A method and device for multi-time-scale coordinated scheduling of low-voltage DC interconnected distribution areas

Abstract

This invention discloses a multi-timescale coordinated scheduling method and device for low-voltage DC interconnected distribution areas. The invention employs a day-ahead long-term optimization scheduling method to obtain the day-ahead scheduling plan, a day-intraday rolling optimization scheduling method to continuously revise the day-ahead scheduling plan, and a real-time optimization scheduling method to schedule converter power, load reduction power, and electric vehicle quota power. This achieves multi-timescale coordinated scheduling of low-voltage DC interconnected distribution areas, enabling flexible scheduling of controllable resources in the distribution area at multiple time scales, reducing the impact of the randomness of photovoltaic power output, and improving the economy and safety of distribution area operation.

Description

Technical Field

[0001] This invention relates to a method and apparatus for multi-timescale coordinated scheduling of low-voltage DC interconnected distribution areas, belonging to the field of power system dispatching. Background Technology

[0002] Against the backdrop of the strategic goals of "carbon peaking" and "carbon neutrality," the large-scale integration of distributed power sources such as photovoltaics and electric vehicles has brought significant challenges to the safe and stable operation of low-voltage distribution substations. Under the existing decentralized operation mode of low-voltage substations, the strong randomness and intermittency of distributed photovoltaics, along with the surge in electric vehicle loads, easily lead to equipment overload and load deficits, thus limiting the photovoltaic absorption capacity of low-voltage substations. How to rationally upgrade existing substations, coordinate controllable resources in each substation, reduce the impact of photovoltaic uncertainties, improve the photovoltaic absorption capacity of substations, and achieve safe and stable operation is an urgent problem to be solved.

[0003] To address the current problems, upgrading existing DC charging stations to achieve low-voltage DC interconnection between distribution substations is an effective solution. Low-voltage DC offers advantages such as flexible networking schemes and wide application scenarios. Interconnecting low-voltage distribution substations through DC charging station upgrades can save on upgrade costs, fully leverage the complementary characteristics of different substations, improve the low-voltage substations' capacity to accommodate electric vehicle loads and photovoltaic power, and reduce operating costs. In actual operation, coordinated scheduling of controllable resources within the substations is required at multiple time scales, but currently, there is no corresponding method. Summary of the Invention

[0004] This invention provides a method and apparatus for multi-timescale coordinated scheduling of low-voltage DC interconnected distribution areas, which solves the problems disclosed in the background art.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0006] A multi-time-scale coordinated scheduling method for low-voltage DC interconnected distribution areas includes:

[0007] Based on the predicted photovoltaic output and photovoltaic prediction error of the low-voltage DC interconnected distribution area, the historical photovoltaic output data is screened for scenarios, and the screened scenarios are clustered. The scenario screening of historical photovoltaic output data includes scenario screening of long-term historical photovoltaic output data and scenario screening of short-term historical photovoltaic output data.

[0008] Based on the scenario clustering results of long-term historical photovoltaic power output data, the day-ahead long-term optimization scheduling method is adopted to obtain the day-ahead scheduling plan. The day-ahead long-term optimization scheduling method includes an upper-level stochastic programming method and a lower-level robust optimization method. The upper-level stochastic programming method aims to minimize the expected cost of the worst scenario in each class, while the lower-level robust optimization method aims to minimize the intraday scheduling cost of the worst scenario in each class.

[0009] Based on the scenario clustering results of short-term historical photovoltaic power output data, an intraday rolling optimization scheduling method is adopted to continuously revise the day-ahead scheduling plan;

[0010] Based on the actual photovoltaic output at the current moment, the day-ahead scheduling plan with rolling correction is used as input, and a real-time optimization scheduling method is adopted to schedule the power of converters in low-voltage DC interconnected areas, load reduction power, and electric vehicle quota power in real time.

[0011] Based on the predicted photovoltaic output and prediction error of low-voltage DC interconnected distribution areas, historical photovoltaic output data is filtered for specific scenarios, and the selected scenarios are clustered, including:

[0012] The photovoltaic fluctuation range is determined based on the photovoltaic power output forecast and the photovoltaic forecast error;

[0013] By filtering historical photovoltaic power output data, scenarios within the photovoltaic fluctuation range can be obtained;

[0014] Clustering is performed on scenarios within the photovoltaic fluctuation range.

[0015] The upper-level stochastic programming method aims to minimize the expected cost of the worst-case scenario in each class and optimizes the day-ahead scheduling plan. The day-ahead scheduling plan optimized by the upper-level stochastic programming optimization method includes the transferable load transfer period, the transferable load transfer power, the load reduction period, the energy storage charging and discharging status, the energy storage charging and discharging power base value, and the distribution transformer power base value.

[0016] The upper-level stochastic programming optimization method used is:

[0017]

[0018] Ax+By k +ζ k ≤c

[0019]

[0020] Where x is the day-ahead scheduling plan, f1(x) is the day-ahead scheduling cost, and K C ρ represents the total number of clusters. k f is the probability of class k. 2,kLet y be the intraday scheduling cost for the worst-case scenario of type k, where A, B, and c are the coefficient matrices or vectors corresponding to the constraints. k For the intraday scheduling plan corresponding to the worst-case scenario of type k, ζ k Provide photovoltaic power for the worst-case scenario of type k.

[0021] The lower-level robust optimization method aims to minimize the intraday scheduling cost in the worst-case scenario among all types of scenarios and optimizes the day-ahead scheduling plan. The day-ahead scheduling plan optimized by the lower-level robust optimization method includes the adjustment of energy storage charging and discharging power, the adjustment of distribution transformer power, the reduction of load that can be reduced, the load limit power of electric vehicles, and the power of converter.

[0022] The optimization model used in the lower-level robust optimization method is:

[0023]

[0024]

[0025] Ax+By k,i +ζ k,i ≤c

[0026] for: k = 1, 2, ..., K C

[0027] Where, ζ k For the photovoltaic power output in the worst-case scenario of type k, f 2,k Let Ω be the intraday scheduling cost for the worst-case scenario of type k, and let A, B, and c be the coefficient matrices or vectors corresponding to the constraints. k Let f be the set of similar scenes in the k-th class. 2,k,i Let K be the intraday scheduling cost of the i-th similar scenario in the k-th class. C y represents the total number of clusters. k,i Let ζ be the intraday scheduling strategy for the i-th similar scenario in the k-th class. k,i Photovoltaic output of the i-th similar scenario in the k-th class.

[0028] The intraday rolling optimization scheduling method uses stochastic programming.

[0029] A low-voltage DC interconnected distribution area multi-timescale coordinated scheduling device includes:

[0030] The clustering module performs scenario filtering on historical photovoltaic output data based on the predicted output and prediction error of photovoltaic power generation in low-voltage DC interconnected distribution areas, and then clusters the selected scenarios. The scenario filtering on historical photovoltaic output data includes scenario filtering on long-term historical photovoltaic output data and scenario filtering on short-term historical photovoltaic output data.

[0031] The day-ahead long-term optimization scheduling module uses the day-ahead long-term optimization scheduling method to obtain the day-ahead scheduling plan based on the scenario clustering results of long-term historical photovoltaic power output data. The day-ahead long-term optimization scheduling method includes an upper-level stochastic programming method and a lower-level robust optimization method. The upper-level stochastic programming method aims to minimize the expected cost of the worst scenario in each class, while the lower-level robust optimization method aims to minimize the intraday scheduling cost of the worst scenario in each class.

[0032] The intraday rolling optimization scheduling module uses the intraday rolling optimization scheduling method to continuously revise the day-ahead scheduling plan based on the scenario clustering results of short-term historical photovoltaic power output data.

[0033] The real-time optimization scheduling module, based on the actual output of photovoltaic power at the current moment, takes the day-ahead scheduling plan with rolling correction as input, and uses the real-time optimization scheduling method to schedule the power of converters in low-voltage DC interconnected distribution areas, load reduction power, and electric vehicle quota power in real time.

[0034] A computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform a multi-timescale coordinated scheduling method for low-voltage DC interconnected distribution areas.

[0035] A computing device includes one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for performing a multi-timescale coordinated scheduling method for low-voltage DC interconnected distribution areas.

[0036] The beneficial effects achieved by this invention are as follows: 1. This invention adopts a day-ahead long-term optimization scheduling method to obtain the day-ahead scheduling plan, adopts an intraday rolling optimization scheduling method to continuously revise the day-ahead scheduling plan, and adopts a real-time optimization scheduling method to schedule converter power, load reduction power, and electric vehicle quota power. This realizes multi-time-scale coordinated scheduling of low-voltage DC interconnected distribution areas, and flexibly schedules controllable resources of the distribution area at multiple time scales, reducing the impact of photovoltaic output randomness and improving the economy and safety of distribution area operation; 2. The stochastic-robust two-layer optimization scheduling adopted by this invention for day-ahead scheduling achieves a good balance between the economy of stochastic programming and the robustness of robust optimization, and significantly improves the solution efficiency. Attached Figure Description

[0037] Figure 1 This is a flowchart of the method of the present invention;

[0038] Figure 2 This is a flowchart illustrating the specific process of the method of the present invention;

[0039] Figure 3 Structural diagram of the flexible transformer demonstration project;

[0040] Figure 4 A graph showing the load rate curves of distribution transformers before and after the interconnection of distribution areas;

[0041] Figure 5(a) shows the flexible load response before the interconnection of the transformer substations;

[0042] Figure 5(b) shows the flexible load response after the interconnection of transformer substations;

[0043] Figure 6 A diagram showing the amount of light curtailment and load deficit in the front-end area for interconnection;

[0044] Figure 7 This is a power curve of the distribution transformers before and after the interconnection. Detailed Implementation

[0045] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0046] like Figure 1 As shown, a multi-time-scale coordinated scheduling method for low-voltage DC interconnected distribution areas includes the following steps:

[0047] Step 1: Based on the photovoltaic power output prediction and photovoltaic prediction error of the low-voltage DC interconnected distribution area, the historical photovoltaic power output data is screened for scenarios, and the screened scenarios are clustered. The scenario screening of historical photovoltaic power output data includes scenario screening of long-term historical photovoltaic power output data and scenario screening of short-term historical photovoltaic power output data.

[0048] Step 2: Based on the scenario clustering results of long-term historical photovoltaic power output data, the day-ahead long-term optimization scheduling method is adopted to obtain the day-ahead scheduling plan. The day-ahead long-term optimization scheduling method includes an upper-level stochastic programming method and a lower-level robust optimization method. The upper-level stochastic programming method aims to minimize the expected cost of the worst scenario in each class, while the lower-level robust optimization method aims to minimize the intraday scheduling cost of the worst scenario in each class.

[0049] Step 3: Based on the scenario clustering results of short-term historical photovoltaic power output data, the intraday rolling optimization scheduling method is used to make rolling corrections to the day-ahead scheduling plan;

[0050] Step 4: Based on the actual photovoltaic output at the current moment, the day-ahead scheduling plan with rolling correction is used as input, and a real-time optimization scheduling method is adopted to schedule the power of the low-voltage DC interconnected transformer area converter, the load reduction power, and the electric vehicle quota power in real time.

[0051] The above method employs a day-ahead long-term optimization scheduling method to obtain the day-ahead scheduling plan, uses an intraday rolling optimization scheduling method to continuously revise the day-ahead scheduling plan, and uses a real-time optimization scheduling method to schedule converter power, load reduction power, and electric vehicle quota power. This achieves multi-time-scale coordinated scheduling of low-voltage DC interconnected distribution areas, enabling flexible scheduling of controllable resources in the distribution area at multiple time scales, reducing the impact of photovoltaic power output randomness, and improving the economy and safety of distribution area operation.

[0052] like Figure 2 As shown, before optimization, modeling can be performed on the terminal equipment and flexible loads in the distribution area to construct a multi-equipment coordinated optimization model for the low-voltage DC interconnected distribution area. This model is essentially a deterministic optimization model. Based on this model, day-ahead, intraday short-term, and intraday real-time scheduling models can be constructed according to the flexibility of different equipment. This allows scheduling decisions for different equipment to be made at the day-ahead, intraday short-term, and intraday real-time stages. The model includes a minimum operating cost objective function, as well as operating constraints for distribution transformers, energy storage, AC / DC converters, flexible loads, and DC buses.

[0053] The objective function specifically aims to minimize the total operating cost of the interconnected distribution network, including the power purchase cost at the distribution transformer port, the depreciation cost of energy storage, the load shedding penalty cost, the load limit penalty cost for electric vehicle charging stations, and the curtailment penalty cost. This can be expressed by the following formula:

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060] Where u is the decision variable for controllable equipment in the distribution radio area, and C tap C is the cost of purchasing electricity at the distribution transformer port. ESS Energy storage depreciation cost, C CL To reduce the cost of load reduction penalties, C Lim C PV The cost of light wastage penalty, T is the total scheduling period, and c is the total cost of light wastage penalty. e,t For time-of-use electricity pricing, N tap P represents the total number of distribution transformers in the interconnected distribution area. tap,l,t Let C be the output power of the distribution transformer l during time period t.fe,l,t For the iron loss of distribution transformer l during time period t, C cu,l,t N represents the copper loss of distribution transformer l during time period t. ESS The number of energy storage devices in the interconnected distribution area. The unit capacity investment cost of energy storage j, For the rated capacity of energy storage j, N life,j For the maximum number of charge-discharge cycles of energy storage j, u ESS,j,t c is the variable for switching the charge / discharge state of energy storage j. LS,t For the period t, the load compensation price can be reduced, P LS,l,t λ is the load reduction capacity of the distribution transformer l during time period t. EV This is the power limit penalty coefficient for electric vehicles. The power limit for AC electric vehicle charging stations connected to distribution transformer l. The power limit for DC electric vehicle charging stations on the DC bus, c cut The price is the penalty for abandoned light units. Let t be the power of abandoned light during time period t.

[0061] The operating constraints of distribution transformers can be expressed by the following formula:

[0062]

[0063] Where: β l,t Let S be the load factor of distribution transformer l during time period t. tap,l C is the rated capacity of distribution transformer l. fe For the total iron loss of the distribution transformer in the interconnected area, C cu,t L represents the total copper loss of the distribution transformer in the interconnected area during time period t. fe,l L is the rated iron loss of the distribution transformer l; cu,l P is the rated copper loss of distribution transformer l. L,l,t P represents the load power carried by distribution transformer l. VSC,l,t For the VSC power connected to distribution transformer l, The required power for an AC electric vehicle charging station connected to distribution transformer l. This refers to the upper limit of the power output of the distribution transformer.

[0064] The operational constraints of energy storage can be expressed by the following formula:

[0065]

[0066] Among them, P ch,j,t and P dis,j,t These represent the charging and discharging power of energy storage j, and u, respectively. ch,j,t and u dis,j,tThese represent the energy storage charging and discharging states, u ch,j,t and u ch,j,t-1 The charging states of energy storage j during time period t and time period t-1 are respectively, u dis,j,t and u dis,j,t-1 P represents the discharge state of energy storage j during time period t and time period t-1, respectively. ch,j,t and P dis,j,t These represent the charging and discharging power of energy storage j, and These are the upper limits for energy storage charging and discharging power, E j,t and E j,t-1 The stored energy j represents the amount of energy stored during time period t and time period t-1, respectively. j Let η be the self-discharge coefficient of energy storage j. ch,j and η dis,j These represent the charging and discharging efficiencies of energy storage j and u, respectively. ESS,j,t This refers to the charging and discharging switching state of energy storage j. Ω represents the maximum number of charge / discharge cycles for energy storage within the scheduling cycle. ESS Energy storage collection.

[0067] The converter capacity constraint can be expressed by the formula:

[0068]

[0069] Among them, P VSC,l,t The active power of VSC connected to distribution transformer l, For P VSC,l,t The upper limit.

[0070] The flexible load response constraint can be expressed by the formula:

[0071]

[0072] Among them, P L,l,t The load power carried by distribution transformer l after flexible load response. P represents the load power carried by distribution transformer l before flexible load response. LT,l,t For the transferable load transfer power, u LS,l,t This represents the load reduction state during time period t. The upper limit of the load reduction capacity of the distribution transformer l during time period t, Ω LS Load set can be reduced, u LT,l,t The response state of the transferable load carried by the distribution transformer l during time period t; This is the upper limit of the transferable load power carried by the distribution transformer l.

[0073] The DC bus power balance constraint can be expressed by the formula:

[0074]

[0075] Where, η VSC,l For the VSC conversion efficiency connected to distribution transformer l, N ESS P represents the number of energy storage devices connected to the DC bus. ch,j,t and P dis,j,t These represent the charging and discharging power of energy storage j, For the power required by DC charging piles, P PV,t The photovoltaic output during period t.

[0076] Based on the above model, and according to the predicted photovoltaic output and prediction error of low-voltage DC interconnected distribution areas, historical photovoltaic output data are filtered for specific scenarios, and the selected scenarios are clustered. The specific process can be as follows:

[0077] 1) Based on the predicted photovoltaic output and prediction error of the low-voltage DC interconnected distribution area, the photovoltaic fluctuation range is determined, which can be expressed by the formula:

[0078] (1-δ t,max )P PV,t,max ≤P PV,t ≤(1+δ t,max )P PV,t,max

[0079] Where, δ t,max P represents the photovoltaic prediction error. PV,t,max For P PV,t The upper limit;

[0080] 2) Filter historical photovoltaic power output data to obtain scenarios within the photovoltaic fluctuation range;

[0081] Among them, the scenario screening of historical photovoltaic power output data includes scenario screening of long-term historical photovoltaic power output data and scenario screening of short-term historical photovoltaic power output data;

[0082] 3) Cluster the scenarios within the photovoltaic fluctuation range;

[0083] The K-means clustering algorithm is used here, and the specific program logic is shown in Table 1.

[0084] Table 1. K-means clustering algorithm program logic

[0085]

[0086] Based on the scenario clustering results, considering the flexibility of each device, and combining stochastic programming and robust optimization, multi-time-scale coordinated scheduling of low-voltage DC interconnected distribution areas is carried out, including day-ahead long-term optimization scheduling, intraday rolling optimization scheduling, and real-time optimization scheduling.

[0087] Based on the scenario clustering results of long-term historical photovoltaic power output data, the day-ahead long-term optimization scheduling method is used to obtain the day-ahead scheduling plan:

[0088] The daytime scheduling is based on a 1-hour interval, optimizing the daytime scheduling plan for the next 24 hours. Based on the cluster analysis results, a stochastic-robust two-level optimization method is used to handle the stochasticity of photovoltaic power output.

[0089] Current long-running optimization scheduling methods include upper-level stochastic programming methods and lower-level robust optimization methods.

[0090] The upper-level stochastic programming method optimizes the day-ahead scheduling plan with the objective of minimizing the expected cost of the worst-case scenario across all classes. The worst-case scenario is defined as the scenario with the highest intraday scheduling cost. The day-ahead scheduling plan optimized by the upper-level stochastic programming method includes the transferable load transfer period, transferable load transfer power, load reduction period, energy storage charging / discharging status, base value of energy storage charging / discharging power, and base value of distribution transformer power. The lower-level robust optimization method optimizes the day-ahead scheduling plan with the objective of minimizing the intraday scheduling cost of the worst-case scenario across all classes. The day-ahead scheduling plan optimized by the lower-level robust optimization method includes the adjustment amount of energy storage charging / discharging power, the adjustment amount of distribution transformer power, the reduction power of loads that can be reduced, the load limit power of electric vehicles, and the converter power.

[0091] Compared to traditional stochastic programming, this stochastic-robust optimization reduces the sample space and improves solution efficiency through scenario classification. Simultaneously, by selecting the worst-case scenario from each category as a representative scenario through lower-level robust optimization, it enhances the model's robustness. Compared to traditional robust optimization, stochastic-robust optimization, by incorporating the probability distribution of stochastic programming, weakens the robustness of the worst-case scenario and improves the model's cost-effectiveness.

[0092] The compact form of the optimization model used in the upper-level stochastic programming method is as follows:

[0093]

[0094] Ax+By k +ζ k ≤c

[0095]

[0096] Where x is the day-ahead scheduling plan, K C ρ represents the total number of clusters. k The probability of being in the k-th class is equal to the number of scenes N in the k-th class. k Divide by the total number of scenes M, f 2,k Let y be the intraday scheduling cost for the worst-case scenario of type k, where A, B, and c are the coefficient matrices or vectors corresponding to the constraints. kFor the intraday scheduling plan corresponding to the worst-case scenario of type k, ζ k For the worst-case scenario k, the photovoltaic output is given, where f1 is the day-ahead dispatch cost, and f1 = C. tap +C ESS .

[0097] The compact form of the optimization model used in the lower-level robust method is as follows:

[0098]

[0099]

[0100] Ax+By k,i +ζ k,i ≤c

[0101] for: k = 1, 2, ..., K C

[0102] Among them, f 2,k Ω represents the intraday scheduling cost for the worst-case scenario of type k. k Let f be the set of similar scenes in the k-th class. 2,k,i Let y be the intraday scheduling cost of the i-th similar scenario in the k-th class. k,i Let ζ be the intraday scheduling strategy for the i-th similar scenario in the k-th class. k,i Photovoltaic output of the i-th similar scenario in the k-th class.

[0103] Intraday scheduling costs Where, N tap This refers to the number of distribution transformers. and These are the penalty coefficients for intraday increases and decreases in distribution transformer power (higher than the time-of-use electricity price). and These represent the upward and downward adjustments of the power output of the distribution transformer within 1 day during time period t.

[0104] The constraints include the aforementioned distribution transformer operation constraints, energy storage operation constraints, converter operation constraints, flexible load response constraints, and DC bus power balance constraints.

[0105] Because the variables are coupled between the upper-level stochastic programming and the lower-level robust optimization, making direct solutions difficult, the difference is divided into the following principal subproblems for iterative solution based on the idea of ​​the CCG algorithm:

[0106] The main problem is to minimize the expected cost of the worst-case scenario in each class to obtain the current optimal day-ahead scheduling strategy. The main problem model is as follows:

[0107]

[0108] Ax+Byk,i +ζ k,i ≤c

[0109] f 2,k ≥f 2,k,i (y k,i ,ζ k,i )

[0110]

[0111] in, Let be the set of current adverse scenarios of class k in the m-th iteration.

[0112] The sub-problem is to minimize the intraday scheduling cost of the worst-case scenario among all types of scenarios based on the current day-ahead scheduling plan, and to identify the worst-case scenario. The sub-problem model is as follows:

[0113]

[0114]

[0115] Ax * +By k,i +ζ k,i ≤c

[0116] for: k = 1, ..., K C

[0117] Where, x * The current-day optimal scheduling strategy obtained from the main problem. The worst-case scenario of the k-th class identified for the m-th iteration subproblem. Let $\frac{ ...

[0118] The specific program logic of the random-robust two-level optimization decomposition iterative solution algorithm is shown in Table 2.

[0119] Table 2. Program Logic of the Random-Robust Two-Level Optimization Decomposition Iterative Solution Algorithm

[0120]

[0121] The stochastic-robust two-layer optimization scheduling method recently adopted achieves a good balance between the economy of stochastic programming and the robustness of robust optimization, and significantly improves the solution efficiency.

[0122] Based on the scenario clustering results of short-term historical photovoltaic power output data, an intraday rolling optimization scheduling method is adopted to continuously adjust the power of distribution transformers and the charging and discharging power of energy storage in the day-ahead scheduling plan:

[0123] The intraday rolling optimization scheduling method employs stochastic programming. Based on short-term photovoltaic forecast data and clustering results, the intraday short-term optimization scheduling uses stochastic programming to continuously optimize the intraday short-term scheduling plan for the next hour at 15-minute intervals. Specifically, it minimizes the expected costs for future time periods t+1, t+2, and t+3 within time period t, then moves to time period t+1, and so on. The intraday short-term scheduling plan includes energy storage charging and discharging adjustment power, distribution transformer adjustment power, converter station power, load reduction power, and charging station quota power.

[0124] Intraday short-term dispatch aims to minimize the expected cost for each scenario, i.e., f2. Dispatch costs include distribution transformer power adjustment penalty costs, reduceable load response compensation costs, and charging station power limit penalty costs, which can be expressed by the formula:

[0125]

[0126] Where t0 is the current scheduling period; s is the scene index, and K in ρ represents the number of scenes. s Let P be the probability of the s-th scene. LS,s,l,t The real-time power reduction within the day for the s-th scenario in time period t. This represents the daily real-time load limit power of the DC electric vehicle charging station corresponding to the s-th scenario in time period t. This represents the daily real-time load limit power of the AC electric vehicle charging station corresponding to the s-th scenario in time period t. This represents the real-time photovoltaic power reduction for the s-th scenario during time period t.

[0127] The constraints include upper limit constraints on the power of distribution transformers, upper limit constraints on the charging and discharging power of energy storage, and power balance constraints, upper limit constraints on the power of converters, upper limit constraints on the power reduction of loads that can be reduced, and power balance constraints on DC buses.

[0128] Based on the actual photovoltaic output at the current moment, and using the day-ahead scheduling plan with rolling corrections as input, a real-time optimization scheduling method is adopted to schedule converter power, load reduction power, and electric vehicle quota power in real time.

[0129] The goal of real-time optimization scheduling is to minimize the operating cost at the current moment, i.e., f real,t At its minimum, based on the actual output of photovoltaic power at the current moment, the converter power, load reduction power, and electric vehicle quota power are flexibly dispatched at the second level. The dispatch cost includes the cost of compensation for the load reduction response and the cost of the charging station power limit penalty, which can be expressed by the formula:

[0130]

[0131] The constraints include upper limit constraints on converter power, upper limit constraints on power reduction for load reduction, and DC bus power balance constraints.

[0132] To verify the above method, set... Figure 3 The flexible distribution transformer demonstration project shown consists of four low-voltage distribution areas interconnected via low-voltage DC. It includes three AC charging piles and three DC charging piles with capacities of 60kW and 120kW respectively. Energy storage and photovoltaic systems are directly connected to the low-voltage DC bus with capacities of 200kW and 500kW respectively. Considering the future load regulation role in the distribution areas, transferable and reduceable loads are included, accounting for 0.2 and 0.1% of the baseline load respectively. The distribution transformer capacities of distribution areas 1 to 4 are 630kVA, 630kVA, 800kVA, and 800kVA respectively, and the AC / DC converter capacities are 200kVA, 200kVA, 200kVA, and 300kVA respectively, with a conversion efficiency of 0.98.

[0133] Scenario 1: Low-voltage distribution areas are not interconnected; DC charging piles are connected to distribution area T2, energy storage is connected to distribution area T3, and photovoltaics are connected to distribution area T4. Scenario 2: Distribution areas are interconnected via low-voltage DC; photovoltaics, energy storage, and distributed photovoltaics are connected to the DC bus.

[0134] 1. Flexible distribution area controllable resource scheduling analysis

[0135] The load rate curves of the distribution transformers before and after the interconnection of low-voltage distribution areas are as follows: Figure 4 As shown, before interconnection (Scenario 1), the load rates of each distribution transformer varied greatly, resulting in power backfeeding. Simultaneously, the limited transformer capacity caused a load deficit of 16.9 kWh. After interconnection (Scenario 2), the load rates of each distribution transformer in the distribution area were relatively balanced, with no load deficit issue. The load transfer dispatch response before and after the distribution area interconnection is as follows: Figure 5(a) , 5(b) As shown, the amount of abandoned light and the load deficit in the interconnection front area are as follows: Figure 6 Show. Depend on Figure 5(a) , 5(b) As shown in point 6, before the interconnection, the T4 distribution area relied solely on flexible load response to promote photovoltaic (PV) consumption, resulting in 206.9 kWh of curtailed PV power, accounting for 5.7% of the total PV power. After the interconnection, the complementary characteristics between different distribution areas promoted PV consumption, and no curtailment occurred. Therefore, the low-voltage DC interconnection of the distribution areas can effectively reduce load deficit, improve PV consumption, and alleviate the problem of tight distribution transformer capacity.

[0136] The power change curves of the distribution transformer two days ago and during the day are as follows: Figure 7 As shown, due to the randomness of photovoltaic output being taken into account during the day-ahead dispatching phase, the power fluctuations of distribution transformers are relatively small during the day-ahead and intraday periods, thus better adapting to the randomness of photovoltaic output.

[0137] 2. Analysis of the current random-robust optimization method

[0138] Table 3 shows a comparison between the stochastic-robust optimization method and stochastic programming and robust optimization methods.

[0139] Table 3 Comparison Table

[0140]

[0141] Table 3 shows that compared to robust optimization, stochastic-robust optimization reduces the day-ahead cost by 49.95 yuan, indicating better economic efficiency. Compared to stochastic programming, stochastic-robust optimization reduces the intra-day average cost and maximum cost by 16.52 yuan and 42.749 yuan, respectively, indicating improved robustness. Therefore, the stochastic-robust method achieves a good balance between the economic efficiency of stochastic programming and the robustness of robust optimization. Furthermore, compared to stochastic programming and robust optimization, the solution time is reduced by 97.7% and 96.6%, respectively, significantly improving solution efficiency.

[0142] After achieving low-voltage DC interconnection in the power distribution area through the transformation of DC charging piles, the above methods can effectively reduce the load deficit of electric vehicles, improve photovoltaic absorption, and alleviate the problem of power distribution transformer capacity shortage.

[0143] Based on the same technical solution, this invention also discloses a software device for the above method, a multi-time-scale coordinated scheduling device for low-voltage DC interconnected distribution areas, comprising:

[0144] The clustering module filters historical photovoltaic output data into scenarios based on photovoltaic power generation forecasts and forecast errors, and then clusters the selected scenarios. The scenario filtering of historical photovoltaic output data includes filtering scenarios from long-term historical photovoltaic output data and filtering scenarios from short-term historical photovoltaic output data.

[0145] The day-ahead long-term optimization scheduling module uses the day-ahead long-term optimization scheduling method to obtain the day-ahead scheduling plan based on the scenario clustering results of long-term historical photovoltaic power output data. The day-ahead long-term optimization scheduling method includes an upper-level stochastic programming method and a lower-level robust optimization method. The upper-level stochastic programming method aims to minimize the expected cost of the worst scenario in each class, while the lower-level robust optimization method aims to minimize the intraday scheduling cost of the worst scenario in each class.

[0146] The intraday rolling optimization scheduling module uses the intraday rolling optimization scheduling method to continuously revise the day-ahead scheduling plan based on the scenario clustering results of short-term historical photovoltaic power output data.

[0147] The real-time optimization scheduling module, based on the actual output of photovoltaic power at the current moment, takes the day-ahead scheduling plan with rolling correction as input, and uses the real-time optimization scheduling method to schedule the power of converters in low-voltage DC interconnected distribution areas, load reduction power, and electric vehicle quota power in real time.

[0148] The processing flow and methods of each module in the above device are consistent, and will not be described again here.

[0149] Based on the same technical solution, the present invention also discloses a computer-readable storage medium for storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform a multi-timescale coordinated scheduling method for low-voltage DC interconnected distribution areas.

[0150] Based on the same technical solution, the present invention also discloses a computing device, including one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for executing a multi-timescale coordinated scheduling method for low-voltage DC interconnected distribution areas.

[0151] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0152] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0153] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0154] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0155] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A low-voltage direct-current interconnected transformer area multi-time scale coordinated scheduling method, characterized in that, The method comprises the following steps: According to the low-voltage direct-current interconnected area photovoltaic predicted power and the photovoltaic prediction error, the historical photovoltaic output data is scene filtered and the filtered scenes are clustered; wherein, the historical photovoltaic output data is scene filtered, including scene filtering of long-time historical photovoltaic output data and scene filtering of short-time historical photovoltaic output data; Based on the scene clustering result of long-time historical photovoltaic output data, a day-ahead long-time optimization scheduling method is used to obtain a day-ahead scheduling plan; wherein, the day-ahead long-time optimization scheduling method includes an upper random programming method and a lower robust optimization method, the upper random programming method takes the minimum expected cost of the worst scene in each type as the target, and the lower robust optimization method takes the minimum intra-day scheduling cost of the worst scene in each type as the target; Based on the scene clustering result of short-time historical photovoltaic output data, an intra-day rolling optimization scheduling method is used to roll correct the day-ahead scheduling plan; Based on the current photovoltaic actual output, the roll-corrected day-ahead scheduling plan is taken as the input, and a real-time optimization scheduling method is used to schedule the low-voltage direct-current interconnected area converter power, load reduction power and electric vehicle limit power in real time; The upper random programming method takes the minimum expected cost of the worst scene in each type as the target to optimize the day-ahead scheduling plan; wherein, the day-ahead scheduling plan optimized by the upper random programming optimization method includes transferable load transfer period, transferable load transfer power, reducible load reduction period, energy storage charging and discharging state, energy storage charging and discharging power base value and distribution transformer power base value; The optimization model used by the upper random programming optimization method is: Ax + By k + ζ k ≤ c where x is the day-ahead dispatch schedule, f1(x) is the day-ahead dispatch cost, K C is the total number of clusters, p k is the probability of the kth cluster, f 2,k is the intra-day dispatch cost of the worst scenario of the kth cluster, A, B, c are the coefficient matrix or vector corresponding to the constraints, y k is the intra-day dispatch schedule corresponding to the worst scenario of the kth cluster, ζ k is the PV power output of the worst scenario of the kth cluster.

2. The low-voltage direct-current interconnected transformer area multi-time scale coordinated dispatching method according to claim 1, characterized in that, According to the low-voltage direct-current interconnected area photovoltaic predicted power and the photovoltaic prediction error, the historical photovoltaic output data is scene filtered and the filtered scenes are clustered, including: According to the photovoltaic predicted power and the photovoltaic prediction error, the photovoltaic fluctuation interval is determined; The historical photovoltaic output data is scene filtered to obtain the scenes in the photovoltaic fluctuation interval; The scenes in the photovoltaic fluctuation interval are clustered.

3. The low-voltage direct-current interconnected transformer area multi-time scale coordinated dispatching method according to claim 1, characterized in that, The lower robust optimization method takes the minimum intra-day scheduling cost of the worst scene in each type as the target to optimize the day-ahead scheduling plan; wherein, the day-ahead scheduling plan optimized by the lower robust optimization method includes energy storage charging and discharging power adjustment, distribution transformer power adjustment, reducible load reduction power, electric vehicle load limit power and converter power.

4. The low-voltage direct-current interconnected transformer area multi-time scale coordinated dispatching method according to claim 3, characterized in that, The optimization model used by the lower robust optimization method is: Ax + By k,i + ζ k,i ≤ c for k = 1, 2,... K C where ζ k is the photovoltaic power output of the kth worst-case scenario, f 2,k is the intra-day scheduling cost of the kth worst-case scenario, A, B, c are the coefficient matrix or vector corresponding to the constraints, Ω k is the set of similar scenarios in the kth cluster, f 2,k,i is the intra-day scheduling cost of the ith similar scenario in the kth cluster, K C is the total number of clusters, y k,i is the intra-day scheduling strategy of the ith similar scenario in the kth cluster, ζ k,i is the photovoltaic power output of the ith similar scenario in the kth cluster.

5. The low-voltage direct-current interconnected transformer area multi-time scale coordinated dispatching method according to claim 1, characterized in that, The intra-day rolling optimization scheduling method uses a random programming method.

6. A low-voltage direct-current interconnected transformer area multi-time scale coordinated scheduling device, characterized in that, The device comprises: A clustering module, according to the low-voltage direct-current interconnected area photovoltaic predicted power and the photovoltaic prediction error, the historical photovoltaic output data is scene filtered and the filtered scenes are clustered; wherein, the historical photovoltaic output data is scene filtered, including scene filtering of long-time historical photovoltaic output data and scene filtering of short-time historical photovoltaic output data; The day-ahead long-time optimization scheduling module adopts a day-ahead long-time optimization scheduling method based on a scene clustering result of long-time historical photovoltaic output data to obtain a day-ahead scheduling plan; the day-ahead long-time optimization scheduling method includes a random programming method of an upper layer and a robust optimization method of a lower layer, the random programming method of the upper layer takes the minimum expected cost of the worst scene in each type as a target, and the robust optimization method of the lower layer takes the minimum intra-day scheduling cost of the worst scene in each type as a target; The intra-day rolling optimization scheduling module adopts an intra-day random rolling optimization scheduling method based on a scene clustering result of short-time historical photovoltaic output data to perform rolling correction on the day-ahead scheduling plan; The real-time optimization scheduling module adopts a real-time optimization scheduling method to perform real-time scheduling on low-voltage direct-current interconnected area converter power, load reduction power and electric vehicle limit power by taking the rolling corrected day-ahead scheduling plan as input based on current photovoltaic actual output.

7. A computer-readable storage medium storing one or more programs, the one or more programs comprising instructions that when executed by a computer cause the computer to perform a method of any of claims 1-6. The one or more programs include instructions that when executed by a computing device, cause the computing device to perform any of the methods of claims 1-5.

8. A computing device, comprising: Comprise: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs include instructions for performing any of the methods of claims 1-5.

Images