KR20190132187A - An incentive-based demand response method and system considering composite DR resources - Google Patents

Abstract

Demand response (DR) has been recognized as a strong tool helping to mitigate power imbalance in a future smart energy system. The present invention constructs an intraday resource trading framework bringing DR resources in various industry fields (large company or small business clients) at a system level from a viewpoint of a grid operator (GO). The GO and generators are evaluated to procure necessary system resources at minimal costs. When considering separate actions of the groups involved, the Stackelberg game theory is adopted to analyze coordination among various decision makers. Identified was a unique Stackelberg equilibrium producing an optimal result for a resource trading game between the GO and demand-side participants. A simulation result shows that the total procurement costs were minimized by applying an approach according to the present invention. The incentive-based demand response method repeats first to fifth steps when a difference between calculated costs and costs calculated by m iterations is greater than a predetermined value, determines that the Stackelberg equilibrium has been reached when the difference between the calculated costs and the costs calculated by m iterations is less than or equal to the predetermined value, and allows demand responses to be made by using grid operator incentives in a Stackelberg equilibrium state, industrial consumer load reductions, and service provider load reductions.

Description

Incentive-based demand response method and system considering composite DR resources

The present invention relates to incentive-based demand response (DR). Specifically, the grid operator (GO) is the least expensive in a market environment with complex demand response resources consisting of industrial consumers (ICs) and service providers (SPs). The present invention relates to an incentive-based demand response method and system.

Conventional electric energy systems have a high penetration rate of renewable energy sources and are undergoing rapid changes due to various energy usage patterns, which may cause large fluctuations between supply and demand. Therefore, grid operators (GOs) need to procure large amounts of resources to balance these fluctuations and increase the operational flexibility of the grid.

In particular, the intermittent nature of renewable energy makes supply-demand balancing work very dynamic and difficult to predict. Efficient ways to mitigate imbalances require intelligent integration of all stakeholders connected to the electrical energy system, enabling the electrical grid to be monitored, managed and optimized in an effective manner. The Internet of Things (IoT) is anticipated as a means of developing intelligent management of electrical energy systems due to its extensive sensing and networking capabilities, a key requirement for functionalizing smart energy systems.

Traditionally, power systems rely on generators to mitigate the imbalance effects caused by high demand or renewable resources. In the past decade, demand response (DR) has been recognized to have significant potential to provide balanced resources in addition to increasing the economics of the entire power system. This economic efficiency is because the services provided by DR resources are much cheaper than the services of flexible generators.

Previous studies have been conducted by researchers in order to model the end consumer's DR in response to grid signals at the consumption level for residential customers as well as the commercial sector. Other studies have attempted to put industrial load into disaster recovery plans. Some studies have brought DR to an aggregate level. A day ahead of the estimated market price, a decision model was developed to optimize the participation of DR collectors in the wholesale energy market. In order to reduce potential overload, coordinators and scheduling frameworks have been proposed in previous studies to allow collectors to drive DR from consumers through monetary incentives. Another study suggests a DR mechanism that aggregates capacity resources from a service provider (SP) subscribed customer by providing incentives in advance.

Recent research shows that the fundamental importance of DR resources is greater if local system operators can dispatch themselves. Several studies have explored DR from this point of view. Previous studies have proposed a two-stage balancing mechanism to maintain power system consistency by procuring load tracking reserves from both power generation and demand side resources through special handling of industrial loads, but large customers were simply treated. In addition, the previous study proposed a power balance plan at the system load level, so that the generation amount and the load were handled integrally, and only the industrial load was regarded as a qualified demand side resource and the load adjustment cost was assumed to be known in advance by the system operator. This model was further developed, where the customer load shift bidding curve was applied to represent the total response of customers submitting bids to GO in the daily load shifting market.

In terms of practical implementation, DR resources can generally participate in the daily market through the role of direct bidding, bilateral contracts or system capacity. This means that DR resources need to be advanced several hours earlier, or GO must predict an emergency ahead of time. Recent experience, however, has shown that daily market decisions have higher uncertainty than rapid decisions that are made in near real time because of the very dynamic and unpredictable nature of intermittent renewable resources. In this regard, the value of DR can be seriously impaired because the vast majority of DR resources may not be available in real time. Therefore, a new resource trading framework is needed, which should increase not only the value of DR but also the competitiveness.

Korean Patent Registration No. 10-1748766

The present invention was devised to solve the above-mentioned problems, and an object of the present invention is to minimize the cost of the grid operator in an environment in which complex DR resources exist, while trading new resources that can maximize the benefit of a subject having DR resources. To provide a framework.

To this end, the incentive-based demand response method according to the present invention is an incentive-based demand response method of a demand response system composed of a grid operator (GO), an industrial consumer (IC) and a service provider (SP), wherein the grid operator is m + 1 applies to industrial consumers who have updated grid operator incentives (π ^m _GO ) from iteration m + 1 to m iterations and based on the updated grid operator incentives (π ^{m + 1} _GO ) A first step of providing a first incentive and a second incentive applied to the service provider to the industrial consumer and the service provider, respectively, and the industrial consumer load reduction amount updated by the industrial consumer based on the first incentive; A second step of transmitting to the grid operator and a service provider load reduction amount updated by the service provider based on the second incentive to the grid operator; A third step of transmitting, a fourth step of the grid operator calculating a cost based on the updated grid operator incentives, an industrial consumer load reduction amount, and a service provider load reduction amount; If C ^{m + 1} _GO ) is less than or equal to the cost calculated in m iterations (C ^m _GO ), the calculated cost is determined as the optimal cost, and the updated grid operator incentive is compared with the maximum value of the grid operator incentive. A fifth step of determining the updated grid operator incentive as an optimal grid operator incentive, and a sixth step of determining, by the grid operator, whether a difference between the calculated cost and the cost calculated in m iterations is equal to or less than a predetermined value. Including the first step to the fifth stage if the difference between the calculated cost and the cost calculated in m iterations greater than a predetermined value Is repeated, and if the difference between the calculated cost and the cost calculated in the m iterations is less than or equal to the predetermined value, it is determined that the Staeckelberg equilibrium has been reached and the Staggerberg equilibrium grid operator incentive, industrial consumer load reduction, service provider It is characterized in that the demand response is carried out using the load reduction amount.

The incentive-based demand response method according to the present invention is an incentive-based demand response method of a demand response system composed of a grid operator, an industrial consumer, and a service provider, wherein the grid operator pays an incentive cost to an industrial consumer according to an industrial consumer load reduction amount. Minimizes procurement costs by adding incentive costs to service providers according to service provider load reductions, generation costs to produce resources that reduce industrial consumer load reductions and service provider load reductions from grid resource shortages. Determine grid operator incentives as much as possible.

The incentive-based demand response method according to the present invention is an incentive-based demand response method of a demand response system composed of a grid operator, an industrial consumer, and a service provider, wherein the grid operator updates a grid operator incentive and based on the updated grid operator incentive. Updating the first incentive applied to the industrial consumer and the second incentive applied to the service provider, wherein the industrial consumer calculates an industrial consumer load reduction based on the first incentive; Estimating service provider load reduction based on incentives; the grid operator calculates costs based on grid operator incentives, industrial consumer load reductions, and service provider load reductions, Includes determining if it has been reached do.

In the method of managing demand response of the grid operator (GO) according to the present invention, in the method of managing demand response of the grid operator (GO) for industrial consumers and service providers, the initial value of the grid operator (GO) incentive ratio is set, and The first GO incentive rate to be applied to the industrial consumer and the second GO incentive rate to be applied to the service provider are determined based on the target value, and the initial value of the demand reduction amount of the industrial consumer and the initial value of the demand reduction amount of the service provider are received. An initial step of calculating the initial value of the procurement cost of the grid operator; and updating the grid operator incentive ratio to calculate an update value of the first GO incentive ratio and an update value of the second GO incentive ratio; An update step sent to the service provider, an update value of the demand reduction amount of the industrial consumer, and an update of the demand reduction amount of the service provider A cost of calculating a grid operator procurement cost based on an update value receiving step of receiving a value, and an update value of the grid operator incentive ratio, an update value of demand reduction amount of an industrial consumer, and an update value of demand reduction amount of a service provider And a difference value comparing step of comparing the difference between the calculated procurement cost and the previously calculated procurement cost with a specific value.

An incentive-based demand response system according to the present invention includes a grid operator server that provides a grid operator incentive, an industrial consumer server that updates its load reduction based on the grid operator incentive, and transmits the load reduction amount to the grid operator server; The grid operator server includes a service provider server that updates its load reduction amount based on operator incentives and sends it to the grid operator server, based on grid operator incentives, industrial consumer load reductions, and service provider load reductions. It is characterized by repeating the process of calculating the procurement cost of the grid operator.

Grid operator server according to the present invention is a grid operator server that manages the demand response to industrial consumers and service providers, the communication unit for performing data communication with the management server of the industrial consumer and the service provider, and applied to the industrial consumer The first incentive to be applied and the second incentive to be applied to the service provider are updated and transmitted to the management server of the industrial consumer and the service provider, respectively, through the communication unit, and the industrial consumer load is reduced from the management server of the industrial consumer through the communication unit. And a controller for receiving the service provider load reduction amount from the service provider and the service provider and calculating a procurement cost based on the first incentive, the second incentive, the industrial consumer load reduction amount, and the service provider load reduction amount.

The incentive-based DR program according to the present invention is a program recorded on a computer-readable recording medium for managing demand response to industrial consumers and service providers. The incentive-based DR program sets the initial value of the grid operator (GO) incentive ratio, Based on the initial value, the ratio of the first GO incentive to be applied to the industrial consumer and the second GO incentive to be applied to the service provider is determined, and the initial value of the reduction in the demand of the industrial consumer and the demand reduction of the service provider is determined. Receiving and calculating the initial value of the procurement cost of the grid operator; and updating the grid operator incentive ratio to calculate the update value of the first GO incentive ratio and the update value of the second GO incentive ratio, and update the update value to the industrial consumer, respectively. And an update step sent to the service provider, an update value of the demand reduction amount of the industrial consumer, An update value receiving step of receiving an update value of the demand reduction amount of the child; and based on the update value of the grid operator incentive ratio, the update value of the demand reduction amount of the industrial consumer, and the update value of the demand reduction amount of the service provider. A cost calculation step of calculating the procurement cost and a difference value comparison step of comparing the difference between the calculated procurement cost and the previously calculated procurement cost with a specific value are performed.

As described above, the present invention can be valued by the grid operator (GO) with generators because the present invention can mitigate system imbalances while minimizing procurement costs by integrating resources from various sectors into the market for a day. . This can enhance system economic efficiency. The main effects of the present invention are as follows.

First, a comprehensive resource trading framework can be established to provide DR resources that engage resources in a variety of sectors, including large industrial and small- and medium-sized customers, through mediators or SPs.

Second, by utilizing two types of incentives, grid operators (GOs) can derive various DR resources from various fields and can be evaluated along with those generated through the proposed trading framework.

Third, the Stackelberg game theory can be adopted to analyze the coordination between hierarchical decision makers, given the unique objectives and the inherent relationships of the parties involved. The existence of the unique Staeckelberg equilibrium (SE) is demonstrated for single leader plural followers games, which can provide theoretically optimal results in a centralized manner.

Fourth, in order to protect the privacy of each demand-side participant, we can propose a fully distributed algorithm for finding the Staeckelberg equilibrium (SE) without disclosing the privacy information of the individual that matches the real life requirements.

Figure 1 shows the configuration of the incentive-based demand response system according to the present invention.
2 shows a graph in which utility values of industrial consumers change according to sensitivity indicators.
Figure 3 shows a Steakberg game structure according to the present invention.
Figure 4 is a graph showing the number of iterations to converge to the Staeckelberg equilibrium according to the present invention.
5 shows a composite resource for compensating for shortage resources.
6 shows the total demand reduction of each service provider.
Figure 7 shows the parameters of the demand participant.
8 shows the algorithm result for another p.
9 shows the signal flow between each subject in the incentive-based demand response system according to the present invention.
10 illustrates an incentive-based demand response process performed by the grid operator according to the present invention.
11 is a server-based incentive-based demand response system according to the present invention.
12 illustrates an internal configuration of a grid operator server according to the present invention.

The description of the present invention is configured as follows. First formulate the problem, then construct a mathematical model for each participant. It then provides a Statenberg game theory analysis of coordination among various market participants. Finally, the simulation results are explained.

Problem formulation

To ensure safe operation of the power system throughout the day, grid operators (GOs) must anticipate shortages of resources in the short term and anticipate them. Typically, this lack of resources is overcome by flexible generators (eg diesel generators and gas turbines) that can be controlled directly and quickly at negligible initial cost. However, these power plants generally have high operating costs, which can lead to a sharp rise in electricity costs and can be a huge financial burden from a GO perspective. The demand side participation in the electricity market is widely recognized as a significant means of absorbing demand imbalances and reducing the financial burden of GO. In this demand side participation, the load reductions provided by the demand side must be compensated for by the prescribed incentive rate. do.

In the present invention, demand side participation in the incentive program of the grid operator is achieved in two ways. 1) how large industrial consumers submit load reductions directly to grid operators; 2) how service providers (SPs) collect load reductions from small and medium-sized consumers (eg, residential or commercial users).

Regarding the actual tariff, the electricity price of industrial consumers is generally lower than that of residential users. Thus, it is necessary for the grid operator to provide various incentives to different demand side participants.

와

는 산업 소비자(IC)와 서비스 제공자(SP)에게 각각 지급되는 인센티브이다. 여기서 인센티브라는 것은 정확히 말하면 산업 소비자와 서비스 제공자에게 지급되는 인센티브 금액이 아니라 인센티브 금액 지급을 위해 적용되는 인센티브 비율을 의미한다. 본 발명의 명세서에서 인센티브와 인센티브 비율은 혼용될 수 있다. 이하, 시장 역할의 목적에 대해 상술한다. 1 shows an incentive-based demand response system according to the present invention. here,

Wow

Are incentives paid to industrial consumers (ICs) and service providers (SPs), respectively. Incentives here, to be exact, refer to the percentage of incentives applied to pay incentives, not the incentives paid to industrial consumers and service providers. In the present specification, incentives and incentive ratios may be used interchangeably. Hereinafter, the purpose of the market role will be described in detail.

1. Industrial Consumer Model

Indeed, industrial consumers (ICs) try to increase profits, which is similar to increasing production to some extent. Here, assume that profit is a function of the increase in energy consumption with marginal profit reduction.

In Equation 1, the function ψ _l (x _l ) is used to quantify the profit of the consumer l (∀l∈ L , L = | L |).

Where x _l is energy consumption, σ _l and ω _l are user dependent parameters that distinguish industrial consumers. In particular, σ _l is an indicator of sensitivity to changes in profit when there is a change in x _l .

가 제공될 때 감당할 수 있는 부하 감축량이다. 따라서 소비된 부하는 D_l ^ava - D_IC,l로 표현될 수 있다. For consumer l with available load D _l ^ava , D _{IC, l} is an incentive from the grid operator

The amount of load reduction that can be afforded when is provided. Therefore, the consumed load can be expressed as D _l ^ava -D _{IC, l} .

To this end, the industrial consumer aims to maximize the utility function, which is the sum of the profits from the product and the profits from participation in the grid operator's program, as defined in Equation 2 (2a).

Here, x _l of the expression (1) is ^ava D _l - D is replaced with the _{IC, l,} D _{IC, l} has a constraint (2b).

In order to understand the objective function of Equation 2 (2a), as shown in Fig. 2, all other parameters are the same to graphically plot the utility values of three industrial consumers according to load reduction. As the load reduction increases, the overall utility value continues to increase, and as the load decreases further, it begins to decrease past its peak. In particular, for consumers with a high sensitivity indicator σ _l , the load reduction is greater after a trade-off between the two terms of (2a). In other words, this class of consumers will benefit more by consuming less energy and selling more load reductions with incentives provided by grid operators.

2. Service Provider Model

)로 도매 전기시장에 참여하여 수집된 수요 감축을 그리드 운영자에게 판매한다. 이를 위해, 서비스 제공자의 목적은 가입 소비자와 그리드 운영자 간 가격 차이를 통해 획득한 수익을 최대화하는 것이다. 그리드 운영자가 조직한 도매 시장에 2개 이상의 서비스 제공자가 참여한다고 가정하면, K는 서비스 제공자의 집합이고, 서비스 제공자의 수 K=|K|이다. 서비스 제공자 k∈K에 대해, 그 목표는 수학식 3과 같이 정의된다. As shown in FIG. 1, the service provider SP is in the collecting phase and acts as an agent to sell demand reductions of subscribing consumers to the grid operator GO. Specifically, service providers operate their sub-programs to provide incentives to subscribing consumers, and subscribing consumers provide demand reductions in exchange for incentive payments. The service provider is responsible for incentives for grid operators (

) Participate in the wholesale electricity market and sell the collected demand reductions to the grid operator. To this end, the service provider's purpose is to maximize the revenue obtained through price differences between subscribing consumers and grid operators. Assuming that two or more service providers participate in the wholesale market organized by grid operators, K is the set of service providers and the number of service providers K = | K | For service provider k ∈ K , the goal is defined as in equation (3).

Where D _{SP, k} represents demand reduction collected by service provider k, and π _{SP, k} is an incentive provided to subscribing consumers. In order to maximize Eq. (3a), two variables, D _{SP, k} and π _{SP, k} , must be optimized. In (3b), D _{i, k} represents a demand reduction amount provided by the consumer i of the service provider SP k. N _k (N _k = | N _k |) represents a set of consumers subscribed to a service provider.

For user i ∈ N _k , given π _SP, k from SP k _{, the} goal is to maximize the utility function that combines incentive income and dissatisfaction cost φ _{i, k} as shown in equation (4).

Here, μ _{i, k} is defined as the weight for φ _{i, k} . Constraint 4b means that D _{i, k} does not exceed the available capacity. In particular, the dissatisfaction cost function φ _{i, k} in (4a) is designed to indicate the degree of discomfort experienced by the consumer in reducing demand. This is defined as a convex function as shown in Equation 5 since the dissatisfaction increases rapidly as demand reduction increases.

In Equation 5, θ _{i, k} and λ _{i, k} are consumer dependent parameters, and θ _{i, k} reflects consumer attitudes toward demand reduction. Larger θ _{i, k} means consumers are more conservative about demand reduction, and smaller is the opposite.

3. Grid Operator Model

When a system resource shortage D _req is expected, the grid operator's goal is to minimize the cost of compensating D _req . Grid operators can obtain resources from demand-side participants, ie industrial consumers and service providers, or from fluid generators. Given the complex system resources coming from the three sectors, the purpose of the grid operator is to minimize total procurement costs, as shown in Equation 6.

In (6a), C _Gen (G) represents the cost of producing power G, and the other two terms represent the cost paid to industrial consumers and service providers. Constraints 6b indicate that incentives provided to industrial consumers should not exceed incentives paid to service providers, and (6c) indicates that the sum of resources in all sectors should be D _req .

In particular, C _Gen (G) of (6a) is pure convex, assuming a monotonically increasing function of power generation amount (G). Without losing generality, C _Gen (G) is expressed as Equation (7).

Here, a, b and c are power generation coefficients and may be used by the grid operator in advance.

Analysis of the Steakenberg Game Theory for Coordination Between Hierarchical Market Participants

1. Formulating the Staelberg Game

와

)가 수요측에 제공될 때, 각 참가자는 수학식 2 및 3의 최대화 문제를 풀어 자신의 부하 감축을 결정할 것이다. 수요측에서 수집된 부하 감축은 조달 비용일 뿐만 아니라 부족 보상을 위한 자원 복합이 되는 것이다. 따라서, 그리드 운영자는 인센티브를 조정하여 수요측에서 부하 감축을 변경하면서 비용을 줄이도록 유도한다. 이와 관련하여, 한 참가자의 선택이 다른 참가자의 결정에 어떠한 영향을 줄 것이다. 이에 따라, 이러한 인자들이 그리드 운영자와 각 수요측 참자가 간의 상호 작용으로 이어진다. From the above problem formulation, the Grid Operator Incentive (

Wow

When) is provided to the demand side, each participant will solve their maximization problem in Equations 2 and 3 to determine their load reduction. Load reduction collected on the demand side is not only a procurement cost, but also a resource mix for undercompensation. Thus, grid operators adjust incentives to drive down costs while changing load reductions on the demand side. In this regard, the choice of one participant will have some influence on the decision of the other participant. As such, these factors lead to the interaction between the grid operator and each demand side participant.

와

)를 공표하면서 리더 역할을 수행한다. 문제를 다루기 쉽게 하기 위해,

는

의 선형 함수라고 가정한다. 이와 관련하여, 그리드 운영자는 시간마다 하나의 전략을 선택하면 된다. π_GO를 기본 전략으로 표시하며,

와

는 수학식 8 및 수학식 9와 같이 표시된다. The Steckenberg game is a class of expansion games that studies the situation of hierarchical determinants, in which the leader acts first according to the strategy and then the followers act in response to the leader's decision. The Staeckelberg game is adequate to be described in detail behind the problem described in Formulating the Problem. In the present invention, one leader and a plurality of follower Stakelberg games are formulated between grid operators, industrial consumers and service providers. As shown in FIG. 3, the grid operator has his strategy, GO incentives, to demand side followers including L industrial consumers and K SPs.

Wow

) Act as a leader while publishing. To make matters easier to deal with,

Is

Assume that is a linear function of. In this regard, grid operators need to select one strategy each time. π _GO as the default strategy,

Wow

Is expressed as in Equation 8 and Equation 9.

Here, the parameter ρ in Equation 8 is adjustable to adjust the incentive size for other participants by providing local GO flexibility.

The strategy of industrial consumer l is load reduction (D _{LC, l} ), and the strategy of SP k is total load reduction (D _{SP, k} ) collected from subscribing consumers.

Given a one-leader plural follower Stakelberg game with a hierarchical decision structure, the solution to the game is the Staeckelberg equilibrium (SE), which is defined as follows.

SE definition: 1 leader plural followers In the Staeckelberg game, the strategy set ( D ^* _IC , D ^* _SP , ð ^* _GO ) constitutes an SE if the set of inequalities in Equations 10-12 is satisfied.

Where D ^* _{IC, -1} = [D ^* _{IC, 1} , D ^* _{IC, 2} , ..., D ^* _{IC, l-1} , D ^* _{IC, l + 1} , ..., D ^* _{IC, L} ] represents the strategy of all industrial consumers except consumer l, whereby D ^* _IC = [D ^* _{IC, l} , D ^* _{IC, -1} ] represents the strategy of all industrial consumers. Similarly, D ^* _SP = [D ^* _{SP, k} , D ^* _{SP, -k} ] contains the strategy of all service providers, where D ^* _{SP, -k} = [D ^* _{SP, 1} , D ^* _{SP, 2} , ..., D ^* _{SP, k-1} , D ^* _{SP, k + 1} , ..., D ^* _{SP, K} ] represents the strategy of all other service providers except SP k.

The physical meaning of equations 10-12 is that no participant can switch to another strategy to further reduce the cost unless all other participant's strategies are changed in the SE state.

2. Presence and Uniqueness of the Staeckelberg Equilibrium

Given the Staeckelberg game formula, verify that there is an SE that satisfies Equation 10-12. In order to verify the existence of this SE, we propose a proposition and related proof.

Proposition: 1 Leader between GO, Industrial Consumer, and SP The Multiple Follower Stakelberg game has a unique SE that satisfies Equations 10-12.

Proof: Backward guidance can be an effective way for game players to choose their strategy in a continuous fashion and to derive equilibrium for games with hierarchical determinants. Thus, the attestation process begins with identifying the best response of the demand participant and then backtracks the grid operator's best strategy.

a) Identify the best response from industrial consumers given GO incentives.

If ρπ _GO is provided from the grid operator, the best response of the industrial consumer l can be obtained according to Equation 2 (2a) for D _{ID, l} .

의 첫 번째 경우를 0으로 하면, D_IC,l는 수학식 14와 같이 된다. In equation (13)

If the first case of 0 is 0, D _{IC, l} is expressed by Equation 14.

이 언제나 양수이다. 따라서, 수학식 15와 같이 나타낼 수 있다. In practice, the grid operator sets a specific market entry criterion for the minimum load reduction requirement (D ^min ). Industrial consumers must meet minimum load reduction requirements in order to be authorized to sell load reductions to grid operators. Looking at Equation 14,

This is always positive. Therefore, it can be expressed as in Equation 15.

To accommodate all industrial consumers, it can be normalized as

The physical meaning in Equation 16 indicates that a reasonable industrial consumer should provide at least a minimum load reduction if he or she decides to participate in the DR operator's DR program. Therefore, the second case of Equation 13 is excluded from the present invention, and in Equation 2 (2b), the minimum load reduction value of the industrial consumer l is replaced with D ^min of Equation 16. Subsequently, the load reduction amount in Equation 14 becomes the only best response function of the industrial consumer l as described later. In addition, the total load reduction amount of all industrial consumers is expressed by Equation 17.

For convenience of notation, each term in Equation 17 is represented as in Equation 18.

Therefore, the best response of all industrial consumers can be expressed by Equation 19 by substituting Equation 18 into Equation 17.

b) Confirm the best response from the service provider provided with the GO incentive.

Indeed, it can be seen that the SP must act as a surrogate for the subscribing consumer or the subcontractor delegates its representative SP to sell demand reductions to the wholesale market. The following describes how a service provider will act as an agent during the collection phase.

In the aggregation phase, the service provider (SP) acts as a commissioner to sell the intended reduction in demand to the wholesale market. Details of the delegation process are as follows:

에 대해, 수학식 4의 (4a)에서 사용자 i의 목적함수

는

에 대하여 순 오목이며, 이것은

의 2차 미분, 예를 들어

을 통해 검증된다. 따라서, 각 사용자 i는

를 최대화하는 고유한

를 가지며, 이것은 (4a)의 1차 도함수 즉,

를 취함으로서 얻어질 수 있다. 따라서 다음과 같은 식을 얻을 수 있다.Given incentives

For i, the objective function of user i in equation (4a)

Is

Against net is concave, which is

Second derivative of, for example

Verified by Thus, each user i

Unique to maximize

Which is the first derivative of (4a)

Can be obtained by taking. Therefore, the following equation can be obtained.

(I.1)

Clause (I.1) represents the behavior of user i in terms of SP k. Substituting (I.1) into (3b) and then into the SP objective function of (3a), it can be reformulated as:

(I.2)

는

에 의해

에 대해 순오목하다. 따라서, SP k의 수익을 극대화하는 고유한 SP 인센티브 값이 존재한다는 것은 명확하다.

의 1차 도함수를 취하면 다음과 같은 식을 얻는다.Of reformulated (I.2)

Is

By

Are lukewarm about. Thus, it is clear that there is a unique SP incentive value that maximizes the return of SP k.

Taking the first derivative of, we obtain

(I.3)

을 0으로 할 경우

은 다음과 같이 나타난다.

Is 0

Appears as follows:

(I.4)

은 (I.4)를 (I.1)에 대입한 다음 사용자N_k에 대해 합산하면, (I.5)와 같은 고유한 형식을 취하게 된다Subsequently, reduce total demand for SP k

Substituting (I.4) into (I.1) and then summing for user N _k takes the form of (I.5):

(I.5)

Accordingly, in response to π _GO , Equation 20 shows that there is a unique strategy of SP k (D _{SP, k} ).

Equation 20 means that each SP k ∈ K with subscribing consumers will act together on the external entity as a whole. In addition, the demand reduction amount of all service providers may be expressed by summing all service providers.

It can be seen that the summation portion of Equation 21 is a constant. For simplicity, as in Equation 22, the best response of all service providers can be expressed as Equation 23 by putting Equation 22 into Equation 21.

c) Verify the existence of GO's only best strategy following the principle of backward guidance.

First, the objective function of the grid operator of Equation 6 (6a) may be expressed as Equation 24 including Equation 6 (6c) and Equation 7-9.

According to the backward derivation principle, substituting the best response of the industrial consumers and service providers of Equations 19 and 23 in the objective function of Equation 24 is reformulated as in Eq.

When the first derivative of Eq. 25 is obtained with respect to? _GO , Eq.

The second derivative of Equation 25 is expressed as Equation 27.

Obviously, the term of equation 27 is always positive by equations 18 and 22. Accordingly, the new objective function of GO in Equation 25 is net convex for π _GO . As a result, the total optimal solution of π ^* _GO can be found solely as in Equation 28 with Equation 26 as zero.

Here, η, γ, and α are defined in equations (18) and (22), respectively.

Accordingly, the best strategies (D ^* _{IC, l} and D ^* _{SP, k} ) of each industry consumer and service provider can be determined by the best response to π ^* _GO . Therefore, the strategy profile ( D ^* _IC , D ^* _SP , SE equilibrium exists in the form ð ^* _GO ).

3. Algorithms for finding SE in a distributed fashion

SE can be derived in a central way, but this requires each demand side participant to disclose their personal information (parameters) to the grid operator. For this reason, a complete distributed algorithm is more desirable. We propose this algorithm with the goal of finding the SE through the exchange of messages between the grid operator and the demand side participants, without revealing each participant's personal information.

The key concept is that the grid operator updates the incentive π ^m _GO repeatedly from m to m + 1 based on the flexible step size Δπ _GO . During each iteration (eg m + 1), the grid operator proposes π ^{m + 1} _GO to the demand side and each industrial consumer and service provider make their best response by sending demand reductions to the grid operator. Determine the reductions (D ^{m + 1} _{IC, l} and D ^{m + 1} _{SP, k} ). Later, the grid operator calculates the total cost (C ^{m + 1} _GO ) based on Equation 31, as shown in the algorithm, and records the current π ^{m + 1} _GO if the result is a lower cost. In this way, lines 1 to 7 are repeated until the termination condition of equation (32) is satisfied. In the termination condition, the cost of the grid operator is no longer reduced. In other words, if the difference between two successive repetitions is less than or equal to a certain small value [epsilon], SE is reached.

In order to accelerate the convergence speed and improve the efficiency of the algorithm, the step size Δπ _GO is appropriately adjusted using equation (29). Where | C ^m _GO -C ^m-1 _GO | represents the difference in the total cost between two consecutive iterations, and δ and ω are system dependent parameters. Clearly, a larger difference between | C ^m _GO -C ^m-1 _GO | results in a larger Δπ _G , and as the difference decreases, the GO cost will saturate and Δπ _GO will be close to zero.

Algorithm

reset:

GO initializes the program by setting π * _GO = ð ^min _GO and selects the value of ρ, and then proposes initial incentives to the demanding participants according to equations (8) and (9).

Each industrial consumer and SP determine D ^* _{IC, l} and D ^* _{SP, k} as best responses based on Equations 14 and 20 and send them to GO.

Therefore, GO calculates the initial cost C ^* _GO based on equation (24).

repeat:

1: In the repetition m → m + 1 do the following:

2: Update π ^m _GO based on the step size Δπ _GO . Δπ _GO is calculated by equation (29), and π ^m _GO is calculated by equation (30).

3: In response to ð ^{m + 1} _GO , the demand participant updates D ^{m + 1} _{IC, l} and D ^{m + 1} _{SP, k} according to equations 19 and 20 and reports to GO.

4: Upon receiving D ^{m + 1} _{IC, l} and D ^{m + 1} _{SP, k} , GO calculates the total cost by equation (31).

5: If C ^{m + 1} _GO ≤C ^* _GO and π ^{m + 1} _GO ≤π ^max _GO , GO updates the optimal incentive and minimum cost (π ^* _GO = ð ^{m + 1} _GO , C ^* _GO = C ^{m + 1} _GO ).

6: Step 5 ends.

7: Steps 1-6 end.

8: Steps 1 to 7 are repeated until the end condition is satisfied as shown in Equation (32).

9: SE ( D ^* _IC , D ^* _SP , π ^* _GO ). The total procurement cost of GO is no longer reduced.

The GO cost function (C _GO ) is net convex for π _GO , which means that increasing π _GO using equations 29 and 30 leads to the minimum cost for GO. Thus, the distribution algorithm according to the present invention ensures that it always converges to a unique SE.

Figure 9 shows the signal flow between each object in the incentive-based demand response system according to the present invention.

)를 산정하여 산업 소비자(20)로 제공한다(S10). Referring to FIG. 9, a grid operator (GO) 10 is a first grid operator incentive applied to an industrial consumer (IC) 20 based on a grid operator incentive, i.e., GO incentive (π _GO ).

) Is calculated and provided to the industrial consumer 20 (S10).

)를 산정하여 서비스 제공자(20)로 제공한다(S12). 본 발명의 실시예에서는 제2 그리드 운영자 인센티브는 그리드 운영자 인센티브와 동일하다. Next, the grid operator 10 receives a second grid operator incentive, which is applied to the service provider (SP) 20 based on the GO incentive.

) Is calculated and provided to the service provider 20 (S12). In an embodiment of the present invention, the second grid operator incentive is the same as the grid operator incentive.

The industrial consumer 20 updates the industrial consumer load reduction amount D _IC based on the first grid operator incentive and transmits it to the grid operator 10 (S14).

In addition, the service provider 30 also updates the service provider load reduction amount D _SP based on the second grid operator incentive and transmits it to the grid operator 10 (S16). Although not shown in FIG. 8, the service provider 30 calculates a service provider load reduction amount by collecting load reduction amount D _i from subscribing customers.

Next, the grid operator 10 calculates the cost of the grid operator based on the grid operator incentive, the industrial consumer load reduction amount, and the service provider load reduction amount (S18).

Specifically, the grid operator 10 may include incentive costs to be paid to industrial consumers according to industrial consumer load reductions, incentive costs to be paid to service providers according to service provider load reductions, industrial consumer load reductions from resource shortages in the grid, and Calculate the procurement cost, which adds the cost of generating power to produce the amount of resources minus the service provider load reduction.

When calculating the procurement cost of the grid operator, it is determined whether the procurement cost has reached the Staeckelberg equilibrium (SE) (S18). Once the Staeckelberg equilibrium is reached, it will no longer be possible to reduce procurement costs. If the procurement cost is determined to have reached the Staeckelberg equilibrium, the grid operator 10 applies the Staggerberg equilibrium grid operator incentive to the industrial consumer 20 and service provider 30 to incentivize the demand response. Perform compensation (S20, S22).

If it is determined that the procurement cost has not yet reached the Staeckelberg equilibrium, the grid operator 10 updates the grid operator incentive and based on the updated grid operator incentive, the first grid operator incentive and the second grid operator incentive. After updating, the process is repeated from step S10 to step S16.

10 shows a demand response process performed in the grid operator according to the present invention.

Referring to FIG. 10, first, the grid operator performs an initialization step for managing demand response (S100).

In the initialization step 100, the grid operator 10 sets the initial value of the grid operator (GO) incentive rate, and the first GO incentive rate and service provider 30 to apply to the industrial consumer 20 based on the initial value. Determine the second GO incentive rate to be applied to, and calculate the initial value of the procurement cost of the grid operator by receiving the initial value of the demand reduction amount of the industrial consumer and the initial value of the demand reduction amount of the service provider.

Next, the grid operator 10 updates the grid operator incentive rate to calculate an update value of the first GO incentive rate and an update value of the second GO incentive rate, and transmits the update value to the industrial consumer and the service provider, respectively ( S102) is performed. When updating the grid operator incentive ratio in the update step (S102), the grid operator incentive ratio is increased by the step size calculated based on the difference in procurement cost and updated.

Next, the grid operator 10 performs an update value receiving step S104 of receiving the update value of the demand reduction amount from the industrial consumer 20 and the update value of the demand reduction amount from the service provider 30.

Thereafter, the grid operator 10 calculates the total procurement cost of the grid operator based on the update value of the grid operator incentive rate, the update value of the demand reduction amount of the industrial consumer, and the update value of the demand reduction amount of the service provider. (S106).

When the total procurement cost is calculated, the grid operator 10 compares the difference between the procurement cost calculated in the cost calculation step S106 and the previously calculated procurement cost to a specific value to determine whether the procurement cost has reached the Staeckelberg equilibrium. (S108).

In other words, if the difference between the procurement cost calculated in the cost calculation step (S106) and the previously calculated procurement cost is greater than a specific value, it is reported that the Steigenberge equilibrium has not yet been reached, and the cost calculation step (S106) from the update step (S102). Repeat to).

However, if the difference between the procurement cost calculated in the cost calculation step (S106) and the previously calculated procurement cost is less than or equal to the specified value, it is determined that the Staeckelberg equilibrium has been reached, and the updated value of the grid operator ratio in the Staeckelberg equilibrium state, The demand response is performed with an update value of the demand reduction amount of the industrial consumer and an update value of the demand reduction amount of the service provider (S110).

11 illustrates a server-based incentive-based demand response system according to the present invention.

Referring to FIG. 11, in the grid operator sector, the GO server 10 is the main agent to manage the demand response, and the industrial consumer sector is the IC server 20 the main agent to perform data communication with the GO server 10. In the service provider section, the SP server 30 is mainly used to perform data communication with the GO server 10.

)를 전송하고 IC 서버(20)는 제1 인센티브(

)에 근거해 산정한 산업 소비자 부하 감축량(D_IC)를 GO 서버(10)로 전송한다. The GO server 10 is an IC server 20 which is a first incentive to apply to industrial consumers.

) And the IC server 20 sends the first incentive (

), The calculated industrial consumer load reduction amount (D _IC ) is transmitted to the GO server 10.

)를 전송하고 SP 서버(30)는 제2 인센티브(

)에 근거해 산정한 산서비스 제공자 부하 감축량(D_SP)를 GO 서버(10)로 전송한다. Similarly, the GO server 10 is the SP server 30 with a second incentive to apply to the service provider.

) And the SP server 30 sends a second incentive (

And the acid service provider load reduction amount (D _SP ) calculated on the basis of the reference) is transmitted to the GO server 10.

GO server 10 provides incentives to IC server 20 and SP server 30 and repeats the process of receiving respective load reduction amounts from IC server 20 and SP server 30 to optimize the three party. Demand response results will be derived.

12 shows an internal configuration of a GO server according to the present invention.

Referring to FIG. 12, the GO server 10 is a communication unit 12. Memory 14, control unit 16, and the like.

The communication unit 12 is a part that performs data communication with the IC server 20 and the SP 30.

The communication unit 12 transmits the first incentive applied to the industrial consumer to the IC server 20 and receives the industrial consumer load reduction amount from the IC server 20. In addition, the communication unit 12 transmits a second incentive applied to the service provider to the SP server 30, and receives the service provider load reduction amount from the SP server 30.

The memory 14 stores a program for incentive-based demand response management according to the present invention. The demand response management program, as described above, allows grid operators to minimize procurement costs and allow industrial consumers and service providers to determine grid operator incentives, industrial consumer load reductions, and service provider load reductions to maximize their benefits. .

The controller 16 is a part that controls the overall operation of the GO server 10. The controller 16 executes a demand response management program stored in the memory 14 to control demand response management for industrial consumers and service providers.

The control unit 16 controls the data communication to the IC server 20 and the SP server 30 through the communication unit 12, and as described above, the Stakelberg equilibrium through the grid operator incentive update and procurement cost calculation Judgment will determine the best grid operator incentives, the best industrial consumer load reductions, and the best service provider load reductions.

Simulation result

는 0.13으로 선택되었고 ε는 10^-4로 설정되었다. 모든 매개 변수값이 본 발명의 시뮬레이션에서 특정되고 지역 전기 시장의 특정 설계, 세대 구성 또는 소비자 특성에 따라 달라질 수 있다는 것이 중요한 점이다. 그러나 이것이 결과의 분석과 해석을 왜곡하지 않는다. Incentive-based DR approach and simulation results for the analysis in accordance with the present invention are provided. In order to have a clear image of demand-side participant behavior in the resource trading process, simulations were performed between the grid operator (GO), three industrial consumers (ICs), and two service providers (SPs) with three subscribing users each. 7 provides the parameters of all demand side participants. The generation coefficient of Equation 7 is set to a = 0.2, b = 0, c = 0. Δ was chosen as 2.5 for the variance algorithm,

Is chosen to be 0.13 and ε is set to 10 ^-4 . It is important that all parameter values are specified in the simulations of the present invention and may vary depending on the specific design, generation configuration or consumer characteristics of the local electricity market. But this does not distort the analysis and interpretation of the results.

이 다음 작업 기간 동안 GO에 의해 예상된다(본 발명에서는 한 시간을 예로 들었음). 수학식 8의 매개변수 ρ는 처음에 0.6으로 설정되었고, 후반부에서는 GO의 의사결정에 대한 ρ의 영향을 조사하기 위해 다른 값이 적용되었다. 이를 위해 위에서 설명한 시나리오를 기반으로 MATLAB R2016b에서 분산된 방식으로 SE를 탐색하기 위해 분산 알고리즘을 수행했다. 알고리즘 결과 및 이에 상응하는 분석은 후술한다.Lack of resources at 96 kWh per day

Expected by GO for this next working period (in the present invention, one hour is taken as an example). The parameter ρ in Equation 8 was initially set to 0.6, and in the latter part, a different value was applied to investigate the influence of ρ on the decision of GO. To this end, based on the scenario described above, a distributed algorithm was performed in MATLAB R2016b to explore the SE in a distributed fashion. Algorithm results and corresponding analyzes are described below.

변화를 보여준다. 최적의 인센티브는 7.27 ￠/ kWh였다. 최적점을 찾는데 분산 알고리즘의 정확도를 고려하여, 수학식 28의 분석 방법을 사용하여 전역 최적 해를 구했고, 7.29 ￠/ kWh로 계산되었으므로 오차는 0.2%에 불과하였다. 4 shows the number of iterations needed to converge to the SE. The algorithm fully converged on the 11th iteration (left graph) when the termination condition of Eq. 32 was met, which means that the total cost of GO could no longer be reduced. Also the right graph shows GO incentives for repeat rounds

Show the change. The optimal incentive was 7.27 kWh / kWh. Considering the accuracy of the variance algorithm in finding the optimal point, the global optimal solution was obtained using the analysis method of Equation 28, and the error was only 0.2% since it was calculated to be 7.29 kWh / kWh.

To intuitively interpret how each demand-side participant responds to the GO's strategy in each iteration, Figure 5 tracks multiple sectors of complex resources for lack of compensation. Clearly, the load reductions submitted by each demand-side participant increase gradually in an iterative process when larger GO incentives are offered. It can be seen that the third industrial consumer with the largest sensitivity indicator σ contributed the most load reduction. The reason is the compromise obtained by solving the optimization problem in Equation 2, which shows that the GO incentives provided as load reduction are more profitable than less consumption.

Furthermore, the collected demand reductions of the SPs of FIG. 5 are further broken down in FIG. 6, which shows end user demand reductions for each SP incentive. By comparing the user group responses belonging to each SP, we observed that SP2 users were reluctant to reduce demand due to larger θ values. This means that a user with a large value of θ, as described above, is more conservative in providing demand reduction.

It is important to evaluate the cost savings achievable with the method according to the invention. For benchmarking, we assume that without demand-side participation, the lack of resources will only be compensated by generator operation. The benchmark cost was calculated to be $ 18.43 using equation (7). As shown in (a) of FIG. 4, after enabling demand side participation, the minimum cost was $ 5.06 (when p was set to 0.6).

For a more comprehensive understanding, we applied various values of ρ from 0.2 to 1.0 and obtained relevant results from the algorithm. FIG. 8 shows the effect of different p on optimal incentive (second row) and total cost (third row), where the fourth row represents the cost savings ratio compared to the benchmark. The larger ρ, the lower the resource shortage compensation costs and the lower the GO incentive. This is because larger ρ can reduce more load from industrial consumers at lower cost compared to the cost required for GO to procure the same quantity at the SP. Therefore, as a policy maker, GO should carefully select the value of ρ based on the characteristics of the DR resource composition in the controlled area.

As such, the present invention presents a comprehensive resource trading framework from the perspective of the GO field, and integrates DR resources from various sectors into the daily market to help alleviate system imbalances with generators. By taking advantage of the two types of incentives, GO can derive a mix of different DR resources to minimize total procurement costs. To better illustrate the implications among different market participants, the Stackelberg game theory was adopted to model the resource trading process in a single leader multiplayer game. The analysis shows that there is a unique SE in the game. This resulted in optimal trading results and minimized procurement costs.

The above description is merely illustrative of the present invention, and various modifications may be made by those skilled in the art without departing from the technical spirit of the present invention.

Therefore, the embodiments disclosed in the specification of the present invention are not intended to limit the present invention. The scope of the present invention should be construed by the claims below, and all techniques within the scope equivalent thereto will be construed as being included in the scope of the present invention.

10: Grid Operator (GO) 20: Industrial Consumer (IC)
30: Service Provider (SP)

Claims (12)

In the incentive-based demand response method of the demand response system consisting of a grid operator (GO), industrial consumers (IC) and service providers (SP),
The grid operator updates the grid operator incentive (π ^m _GO ) from m + 1 iteration m + 1 to m iteration m and updates based on the updated grid operator incentive (π ^{m + 1} _GO ). Providing a first incentive applied to an industrial consumer and a second incentive applied to a service provider to the industrial consumer and the service provider, respectively;
Transmitting, by the industrial consumer, the industrial consumer load reduction amount updated based on a first incentive to the grid operator;
Transmitting a service provider load reduction amount updated by the service provider based on a second incentive to the grid operator;
A fourth step of the grid operator calculating a cost based on the updated grid operator incentive, industrial consumer load reduction, and service provider load reduction;
If the grid operator compares the calculated cost (C ^{m + 1} _GO ) with the cost (C ^m _GO ) calculated in m iterations, determine the calculated cost as an optimal cost and determine the updated grid operator incentive. Comparing the maximum value of the grid operator incentive to the maximum value of the grid operator incentive and determining the updated grid operator incentive as an optimal grid operator incentive;
And a sixth step of determining, by the grid operator, whether a difference between the calculated cost and the cost calculated in m repetitions is equal to or less than a predetermined value.
If the difference between the calculated cost and the cost calculated in m iterations is greater than a predetermined value, the first to fifth steps are repeated, and if the difference between the calculated cost and the cost calculated in m iterations is less than or equal to a predetermined value An incentive-based demand response method characterized by determining that the balance of the balance has been reached and utilizing the Staggerberg equilibrium grid operator incentive, industrial consumer load reduction, and service provider load reduction. The method of claim 1,
In the first step, the grid operator incentive is increased to the step size calculated based on the difference between the cost calculated at m iterations (C ^m _GO ) and the cost calculated at m -1 iterations (C ^m-1 _GO ). Incentive-based demand response method. The method of claim 2,
The step size (Δπ _GO ) is an incentive-based demand response method, characterized in that calculated by equation (29).

The method of claim 1,
Incentive-based demand response method characterized in that the industrial consumer load reduction in the second step is calculated by the equation (14).
The method of claim 1,
Incentive-based demand response method, characterized in that the service provider load reduction amount is calculated by Equation 20 in the third step.
The method of claim 1,
Incentive-based demand response method, characterized in that the cost is calculated by Equation 31 in the fourth step.
Incentive-based demand response method of demand response system composed of grid operators, industrial consumers and service providers,
The grid operator is responsible for incentive costs to be paid to industrial consumers according to industrial consumer load reductions, incentives to be paid to service providers according to service provider load reductions, industrial consumer load reductions and service provider load reductions from grid resource shortages. Incentive-based demand response method characterized in that the grid operator incentives are determined to minimize the procurement costs summed up the generation cost to produce the amount of resources subtracted. The method of claim 7, wherein
The industrial consumer determines the industrial consumer load reduction amount to maximize the sum of the incentive income according to the industrial consumer load reduction amount and the profit obtained by subtracting the load reduction amount from the resource amount to be used. Based demand response method. The method of claim 7, wherein
Wherein the service provider determines the service provider load reduction amount to maximize the value of the incentive income according to the service provider load reduction amount subtracted from the cost to be paid to the subscribing customer. In the grid response (GO) demand response management method for industrial consumers and service providers,
Set the initial value of the grid operator (GO) incentive rate, determine the first GO incentive rate to apply to the industrial consumer and the second GO incentive rate to apply to the service provider based on the initial value, and reduce the demand reduction of the industrial consumer An initial step of calculating the initial value of the procurement cost of the grid operator by receiving the initial value of the service provider and the initial value of the demand reduction amount of the service provider;
An update step of updating the grid operator incentive rate to calculate an update value of the first GO incentive rate and an update value of the second GO incentive rate and transmitting the update value to the industrial consumer and the service provider, respectively;
An update value receiving step of receiving an update value of the demand reduction amount of the industrial consumer and an update value of the demand reduction amount of the service provider;
A cost calculation step of calculating a procurement cost of the grid operator based on the update value of the grid operator incentive ratio, the update value of the demand reduction amount of the industrial consumer, and the update value of the demand reduction amount of the service provider;
And comparing the difference between the calculated procurement cost and the previously calculated procurement cost with a specific value. The method of claim 10,
If the difference value is greater than the specified value in the difference value comparison step, the grid operator demand response management method characterized in that it is repeated from the update step to the cost calculation step. The method of claim 10,
In the difference comparison step, if the difference is less than the specified value, it is determined that the Staeckelberg equilibrium is reached, and the grid operator is compensated for the demand response with the updated value of the grid operator ratio in the Staeckelberg equilibrium state. How to manage demand response.

Images