A multi-dimensional dispatching method for smart energy systems serving the needs of upper power grids

Abstract

The present invention discloses a multidimensional dispatching method for a smart energy system that serves the needs of a superior power grid, which belongs to the field of power systems. First, a multidimensional market dispatching model for a smart energy system that takes capacity prices into account is constructed, with the minimum dispatching cost as the objective function, and the power deviation and carbon emission deviation as constraints; then, an induced dispatching model for a smart energy system based on meta-learning is constructed to improve the effectiveness of dispatching; further, a load-side resource response characteristic curve analogous to the coal consumption characteristics of a generator is formulated to provide a dispatching basis for the dispatching of load-side resources by the smart energy system; finally, a multidimensional dispatching model for a smart energy system that serves the needs of a superior power grid is constructed, and multidimensional dispatching such as power, low carbon, and cleanliness is performed to serve the power dispatching needs, carbon emission dispatching needs, and pollutant emission dispatching needs of the superior power grid. The present invention can realize multidimensional dispatching of smart energy systems, meet the multidimensional dispatching needs of the superior power grid such as economy, low carbon, and cleanliness, and improve the regulation capability and dispatching flexibility of the system.

Description

Technical Field

[0001] The present invention belongs to the field of power systems, and specifically relates to a multi-dimensional dispatching method for a smart energy system serving the needs of a superior power grid. Background Art

[0002] Multi-dimensional dispatching of loads (economic, low-carbon, and clean) is to meet the multiple challenges faced by modern power systems and improve the overall efficiency and sustainability of power systems. The traditional power grid dispatching method has the following problems: 1) The dispatching target is too single. Traditional dispatching is usually based on economic efficiency and may ignore environmental and carbon emission issues; it fails to fully consider environmental protection and low-carbon goals, resulting in the frequent use of high-pollution, high-carbon emission power sources. 2) Insufficient utilization of renewable energy. Renewable energy such as wind and solar energy has large volatility and intermittency, and traditional dispatching methods are difficult to fully utilize their potential; traditional dispatching may give priority to fossil fuel power generation with high stability rather than renewable energy. 3) Insufficient system flexibility. Traditional power grid dispatching is slow to respond to load changes and it is difficult to flexibly adjust power output; there is a lack of effective dispatching and management of demand-side resources, such as demand response and load management. 4) Environmental and health costs are not considered. Traditional dispatching does not consider the external environmental and health costs generated during the power generation process, such as the impact of air pollution on health. In summary, a multi-dimensional dispatching method for smart energy systems that serves the needs of the upper power grid is needed to meet the multi-dimensional dispatching needs of the upper power grid. Summary of the invention

[0003] Based on the above shortcomings, the present invention proposes a multi-dimensional dispatching method for a smart energy system that serves the needs of the superior power grid, meets the economic, low-carbon and clean dispatching needs of the superior power grid, provides technical support for the load dispatching of the smart energy system, and conforms to the development and dispatching needs of the new power system.

[0004] The technical solution adopted by the present invention is as follows: a multi-dimensional dispatching method of a smart energy system serving the needs of a superior power grid, comprising the following steps:

[0005] Step 1: Construct a multi-dimensional market dispatch model for smart energy systems taking capacity prices into account: Taking the minimization of the dispatch cost of the smart energy system as the objective function, including the electricity market clearing cost, the carbon trading market clearing cost, the actual electricity operation and clearing deviation cost, the actual carbon emission and clearing deviation cost, the electricity capacity cost, and the carbon capacity cost, taking the smart energy system electricity deviation and carbon emission deviation as constraints, realize the multi-dimensional market dispatch of the smart energy system electricity dispatch and carbon dispatch, reduce the operation cost of the smart energy system, and reduce carbon emissions;

[0006] Step 2: Construct an induced dispatch model of a smart energy system based on meta-learning, issue different incentives to different users in the smart energy system, induce users to participate in grid dispatch, and each user decides to respond to grid power and capacity according to the incentives, so that the dispatch center can grasp the actual response capability of the smart energy system under different incentives: including power, carbon emissions and pollutant emissions; the induced dispatch model of a smart energy system based on meta-learning includes a task feature extractor, a meta-learner, a task adapter and a model predictor. By training different load-side resource incentives and response powers, a meta-learning model is constructed to accurately predict the response prediction functions of power, carbon emissions and pollutant emissions of different users under different incentive inductions, providing support for smart energy system dispatch;

[0007] Step 3: Formulate a load-side resource response characteristic curve analogous to the coal consumption characteristics of the generator: Based on the power, carbon emission, and pollutant emission response prediction functions given by the meta-learning smart energy system induced scheduling method, a load-side resource response characteristic curve based on the analogy of the coal consumption characteristics of the generator is constructed to provide a scheduling basis for the smart energy system to schedule load-side resources, and give priority to users with large response power, small response carbon emissions, and small response pollutant emissions under the incentive of the scheduling unit;

[0008] Step 4. Construct a multi-dimensional dispatching model for the smart energy system that serves the needs of the upper-level power grid: take the minimum total incentive, minimum total power deviation, minimum total carbon emission deviation, and minimum total pollutant emission of the smart energy system as the objective function, and take the total incentive cost constraint, total power deviation constraint, total carbon emission deviation constraint, and total pollutant emission constraint as constraints; by setting different dispatching factors, carry out dispatching in different dimensions with emphasis, thereby realizing multi-dimensional dispatching of the smart energy system.

[0009] Furthermore, in step 1, the objective function is to minimize the dispatching cost of the smart energy system. The objective function is shown in formula (1), and the constraint condition is shown in formula (2).

[0010] miny=y1+y2+y3+y4+y5+y6

[0011]

[0012] Where: y1 is the clearing cost of the smart energy system in the electricity market; y2 is the clearing cost of the smart energy system in the carbon trading market; y3 is the actual electricity operation and clearing deviation cost of the smart energy system; y4 is the actual carbon emission and clearing deviation cost of the smart energy system; y5 is the electricity capacity cost of the smart energy system; y6 is the carbon capacity cost of the smart energy system; M is the number of electricity users participating in the operation of the smart energy system; N is the number of carbon trading users participating in the operation of the smart energy system; k D,i is the electricity trading coefficient of user i; k C,jis the carbon trading coefficient of user j; a SD,i is the electricity clearing price of user i; a SC,j is the carbon trading clearing price of user j; a FD Penalize price for power deviation; a FC Penalize prices for carbon trading deviations; a RD,i is the electricity capacity price of user i; a RC,j is the carbon capacity price of user j; P S2,i (t) is the power clearing power of user i at time t; C S2,j (t) is the carbon clearance rate of user j at time t; t1 and t2 are the start and end times of market clearing respectively; P Sq (t) is the actual power operation and clearing deviation of the smart energy system at time t; C Sq (t) The actual carbon emission and clearing deviation of the smart energy system at time t; P S1 (t) is the actual power of the smart energy system at time t; C S1 (t) is the actual carbon emission rate of the smart energy system at time t; ΔP S is the power deviation threshold of the smart energy system; ΔC S It is the carbon emission deviation threshold of the smart energy system.

[0013] Furthermore, in step 2, the meta-learning process is divided into two stages: 1) meta-training stage, for each task in the task set, first use the task feature extractor to extract the task features, then use the meta-learner to learn the common patterns and rules between tasks, and finally, use the task adapter to quickly adapt each task to obtain the adapted model parameters; 2) meta-testing stage, given the task features of a new task and the parameters output by the meta-learner, use the task adapter to quickly adapt the new task to obtain the adapted model parameters, and then use the model predictor to predict the new task;

[0014] The task feature extractor is shown in formula (3), based on which the response features of different users under different incentives are extracted.

[0015] Z D,i =f D,i (a XD,i ,P S3,i ) (3)

[0016] Where: f D,i () is the power response feature extraction function of user i; Z D,i is the power response characteristic of user i; a XD,i is the power incentive of user i; P S3,i is the power response power of user i;

[0017] The meta-learner is shown in formula (4), based on which the law between the power characteristics of different users and the actual power response power is extracted.

[0018]

[0019] Where: g D,i () is the power response law extraction function of user i; is the output of the meta-learner of user i; it represents the law between power characteristics and actual power response power; Z D,i is the power response characteristic of user i; P S3,i is the power response power of user i;

[0020] The task adapter is shown in formula (5), based on which the regularity function is modified to improve the adaptability of the model to new samples.

[0021]

[0022] Where: h D,i () is the power response adaptation function of user i; θ D,i is the model parameter after adaptation by user i; is the output of the meta-learner of user i; it represents the law between power characteristics and actual power response power; Z D,i is the power response characteristic of user i;

[0023] The model predictor is shown in formula (6), which constructs a user power response prediction function model to predict the actual power response power of different users under a given incentive;

[0024] θ D,i =f YD,i (a XD,i ) (6)

[0025] Where: f YD,i () is the power response prediction function model of user i, θ D,i is the power response power of user i output by the model, a XD,i Power incentive for user i;

[0026] A carbon emission response prediction function model of user i is constructed, as shown in formula (7), which predicts the user's carbon emission response under different incentives;

[0027] θ C,i =f YC,i (a XC,i ) (7)

[0028] Where: f YC,i () is the carbon emission response prediction function of user i; θ C,iis the carbon emission of user i output by the model, a XC,i Low-carbon incentive for user i;

[0029] The pollutant emission response prediction function model of user i is constructed, as shown in formula (8), which predicts the user's pollutant emission response under different incentives;

[0030] θ W,i =f YW,i (a XW,i ) (8)

[0031] Where: f YW,i () is the pollutant emission response prediction function of user i; θ W,i is the pollutant emission of user i output by the model, a XW,i Cleaning incentive for user i.

[0032] Furthermore, in step 3, the load-side resources are scheduled based on the load-side resource response characteristic curve analogous to the coal consumption characteristics of the generator, and users with large response power, small response carbon emissions, and small response pollutant emissions under unit incentives are prioritized, as shown in formula (9). If the response power of user i under the unit price is higher than that of user j, user i is prioritized;

[0033]

[0034] Where: Δθ D,i and Δθ D,j are the power changes of users i and j respectively; Δa XD,i and Δa XD,j are the incentive changes of users i and j respectively;

[0035] As shown in formula (10), if the carbon emissions of user i under the unit price are lower than those of user j, user i will be called first;

[0036]

[0037] Where: Δθ C,i and Δθ C,j are the carbon emission changes of users i and j respectively; Δa XC,i and Δa XC,j are the incentive changes of users i and j respectively;

[0038] As shown in formula (11), if the pollutant emission of user i under the unit price is lower than that of user j, user i will be called first;

[0039]

[0040] Where: Δθ W,i and Δθ W,jare the changes in pollutant emissions of users i and j respectively; Δa XW,i and Δa XW,j are the incentive changes of users i and j respectively.

[0041] Furthermore, in step 4, the objective function is shown in formula (12), including the minimum total incentive of the smart energy system, the minimum total power deviation, the minimum total carbon emission deviation, and the minimum total pollutant emission;

[0042] min k D (a XD +Δε D )+k C (a XC +Δε C )+k W (a XW +Δε W )

[0043]

[0044] Where: k D , k C , k W They are the power dispatch factor, low-carbon dispatch factor and clean dispatch factor of the smart energy system respectively; a XD 、a XC 、a XW They are the total incentive cost of power dispatch, total incentive cost of low-carbon dispatch, and total incentive cost of clean dispatch of smart energy system; Δε D , Δε C , Δε W are the power dispatch deviation cost, low-carbon dispatch deviation cost, and clean dispatch deviation cost of the smart energy system respectively; A, B, and K are the number of users participating in power dispatch, the number of users participating in low-carbon dispatch, and the number of users participating in clean dispatch in the smart energy system respectively; a XD,i is the power incentive of user i, a XC,i is the low-carbon incentive for user i, a XW,i is the cleaning incentive of user i; δ D , δ C , δ W are the power dispatch deviation penalty coefficient, low-carbon dispatch deviation penalty coefficient, and clean dispatch deviation penalty coefficient of the smart energy system respectively; θ XD ,θ XC ,θ XW are the power dispatch target, low-carbon dispatch target, and clean dispatch target issued by the upper power grid; θ SD,i is the actual response power of user i, θ SC,i is the actual carbon emissions of user i, θ SW,i is the actual pollutant emission of user i;

[0045] The constraints are shown in formula (13), including total incentive cost constraint, total power deviation constraint, total carbon emission deviation constraint, and total pollutant emission constraint;

[0046]

[0047] Where: a XD 、a XC 、a XW They are the total incentive cost of power dispatch, total incentive cost of low-carbon dispatch, and total incentive cost of clean dispatch of smart energy system respectively; a ξ is the total incentive cost of the smart energy system; θ XD ,θ XC ,θ XW are the power dispatch target, low-carbon dispatch target, and clean dispatch target issued by the smart energy system; θ SD,i is the actual response power of user i; θ SC,i is the actual carbon emissions of user i; θ SW,i is the actual pollutant emission of user i; ξ D , C , W They are the power dispatch threshold, low-carbon dispatch threshold, and clean dispatch threshold of the smart energy system respectively.

[0048] Furthermore, another object of the present invention is achieved through a multidimensional scheduling system for a smart energy system that serves the needs of a superior power grid, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the computer program is executed by the processor, it implements a multidimensional scheduling method for a smart energy system that serves the needs of a superior power grid as described above.

[0049] Furthermore, another object of the present invention is achieved through a computer-readable storage medium, on which a program for implementing information transmission is stored, and when the program is executed by a processor, the steps of a multi-dimensional scheduling method for a smart energy system that serves the needs of a superior power grid are implemented as described above.

[0050] The present invention has the following advantages and beneficial effects: the present invention can effectively reduce the operating costs of the power system and promote fair competition in the market; it can improve the utilization rate of renewable energy such as wind energy and solar energy, and help reduce the use of fossil fuels; it can reduce pollutant emissions (such as SO2, NOx, etc.), improve air quality and environmental conditions; and thereby reduce the negative impact of pollutant emissions on public health. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a flow chart of the present invention;

[0052] Figure 2 The load-side resource response characteristic curve is analogous to the coal consumption characteristics of the generator. DETAILED DESCRIPTION

[0053] The present invention is further described below with reference to the accompanying drawings:

[0054] Example 1

[0055] like Figure 1 As shown, a multi-dimensional dispatching method of a smart energy system serving the needs of a superior power grid includes the following steps:

[0056] Step 1: Construct a multi-dimensional market dispatch model for a smart energy system taking into account capacity prices: In order to cope with the uncertainty and long-cycle intermittency of new energy, the new power system requires a large amount of power reserve capacity. The current reserve on the power supply side is mainly thermal power units. Due to the restriction of carbon emissions, the overall power generation of thermal power units is strictly limited; if the thermal power units are set with too much reserve capacity, their operating hours will be low, and they need to be compensated for their capacity. The smart energy system takes the load-side resources as the dispatching subject and provides a large amount of reserve for the main grid. When dispatching it, it also needs to be compensated for the capacity price, while the existing load dispatching methods pay less attention to the capacity price of load-side resources. This embodiment proposes a multi-dimensional market dispatching method for a smart energy system taking into account capacity prices. In the process of market dispatching the smart energy system, the power capacity and carbon capacity that can be provided by its internal resources are considered to make the dispatch more in line with reality and improve the effectiveness of the dispatch. The objective function is to minimize the dispatching cost of the smart energy system, including the power market clearing cost, the carbon trading market clearing cost, the actual power operation and clearing deviation cost, the actual carbon emission and clearing deviation cost, the power capacity cost, and the carbon capacity cost. Taking the power deviation and carbon emission deviation of the smart energy system as constraints, multi-dimensional market scheduling of power dispatch and carbon dispatch of the smart energy system is realized, which reduces the operating cost of the smart energy system and reduces carbon emissions.

[0057] The objective function is shown in formula (1), which takes the minimization of the dispatching cost of the smart energy system as the objective function, including the electricity market clearing cost, the carbon trading market clearing cost, the actual electricity operation and clearing deviation cost, the actual carbon emission and clearing deviation cost, the electricity capacity cost, and the carbon capacity cost.

[0058] The constraints are shown in formula (2), which mainly include the power deviation constraint and carbon emission deviation constraint of the smart energy system, realizing the multi-dimensional market dispatch of power dispatch and carbon dispatch of the smart energy system, reducing the operating cost of the smart energy system and reducing carbon emissions.

[0059] miny=y1+y2+y3+y4+y5+y6

[0060]

[0061] Where: y1 is the clearing cost of the smart energy system in the electricity market; y2 is the clearing cost of the smart energy system in the carbon trading market; y3 is the actual electricity operation and clearing deviation cost of the smart energy system; y4 is the actual carbon emission and clearing deviation cost of the smart energy system; y5 is the electricity capacity cost of the smart energy system; y6 is the carbon capacity cost of the smart energy system; M is the number of electricity users participating in the operation of the smart energy system; N is the number of carbon trading users participating in the operation of the smart energy system; k D,i is the electricity trading coefficient of user i; k C,j is the carbon trading coefficient of user j; a SD,i is the electricity clearing price of user i; a SC,j is the carbon trading clearing price of user j; a FD Penalize price for power deviation; a FC Penalize prices for carbon trading deviations; a RD,i is the electricity capacity price of user i; a RC,j is the carbon capacity price of user j; P S2,i (t) is the power clearing power of user i at time t; C S2,j (t) is the carbon clearance rate of user j at time t; t1 and t2 are the start and end times of market clearing respectively; P Sq (t) is the actual power operation and clearing deviation of the smart energy system at time t; C Sq (t) The actual carbon emission and clearing deviation of the smart energy system at time t; P S1 (t) is the actual power of the smart energy system at time t; C S1 (t) is the actual carbon emission rate of the smart energy system at time t; ΔP S is the power deviation threshold of the smart energy system; ΔC S It is the carbon emission deviation threshold of the smart energy system.

[0062] Step 2: Construct an induced dispatch model for a smart energy system based on meta-learning: Different from the power supply side units, whose operating power is strictly executed according to the dispatch instructions issued by the main grid, the load-side resources of the smart energy system have different responses to grid instructions due to their own operating limitations and willingness to participate in response. Taking into account the consumer role and operating limitations of load-side resources, this embodiment uses an induced method to dispatch the smart energy system, that is, different incentives are issued to different users in the smart energy system to induce users to participate in grid dispatch, and each user decides to respond to the grid power and capacity based on the incentives. Based on meta-learning, the power, carbon emissions, and pollutant emissions of different load-side resource incentives and responses of the smart energy system are trained, and an induced dispatch model for a smart energy system based on meta-learning is constructed, which facilitates the dispatch center to grasp the actual response capabilities of the smart energy system under different incentives (including power, carbon emissions, and pollutant emissions) and improve the effectiveness of dispatch. The meta-learning-based induced scheduling model for smart energy systems includes task feature extractors, meta-learners, task adapters, model predictors, and other parts. By training the incentives and response powers of different load-side resources, a meta-learning model is constructed to accurately predict the response power, carbon emissions, and pollutant emissions of different users under different incentive inductions, providing support for smart energy system scheduling.

[0063] The meta-learning process is divided into two stages: 1) Meta-training stage: for each task in the task set, the task feature extractor is first used to extract the task features, and then the meta-learner is used to learn the common patterns and rules between the tasks. Finally, each task is quickly adapted by the task adapter to obtain the adapted model parameters; 2) Meta-testing stage: given the task features of a new task and the parameters output by the meta-learner, the task adapter is used to quickly adapt the new task to obtain the adapted model parameters, and then the model predictor is used to predict the new task;

[0064] The task feature extractor is shown in formula (3), based on which the response features of different users under different incentives are extracted.

[0065] Z D,i =f D,i (a XD,i ,P S3,i ) (3)

[0066] Where: f D,i () is the power response feature extraction function of user i; Z D,i is the power response characteristic of user i; a XD,i is the power incentive of user i; P S3,i is the power response power of user i;

[0067] The meta-learner is shown in formula (4), based on which the law between the power characteristics of different users and the actual power response power is extracted.

[0068]

[0069] Where: g D,i () is the power response law extraction function of user i; is the output of the meta-learner of user i; it represents the law between power characteristics and actual power response power; Z D,i is the power response characteristic of user i; P S3,i is the power response power of user i;

[0070] The task adapter is shown in formula (5), based on which the regularity function is modified to improve the adaptability of the model to new samples.

[0071]

[0072] Where: h D,i () is the power response adaptation function of user i; θ D,i is the model parameter after adaptation by user i; is the output of the meta-learner of user i; it represents the law between power characteristics and actual power response power; Z D,i is the power response characteristic of user i;

[0073] The model predictor is shown in formula (6), which constructs a user power response prediction function model to predict the actual power response power of different users under a given incentive;

[0074] θ D,i =f YD,i (a XD,i ) (6)

[0075] Where: f YD,i () is the power response prediction function model of user i, θ D,i is the power response power of user i output by the model, a XD,i Power incentive for user i;

[0076] A carbon emission response prediction function model of user i is constructed, as shown in formula (7), which predicts the user's carbon emission response under different incentives;

[0077] θ C,i =f YC,i (a XC,i ) (7)

[0078] Where: f YC,i () is the carbon emission response prediction function of user i; θ C,iis the carbon emission of user i output by the model, a XC,i Low-carbon incentive for user i;

[0079] The pollutant emission response prediction function model of user i is constructed, as shown in formula (8), which predicts the user's pollutant emission response under different incentives;

[0080] θ W,i =f YW,i (a XW,i ) (8)

[0081] Where: f YW,i () is the pollutant emission response prediction function of user i; θ W,i is the pollutant emission of user i output by the model, a XW,i Cleaning incentive for user i.

[0082] Step 3. Develop a load-side resource response characteristic curve that is analogous to the coal consumption characteristics of the generator: In the traditional power unit scheduling method, the generator is generally scheduled based on the coal consumption characteristics of the generator, and the units with high unit coal consumption and power generation efficiency are generally scheduled. This embodiment is analogous to the coal consumption characteristics of the generator, and based on the power, carbon emissions, and pollutant emission response prediction functions given by the meta-learning smart energy system induced scheduling method, it constructs the load-side resource response characteristics, and provides a scheduling basis for the smart energy system to schedule load-side resources. Analogous to the scheduling principle of the coal consumption characteristics of the generator set, this embodiment gives priority to users with large response power, small response carbon emissions, and small response pollutant emissions under unit incentives.

[0083] like Figure 2 As shown in the figure, the load-side resources are dispatched based on the load-side resource response characteristic curve analogous to the coal consumption characteristics of the generator, as follows:

[0084] As shown in formula (9), if the response power per unit price of user i is higher than that of user j, user i will be called first.

[0085]

[0086] Where: Δθ D,i and Δθ D,j are the power changes of users i and j, Δa XD,i and Δa XD,j are the incentive changes of users i and j respectively.

[0087] As shown in formula (10), if the carbon emissions per unit price of user i are lower than that of user j, user i will be called first.

[0088]

[0089] Where: ΔθC,i and Δθ C,j are the carbon emission changes of users i and j, Δa XC,i and Δa XC,j are the incentive changes of users i and j respectively.

[0090] As shown in formula (11), if the pollutant emissions of user i under the unit price are lower than those of user j, user i will be called first.

[0091]

[0092] Where: Δθ W,i and Δθ W,j are the pollutant emission changes of users i and j, Δa XW,i and Δa XW,j are the incentive changes of users i and j respectively.

[0093] Step 4: Construct a multi-dimensional dispatching model for a smart energy system that serves the needs of the upper power grid: Based on the load-side resource response characteristic curve of the coal consumption characteristics of a generator-like system, a service demand-oriented multi-dimensional dispatching method for a smart energy system is proposed to perform multi-dimensional dispatching such as power, low carbon, and cleanliness to serve the power dispatching needs, carbon emission dispatching needs, and pollutant emission dispatching needs of the upper power grid. This method takes the minimum total incentive, minimum total power deviation, minimum total carbon emission deviation, and minimum total pollutant emission of the smart energy system as the objective function, and takes the total incentive cost constraint, total power deviation constraint, total carbon emission deviation constraint, and total pollutant emission constraint as the constraint conditions to dispatch different users of the smart energy system. This embodiment sets different dispatching factors to focus on dispatching in different dimensions. For example, when the power grid needs to give priority to carbon emission reduction, the low-carbon dispatching factor can be appropriately increased in the objective function to increase the proportion of the carbon emission reduction target. Thereby realizing the multi-dimensional dispatching of the smart energy system.

[0094] The objective function is shown in formula (12), which mainly includes the minimum total incentive of the smart energy system, the minimum total power deviation, the minimum total carbon emission deviation, and the minimum total pollutant emission.

[0095] min k D (a XD +Δε D )+k C (a XC +Δε C )+k W (a XW +Δε W )

[0096]

[0097] Where: k D , kC , k W They are the power dispatch factor, low-carbon dispatch factor and clean dispatch factor of the smart energy system; a XD 、a XC 、a XW They are the total incentive cost of power dispatch, total incentive cost of low-carbon dispatch, and total incentive cost of clean dispatch of smart energy system; Δε D , Δε C , Δε W are the power dispatch deviation cost, low-carbon dispatch deviation cost, and clean dispatch deviation cost of the smart energy system respectively; A, B, and K are the number of users participating in power dispatch, the number of users participating in low-carbon dispatch, and the number of users participating in clean dispatch in the smart energy system respectively; a XD,i is the power incentive of user i, a XC,i is the low-carbon incentive for user i, a XW,i is the cleaning incentive of user i; δ D , δ C , δ W are the power dispatch deviation penalty coefficient, low-carbon dispatch deviation penalty coefficient, and clean dispatch deviation penalty coefficient of the smart energy system respectively; θ XD ,θ XC ,θ XW are the power dispatch target, low-carbon dispatch target, and clean dispatch target issued by the upper power grid; θ SD,i is the actual response power of user i, θ SC,i is the actual carbon emissions of user i, θ SW,i is the actual pollutant emission of user i;

[0098] The constraints are shown in formula (13), which mainly include total incentive cost constraint, total power deviation constraint, total carbon emission deviation constraint, and total pollutant emission constraint.

[0099]

[0100] Where: a XD 、a XC 、a XW They are the total incentive cost of power dispatch, total incentive cost of low-carbon dispatch, and total incentive cost of clean dispatch of smart energy system respectively; a ξ is the total incentive cost of the smart energy system; θ XD ,θ XC ,θ XW are the power dispatch target, low-carbon dispatch target, and clean dispatch target issued by the smart energy system; θ SD,i is the actual response power of user i; θ SC,i is the actual carbon emissions of user i; θ SW,i is the actual pollutant emission of user i; ξD , C , W They are the power dispatch threshold, low-carbon dispatch threshold, and clean dispatch threshold of the smart energy system respectively.

[0101] This embodiment can meet the economic, low-carbon and clean dispatching requirements of the superior power grid, provide technical support for the load dispatching of the smart energy system, enhance the system's regulation capability and dispatching flexibility, and conform to the development and dispatching requirements of the new power system.

Claims

1. A multi-dimensional dispatching method for a smart energy system serving the needs of a superior power grid, characterized in that: The steps include: Step 1: Construct a multi-dimensional market dispatch model for smart energy systems taking capacity prices into account: Taking the minimization of the dispatch cost of the smart energy system as the objective function, including the electricity market clearing cost, the carbon trading market clearing cost, the actual electricity operation and clearing deviation cost, the actual carbon emission and clearing deviation cost, the electricity capacity cost, and the carbon capacity cost, taking the smart energy system electricity deviation and carbon emission deviation as constraints, realize the multi-dimensional market dispatch of the smart energy system electricity dispatch and carbon dispatch, reduce the operation cost of the smart energy system, and reduce carbon emissions; The objective function is to minimize the dispatching cost of the smart energy system. The objective function is shown in formula (1), and the constraint condition is shown in formula (2). Where: y1 is the clearing cost of the smart energy system in the electricity market; y2 is the clearing cost of the smart energy system in the carbon trading market; y3 is the actual electricity operation and clearing deviation cost of the smart energy system; y4 is the actual carbon emission and clearing deviation cost of the smart energy system; y5 is the electricity capacity cost of the smart energy system; y6 is the carbon capacity cost of the smart energy system; M is the number of electricity users participating in the operation of the smart energy system; N is the number of carbon trading users participating in the operation of the smart energy system; k D,i is the electricity trading coefficient of user i; k C,j is the carbon trading coefficient of user j; a SD,i is the electricity clearing price of user i; a SC,j is the carbon trading clearing price of user j; a FD Penalize price for power deviation; a FC Penalize prices for carbon trading deviations; a RD,i is the electricity capacity price of user i; a RC,j is the carbon capacity price of user j; P S2,i (t) is the power clearing power of user i at time t; C S2,j (t) is the carbon clearance rate of user j at time t; t1 and t2 are the start and end times of market clearing respectively; P Sq (t) is the actual power operation and clearing deviation of the smart energy system at time t; C Sq (t) is the actual carbon emission and clearing deviation of the smart energy system at time t; P S1 (t) is the actual power of the smart energy system at time t; C S1 (t) is the actual carbon emission rate of the smart energy system at time t; ΔP S is the power deviation threshold of the smart energy system; ΔC S It is the carbon emission deviation threshold of the smart energy system; Step 2: Construct an induced dispatch model of a smart energy system based on meta-learning, issue different incentives to different users in the smart energy system, induce users to participate in grid dispatch, and each user decides to respond to grid power and capacity according to the incentives, so that the dispatch center can grasp the actual response capability of the smart energy system under different incentives: including power, carbon emissions and pollutant emissions; the induced dispatch model of a smart energy system based on meta-learning includes a task feature extractor, a meta-learner, a task adapter and a model predictor. By training different load-side resource incentives and response powers, a meta-learning model is constructed to accurately predict the response prediction functions of power, carbon emissions and pollutant emissions of different users under different incentive inductions, providing support for smart energy system dispatch; The meta-learning process is divided into two stages: 1) Meta-training stage: for each task in the task set, the task feature extractor is first used to extract the task features, and then the meta-learner is used to learn the common patterns and rules between the tasks. Finally, each task is quickly adapted by the task adapter to obtain the adapted model parameters; 2) Meta-testing stage: given the task features of a new task and the parameters output by the meta-learner, the task adapter is used to quickly adapt the new task to obtain the adapted model parameters, and then the model predictor is used to predict the new task; The task feature extractor is shown in formula (3), based on which the response features of different users under different incentives are extracted. Z D,i =f D,i (a XD,i ,P S3,i ) (3) Where: f D,i () is the power response feature extraction function of user i; Z D,i is the power response characteristic of user i; a XD,i is the power incentive of user i; P S3,i is the power response power of user i; The meta-learner is shown in formula (4), based on which the law between the power characteristics of different users and the actual power response power is extracted. Where: g D,i () is the power response law extraction function of user i; is the output of the meta-learner of user i; it represents the law between power characteristics and actual power response power; Z D,i is the power response characteristic of user i; P S3,i is the power response power of user i; The task adapter is shown in formula (5), based on which the regularity function is modified to improve the adaptability of the model to new samples. Where: h D,i () is the power response adaptation function of user i; θ D,i is the model parameter after adaptation by user i; is the output of the meta-learner of user i; it represents the law between power characteristics and actual power response power; Z D,i is the power response characteristic of user i; The model predictor is shown in formula (6), which constructs a user power response prediction function model to predict the actual power response power of different users under a given incentive; i D,i =f YD,i (a XD,i ) (6) Where: f YD,i () is the power response prediction function model of user i, θ D,i is the power response power of user i output by the model, a XD,i Power incentive for user i; A carbon emission response prediction function model of user i is constructed, as shown in formula (7), which predicts the user's carbon emission response under different incentives; i C,i =f YC,i (a XC,i ) (7) Where: f YC,i () is the carbon emission response prediction function of user i; θ C,i is the carbon emission of user i output by the model, a XC,i Low-carbon incentive for user i; The pollutant emission response prediction function model of user i is constructed, as shown in formula (8), which predicts the user's pollutant emission response under different incentives; i W,i =f YW,i (a XW,i ) (8) Where: f YW,i () is the pollutant emission response prediction function of user i; θ W,i is the pollutant emission of user i output by the model, a XW,i Cleaning incentive for user i; Step 3: Formulate a load-side resource response characteristic curve analogous to the coal consumption characteristics of the generator: Based on the power, carbon emission, and pollutant emission response prediction functions given by the meta-learning smart energy system induced scheduling model, construct a load-side resource response characteristic curve based on the analogy of the coal consumption characteristics of the generator, provide a scheduling basis for the smart energy system to schedule load-side resources, and give priority to users with large response power, small response carbon emissions, and small response pollutant emissions under the incentive of the scheduling unit; Step 4. Construct a multi-dimensional dispatching model for the smart energy system that serves the needs of the upper-level power grid: take the minimum total incentive, minimum total power deviation, minimum total carbon emission deviation, and minimum total pollutant emission of the smart energy system as the objective function, and take the total incentive cost constraint, total power deviation constraint, total carbon emission deviation constraint, and total pollutant emission constraint as constraints; by setting different dispatching factors, carry out dispatching in different dimensions with emphasis, thereby realizing multi-dimensional dispatching of the smart energy system.

2. A multi-dimensional dispatching method for a smart energy system serving the needs of a superior power grid according to claim 1, characterized in that: In step 3, the load-side resources are dispatched based on the load-side resource response characteristic curve analogous to the coal consumption characteristics of the generator, and users with large response power, small response carbon emissions, and small response pollutant emissions under unit incentives are prioritized, as shown in formula (9). If the response power of user i under the unit price is higher than that of user j, user i is prioritized; Where: Δθ D,i and Δθ D,j are the power changes of users i and j respectively; Δa XD,i and Δa XD,j are the changes in power incentives for users i and j respectively; As shown in formula (10), if the carbon emissions of user i under the unit price are lower than those of user j, user i will be called first; Where: Δθ C,i and Δθ C,j are the carbon emission changes of users i and j respectively; Δa XC,i and Δa XC,j are the changes in low-carbon incentives for users i and j respectively; As shown in formula (11), if the pollutant emission of user i under the unit price is lower than that of user j, user i will be called first; Where: Δθ W,i and Δθ W,j are the changes in pollutant emissions of users i and j respectively; Δa XW,i and Δa XW,j are the cleaning incentive changes for users i and j respectively.

3. A multi-dimensional dispatching method for a smart energy system serving the needs of a superior power grid according to claim 2, characterized in that: In step 4, the objective function is shown in formula (12), including the minimum total incentive of the smart energy system, the minimum total power deviation, the minimum total carbon emission deviation, and the minimum total pollutant emission; Where: k D , k C , k W They are the power dispatch factor, low-carbon dispatch factor and clean dispatch factor of the smart energy system; a XD 、a XC 、a XW They are the total incentive cost of power dispatch, the total incentive cost of low-carbon dispatch, and the total incentive cost of clean dispatch of the smart energy system; Δε D , Δε C , Δε W are the power dispatch deviation cost, low-carbon dispatch deviation cost, and clean dispatch deviation cost of the smart energy system respectively; A, B, and K are the number of users participating in power dispatch, the number of users participating in low-carbon dispatch, and the number of users participating in clean dispatch in the smart energy system respectively; a XD,i is the power incentive of user i, a XC,i is the low-carbon incentive for user i, a XW,i is the cleaning incentive of user i; δ D , δ C , δ W are the power dispatch deviation penalty coefficient, low-carbon dispatch deviation penalty coefficient, and clean dispatch deviation penalty coefficient of the smart energy system respectively; θ XD ,θ XC ,θ XW are the power dispatch target, low-carbon dispatch target, and clean dispatch target issued by the upper power grid; θ SD,i is the actual response power of user i, θ SC,i is the actual carbon emissions of user i, θ SW,i is the actual pollutant emission of user i; The constraints are shown in formula (13), including total incentive cost constraint, total power deviation constraint, total carbon emission deviation constraint, and total pollutant emission constraint; Where: a XD 、a XC 、a XW They are the total incentive cost of power dispatch, total incentive cost of low-carbon dispatch, and total incentive cost of clean dispatch of smart energy system respectively; a ξ is the total incentive cost of the smart energy system; θ XD ,θ XC ,θ XW are the power dispatch target, low-carbon dispatch target, and clean dispatch target issued by the smart energy system; θ SD,i is the actual response power of user i; θ SC,i is the actual carbon emissions of user i; θ SW,i is the actual pollutant emission of user i; ξ D , C , W They are the power dispatch threshold, low-carbon dispatch threshold, and clean dispatch threshold of the smart energy system respectively.

4. A multi-dimensional dispatching system for a smart energy system serving the needs of a superior power grid, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, a multi-dimensional scheduling method for a smart energy system serving the needs of a superior power grid as described in any one of claims 1 to 3 is implemented.

5. A computer-readable storage medium having stored thereon a program for implementing information transmission, characterized in that: When the program is executed by the processor, the steps of a multi-dimensional scheduling method for a smart energy system serving the needs of a superior power grid are implemented as described in any one of claims 1-3.

Images