A Demand-Side Resource Scheduling Method and Device for Virtual Capacity Expansion in Substations

Abstract

The present invention discloses a method and device for demand-side resource scheduling for virtual capacity increase in a distribution area. By considering different types of demand responses, the total load curve of the distribution area is adjusted. According to the construction of different objective functions, it can meet the load adjustment of different demand responses in different distribution areas, and establish different load response constraint conditions to constrain different objective functions. By combining each objective function with the corresponding constraint conditions, the prediction effect of each objective function is improved, thereby greatly improving the utilization efficiency of demand-side resources in different distribution areas and alleviating the problem of short-term shortage of distribution area capacity.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid planning, and particularly to a demand-side resource scheduling method and device for virtual capacity increase of a distribution transformer area. Background Art

[0002] Under the influence of many factors in the new energy supply situation, multi-type distribution transformers in urban communities will face the pressure of full load capacity and the need for optimized operation. For example, with the explosive growth of new energy supply electric vehicles and battery cars, and the rapid advancement of all-electric kitchen and all-electric block projects, the proportion of terminal electricity consumption is increasing day by day, and the capacity of the distribution transformer area. At the same time, with the trend of working from home and online shopping, the production and lifestyle in the city are also facing important changes, which further affect the time-series distribution characteristics of the power load and bring intermittent tension in the capacity of the distribution transformer. In view of the complex and diverse scenarios of shortage in the capacity of the distribution transformer, the decision-making method based on a single peak load analysis dimension and the capacity increase method simply relying on transformer replacement are increasingly exposed with their economic defects, and there is an urgent need to find a more economical and efficient solution.

[0003] Looking at the wide area of the power demand side, a large number of flexible loads have great potential for flexible adjustment, such as electric vehicles, air conditioners, water heaters, industrial production lines, etc. On the one hand, the proportion of such loads is very large, and the air conditioner load alone accounts for more than 50% of the peak load in many areas. At the same time, they also have the characteristics of energy storage. If these flexible loads can be coordinated for normal adjustment, the pressure of shortage in the capacity of the distribution transformer area can be effectively alleviated, and the shortage of adjustment resources under the high proportion of new energy grid connection can be fundamentally compensated.

[0004] The capacity increase of the distribution transformer area mainly refers to the capacity increase of the distribution network transformer. At present, the research on the capacity increase method of the transformer mainly focuses on replacing or transforming the transformer, installing a new transformer or load transfer, using energy storage devices, dynamic capacity increase, etc. The prior art "Load Transfer Strategy in the Capacity Increase and Transformation of 110kV Transformers" takes 110kV transformers as an example and mainly studies and compares several in-station load transfer methods, and analyzes the constraint conditions in the in-station load transfer process. The prior art "Distribution Network Load Transfer Control Method Based on Deep Reinforcement Learning" proposes a new load transfer control method based on deep reinforcement learning, adds a pre-simulation mechanism to the algorithm, adjusts the ratio of actions to learning, and uses an adaptive optimization algorithm for solution.

[0005] The prior art "Feasibility Study on Oil Replacement and Capacity Increase of 110kV Transformer" proposed a method for oil replacement and capacity increase of transformers, and carried out electrical and insulation transformation on old transformers that met the requirements, increasing their rated capacity by about 25%. The prior art "Capacity Expansion of Distribution Network with Control of Multi-function Battery Storage" proposed a multifunctional control strategy for battery energy storage, which achieved capacity expansion, power quality management, i.e. reactive compensation and negative sequence current compensation under unbalanced conditions, and uninterruptible power supply (UPS) at a relatively low cost. The prior art "Overload strategy of transmission and transformation equipment for safety operation" and "Transformer health status evaluation and short-term capacity increase research" developed a transformer health evaluation method, and evaluated the transformer load capacity under the conditions of transformer resistance temperature constraints, transformer failure rate constraints, transformer overload risks, and capacity increase economy, and developed a transformer short-term dynamic capacity increase strategy. The prior art "A method for economical virtual capacity expansion of substations based on distributed flexible resources" proposes a method for economical virtual capacity expansion of substations based on distributed flexible resources, but only considers one flexible resource, electric vehicles, and only performs optimal scheduling for a single substation.

[0006] In summary, most of the existing transformer capacity expansion strategies involve the transformation and replacement of transformers or the addition of new transformers. This is less economical in solving short-term overload problems and is labor-intensive and resource-intensive. Load transfer or the use of energy storage devices is less practical, and improper operation can easily weaken the stability of the power grid and lead to a decline in power quality. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide a demand-side resource scheduling method and device for virtual capacity increase of a transformer area, thereby alleviating the problem of short-term capacity shortage of the transformer area.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0009] A demand-side resource scheduling method for virtual capacity increase in a substation area includes the following steps:

[0010] Obtain different types of load curves of the substations and obtain the total load curve of the substations;

[0011] The total load curve of the substation is adjusted according to the demand response to obtain a demand-side dispatch capacity curve;

[0012] Construct different objective functions according to the demand-side scheduling capacity curve;

[0013] Establish multiple load response constraint conditions, and use the multiple load response constraint conditions to constrain different objective functions;

[0014] Solve different objective functions to obtain the best load prediction values corresponding to different types of substation area loads;

[0015] Schedule the grid load according to the best load prediction value.

[0016] To solve the above technical problems, another technical solution adopted by the present invention is:

[0017] A demand-side resource scheduling device for virtual capacity increase of substation areas includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it realizes each step in the above-mentioned demand-side resource scheduling method for virtual capacity increase of substation areas.

[0018] The beneficial effects of the present invention are as follows: By considering different types of demand responses to adjust the total load curve of the substation area, and according to the construction of different objective functions, it can meet the load adjustment of different demand responses in different substation areas, and establish different load response constraint conditions to constrain different objective functions. By combining each objective function with the corresponding constraint conditions, the prediction effect of each objective function is improved, thereby greatly improving the utilization efficiency of demand-side resources in different substation areas and alleviating the problem of short-term shortage of substation area capacity. Description of the Drawings

[0019] Figure 1 It is a step flow chart of a demand-side resource scheduling method for virtual capacity increase of substation areas in an embodiment of the present invention;

[0020] Figure 2 It is a schematic diagram of a multi-substation area topological structure in an embodiment of the present invention;

[0021] Figure 3 It is a scheduling result diagram obtained by optimizing a demand-side resource scheduling method for virtual capacity increase of substation areas in an embodiment of the present invention in a manner where the weights of four optimization objectives are the same;

[0022] Figure 4 It is a scheduling result diagram obtained by optimizing a demand-side resource scheduling method for virtual capacity increase of substation areas in an embodiment of the present invention in a manner where the weight of virtual capacity increase of the upper-level substation area is relatively large;

[0023] Figure 5 It is a structural schematic diagram of a demand-side resource scheduling device for virtual capacity increase of substation areas in an embodiment of the present invention. Detailed implementation manners

[0024] To describe in detail the technical content, achieved object and effects of the present invention, the following is described in conjunction with the implementation manners and with reference to the drawings.

[0025] Please refer to Figure 1 , a demand-side resource scheduling method for virtual capacity increase of a substation area, including the steps of:

[0026] Obtain load curves of different types of substation areas to obtain the total load curve of the substation area;

[0027] Adjust the total load curve of the substation area according to the demand response to obtain a demand-side scheduling capacity curve;

[0028] Construct different objective functions according to the demand-side scheduling capacity curve;

[0029] Establish multiple load response constraint conditions, and constrain different objective functions through the multiple load response constraint conditions;

[0030] Solve different objective functions to obtain the best load prediction values corresponding to different types of substation area loads;

[0031] Schedule the grid load according to the best load prediction value.

[0032] As can be seen from the above description, the beneficial effects of the present invention are as follows: By considering different types of demand responses to adjust the total load curve of the substation area, and by constructing different objective functions, it is possible to meet the load adjustment of different demand responses in different substations, and to establish different load response constraint conditions to constrain different objective functions. By combining each objective function with the corresponding constraint conditions, the prediction effect of each objective function is improved, thereby greatly improving the utilization efficiency of demand-side resources in different substations and alleviating the problem of short-term shortage of substation capacity.

[0033] Furthermore, the step of solving different objective functions to obtain the best load prediction values corresponding to different types of substation area loads further includes:

[0034] Obtain the objective functions determined for different substations;

[0035] Combine the objective functions corresponding to different substations to obtain a multi-substation objective function;

[0036] Solve the multi-substation objective function to obtain the best load prediction values corresponding to different substations.

[0037] As described above, after determining the objective functions corresponding to different substations, multiple objective functions are combined to obtain the multi-substation objective function, thereby forming a relationship of mutual connection and mutual restriction between different substations, realizing the joint scheduling of multiple substations, and effectively optimizing the resource scheduling between substations.

[0038] Further, solving the multi-substation objective function to obtain the optimal load prediction values corresponding to different substations includes:

[0039] Writing the multi-substation objective function and the corresponding constraint conditions in a compact form;

[0040] Calculating the maximum and minimum values of different objective functions and normalizing the results;

[0041] Generating a single objective function by weighting the normalized different objective functions;

[0042] Solving the single objective function to obtain the optimal load prediction values corresponding to different substations.

[0043] As described above, by converting the multi-objective optimization problem into a series of single-objective optimization problems for solution, by normalizing the objective functions, eliminating the differences in the orders of magnitude of different objective functions, and weighting different objective functions to obtain a single objective function, the final calculation results are evenly distributed.

[0044] Further, the demand-side scheduling capacity curve includes a new energy output load scheduling curve and a power consumption load scheduling curve;

[0045] Constructing different objective functions according to the demand-side scheduling capacity curve includes:

[0046] Constructing a virtual capacity increase objective function according to the demand-side scheduling capacity curve;

[0047] Constructing a new energy curtailment objective function according to the new energy output load scheduling curve;

[0048] Constructing a user satisfaction objective function according to the power consumption load scheduling curve.

[0049] As described above, by using the new energy output load scheduling curve and the power consumption load scheduling curve to establish a virtual capacity increase objective function, a new energy curtailment objective function, and a user satisfaction objective function respectively, it is possible to select different objective functions for scheduling optimization according to different substation scenarios.

[0050] Further, constructing the virtual capacity increase objective function according to the demand-side scheduling capacity curve includes:

[0051]

[0052] Where T is the number of time periods, α is a constant between 0 and 1, PD(t) is the demand-side dispatch capacity curve, is the maximum value of PD(t); Var max is the constraint value, and P is the target curve.

[0053] As can be seen from the above description, by combining the demand-side dispatch capacity curve with the constraint value, a virtual capacity increase target curve is obtained, and a demand-side dispatch model for a single transformer area is constructed to realize the optimization of resource dispatch for a single transformer area.

[0054] Furthermore, it further includes:

[0055]

[0056] Where Var max is the maximum value variance.

[0057] As can be seen from the above description, if only the minimum of the maximum value of the adjusted capacity curve is considered, since there are not enough constraints on other non-maximum value points, the calculated target curve may have a large number of reasonable solutions, and there are also difficulties in the implementation process. Therefore, the maximum value variance is used to limit other points with larger values, fully considering the discreteness of the capacity curve to ensure that the values of other points are as small as possible.

[0058] Furthermore, the construction of the new energy curtailment target function according to the new energy output load dispatch curve includes:

[0059]

[0060] Where P PV,max (t) is the predicted maximum output of new energy, and PD PV (t) is the new energy output load dispatch curve.

[0061] As can be seen from the above description, when there is a large amount of output from distributed new energy such as photovoltaic and wind power in the transformer area, it is considered to dispatch flexible loads at other times to this time through demand response to increase the load to promote the consumption of photovoltaic power; at the same time, the distributed new energy can also be connected to the grid through the distribution transformer and transmitted to other transformer areas at the same level for consumption; that is, the capacity of this transformer area is negative at this time; therefore, in order to minimize the waste of distributed new energy output in the transformer area and make the new energy output in the transformer area be consumed by the transformer area itself as much as possible, so as to achieve the precise dispatch of new energy output.

[0062] Furthermore, the power consumption load dispatch curve includes an air-conditioning load curve and an electric vehicle load curve;

[0063] Constructing a user satisfaction objective function based on the described power consumption load scheduling curve includes:

[0064]

[0065] In the formula, PD AC,i (t) and PD EV,j (t) are the air-conditioning load curve and the electric vehicle load curve respectively; PS AC (t) and PS EV,j (t) are the air-conditioning predicted load curve and the electric vehicle predicted load curve respectively.

[0066] As can be seen from the above description, by combining the usage situations of users' air conditioners and electric vehicles to construct a user satisfaction objective function, making each flexible load unit close to the standard load curve, the requirements of users for power quality or comfort can be met.

[0067] Furthermore, establishing multiple load response constraint conditions to constrain different objective functions through the multiple load response constraint conditions includes:

[0068] Establishing a transformer capacity constraint to constrain the virtual capacity increase objective function;

[0069] Establishing a new energy output constraint to constrain the new energy abandonment objective function;

[0070] Establishing a power consumption load constraint to constrain the user satisfaction objective function.

[0071] As can be seen from the above description, by establishing a transformer capacity constraint, a new energy output constraint, and a power consumption load constraint, and respectively constraining different objective functions, the accuracy of the output of each objective function is improved.

[0072] Another embodiment of the present invention provides a demand-side resource scheduling device for virtual capacity increase of a distribution area, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it realizes each step in the above-mentioned demand-side resource scheduling method for virtual capacity increase of a distribution area.

[0073] The above-mentioned demand-side resource scheduling method and device for virtual capacity increase of a distribution area of the present invention are applicable to the flexible scheduling of different types of demand-side resources. In particular, through multi-distribution area joint scheduling, the utilization efficiency of demand-side resources in the distribution area is greatly improved, and while taking into account new energy consumption and user satisfaction, the short-term shortage problem of distribution area capacity is alleviated with high quality. The following is illustrated through specific embodiments:

[0074] Embodiment 1

[0075] Please refer toFigure 1 , a demand-side resource scheduling method for virtual capacity increase in a distribution area, including the steps of:

[0076] S1. Obtain load curves of different types in the distribution area to obtain the total load curve of the distribution area; wherein, the load curves of the distribution area include: new energy output load curves such as the load curve of distributed photovoltaic output in the distribution area, and power consumption load curves such as the load curve of air-conditioning load in the distribution area and the load curve of electric vehicle load in the distribution area;

[0077] If the capacity of the distribution area is defined as:

[0078] P(t) = P UL (t) + P AC (t) + P EV (t) - P PV (t);

[0079]

[0080]

[0081]

[0082] Among them, P UL (t) represents the load value of the rigid load curve in the distribution area at the t-th moment, P AC (t) represents the total load curve of the air-conditioning load in the distribution area, P AC,i (t) represents the load curve of the i-th air conditioner; P EV (t) represents the total load curve of the electric vehicle load in the distribution area, P EV,j (t) represents the load curve of the j-th electric vehicle; P PV (t) represents the total load curve of the distributed photovoltaic output in the distribution area, P PV,n (t) represents the output curve of the n-th distributed photovoltaic power source;

[0083] Thus, the remaining capacity P left of the distribution area can be defined as:

[0084] P left (t) = P N (t) - P(t);

[0085] Among them, P N (t) represents the rated capacity of the transformer;

[0086] S2. Adjust the total load curve of the distribution area according to the demand response to obtain the demand-side scheduling capacity curve; since when the peak load in the distribution area is relatively high, the margin in the distribution area is insufficient, that is, P left is small, resulting in poor anti-risk ability; in particular, when the capacity of the distribution area is greater than the rated capacity of the transformer, the transformer is overloaded, that is, Pleft is negative, and at this time the power grid system is in a very unsafe operating state; therefore, to ensure the safe operation of the transformer, that is, to always make P left >0, it is necessary to ensure that P N is greater than the maximum capacity P max of the substation area. Through demand-side response, flexible loads are involved in power grid dispatching to cut load peaks, so that the maximum capacity P max of the substation area is reduced, and then it is ensured that P left >0. Without changing the rated capacity of the transformer, through demand-side response, it can bear the original load, realizing the virtual capacity increase of the transformer; among them, the virtual capacity increase VP of the transformer is as follows:

[0087] VP = P left,min - P left0,min ;

[0088] In the formula: P left，min represents the minimum value of the remaining capacity of the substation area after demand-side resource dispatching, and P left0，min represents the minimum value of the remaining capacity of the original substation area load curve; and the substation area capacity curve after demand-side dispatching is denoted as PD(t);

[0089] S3. Construct different objective functions according to the demand-side dispatching capacity curve, specifically:

[0090] S31. Construct a virtual capacity increase objective function according to the demand-side dispatching capacity curve;

[0091]

[0092] In the formula, T is the number of time periods, α is a constant between 0 and 1, PD(t) is the demand-side dispatching capacity curve, is the maximum value of PD(t); Var max is the constraint value, and P is the target curve;

[0093] If only considering that the maximum value of the adjusted capacity curve is the smallest, since there are not enough constraints on other non-maximum value points, the calculated flexible load target curve may have a large number of reasonable solutions, and there are also difficulties in the implementation process. By simultaneously considering the discreteness of the capacity curve and restricting other non-maximum values, the maximum variance is defined:

[0094]

[0095] In the formula, Var max is the maximum value variance;

[0096] Please refer to Figure 2, for the virtual capacity increase problem of the upper-level power distribution area, a simplified method is adopted for calculation, that is, the situation where the upper-level power distribution area is directly connected to the user load side is not considered. Therefore, the capacity of the upper-level power distribution area is the sum of the capacities of each lower-level power distribution area. The virtual capacity increase can be achieved by adjusting the capacity curve of each lower-level power distribution area on the demand side. The objective function is:

[0097]

[0098] Among them, N in the formula represents the number of power distribution areas; since the scheduling problems of multiple power distribution areas are involved, that is, it is necessary to minimize the maximum value of the sum of the capacity curves of all power distribution areas while considering the conditions that each power distribution area itself needs to meet. For example, if a lower-level power distribution area also needs to perform virtual capacity increase, it is necessary to simultaneously consider the minimum value of the maximum value of the capacity curve of this power distribution area. These two objectives may either be unified, that is, the peak time of this power distribution area is the same as the peak time of the upper-level power distribution area capacity; or they may be opposite, that is, the valley time of this power distribution area capacity is the same as the peak time of the upper-level power distribution area capacity.

[0099] S32. Construct an objective function for new energy curtailment according to the new energy output load scheduling curve;

[0100] When there is a large amount of output from distributed new energy sources such as photovoltaic and wind power in the power distribution area, it is considered to schedule flexible loads at other times to this time through demand response to increase the load to promote the consumption of photovoltaic power; at the same time, the distributed new energy can also be connected to the grid through the distribution transformer and transmitted to other power distribution areas at the same level for consumption; that is, the capacity of this power distribution area is negative at this time. Therefore, in order to minimize the waste of distributed new energy output in the power distribution area and make the new energy output in the power distribution area be consumed by the power distribution area itself as much as possible, and the substation should be consumed by the substation itself as much as possible, an objective function for new energy curtailment is defined:

[0101]

[0102] In the formula, P PV,max (t) is the predicted maximum output of new energy, and PD PV (t) is the new energy output load scheduling curve; taking photovoltaic power generation as an example, that is, P PV,max (t) is the predicted maximum output of photovoltaic power, and PD PV (t) is the scheduled photovoltaic output value;

[0103] S33. Construct an objective function for user satisfaction according to the power consumption load scheduling curve; in the case of high requirements for power quality or comfort, each flexible load unit needs to be as close as possible to the standard load curve. The difference between the flexible load curve and the standard load curve or the minimum user adjustment amount f is described by defining the objective function for user satisfaction 3 :

[0104]

[0105] In the formula, PD AC,i (t) and PD EV,j (t) are the air-conditioning load curve and the electric vehicle load curve respectively; PS AC (t) and PS EV,j (t) are the air-conditioning predicted load curve and the electric vehicle predicted load curve respectively; in different application scenarios, the selection of the load curve can be changed, such as the electrical appliance load curve, etc.;

[0106] S4. Establish multiple load response constraint conditions, and use the multiple load response constraint conditions to constrain different target functions;

[0107] S5. Solve different target functions to obtain the optimal load prediction values corresponding to different types of substation area loads;

[0108] S6. Schedule the grid load according to the optimal load prediction value.

[0109] Embodiment 2

[0110] This embodiment specifically defines how to constrain the target function through constraint conditions;

[0111] S4. The establishment of multiple load response constraint conditions and the use of the multiple load response constraint conditions to constrain different target functions include:

[0112] S41. Establish a transformer capacity constraint to constrain the virtual capacity increase target function;

[0113] Since the rated capacity of the transformer is limited, if the operating capacity exceeds the rated capacity of the transformer, it will cause safety problems in the operation of the power grid; by setting the transformer capacity constraint, the above problems can be avoided:

[0114] 0 ≤ PD(t) ≤ P N ;

[0115] S42. Establish a new energy output constraint to constrain the new energy abandonment target function;

[0116] Taking photovoltaic power generation as an example: Since the power generation capacity of distributed photovoltaics is limited, it is necessary to constrain the upper and lower limits of the output of distributed photovoltaics:

[0117] 0 ≤ PD PV (t) ≤ Capa PV *P PV,max (t);

[0118] P PV,max(t) represents the normalized curve of the maximum photovoltaic output predicted based on the data of the day before yesterday (historical data), and Capa PV represents the total installed capacity of distributed photovoltaics in the distribution area; if it is assumed that users prefer to select the output of distributed power sources in the distribution area, the constraint formula is obtained:

[0119] PD PV(t) ≥min(P UL (t)+PD AC (t)+PD EV (t),Capa PV *P PV,max (t));

[0120] S43. Establish the power consumption load constraint to constrain the user satisfaction objective function; taking the air conditioner and electric vehicle loads as examples:

[0121] S431. Establish the air conditioner load response constraint;

[0122] For the air conditioner load, it is mainly constrained by the degree of freedom ν AC and the scheduling allowable error α; the degree of freedom ν AC represents the number of free adjustment periods of the air conditioner. Ideally, as long as it is ensured that the total energy consumption of the air conditioner within this time period is the same as the total energy consumption of the standard load curve of the air conditioner load within this time period, it can be ensured that the user is always in a comfortable environment. In this embodiment, ν AC is set to 1-2, and the temperature is always within the comfortable range without affecting the user experience. In order to make the air conditioner load more flexible in participating in the demand-side response, the scheduling allowable error α is defined, that is, the maximum value of the ratio of the absolute value of the difference between the total energy consumption of the variable-frequency air conditioner load and the total energy consumption of the standard load curve within the specified degree of freedom to the total energy consumption of the standard load curve; considering that too many air conditioners may lead to a higher solution complexity and thus may lead to the curse of dimensionality, it is therefore assumed that each user in the distribution area has the same satisfaction with the air conditioner when using it. In this way, all air conditioner loads are randomly divided into three levels, and the number of air conditioners in each level is the same. The air conditioners within each level are regarded as a whole for unified scheduling:

[0123]

[0124] t 0 represents the starting time of scheduling the air conditioner, and PS AC (t) represents the standard load curve of the AC load, and N AC represents the number of AC loads in each layer; in addition, to ensure the user comfort as much as possible, the total power consumption of the air conditioner must be the same as the total power consumption of the standard load curve, that is:

[0125]

[0126] S432. Establish the load response constraints for electric vehicles;

[0127] Since electric vehicles can only be charged when parked, and the charging power remains constant during charging, the corresponding constraints can be expressed as:

[0128]

[0129] status(j,t) is the state variable of the electric vehicle, where status(j,t)=0 indicates that the electric vehicle is parked at time t, and status(j,t)=1 indicates that it is in the driving state; P Charge represents that the charging power is a constant value;

[0130] Furthermore, in order to ensure that the electric vehicle is fully charged when the user travels, it is necessary to ensure that the electric vehicle is fully charged during the parking period, and the corresponding constraints can be expressed as:

[0131]

[0132] Demand EV,j refers to the charging amount required by the electric vehicle.

[0133] Embodiment III

[0134] This embodiment specifically provides a solution method for multi-objective functions;

[0135] Since in the multi-substation area joint optimization scheduling model, each substation area has its own optimization objective; and there are interrelated and restrictive relationships among these objectives, which are neither completely consistent nor completely deviated, and there is a certain "competition - cooperation" relationship, making it difficult to simultaneously reach their optimal solutions; therefore, when facing multi-objective problems, it is usually chosen to take into account each objective and obtain its Pareto optimal solution set; the Normal-Boundary Intersection (NBI) method solves the multi-objective optimization problem by converting it into a series of single-objective optimization problems, and finally obtains non-dominated solutions evenly distributed on the Pareto front; the traditional weighted method converts the multi-objective optimization into a single-objective optimization problem, and due to the different magnitudes of each objective, etc., the final result obtained after weighting is very unevenly distributed on the Pareto front, while the NBI algorithm normalizes the objective function, eliminating the difference in its magnitude, so that the final obtained solutions can be evenly distributed on the Pareto front;

[0136] Please refer to Figure 2 , S5. Solve different said objective functions to obtain the best load prediction values corresponding to different types of substation area loads, including:

[0137] S51. Obtain the objective functions determined for different substations; that is, obtain the n objective functions constructed in Example 1, including the virtual capacity increase objective function and the new energy curtailment objective function, and obtain the n constraint conditions constructed in Example 2, including the transformer capacity constraint, the new energy output constraint, and the power consumption load constraint;

[0138] S52. Combine the objective functions corresponding to different substations to obtain a multi-substation objective function;

[0139] S53. Solve the multi-substation objective function to obtain the optimal load prediction values corresponding to different substations;

[0140] S531. Write the multi-substation objective function and the corresponding constraint conditions in a compact form; as follows:

[0141] minF(x) = {f 1 (x), f 2 (x), f 3 (x)};

[0142]

[0143] In the formula, g(x) represents the equality constraint, h(x) represents the inequality constraint, and x is the vector composed of all decision variables;

[0144] S532. To avoid the differences in dimension and order of magnitude between objective functions, it is necessary to normalize the objective functions so that they are controlled within the interval [0, 1]. The specific method is: separately solve the maximum and minimum values of each objective function, and normalize each objective function according to the following formula:

[0145]

[0146] In the formula, is the k-th normalized objective function, f k is the k-th objective function, f kmin , f kmax are the minimum and maximum values of the k-th objective function respectively.

[0147] S533. When only considering the minimization of f 1 (x) for single-objective optimization, the optimal solution x 1* is obtained, corresponding to the point f 1* (f 1 (x 1* ), f 2 (x 1* ), f 3 (x 1*)); Similarly, the optimal solutions x and x that only consider f(x) and f(x) at their minimum values can be obtained respectively, corresponding to the points f(f(x), f(x), f(x)) and f(f(x), f(x), f(x)). In the coordinate space formed by each objective function, the points f, f, and f form the endpoints of the Pareto front, and the plane determined by them is called the utopia plane. 2 (x) and f 3 (x) at their minimum values can be obtained respectively, corresponding to the points f 2* and x 3* , corresponding to the points f 2* (f 1 (x 2* ), f 2 (x 2* ), f 3 (x 2* )) and f 3* (f 1 (x 3* ), f 2 (x 3* ), f 3 (x 3* ))。In the coordinate space formed by each objective function, the points f 1* , f 2* and f 3* form the endpoints of the Pareto front, and the plane determined by them is called the utopia plane.

[0148] S534. Generate uniformly distributed points on the utopia plane. Assume that the vector from point f to point f is N1, the vector from point f to point f is N2, and the vector from point f to point f is N3. Nk is divided into m equal parts, then the unit length of each part δ = 1 / m, where k ∈ {1, 2, 3}. Any point on the utopia plane can be represented by a linear combination of the endpoints f, f, and f. Taking the jth point as an example, its coordinates p are: 1* to point f 3* is N1, the vector from point f 2* to point f 3* is N2, and the vector from point f 1* to point f 2* is N3. Nk is divided into m equal parts, then the unit length of each part δ k = 1 / m k , where k ∈ {1, 2, 3}. Any point on the utopia plane can be represented by a linear combination of the endpoints f k , f 1* , and f 2* . Taking the jth point as an example, its coordinates p 3* are: j as follows:

[0149]

[0150] where β 1j = [0, 1,..., m 1 δ 1 , β 2j = [0, 1,..., m' 2 δ 2 , β 3j = 1 - β 1j - β 2j , m' 2 = Ψ[(1 - β 1j ) / δ 2, Ψ[·] is to round the function. The parameter β 1j 's value determines the distribution of points on the utopia surface (which can be selected artificially. The more the number of selected points, the more accurate the obtained Pareto front).

[0151] S535. The multi-objective problem is transformed into a single-objective problem. After the equal division point vector β is given, the multi-objective problem can be respectively transformed into single-objective optimization problems:

[0152] min-D

[0153]

[0154] where D is the distance that the selected point on the utopia e can reach along the normal direction, n is the quasi-normal vector, and e = (1, 1, L, 1) T . As D increases, the objective functions corresponding to the feasible solutions determined by D are gradually improved. When D increases to the maximum value, the objective functions reach Pareto optimality. Therefore, the meaning of the above formula is a single-objective optimization problem with the maximum distance between the point on the utopia surface and the point on the corresponding Pareto front as the objective. As different values of β are traversed, the multi-objective optimization problem is converted into a series of single-objective optimization problems, which can be directly solved by the solver Gurobi, and then the best load prediction values corresponding to different substations can be obtained.

[0155] This embodiment provides a specific example. For example, in a specific application scenario, an optimal scheduling problem of three lower-level substations and one upper-level substation is given. The basic information of the substations is shown in Table (1), the simulation data is shown in Table (2), and the optimization results are shown in Table (3).

[0156] Figure 3 represents the optimal scheduling result with the same optimization objective weights; Figure 4It represents the optimal scheduling result with a relatively large virtual capacity increase weight for the upper-level power distribution area. Analysis shows that: (1) The impacts among various objectives between multiple power distribution areas are relatively small. The main reasons are as follows: First, the mutual influences and games among the lower-level power distribution areas mainly depend on the constraints of the upper-level power distribution area. For example, since the upper-level power distribution area cannot transmit the distributed photovoltaic power output of the power distribution area upward anymore, when the distributed photovoltaic power output of a certain power distribution area is excessive at a certain time period and its own demand-side resource scheduling ability is insufficient, in order to ensure the consumption of photovoltaic power, it may schedule the flexible loads of other power distribution areas to concentrate at this time period. If the constraints of the upper-level power distribution area are not considered, that is, there are no upper and lower limits for the total capacity of each power distribution area, then each power distribution area only needs to achieve its own objectives and there will be no mutual restrictions among the power distribution areas. (2) Photovoltaic power consumption should be ensured first among various objectives. It can be seen from the data in the above table that no matter which objective has a large weight, the distributed photovoltaic power output can always be completely consumed, because a sufficiently large weight has been given to this objective in the model construction to ensure its priority. (3) The objectives of the upper-level power distribution area have a greater impact on the lower-level power distribution areas. When the weight of the virtual capacity increase objective of the upper-level power distribution area is large, each lower-level power distribution area responds. When the weights of other objectives are large, in order to consume the distributed photovoltaic power output of its own power distribution area as much as possible, Power Distribution Area 2 transfers the flexible loads at other times to the moment when the distributed photovoltaic power output is the largest, and coincidentally this moment is also the time when the sum of the capacities of each power distribution area is the largest. Therefore, when the virtual capacity increase requirement of the upper-level power distribution area is high, Power Distribution Area 2 has to abandon this objective and instead transfer more photovoltaic power to Power Distribution Area 1 and Power Distribution Area 3. To ensure the satisfaction of its own users, in the first experiment, the optimal curves of each flexible load have little difference from the standard curve, but when the virtual capacity increase requirement of the upper-level power distribution area is high, Power Distribution Area 3 has to abandon its own optimization objective and transfer the loads during the time period when the sum of the capacities of the power distribution area is the largest, resulting in an increase in the user adjustment amount of this power distribution area. And Power Distribution Area 1 has no change because its own objective is virtual capacity increase, which is basically consistent with the optimization objective of the upper-level power distribution area. (4) The above examples prove that in the case of joint scheduling of multiple power distribution areas, this technology can effectively schedule demand-side resources to achieve virtual capacity increase of this power distribution area or the upper-level power distribution area, and at the same time each power distribution area can also take into account its own new energy consumption and user power consumption satisfaction. Among them, all the figures have curves for 6 days. The reason why some only show 5 curves is that the curves before and after the 3DR of the power distribution area overlap.

[0157] Table 1. Basic information of the power distribution area e

[0158] Transformer substation area Virtual capacity increase Photovoltaic power consumption User satisfaction 1 Yes No No 2 No Yes No 3 No No Yes Upper-level transformer substation area Yes / /

[0159] Table 2. Simulation data

[0160]

[0161] Table 3. Optimization results

[0162] Optimization result PD1,max (kW) Loss (kW) Reg (kW) PDmax (kW) 0 (before optimization) 217.29 37.56 143.19 663.78 1 186.05 0 1.67 621.61 2 186.05 0 128.64 600.42

[0163] Embodiment 4

[0164] Please refer to Figure 5 , a demand-side resource scheduling device for virtual capacity augmentation of a substation area, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it realizes each step in a demand-side resource scheduling method for virtual capacity augmentation of a substation area as described in any one of Embodiments 1, 2, or 3.

[0165] In summary, the demand-side resource scheduling method and device for virtual capacity augmentation of a substation area provided by the present invention start from flexible resources on the demand side. First, a demand-side resource scheduling method with the goal of virtual capacity augmentation of a substation area is proposed. Further, multiple goals such as virtual capacity augmentation of a substation area, new energy consumption, and user satisfaction are considered, and different load response constraint conditions are established to constrain different objective functions. A demand-side resource scheduling model for multi-substation area joint is proposed. The boundary normal intersection method is used to convert the multi-objective optimization problem into a series of single-objective optimization problems, and the Pareto curve is obtained by solving respectively, improving the prediction effect of each objective function, thereby greatly improving the utilization efficiency of demand-side resources in different substation areas and alleviating the problem of short-term shortage of substation area capacity.

[0166] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in the related technical field, shall be similarly included in the patent protection scope of the present invention.

Claims

1. A demand - side resource scheduling method for virtual capacity increase in the sub - station area, characterized in that, it includes the steps: Obtain the load curves of different types in the sub - station area to get the total load curve of the sub - station area; Adjust the total load curve of the sub - station area according to the demand response to obtain the demand - side scheduling capacity curve; Construct different objective functions according to the demand - side scheduling capacity curve; Establish multiple load response constraint conditions, and use the multiple load response constraint conditions to constrain different objective functions; Solve different objective functions to obtain the optimal load prediction values corresponding to different types of sub - station area loads; Schedule the grid load according to the optimal load prediction value; The step of solving different objective functions to obtain the optimal load prediction values corresponding to different types of sub - station area loads further includes: Obtain the determined objective functions of different sub - station areas; Combine the objective functions corresponding to different sub - station areas to obtain a multi - sub - station area objective function; Solve the multi - sub - station area objective function to obtain the optimal load prediction values corresponding to different sub - station areas; The step of solving the multi - sub - station area objective function to obtain the optimal load prediction values corresponding to different sub - station areas includes: Write the multi - sub - station area objective function and the corresponding constraint conditions in a compact form; Calculate the maximum and minimum values of different objective functions and normalize the results; Generate a single objective function by weighting the normalized different objective functions; Solve the single objective function to obtain the optimal load prediction values corresponding to different sub - station areas; The demand - side scheduling capacity curve includes a new - energy output load scheduling curve and a power consumption load scheduling curve; The step of constructing different objective functions according to the demand - side scheduling capacity curve includes: Construct a virtual capacity - increase objective function according to the demand - side scheduling capacity curve; Construct a new - energy abandonment objective function according to the new - energy output load scheduling curve; Construct a user satisfaction objective function according to the power consumption load scheduling curve; The step of establishing multiple load response constraint conditions and using the multiple load response constraint conditions to constrain different objective functions includes: Establish a transformer capacity constraint to constrain the virtual capacity - increase objective function; Establish a new - energy output constraint to constrain the new - energy abandonment objective function; Establish a power consumption load constraint to constrain the user satisfaction objective function.

2. The demand - side resource scheduling method for virtual capacity increase in the sub - station area according to claim 1, characterized in that, the step of constructing a virtual capacity - increase objective function according to the demand - side scheduling capacity curve includes: ； In the formula, is the number of time periods, is a constant between 0 and 1, is the demand-side dispatch capacity curve, is the square of the maximum value of; is the constraint value, and P is the target curve.

3. The demand - side resource scheduling method for virtual capacity increase in the sub - station area according to claim 2, characterized in that, it further includes: ； In the formula, is the maximum variance.

4. The demand - side resource scheduling method for virtual capacity increase in the sub - station area according to claim 1, characterized in that, the step of constructing a new - energy abandonment objective function according to the new - energy output load scheduling curve includes: ； In the formula, is the maximum predicted output of new energy, is the load scheduling curve of new energy output.

5. The demand - side resource scheduling method for virtual capacity increase in the sub - station area according to claim 1, characterized in that, the power consumption load scheduling curve includes an air - conditioner load curve and an electric - vehicle load curve; Constructing a user satisfaction objective function based on the power consumption load scheduling curve includes: ； In the formula, and are the air-conditioning load curve and the electric vehicle load curve respectively; and are the predicted air-conditioning load curve and the predicted electric vehicle load curve respectively.

6. A demand-side resource scheduling device for virtual capacity increase in a distribution area, comprising a memory, a processor, and a computer program stored on the memory and capable of running on the processor, characterized in that, when the processor executes the computer program, it realizes each step in a demand-side resource scheduling method for virtual capacity increase in a distribution area as described in any one of claims 1-5.

Images