我們利用多代理人的溝通以及合作技術，在奈米電網中設計一套能源管理系統。讓多代理人理解運作限制與目標，透過彼此溝通和合作找到適合運作的電壓。由於在奈米電網中，電力供應以及需求是隨機且間歇的，因此平衡供給以及需求是很困難的。能源管理系統必須維持一個穩定的電壓，在這個電壓之下，電力供給以及電力需求是平衡的。我們利用市場機制模組作為多代理人之前的溝通機制，用作平衡需求以及供給端的電力。我們也設計了一個對外溝通的通道，系統可以透過這個通道與外界電力公司作交易。此外我們設計兩套預測機制，讓系統預測在一段特定的時間內，電力供給以及電力需求的間歇性變化，讓能源管理系統能夠在問題發生之前採取應對措施。最後，我們利用高斯分布以及卜瓦松分布，產生十天的隨機用電模擬資料，並輸入資料進行若干實驗去驗證我們的系統運作效率。

We design and implement a distributed control energy management system (EMS) based on the multi-agent platform for nanogrids in which multi-agents can coordinate by negotiation to maintain power stability with minimal cost. Specifically, we model the decision making and negotiation of multi-agents by allowing them to know constraints and objectives, to receive messages of power signals and can collaborate together to find a compromised bus voltage via communication and negotiation with other agents in nanogrids. Since the demand power from loads and supply power from solar cell are both intermittent, it is hard to balance supply and demand power. We use market-oriented model to implement the negotiation protocol between every devices in EMS. The market-oriented model is to balance the current of demand side and supply side. The EMS in a nanogrid must maintain stable bus voltages so that power supply and power demand are balanced and the maximal power can be delivered from power supply to the consumption devices. In terms of currents, it is that the total supply currents must be equal to the total consumption currents, e.g. Ig + Is + Ib + Ip = Il1 + Il2 + … Iln. We also investigate a prediction system for EMS to predict the behavior of intermittent supply and demand of power under a specific window time period. Furthermore, we build a protocol for EMS to make transaction with an electric company in which the system would buy or sell power from and to the electronic company respectively. Finally, we measure the performance of gain/cost over the 10 days simulated data that are modeled the intermittent behaviors of power generation and consumption using Poisson and Gaussian probability models.

1. Boming Zhang, A Multi-Agent based distributed computing platform for new generation of EMS, Power Systems Conference and Exposition, 2009. PSCE '09. IEEE/PES.
2. G. Weiss, Multiagent Systems, A Modern Approach to DAI. Cambridge, MA: MIT Press, 1999, p. 100–104, 233-236.
3. Scott H. Clearwater, Market-based control: a paradigm for distributed resource allocation, World Scientific Publishing Co. 1996.
4. Vincenzo Loia, Alfredo Vaccaro, and K. Vaisakh, A Self-Organizing Architecture Based on Cooperative Fuzzy Agents for Smart Grid Voltage Control, IEEE Transactions on Industrial Informatics, Vol. 9, No. 3, August 2013.
5. Bruce Norman, Nanogrids-evolving our electricity systems from the bottom-up, Green Computing Conference (IGCC), 2013 International.
6. Fred C. Lee, Dushan Boroyevich, Paolo Mattavelli, Khai Ngo, Prospectus of CPES Mini‐Consortium on Renewable Energy and Nanogrids (REN), 2010, CPES report.
7. Kai-Wei Hu, and Chang-Ming Liaw, Incorporated Operation Control of DC Microgrid and Electric Vehicle, IEEE Transactions on Industrial Electronics, Vol. 63, No. 1, September 2015.
8. Igor Cvetkovic, Dushan Boroyevich, Fred C. Lee, Paolo Mattavelli, Dong Dong, Li Jiang, and Yue Chang, Future Home DC-based Renewable Energy nanoGrid System, Proceedings of the 2010 Conference on Grand Challenges in Modeling & Simulation 2010.
9. John Schönberger, Richard Duke, and Simon D. Round, DC-Bus Signaling: A Distributed Control Strategy for a Hybrid Renewable Nanogrid, IEEE Transaction on Industrial Electronics, Vol. 53, No. 5, October 2006.
10. M. Moreira, and E. Fiesler, Neural Networks with Adaptive Learning Rate and Momentum Terms, IDIAP 1995.
11. Preshing on programming Web Site, (http://preshing.com/20111007/how-to-generate-random-timings-for-a-poisson-process/)
12. Wikipedia Web Site, (https://en.wikipedia.org/wiki/Marsaglia_polar_method)
13. Taiwan Power Company Web Site, (http://www.taipower.com.tw/content/new_info/new_info_in.aspx?LinkID=11)