Load regulation and control strategy and system considering superposition of multiple markets

Abstract

The invention provides a load regulation and control strategy considering multiple market superposition, and relates to the technical field of power systems. The method comprises the following steps: S1, inputting load side resource feature data of multiple markets; s2, calculating indexes of the fluctuation rate, the peak-valley difference ratio, the discharge load support rate, the load transfer capability and the load interruption capability of each resource main body as feature vectors of each resource main body, constructing a feature vector set of each resource main body, and hierarchically aggregating the feature vectors in the feature vector set of all the resource main bodies by adopting a DBSCAN clustering algorithm; s3, on the basis of clustering results of all the feature vectors, in combination with constraint conditions, establishing a load side resource aggregation model with the lowest operation cost as an objective function; and S4, calling a commercial solver to solve the load side resource aggregation model to obtain an optimal load side resource hierarchical aggregation scheme. According to the invention, the problems of heterogeneity of each load resource, difference in response capability and low enthusiasm of users participating in demand response in the existing load management can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a load regulation strategy and system that considers the superposition of multiple markets. Background Art

[0002] With the rapid development of the power market and the widespread application of renewable energy, demand response (DR) as an effective load management strategy is attracting more and more attention. As a strategy to ensure the stable operation of the power system, demand response incentivizes and regulates the load based on price signals and incentive mechanisms. Among them, price-based demand response strategies include time-of-use electricity prices, peak electricity prices and real-time electricity prices. Incentive demand response can be divided into two types: plan-based and market-based.

[0003] Demand response not only helps balance electricity supply and demand and improve the reliability of the power grid, but also promotes efficient use of energy and reduces greenhouse gas emissions. In this context, the effective scheduling and management of load-side resources has become the key to achieving demand response goals.

[0004] However, traditional load management methods often face many challenges, such as the heterogeneity of various load resources, differences in response capabilities, and difficulty in user participation. Application number 202311076252.8 discloses a distributed control method and system for load-side resources of a virtual power plant. It builds a load-side dynamic frequency regulation model based on the operating characteristics of the load-side heterogeneous resources participating in demand response, realizes unified modeling of heterogeneous resources, and obtains the frequency regulation service price of the virtual power plant according to the historical operating parameters of the virtual power plant and the frequency regulation demand of the power grid; according to the frequency regulation service price of the virtual power plant, the optimal real-time regulation strategy and optimal bidding method of each load-side heterogeneous resource for the frequency regulation service price signal are solved in parallel, and the optimal bidding method is determined according to the optimal real-time regulation strategy, which can solve the problem of heterogeneity of various load resources to a certain extent, but it still has the following shortcomings: it does not solve the problem of differences in response capabilities of various load resources and low enthusiasm of users to participate in demand response in traditional load management methods. Summary of the invention

[0005] The technical problem to be solved by the present invention is how to simultaneously solve the problems of heterogeneity of various types of load resources, differences in response capabilities, and low enthusiasm of users to participate in demand response in load-side resource regulation under the superposition of multiple markets.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a load regulation strategy considering the superposition of multiple markets, comprising the following steps:

[0007] S1: Input multiple markets, including load-side resource characteristic data under the demand response market and the orderly electricity market;

[0008] S2: Based on the resource characteristic data on the load side, the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource entity are calculated as the characteristic vector of each resource entity, and the characteristic vector set of each resource entity is constructed. The DBSCAN clustering algorithm is used to hierarchically aggregate the characteristic vectors in the characteristic vector set of all resource entities;

[0009] S3: Based on the clustering results of all feature vectors and combined with constraints, a load-side resource aggregation model with the lowest operating cost as the objective function is established;

[0010] S4: calling a commercial solver to solve the load-side resource aggregation model to obtain an optimal load-side resource hierarchical aggregation solution under the superposition of multiple markets.

[0011] Preferably, the load-side resource characteristic data in step S1 includes distributed unit capacity, supplier capacity, elastic load capacity and resource incentive price;

[0012] The distributed unit capacity is: the maximum power generation capacity of all distributed units involved in power generation, including wind power, photovoltaics and micro-turbines; the supplier capacity is: the maximum power supply that each power supplier can provide; the flexible load capacity is: the ability and amplitude of load adjustment that users participating in demand response can achieve; the resource incentive price is: the market incentive price corresponding to distributed units, suppliers and flexible loads.

[0013] Preferably, in step S2,

[0014] Volatility index of resource entity i i∈I, where is the demand response reaching target time, T s is the response start time, T f To respond to the end moment;

[0015] Peak-to-valley ratio indicator i∈I, where P t i is the load of resource subject i at time t;

[0016] Discharge load support rate index i∈I, where is the discharge load capacity of resource subject i at time t, is the daily average rigid load of resource entity i, and the calculation formula is: is the rigid load of resource subject i at time t;

[0017] Load transfer capability index i∈I, where is the transferable load of resource subject i at time t, is the total load of resource subject i at time t;

[0018] Load interruption capability index i∈I, where is the interruptible load of resource subject i at time t, is the total load of resource entity i at time t.

[0019] Preferably, in step S2, the resource subject feature vector set constructed is X, and the resource subject feature vector set of resource subject i is

[0020] The present invention calculates the volatility and other indicators for each resource subject on the load side, constructs a feature vector set for each resource subject, classifies and manages the resources, and effectively solves the problem of heterogeneity of various resources in load-side resource management.

[0021] Preferably, in step S2, the specific process of hierarchically aggregating the feature vectors in the feature vector set of all resource subjects using the DBSCAN clustering algorithm is as follows:

[0022] S2031: Set the minimum sample number minPts and radius parameter ε of the DBSCAN clustering algorithm;

[0023] S2032: Take the feature vectors in the feature vector set of all resource entities as sample points, extract a sample point p, and group all sample points whose distance from the sample point is less than or equal to ε into a set;

[0024] S2033: Determine whether the sample point p is a core sample point. The specific determination process is as follows:

[0025] (1) With sample point p as the center, if there are more than MinPts sample points in the area with ε as the radius, that is, if |N ò (p)|≥MinPts,|N ò (p)| is the number of sample points in the ε-domain of sample point p, then sample point p is the core sample point;

[0026] (2) If the sample point p is not a core sample point, but is within the region with a radius of ε of other core sample points, then the sample point is a boundary sample point;

[0027] (3) If the sample point p is neither a core sample point nor a boundary sample point, then the sample point is a noise sample point;

[0028] S2034: if the sample point p is a core sample point, create a cluster of the core sample point p; with the core sample point p as the center, continuously propagate through density direct access until the density reachable boundary of the cluster, classify each sample point a in the ε-domain of the core sample point p as a core sample point, a boundary sample point or a noise sample point, and determine whether to add the sample point a or all the sample points in the sample point a and its ε-domain to the cluster of the core sample point p;

[0029] If the sample point p is not a core sample point, then re-extract the sample point from all resource subject feature vector sets to determine whether it is a core sample point;

[0030] S2035: repeating the above steps S2032-S2034 until all noise sample points are excluded, all sample points in the resource main feature vector set are added to each clustering cluster, and the clustering results of all feature vectors are obtained;

[0031] The present invention implements a hierarchical aggregation strategy based on the load characteristics and response capabilities of different resource entities, thereby solving the problem of differences in response capabilities of various resources.

[0032] Preferably, in step S2032, the distance between the sample point p and other sample points a is d(p,a), and the comprehensive distance d(X i ,X j ) characterization, d(X i ,X j )=λ 1 EU i,j +λ 2 (1-r i,j ), λ 1 +λ 2 =1; where λ 1 , 2 are the weight coefficients of Euclidean distance and Pearson distance, EU i,j is the Euclidean distance between resource entities i and j, r i,j is the Pearson similarity between resource subjects i and j,

[0033] Preferably, in step S2034, the specific process of determining whether to add sample point a or all sample points in sample point a and its ε-domain to the clustering cluster of core sample point p is as follows: if a is a boundary sample point, a is added to the clustering cluster of core sample point p; if a is a core sample point, a and all sample points in its ε-domain are added to the clustering cluster of core sample point p; if a is a noise sample point, a is not added and is excluded.

[0034] Preferably, in step S3, the objective function is:

[0035]

[0036] Where C is the operating cost; T is a power generation cycle; P t DG,m is the power generation of the mth generator at time t; is the unit operating cost of the mth generator at time t; P t NSP is the non-power supply at time t; is the unit cost of non-power supply at time t; is the power supply of the nth supplier at time t; is the power supply cost of the nth supplier at time t; P t IDR,q The energy for the qth user to participate in the incentive demand response at time t; is the unit incentive for the qth user to participate in the incentive demand response at time t; are the initial power and load reduction of the qth user participating in real-time pricing at time t, respectively; They are the initial compensation cost and additional compensation cost of the qth user participating in real-time pricing at time t respectively.

[0037] Preferably, in step S3, the constraint conditions include:

[0038]

[0039]

[0040] in, is the initial power consumption of the qth user at time t; For the mth supplier The maximum value of the power provided by the nth supplier; is the maximum outsourced electricity from the nth supplier; is the maximum incentive demand response power of the qth user; The qth user participates in real-time pricing of maximum power; The maximum unit price for the qth user to participate in real-time pricing;

[0041] The present invention establishes a load-side resource aggregation model with the lowest operating cost as the objective function, which can mobilize the enthusiasm of users to participate in demand response.

[0042] Corresponding to the above method, the present invention also provides a load control system considering the superposition of multiple markets, including the following modules:

[0043] Input module: used to input multiple markets, including load-side resource characteristic data in demand response market and orderly electricity market;

[0044] Hierarchical aggregation module: It is used to calculate the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource entity based on the resource characteristic data on the load side as the characteristic vector of each resource entity, construct the characteristic vector set of each resource entity, and use the DBSCAN clustering algorithm to hierarchically aggregate the characteristic vectors in the characteristic vector set of all resource entities;

[0045] Establish resource aggregation model module: It is used to establish a load-side resource aggregation model with the lowest operating cost as the objective function based on the clustering results of all feature vectors and combined with constraints;

[0046] Obtaining the optimal solution module: used to call the commercial solver to solve the load-side resource aggregation model and obtain the optimal load-side resource layered aggregation solution under the superposition of multiple markets.

[0047] Compared with the prior art, the advantages of the present invention are:

[0048] (1) Classified management of load-side resources can effectively solve the problem of heterogeneity among various types of load resources. Based on the load characteristics and response capabilities of different resource entities, a hierarchical aggregation strategy is implemented to solve the problem of differences in the response capabilities of load resources;

[0049] (2) Through scientific data analysis and intelligent decision support, a load-side resource aggregation model with the lowest operating cost as the objective function is established, which can increase users' enthusiasm for participating in demand response, make user behavior more controllable, and provide effective guarantees for the sustainable management and operation of the power system;

[0050] (3) The present invention develops a load-side resource hierarchical aggregation technology for demand response, which can obtain the optimal load control strategy when multiple markets are superimposed. It can not only achieve maximum economic and environmental benefits in the power market and promote the construction of smart grids, but also provide strong support for achieving low-carbon economy and sustainable development goals. It can improve the intelligence level of the power system, optimize the allocation of energy resources, and realize efficient, economical and environmentally friendly operation of the power market. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of Example 1 of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Example 1

[0054] like Figure 1 As shown, it is a flow chart of this embodiment 1. The present invention provides a load regulation strategy considering the superposition of multiple markets, including the following steps:

[0055] S1: Input the load-side resource characteristic data such as the distributed unit capacity, supplier capacity, flexible load capacity and resource incentive price of various markets, such as the demand response market and the orderly power consumption market;

[0056] Specifically, the distributed unit capacity is: the maximum power generation capacity of all distributed units involved in power generation, such as wind power, photovoltaics, micro-turbines, etc.; the supplier capacity is: the maximum power supply power that each power supplier can provide; the flexible load capacity is: the ability and amplitude of load adjustment that users participating in demand response can achieve, including load reduction and load transfer; the resource incentive price is: the market incentive price corresponding to distributed units, suppliers and flexible loads.

[0057] S2: Based on the resource characteristic data on the load side, the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource subject, i.e., each user, generator and supplier, are calculated as the characteristic vector of each resource subject, and the characteristic vector set of each resource subject is constructed. The DBSCAN clustering algorithm is used to hierarchically aggregate the characteristic vectors in the characteristic vector set of all resource subjects, which specifically includes the following steps:

[0058] S201: Taking resource subject i as an example, calculate the volatility index of resource subject i i∈I, where is the demand response reaching target time, T s is the response start time, T f To respond to the end moment;

[0059] Peak-to-valley ratio indicator i∈I, where P t i is the load of resource subject i at time t;

[0060] Discharge load support rate index i∈I, where is the discharge load capacity of resource subject i at time t, is the daily average rigid load of resource entity i, and the calculation formula is: is the rigid load of resource subject i at time t;

[0061] Load transfer capability index i∈I, where is the transferable load of resource subject i at time t, is the total load of resource subject i at time t;

[0062] Load interruption capability index i∈I, where is the interruptible load of resource subject i at time t, is the total load of resource entity i at time t.

[0063] S202: Calculate and obtain the volatility, peak-to-valley ratio, discharge load support rate, load transfer capability, and load interruption capability indicators of each resource subject as the feature vector of each resource subject, and construct the feature vector set X of each resource subject on the load side. The resource subject feature vector set of resource subject i is

[0064] S203: hierarchically aggregate the resource subject feature vector sets of all resource subjects using the DBSCAN clustering algorithm. The specific process is as follows:

[0065] S2031: Set the minimum sample number minPts and radius parameter ε of the DBSCAN clustering algorithm;

[0066] S2032: Take the feature vectors in the feature vector set of all resource subjects as sample points, extract a sample point p, and classify all sample points whose distance from the sample point is less than or equal to ε into a set; the distance between the sample point p and other sample points a is d(p,a), and this embodiment uses the comprehensive distance d(X) after Euclidean distance and Pearson similarity weighting. i ,X j ) characterization, d(X i ,X j )=λ 1 EU i,j +λ 2 (1-r i,j ), λ 1 +λ 2 =1; where λ 1 , 2 are the weight coefficients of Euclidean distance and Pearson distance respectively; EU i,j is the Euclidean distance between the feature vectors of resource subjects i and j, ri,j is the Pearson similarity between the feature vectors of resource subjects i and j,

[0067] S2033: Determine whether the sample point p is a core sample point. The specific determination process is as follows:

[0068] (1) With sample point p as the center, if there are more than MinPts sample points in the area with ε as the radius, that is, if |N ò (p)|≥MinPts, then the sample point p is the core sample point, |N ò (p)| is the number of sample points in the ε-domain of sample point p;

[0069] (2) If the sample point p is not a core sample point, but is within the region with a radius of ε of other core sample points, then the sample point is a boundary sample point;

[0070] (3) If the sample point p is neither a core sample point nor a boundary sample point, then the sample point is a noise sample point.

[0071] S2034: if the sample point p is a core sample point, create a cluster of the core sample point p; with the core sample point p as the center, continuously propagate through density direct access until the density reachable boundary of the cluster is reached, classify each sample point a in the ε-domain of the core sample point p as a core sample point, a boundary sample point or a noise sample point, and judge whether to add the sample point a or the sample point a and all the sample points in its ε-domain to the cluster of the core sample point p. The specific judgment process is: if a is a boundary sample point, add a to the cluster of the core sample point p; if a is a core sample point, add a and all the sample points in its ε-domain to the cluster of the core sample point p; if a is neither a boundary sample point nor a core sample point, mark a as a noise sample point and exclude it;

[0072] Each sample point in the ε-domain of the core sample point p is classified and judged until all boundary sample points in the dense area and all sample points in the core sample point p and its ε-domain are added to the cluster of the core sample point p, and all noise sample points are excluded, and the cluster of the core sample point p is obtained;

[0073] If the sample point p is not a core sample point, then re-extract the sample point from the resource main feature vector set to determine whether it is a core sample point;

[0074] S2035: Repeat the above steps S2032-S2034 until all noise sample points are excluded, and the feature vectors in the remaining resource subject feature vector set, that is, the remaining sample points are added to the clustering clusters of each core sample point to obtain the clustering results of all resource subject feature vectors on the load side.

[0075] S3: Based on the clustering results of the resource subject feature vector set and considering the constraints, a load-side resource aggregation model with the lowest operating cost as the objective function is established;

[0076] The operating costs include distributed units and user elastic loads, including the power generation and cost of distributed units throughout the entire cycle, the elastic load of users participating in demand response and the incentive compensation they deserve;

[0077] The objective function is:

[0078]

[0079] Where C is the operating cost; T is a power generation cycle; P t DG,m is the power generation of the mth generator at time t; is the unit operating cost of the mth generator at time t; P t NSP is the non-power supply at time t; is the unit cost of non-power supply at time t; is the power supply of the nth supplier at time t; is the power supply cost of the nth supplier at time t; P t IDR,q The energy of the qth user participating in the incentive demand response at time t; c t IDR,q is the unit incentive for the qth user to participate in the incentive demand response at time t; are the initial power and load reduction of the qth user participating in real-time pricing at time t, respectively; They are the initial compensation cost and additional compensation cost of the qth user participating in real-time pricing at time t, respectively;

[0080] The constraints include energy balance and external output restrictions during the entire demand response event, specifically, the power generation restrictions of distributed units, the external purchase amount restrictions, the maximum load reduction under the elastic load range and real-time pricing constraints, and the cost of the pricing scheme, etc., specifically:

[0081]

[0082] in, is the initial power consumption of the qth user at time t; For the mth supplier The maximum value of the power provided by the nth supplier; is the maximum outsourced electricity from the nth supplier; is the maximum incentive demand response power of the qth user; The qth user participates in real-time pricing of maximum power; The maximum unit price for the qth user to participate in real-time pricing.

[0083] S4: Call the commercial solver to solve the load-side resource aggregation model in step S3 to obtain the optimal load-side resource hierarchical aggregation solution under the superposition of multiple markets.

[0084] This embodiment can effectively solve the problem of heterogeneity among various load resources in the prior art by classifying and managing load-side resources. By calculating indicators such as volatility of each resource entity and constructing a feature vector to form a feature vector set, different scheduling schemes can be considered based on the load characteristics and response capabilities of different resource entities, thereby solving the problem of differences in response capabilities among various load resources. In resource aggregation, the DBSCAN algorithm based on the comprehensive similarity measurement of Euclidean distance and Pearson correlation coefficient is used to hierarchically aggregate resource entities, making the aggregation results more accurate. Through scientific data analysis and intelligent decision support, the enthusiasm of users to participate in demand response can be improved, user behavior can be made controllable, and effective guarantees can be provided for the sustainable management and operation of the power system.

[0085] Example 2

[0086] Corresponding to the first embodiment of the present invention, the second embodiment of the present invention further provides a load control system considering the superposition of multiple markets, including the following modules:

[0087] Input module: used to input load-side resource characteristic data such as distributed unit capacity, supplier capacity, flexible load capacity and resource incentive price in various markets, such as demand response market and orderly power consumption market;

[0088] Specifically, the distributed unit capacity is: the maximum power generation capacity of all distributed units involved in power generation, such as wind power, photovoltaics, micro-turbines, etc.; the supplier capacity is: the maximum power supply power that each power supplier can provide; the flexible load capacity is: the ability and amplitude of load adjustment that users participating in demand response can achieve, including load reduction and load transfer; the resource incentive price is: the market incentive price corresponding to distributed units, suppliers and flexible loads.

[0089] Hierarchical aggregation module: It is used to calculate the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource subject based on the resource characteristic data on the load side as the characteristic vector of each resource subject, construct the characteristic vector set of each resource subject, and hierarchically aggregate the characteristic vectors in the characteristic vector set of all resource subjects using the DBSCAN clustering algorithm. Specifically, it includes the following units:

[0090] Calculation unit: used to calculate the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource entity; taking resource entity i as an example, calculate the volatility index of resource entity i i∈I, where is the demand response reaching target time, T s is the response start time, T f To respond to the end moment;

[0091] Peak-to-valley ratio indicator i∈I, where P t i is the load of resource subject i at time t;

[0092] Discharge load support rate index i∈I, where is the discharge load capacity of resource subject i at time t, is the daily average rigid load of resource entity i, and the calculation formula is: is the rigid load of resource subject i at time t;

[0093] Load transfer capability index i∈I, where is the transferable load of resource subject i at time t, is the total load of resource subject i at time t;

[0094] Load interruption capability index i∈I, where is the interruptible load of resource subject i at time t, is the total load of resource entity i at time t.

[0095] Construct resource subject feature vector set unit: used to use the volatility, peak-to-valley ratio, load support rate, load transfer capacity and load interruption capacity indicators of each resource subject calculated by the calculation unit as the feature vector of each resource subject, and construct the resource subject feature vector set X on the load side. The resource subject feature vector set of resource subject i is

[0096] Hierarchical Aggregate Feature Vector Unit: Used to perform the following steps:

[0097] First, set the minimum sample size minPts and radius parameter ε of the DBSCAN clustering algorithm;

[0098] Then extract a sample point p from all resource main feature vector sets, and classify all sample points whose distance from the sample point is less than or equal to ε into a set. The distance between the sample point p and other sample points a is d(p,a). In this embodiment, the comprehensive distance d(X i ,X j ) characterization, d(X i ,X j )=λ 1 EU i,j +λ 2 (1-r i,j ), λ 1 +λ 2 =1; where λ 1 , 2 are the weight coefficients of Euclidean distance and Pearson distance respectively; EU i,j is the Euclidean distance between the feature vectors of resource subjects i and j, r i,j is the Pearson similarity between the feature vectors of resource subjects i and j,

[0099] Determine whether the sample point p is a core sample point. The specific judgment process is as follows:

[0100] (1) With sample point p as the center, if there are more than MinPts sample points in the area with ε as the radius, that is, if |N ò (p)|≥MinPts, then the sample point p is the core sample point, |N ò (p)| is the number of sample points in the ε-domain of sample point p;

[0101] (2) If the sample point p is not a core sample point, but is within the region with a radius of ε of other core sample points, then the sample point is a boundary sample point;

[0102] (3) If the sample point p is neither a core sample point nor a boundary sample point, then the sample point is a noise sample point.

[0103] S2034: if the sample point p is a core sample point, create a cluster of the core sample point p; with the core sample point p as the center, continuously propagate through density direct access until the density reachable boundary of the cluster is reached, classify each sample point a in the ε-domain of the core sample point p as a core sample point, a boundary sample point or a noise sample point, and judge whether to add the sample point a or the sample point a and all the sample points in its ε-domain to the cluster of the core sample point p. The specific judgment process is: if a is a boundary sample point, add a to the cluster of the core sample point p; if a is a core sample point, add a and all the sample points in its ε-domain to the cluster of the core sample point p; if a is neither a boundary sample point nor a core sample point, mark a as a noise sample point and exclude it;

[0104] Each sample point in the ε-domain of the core sample point p is classified and judged until all boundary sample points in the dense area and all sample points in the core sample point p and its ε-domain are added to the cluster of the core sample point p, and all noise sample points are excluded, and the cluster of the core sample point p is obtained;

[0105] If the sample point p is not a core sample point, then re-extract the sample point from the resource main feature vector set to determine whether it is a core sample point;

[0106] The hierarchical aggregation feature vector unit is repeatedly operated until all noise sample points are excluded, and the feature vectors in the remaining resource subject feature vector set, that is, the remaining sample points are added to the clustering clusters of each core sample point to obtain the clustering results of all resource subject feature vectors on the load side.

[0107] Establish resource aggregation model module: It is used to establish a load-side resource aggregation model with the lowest operating cost as the objective function based on the clustering results of all feature vectors and combined with constraints. It specifically includes the following units:

[0108] Objective function unit: used to execute the objective function minC,

[0109]

[0110] Where C is the operating cost; T is a power generation cycle; P t DG,m is the power generation of the mth generator at time t; is the unit operating cost of the mth generator at time t; P t NSP is the non-power supply at time t; is the unit cost of non-power supply at time t; is the power supply of the nth supplier at time t; is the power supply cost of the nth supplier at time t; Pt IDR,q The energy for the qth user to participate in the incentive demand response at time t; is the unit incentive for the qth user to participate in the incentive demand response at time t; are the initial power and load reduction of the qth user participating in real-time pricing at time t, respectively; They are the initial compensation cost and additional compensation cost of the qth user participating in real-time pricing at time t, respectively;

[0111] Constraint unit: used to provide constraints, including:

[0112]

[0113] in, is the initial power consumption of the qth user at time t; For the mth supplier The maximum value of the power provided by the nth supplier; is the maximum outsourced electricity from the nth supplier; is the maximum incentive demand response power of the qth user; The qth user participates in real-time pricing of maximum power; The maximum unit price for the qth user to participate in real-time pricing.

[0114] Obtaining the optimal solution module: calling the commercial solver to solve the load-side resource aggregation model, and obtaining the optimal load-side resource layered aggregation solution under the superposition of multiple markets.

[0115] In this embodiment, a load control system considering the superposition of multiple markets is provided. First, the hierarchical aggregation module calculates the volatility and other indicators of each resource subject based on the load-side resource characteristic data of each market in the input module, constructs the characteristic vector of each resource subject on the load side, classifies and manages the load-side resources, and solves the problem of heterogeneity between various load resources. The hierarchical aggregation module also hierarchically aggregates the resource subjects using the DBSCAN clustering algorithm based on the comprehensive similarity metric, implements a hierarchical aggregation strategy based on the load characteristics and response capabilities of different resource subjects, and effectively solves the problem of differences in response capabilities between various resource subjects. Then, a load-side resource aggregation model with the lowest operating cost as the objective function is established by establishing a resource aggregation model module. Through scientific data analysis and intelligent decision support, the enthusiasm of users to participate in demand response can be improved, user behavior can be controlled, and effective guarantee can be provided for the sustainable management and operation of the power system. Finally, the optimal solution module is used to obtain the optimal load-side resource hierarchical aggregation solution under multiple markets, which effectively improves the intelligence level of the power system, optimizes the allocation of energy resources, and realizes the efficient, economical and environmentally friendly operation of the power market.

[0116] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A load regulation strategy considering multiple market superpositions, characterized in that: The following steps are involved: S1: Input multiple markets, including load-side resource characteristic data under the demand response market and the orderly electricity market; S2: Based on the resource characteristic data on the load side, the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource entity are calculated as the characteristic vector of each resource entity, and the characteristic vector set of each resource entity is constructed. The DBSCAN clustering algorithm is used to hierarchically aggregate the characteristic vectors in the characteristic vector set of all resource entities; S3: Based on the clustering results of all feature vectors and combined with constraints, a load-side resource aggregation model with the lowest operating cost as the objective function is established; S4: calling a commercial solver to solve the load-side resource aggregation model to obtain an optimal load-side resource hierarchical aggregation solution under the superposition of multiple markets.

2. A load regulation strategy considering multiple market superposition according to claim 1, characterized in that: The load-side resource characteristic data in step S1 includes distributed unit capacity, supplier capacity, elastic load capacity and resource incentive price; The distributed unit capacity is: the maximum power generation capacity of all distributed units involved in power generation, including wind power, photovoltaics and micro-turbines; the supplier capacity is: the maximum power supply that each power supplier can provide; the flexible load capacity is: the ability and amplitude of load adjustment that users participating in demand response can achieve; the resource incentive price is: the market incentive price corresponding to distributed units, suppliers and flexible loads.

3. A load regulation strategy considering multiple market superpositions according to claim 1, characterized in that: In step S2, Volatility index of resource entity i in, is the demand response reaching target time, T s is the response start time, T f To respond to the end moment; Peak-to-valley ratio indicator Among them, P t i is the load of resource subject i at time t; Discharge load support rate index in, is the discharge load capacity of resource subject i at time t, is the daily average rigid load of resource entity i, and the calculation formula is: is the rigid load of resource subject i at time t; Load transfer capability index in, is the transferable load of resource subject i at time t, is the total load of resource subject i at time t; Load interruption capability index in, is the interruptible load of resource subject i at time t, is the total load of resource entity i at time t.

4. A load regulation strategy considering multiple market superpositions according to claim 3, characterized in that: In step S2, the resource subject feature vector set constructed is X, and the resource subject feature vector set of resource subject i is 5. A load regulation strategy considering multiple market superpositions according to claim 4, characterized in that: In step S2, the specific process of hierarchically aggregating the feature vectors in the feature vector set of all resource entities using the DBSCAN clustering algorithm is as follows: S2031: Set the minimum sample number minPts and radius parameter ε of the DBSCAN clustering algorithm; S2032: Take the feature vectors in the feature vector set of all resource entities as sample points, extract a sample point p, and group all sample points whose distance from the sample point is less than or equal to ε into a set; S2033: Determine whether the sample point p is a core sample point. The specific determination process is as follows: (1) With sample point p as the center, if there are more than MinPts sample points in the area with ε as the radius, that is, if |N ò (p)|≥MinPts,|N ò (p)| is the number of sample points in the ε-domain of sample point p, then sample point p is the core sample point; (2) If the sample point p is not a core sample point, but is within the region with a radius of ε of other core sample points, then the sample point is a boundary sample point; (3) If the sample point p is neither a core sample point nor a boundary sample point, then the sample point is a noise sample point; S2034: If the sample point p is a core sample point, create a cluster of the core sample point p; Taking the core sample point p as the center, the density is directly propagated until the density reachable boundary of the cluster is reached, and each sample point a in the ε-domain of the core sample point p is classified as a core sample point, a boundary sample point or a noise sample point, and it is determined whether to add the sample point a or all the sample points in the sample point a and its ε-domain to the cluster of the core sample point p; If the sample point p is not a core sample point, then re-extract the sample point from all resource subject feature vector sets to determine whether it is a core sample point; S2035: Repeat the above steps S2032-S2034 until all noise sample points are excluded, and all sample points in the resource main feature vector set are added to each clustering cluster to obtain the clustering results of all feature vectors.

6. A load regulation strategy considering multiple market superpositions according to claim 5, characterized in that: In step S2032, the distance between sample point p and other sample points a is d(p,a), and the comprehensive distance d(X) weighted by Euclidean distance and Pearson similarity is i ,X j ) characterization, d(X i ,X j )=λ1EU i,j +λ2(1-r i,j ), λ1+λ2=1; where λ1 and λ2 are the weight coefficients of Euclidean distance and Pearson distance respectively, EU i,j is the Euclidean distance between resource entities i and j, r i,j is the Pearson similarity between resource subjects i and j, 7. A load regulation strategy considering multiple market superpositions according to claim 5, characterized in that: In step S2034, the specific process of determining whether to add sample point a to the cluster of core sample point p is as follows: if a is a boundary sample point, a is added to the cluster of core sample point p; if a is a core sample point, a and all sample points in its ε-domain are added to the cluster of core sample point p; if a is a noise sample point, a is not added and is excluded.

8. The load regulation strategy considering multiple market superposition according to claim 1 is characterized in that: In step S3, the objective function is: Where C is the operating cost; T is a power generation cycle; P t DG,m is the power generation of the mth generator at time t; is the unit operating cost of the mth generator at time t; P t NSP is the non-power supply at time t; is the unit cost of non-power supply at time t; is the power supply of the nth supplier at time t; is the power supply cost of the nth supplier at time t; P t IDR,q The energy for the qth user to participate in the incentive demand response at time t; is the unit incentive for the qth user to participate in the incentive demand response at time t; are the initial power and load reduction of the qth user participating in real-time pricing at time t, respectively; They are the initial compensation cost and additional compensation cost of the qth user participating in real-time pricing at time t respectively.

9. A load regulation strategy considering multiple market superpositions according to claim 8, characterized in that: In step S3, the constraints include: in, is the initial power consumption of the qth user at time t; For the mth supplier The maximum value of the power provided by the nth supplier; is the maximum outsourced electricity from the nth supplier; is the maximum incentive demand response power of the qth user; The qth user participates in real-time pricing of maximum power; The maximum unit price for the qth user to participate in real-time pricing.

10. A load control system considering multiple market superpositions, characterized in that: Includes the following modules: Input module: used to input multiple markets, including load-side resource characteristic data in demand response market and orderly electricity market; Hierarchical aggregation module: It is used to calculate the volatility, peak-to-valley ratio, discharge load support rate, load transfer capacity and load interruption capacity of each resource entity based on the resource characteristic data on the load side as the characteristic vector of each resource entity, construct the characteristic vector set of each resource entity, and use the DBSCAN clustering algorithm to hierarchically aggregate the characteristic vectors in the characteristic vector set of all resource entities; Establish resource aggregation model module: It is used to establish a load-side resource aggregation model with the lowest operating cost as the objective function based on the clustering results of all feature vectors and combined with constraints; Obtaining the optimal solution module: used to call the commercial solver to solve the load-side resource aggregation model and obtain the optimal load-side resource layered aggregation solution under the superposition of multiple markets.

Images