CN113592553B - Cloud energy storage double-layer optimization control method for competitive condition generation type countermeasure network - Google Patents
Cloud energy storage double-layer optimization control method for competitive condition generation type countermeasure network Download PDFInfo
- Publication number
- CN113592553B CN113592553B CN202110881691.0A CN202110881691A CN113592553B CN 113592553 B CN113592553 B CN 113592553B CN 202110881691 A CN202110881691 A CN 202110881691A CN 113592553 B CN113592553 B CN 113592553B
- Authority
- CN
- China
- Prior art keywords
- energy storage
- countermeasure network
- generation type
- representing
- type countermeasure
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004146 energy storage Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000005457 optimization Methods 0.000 title claims abstract description 29
- 230000002860 competitive effect Effects 0.000 title claims abstract description 11
- 238000005096 rolling process Methods 0.000 claims abstract description 13
- 230000035772 mutation Effects 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 17
- 238000009826 distribution Methods 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 7
- 230000005611 electricity Effects 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 230000007774 longterm Effects 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 238000005265 energy consumption Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003042 antagnostic effect Effects 0.000 description 1
- 230000008485 antagonism Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Finance (AREA)
- General Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Public Health (AREA)
- Data Mining & Analysis (AREA)
- Game Theory and Decision Science (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a cloud energy storage double-layer optimization control method of a competitive condition generation type countermeasure network. According to the method, a competition learning method is integrated into a condition generation type countermeasure network method, and the method is used for double-layer optimization control of cloud energy storage. Firstly, expanding a competition learning method into a condition generation type countermeasure network training for predicting load power of a future day; secondly, based on the predicted load, establishing a double-layer model of an optimal cloud energy storage decision; and finally, adopting rolling optimization to solve the cloud energy storage optimal control decision. The method can better solve the problems of unstable network and mode collapse of the existing condition generation type countermeasure network, realize accurate prediction of power load, minimize the total cost of cloud energy storage of consumers, and reasonably distribute and charge and discharge energy stored by cloud energy storage providers.
Description
Technical Field
The invention belongs to the field of optimal control and operation of power systems, and relates to an optimal control method based on an artificial intelligence technology, which is suitable for double-layer optimal control of cloud energy storage of a power system.
Background
The reduction in the price of stored energy makes possible the large-scale use of stored energy. However, distributed energy storage is still expensive for the consumer. In order to meet consumer demand, a cloud energy storage based on shared energy storage technology is proposed.
The generative antagonism network in artificial intelligence technology is one of the main methods for learning generative models in complex data. The generated countermeasure network not only can learn the distribution of the real samples, but also has strong prediction capability. The generation type countermeasure network is widely applied to the fields of image generation, text generation and voice generation. The generated countermeasure network is sometimes difficult to train. If the data distribution and the generated distribution do not substantially overlap, the generator gradient may point in more or less random directions, even causing problems with gradient disappearance and model collapse.
Therefore, the invention provides a cloud energy storage double-layer optimization control method for a competitive condition generation type countermeasure network. The proposed method extends the competition learning method to be used in countermeasure training for predicting the daily electrical load. The competition generation type countermeasure network takes the historical power load as a condition, so that different training target training generators are kept all the time, the limitation of the training targets is overcome, the training problems of network instability, mode collapse and the like are reduced, and the accurate prediction of the power load is realized. The bi-layer model of optimal energy storage decisions minimizes the overall cost of consumer energy storage and pricing and capacity decisions for energy storage built by cloud energy storage providers.
Disclosure of Invention
The invention provides a cloud energy storage double-layer optimization control method of a competitive condition generation type countermeasure network, which integrates a competitive learning method into a condition generation type countermeasure network method and is used for double-layer optimization control of cloud energy storage; the cloud energy storage double-layer optimization control method of the competitive condition generation type countermeasure network comprises the following steps:
step (1): expanding the competition learning method to a condition generation type countermeasure network training for predicting load power of a future day, wherein a discriminator serves as a competition environment, and a generator generates competition mutation in a given discriminator; training generators using the maximum and minimum mutations, heuristic mutations and least squares mutations during competition, thereby generating a competition population comprising a plurality of different generators, and finally selecting only the generator with the maximum quality fitness for further training; the generator gradually generates a prediction sample with higher quality fitness under the condition of historical power load, and the optimal generator can generate the prediction sample with higher precision after further training; the condition generation type countermeasure network fused with the competition method can overcome the problems of unstable condition generation type countermeasure network and mode collapse, improves the prediction precision of the condition generation type countermeasure network, and comprises an objective function, a maximum and minimum mutation objective function, a heuristic mutation objective function, a least square mutation objective function and a quality fitness function of the competition generation type countermeasure network as follows:
wherein x is the input sample; y is a predicted value; z is the noise vector; c is the condition; g represents a generator; d represents a discriminator; p (P) data The true probability distribution of the sample; p (P) s Probability distribution for a historical power load sequence; p (P) z Is the probability distribution of the noise vector; g (z) represents mapping the input noise z into data; d (G (z)) represents the probability that the arbiter can recognize that the generator generates data by noise z; d (x|c) represents the probability that x comes from the true data distribution under condition c; d (y|c) represents the probability that y comes from the noise data distribution under condition c; 1g represents the base 10 logarithm;representing the expectation under a real sample;representing the expectations under the prediction samples; />Representing expectations under noise vectors;/>Representing a generator and a arbiter maximum and minimum game; />Representing generator maximum and minimum mutations; />Representing generator heuristic mutations; />Representing generator least squares mutation; />Representing a quality fitness function;
step (2): establishing a double-layer model of an optimal energy storage decision based on competition generation type countermeasure network prediction; the upper model maximizes the rental price of the cloud energy storage provider, and the lower model minimizes the total cost of consumer energy storage; aiming at the long-term planning problem, the proposed double-layer model explores the optimal scale capacity and lease price from the perspective of a cloud energy storage operator so as to maximize the net profit; exploring optimal lease and business decisions from the consumer's perspective to minimize their storage energy costs; the proposed objective function of the bilayer model is:
wherein NP CES Net profit for cloud energy storage providers; i Rent Lease revenue for cloud energy storage providers; c (C) Inv The cost is stored for energy storage;indicating the income under the predicted load y at the time t; />Representing the running cost under the predicted load at time t; />Representing the total cost of consumer i; />Representing the capacity of consumer i to lease energy under a predicted load; />Representing the stored energy power purchased by the consumer i under the predicted load at the time t; />Representing the stored energy power consumed by consumer i under predicted load at time t; Δt is the load predicted time interval; beta is the consumer leasing energy storage capacity proportionality coefficient; lambda (lambda) t The time coefficient of the energy storage purchased by the consumer at the moment t; θ t Is the time coefficient of energy storage consumed by the consumer at the moment t;
step (3): converting the double-layer model into a linear programming problem which is convenient to solve by adopting rolling optimization, and solving a cloud energy storage optimal control decision; the time scale of the rolling optimization is 1 hour; in order to eliminate fluctuation caused by prediction errors, the input of the rolling optimization is a load predicted value of the period, and the output is the optimal scale capacity and lease price of the Duan Yun energy storage operator at the next time; after each optimization, selecting a final optimization result as an initial value of rolling optimization in the next period, wherein the objective function of the rolling optimization is as follows:
wherein, gamma t The electricity price of the power grid in the t period;predicting the total power purchasing power under load for a cloud energy storage provider at the time t; />And->Maximum limits of energy storage power and energy storage capacity based on predicted load, respectively; alpha and eta are the scaling factors of the purchased stored energy power and stored energy capacity, respectively.
Compared with the prior art, the invention has the following advantages and effects:
(1) According to the invention, the competition learning method is expanded to the condition generation type countermeasure network training to predict the power load of one day, so that the training problems of unstable generation type countermeasure network, mode collapse and the like in the prior art are reduced, and the prediction precision of the generation type countermeasure network is improved.
(2) According to the method, the optimal energy storage decision-making double-layer model is built on the basis of prediction, so that profit of a cloud energy storage provider is maximized, energy storage and storage cost of a consumer is minimized, and energy storage utilization efficiency is improved.
(3) According to the invention, the rolling optimization is used for solving the double-layer model of the optimal energy storage decision, the whole model is converted into a linear programming problem which is convenient to solve, and the solving efficiency of the double-layer model of the optimal energy storage decision is improved.
Drawings
Fig. 1 is a diagram of a contention-generating type countermeasure network framework in accordance with the method of the present invention.
Fig. 2 is a double-layer control flow chart of the optimal energy storage decision of the method of the present invention.
Detailed Description
The invention provides a cloud energy storage double-layer optimization control method for a competitive condition generation type countermeasure network, which is described in detail below with reference to the accompanying drawings.
Fig. 1 is a diagram of a contention-generating type countermeasure network framework in accordance with the method of the present invention. First, extending competition learning into conditional generation type countermeasure network training, a plurality of generators are regarded as one competition population, and a discriminator serves as a competition environment. The input of the generator is a gaussian distributed noise vector with a mean value of 0 and a standard deviation of 1. For each mutation step, the generator is updated with least squares mutation, maximum and minimum mutation and heuristic mutation targets to adapt to the current environment, i.e., to the arbiter. Different antagonistic objective functions aim at minimizing different distances between the generated distribution and the actual data distribution, resulting in different mutations. At the same time, the quality of the generated samples is evaluated by means of a discriminator. Finally, according to the principle of "survival of the fittest", poorly performing generators are culled, and the best performing generators are selected for further training. The historical power load data is provided to the generator and the arbiter as a condition to approximate the data generated by the generator to the historical load sequence.
Fig. 2 is a double-layer control flow chart of the optimal energy storage decision of the method of the present invention. And constructing an external electricity purchasing model of a cloud energy storage provider at the upper layer of the double-layer model of the optimal energy storage decision, and optimizing the lease price. And constructing a consumer energy storage behavior model at the lower layer of the double-layer model of the optimal energy storage decision, and optimizing the total cost. And adopting rolling optimization to obtain a linearization model so as to solve optimal pricing and capacity decisions of the cloud energy storage operators.
Claims (1)
1. A cloud energy storage double-layer optimization control method of a competitive condition generation type countermeasure network is characterized in that the method integrates a competitive learning method into the condition generation type countermeasure network method and is used for double-layer optimization control of cloud energy storage; the cloud energy storage double-layer optimization control method of the competitive condition generation type countermeasure network comprises the following steps:
step (1): expanding the competition learning method to a condition generation type countermeasure network training for predicting load power of a future day, wherein a discriminator serves as a competition environment, and a generator generates competition mutation in a given discriminator; training generators using the maximum and minimum mutations, heuristic mutations and least squares mutations during competition, thereby generating a competition population comprising a plurality of different generators, and finally selecting only the generator with the maximum quality fitness for further training; the generator gradually generates a prediction sample with higher quality fitness under the condition of historical power load, and the optimal generator can generate the prediction sample with higher precision after further training; the condition generation type countermeasure network fused with the competition method can overcome the problems of unstable condition generation type countermeasure network and mode collapse, improves the prediction precision of the condition generation type countermeasure network, and comprises an objective function, a maximum and minimum mutation objective function, a heuristic mutation objective function, a least square mutation objective function and a quality fitness function of the competition generation type countermeasure network as follows:
wherein x is the input sample; y is a predicted value; z is the noise vector; c is the condition; g represents a generator; d represents a discriminator; p (P) data The true probability distribution of the sample; p (P) s Probability distribution for a historical power load sequence; p (P) z Is the probability distribution of the noise vector; g (z) represents mapping the input noise z into data; d (G (z)) represents the probability that the arbiter can recognize that the generator generates data by noise z; d (x|c) represents the probability that x comes from the true data distribution under condition c;d (y|c) represents the probability that y comes from the noise data distribution under condition c; lg represents the base 10 logarithm;representing the expectation under a real sample;representing the expectations under the prediction samples; />Representing the expectation under the noise vector; />Representing a generator and a arbiter maximum and minimum game; />Representing generator maximum and minimum mutations; />Representing generator heuristic mutations; />Representing generator least squares mutation; />Representing a quality fitness function;
step (2): establishing a double-layer model of an optimal energy storage decision based on competition generation type countermeasure network prediction; the upper model maximizes the rental price of the cloud energy storage provider, and the lower model minimizes the total cost of consumer energy storage; aiming at the long-term planning problem, the proposed double-layer model explores the optimal scale capacity and lease price from the perspective of a cloud energy storage operator so as to maximize the net profit; exploring optimal lease and business decisions from the consumer's perspective to minimize their storage energy costs; the proposed objective function of the bilayer model is:
wherein NP CES Net profit for cloud energy storage providers; i Rent Lease revenue for cloud energy storage providers; c (C) Inv The cost is stored for energy storage;indicating the income under the predicted load y at the time t; />Representing the running cost under the predicted load at time t; />Representing the total cost of consumer i; />Representing the capacity of consumer i to lease energy under a predicted load; />Representing the stored energy power purchased by the consumer i under the predicted load at the time t; />Representing the stored energy power consumed by consumer i under predicted load at time t; Δt is the load predicted time interval; beta is the consumer leasing energy storage capacity proportionality coefficient; lambda (lambda) t The time coefficient of the energy storage purchased by the consumer at the moment t; θ t Is the consumer t moment cancellationTime coefficient of energy consumption;
step (3): converting the double-layer model into a linear programming problem which is convenient to solve by adopting rolling optimization, and solving a cloud energy storage optimal control decision; the time scale of the rolling optimization is 1 hour; in order to eliminate fluctuation caused by prediction errors, the input of the rolling optimization is a load predicted value of the period, and the output is the optimal scale capacity and lease price of the Duan Yun energy storage operator at the next time; after each optimization, selecting a final optimization result as an initial value of rolling optimization in the next period, wherein the objective function of the rolling optimization is as follows:
wherein, gamma t The electricity price of the power grid in the t period; p (P) t y Predicting the total power purchasing power under load for a cloud energy storage provider at the time t;and->Maximum limits of energy storage power and energy storage capacity based on predicted load, respectively; alpha and eta are the scaling factors of the purchased stored energy power and stored energy capacity, respectively.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110881691.0A CN113592553B (en) | 2021-08-02 | 2021-08-02 | Cloud energy storage double-layer optimization control method for competitive condition generation type countermeasure network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110881691.0A CN113592553B (en) | 2021-08-02 | 2021-08-02 | Cloud energy storage double-layer optimization control method for competitive condition generation type countermeasure network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113592553A CN113592553A (en) | 2021-11-02 |
CN113592553B true CN113592553B (en) | 2023-10-17 |
Family
ID=78253889
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110881691.0A Active CN113592553B (en) | 2021-08-02 | 2021-08-02 | Cloud energy storage double-layer optimization control method for competitive condition generation type countermeasure network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113592553B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114219161A (en) * | 2021-12-21 | 2022-03-22 | 广东纬德信息科技股份有限公司 | Electric load prediction method and system for electric vehicle charging pile |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109902824A (en) * | 2019-03-02 | 2019-06-18 | 天津工业大学 | It is a kind of to generate confrontation network method with self adaptive control learning improvement |
CN110034561A (en) * | 2019-05-15 | 2019-07-19 | 长沙理工大学 | A kind of wind farm energy storage capacity optimization method based on cloud energy storage lease service |
WO2020143104A1 (en) * | 2019-01-08 | 2020-07-16 | 南京工程学院 | Power grid mixing and rolling scheduling method that considers clogging and energy-storing time-of-use price |
CN112134300A (en) * | 2020-10-09 | 2020-12-25 | 国网江苏省电力有限公司无锡供电分公司 | Reservation-based rolling optimization operation method and system for electric vehicle light storage charging station |
CN112381271A (en) * | 2020-10-30 | 2021-02-19 | 广西大学 | Distributed multi-objective optimization acceleration method for rapidly resisting deep belief network |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20210075221A1 (en) * | 2019-09-06 | 2021-03-11 | King Fahd University Of Petroleum And Minerals | Real time energy management and control of renewable energy based microgrid in grid-connected and island modes |
-
2021
- 2021-08-02 CN CN202110881691.0A patent/CN113592553B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020143104A1 (en) * | 2019-01-08 | 2020-07-16 | 南京工程学院 | Power grid mixing and rolling scheduling method that considers clogging and energy-storing time-of-use price |
CN109902824A (en) * | 2019-03-02 | 2019-06-18 | 天津工业大学 | It is a kind of to generate confrontation network method with self adaptive control learning improvement |
CN110034561A (en) * | 2019-05-15 | 2019-07-19 | 长沙理工大学 | A kind of wind farm energy storage capacity optimization method based on cloud energy storage lease service |
CN112134300A (en) * | 2020-10-09 | 2020-12-25 | 国网江苏省电力有限公司无锡供电分公司 | Reservation-based rolling optimization operation method and system for electric vehicle light storage charging station |
CN112381271A (en) * | 2020-10-30 | 2021-02-19 | 广西大学 | Distributed multi-objective optimization acceleration method for rapidly resisting deep belief network |
Non-Patent Citations (1)
Title |
---|
基于深度自适应动态规划的孤岛主动配电网发电控制与优化一体化算法;殷林飞;余涛;张泽宇;张孝顺;;控制理论与应用(第02期);44-58 * |
Also Published As
Publication number | Publication date |
---|---|
CN113592553A (en) | 2021-11-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Younes et al. | A memory-based gravitational search algorithm for solving economic dispatch problem in micro-grid | |
Wang et al. | Vehicle to grid frequency regulation capacity optimal scheduling for battery swapping station using deep Q-network | |
CN109508857B (en) | Multi-stage planning method for active power distribution network | |
Liu et al. | Dynamic pricing for decentralized energy trading in micro-grids | |
Rajesh et al. | Diminishing energy consumption cost and optimal energy management of photovoltaic aided electric vehicle (PV-EV) by GFO-VITG approach | |
Luo et al. | Short‐term operational planning framework for virtual power plants with high renewable penetrations | |
Soares et al. | Distributed energy resource short-term scheduling using Signaled Particle Swarm Optimization | |
Ghadimi et al. | PSO based fuzzy stochastic long-term model for deployment of distributed energy resources in distribution systems with several objectives | |
Jasmin et al. | Reinforcement learning approaches to economic dispatch problem | |
CN109190802B (en) | Multi-microgrid game optimization method based on power generation prediction in cloud energy storage environment | |
Chen et al. | A battery management strategy in microgrid for personalized customer requirements | |
CN113098007B (en) | Distributed online micro-grid scheduling method and system based on layered reinforcement learning | |
Hong et al. | Locational marginal price forecasting using deep learning network optimized by mapping-based genetic algorithm | |
Taylor et al. | Consolidated dynamic pricing of power system regulation | |
Hassler | Heuristic decision rules for short-term trading of renewable energy with co-located energy storage | |
Feng et al. | Scenario reduction for stochastic unit commitment with wind penetration | |
CN113592553B (en) | Cloud energy storage double-layer optimization control method for competitive condition generation type countermeasure network | |
CN117578409A (en) | Multi-energy complementary optimization scheduling method and system in power market environment | |
Kosana et al. | A novel reinforced online model selection using Q-learning technique for wind speed prediction | |
CN113794199A (en) | Maximum profit optimization method of wind power energy storage system considering electric power market fluctuation | |
Zhang et al. | Distributionally robust co-optimized offering for transactive multi-energy microgrids | |
Yang et al. | Deep reinforcement learning for economic energy scheduling in data center microgrids | |
CN116544934B (en) | Power scheduling method and system based on power load prediction | |
Rahman et al. | On efficient operation of a V2G-enabled virtual power plant: when solar power meets bidirectional electric vehicle charging | |
CN109615151B (en) | Load energy storage double-optimization prediction method, device and medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |