g(θ) = arg max a A v i (a, θ i ) (1) i p i (θ) = v j (g(θ i ), θ j ) v j (g(θ), θ j ) (2) u i (θ i ) = v i (a, θ i ) p i (θ) (3) 1 CDR model flow,, CD

Size: px

Start display at page:

Download "g(θ) = arg max a A v i (a, θ i ) (1) i p i (θ) = v j (g(θ i ), θ j ) v j (g(θ), θ j ) (2) u i (θ i ) = v i (a, θ i ) p i (θ) (3) 1 CDR model flow,, CD"

あかりやまのかみしゃ
5 years ago
Views:

1 1,a) 1,b),, (DR), HEMS,, DR,,,, VCG, (CDR: Cooperative Demand Response),,,,,,, 1.,, (DR) [1]., HEMS,, DR,, HEMS, [2]., HEMS DR [3], [4], [5]. DR, (RTP: Real Time Pricing)[6], [7], [8], [9] RTP,,, [10], RTP,, 1, Nagoya Institute of Technology, Gokiso-cho, Showa-ku, Nagoya , Japan a) b) DR,,,, (CDR: Cooperative Demand Response),, bid,,,,, [11]. [12], [13], [14]., Continuous Ranked Probability Score(CRPS) [15] VCG,,,,,, CDR c 2014 Information Processing Society of Japan 1

2 g(θ) = arg max a A v i (a, θ i ) (1) i p i (θ) = v j (g(θ i ), θ j ) v j (g(θ), θ j ) (2) u i (θ i ) = v i (a, θ i ) p i (θ) (3) 1 CDR model flow,, CDR 2, CDR 3,, 4 2. Cooperative Demand Response Model 2.1 N (CAs: Consumer Agents) (Genco: Generation Company), CAs CDRS(Cooperative demand Response System),, Genco CAs 1 [1] CAs, AI 2. Genco, CAs, CAs,, CAs CAs, CAs ( ) θ i Θ i A = {a, b,...} θ i CA i, a A v i (a, θ i ) R CA i v i i θ i CDR VCG, g : Θ A, p : Θ R (1), (2) (3) CDR Step1:, Genco, T h, t h, price h, t l, price l,, [16]., Genco price h > price l, CAs t h t l price t = { priceh, if T otaldemand t T h price l, if T otaldemand t < T h (4) Step2: CAs, sd p i,t CAs, t h i t (5), DV i i sd p i,t = j DV i ShiftDevicesDemand p j,t (5) CAs, Genco,, (SR: Scoring Rule), SR,, [17], [18]., SR, Brier score SR,, Continuous Ranked Probability Score(CRPS) SR [15]. CA i c 2014 Information Processing Society of Japan 2

3 Step4: Accept CDRS, CAs bid Accept., CDRS CAs bid sd p i,t (1 ), bid, Accept, CDRS 2, bid Accept [Accept ] T otaldemand th sd b,t T h (8) b Bids t T otaldemand tl + sd b,t < T h (9) b Bids t Fig. 2 2 CRPS CRPS scoring mechanism for different errors e i,t (6) e i,t = sda i,t sd p i,t sd p i,t (6), d a i,t CA i, CRPS (7) CRP S(N(µ = 0, σ 2 i,t ), e i,t) = [ 1 π 2φ ( ei,t ) e i,t ( 2Φ ( ei,t ) )] 1 (7), φ Φ, CRPS 2 CRPS, ( = 0), (e i,t = 0),, CAs CRP S(N(µ = 0, σ 2 i,t ), e i,t) CRP S i,t (θ i ) Step3: bid, CAs, CDRS bid, t h t l, bid, t h, t l, sd c i,t, 4 1 bid, CAs t h t l Symbol t h t l sd p i,t 1 Table 1 A bid composition bid Description beginning of shift demand end of shift demand predicted shift demand confidence of prediction (1), t h, (2), t l, Step5: CDRS, CAs, Genco CDRS, sd p c,t = i N sdp i,t, sda c,t = i N sda i,t, CDRS, e c,t = sda c,t sdp c,t, σ 2 sd p c,t = i N (sdp i,t ) 2 c,t ( i N sdp i,t) 2., CA CDRS, CA CDRS Step6:,,,, [19].,,,,,,., c h c l, c h c l costv alue t, (10) profit G t = costv alue t sd a c,t (10), [19], [20] Step7: Genco, CDRS, t l p h p l value t, Genco CDRS (11) V c,t (a, θ) = CRP S c,t (θ) value t sd a c,t λ profitg t (11),,, c 2014 Information Processing Society of Japan 3

4 ,, CAs,, Step8: CAs,,, CDRS CA i, (12) CRP S i,t (θ i ) sd a i,t v i,t (a, θ i ) = j N CRP S v j,t(θ j ) sd a c,t (a, θ)(12) j,t, CAs, 2.3, VCG CRPS,, 1. CDR CDR, CA i ˆθ i, CA i u i ( ˆθ i ) = v i (g(ˆθ), θ i ) p i (ˆθ) = v i (g(ˆθ), θ i ) + v j (g(ˆθ), ˆθ j ) v j (g( θ ˆ i ), ˆθ j ) v j(g(θ i ), θ j ), CA i, CA i (13) max ˆθ i Θ i v i (g( ˆθ i, θ i ˆ ), θ i ) + v j (g( ˆθ i, θ i ˆ ), ˆθ j ) (13) CA i ˆθ i g( ˆθ i, ˆ θ i )., CA i, ˆθ i, v i, (11) (12) (11), CRPS strictly proper scoring rule, (11) CRPS, (11) (12) strictly proper scoring rule, (1) i v i(a, θ i ) a (1),, g( ˆθ i, ˆ θ i ) = a max v i(a, θ i ) + v j (a, ˆθ j ) a A = max a A CRP S i,t (θ i ) sd a i,t j N CRP S j,t(θ j ) sd a j,t v c,t + v j (a, ˆθ j ) strictly proper scoring rule, CA i θ i 2. CDR CDR, CAs, CA i (14) u i (θ i, θ i ) = v i (g(θ), θ i ) v j (g(θ i ), θ j ) v j (g(θ), θ j ) = i v i (g(θ), θ i ) v j (g(θ i ), θ j ) (14) Eq.(14) VCG, CA i, CDR,, CRPS,,, [21], (CDRS), , 62 CAs, [22] , [23],, CAs CDRS CDRS 3 CDRS, (Step3, Step4) Genco, CAs Genco, (11) 3.2,,, [ ] ,, [22], c 2014 Information Processing Society of Japan 4

5 3 Singleton model flow peak demand bid 20 λ 0.1 p h [ /kwh] 20 p l [ /kwh] 15 c h [ /kwh] 25.1 c l [ /kwh] (24 ) Fig. 5 Energy usage before and after using of CDR algorithm: 24 hours, [CA CDRS vs ] 6, CA ( CA ) 6, CDRS, VCG, CDRS,, CAs CDRS, CDRS [CA ] 2, CA, CA CA, CDRS, CDRS, CA, CA CDRS Fig. 4 4 (31 ) Energy usage before (black dotted line) and after (red solid line) using of CDR algorithm: 31 days 3 CA CA Table 3 Scalability and number of CAs number of CA CDRS AVE [ ] Singleton AVE[ ] , t 10, t 11, t 12, t 14, t 15, t 16, t 17, t 20 t h, t l, kwh, kwh,, kwh 5,,, DR,, VCG CDR CDR c 2014 Information Processing Society of Japan 5

6 Fig. 6 6 CA : CDRS vs CA s average utility with CDRS vs Singleton,,, CDRS,,,,,,,, [1] Albadi, M. and El-Saadany, E.: A summary of demand response in electricity markets, Electric Power Systems Research, Vol. 78, pp (2008). [2] Doostizadeh, M. and Ghasemi, H.: A day-ahead electricity pricing model based on smart metering and demand-side management, Energy, Vol. 46, pp (2012). [3] Chao, H.: Peak load pricing and capacity planning with demand and supply uncertainty, The Bell Journal of Economics, Vol. 14, pp (1983). [4] Du, P. and Lu, N.: Appliance commitment for household load scheduling, IEEE Trans. Smart Grid, Vol. 2, No. 2, pp (2011). [5] Roozbehani, M., Dahleh, M. and Mitter, S.: Dynamic pricing and stabilization of supply and demand in modern electric power grids, In IEEE International Conference on Smart Grid Communications, pp (2010). [6] Samadi, P., Mohsenian-Rad, A., Schober, R., Wong, V. W. S. and Jatskevich, J.: Optimal real-time pricing algorithm based on utility maximization for smart grid, In IEEE Smart Grid Communications, pp (2010). [7] Li, N., Chen, L. and Low, S.: Optimal demand response based on utility maximization in power networks, IEEE Power and Energy Society General Meeting (2011). [8] Chen, L., N. Li, S. L. and Doyle, J.: Two market models for demand response in power networks, In IEEE International Conference on Smart Grid Communications, pp (2010). [9] Caron, S. and Kesidis, G.: Incentive-based energy consumption scheduling algorithms for the smart grid, In IEEE International Conference on Smart Grid Communications, pp (2010). [10] A.-H., M.-R. and A, L.-G.: Optimal residential load control with price prediction in real-time electricity pricing environments, In IEEE Transactions on Smart Grid, Vol. 1, pp (2010). [11] Chakraborty, S., Ito., T. and Senju, T.: Smart Pricing Scheme: A Multi-layered Scoring Rule Application, Expert Systems with Applications, Vol. 41, pp (2014). [12] Ibars, C., Navarro, M. and Giupponi, L.: Distributed demand management in smart grid with a congestion game, In IEEE International Conference on Smart Grid Communications, pp (2010). [13] Vytelingum, P., Ramchurn, S. D., Voice, T. D., Rogers, A. and Jennings, N. R.: Trading agents for the smart electricity grid, In Proceedings of the 9th International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS-2010), pp (2010). [14] Robu, V., Kota, R., Chalkiadakis, G., Rogers, A. and Jennings., N. R.: Cooperative Virtual Power Plant Formation using Scoring Rules, In Proc. of the 26th Conference on Artificial Intelligence (AAAI-12), pp (2012). [15] Gneiting, T. and Raftery, A.: Strictly proper scoring rules, prediction and estimation, Journal of the American Statistical Association, Vol. 102, pp (2007). [16] Borenstein, S.: The long-run efficiency of real-time electricity pricing, The Energy Journal, Vol. 26, No. 3, pp (2005). [17] Boutilier, C.: Eliciting forecasts from self-interested experts: scoring rules for decision makers, In Proceedings of the 11th International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS-2012) (2012). [18] Rose, H., Rogers, A. and Gerding, E. H.: A scoring rule-based mechanism for aggregate demand prediction in the smart grid, In Proceedings of the 11th International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS-2012) (2012). [19] Company, C. E. P.: Relationship between power resources and usage of electricity in a day (2013). [20] Unit, N. P.: Verification committee s reports in terms of cost (2011). [21] Surowiecki, J.: The Wisdom of Crowds: Why the Many are Smarter Than the Few and how Collective Wisdom Shapes Business, Economies, Societies, and Nations, Doubleday (2004). [22] Company, C. E. P.: electricity demand performance (2013) (2013). [23] the Agency of Natural Resources and Energy: Demand structure estimation of a summer maximum dissipation use day (2011). c 2014 Information Processing Society of Japan 6

IPSJ SIG Technical Report Vol.2015-ICS-179 No /3/20 Cooperative Demand Response 1,a) 1,b),, (DR)., HEMS,, DR.,,,., VCG, (CDR: Cooperative Deman

IPSJ SIG Technical Report Vol.2015-ICS-179 No /3/20 Cooperative Demand Response 1,a) 1,b),, (DR)., HEMS,, DR.,,,., VCG, (CDR: Cooperative Deman Cooperative Demand Response 1,a) 1,b),, (DR), HEMS,, DR.,,,., VCG, (CDR: Cooperative Demand Response),,,,.,,, 1.,, (DR) [1]., HEMS,, DR,, HEMS, [2]., HEMS. DR [3], [4], [5]. DR, (RTP: Real Time Pricing)[6],