«上一篇
 文章快速检索 高级检索

 智能系统学报  2018, Vol. 13 Issue (1): 125-130  DOI: 10.11992/tis.201705030 0

### 引用本文

BAO Yi, LOU Fengdan, WANG Wanliang. Multiobjective optimization control of intelligent household electricity with demand management[J]. CAAI Transactions on Intelligent Systems, 2018, 13(1), 125-130. DOI: 10.11992/tis.201705030.

### 文章历史

1. 杭州天丽科技有限公司，浙江 杭州 310051;
2. 国网浙江省电力公司信息通信分公司，浙江 杭州 310073;
3. 浙江工业大学 计算机科学与技术学院，浙江 杭州 310023

Multiobjective optimization control of intelligent household electricity with demand management
BAO Yi1, LOU Fengdan2, WANG Wanliang3
1. Hangzhou TianLi Electronic Technology co., LTD, Hangzhou 310051, China;
2. State Network Zhejiang Electric Power Corporation Information Communications Branch, Hangzhou 3100073, China;
3. School of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China
Abstract: In this paper, we propose a home electricity control system with a smart grid for managing the demand of home appliances and to ease the peak-time grid pressure. We designed an intelligent controller that can obtain user power information and provide users with time-sharing electricity metering, while also being convenient for suppliers to apply the demand management system. To reduce the load power and demand-response delay time, we propose a multi-objective optimization technique. Its convergence rate is rapid and it can satisfy the immediate response requirement. The results for 500 families taking part in the experiment show that the proposed household electricity control system is reasonable, reduces user electricity costs, and reduces the response time delay due to its fast calculation speed, thereby effectively alleviating the peak time of the power grid.
Key words: demand management    smart grid    multi-objective decision    optimal control    dragonfly algorithm    household electricity    intelligent control    load classification

1 智能电网环境下家电系统运行方式

 Download: 图 1 智能家庭用电管理系统 Fig. 1 Diagram of home energy management system

2 多目标算法在需求侧管理中的应用

2.1 附加型负载系统模型

 ${Q_t} = (\sum\nolimits_i {{\beta _d}{P_k}} + \sum\nolimits_j {{\beta _e}{P_r}} ) \cdot {q_t}$ (1)

 ${W_t} = {\beta _d}{P_k} + {\beta _e}{P_r}$ (2)

 $Z = \Delta W \cdot T$ (3)
 $\Delta W = {W_t} - {W_x}$ (4)

2.2 多目标蜻蜓算法

1) 分散行为：

 ${S_i} = - \sum\limits_{j = 1}^n {X - {X_j}}$ (5)

2) 同步行为：

 ${A_i} = \frac{{\sum\nolimits_{j = 1}^n {{V_j}} }}{N}$ (6)

3) 汇聚行为：

 ${C_i} = \frac{{\sum\nolimits_{j = 1}^n {{X_j}} }}{N} - X$ (7)

4) 优化过程：

 ${F_i} = \overline X - X$ (8)

5) 避敌行为：

 ${E_i} = \widetilde X + X$ (9)

 $\Delta {X_{t + 1}} = (s{S_i} + a{A_i} + c{C_i} + f{F_i} + e{E_i}) + w\Delta {X_i}$ (10)

 ${{\mathit{\boldsymbol{X}}}_{t + 1}} = {{\mathit{\boldsymbol{X}}}_t} + \Delta {{\mathit{\boldsymbol{X}}}_{t + 1}}$ (11)

1) 算法读取当前系统耗电量，Si=Qt

2) 读取当前系统响应时间，Ei=Z

3) 输入待优化数目，初始设为m，设置空调和热水器权值相等，αk=αr

4) 根据式(10)(位置和加速度)对系统的耗电量和响应时间进行迭代计算，初始化迭代步数为30；

5) 进行响应管理时刻，优先对权值较小的负载进行处理；

6) 选取收敛标准，当系统整体耗电量降为历史数据的80%时，即认为算法运行有效，选取所受影响时间最短的值作为输出结果。

3 仿真实验结果分析