Flexible resource adjustment characteristic modeling and aggregation method

Abstract

The invention discloses a flexibility resource adjustment characteristic modeling and aggregation method, which comprises the following steps of: collecting a virtual battery model coefficient matrix reported by each flexibility resource individual by a similar resource aggregation platform, then calculating an element polyhedron coefficient and sending the element polyhedron coefficient to each individual; calculating parameters by each individual, calculating a final equivalent aggregate by the similar resource aggregation platform according to the parameters, and sending the final equivalent aggregate to a scheduling center for decision making; and finally, deaggregating the decision result, and issuing the decision result to each individual for execution. According to the method, the consistency of group behaviors is ensured, and each individual can be controlled to meet the constraint as long as the final equivalent polymer is controlled to be charged and discharged asynchronously. Therefore, the actual behavior of the energy storage system is reflected more accurately, the overall utilization efficiency of flexible resources is improved, and the coping capacity of a power system for the fluctuation and uncertainty of renewable energy sources is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of electrical engineering and automation thereof, and in particular to a flexibility resource regulation characteristic modeling and aggregation method. Background Art

[0002] As the penetration of renewable energy in the power system continues to increase, the role of flexible resources such as energy storage systems and electric vehicles in maintaining grid stability and reliability becomes increasingly important. These resources provide important support for the balance of power supply and demand, and effectively alleviate the volatility and uncertainty brought by renewable energy. However, in practical applications, how to effectively integrate and utilize these flexible resources, especially in terms of their regulation characteristics and aggregation methods, still faces many challenges.

[0003] Existing research emphasizes the importance of detailed modeling of the regulatory characteristics of individual flexibility resources to optimize their collective performance in the power system. For example, some studies have suggested that the charging and discharging efficiency of energy storage systems directly affects the accurate prediction of their behavior and their potential contribution to grid services. Accurately modeling the charging and discharging process of energy storage systems can help power system operators better maintain the stability of the grid and improve the absorption capacity of renewable energy. In addition, aggregating many flexibility resources into virtual power plants or similar structures can also effectively improve their operating efficiency and economic feasibility. This aggregation method can integrate scattered resources into a whole, making them more competitive in the electricity market.

[0004] Therefore, at present, the configuration and control of distributed energy storage has become a research focus. Some studies have suggested that in order to achieve the best configuration strategy, the high cost of energy storage and the need to increase grid flexibility must be taken into account, which is the key to the efficient integration of renewable energy. At the same time, some studies have proposed real-time clustering algorithms that can dynamically aggregate energy storage systems into virtual power plants to improve operating efficiency and flexibility. These studies have shown that the overall performance of the power system can be significantly improved by optimizing the configuration and control of energy storage systems.

[0005] However, the aggregation methods proposed by existing studies usually rely on the use of polyhedral models to approximate the behavior of flexible resources. These methods often ignore the non-simultaneous charging and discharging constraints that energy storage systems must comply with during charging and discharging, which may lead to suboptimal or even infeasible solutions in practical applications. That is, when dealing with energy storage systems, traditional methods usually ignore the constraint that they cannot charge and discharge simultaneously, which is unrealistic in actual operation. Therefore, existing aggregation methods have obvious limitations when dealing with energy storage systems.

[0006] Research progress in recent years has focused on developing more complex aggregation technologies, which have improved the overall utility of flexibility resources to a certain extent, but have not yet solved the problem of non-simultaneous charging and discharging constraints of energy storage systems. Therefore, there is an urgent need for a new method that can ensure that each flexibility resource can comply with this key constraint during the aggregation process. Summary of the invention

[0007] In order to solve the technical problem that the constraint of not being able to charge and discharge simultaneously is ignored when aggregating flexibility resources, resulting in suboptimal or even unfeasible solutions, the present invention provides a flexibility resource regulation characteristic modeling and aggregation method that can ensure that each flexibility resource can comply with the non-simultaneous charging and discharging constraint during the aggregation process.

[0008] In order to achieve the above technical purpose, the technical solution of the present invention is:

[0009] A flexibility resource adjustment characteristic modeling and aggregation method includes the following steps:

[0010] Step 1: The virtual battery model coefficient matrix reported by each individual flexibility resource is collected by a similar resource aggregation platform, wherein the category of the flexibility resource to which each individual belongs is an energy storage system ES or an electric vehicle EV;

[0011] Step 2: The similar resource aggregation platform uses the collected coefficient matrix of each individual as the polyhedron expression, and then calculates the primitive polyhedron based on the maximum inner approximation algorithm of similar polyhedrons, and sends the calculated coefficients of the primitive polyhedron to each individual;

[0012] Step 3: The same resource aggregation platform receives the parameters independently calculated by each individual, where the parameters are calculated by each individual according to the category of the flexibility resource to which the individual belongs, using the corresponding parameter calculation method;

[0013] Step 4: The similar resource aggregation platform calculates the final equivalent aggregate based on the parameters of each individual, and sends it to the dispatch center to make a decision based on the preset decision algorithm, and then receives the decision result;

[0014] Step 5: The same resource aggregation platform de-aggregates the decision results and finally sends them to each individual for execution.

[0015] The method, in step 1, the virtual battery model coefficient matrix reported by the i-th flexibility resource individual for:

[0016]

[0017] Where K is the category of resource individuals, K∈{ES,EV}, is the decision variable, F i K , is the coefficient matrix, represents the real number space, and τ is the maximum value of the offline time interval.

[0018] In the method, in step 1, the virtual battery model coefficient matrix is ​​obtained based on the virtual battery model issued by the same resource aggregation platform, and the virtual battery model is:

[0019]

[0020] in, represents the matrix transpose, represents the charging power column vector of the i-th individual, is the charging power of the ith individual at time t, P i K,dis represents the charging power column vector of the i-th individual, is the charging power of the ith individual at time t, represents the upper bound of the charging power of the ith individual, T represents the offline time interval, T={1,2,…,τ}, represents the upper bound of the discharge power of the ith individual, represents the maximum downward climbing rate of the i-th individual, represents the power of the ith individual at time t, represents the maximum upward climbing rate of the i-th individual, represents the lower bound of the energy value of the ith individual at time t+1, represents the energy value of the i-th individual at time t+1, Represents the upper bound of the energy value of the i-th individual at time t+1.

[0021] In the method, in step 1, the virtual battery model is obtained by unified modeling based on individual models of ES and EV and adjustment characteristics.

[0022] The individual model of ES is:

[0023]

[0024]

[0025]

[0026]

[0027]

[0028] in, is the charging power of the i-th ES at time t, is the discharge power of the i-th ES at time t, is the upper bound of the charging power of the i-th ES at time t, is the upper bound of the discharge power of the i-th ES at time t; is the external power of the i-th ES; is the maximum downward climbing rate of the i-th ES, is the maximum upward climbing rate of the i-th ES; is the energy stored in the i-th ES, is the self-discharge rate, Δt is the time interval, is the charging efficiency of the i-th ES, is the discharge efficiency of the i-th ES; is the upper bound of the energy allowed for the i-th ES, is the lower bound of the energy allowed for the i-th ES;

[0029] The individual model of EV is:

[0030] When the EV is charging,

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] in, is the charging power of the i-th EV at time t, is the discharge power of the i-th EV at time t; is the upper bound of the charging power of the i-th EV, is the upper bound of the discharge power of the i-th EV; is the external power of the i-th EV; is the maximum downward climbing rate, is the maximum upward climbing rate; is the energy of the i-th EV at time t, is the self-discharge rate of the i-th EV at time t, and Δt is the time interval; is the charging efficiency of the i-th EV at time t, is the discharge efficiency of the i-th EV at time t; is the upper bound of the energy allowed for the i-th EV, is the lower bound of the energy allowed for the i-th EV, is the expected energy value of the i-th EV at time t+1;

[0037] When the EV is not charging,

[0038]

[0039] In the method, in step 2, calculating the primitive polyhedron includes:

[0040] The coefficient matrix of the virtual battery model As a polyhedral representation, take the average value of each category of flexibility resources As a primitive polyhedron:

[0041]

[0042] in, F i K The average value of N K is the number of individuals of a certain type of flexibility resource, for The average value of Decision variables representing primitive polyhedra;

[0043] Then the same resource aggregation platform will and Sent to each individual.

[0044] In the method, in step 3, when each individual calculates the parameter independently, when the individual is ES, the parameter is calculated by the following formula:

[0045]

[0046] in Minimize st represents the constraint condition, represents the scaling factor conversion coefficient, G represents a non-negative matrix, ri K represents the translation factor conversion vector;

[0047] When the individual is EV, the parameter is calculated by the following formula:

[0048]

[0049] In the method, in step 3, the conversion between the scaling factor and the translation factor is performed by the following variable substitution:

[0050]

[0051]

[0052] in, represents the translation factor of the ith individual, Represents the scaling factor of the i-th individual; and after the calculation is completed, the individual sends its own scaling factor or translation factor and scaling factor to the same resource aggregation platform.

[0053] In the method, in step 4, the similar resource aggregation platform calculates the final equivalent aggregate according to the parameters of each individual:

[0054]

[0055] In the method, in step 5, the similar resource aggregation platform de-aggregates the decision results, including:

[0056] The same resource aggregation platform performs de-aggregation based on the following formula:

[0057]

[0058] in is the scaling factor, β K represents the sum of scaling factors, U K represents the aggregate decision variable, represents the sum of translation factors, is the translation factor,

[0059] Then the same resource aggregation platform will Distributed to each individual.

[0060] The technical effects of the present invention are as follows: first, by introducing non-simultaneous charging and discharging constraints, the present invention can more accurately reflect the actual behavior of the energy storage system and improve the accuracy and reliability of the model; second, the present invention adopts a de-translation approximation method to ensure that each flexibility resource meets the key constraints and achieves the optimal scheduling and utilization of resources, thereby significantly improving the overall utilization efficiency of flexibility resources and optimizing the operation of the power system; further, by aggregating flexibility resources, the present invention can effectively balance the supply and demand relationship of the power system, alleviate the volatility and uncertainty brought by renewable energy, enhance the stability and reliability of the power grid, and improve the economic benefits of the system. The present invention proposes a new de-translation approximation aggregation method based on similar polyhedrons for energy storage resources. By ensuring the consistency of group behavior, as long as the final equivalent aggregate is controlled not to charge and discharge simultaneously, each individual can be controlled to meet the constraints. Through this method, the actual behavior of the energy storage system can be more accurately reflected, the overall utilization efficiency of flexibility resources can be improved, and the ability of the power system to cope with the volatility and uncertainty of renewable energy can be enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of a flow chart of an embodiment of the present invention.

[0062] Figure 2 Schematic diagram of the upper and lower limits of power and energy of the corresponding energy storage units in the prior art; (a) is a schematic diagram of the upper limits of power and energy, and (b) is a schematic diagram of the lower limits of power and energy.

[0063] Figure 3 Schematic diagram of the upper and lower limits of power and energy of the corresponding energy storage individuals in the present invention; (a) is a schematic diagram of the upper limits of power and energy, and (b) is a schematic diagram of the lower limits of power and energy. DETAILED DESCRIPTION

[0064] In this embodiment, firstly, a single individual is modeled according to different types of flexible resources (FR), where flexible resources mainly include energy storage (ES) and electric vehicles (EV). K , K∈{ES,EV}, and given the offline time interval T={1,2,…,τ}, the models of these two types of flexibility resources are:

[0065] 1) ES:

[0066]

[0067]

[0068]

[0069]

[0070]

[0071] In the formula t∈T. Formula (1.a) represents the ES charging (discharging) power constraint, is the charging (discharging) power of the i-th ES at time t, is the upper limit of charging (discharging) power. Formula (1.b) represents the external power of ES individual Equation (1.c) represents the ramp rate constraint of ES. are the maximum downward and upward climbing rates respectively. Equation (1.d) represents the equation relationship between ES energy and power, is the energy stored in ES, is the self-discharge rate, Δt is the time interval, is the charging (discharging) efficiency. Formula (1.e) represents the energy constraint of ES, are the upper and lower bounds of the allowed energy, respectively.

[0072] 2)EV:

[0073] if t≤t i,arrive or t≥t i,leave :

[0074]

[0075] else:

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] In the formula t∈T. When the EV is not charging, t≤t i,arrive or t ≥ t i,leave When , the charge and discharge power can only be 0, as described in formula (2.a). When the EV is charging, formula (2.b) represents the EV charge (discharge) power constraint, is the charging (discharging) power of the i-th EV at time t, is the upper limit of charging (discharging) power. Formula (2.c) represents the external power of EV Equation (2.d) represents the ramp rate constraint of EV. are the maximum downward and upward climbing rates respectively. Equation (2.e) represents the equation relationship between EV energy and power, For energy, is the self-discharge rate, Δt is the time interval, is the charging (discharging) efficiency. Formula (2.f) represents the energy constraint of EV, are the upper and lower bounds of the allowed energy, respectively. Since EV leaves at time t i,leave It is necessary to ensure that its energy is greater than or equal to Therefore, we also need to consider the time from t to t i,leave At maximum power When charging, The minimum energy required to convert to time t+1

[0082] Next, the individual regulation characteristics of flexibility resources are uniformly modeled and the individual models are converted into aggregate reference models, that is, virtual battery models.

[0083] The virtual battery model is described as follows:

[0084]

[0085] In the formula are the column vectors of charge and discharge power respectively. It describes the charge and discharge power constraints, ramp constraints and energy state constraints. It can be further written in matrix form.

[0086]

[0087] Where: is the decision variable, is the coefficient matrix.

[0088] Then, we aggregate each individual. Since the general Minkowski sum is an NP-hard problem, there is no exact algorithm with polynomial time complexity, and it can only be solved approximately. The characteristics of the same type of FR are relatively similar, except for the differences in parameters. The difficulty of aggregation is mainly the high-dimensional centralized computing cost caused by the large number of individuals. Therefore, an efficient aggregation method that supports parallel computing is needed, and the maximum inner approximation method based on similar polyhedrons can meet this requirement well.

[0089] As shown in formula (4), the FR individuals are represented in the form of polyhedrons. When selecting the primitive polyhedron of each type of flexibility resource, the average value of each type is usually taken:

[0090]

[0091] Where:

[0092] And approximate each individual as a similar polyhedron of the primitive polyhedron:

[0093]

[0094] Where: is the scaling factor, is the translation factor.

[0095] Then the Minkowski sum of the same type of FR is simplified to:

[0096]

[0097] By Use Ω K express, Substitution Constraint, Ω K It can be further expressed as:

[0098]

[0099] If the similar polyhedron belongs to the original polyhedron, that is If the original polyhedron belongs to a similar polyhedron, that is, It is an external approximation.

[0100] In order to ensure the feasibility of scheduling, the inner approximation is adopted. The maximum inner approximation contained in The mathematical expression is:

[0101]

[0102] By variable substitution, The above problem can be equivalent to solving a linear programming problem:

[0103]

[0104] Among them: G ≥ 0 means that all elements are non-negative.

[0105] This aggregation method supports parallel computing. and After that, each individual can independently calculate its corresponding Then report to the aggregation platform for simple addition, which is very suitable for a large number of similar FR aggregations. At the same time, this aggregation method has its own de-aggregation attribute. After obtaining the overall decision variables, there is no need to calculate the individual decision variables through optimization or collaborative control methods. The de-aggregation method is as follows:

[0106]

[0107] Taking ES as an example, a very important constraint is that the storage energy cannot be charged and discharged at the same time, that is, If the decision variables introduce 0-1 variables, the feasible domain will be discrete and non-convex, which will bring great difficulty to aggregation. In addition, in order to ensure this constraint, some studies only use the same To represent ES power, greater than zero means charging, and less than zero means discharging. However, this approach ignores the charging and discharging efficiency, assuming that the charging and discharging efficiency is 1, which is a large deviation in current practice. Other studies have used the cost function to represent ES power. This nonlinear constraint can be relaxed by placing it in the objective function as a penalty term. However, this method has its applicable scenarios. In essence, it is to increase the net discharge power of the ES under the condition of equal energy change, or to reduce the net charging power of the ES. Only when the original goal is consistent with the penalty item goal can better results be achieved. When the original goal hopes to increase the energy storage charging power, it will be contrary to the penalty item goal and the expected solution cannot be obtained. Therefore, it is difficult for the existing aggregation method to ensure that when scheduling the ES equivalent aggregate, each ES individual can meet the constraint of charging and discharging at different times.

[0108] In the original approximation method, without considering simultaneous charging and discharging, the maximum inner approximation of the original polyhedron is obtained by solving the optimization problem of equation (10):

[0109]

[0110] Although The lower bound of the charge and discharge power is 0, but because it is an inner approximation, The existence of will make the lower bound of the charge and discharge power greater than 0 at certain moments. The originally expected maximum inner approximation will make it so that the constraint of non-simultaneous charge and discharge will be violated at certain moments, and the obtained approximate feasible domain is not actually usable.

[0111] Based on the above understanding, the present invention proposes a de-translation approximation method to solve this problem. In the inner approximation, the translation factor is removed and only the scaling factor is retained:

[0112]

[0113] This method not only ensures that the lower bound of the charging and discharging power is 0, but also ensures the consistency of the overall behavior of the ES cluster. As long as the ES after the overall equivalent aggregation of the cluster is controlled not to charge and discharge at the same time, each EV will not charge and discharge at the same time. This will constrain the ES cluster as a whole to charge and discharge at different times when different types of flexible resources are aggregated. The corresponding optimization problem to be solved is modified as follows:

[0114]

[0115] This method can be used not only for energy storage, but also for any flexible resource that considers charging and discharging efficiency and has different charging and discharging constraints. In addition, since this method ensures the consistency of charging and discharging, it is also convenient to add constraints on the number of charging and discharging times.

[0116] See also Figure 1 Based on the above content, the implementation steps of this embodiment are as follows:

[0117] Step 1: The same resource aggregation platform sends the template and calculation method of formula (4) to each resource individual, requiring them to report the corresponding model coefficient matrix.

[0118] Step 2: Each resource individual needs to calculate its model coefficient matrix based on the template and report it to the similar resource aggregation platform.

[0119] Step 3: The resource aggregation platform of the same type calculates the primitive polyhedron according to formula (5) and sends the coefficients to each resource individual. Depending on whether it is an energy storage resource, the parameter calculation method of formula (10) or formula (14) is sent to each resource individual.

[0120] Step 4: Each resource individual calculates parameters independently according to formula (10) or formula (14) and reports them to the same resource aggregation platform.

[0121] Step 5: The final equivalent aggregate calculated by the same aggregation platform according to formula (8) is reported to the dispatch center.

[0122] Step 6: The dispatch center makes a decision based on the final equivalent aggregate and sends the decision result to the same type of aggregation platform.

[0123] Step 7: The same aggregation platform de-aggregates the decision results according to formula (11) and sends them to each resource individual for execution.

[0124] See also Figure 2 In the prior art, both the upper and lower bounds of the energy storage device are charged and discharged simultaneously to relax the energy limit and thus achieve a higher power limit. However, in reality, the energy storage device cannot be charged and discharged simultaneously. Figure 2 The power cap in the example is not achievable in actual application scenarios.

[0125] See also Figure 3 The de-translation approximation method proposed in the present invention can effectively ensure that each individual satisfies the constraint of charging and discharging at different times, thereby accurately giving the power upper limit in actual application scenarios.

Claims

1. A method for modeling and aggregating flexibility resource regulation characteristics, characterized in that: The following steps are involved: Step 1: The virtual battery model coefficient matrix reported by each individual flexibility resource is collected by a similar resource aggregation platform, wherein the category of the flexibility resource to which each individual belongs is an energy storage system ES or an electric vehicle EV; Step 2: The similar resource aggregation platform uses the collected coefficient matrix of each individual as the polyhedron expression, and then calculates the primitive polyhedron based on the maximum inner approximation algorithm of similar polyhedrons, and sends the calculated coefficients of the primitive polyhedron to each individual; Step 3: The same resource aggregation platform receives the parameters independently calculated by each individual, where the parameters are calculated by each individual according to the category of the flexibility resource to which the individual belongs, using the corresponding parameter calculation method; Step 4: The similar resource aggregation platform calculates the final equivalent aggregate based on the parameters of each individual, and sends it to the dispatch center to make a decision based on the preset decision algorithm, and then receives the decision result; Step 5: The same resource aggregation platform de-aggregates the decision results and finally sends them to each individual for execution.

2. The method according to claim 1, characterized in that In step 1, the virtual battery model coefficient matrix reported by the i-th flexibility resource individual is for: Where K is the category of resource individuals, K∈{ES,EV}, is the decision variable, is the coefficient matrix, represents the real number space, and τ is the maximum value of the offline time interval.

3. The method according to claim 2, characterized in that In step 1, the virtual battery model coefficient matrix is ​​obtained based on the virtual battery model issued by the same resource aggregation platform, and the virtual battery model is: Where T represents the matrix transpose, represents the charging power column vector of the i-th individual, is the charging power of the ith individual at time t, P i K,dis represents the charging power column vector of the i-th individual, is the charging power of the ith individual at time t, represents the upper bound of the charging power of the ith individual, T represents the offline time interval, T={1,2,…,τ}, represents the upper bound of the discharge power of the ith individual, represents the maximum downward climbing rate of the i-th individual, represents the power of the ith individual at time t, represents the maximum upward climbing rate of the i-th individual, represents the lower bound of the energy value of the ith individual at time t+1, represents the energy value of the i-th individual at time t+1, Represents the upper bound of the energy value of the i-th individual at time t+1.

4. The method according to claim 3, characterized in that In the step 1, the virtual battery model is obtained by unified modeling based on the individual models of ES and EV and adjustment characteristics.

5. The method according to claim 4, characterized in that The individual model of ES is: in, is the charging power of the i-th ES at time t, is the discharge power of the i-th ES at time t, is the upper bound of the charging power of the i-th ES at time t, is the upper bound of the discharge power of the i-th ES at time t; is the external power of the i-th ES; is the maximum downward climbing rate of the i-th ES, is the maximum upward climbing rate of the i-th ES; is the energy stored in the i-th ES, is the self-discharge rate, Δt is the time interval, is the charging efficiency of the i-th ES, is the discharge efficiency of the i-th ES; is the upper bound of the energy allowed for the i-th ES, is the lower bound of the energy allowed for the i-th ES; The individual model of EV is: When the EV is charging, in, is the charging power of the i-th EV at time t, is the discharge power of the i-th EV at time t; is the upper bound of the charging power of the i-th EV, is the upper bound of the discharge power of the i-th EV; is the external power of the i-th EV; is the maximum downward climbing rate, is the maximum upward climbing rate; is the energy of the i-th EV at time t, is the self-discharge rate of the i-th EV at time t, and Δt is the time interval; is the charging efficiency of the i-th EV at time t, is the discharge efficiency of the i-th EV at time t; is the upper bound of the energy allowed for the i-th EV, is the lower bound of the energy allowed for the i-th EV, is the expected energy value of the i-th EV at time t+1; When the EV is not charging, 6. The method according to claim 2, characterized in that In the step 2, calculating the primitive polyhedron includes: The coefficient matrix of the virtual battery model As a polyhedral representation, take the average value of each category of flexibility resources As a primitive polyhedron: in, F i K The average value of N K is the number of individuals of a certain type of flexibility resource, for The average value of Decision variables representing primitive polyhedra; Then the same resource aggregation platform will and Sent to each individual.

7. The method according to claim 4, characterized in that In step 3, when each individual calculates the parameters independently, when the individual is ES, the parameters are calculated by the following formula: in Minimize st represents the constraint condition, represents the scaling factor conversion coefficient, G represents a non-negative matrix, ri K represents the translation factor conversion vector; When the individual is EV, the parameter is calculated by the following formula:

8. The method according to claim 7, characterized in that In step 3, the conversion between the scaling factor and the translation factor is performed by the following variable substitution: in, represents the translation factor of the ith individual, Represents the scaling factor of the i-th individual; and after the calculation is completed, the individual sends its own scaling factor or translation factor and scaling factor to the same resource aggregation platform.

9. The method according to claim 8, characterized in that In step 4, the same resource aggregation platform calculates the final equivalent aggregate according to the parameters of each individual:

10. The method according to claim 9, characterized in that In step 5, the similar resource aggregation platform de-aggregates the decision results including: The same resource aggregation platform performs de-aggregation based on the following formula: in is the scaling factor, β K represents the sum of scaling factors, U K represents the aggregate decision variable, represents the sum of translation factors, is the translation factor, Then the same resource aggregation platform will Distributed to each individual.

Images