New energy and electric vehicle charging pile inertia aggregation method, system and computer program product

Abstract

This application discloses a method, system, and computer program product for inertia aggregation of new energy and electric vehicle charging piles, belonging to the field of smart grid technology. The method includes obtaining the inertia aggregation upper limit of the inertia aggregator at each node; constructing an optimal aggregation model for the inertia aggregator with the goal of minimizing the inertia aggregation cost of the inertia aggregator, and obtaining the optimal inertia aggregation cost of the inertia aggregator under a known winning bid; obtaining a quotation sample dataset based on the optimal aggregation model for the inertia aggregator, establishing a curve fitting model based on a deep residual network combined with an adaptive activation function, and inputting the quotation sample dataset into the curve fitting model to calculate the quotation curve of the inertia aggregator under an unknown winning bid. This application considers the influence of distributed electric vehicles, more accurately predicts the quotation curve of the inertia aggregator, and reduces the amount of computation, effectively improving the reliability and efficiency of inertia services, thus having significant practical value.

Description

Technical Field

[0001] The present application relates to the field of smart grids, and in particular to a method, system, and computer program product for aggregating inertia of new energy and electric vehicle charging piles. Background Art

[0002] With the integration of a high proportion of renewable energy, the synchronous inertia of the power system continues to decrease, and frequency stability issues are becoming increasingly prominent. Traditional inertia support relies on synchronous generators, but new energy generators and distributed resources (such as electric vehicles) cannot directly provide physical inertia and require virtual inertia control technology to achieve equivalent inertia response. During inertia service events, the power grid often contracts with multiple inertia aggregators to provide indirect inertia supply services to the underlying electric vehicles and distributed photovoltaic systems. The underlying electric vehicles and distributed photovoltaic systems provide their own inertia supply unit prices to the inertia aggregators based on demand. The inertia aggregators then calculate the quoted costs based on optimization algorithms and participate in inertia market transactions. In recent years, electric vehicles, as large-scale distributed energy storage resources, have been able to participate in system inertia support through vehicle-to-grid (V2G) interaction. However, their decentralized and dynamic nature, coupled with the multi-tiered grid structure, pose challenges to the efficient aggregation of inertia resources.

[0003] Currently, in-depth research has been conducted on energy storage control methods to improve grid frequency stability, suppress frequency disturbances, and reduce steady-state frequency deviations. This provides theoretical support for the participation of electric vehicles and their supporting photovoltaic systems in grid inertia services. When participating in grid inertia services, electric vehicle clusters must meet user charging requirements. Unlike the state-of-charge (SOC) control targets of energy storage batteries, the virtual inertia operation control of an electric vehicle cluster essentially aggregates the charge and discharge power of a large number of individual electric vehicles. Because the charging and discharging requirements of individual electric vehicles vary, appropriate control strategies for participating in inertia services should be developed within the controllable range of each vehicle's charge and discharge. However, due to the large number of nodes in multi-tiered power grids and the interconnection of numerous electric vehicle charging stations and distributed photovoltaic systems at the bottom voltage level, traditional inertia supply does not consider the distributed electric vehicle inertia resources. Optimizing the bidding cost for each node would be labor-intensive and prone to data errors. Summary of the Invention

[0004] The present application aims to provide a method, system and computer program product for aggregating the inertia of new energy and electric vehicle charging piles, which solves the problem of distributed electric vehicle inertia resources not being considered in traditional inertia supply.

[0005] To achieve the above objectives, the technical solution of this application is:

[0006] A method for aggregating inertia of new energy and electric vehicle charging piles, comprising:

[0007] Get the inertia aggregation upper limit of the inertia aggregator at each node;

[0008] With the goal of minimizing the inertia aggregation cost of the inertia aggregator, an optimal aggregation model for the inertia aggregator is constructed to obtain the optimal inertia aggregation cost of the inertia aggregator under known winning bids.

[0009] According to the optimal aggregation model of the inertia aggregator, a quotation sample dataset is obtained, and a curve fitting model based on a deep residual network combined with an adaptive activation function is established. The quotation sample dataset is input into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities.

[0010] Optionally, obtain the inertia aggregation upper limit of the inertia aggregator at each node, including:

[0011] Calculate the upper limit of inertia supply for power shortage at each node;

[0012] Calculate the upper limit of inertia supply for power redundancy at each node;

[0013] The minimum value between the upper limit of the aggregated inertia of each node facing power shortage and the upper limit of the aggregated inertia facing power redundancy is taken as the inertia aggregation upper limit of the inertia aggregator under each node.

[0014] Optionally, the objective function for calculating the upper limit of inertia supply for power shortage at each node is expressed as follows:

[0015]

[0016] in, Indicates the voltage level, The lower the value, the lower the voltage level; b represents the node number connecting the upper and lower power grids; t represents the time sequence number; H aggEV.up,i-1,b,t for time The lower voltage level connected at the node is aggregated upward to the voltage level The inertia is converted into the upper limit of the aggregated inertia for power shortage ; for Voltage level at any moment of The injected power of the node; is the maximum frequency change rate constraint for node b; for time Voltage level connected to the node The non-adjustable load in is a fixed value; for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent load is an adjustable load for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent power generation power of hour, for time The node is connected to the non-adjustable load of the lowest voltage level. for time The equivalent load of the cluster connected to the lowest voltage level at the node, for time The equivalent power generation of the cluster at the lowest voltage level connected to the node;

[0017] The objective function for calculating the upper limit of inertia supply for power redundancy at each node is expressed as follows:

[0018]

[0019] in, H aggEV.down,i-1,b,t for time The lower voltage level connected at the node is aggregated upward to the voltage level The inertia is converted into the upper limit of the aggregated inertia for power redundancy ; for Voltage level at any moment of The injected power of the node; for time Voltage level connected to the node The non-adjustable load in is a fixed value; for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent load is an adjustable load; when hour, for time The node is connected to the non-adjustable load of the lowest voltage level. for time The equivalent load of electric vehicles connected to the lowest voltage level at the node;

[0020] The constraints for calculating the upper limit of inertia supply for power shortage and the upper limit of inertia supply for power redundancy at each node include:

[0021] Active power balance constraints of the power grid, reactive power constraints of the power grid, and current relationship constraints between power grid nodes 、 Voltage relationship constraints between power grid nodes 、 Charging and discharging power constraints of distributed electric vehicles participating in inertia control 、 Power generation constraints for distributed photovoltaic systems participating in inertia control 、 Active power output constraints of generators at each level 、 Reactive output constraints of generators at each level 、 Active power constraints on branches between nodes at each level 、 Reactive power constraints between nodes at each level 、 Voltage constraints at each node 、 Branch current constraints between nodes at each level.

[0022] Optionally, the objective function of the optimal aggregation model of the inertia aggregator is expressed as follows:

[0023]

[0024] in, is the price quote of the kth distributed electric vehicle, represents the adjustable power generation or consumption provided by the kth distributed electric vehicle; is the quotation of the kth supporting distributed photovoltaic, represents the adjustable power generation or consumption provided by the kth supporting distributed photovoltaic system; Indicates the maximum frequency change rate required by node b; represents the supply inertia of the kth distributed electric vehicle at time t, represents the supply inertia of the kth supporting distributed photovoltaic at time t;

[0025] The constraints of the inertia aggregator's optimal aggregation model include: the upper limit constraint of the adjustable inertia of each distributed electric vehicle and each supporting distributed photovoltaic, the inertia aggregation constraint balanced with the winning bid, the lower-level power grid flow constraint, and the maximum frequency change rate constraint of the new power system.

[0026] Optionally, modify the adopted deep residual network, including:

[0027] Use self-gating activation function instead of linear rectification function;

[0028] Add a batch normalization layer after residual addition;

[0029] Use fully connected layers instead of convolution operations;

[0030] The curve fitting accuracy is autonomously optimized by iteratively updating hyperparameters through genetic algorithms.

[0031] Optionally, obtain a sample quotation dataset based on the optimal aggregation model of the inertia aggregator, establish a curve fitting model based on a deep residual network combined with an adaptive activation function, input the sample quotation dataset into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities, including:

[0032] Step S31: obtaining a quotation sample data set according to the inertia aggregator's optimal aggregation model, and preprocessing the quotation sample data set;

[0033] Step S32: Initialize the model structure and parameters of the deep residual network;

[0034] Step S33: inputting the quotation sample data set into the curve fitting model to perform quotation curve fitting;

[0035] Step S34: Determine whether the quote curve is fitted. If so, proceed to step S36; if not, proceed to step S35;

[0036] Step S35: Using a genetic algorithm to change the number of residual blocks N and the hidden layer dimension d for iterative optimization, and then returning to step S32;

[0037] Step S36: Output the final quotation curve.

[0038] Optionally, the quotation sample dataset is input into a curve fitting model to perform quotation curve fitting, including:

[0039] After processing by the first fully connected layer, batch normalization and self-gated activation function, the intermediate feature state inside the residual block is obtained, which is expressed as follows:

[0040]

[0041] in, W t,1 Residual block t The weight matrix of the first fully connected layer; b t,1 Residual block t The bias vector of the first fully connected layer; h t-1 represents the t-1th residual block; BN represents the batch normalization operation; h mid Residual block t The intermediate characteristic state within is the bridge connecting the two layers;

[0042] After the second fully connected layer, the output features of the second fully connected layer are obtained, which are expressed as follows:

[0043]

[0044] in, h res Residual block t Output features of the second fully connected layer; W t,2 Residual block t The weight matrix of the second fully connected layer; b t,2 Residual block t The bias vector of the second fully connected layer;

[0045] Connect the residual blocks to obtain the quotation curve, which is expressed as follows:

[0046]

[0047] in, y pred Represents the quote curve after a deep residual network processing; h N represents the final residual block; W out represents the weight matrix of the output layer; b out Represents the bias vector of the output layer.

[0048] Optionally, the conditions for completing the quote curve fitting include: the loss on the validation set is less than or equal to the set threshold , which is expressed as follows:

[0049]

[0050] in, represents the loss on the validation set, Indicates the set threshold.

[0051] A new energy and electric vehicle charging pile inertia aggregation system, for executing any of the new energy and electric vehicle charging pile inertia aggregation methods described above, comprising: an inertia aggregation upper limit calculation module for an inertia aggregator, an inertia aggregator optimal aggregation cost module, and an inertia aggregator quotation curve calculation module;

[0052] The inertia aggregation upper limit calculation module of the inertia aggregator, the optimal aggregation cost module of the inertia aggregator and the inertia aggregator quotation curve calculation module are connected in sequence.

[0053] A computer program product includes a computer program, which, when executed by a processor, implements the steps of the method for aggregating inertia of new energy and electric vehicle charging piles as described in any one of the above.

[0054] The present application provides a method, system, and computer program product for aggregating inertia of new energy and electric vehicle charging piles, which solves the problem that distributed electric vehicle inertia resources are not considered in traditional inertia supply. The maximum inertia supply upper limit of each node is calculated through an optimization algorithm that considers the influence of distributed electric vehicles. Then, from the perspective of the inertia aggregator, the optimal inertia aggregation is performed under a known winning bid amount. Finally, based on an artificial intelligence method, a quotation curve for the inertia aggregator to participate in the inertia market is fitted. In the problem of maximizing the benefits of inertia aggregators participating in inertia transactions, an optimal aggregation model for inertia aggregators is constructed to calculate the winning bid amount for inertia aggregators to achieve inertia transactions at the lowest cost, thereby more accurately predicting the quotation of inertia aggregators. In the problem of calculating the quotation of inertia aggregators participating in inertia transactions, several key nodes are first selected for traditional quotation calculation, and then the deep residual network improved by the present application is used to perform curve fitting using the key nodes to obtain the quotation curve. This solves the problem that the traditional method requires optimization calculation of each node, which leads to a huge amount of calculation, significantly reduces the amount of calculation, and improves the accuracy of the quotation. This application considers the influence of distributed electric vehicles to more accurately predict the quotation curve of inertia aggregators while reducing the amount of calculation, effectively improving the reliability and efficiency of inertia services, and has significant practical value.

[0055] In order to make the above features and advantages of the application more obvious and easy to understand, the following embodiments are given and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of the new energy and electric vehicle charging pile inertia aggregation method provided in this application.

[0057] Figure 2 Flowchart for obtaining the inertia aggregation upper limit of the inertia aggregator at each node.

[0058] Figure 3 This is a flowchart of step S3.

[0059] Figure 4 Module diagram of the new energy and electric vehicle charging pile inertia aggregation system provided for this application. DETAILED DESCRIPTION

[0060] To make the purpose and technical solutions of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be clearly and completely described below in conjunction with the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the described embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0061] Because electric vehicles are decentralized, it's necessary to use charging stations or parking areas (such as parking lots) and supporting distributed photovoltaic systems at the lowest voltage level to aggregate all resources participating in inertial services into an electric vehicle cluster, which is then connected to the grid. This electric vehicle cluster includes distributed electric vehicles and supporting distributed photovoltaic systems.

[0062] In a specific embodiment of this application, please refer to Figure 1 , Figure 1 This is a flow chart of the method for aggregating the inertia of new energy and electric vehicle charging piles provided in this application. The method for aggregating the inertia of new energy and electric vehicle charging piles provided in this application includes: steps S1 to S3.

[0063] Step S1: Obtain the inertia aggregation upper limit of the inertia aggregator at each node;

[0064] Step S2: With the goal of minimizing the inertia aggregation cost of the inertia aggregator, an optimal aggregation model for the inertia aggregator is constructed to obtain the optimal inertia aggregation cost of the inertia aggregator under a known winning bid;

[0065] Step S3: Obtain a quotation sample dataset based on the optimal aggregation model of the inertia aggregator, establish a curve fitting model based on a deep residual network (ResNet) combined with an adaptive activation function, and input the quotation sample dataset into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities.

[0066] Among them, new energy includes photovoltaic energy.

[0067] In step S1, see Figure 1 In step S1, the inertia aggregation upper limit of the inertia aggregator at each node is obtained.

[0068] In order to realize the grid's inertia incentives for load-side clusters such as distributed electric vehicle charging piles and supporting distributed photovoltaics, and at the same time facilitate the participation of load-side inertia aggregators in the inertia service market, it is necessary to first obtain the maximum supportable inertia supply of all clusters.

[0069] When a disturbance occurs, traditional power systems achieve mechanical-electromagnetic power conservation in the power system before and after the disturbance by responding to the mechanical motion of the synchronous generator's rotor. This means that the disturbance power is compensated by the response of the rotor's mechanical motion. This process is called inertia response. In new power systems, however, distributed power sources can achieve equivalent inertia response through virtual inertia control technology. This means that the disturbance power is compensated by the response of generated or consumed power. Therefore, the virtual inertia of new power system resources is related to the occupied generation or consumption of electricity, as shown below:

[0070] (1)

[0071] in,P c,n Indicates the generated or consumed power when the nth disturbance occurs, H agg represents inertia, P c,max Indicates supply inertia H agg The maximum power generated or consumed, represents the rate of change of angular frequency under the historical maximum disturbance, and Indicates the frequency change rate under the historical maximum disturbance.

[0072] Formula (1) shows that the upper limit of the inertia aggregation required for the inertia aggregator at each node is required, that is, the upper limit of the maximum power generation / power consumption that can be occupied by each node is required, that is, the maximum adjustable load of the equivalent electric vehicle cluster at each node is required. The upper limit of inertia aggregation is obtained based on the output power required to control the virtual inertia provided by the cluster connected to each node, that is, the electric load of the electric vehicle and the photovoltaic output power are changed. By controlling the distributed electric vehicle to reduce the power consumption and increase the photovoltaic output power to achieve equivalent power generation power, and by controlling the distributed electric vehicle to increase the power consumption and reduce the photovoltaic output power to achieve the power consumption, the effect of virtual inertia is achieved.

[0073] Since the grid flow will vary with the number of electric vehicle clusters connected to the grid or the power generation reported by users, the more connected electric vehicles are not necessarily better. The optimal value of the inertia aggregation upper limit of the inertia aggregator at each node is obtained by solving the objective function of the optimization problem.

[0074] Specifically, see Figure 2 , Figure 2 The flowchart for obtaining the inertia aggregation upper limit of the inertia aggregation quotient at each node includes steps S11 to S13.

[0075] In step S11 , the upper limit of inertia supply for power shortage at each node is calculated.

[0076] As an example, we can find the upper limit of inertia supply for power shortage at each node, that is, the maximum power consumption that can be reduced by the electric vehicle cluster at each node, that is, the minimum active power injected into the node by the upper voltage level. The objective function is expressed as follows:

[0077] (2)

[0078] in, Indicates the voltage level, The lower the value, the lower the voltage level; b represents the node number connecting the upper and lower power grids; t represents the time sequence number; HaggEV.up,i-1,b,t for time The lower voltage level connected at the node is aggregated upward to the voltage level The inertia is converted into the upper limit of the aggregated inertia for power shortage by formula (1): ; for Voltage level at any moment of The injected power of the node; is the maximum frequency change rate constraint for node b; for time Voltage level connected to the node The non-adjustable load in is a fixed value; for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent load is an adjustable load for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent power generation power. hour, for time The node is connected to the non-adjustable load of the lowest voltage level. for time The equivalent load of the cluster connected to the lowest voltage level at the node, for time The equivalent power generation of the cluster at the lowest voltage level connected to the node.

[0079] Furthermore, the constraints on inertia supply for power shortage at each node are established. con1.1 ~Constraints con1.13 .

[0080] As an example, the active power balance constraint of the grid con1.1 It is expressed as follows:

[0081] (3)

[0082] in, and express Node number at the network level; for time The injected active power of the node, for time The node's own non-adjustable active load, for time The active output of the generator on the node, for time The equivalent aggregate load on the node, for time Equivalent aggregated generated power at the node.

[0083] Reactive power constraints of the power grid con1.2 It is expressed as follows:

[0084] (4)

[0085] in, express Node number at the network level; for time The node injects reactive power, for time The node's own non-adjustable reactive load, for time The reactive power output of the generator on the node, for time Equivalent aggregate reactive load at the node, for time Equivalent aggregate reactive power at the node.

[0086] Current relationship constraints between power grid nodes con1.3 It is expressed as follows:

[0087] (5)

[0088] in, for time 、 The current value between nodes, 、 for time 、 Active and reactive power flowing between nodes.

[0089] Voltage relationship constraints between power grid nodes con1.4 It is expressed as follows:

[0090] (6)

[0091] in, for time The voltage value of the node, for time The voltage value of the node, 、 for 、 The resistance and reactance values ​​between nodes.

[0092] Charging and discharging power constraints of electric vehicle clusters participating in inertia control con1.5 It is expressed as follows:

[0093] (7)

[0094] in, is the charging and discharging power of the electric vehicle cluster, are the minimum and maximum charging and discharging powers of the electric vehicle cluster.

[0095] Power generation constraints for distributed photovoltaic systems participating in inertia control con1.6 It is expressed as follows:

[0096] (8)

[0097] in, To support distributed photovoltaic power generation, It is the minimum and maximum value of the supporting distributed photovoltaic power generation power.

[0098] Active power output constraints of generators at each level con1.7 It is expressed as follows:

[0099] (9)

[0100] in, for i The active output of the generators at each level, for i The lower limit of the active output of the generator at each level, for i The upper limit of the active output of the generator at the level.

[0101] Reactive output constraints of generators at each level con1.8 It is expressed as follows:

[0102] (10)

[0103] in, fori The reactive power output of the generator at each level, for i The lower limit of reactive power output of the generator at each level, for i The upper limit of reactive power output of generators at each level.

[0104] Active power constraints on branches between nodes at each level con1.9 It is expressed as follows:

[0105] (11)

[0106] in, for time i Hierarchical nodes ,node The active power of the branch between for i Hierarchical nodes ,node The lower limit of the branch active power between for i Hierarchical nodes ,node The upper limit of the active power of the branches between them.

[0107] Reactive power constraints between nodes at each level con1.10 It is expressed as follows:

[0108] (12)

[0109] in, for time i Hierarchical nodes ,node The branch reactive power between for i Hierarchical nodes ,node The lower limit of the branch reactive power between for i Hierarchical nodes ,node The reactive power upper limit of the branches between them.

[0110] node k Voltage Constraint con1.11 It is expressed as follows:

[0111] (13)

[0112] in, for Time Node The voltage, For nodes The lower voltage limit, For nodes upper voltage limit.

[0113] node Voltage Constraint con1.12 It is expressed as follows:

[0114] (14)

[0115] in, for Time Node The voltage, For nodes The lower voltage limit, For nodes upper voltage limit.

[0116] Branch current constraints between nodes at each level con1.13 It is expressed as follows:

[0117] (16)

[0118] in, for Time Node ,node The branch current between For nodes ,node The lower limit of the branch current between For nodes ,node The upper limit of the branch current between .

[0119] In step S12, the upper limit of inertia supply for power redundancy at each node is calculated.

[0120] As an example, we can find the upper limit of the inertia supply for power redundancy at each node, that is, the equivalent maximum power load of the electric vehicle cluster at each node, that is, the maximum active power injected into each node by the upper voltage level. The objective function is expressed as follows:

[0121] (17)

[0122] in, H aggEV.down,i-1,b,t for time The lower voltage level connected at the node is aggregated upward to the voltage level The inertia is converted into the upper limit of the aggregated inertia for power redundancy by formula (1): ; for Voltage level at any moment of The injected power of the node; for time Voltage level connected to the node The non-adjustable load in is a fixed value; for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent load is an adjustable load. hour, for time The node is connected to the non-adjustable load of the lowest voltage level. for time The equivalent load of the electric vehicle cluster at the lowest voltage level is connected at the node.

[0123] Furthermore, the constraints on inertia supply for power redundancy at each node are established. con2.1 ~Constraints con2.13 It is worth noting that in addition to the objective function being different from the objective function of finding the upper limit of inertia supply for power redundancy at each node and finding the upper limit of inertia supply for power shortage at each node, in the constraints, since finding the upper limit of inertia supply for power redundancy at each node is to find the maximum power load of the equivalent electric vehicle cluster at each node, the increase in load may lead to an increase in the power flow burden of the lower power grid. Therefore, it is necessary to consider not only the power flow constraints of the upper power grid but also the power flow constraints of the lower power grid.

[0124] As an example, the active power balance constraint of the grid con2.1 It is expressed as follows:

[0125] (18)

[0126] in, and Indicates the power grid i The node number of the level; for time The injected active power of the node, for time The node's own non-adjustable active load, for time The active output of the generator on the node, for time The equivalent aggregate load on the node, for time Equivalent aggregated generated power at the node.

[0127] Reactive power constraints of the power grid con2.2 It is expressed as follows:

[0128] (19)

[0129] in, Indicates the power grid i The node number of the level; for time The node injects reactive power, for time The node's own non-adjustable reactive load, for time The reactive power output of the generator on the node, for time Equivalent aggregate reactive load at the node, for time Equivalent aggregate reactive power at the node.

[0130] Current relationship constraints between power grid nodes con2.3 It is expressed as follows:

[0131] (20)

[0132] in, for time 、 The current value between nodes, 、 for time 、 Active and reactive power flowing between nodes.

[0133] Voltage relationship constraints between power grid nodes con2.4 It is expressed as follows:

[0134] (twenty one)

[0135] in, for time The voltage value of the node, for time The voltage value of the node, 、 for 、 The resistance and reactance values ​​between nodes.

[0136] Charging and discharging power constraints of electric vehicle clusters participating in inertia control con2.5 It is expressed as follows:

[0137] (twenty two)

[0138] in, is the charging and discharging power of the electric vehicle cluster, are the minimum and maximum charging and discharging powers of the electric vehicle cluster.

[0139] Power generation constraints for distributed photovoltaic systems participating in inertia control con2.6 It is expressed as follows:

[0140] (twenty three)

[0141] in, To support distributed photovoltaic power generation, It is the minimum and maximum value of the supporting distributed photovoltaic power generation power.

[0142] Active power output constraints of generators at each level con2.7 It is expressed as follows:

[0143] (twenty four)

[0144] in, for i The active output of the generators at each level, for i The lower limit of the active output of the generator at each level, for i The upper limit of the active output of the generator at the level.

[0145] Reactive output constraints of generators at each level con2.8 It is expressed as follows:

[0146] (25)

[0147] in, for i The reactive power output of the generator at each level, for i The lower limit of reactive power output of the generator at each level, for i The upper limit of reactive power output of generators at each level.

[0148] Active power constraints on branches between nodes at each level con2.9 It is expressed as follows:

[0149] (26)

[0150] in, for time i Hierarchical nodes ,node The active power of the branch between for i Hierarchical nodes ,node The lower limit of the branch active power between for i Hierarchical nodes ,node Furthermore, the power flow constraint of the lower power grid is calculated using formula (26).

[0151] Reactive power constraints between nodes at each level con2.10 It is expressed as follows:

[0152] (27)

[0153] in, for time i Hierarchical nodes ,node The branch reactive power between for i Hierarchical nodes ,node The lower limit of the branch reactive power between for i Hierarchical nodes ,node Furthermore, the power flow constraint of the lower power grid is calculated using formula (27).

[0154] node k Voltage Constraint con2.11 It is expressed as follows:

[0155] (28)

[0156] in, for Time Node The voltage, For nodes The lower voltage limit, For nodes upper voltage limit.

[0157] node Voltage Constraint con2.12 It is expressed as follows:

[0158] (29)

[0159] in, for Time Node The voltage, For nodes The lower voltage limit, For nodes upper voltage limit.

[0160] Branch current constraints between nodes at each level con2.13 It is expressed as follows:

[0161] (30)

[0162] in, for Time Node ,node The branch current between For nodes ,node The lower limit of the branch current between For nodes ,node The upper limit of the branch current between .

[0163] In step S13, the minimum value between the upper limit of the inertia that can be aggregated for power shortage and the upper limit of the inertia that can be aggregated for power redundancy of each node is taken as the upper limit of the inertia aggregation of the inertia aggregation quotient under each node.

[0164] As an example, the maximum power consumption that can be reduced and the maximum power consumption that can be increased by an electric vehicle cluster are usually inconsistent. When calculating the upper limit of the aggregated inertia, the maximum power consumption that can be reduced and the maximum power consumption that can be increased by the electric vehicle cluster are calculated separately. The minimum value between the maximum power consumption that can be reduced and the maximum power consumption that can be increased at each node, that is, the minimum value between the upper limit of the aggregated inertia for power shortage and the upper limit of the aggregated inertia for power redundancy at each node, is taken as the upper limit of the maximum power generation or consumption that can be occupied, that is, the inertia aggregation upper limit of the inertia aggregator under each node, which is specifically expressed as follows:

[0165] (31)

[0166] in, Indicates the grid voltage level, The lower the value, the lower the voltage level; b represents the node number; t represents the time sequence number; and They represent the upper limit of the aggregated inertia for power shortage and the upper limit of the aggregated inertia for power redundancy respectively.

[0167] In step S2, see Figure 1 In step S2, with the goal of minimizing the inertia aggregation cost of the inertia aggregator, an optimal aggregation model for the inertia aggregator is constructed to obtain the optimal inertia aggregation cost of the inertia aggregator under known bidding quantities.

[0168] As an example, different electric vehicle charging piles and supporting photovoltaic systems will have different virtual inertias, and the quotations provided will also be different. The winning bid for inertia transactions will be provided at the lowest cost, ensuring that inertia aggregators can efficiently obtain the benefits of inertia transactions.

[0169] As an example, taking the minimum inertia aggregation cost of the inertia aggregator as the objective function, an optimal aggregation model for the inertia aggregator is constructed. The objective function is expressed as follows:

[0170] (31)

[0171] in, is the price quote of the kth distributed electric vehicle, represents the adjustable power generation or consumption provided by the kth distributed electric vehicle; is the quotation of the kth supporting distributed photovoltaic, represents the adjustable power generation or consumption provided by the kth supporting distributed photovoltaic system; Indicates the maximum frequency change rate required by node b; represents the supply inertia of the kth distributed electric vehicle at time t, It represents the supply inertia of the kth supporting distributed photovoltaic at time t.

[0172] Furthermore, the constraints for constructing the optimal aggregation model of the inertia aggregator include: the upper limit constraint of the adjustable inertia of each distributed electric vehicle and each supporting distributed photovoltaic, the inertia aggregation constraint balanced with the winning bid, the lower-level power grid flow constraint, and the maximum frequency change rate constraint of the new power system.

[0173] As an example, when aggregating distributed electric vehicle groups, the inertia aggregator must consider the upper limit of their adjustable inertia. The adjustment amount of each distributed electric vehicle must be within this upper limit to ensure the normal operation of the power supply. The upper limit constraint of the adjustable inertia of each distributed electric vehicle is: con3.1 It is expressed as follows:

[0174] (32)

[0175] in, is the upper limit of the adjustable capacity of the kth distributed electric vehicle at time t, which is determined by the inertia aggregation upper limit of the inertia aggregator under each node obtained in step S1; is the lower limit of the adjustable capacity of the kth distributed electric vehicle group at time t.

[0176] As an example, when aggregating distributed photovoltaic power plants, the inertia aggregator must consider the upper limit of the adjustable inertia. The adjustment amount of each distributed photovoltaic power plant must be within this upper limit to ensure the normal operation of the power supply. The upper limit constraint of the adjustable inertia of each distributed photovoltaic power plant is: con3.2 It is expressed as follows:

[0177] (33)

[0178] in, is the upper limit of the adjustable capacity of the kth supporting distributed photovoltaic at time t, which is determined by the inertia aggregation upper limit of the inertia aggregator under each node obtained in step S1; is the lower limit of the adjustable capacity of the kth supporting distributed photovoltaic at time t.

[0179] As an example, when an inertia aggregator participates in the spot market bidding, it is required to provide an inertia supply that is balanced with the inertia of the inertia aggregator through the integration and coordination of multiple distributed electric vehicle clusters, and the inertia aggregation constraint that is balanced with the inertia of the inertia of the inertia aggregator is con3.3 It is expressed as follows:

[0180] (34)

[0181] in, The adjustable power generation or power consumption supplied to the kth electric vehicle cluster, including: the adjustable power generation or power consumption supplied by the kth distributed electric vehicle and the adjustable power generation or power consumption provided by the kth supporting distributed photovoltaic system ; Indicates i -1 level inertia aggregators participate in the winning bid volume of inertia market transactions.

[0182] As an example, when the inertia aggregator aggregates inertia in the lower grid, k ,node The active power and reactive power of the branches between them are constrained, and the power flow constraints of the lower power grid are con3.4 ~Constraints con3.5 It is expressed as follows:

[0183] (35)

[0184] (36)

[0185] in, express t Time Node k ,node The active power of the branch between Representation node k ,node The lower limit of the branch active power between Representation node k ,node The upper limit of the active power of the branches between express t Time Node k ,node The branch reactive power between Representation node k ,node The lower limit of the branch reactive power between Representation node k ,node The reactive power upper limit of the branches between them.

[0186] As an example, the inertia provided by the inertia supplier and the inertia of the new power system need to be able to meet the frequency stability requirements of the new power system and the maximum frequency change rate constraint of the new power system. con3.6 It is expressed as follows:

[0187] (37)

[0188] in, represents the system disturbance, Indicates the system rated frequency, Indicates the total active power output of the system. represents the original inertia of the system, Indicates the maximum frequency change rate set by the system.

[0189] As an example, after constructing the optimal aggregation model of the inertia aggregator, a genetic algorithm is used to solve the optimal aggregation model of the inertia aggregator to obtain the optimal inertia aggregation cost of the inertia aggregator under a known winning bid.

[0190] In step S3, see Figure 1 In step S3, a quotation sample dataset is obtained based on the optimal aggregation model of the inertia aggregator. A curve fitting model based on a deep residual network combined with an adaptive activation function is established. The quotation sample dataset is input into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities.

[0191] Before conducting an inertia transaction, an inertia aggregator needs to quote inertia. However, the cost calculation of the inertia aggregator's optimal aggregation model requires knowing the actual standard volume in the inertia transaction. Therefore, the inertia aggregator needs to obtain a cost-inertia supply curve to support the quote of the inertia transaction. The cost-inertia supply curve is obtained by optimizing the optimal aggregation cost for each time. Theoretically, the curve has countless points, and it is impossible to optimize the inertia aggregation cost for each time. Therefore, in this application, an improved deep residual network is used. Some optimized inertia aggregation cost values ​​are selected as key points to form a quotation sample data set as training data. The improved deep residual network is used to establish a curve fitting model to obtain the cost-inertia supply quotation curve.

[0192] As an example, in the curve fitting scenario, this application makes the following improvements to the standard deep residual network:

[0193] Activation function replacement: Use the self-gated activation function (Swish function) to replace the linear rectification function (ReLU function);

[0194] Add a batch normalization (BN) layer after the residual addition to avoid unstable training caused by the dimensionality difference between the input and the residual;

[0195] Remove the convolution operation and replace it with a fully connected layer to adapt to the one-dimensional input and output structure;

[0196] The hyperparameters are iteratively updated through genetic algorithms to autonomously optimize the curve fitting accuracy.

[0197] Specifically, see Figure 3 , Figure 3 This is a flowchart of step S3, based on the improved deep residual network combined with the curve fitting model of the adaptive activation function, step S3 includes: steps S31 to S36.

[0198] Step S31: obtaining a quotation sample data set according to the inertia aggregator's optimal aggregation model, and preprocessing the quotation sample data set.

[0199] As an example, the quote sample dataset is represented as follows:

[0200] (38)

[0201] in, Indicates the q scalar quantity in group inertia, Indicates the inertia aggregator in the q scalar value of group inertia The following quote, Q Represents the set of all data points.

[0202] Furthermore, the quotation sample data set is preprocessed, including normalization processing. The normalization processing is expressed as follows:

[0203] (39)

[0204] in, represents the normalized inertia scalar, represents the normalized quote, μ x represents the mean value of the scalar quantity in inertia, μ y Indicates the average price of the quote; δ x Indicates the maximum value of the scalar inertia; μ y Indicates the maximum quote value.

[0205] Furthermore, the preprocessed quotation sample dataset is divided into a training set, a validation set, and a test set in proportion.

[0206] In one embodiment of the present application, the preprocessed quotation sample data set is divided into a training set, a validation set, and a test set in a ratio of 8:1:1.

[0207] Step S32: Initialize the model structure and parameters of the deep residual network.

[0208] As an example, the network hyperparameters of the deep residual network are input, including: the number of input residual blocks N, the input hidden layer dimension d, and the input adaptive activation function. .

[0209] As an example, a genetic algorithm is used to optimize the network hyperparameters by changing the number of residual blocks N and the hidden layer dimension d.

[0210] As an example, parameter initialization includes initializing the parameters of the deep residual network using HE initialization, which is expressed as follows:

[0211] (40)

[0212] Where W represents the weight parameter of the deep residual network; N represents the normal distribution; 0 represents the mean of the normal distribution, that is, the mean of the initialized weight parameter is 0; represents the standard deviation of the normal distribution, d in Represents the input hidden layer dimension.

[0213] Step S33: inputting the quotation sample data set into the curve fitting model to perform quotation curve fitting.

[0214] As an example, the input layer inputs the scalar inertia after preprocessing Mapped to high-dimensional space, it is expressed as follows:

[0215] (41)

[0216] in, W 0 Represents the weight matrix, which converts the preprocessed inertia into a scalar The input is mapped to a d-dimensional feature space; b 0 Represents the bias vector.

[0217] Furthermore, residual block calculation is performed. Taking the t-th residual block as an example (t=1, 2, ..., N), the internal calculation process of the t-th residual block includes: Step S331 to Step S333:

[0218] Step S331: After processing by the first fully connected layer, batch normalization, and self-gated activation function, the intermediate feature state inside the residual block is obtained, which is expressed as follows:

[0219] (42)

[0220] in, W t,1 Residual block t The weight matrix of the first fully connected layer; b t,1 Residual block t The bias vector of the first fully connected layer; h t-1 represents the t-1th residual block; BN represents the batch normalization operation; h mid Residual block t The intermediate feature state within is the bridge connecting the two layers.

[0221] Step S332: After passing through the second fully connected layer, the output features of the second fully connected layer are obtained, which are expressed as follows:

[0222] (43)

[0223] in, h res Residual block t Output features of the second fully connected layer; W t,2 Residual block t The weight matrix of the second fully connected layer; b t,2 Residual block t The bias vector of the second fully connected layer.

[0224] Step S333: Connect the residual blocks to obtain a quotation curve.

[0225] As an example, the t-th residual block h t It is expressed as follows:

[0226] (44)

[0227] Furthermore, the output layer outputs a quote curve, which is expressed as follows:

[0228] (45)

[0229] in, y pred Represents the quote curve after a deep residual network processing; h N represents the final residual block; W out represents the weight matrix of the output layer; b out Represents the bias vector of the output layer.

[0230] Step S34: Determine whether the quotation curve is fitted. If so, proceed to step S36; if not, proceed to step S35.

[0231] As an example, the mean square error is used for loss calculation, which is expressed as follows:

[0232] (46)

[0233] in, y pred,i Represents the predicted value of the i-th quotation curve output by the deep residual network; y norm,i represents the standard value of the i-th quotation curve output by the deep residual network in the validation set; M represents the number of samples; represents the loss on the validation set.

[0234] Furthermore, the conditions for completing the fitting of the quote curve include: The loss is less than or equal to the set threshold , which is expressed as follows:

[0235] (46)

[0236] Step S35: Use a genetic algorithm to change the number of residual blocks N and the hidden layer dimension d for iterative optimization, and return to step S32.

[0237] Step S36: Output the final quotation curve.

[0238] This application also provides a new energy and electric vehicle charging pile inertia aggregation system for executing the above-mentioned new energy and electric vehicle charging pile inertia aggregation method. Figure 4 , Figure 4 This is a module diagram of the new energy and electric vehicle charging pile inertia aggregation system provided in this application. The new energy and electric vehicle charging pile inertia aggregation system provided in this application includes: an inertia aggregation upper limit calculation module 41 of the inertia aggregator, an inertia aggregator optimal aggregation cost module 42, and an inertia aggregator quotation curve calculation module 43; the inertia aggregation upper limit calculation module 41 of the inertia aggregator, the inertia aggregator optimal aggregation cost module 42 and the inertia aggregator quotation curve calculation module 43 are connected in sequence.

[0239] As an example, the inertia aggregation upper limit calculation module 41 of the inertia aggregation provider is used to obtain the inertia aggregation upper limit of the inertia aggregation provider at each node;

[0240] The inertia aggregator optimal aggregation cost module 42 is used to build an inertia aggregator optimal aggregation model with the goal of minimizing the inertia aggregation cost of the inertia aggregator, and obtain the inertia optimal aggregation cost of the inertia aggregator under a known winning bid amount;

[0241] The inertia aggregator quotation curve calculation module 43 is used to obtain a quotation sample data set based on the inertia aggregator's optimal aggregation model, establish a curve fitting model based on a deep residual network combined with an adaptive activation function, and input the quotation sample data set into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities.

[0242] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for aggregating inertia of new energy and electric vehicle charging piles.

[0243] The present application provides a method, system, and computer program product for aggregating inertia of new energy and electric vehicle charging piles, which solves the problem that distributed electric vehicle inertia resources are not considered in traditional inertia supply. The maximum inertia supply upper limit of each node is calculated through an optimization algorithm that considers the influence of distributed electric vehicles. Then, from the perspective of the inertia aggregator, the optimal inertia aggregation is performed under a known winning bid amount. Finally, based on an artificial intelligence method, a quotation curve for the inertia aggregator to participate in the inertia market is fitted. In the problem of maximizing the benefits of inertia aggregators participating in inertia transactions, an optimal aggregation model for inertia aggregators is constructed to calculate the winning bid amount for inertia aggregators to achieve inertia transactions at the lowest cost, thereby more accurately predicting the quotation of inertia aggregators. In the problem of calculating the quotation of inertia aggregators participating in inertia transactions, several key nodes are first selected for traditional quotation calculation, and then the deep residual network improved by the present application is used to perform curve fitting using the key nodes to obtain the quotation curve. This solves the problem that the traditional method requires optimization calculation of each node, which leads to a huge amount of calculation, significantly reduces the amount of calculation, and improves the accuracy of the quotation. This application considers the influence of distributed electric vehicles to more accurately predict the quotation curve of inertia aggregators while reducing the amount of calculation, effectively improving the reliability and efficiency of inertia services, and has significant practical value.

[0244] Although the present application has been disclosed above with reference to the embodiments, they are not intended to limit the present application. Anyone with ordinary knowledge in the technical field may make slight changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be determined by the scope of the appended patent application.

Claims

1. A method for aggregating the inertia of new energy and electric vehicle charging piles, characterized in that: include, Get the inertia aggregation upper limit of the inertia aggregator at each node; With the goal of minimizing the inertia aggregation cost of the inertia aggregator, an optimal aggregation model for the inertia aggregator is constructed to obtain the optimal inertia aggregation cost of the inertia aggregator under known winning bids. Based on the optimal aggregation model of the inertia aggregator, a sample quotation dataset is obtained. A curve fitting model based on a deep residual network combined with an adaptive activation function is established. The sample quotation dataset is input into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities. Get the inertia aggregation upper limit of the inertia aggregator at each node, including: Calculate the upper limit of inertia supply for power shortage at each node; Calculate the upper limit of inertia supply for power redundancy at each node; The minimum value between the upper limit of the aggregated inertia for power shortage and the upper limit of the aggregated inertia for power redundancy of each node is taken as the upper limit of the inertia aggregation of the inertia aggregator under each node; The objective function of the optimal aggregation model of the inertia aggregator is expressed as follows: in, is the price quote of the kth distributed electric vehicle, represents the adjustable power generation or consumption provided by the kth distributed electric vehicle; is the quotation of the kth supporting distributed photovoltaic, represents the adjustable power generation or consumption provided by the kth supporting distributed photovoltaic system; Indicates the maximum frequency change rate required by node b; represents the supply inertia of the kth distributed electric vehicle at time t, represents the supply inertia of the kth supporting distributed photovoltaic at time t; The constraints of the inertia aggregator's optimal aggregation model include: the upper limit constraint on the adjustable inertia of each distributed electric vehicle and each supporting distributed photovoltaic, the inertia aggregation constraint balanced with the winning bid, the lower-level power grid flow constraint, and the maximum frequency change rate constraint of the new power system; Modifications to the adopted deep residual network include: Use self-gating activation function instead of linear rectification function; Add a batch normalization layer after the residual addition; Use fully connected layers instead of convolution operations; Iteratively update hyperparameters through genetic algorithms to autonomously optimize curve fitting accuracy; Based on the optimal aggregation model of the inertia aggregator, a quotation sample dataset is obtained. A curve fitting model based on a deep residual network combined with an adaptive activation function is established. The quotation sample dataset is input into the curve fitting model to calculate the quotation curve of the inertia aggregator under unknown intermediate quantities, including: Step S31: obtaining a quotation sample data set according to the inertia aggregator's optimal aggregation model, and preprocessing the quotation sample data set; Step S32: Initialize the model structure and parameters of the deep residual network; Step S33: inputting the quotation sample data set into the curve fitting model to perform quotation curve fitting; Step S34: Determine whether the quote curve is fitted. If so, proceed to step S36; if not, proceed to step S35; Step S35: Using a genetic algorithm to change the number of residual blocks N and the hidden layer dimension d for iterative optimization, and then returning to step S32; Step S36: Output the final quotation curve.

2. The method for aggregating the inertia of new energy and electric vehicle charging piles according to claim 1, characterized in that: The objective function for calculating the upper limit of inertia supply for power shortage at each node is expressed as follows: in, Indicates the voltage level, The lower the value, the lower the voltage level; b represents the node number connecting the upper and lower power grids; t represents the time sequence number; for time The lower voltage level connected at the node is aggregated upward to the voltage level The inertia is converted into the upper limit of the aggregated inertia for power shortage ; for Voltage level at any moment of The injected power of the node; is the maximum frequency change rate constraint for node b; for time Voltage level connected to the node The non-adjustable load in is a fixed value; for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent load is an adjustable load for time The lower voltage level connected at the node is aggregated upward to the voltage level Equivalent power generation; The objective function for calculating the upper limit of inertia supply for power redundancy at each node is expressed as follows: in, for time The lower voltage level connected at the node is aggregated upward to the voltage level The inertia is converted into the upper limit of the aggregated inertia for power redundancy ; for Voltage level at any moment of The injected power of the node; for time Voltage level connected to the node The non-adjustable load in is a fixed value; for time The lower voltage level connected at the node is aggregated upward to the voltage level The equivalent load is an adjustable load; The constraints for calculating the upper limit of inertia supply for power shortage and the upper limit of inertia supply for power redundancy at each node include: Active power balance constraints of the power grid, reactive power constraints of the power grid, and current relationship constraints between power grid nodes 、 Voltage relationship constraints between power grid nodes 、 Charging and discharging power constraints of distributed electric vehicles participating in inertia control 、 Power generation constraints for distributed photovoltaic systems participating in inertia control 、 Active power output constraints of generators at each level 、 Reactive output constraints of generators at each level 、 Active power constraints on branches between nodes at each level 、 Reactive power constraints between nodes at each level 、 Voltage constraints at each node 、 Branch current constraints between nodes at each level.

3. The method for aggregating the inertia of new energy and electric vehicle charging piles according to claim 1, characterized in that: Input the quotation sample data set into the curve fitting model to perform quotation curve fitting, including: After processing by the first fully connected layer, batch normalization and self-gated activation function, the intermediate feature state inside the residual block is obtained, which is expressed as follows: in, W t,1 Residual block t The weight matrix of the first fully connected layer; b t,1 Residual block t The bias vector of the first fully connected layer; h t-1 represents the t-1th residual block; BN represents the batch normalization operation; h mid Residual block t The intermediate characteristic state within is the bridge connecting the two layers; After the second fully connected layer, the output features of the second fully connected layer are obtained, which are expressed as follows: in, h res Residual block t Output features of the second fully connected layer; W t,2 Residual block t The weight matrix of the second fully connected layer; b t,2 Residual block t The bias vector of the second fully connected layer; Connect the residual blocks to obtain the quotation curve, which is expressed as follows: in, y pred Represents the quote curve after a deep residual network processing; h N represents the final residual block; W out represents the weight matrix of the output layer; b out Represents the bias vector of the output layer.

4. The method for aggregating inertia of new energy and electric vehicle charging piles according to claim 1, characterized in that: The conditions for completing the quote curve fitting include: the loss on the validation set is less than or equal to the set threshold , which is expressed as follows: in, represents the loss on the validation set, Indicates the set threshold.

5. A new energy and electric vehicle charging pile inertia aggregation system, used to execute the new energy and electric vehicle charging pile inertia aggregation method according to any one of claims 1 to 4, characterized in that: include: Inertia aggregation upper limit calculation module, inertia aggregation optimal aggregation cost module, and inertia aggregator quotation curve calculation module; The inertia aggregation upper limit calculation module of the inertia aggregator, the optimal aggregation cost module of the inertia aggregator and the inertia aggregator quotation curve calculation module are connected in sequence.

6. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the new energy and electric vehicle charging pile inertia aggregation method according to any one of claims 1 to 4 are implemented.

Images