A provincial and local coordinated energy storage dispatching method, device, equipment, medium and product

Abstract

The present application discloses a provincial-local collaborative energy storage scheduling method, device, equipment, medium and product, and relates to the field of energy scheduling technology. The method comprises: aggregating a distributed energy storage operation scheduling model to obtain a regional energy storage peak capacity characterization model; redundancy identification and elimination of regional cost function constraints and regional safety operation constraints to obtain redundancy-removed constraints; regional safety operation constraints include a distributed energy storage operation scheduling model and a regional energy storage peak capacity characterization model; based on the redundancy-removed constraints, a dual outer cutting algorithm is used to perform polyhedron projection to obtain a feasible domain set; according to the feasible domain set and the provincial power grid optimization operation scheduling model, a scheduling strategy for each moment is obtained; and a KA-DDPG algorithm is used to train a neural network model according to the feasible domain set and the scheduling strategy for each moment to obtain a scheduling model. The present application can reduce the time of provincial-local collaborative energy storage scheduling.

Description

Technical Field

[0001] The present application relates to the field of energy dispatch technology, and in particular to a provincial and local collaborative energy storage dispatch method, device, equipment, medium and product. Background Art

[0002] As the proportion of renewable energy power generation increases year by year, the installed capacity of conventional supporting power sources is insufficient and growing slowly. In extreme weather conditions such as extremely hot and windless weather, and extremely cold and lightless weather, the power grid faces a tense situation of imbalance between power supply and demand. Energy storage, as an important regulating resource, is mainly concentrated at the moment of the largest power gap to solve the problem of the largest power gap. However, all energy storage power stations are operated at maximum power discharge, which cannot cover the entire power gap period, and cause waste of energy storage power, resulting in a mismatch between the size and duration of the power gap and the spatial and temporal distribution of new energy and energy storage facilities.

[0003] In the prior art, for the distributed energy storage scheduling problem, the coordinated control mode between multi-level energy storage control is not taken into account. The capacity of a single energy storage power station is small and it is difficult to form an effective control resource. Therefore, the balance analysis and scheduling control of the power grid partition should focus on the overall charging and discharging characteristics of multiple power stations after aggregation in the entire partition or section. At present, considering the transmission capacity limitations of multi-level sections, there is a lack of effective aggregation means to fit the overall charging and discharging capacity of energy storage resources, which leads to a lack of effective scheduling means to achieve centralized optimization of large-scale distributed energy storage.

[0004] As a means of aggregation, feasible domain projection usually projects the internal safe operation feasible range to the outside when constructing the feasible domain. Minkowski method is used as an aggregation method to characterize the feasible domain, and the vertices of the sub-constraints are summed up to find the outer vertices; vertex enumeration method is used as an aggregation method to obtain the boundary of the feasible domain in low-dimensional space by obtaining the extreme value of each projection variable in the extreme direction. The above methods are not suitable for provincial and local coordinated energy storage aggregation, which will lead to a long aggregation time for energy storage and thus a long scheduling time. Summary of the invention

[0005] The purpose of this application is to provide a provincial and local coordinated energy storage scheduling method, device, equipment, medium and product, which can reduce the time of provincial and local coordinated energy storage scheduling.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides a provincial and local coordinated energy storage scheduling method, comprising:

[0008] Construct a distributed energy storage operation and dispatching model;

[0009] Aggregating the distributed energy storage operation and scheduling model to obtain a peak capacity characterization model for energy storage in the region;

[0010] Constructing the cost function constraints and the safe operation constraints within the region; the safe operation constraints within the region include the power balance constraints within the region, the safety constraints of the interconnection lines, the operation constraints of the thermal power units, the output constraints of new energy sources and the energy storage operation constraints; the energy storage operation constraints include the distributed energy storage operation scheduling model and the peak capacity characterization model of the energy storage within the region;

[0011] Redundancy identification and elimination of the cost function constraints within the area and the safety operation constraints within the area are performed to obtain the constraints after redundancy is removed;

[0012] Based on the constraints after removing the redundancy, the dual outer cutting algorithm is used to perform polyhedron projection to obtain a feasible domain set, which includes the feasible domains after projection at multiple moments; the projected feasible domain includes the peak capacity of energy storage in the region and the power input to the region through the interconnection line;

[0013] Construct a provincial power grid optimization operation and dispatching model;

[0014] The dispatching strategy at each moment is obtained according to the feasible domain set and the provincial power grid optimal operation dispatching model;

[0015] The KA-DDPG algorithm is used to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model, which is used to schedule provincial and local collaborative energy storage.

[0016] In a second aspect, the present application provides a provincial-region coordinated energy storage scheduling device, the provincial-region coordinated energy storage scheduling device comprising:

[0017] Distributed energy storage operation and scheduling model construction module, used to construct a distributed energy storage operation and scheduling model;

[0018] An aggregation module, used for aggregating the distributed energy storage operation scheduling model to obtain a peak capacity characterization model of energy storage in the region;

[0019] A constraint construction module is used to construct a cost function constraint within the region and a safe operation constraint within the region; the safe operation constraint within the region includes a power balance constraint within the region, a tie line safety constraint, a thermal power unit operation constraint, a new energy output constraint, and an energy storage operation constraint; the energy storage operation constraint includes a distributed energy storage operation scheduling model and an energy storage peak capacity characterization model within the region;

[0020] A redundancy identification and elimination module, used for redundancy identification and elimination of the cost function constraints within the area and the safety operation constraints within the area to obtain constraints after redundancy is removed;

[0021] A projection module is used to perform polyhedron projection based on the constraints after removing redundancy by using a dual outer cutting algorithm to obtain a feasible domain set, wherein the feasible domain set includes feasible domains after projection at multiple moments; the projected feasible domain includes the peak energy storage capacity in the region and the power input to the region through the interconnection line;

[0022] A provincial power grid optimization operation and dispatching model construction module is used to construct a provincial power grid optimization operation and dispatching model;

[0023] The dispatch strategy generation module is used to obtain the dispatch strategy at each moment based on the feasible domain set and the provincial power grid optimization operation dispatch model;

[0024] The training module is used to use the KA-DDPG algorithm to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model, and the scheduling model is used to schedule provincial and local collaborative energy storage.

[0025] In a third aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the provincial and local collaborative energy storage scheduling methods described above.

[0026] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the provincial and local collaborative energy storage scheduling methods described above.

[0027] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the provincial and local collaborative energy storage scheduling methods described above.

[0028] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0029] The present application provides a provincial-local collaborative energy storage scheduling method, device, equipment, medium and product, which obtains the constraints after removing the redundancy by redundancy identification and elimination of the cost function constraints and the safe operation constraints within the area, and adopts the dual outer cutting algorithm to perform polyhedron projection based on the constraints after removing the redundancy, thereby solving the problem that the existing distributed energy storage aggregation problem contains a large number of redundant constraints, which increases the computational complexity and difficulty, and adopts the dual outer cutting algorithm to perform polyhedron projection to further reduce the energy storage aggregation time, thereby reducing the time of provincial-local collaborative energy storage scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0031] Figure 1 This is an application environment diagram of a provincial and local collaborative energy storage scheduling method in an embodiment of the present application;

[0032] Figure 2 A flow chart of a provincial and local collaborative energy storage scheduling method provided in one embodiment of the present application;

[0033] Figure 3 A schematic diagram of the principle of a provincial and regional coordinated energy storage scheduling method provided in one embodiment of the present application;

[0034] Figure 4 A general flow chart of a provincial and local coordinated energy storage scheduling method provided in another embodiment of the present application;

[0035] Figure 5 A framework diagram of the KA-DDPG algorithm provided in one embodiment of the present application;

[0036] Figure 6 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in 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.

[0038] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0039] The provincial and local coordinated energy storage scheduling method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store data that the server 104 needs to process. The data storage system can be set up separately, or it can be integrated on the server 104, or it can be set up on the cloud or other servers. Server 104 constructs a distributed energy storage operation scheduling model; aggregates the distributed energy storage operation scheduling model to obtain a regional energy storage peak capacity characterization model; constructs regional cost function constraints and regional safety operation constraints; the regional safety operation constraints include regional power balance constraints, interconnection line safety constraints, thermal power unit operation constraints, new energy output constraints and energy storage operation constraints; the energy storage operation constraints include distributed energy storage operation scheduling models and regional energy storage peak capacity characterization models; the regional cost function constraints and the regional safety operation constraints are redundantly identified and eliminated to obtain the estimated bundle; based on the constraints after removing the redundancy, the dual outer cutting algorithm is used to perform polyhedron projection to obtain a feasible domain set, and the feasible domain set includes the feasible domains after projection at multiple moments; the projected feasible domain includes the peak capacity of energy storage in the region and the power input to the region through the interconnection line; a provincial power grid optimization operation and dispatching model is constructed; the dispatching strategy at each moment is obtained according to the feasible domain set and the provincial power grid optimization operation and dispatching model; the KA-DDPG algorithm is used to train the neural network model according to the feasible domain set and the dispatching strategy at each moment to obtain a dispatching model, and the dispatching model is used to dispatch provincial and local collaborative energy storage. In addition, in some embodiments, the provincial and local collaborative energy storage scheduling method can also be implemented separately by the server 104 or the terminal 102, such as the terminal 102 can directly schedule the provincial and local collaborative energy storage, or the server 104 can obtain the provincial and local collaborative energy storage from the data storage system and schedule the provincial and local collaborative energy storage.

[0040] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0041] In an exemplary embodiment, Figure 2 As shown, considering the provincial and local coordinated dispatching mode of the power grid, it is necessary to aggregate the energy storage under the jurisdiction of the local dispatching into equivalent centralized energy storage to participate in the dispatching optimization of the provincial dispatching. Based on this, a provincial and local coordinated energy storage dispatching method is provided. The method is executed by a computer device, which can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1The server 104 in the example is used as an example to illustrate the method, which includes the following steps 201 to 208. Among them:

[0042] Step 201: Construct a distributed energy storage operation and scheduling model.

[0043] Step 202: Aggregate the distributed energy storage operation scheduling model to obtain a model representing the peak capacity of energy storage in the region.

[0044] Step 203: Constructing the intra-region cost function constraints and intra-region safety operation constraints. The intra-region safety operation constraints include intra-region power balance constraints, interconnection line safety constraints, thermal power unit operation constraints, new energy output constraints and energy storage operation constraints; the energy storage operation constraints include a distributed energy storage operation scheduling model and an intra-region energy storage peak capacity characterization model.

[0045] Step 204: Redundancy identification and elimination of the intra-region cost function constraints and the intra-region safe operation constraints are performed to obtain redundancy-free constraints. Because the operation constraints within the power grid partition can be expressed in linear form, and the power grid optimization problem has a large number of optimization variables, and also contains a large number of redundant constraints, which increases the complexity and difficulty of the calculation, making the time for energy storage aggregation long, and thus leading to a long scheduling time. Therefore, the present application performs redundancy identification and elimination of the intra-region cost function constraints and the intra-region safe operation constraints to obtain redundancy-free constraints, and realizes the dimensionality reduction preprocessing of the constraints within the power grid partition, which can reduce the time for provincial and local collaborative energy storage aggregation, and thus save the time for provincial and local collaborative energy storage scheduling.

[0046] Step 205: Based on the constraints after removing redundancy, a dual outer cutting algorithm is used to perform polyhedron projection to obtain a feasible domain set. The feasible domain set includes feasible domains after projection at multiple moments; the projected feasible domain includes the peak energy storage capacity in the region and the power input to the region through the interconnection line. The present application uses a dual outer cutting algorithm to perform polyhedron projection based on the constraints after removing redundancy to obtain the projected feasible domain, project the internal safe operating conditions of the power grid partition into the interval interconnection line, and characterize the peak energy storage capacity in the region, avoiding the problems of repeated boundary generation and missing vertices caused by the traditional fixed-point search method.

[0047] Step 206: Construct a provincial power grid optimization operation and dispatching model.

[0048] Step 207: Obtain the dispatch strategy at each moment according to the feasible domain set and the provincial power grid optimization operation dispatch model.

[0049] Step 208: Use the KA-DDPG algorithm to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model, which is used to schedule the provincial and local coordinated energy storage. Due to the timing coupling between the energy storage operation and the climbing constraints of the thermal power units, this application solves the optimization problem in the continuous space based on the KA-DDPG algorithm, which solves the problem of algorithm convergence and even optimization failure.

[0050] In practical applications, the distributed energy storage operation and scheduling models include:

[0051] The charging and discharging capacity of energy storage is subject to the maximum and minimum output technical constraints, as shown in the following formula:

[0052]

[0053] In the formula, and They are respectively the energy storage devices i in region g. ess Minimum and maximum technical output limits; and They are respectively the energy storage devices i in region g. ess The discharge and charge amounts at time t.

[0054] At the same time, energy storage needs to meet the constraints of simultaneous charging and discharging:

[0055]

[0056] The energy storage SOC is limited by the energy storage capacity. When the energy storage is charged, it absorbs active power, and the SOC increases; when it is discharged, it emits active power, and the SOC decreases. The energy storage SOC value calculation formula at time t in area g is:

[0057]

[0058] In the formula, The energy storage device i in the g area ess SOC value at time t; and They are respectively the energy storage devices i in region g. ess The charging and discharging efficiency; The energy storage device i in the g area ess capacity.

[0059] The energy storage SOC constraint is as follows:

[0060]

[0061] In the formula, and They are respectively the energy storage devices i in region g. ess The lower and upper limits of SOC.

[0062] The starting and ending SOC of short-term energy storage within a day need to be equal, which can be expressed as follows:

[0063]

[0064] In another exemplary embodiment of the present application, due to the existence of nonlinear constraints such as formula (3), in order to facilitate the subsequent feasible domain projection, the nonlinear constraints need to be simplified. When considering the intra-regional collaborative optimization problem, the arbitrage profit of energy storage is taken into account. For energy storage, it is usually discharged when the grid-connected electricity price is high and charged when the grid-connected electricity price is low. If the energy storage does not satisfy formula (3), it means that the energy storage buys electricity at a high price when the electricity price is high and sells electricity at a low price when the electricity price is low, which does not meet the profit-seeking nature of the energy storage operator. Therefore, according to the optimality of the model, formula (3) is always satisfied and can be directly simplified. Therefore, the distributed energy storage operation and scheduling model is aggregated to obtain the intra-regional energy storage peak capacity characterization model, which specifically includes:

[0065] The nonlinear constraints of the distributed energy storage operation scheduling model are simplified to obtain a simplified distributed energy storage operation scheduling model.

[0066] The simplified distributed energy storage operation scheduling model is aggregated to obtain a model representing the peak capacity of energy storage in the region.

[0067] In practical applications, the distributed energy storage operation scheduling model is aggregated to obtain a peak capacity characterization model for energy storage in the region, specifically:

[0068] Based on the aggregation effect, distributed energy storage devices are aggregated into energy storage represented by centralized parameters, forming four equivalent parameters: equivalent centralized rated power, equivalent centralized rated capacity, and equivalent centralized charging and discharging efficiency. Assuming that all distributed energy storage devices have the same charging and discharging time at rated power, the corresponding equivalent centralized energy storage parameter representation is as follows:

[0069] Equivalent concentrated rated power:

[0070]

[0071] Where: The energy storage device i in the g area ess Rated power, is the equivalent concentrated energy storage rated power in area g, Ω g,ESS It is the collection of all energy storage devices in area g.

[0072] Equivalent concentrated rated capacity:

[0073]

[0074] Where: Q g,conis the equivalent centralized energy storage rated capacity in area g.

[0075] Equivalent centralized charging and discharging efficiency:

[0076]

[0077] Where: and are the equivalent concentrated energy storage charging and discharging efficiency parameters in region g, respectively.

[0078] Therefore, the peak capacity of energy storage in region g at time t can be expressed as follows:

[0079]

[0080] Where: is the peak capacity of energy storage in region g at time t.

[0081] Introducing slack variables Convert the original maximization problem into a constrained form:

[0082]

[0083] Where: is the relaxation variable that characterizes the peak capacity in region g at time t.

[0084] In practical applications, the process of determining the cost function constraints within the region specifically includes:

[0085] First, construct the cost function F within the region g , which can be expressed as follows:

[0086] F g =C(P th )+C(P ESS_ch ,P ESS_dis ) (12)

[0087]

[0088] Where: C(P th ) is the power generation cost of thermal power units, C(P ESS_ch ,P ESS_dis ) is the energy storage operation cost, They are respectively thermal power units i in region g th The first, second and third coefficients of power generation cost, Ω g,th is the thermal power unit and photovoltaic unit in region g, is the thermal power unit i in region g at time t th The output; λ buy With λ sell are the buying and selling electricity prices of the power grid respectively.

[0089] C(P th ) is linearized by Taylor series expansion, C(P th )exist The Taylor expansion of is as follows:

[0090]

[0091] In the formula, Indicates that Taylor series expansion is performed at this point.

[0092] Then introduce the slack variable Convert the original minimization problem into a constraint expression to obtain the cost function constraint within the region:

[0093]

[0094] Where: is the slack variable that represents the minimum cost in region g at time t.

[0095] In practical applications,

[0096] The power balance constraint within the area is expressed as follows:

[0097]

[0098] Where: is the thermal power unit i in region g at time t th contribution; is the photovoltaic power i in region g at time t pv contribution; is the wind turbine i in region g at time t wt contribution; is the load node i in region g at time t load load demand; is the regional tie line node i in region g at time t bus The power transmitted by the section interconnection line; Ω g,th ,Ω g,PV ,Ω g,WT ,Ω g,load ,Ω g,B They are the thermal power group set and photovoltaic group set in area g, the wind power group set, the load node set, and the section interconnection line node set.

[0099] Tie line safety constraints are expressed in the following form:

[0100]

[0101] Where: They are respectively thermal power units i in region g thFor the power transfer distribution factor of transmission line k, the energy storage device i in region g ess For the power transfer distribution factor of transmission line k, the photovoltaic power generation capacity of region g is pv For the power transfer distribution factor of transmission line k, the wind turbine i in region g wt For the power transfer distribution factor of transmission line k, the load node i in region g load For the power transfer distribution factor of transmission line k, the g area interval tie line node i bus The power transfer distribution factor for transmission line k. P g,k and are the lower and upper limits of power transmission of transmission line k respectively.

[0102] The operation constraints of thermal power units are expressed as follows, where equation (19) is the output constraint of the thermal power unit and equation (20) is the ramp constraint of the thermal power unit:

[0103]

[0104] Where: and They are respectively thermal power units i in region g th The lower and upper limits of output; and They are respectively thermal power units i in region g th The maximum uphill climbing power and downhill climbing power limit.

[0105] The output constraints of new energy sources are as follows:

[0106]

[0107] Where: is the photovoltaic area i pv Output limit; is the wind turbine i in region g wt output limit.

[0108] In another exemplary embodiment of the present application, redundancy identification and elimination of the intra-area cost function constraint and the intra-area safe operation constraint are performed to obtain the constraints after redundancy removal, specifically including:

[0109] The Llewellyn method is used to identify and eliminate redundancy of the cost function constraints and the safe operation constraints within the area to obtain the constraints after redundancy is removed.

[0110] In practical applications, S can be used to represent the feasible domain of linear programming problems:

[0111]

[0112] x is the projected vector set R n represents an n-dimensional real number set, A represents the coefficient matrix before the variables in the constraint after removing redundancy, and b represents the constant matrix of the constraint. b, A are -1 and 1 respectively; thermal power unit climbing constraint b, A are -1 and 1 respectively; in the new energy output constraint b, A are 1 and 1 respectively; b in the cost function constraint within the region is:

[0113]

[0114] , A is In the power balance constraint within the area, b is 0, A is 1, and in the tie line safety constraint, P g,k and is b, A is -1 and 1.

[0115] Defining the yth constraint is redundant, so the constraint form after removing the redundancy can be expressed as follows:

[0116]

[0117] Among them, A i represents the coefficient matrix before the variable in the i-th constraint, x j represents the jth variable, b i represents the constant matrix of the i-th constraint, m represents the number of constraints, and n represents the number of variables.

[0118] Redundant constraints take the following forms:

[0119]

[0120] Among them, a y represents the pre-variable coefficient matrix of the y-th constraint, b y Represents the constant term for the yth constraint.

[0121] Therefore, the Llewellyn method is used to identify and eliminate the redundancy of the cost function constraints and the safety operation constraints in the area to obtain the constraints after removing the redundancy. The specific steps are:

[0122] 1) Compare the yth constraint with the rth constraint. If b y and b r are all greater than or equal to 0, y=1,…,n,r=1,…,n, and y≠r, then go to 2), otherwise go to 3).

[0123] 2) but

[0124] 3) but

[0125] Among them, a yj represents the coefficient before the jth variable of the yth constraint, a rj represents the j-th variable pre-coefficient of the r-th constraint.

[0126] 4) If all constraints are identified, output the constraints S after removing the redundancies. y =SS r ; If not all constraints have been identified, return to 1).

[0127] In practical applications, based on the constraints after removing the redundancy, the dual outer cutting algorithm is used to perform polyhedron projection to obtain the feasible domain after projection at time t, which specifically includes:

[0128] According to equations (1) to (11), (18) to (25), the feasible region for safe operation at time t in region g can be expressed as follows:

[0129]

[0130] Since the above constraints are all linear constraints, they can be simplified into the following form:

[0131]

[0132] Where: A and b are obtained by projecting through equation (26).

[0133] Construct an optimal problem to characterize the above feasible domain:

[0134]

[0135] Where: 1 T represents the transpose of 1, 1 represents the identity matrix, x and y are meaningless variables, z is the slack variable vector, B′ represents the coefficient matrix before variable y; I is a 1 vector with compatible dimensions, and b′ represents a constant matrix; is the dual variable, x is the set of projected vectors y is the set of variables in the region If the optimization result of formula (28) is 0, the feasible region (27) exists, and the vertex Λ(x) is not empty, then the dual problem of formula (28) can be expressed as follows:

[0136]

[0137] In the formula, Express Find the transpose, Ξ represents the feasible domain set, B′ TIt means to transpose B′.

[0138] Therefore, the original feasible domain can be expressed as follows:

[0139]

[0140] Λ(Ξ) represents any feasible domain set, if the optimal solution for x Then it must exist Make The corresponding hyperplane is Based on this, we only need to find And remove it from the feasible region, so that the feasible region of region g is obtained.

[0141] So first construct a large enough set Then identify all and from it Remove from the , and the final result is

[0142]

[0143] If the maximum value If it is greater than 0, it means So x * from Remove it and add a feasible cut to the original problem (31) Constrain until the maximum value of equation (31) is 0, then the feasible domain after projection is obtained That is, solving equation (31) yields the feasible domain after projection of region g at time t:

[0144] In practical applications, facing the peak demand of the power grid, a provincial power grid optimization operation and dispatching model considering cross-regional mutual assistance is constructed, which can be expressed as follows:

[0145]

[0146] Where: C(P load_cut ) is about the variable The cost function, Ω B is the set of all regions, λ load_cut Unit cost of shedding load for the system; is the load shedding amount in area g at time t; is the actual load demand in area g at time t; P k and are the lower and upper limits of the transmission power of the section tie line k respectively; is the power transfer distribution factor of region g for the section tie line k; is the power input to region g through the interconnection line at time t.

[0147] In practical applications, the dispatch strategy at time t is obtained based on the projected feasible domain at time t and the provincial power grid optimization operation dispatch model. Specifically, the initial state is input, that is, the peak capacity of energy storage in the area at the current time ξ g,t , select a proxy action, that is Solve the optimal solution through the optimization model (provincial power grid optimization operation and dispatch model) That is, the scheduling strategy.

[0148] In practical applications, a well-known algorithm, Kinematic Awareness Deep Deterministic Policy Gradients (KA-DDPG) algorithm, is used to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model. The scheduling model is used to schedule provincial and local collaborative energy storage, specifically including:

[0149] Since the operation of energy storage and the ramp constraints of thermal power units are coupled in time series, the cross-regional scheduling problem of energy storage can be modeled as a dynamic programming problem on a daily time scale. This application uses the KA-DDPG algorithm to solve the dynamic programming problem of continuous action space, where the KA-DDPG algorithm framework diagram is shown in the figure below: Figure 5 As shown:

[0150] Establish the Bellman equation for the optimal action-value function:

[0151]

[0152] Where: Q * (s t ,a t ) is the optimal action value function at time t, Q * (s t+1 ,a t+1 ) is the optimal action value function at time t+1, Represents the state s at time t+1 t+1 Satisfies the expectation of P distribution, r(s t ,a t ) is the state s at time t t Take action a at time t t The reward function is the peak energy storage capacity ξ in the g region. g,t , action is the contact line is the standard action in the KA-DDPG algorithm; γ is the discount factor; the action-value function value Q at time t is calculated using a neural network θ(s t ,a t ) for estimation.

[0153] The following is the algorithm flow:

[0154] get Then, select Execute And calculate the guidance reward. The protector determines whether the standard action is safe based on the standard action and the guidance reward. If the protector determines that the action is unsafe or has too high a risk, it refuses to perform the action and imposes a negative reward γ t , or the protector accepts this execution action and gets the final reward value γ t .

[0155] The following describes the offline training and update process of the Critic network and the Actor network:

[0156] Input the initial Critic network loss function parameter θ and the Actor network function The state is the peak energy storage capacity of the area at the current time ξ g,t Enter the current Actor network to obtain the action, that is, the current contact line Plan, interact the state and action input environment to obtain the next state of the agent ξ g,t+1 , reward value γ t ', termination indicator value α, Put it into experience replay pool D.

[0157] Randomly sample m from the experience replay pool Tuple, defining the mean squared Bellman error function to measure Q θ (s t ,a t ) satisfies the Bellman equation, and the parameter θ of the Q function formula (33) is updated with the goal of minimizing the loss function L, where the mean square Bellman error function is used as the loss function L, which is expressed as follows:

[0158]

[0159] Where: E[] represents the expectation, r represents the reward, α is a 0-1 variable, indicating s t+1 Whether the terminal state is reached, Q θ (s t ,a t ) represents the action-value function value at time t, Q θ (s t+1 ,a t+1 ) represents the action-value function value at time t+1.

[0160] Find the minimum value of the loss function L in the gradient direction of θ, calculate the gradient of L in the direction of θ, and update the parameter θ along the gradient direction.

[0161] In order to learn to maximize Q θ (s t ,a t ) deterministic strategy, using the gradient descent method to update the action network parameters, then the current Actor network for the parameters The gradient of is:

[0162]

[0163] For the Critic network loss function parameter θ, and the Actor network function Update in the gradient direction:

[0164]

[0165] Where: ρ∈[0,1] is the update coefficient, which is the weighted average of the original target network parameters and the current network parameters.

[0166] When t reaches T, a training set ends. Repeat the above offline training steps until the number of training sets reaches the maximum training set umax, the offline training process ends, and the optimal neural network model at this time is saved.

[0167] After training is completed, the optimal neural network model can be applied online. The intelligent agent directly interacts with the environment according to the learned strategy and performs actions to achieve the task objectives, that is, energy storage scheduling is achieved according to the optimal neural network model.

[0168] like Figure 3 and Figure 4 As shown, this application first proposes a method for characterizing the peak capacity of energy storage under the aggregation effect, and simplifies the energy storage constraints through optimality conditions; secondly, a redundant constraint identification method based on the Llewellyn method is proposed to perform dimensionality reduction preprocessing on the constraints within the power grid partition; then, polyhedron projection is performed based on the dual outer cutting algorithm, and the internal safe operating conditions of the power grid partition and the peak capacity of energy storage in the area are projected into the interval interconnection line; finally, based on the KA-DDPG algorithm, a grid-side energy storage intraday optimization scheduling method for power grid peak demand is proposed.

[0169] The present application also provides an application scenario, which applies the above-mentioned provincial-local collaborative energy storage scheduling method. Specifically: The provincial-local collaborative energy storage scheduling method provided in this embodiment can be applied in the provincial-local collaborative energy storage scheduling scenario facing the peak demand of the provincial network. The provincial-local collaborative energy storage scheduling method provided in this embodiment belongs to the process of pre-training the scheduling model, and then energy storage scheduling is performed according to the scheduling model.

[0170] Based on the same inventive concept, the embodiment of the present application also provides a provincial-local coordinated energy storage scheduling device for implementing the provincial-local coordinated energy storage scheduling method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more provincial-local coordinated energy storage scheduling device embodiments provided below can refer to the limitations of the provincial-local coordinated energy storage scheduling method above, and will not be repeated here.

[0171] In an exemplary embodiment, a land-saving coordinated energy storage scheduling device is provided, comprising:

[0172] The distributed energy storage operation and scheduling model construction module is used to construct the distributed energy storage operation and scheduling model.

[0173] The aggregation module is used to aggregate the distributed energy storage operation scheduling model to obtain a peak capacity characterization model of energy storage in the area.

[0174] The constraint construction module is used to construct the cost function constraints within the area and the safe operation constraints within the area; the safe operation constraints within the area include the power balance constraints within the area, the safety constraints of the interconnection lines, the operation constraints of the thermal power units, the output constraints of new energy sources and the energy storage operation constraints; the energy storage operation constraints include the distributed energy storage operation scheduling model and the energy storage peak capacity characterization model within the area.

[0175] The redundancy identification and elimination module is used to identify and eliminate the redundancy of the cost function constraints within the area and the safe operation constraints within the area to obtain the constraints after the redundancy is removed.

[0176] The projection module is used to obtain a feasible domain set by polyhedron projection using a dual outer cutting algorithm based on the constraints after removing redundancy, wherein the feasible domain set includes feasible domains after projection at multiple moments; the projected feasible domain includes the peak energy storage capacity in the region and the power input to the region through the interconnection line.

[0177] The provincial power grid optimization operation and dispatching model construction module is used to construct the provincial power grid optimization operation and dispatching model.

[0178] The dispatch strategy generation module is used to obtain the dispatch strategy at each moment based on the feasible domain set and the provincial power grid optimization operation dispatch model.

[0179] The training module is used to use the KA-DDPG algorithm to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model, and the scheduling model is used to schedule provincial and local collaborative energy storage.

[0180] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 6 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store provincial and local collaborative energy storage scheduling data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a provincial and local collaborative energy storage scheduling method is implemented.

[0181] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0182] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0183] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0184] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0185] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0186] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0187] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0188] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0189] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A provincial and local coordinated energy storage scheduling method, characterized in that: The provincial and local coordinated energy storage scheduling method comprises: Construct a distributed energy storage operation and dispatching model; Aggregating the distributed energy storage operation and scheduling model to obtain a peak capacity characterization model for energy storage in the region; Constructing the cost function constraints and the safe operation constraints within the region; the safe operation constraints within the region include the power balance constraints within the region, the safety constraints of the interconnection lines, the operation constraints of the thermal power units, the output constraints of new energy sources and the energy storage operation constraints; the energy storage operation constraints include the distributed energy storage operation scheduling model and the peak capacity characterization model of the energy storage within the region; Redundancy identification and elimination of the cost function constraints within the area and the safety operation constraints within the area are performed to obtain the constraints after redundancy is removed; Based on the constraints after removing the redundancy, the dual outer cutting algorithm is used to perform polyhedron projection to obtain a feasible domain set, which includes the feasible domains after projection at multiple moments; the projected feasible domain includes the peak capacity of energy storage in the region and the power input to the region through the interconnection line; Construct a provincial power grid optimization operation and dispatching model; The dispatching strategy at each moment is obtained according to the feasible domain set and the provincial power grid optimal operation dispatching model; The KA-DDPG algorithm is used to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model, which is used to schedule provincial and local collaborative energy storage.

2. The provincial and local coordinated energy storage dispatching method according to claim 1 is characterized in that: The distributed energy storage operation scheduling model is specifically: in, The energy storage device i in the g area ess Minimum technical output limit, The energy storage device i in the g area ess Maximum technical output limit, is the energy storage device i in area g ess The discharge amount at time t is The energy storage device i in the g area ess The charge at time t is The energy storage device i in the g area ess The SOC value at time t is: The energy storage device i in the g area ess The charging efficiency, is the energy storage device i in area g ess The discharge efficiency, The energy storage device i in the g area ess The capacity, is the energy storage device i in area g ess The lower limit of SOC, The energy storage device i in the g area ess SOC upper limit.

3. The provincial and local coordinated energy storage dispatching method according to claim 1, characterized in that: The peak capacity characterization model of energy storage in the region is specifically: in, is the equivalent concentrated energy storage rated power in area g, Ω g,ESS is the set of all energy storage devices in area g, The energy storage device i in the g area ess Rated power, Q g,con is the equivalent centralized energy storage rated capacity in region g, The energy storage device i in the g area ess The capacity, is the equivalent concentrated energy storage discharge efficiency parameter in region g, The energy storage device i in the g area ess The discharge efficiency, is the equivalent centralized energy storage charging efficiency parameter in region g, The energy storage device i in the g area ess The charging efficiency, is the peak capacity of energy storage in region g at time t, is the relaxation variable that characterizes the peak capacity in region g at time t.

4. The provincial and local coordinated energy storage dispatching method according to claim 1 is characterized in that: The provincial power grid optimization operation and dispatching model is specifically as follows: quantity The cost function, Ω B is the set of all regions, λ load_cut is the unit cost of system load shedding, is the load shedding amount in area g at time t, is the power input to region g through the tie line at time t, ξ g,t is the peak energy storage capacity in region g at time t, represents the feasible domain after projection of region g at time t, P k is the lower limit of transmission power of the section tie line k, is the power transfer distribution factor of area g for the section tie line k at time t, is the upper limit of transmission power of the section tie line k, is the actual load in region g at time t, is the relaxation variable that characterizes the peak capacity in region g at time t.

5. The provincial and local coordinated energy storage dispatching method according to claim 1, characterized in that: The distributed energy storage operation scheduling model is aggregated to obtain the peak capacity characterization model of energy storage in the region, which specifically includes: Simplifying the nonlinear constraints of the distributed energy storage operation scheduling model to obtain a simplified distributed energy storage operation scheduling model; The simplified distributed energy storage operation scheduling model is aggregated to obtain a model representing the peak capacity of energy storage in the region.

6. The provincial and local coordinated energy storage dispatching method according to claim 1, characterized in that: The redundancy identification and elimination of the cost function constraint within the area and the safe operation constraint within the area to obtain the redundancy-removed constraint specifically includes: The Llewellyn method is used to identify and eliminate redundancy of the cost function constraints and the safe operation constraints within the area to obtain the constraints after redundancy is removed.

7. A provincial and local coordinated energy storage dispatching device, characterized in that: The provincial and local coordinated energy storage dispatching device comprises: Distributed energy storage operation and scheduling model construction module, used to construct a distributed energy storage operation and scheduling model; An aggregation module, used for aggregating the distributed energy storage operation scheduling model to obtain a peak capacity characterization model of energy storage in the region; A constraint construction module is used to construct a cost function constraint within the region and a safe operation constraint within the region; the safe operation constraint within the region includes a power balance constraint within the region, a tie line safety constraint, a thermal power unit operation constraint, a new energy output constraint, and an energy storage operation constraint; the energy storage operation constraint includes a distributed energy storage operation scheduling model and an energy storage peak capacity characterization model within the region; A redundancy identification and elimination module, used for redundancy identification and elimination of the cost function constraints within the area and the safety operation constraints within the area to obtain constraints after redundancy is removed; A projection module is used to perform polyhedron projection based on the constraints after removing redundancy by using a dual outer cutting algorithm to obtain a feasible domain set, wherein the feasible domain set includes feasible domains after projection at multiple moments; the projected feasible domain includes the peak energy storage capacity in the region and the power input to the region through the interconnection line; A provincial power grid optimization operation and dispatching model construction module is used to construct a provincial power grid optimization operation and dispatching model; The dispatch strategy generation module is used to obtain the dispatch strategy at each moment based on the feasible domain set and the provincial power grid optimization operation dispatch model; The training module is used to use the KA-DDPG algorithm to train the neural network model according to the feasible domain set and the scheduling strategy at each moment to obtain a scheduling model, and the scheduling model is used to schedule provincial and local collaborative energy storage.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the provincial and local collaborative energy storage scheduling method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the provincial and local coordinated energy storage scheduling method described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the provincial and local coordinated energy storage scheduling method described in any one of claims 1 to 6 is implemented.

Images