output_min'])
|
|
ramping_up.append(gen_info['ramping_up'])
|
|
ramping_down.append(gen_info['ramping_down'])
|
|
a_para.append(gen_info['a'])
|
|
b_para.append(gen_info['b'])
|
|
c_para.append(gen_info['c'])
|
|
return p_max, p_min, ramping_up, ramping_down, a_para, b_para, c_para
|
|
|
|
|
|
p_max, p_min, ramping_up, ramping_down, a_para, b_para, c_para = get_dg_info(parameters=DG_parameters)
|
|
|
|
|
|
# 目标函数
|
|
def objective(x, period_batch, start, initial_states):
|
|
total_cost = 0
|
|
soc, sg1, sg2, sg3 = initial_states
|
|
for t in range(period_batch):
|
|
Ac, Ag1, Ag2, Ag3, Av = x[t * 5:(t + 1) * 5]
|
|
Pc = max(0.2, min(0.8, soc + 0.2 * Ac * 0.9))
|
|
Pg1 = max(0, min(150, 100 * Ag1 + sg1))
|
|
Pg2 = max(0, min(375, 100 * Ag2 + sg2))
|
|
Pg3 = max(0, min(500, 200 * Ag3 + sg3))
|
|
Pso = max(0, (0.2 * data['irradiance'][start + t] + 0.05 * data['temperature'][start + t] - 9.25) * (1 + Av))
|
|
|
|
if 3 <= data['wind_speed'][start + t] < 8:
|
|
Pw = data['wind_speed'][start + t] ** 3 * 172.2625 / 1000
|
|
elif 8 <= data['wind_speed'][start + t] < 12:
|
|
Pw = 64 * 172.2625 / 125
|
|
else:
|
|
Pw = 0
|
|
|
|
P = Pc + Pg1 + Pg2 + Pg3 + Pso + Pw
|
|
|
|
if P >= data['load'][start + t]:
|
|
Ee = P - data['load'][start + t]
|
|
Es = 0
|
|
else:
|
|
Es = data['load'][start + t] - P
|
|
Ee = 0
|
|
|
|
Cb = 0.01 * battery_capacity + 0.1 * Pc
|
|
Cg1 = 0.0034 * Pg1 ** 2 + 3 * Pg1 + 30
|
|
Cg2 = 0.001 * Pg2 ** 2 + 10 * Pg2 + 40
|
|
Cg3 = 0.001 * Pg3 ** 2 + 15 * Pg3 + 70
|
|
Cs = 0.01 * Pso
|
|
Cw = 0.01 * Pw
|
|
Rs = 0.5 * data['price'][start + t] * Ee
|
|
Cp = data['price'][start + t] * Es
|
|
Pe = (Ee - 100) * 50 if Ee > 100 else 0
|
|
Ps = (Es - 100) * 50 if Es > 100 else 0
|
|
|
|
total_cost += (Cb + Cg1 + Cg2 + Cg3 + Cs + Cw + Pe + Ps - Rs + Cp) / 1000
|
|
|
|
soc = Pc
|
|
sg1 = Pg1
|
|
sg2 = Pg2
|
|
sg3 = Pg3
|
|
|
|
return total_cost
|
|
|
|
|
|
# 约束条件
|
|
def constraint(x, period_batch, start, initial_states):
|
|
constraints = []
|
|
soc, sg1, sg2, sg3 = initial_states
|
|
for t in range(period_batch):
|
|
Ac, Ag1, Ag2, Ag3, Av = x[t * 5:(t + 1) * 5]
|
|
Pc = max(0.2, min(0.8, soc + 0.2 * Ac * 0.9))
|
|
Pg1 = max(10, min(150, 100 * Ag1 + sg1))
|
|
Pg2 = max(50, min(375, 100 * Ag2 + sg2))
|
|
Pg3 = max(100, min(500, 200 * Ag3 + sg3))
|
|
Pso = max(0, (0.2 * data['irradiance'][start + t] + 0.05 * data['temperature'][start + t] - 9.25) * (1 + Av))
|
|
|
|
if 3 <= data['wind_speed'][start + t] < 8:
|
|
Pw = data['wind_speed'][start + t] ** 3 * 172.2625 / 1000
|
|
elif 8 <= data['wind_speed'][start + t] < 12:
|
|
Pw = 64 * 172.2625 / 125
|
|
else:
|
|
Pw = 0
|
|
|
|
P = Pc + Pg1 + Pg2 + Pg3 + Pso + Pw
|
|
|
|
# 保证变化后的值不小于最小值
|
|
constraints.append(Pc - 0.2)
|
|
constraints.append(Pg1 - 10)
|
|
constraints.append(Pg2 - 50)
|
|
constraints.append(Pg3 - 100)
|
|
|
|
constraints.append(P - data['load'][start + t])
|
|
soc = Pc
|
|
sg1 = Pg1
|
|
sg2 = Pg2
|
|
sg3 = Pg3
|
|
return constraints
|
|
|
|
|
|
actions = []
|
|
batch_size = 10 # 每次处理 10 个数据
|
|
total_batches = 10 # 总共处理 100 个数据
|
|
start = 0
|
|
|
|
for batch_num in range(total_batches):
|
|
end = min(start + batch_size, period)
|
|
period_batch = end - start
|
|
|
|
# 初始化初始状态
|
|
initial_states = (initial_soc, initial_sg1, initial_sg2, initial_sg3)
|
|
|
|
# # 使用随机初始化更接近实际情况
|
|
# x0 = np.random.uniform(0, 1, period_batch * 5)
|
|
# bounds = [(-1, 1)] * (period_batch * 5)
|
|
# cons = {'type': 'ineq', 'fun': lambda x: constraint(x, period_batch, start, initial_states)}
|
|
# result = minimize(lambda x: objective(x, period_batch, start, initial_states), x0, method='SLSQP', bounds=bounds,
|
|
# constraints=cons)
|
|
#
|
|
# actions.extend(result.x.reshape((period_batch, 5)).tolist())
|
|
|
|
# 使用粒子群优化算法
|
|
lb = [-1] * (period_batch * 5)
|
|
ub = [1] * (period_batch * 5)
|
|
xopt, fopt = pso(lambda x: objective(x, period_batch, start, initial_states), lb, ub,
|
|
ieqcons=[lambda x: constraint(x, period_batch, start, initial_states)], maxiter=100, swarmsize=50)
|
|
|
|
actions.extend(xopt.reshape((period_batch, 5)).tolist())
|
|
|
|
# 更新初始状态为当前批次最后一个状态
|
|
last_action = actions[-1]
|
|
initial_soc = max(0.2, min(0.8, initial_soc + 0.2 * last_action[0] * 0.9))
|
|
initial_sg1 = max(10, min(150, 100 * last_action[1] + initial_sg1))
|
|
initial_sg2 = max(50, min(375, 100 * last_action[2] + initial_sg2))
|
|
initial_sg3 = max(100, min(500, 200 * last_action[3] + initial_sg3))
|
|
|
|
start += batch_size
|
|
|
|
actions_json_path = './actions_batch_p.json'
|
|
try:
|
|
with open(actions_json_path, 'w') as f:
|
|
json.dump(actions, f)
|
|
print(f"文件保存成功：{actions_json_path}")
|
|
except Exception as e:
|
|
print(f"文件保存失败：{e}")
|