Active distribution network optimization dispatching method and system based on stochastic model predictive control

Abstract

The present disclosure provides an active distribution network optimization scheduling method and system based on stochastic model predictive control, comprising the following steps: obtaining a backup configuration interval of renewable energy; obtaining an optimal scheduling plan for the active distribution network according to the obtained backup configuration interval and a preset multi-time scale active distribution network optimization scheduling model; wherein the multi-time scale active distribution network optimization scheduling model includes a day-ahead scheduling stage, an intra-day scheduling stage and a real-time scheduling stage, adopts stochastic model predictive control, takes minimum operating cost as a target, and takes backup constraints as constraints.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of distribution network optimization and scheduling, and particularly relates to an active distribution network optimization and scheduling method and system based on stochastic model predictive control. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] The high penetration of renewable energy sources (RES) exposes active distribution networks (ADNs) to increased uncertainty and randomness, making their scheduling more complex. In particular, the proper integration of wind turbines and photovoltaics is considered critical for stabilizing power fluctuations and arranging backup capacity. Furthermore, large-scale fluctuations in renewable energy within ADNs pose a serious threat to operational safety and economic viability and can also impact upstream transmission networks.

[0004] Improving the forecast accuracy of renewable energy has long been a hot topic and an effective measure to address uncertainty and randomness in optimized operation. Given the varying probability distributions of forecast errors across different time periods, several related studies exist. One paper statistically analyzed forecast error data for different time periods and proposed a time series segmentation fitting method. Further analysis revealed strong temporal correlation between forecast errors in adjacent time periods. A multivariate joint cumulative probability distribution function was established based on copula theory, finding that the closer the wind farms are, the stronger the spatial correlation of their forecast errors. Based on this, accounting for the spatiotemporal correlation of forecast errors can lead to more realistic forecasts for renewable energy. However, in most existing forecast error studies, the results are primarily used for day-ahead unit commitment decisions and capacity optimization of energy storage systems. Few studies have considered copula theory in active distribution network scheduling, and even fewer have used stochastic model predictive control frameworks. This can lead to significant discrepancies in forecasts for multiple wind turbines or photovoltaic power plants, resulting in excessively fluctuating scenarios and directly impacting the applicability of active distribution network scheduling results.

[0005] Properly arranging reserve capacity is also an effective means of addressing the uncertainty and randomness of renewable energy. As renewable energy penetration continues to rise, increasing the reserve capacity of the entire power system is a conservative solution to this uncertainty, which will incur higher operating costs. Therefore, optimizing reserve capacity is crucial.

[0006] In terms of scheduling frameworks, Model Predictive Control (MPC) has achieved significant success in addressing the uncertainty of renewable energy. To more effectively address the uncertainty of renewable energy, deterministic optimization methods are no longer applicable. Uncertainty optimization methods are gaining increasing attention, with stochastic optimization being widely used. Existing research does not conduct statistical and correlation analysis across time periods, nor does it consider the uncertainty of prediction errors. This results in significant deviations between scenarios and actual conditions, and can even render scheduling results inapplicable. The lack of integration of stochastic model predictive control with multi-timescale scheduling results in a lack of coordination among distributed resources at each scheduling stage, making it impossible to provide the most up-to-date data to the active distribution network.

[0007] Therefore, in the context of large-scale access of renewable energy to distribution networks, there is an urgent need to conduct research on active distribution network optimization and scheduling. Summary of the Invention

[0008] In order to solve the above problems, the present disclosure proposes a method and system for optimizing and dispatching active distribution networks based on stochastic model predictive control, which optimizes and dispatches active distribution networks at multiple time scales based on stochastic model predictive control, including a backup arrangement strategy and a scheduling framework; a multi-time scale active distribution network optimization scheduling model based on stochastic model predictive control is used to track the random fluctuations of renewable energy and solve operational risk problems.

[0009] According to some embodiments, a first solution of the present disclosure provides an active distribution network optimization scheduling method based on stochastic model predictive control, which adopts the following technical solutions:

[0010] An active distribution network optimization scheduling method based on stochastic model predictive control includes the following steps:

[0011] Obtaining the backup configuration range of renewable energy;

[0012] According to the obtained backup configuration interval and the preset multi-time-scale active distribution network optimization scheduling model, the optimal scheduling plan of the active distribution network is obtained;

[0013] Among them, the multi-time-scale active distribution network optimization scheduling model includes a day-ahead scheduling stage, an intraday scheduling stage and a real-time scheduling stage, adopts a stochastic model predictive control, takes the minimum operating cost as the goal, and takes the reserve constraint as the constraint condition.

[0014] As a further technical limitation, considering the trade-off between backup cost and operational risk cost, the load loss risk and renewable energy curtailment risk are quantified based on the conditional value at risk of probability distribution to obtain the optimal error range of renewable energy, that is, the backup configuration range of renewable energy.

[0015] As a further technical limitation, in the day-ahead scheduling phase, the multi-time-scale active distribution network optimization scheduling model aims to minimize the operating cost and backup cost of the active distribution network, while taking into account the backup capacity provided by controllable resources including controllable distributed generators, energy storage systems, switching capacitor banks, static VAR compensators, on-load tap-changing transformers, and electricity purchased from the transmission network.

[0016] Furthermore, in the intraday scheduling stage, the multi-time-scale active distribution network optimization scheduling model aims to minimize the operating cost and backup cost of the active distribution network, while taking into account the backup capacity provided by controllable resources including controllable distributed generators, energy storage systems, static VAR compensators, on-load tap-changing transformers and electricity purchased from the transmission network.

[0017] Furthermore, in the real-time scheduling stage, the multi-time-scale active distribution network optimization scheduling model updates the prediction information of the active distribution network scenario in each stage based on the prediction information of the active distribution network scenario, with the goal of minimizing power fluctuation and operating cost in the active distribution network scenario, combined with the actual measured values ​​of renewable energy, and performs closed-loop control on the scheduling of the active distribution network.

[0018] Furthermore, the correlation analysis of the prediction information of the active distribution network scenario is performed based on the copula theory to reduce the volatility and prediction error of renewable energy processing in the generation scenario.

[0019] As a further technical limitation, the constraints of the multi-time-scale active distribution network optimization scheduling model also include flow constraints, branch capacity constraints, node voltage constraints, renewable energy power generation operation constraints, controllable distributed power generation operation constraints, on-load tap-changing transformer operation constraints, static VAR compensation device operation constraints, compensation capacitor group operation constraints and energy storage device operation constraints.

[0020] According to some embodiments, a second solution of the present disclosure provides an active distribution network optimization scheduling system based on stochastic model predictive control, which adopts the following technical solutions:

[0021] An active distribution network optimization dispatching system based on stochastic model predictive control, comprising:

[0022] An acquisition module, used to obtain a backup configuration interval of renewable energy;

[0023] The optimization module obtains the optimal scheduling plan of the active distribution network based on the obtained backup configuration interval and the preset multi-time scale active distribution network optimization scheduling model;

[0024] Among them, the multi-time-scale active distribution network optimization scheduling model includes a day-ahead scheduling stage, an intraday scheduling stage and a real-time scheduling stage, adopts a stochastic model predictive control, takes the minimum operating cost as the goal, and takes the reserve constraint as the constraint condition.

[0025] According to some embodiments, a third solution of the present disclosure provides a computer-readable storage medium, which adopts the following technical solution:

[0026] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps of the active distribution network optimization scheduling method based on stochastic model predictive control as described in the first aspect of the present disclosure.

[0027] According to some embodiments, a fourth solution of the present disclosure provides an electronic device, which adopts the following technical solution:

[0028] An electronic device comprises a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the active distribution network optimization scheduling method based on stochastic model predictive control as described in the first aspect of the present disclosure are implemented.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1) This paper considers the trade-off between backup and operational risk costs and introduces conditional value-at-risk (CVR) to determine the optimal backup configuration. Based on the probability distribution, the CVR quantifies the risk of load loss and RES curtailment, thereby minimizing the operational risk and backup costs, finding the optimal error range for renewable energy, and determining the optimal backup configuration range.

[0031] 2) To accurately track the volatility of renewable energy, this paper applies copula theory to establish a high-dimensional prediction error model and capture spatiotemporal characteristics. At the same time, based on the updated prediction error curve, it generates typical renewable energy output scenarios for each scheduling stage, ensuring the effectiveness and accuracy of operation.

[0032] 3) This disclosure combines stochastic model predictive control with multi-time scale scheduling to perform scheduling of active distribution networks, using stochastic optimization methods in all three stages. In the real-time stage, deviations from the intraday stage commitments are promptly corrected based on the forecast scenario to reduce power fluctuations at the interface with the transmission system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings, which constitute a part of the present disclosure, are used to provide a further understanding of the present disclosure. The exemplary embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation to the present disclosure.

[0034] Figure 1This is a flow chart of the active distribution network optimization scheduling method based on stochastic model predictive control in the first embodiment of the present disclosure;

[0035] Figure 2 This is the overall scheduling framework diagram in the first embodiment of the present disclosure;

[0036] Figure 3 is the backup scheduling strategy diagram in the first embodiment of the present disclosure;

[0037] Figure 4 This is a structural diagram of the improved IEEE 33-node power distribution network in the first embodiment of the present disclosure;

[0038] FIG5( a ) is a scene graph generated based on Copula theory in the first embodiment of the present disclosure;

[0039] FIG5( b ) is a scenario in the first embodiment of the present disclosure without considering the copula theory;

[0040] Figure 6 is the wind power output diagram predicted in the first embodiment of the present disclosure;

[0041] Figure 7 is the photovoltaic power generation diagram predicted in the first embodiment of the present disclosure;

[0042] Figure 8 is the active power output diagram of the controllable distributed generator in the first embodiment of the present disclosure;

[0043] FIG9( a ) is a power fluctuation diagram based on method 0 in embodiment 1 of the present disclosure;

[0044] FIG9( b ) is a power fluctuation diagram based on method 4 in embodiment 1 of the present disclosure;

[0045] Figure 10 Schematic diagram of power fluctuation distribution based on method 0 and method 4 in embodiment 1 of the present disclosure;

[0046] FIG11( a ) is a diagram of active power output of a controllable distributed generator and an energy storage system based on method 0 in embodiment 1 of the present disclosure;

[0047] FIG11( b ) is a diagram of active power output of a controllable distributed generator and energy storage system based on method 4 in embodiment 1 of the present disclosure;

[0048] Figure 12 This is a structural block diagram of the active distribution network optimization and scheduling system based on stochastic model predictive control in the second embodiment of the present disclosure. DETAILED DESCRIPTION

[0049] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0050] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present disclosure belongs.

[0051] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0052] In the absence of conflict, the embodiments of the present disclosure and the features thereof may be combined with each other.

[0053] Example 1

[0054] Embodiment 1 of the present disclosure introduces an active distribution network optimization scheduling method based on stochastic model predictive control.

[0055] like Figure 1 The active distribution network optimization scheduling method based on stochastic model predictive control shown in the figure includes the following steps:

[0056] Obtaining the backup configuration range of renewable energy;

[0057] According to the obtained backup configuration interval and the preset multi-time-scale active distribution network optimization scheduling model, the optimal scheduling plan of the active distribution network is obtained.

[0058] Among them, the multi-time scale active distribution network optimization scheduling model includes the day-ahead scheduling stage, the intraday scheduling stage and the real-time scheduling stage; in different scheduling stages, scheduling is carried out according to the characteristics of distributed resources for different goals. The overall scheduling framework is as follows Figure 2 As shown, the alternative scheduling strategy is as follows Figure 3 shown.

[0059] (1) Day-ahead scheduling stage

[0060] Monte Carlo sampling and scenario reduction techniques are used to predict typical RES scenarios for the next 24 hours. The distribution network is optimized to minimize operating and backup costs, while also considering the backup capabilities of controllable resources. Existing controllable resources include controllable distributed generators (CDGs), energy storage systems (ESSs), capacitor banks (CBs), static var compensators (SVCs), and on-load tap changers (OLTCs), while also considering the purchase of power and backup from the transmission grid. Capacitor banks and on-load tap changers are slow-response resources, and their operational status and amount for the next day are determined during this phase.

[0061] (2) Intraday scheduling stage

[0062] Based on a typical scenario of renewable energy output over the next four hours, distributed resources are dispatched using distribution network operating costs and backup costs as objective functions, while also considering the backup of controllable resources. Notably, controllable resources during this phase do not involve switching capacitor banks. Furthermore, this phase "commits" the exchange power at the interface between the distribution and transmission networks.

[0063] (3) Real-time scheduling stage

[0064] During this phase, the distribution network is optimized based on typical scenarios predicted for the next hour, with power fluctuations and operating costs as the objective functions. Simultaneously, the scenario forecast for each time period is updated using measured wind turbine and photovoltaic output, forming a closed-loop control system.

[0065] In this embodiment, the load shedding risk and wind curtailment risk are measured using the conditional value at risk. By weighing the reserve cost, load shedding risk, and wind curtailment risk, an optimal prediction error interval is obtained, which serves as a reference for the reserve interval.

[0066] The CVaR of wind power curtailment can be expressed as:

[0067]

[0068]

[0069] in, and is the wind power prediction error, P(e f ) corresponds to e f The probability density of .

[0070] In this embodiment, copula theory is used to analyze the correlation of prediction errors of multiple wind turbines (photovoltaic plants) to generate scenarios. Sklar's theorem laid the foundation for the development of copula theory.

[0071] According to Sklar's theorem, there exists a Copula function C that satisfies:

[0072] H(x1,x2,...,x n )=C(F1(x1),F2(x2),...F n (x n )) (3)

[0073] In this paper, the t copula function is used because it is applicable to random variables with tail correlation. According to the prediction error matrix (4) of N wind turbines, the marginal cumulative distribution function (5) of each wind turbine in each period is calculated.

[0074]

[0075] Where, represents the R-th prediction error of the N-th wind turbine in the t-th period. The fitted marginal cumulative distribution function expression is:

[0076]

[0077] in, represents the marginal cumulative distribution function of the prediction error of the Nth wind turbine in the Tth period.

[0078] The kernel density estimation method is used to fit the probability distribution because the non-parametric fitting method is more suitable for solving the "fat tail" effect of the error distribution.

[0079] The marginal cumulative distribution functions are connected by the t copula function. The t copula can be expressed as

[0080]

[0081] Among them, t ρ,k is the t copula function with k degrees of freedom and ρ correlation matrix.

[0082] The maximum likelihood estimation method is used to obtain the degrees of freedom and the correlation matrix. Therefore, given the marginal cumulative distribution function of the forecast error and the t copula parameters, a high-dimensional model of the forecast error can be obtained, expressed as:

[0083]

[0084] Among them, f e (e ij) is e ij The marginal distribution function of .

[0085] In order to improve the economic efficiency of active distribution network operation, the backup arrangement strategy needs to be considered before dispatching at each stage.

[0086] Firstly, according to the probability distribution of renewable energy forecast error in each period within the time window, the conditional value at risk is used to express the load loss risk and renewable energy curtailment risk.

[0087]

[0088] Among them, CVaR LS,P and CVaR LS,W CVaR is the risk value of load loss caused by uncertainty of photovoltaic and wind power. PC and CVaR WC are the conditional value at risk of PV curtailment and wind power curtailment, respectively.

[0089] Then, the conditional value at risk is optimized to balance various backup costs, load loss risks, and renewable energy curtailment risks, and the optimal error range of renewable energy output is obtained:

[0090]

[0091]

[0092] Among them, R t For the arranged spare, c LS ,c PC ,c WC , and c WC They are load loss cost, photovoltaic reduction cost, wind power reduction cost and standby cost.

[0093] Finally, the backup of the active distribution network is arranged according to the obtained interval.

[0094] 1. Objective Function of Multi-Time-Scale Active Distribution Network Optimal Scheduling Model

[0095] 1) Day-ahead scheduling and intraday scheduling

[0096] In order to ensure economic operation, minimizing the operating cost is taken as the optimization goal in the day-ahead and intra-day stages, that is,

[0097]

[0098] Where DA(ID) stands for the day-ahead (intraday) phase, T and t represent the time window (24 hours for the day-ahead phase and 4 hours for the intraday phase) and the time period (1 hour resolution for the day-ahead phase and 15 minutes resolution for the intraday phase), respectively. represents the electricity cost of the transmission network. To maintain generality, the peak and valley electricity prices are used. are the charging and discharging cost of ESS and the fuel cost of controllable DG respectively. s is the probability of scenario s.

[0099] 2) Real-time scheduling stage

[0100] Real-time dispatch is to minimize the deviation from the exchange power commitment for the next hour proposed during the intraday phase while ensuring economic operation. Its objective function is as follows:

[0101]

[0102]

[0103] The superscript RT stands for real-time phase, T and t stand for time window (1 hour ahead) and time period (5-minute resolution), respectively. is the penalty term for power fluctuation at the transmission network interface, ρ penalty is the weight coefficient, represents the power at the transmission network interface in scenario s, and is the power exchange value promised at the interface between the distribution network and the transmission network during the intraday dispatch phase, p s is the probability of scene s.

[0104] 2. Constraints of the Multi-time-scale Active Distribution Network Optimal Scheduling Model

[0105] 1) Backup constraints

[0106] Reserves are scheduled during the day-ahead and intraday phases. Reserve capacity is determined by the optimal error range for renewable energy, derived through pre-optimized conditional value at risk. The coordination of reserves across these phases takes into account the spatiotemporal characteristics of controllable resources. Specifically, capacitor banks are only available during the day-ahead phase, while the remaining resources can provide reserves during the intraday phase.

[0107] The alternative constraints are as follows:

[0108] Active reserve constraints:

[0109]

[0110] in, and They are the up / down backup provided by the transmission grid, controllable distributed generators and energy storage devices, and is the reserve requirement determined by the optimal error interval of RES, and s represents the scenario they are in. That is, a reserve is arranged in each scenario. In the following constraints, s represents the scenario.

[0111] Reactive power reserve constraints:

[0112]

[0113] in, Indicates the up / down reserve provided by switching capacitor banks and static VAR compensators.

[0114] 2) Power flow constraints

[0115] According to the characteristics of the distribution network, this paper adopts the power flow constraint equation in the existing technology to ensure the accuracy of the results on the basis of easy solution. The power flow constraint equation is as follows:

[0116]

[0117] Among them, α(j) is the set of first nodes of the branch with node j as the end node; β(j) is the set of last nodes of the branch with node j as the first node; P ij,t and Q ij,t are respectively the active power and reactive power at the head end of branch ij; P j,t and Q j,t are the active power and reactive power injection values ​​of node j respectively; r ij +jx ij is the impedance of branch ij; is the square of the current amplitude flowing through branch ij; k ij,t is the transformation ratio of the on-load tap-changing transformer at branch ij; is the square of the voltage amplitude at node i.

[0118] 3) Branch capacity constraints

[0119]

[0120] Where: is the upper limit of the current amplitude of branch ij.

[0121] 4) Node voltage constraints

[0122]

[0123] Where, and are the upper and lower limits of the voltage amplitude at node i, respectively.

[0124] 5) Operational constraints of renewable energy power generation

[0125]

[0126] Among them, P RES,i,s,t , They are the dispatch value and predicted value of active power output respectively. Since renewable resources are connected to the grid through inverters, they can provide a certain amount of reactive power. RES,i,s,t is the reactive power output of renewable resources, S i,inverter Refers to the capacity of inverter i.

[0127] 6) Controllable distributed generation operation constraints

[0128]

[0129] Where, P CDG,i,t and Q CDG,i,t Represents the active and reactive output of controllable distributed generation; P CDG,min , P CDG,max , Q CDG,min and Q CDG,max Indicates the upper and lower limits of the active and reactive output of controllable distributed generation; S CDG,i,t The installed capacity of controllable distributed generation; These are the upper and lower limits of the climbing power to ensure safe operation. It is the up / down backup provided by controllable distributed generation.

[0130] 7) Operation constraints of on-load tap-changing transformers

[0131] Accurately linearize and model on-load tap-changing transformers:

[0132]

[0133] Where Δk ij is the adjustment step length; k ij,s,t , They are the adjustable gears of the on-load tap-changing transformer and their upper and lower limits.

[0134] 8) Operation constraints of static VAR compensation devices

[0135]

[0136] Where, and are the upper and lower limits of the adjustable reactive output of the static VAR compensation device respectively; and It is the up / down backup provided by the static VAR compensator.

[0137] 9) Compensation capacitor bank operation constraints

[0138]

[0139] Where Hi,t is the adjustable gear; ΔQ cb,i is the adjustment step length; H max The maximum adjustable gear for the compensation capacitor bank; and It is the upper / lower backup provided by the switching capacitor bank.

[0140] 10) Energy storage device operation constraints

[0141]

[0142] Where, P ch,i,t and P dis,i,t are the charging and discharging power of the energy storage device respectively; η ch and η dis are the charging and discharging efficiencies of the energy storage device, respectively; and are the maximum charge and discharge power of the energy storage device respectively; D ch,i,t and D dis,i,t It is a 0-1 variable that indicates the charge and discharge status of the energy storage device to ensure that the charge and discharge of the energy storage device do not occur at the same time; The capacity of the energy storage device is limited to 20%-80% in consideration of the service life of the energy storage; and The upper and lower reserves provided by the ESS are considered. When determining the reserve, the upper reserve should minimize the quotient of the dischargeable power and the remaining charge / time, while the lower reserve should minimize the quotient of the rechargeable power and the rechargeable capacity / time. This allows for energy constraints to be considered during reserve coordination.

[0143] Next, we will verify the effectiveness of the method with examples.

[0144] like Figure 4 The diagram shows an improved IEEE 33-node distribution network structure, including OLTC, controllable DG (GT, located at nodes 4 and 24), wind turbine (WT, located at nodes 13 and 29), photovoltaic (PV, located at nodes 16 and 31), energy storage device (ESS, located at node 10), switching capacitor bank (CB, located at node 21) and static VAR compensator (SVC, located at node 27).

[0145] In the model of this embodiment, the distribution network can purchase electricity and reserve from the transmission network at a certain price. Without loss of generality, during peak hours (9:00-12:00 and 17:00-20:00), normal hours (7:00-9:00, 12:00-17:00 and 20:00-22:00), and off-peak hours (22:00-24:00-0:00), the electricity prices are 0.566$ / kWh, 0.548$ / kWh, and 0.527$ / kWh, respectively. The reserve cost is 0.2 times the power generation cost. The power generation cost of the controllable distributed generator set is 0.66 US dollars / kWh. The example simulation is compiled in the MATLAB R2018a environment, using the Yalmip optimization tool for modeling and calling Gurobi for solution. The computer configuration for compiling the algorithm is Intel(R) Core(TM) i7-5500 2.40GHz, 8G memory.

[0146] Method 0, the proposed method;

[0147] Method 1: a multi-time-scale scheduling method that does not consider spatiotemporal correlation;

[0148] Method 2: Multi-time scale scheduling method with low reserve level;

[0149] Method 3: Multi-time scale scheduling method with high reserve level;

[0150] Method 4: Multi-time-scale scheduling method based on model predictive control.

[0151] Method 1 does not consider the spatiotemporal correlation of forecast errors in scenario generation. Methods 2 and 3 differ from Method 0 only in the allocation of reserves during the day-ahead and intraday phases. Method 2 allocates half the reserves of the proposed method, while Method 3 allocates 1.5 times the reserves of the proposed method. In Method 4, the scheduling process is still divided into three phases, but stochastic optimization is not used.

[0152] Stochastic model predictive control (SMPC) is characterized by optimizing scheduling for a variety of typical scenarios, ultimately deriving the output values ​​of distributed resources through a probability-weighted approach. Therefore, capturing these typical scenarios is a crucial step in this method and explains its advantages over other approaches.

[0153] To investigate the impact of copula theory on scenario generation, Figures 5(a) and 5(b) show scenarios generated using copula theory and without it, respectively. For simplicity, only wind power is involved in the scenarios. A separate analysis of PV output yields the following conclusions. Note that hours 1-24 represent the predicted output sequence for WT 1, while hours 25-48 represent the predicted output sequence for WT 2. After accounting for correlation, wind power output changes across time periods tend to be moderate, indicating temporal correlation, while the predicted outputs of the two WTs tend to follow the same changing trend, indicating spatial correlation.

[0154] To further study the ability of the two methods to simulate real scenarios, the fluctuation index, total error, power peak-to-valley difference, and output difference between the two wind turbines are defined as (24), (25), (26), and (27). The calculation results of the two methods and the real scenario are shown in Table 1:

[0155]

[0156]

[0157] γ=max t∈T (P wind,t )-min t∈T (P wind,t ) (27)

[0158]

[0159] Where, To predict wind power output, is the wind power output under actual conditions, P WT1,t For WT1's contribution.

[0160] As can be seen from Table 1, compared to Method 1, the method proposed in this embodiment is closer to the real-world scenario in terms of fluctuation index, power peak-to-valley difference, and output difference between the two wind turbines. However, the difference in total error between the two methods is not significant. This is because correlation analysis simply establishes a multivariate joint cumulative probability distribution model, which reasonably expresses the correlation between the prediction errors of multiple wind turbines at different time periods, which is consistent with the actual situation in the distribution network. However, considering correlation does not fundamentally improve the prediction technology and is therefore not very effective in reducing prediction error.

[0161] Table 1. Results of Method 0, Method 1, and Actual Scenario

[0162]

[0163] To investigate the effectiveness of this method in balancing fuel costs, various reserve costs, load loss risk, and wind curtailment risk, Table 2 presents the total cost, reserve cost, and power generation cost results for the proposed method and two other methods. Reserve cost refers to the cost of purchasing reserve power from the transmission grid. Power generation cost is the sum of the cost of controllable distributed generation, the cost of purchasing power from the transmission grid, and the operating costs of the energy storage device.

[0164] Table 2 shows that the proposed method significantly reduces total costs compared to Methods 2 and 3. While the proposed method does not minimize both standby and power generation costs compared to Methods 2 and 3, it does achieve the lowest total cost by balancing standby and power generation costs. This is because a large standby interval significantly reduces wind and solar curtailment, thereby reducing power generation costs. However, this increases standby costs and the total cost of distribution network operation. A small standby interval, while lowering standby costs, is more likely to result in wind and solar curtailment, increasing power generation costs and the total cost of distribution network operation.

[0165] Table 2. Cost of each method

[0166]

[0167] In order to study the impact of the reserve interval on limiting wind and solar curtailment, the wind power output and photovoltaic output of Method 0, Method 2, and Method 3, as well as the predicted wind power output and photovoltaic power generation are shown in Figure 2. Figure 6 、 Figure 7 At the same time, the wind (solar) curtailment index is defined as shown in formula (29), and the calculation results of the wind (solar) curtailment index of the three methods are shown in Table 3.

[0168]

[0169] Table 3. Statistical results of wind and solar curtailment

[0170]

[0171] It is not difficult to see that when the reserve interval is lower than the interval of the method in this paper, the phenomenon of wind and solar power abandonment is common, and the wind (solar power abandonment) index is high. Therefore, the electricity obtained from the transmission network and controllable distributed generation increases, resulting in an increase in power generation costs, which in turn increases the total cost; when the reserve interval is higher than the interval of the method in this paper, the phenomenon of wind and solar power abandonment is not very obvious, and the wind (solar power abandonment) index difference is not large. Therefore, the power generation cost difference is not large at this time, but the increase in the reserve interval leads to an increase in the reserve cost, which in turn increases the total cost. It can be seen from this that the method in this embodiment can reasonably set the reserve interval for the distribution network, balance the power generation cost and the reserve cost, and achieve the goal of minimizing the total operating cost.

[0172] In addition, the phenomenon of abandoning renewable energy mainly occurs in the early morning and midnight, while renewable energy can basically be fully consumed at other times. Figure 8 As shown in the figure, this phenomenon can be explained by the fact that electricity prices are at a low point in the early morning and midnight, when the controllable distributed generators basically do not generate active power. When the output of renewable energy increases, it can only be absorbed by charging renewable resources and reducing the power sent from the distribution network to the transmission network. However, the charging power of renewable resources is small, and the power fluctuation between the transmission and distribution networks must be minimized in real time, so renewable resources can only be abandoned; at other times, electricity prices are high, and controllable distributed generators will operate close to full load to reduce power generation costs, while renewable resources have no power generation costs. When the output of renewable energy increases, the controllable distributed generators can reduce their active output to absorb renewable resources.

[0173] To verify the effectiveness of this method in limiting power fluctuations at the transmission network interface, the power fluctuations of Method 0 and Method 4 are shown in Figure 9(a) and Figure 9(b). To further highlight the ability of the two methods to stabilize power fluctuations, renewable energy is connected to the distribution network at maximum output in the real-time stage, that is, power fluctuations can only be limited by adjusting the active power of controllable distributed generators and renewable resources. The active output of controllable distributed generators and renewable resources for the two methods is shown in Figure 9(a). Figure 12 It should be noted that in model predictive control, the randomness of the prediction scenario is not considered, so it is impossible to provide a backup interval for the active distribution network by optimizing the conditional risk value.

[0174] In order to further study the effectiveness of the proposed method in limiting power fluctuation, the power fluctuation is defined as (30), and the distribution of power fluctuation is as follows Figure 10 As shown. In (30), is the actual active power at the interface between the distribution network and the transmission network, The commitment value for the intraday period is:

[0175]

[0176] From Figure 9(a), Figure 9(b) and Figure 10 It can be seen that compared with Method 4 based on model predictive control, the proposed method has better performance in suppressing power fluctuations. This is because stochastic model predictive control optimizes scheduling for typical scenarios, fully considering the uncertainty of prediction results, so its scheduling results are adaptable to various typical scenarios. However, due to large prediction errors and the failure to consider the stochastic nature of renewable energy, model predictive control can lead to inappropriate power commitment values ​​in the daily phase and the lack of appropriate backup capacity, resulting in large power fluctuations in the real-time phase.

[0177] As shown in Figure 11(a), in Method 0, the active power of the controllable distributed generators and renewable resources varies flexibly to maintain minimal power fluctuations. Figure 11(b) shows that in Method 4, the controllable distributed generators are sometimes fully loaded, meaning they lose some of their regulation capabilities, resulting in significant power fluctuations. Therefore, because the randomness of renewable resources is taken into account, the method proposed in this embodiment is more suitable for coping with power fluctuations.

[0178] Example 2

[0179] The second embodiment of the present disclosure introduces an active distribution network optimization scheduling system based on stochastic model predictive control.

[0180] like Figure 12 An active distribution network optimization scheduling system based on stochastic model predictive control is shown, comprising:

[0181] An acquisition module, used to obtain a backup configuration interval of renewable energy;

[0182] The optimization module obtains the optimal scheduling plan of the active distribution network based on the obtained backup configuration interval and the preset multi-time scale active distribution network optimization scheduling model;

[0183] Among them, the multi-time-scale active distribution network optimization scheduling model includes a day-ahead scheduling stage, an intraday scheduling stage and a real-time scheduling stage, adopts a stochastic model predictive control, takes the minimum operating cost as the goal, and takes the reserve constraint as the constraint condition.

[0184] The detailed steps are the same as those of the active distribution network optimization scheduling method based on stochastic model predictive control provided in Example 1, and will not be repeated here.

[0185] Example 3

[0186] A third embodiment of the present disclosure provides a computer-readable storage medium.

[0187] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps of the active distribution network optimization scheduling method based on stochastic model predictive control as described in the first embodiment of the present disclosure.

[0188] The detailed steps are the same as those of the active distribution network optimization scheduling method based on stochastic model predictive control provided in Example 1, and will not be repeated here.

[0189] Example 4

[0190] A fourth embodiment of the present disclosure provides an electronic device.

[0191] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps of the active distribution network optimization and scheduling method based on stochastic model predictive control as described in the first embodiment of the present disclosure are implemented.

[0192] The detailed steps are the same as those of the active distribution network optimization scheduling method based on stochastic model predictive control provided in Example 1, and will not be repeated here.

[0193] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Those skilled in the art should understand that, based on the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. An active distribution network optimization scheduling method based on stochastic model predictive control, characterized in that: The following steps are involved: Obtaining the backup configuration range of renewable energy; According to the obtained backup configuration interval and the preset multi-time scale active distribution network optimization dispatching model, the optimal dispatching plan of the active distribution network is obtained; Considering the trade-off between reserve and operation risk cost, conditional risk value is introduced to determine the optimal reserve configuration; the conditional risk value based on probability distribution is used to quantify the risk of load loss and RES reduction, thereby minimizing the operation risk cost and reserve cost, finding the optimal error range of renewable energy, and determining the optimal reserve configuration range; The multi-time-scale active distribution network optimization dispatching model includes the day-ahead dispatching stage, the intraday dispatching stage and the real-time dispatching stage, adopts the random model predictive control, takes the minimum operation cost as the goal, and takes the reserve constraint as the constraint condition; in different dispatching stages, for different goals, dispatching is carried out according to the characteristics of distributed resources; Based on the copula theory, the correlation analysis of the forecast information of the active distribution network scenario is carried out to reduce the volatility and forecast error of renewable energy processing in the generation scenario; The stochastic model predictive control includes using stochastic optimization methods in all three stages, and in the real-time stage, timely correcting the deviation from the intraday stage commitment according to the forecast scenario to reduce the power fluctuations at the interface with the transmission system; The alternate constraints are as follows: Active reserve constraints: in, and They are the up / down backup provided by the transmission grid, controllable distributed generators and energy storage devices. and is the reserve demand determined by the optimal error interval of RES, s represents the scenario they are in, and reserves are arranged in each scenario. In the following constraints, s represents the scenario; Reactive power reserve constraints: in, Indicates the upper / lower backup provided by switched capacitor banks and static VAR compensators.

2. The method for optimizing the dispatching of active distribution networks based on stochastic model predictive control as claimed in claim 1, characterized in that: In the day-ahead scheduling stage, the multi-time-scale active distribution network optimization scheduling model aims to minimize the operating cost and backup cost of the active distribution network, while considering the backup capacity provided by controllable resources including controllable distributed generators, energy storage systems, switched capacitor banks, static VAR compensators, on-load tap-changing transformers and electric energy purchased from the transmission network.

3. The method for optimizing the dispatching of active distribution networks based on stochastic model predictive control as claimed in claim 2, characterized in that: In the intraday scheduling stage, the multi-time-scale active distribution network optimization scheduling model aims to minimize the operating cost and backup cost of the active distribution network, while considering the backup capacity provided by controllable resources including controllable distributed generators, energy storage systems, static VAR compensators, on-load tap-changing transformers and electricity purchased from the transmission network.

4. The method for optimizing the dispatching of active distribution networks based on stochastic model predictive control as claimed in claim 3, characterized in that: In the real-time scheduling stage, the multi-time-scale active distribution network optimization scheduling model updates the prediction information of the active distribution network scenario in each stage based on the prediction information of the active distribution network scenario, with the goal of minimizing power fluctuation and operating cost in the active distribution network scenario, combined with the measured values ​​of renewable energy, and performs closed-loop control on the scheduling of the active distribution network.

5. The method for optimizing the dispatching of active distribution networks based on stochastic model predictive control as claimed in claim 1, characterized in that: The constraints of the multi-time-scale active distribution network optimization scheduling model also include power flow constraints, branch capacity constraints, node voltage constraints, renewable energy generation operation constraints, controllable distributed generation operation constraints, on-load tap-changing transformer operation constraints, static VAR compensation device operation constraints, compensation capacitor group operation constraints and energy storage device operation constraints.

6. An active distribution network optimization dispatching system based on stochastic model predictive control, characterized in that: include: An acquisition module, used for acquiring a backup configuration interval of renewable energy; Considering the trade-off between backup and operational risk costs, conditional risk value is introduced to determine the optimal backup configuration; The load loss risk and RES curtailment risk were quantified based on the conditional value at risk of probability distribution, and the operation risk cost and reserve cost were minimized to obtain the optimal error interval of renewable energy and determine the optimal reserve configuration interval. The optimization module obtains the optimal dispatching scheme of the active distribution network according to the obtained backup configuration interval and the preset multi-time scale active distribution network optimization dispatching model; The multi-time-scale active distribution network optimization dispatching model includes the day-ahead dispatching stage, the intraday dispatching stage and the real-time dispatching stage, adopts the random model predictive control, takes the minimum operation cost as the goal, and takes the reserve constraint as the constraint condition; in different dispatching stages, for different goals, dispatching is carried out according to the characteristics of distributed resources; Based on the copula theory, the correlation analysis of the forecast information of the active distribution network scenario is carried out to reduce the volatility and forecast error of renewable energy processing in the generation scenario; The stochastic model predictive control includes using stochastic optimization methods in all three stages, and in the real-time stage, timely correcting the deviation from the intraday stage commitment according to the forecast scenario to reduce the power fluctuations at the interface with the transmission system; The alternate constraints are as follows: Active reserve constraints: in, and They are the up / down backup provided by the transmission grid, controllable distributed generators and energy storage devices. and is the reserve demand determined by the optimal error interval of RES, s represents the scenario they are in, and reserves are arranged in each scenario. In the following constraints, s represents the scenario; Reactive power reserve constraints: in, Indicates the upper / lower backup provided by switched capacitor banks and static VAR compensators.

7. An active distribution network optimization dispatching system based on stochastic model predictive control as described in claim 6, characterized in that: In the day-ahead scheduling stage, the multi-time-scale active distribution network optimization scheduling model aims to minimize the operating cost and backup cost of the active distribution network, while considering the backup capacity provided by controllable resources including controllable distributed generators, energy storage systems, switched capacitor banks, static VAR compensators, on-load tap-changing transformers and electric energy purchased from the transmission network.

8. An active distribution network optimization dispatching system based on stochastic model predictive control as described in claim 7, characterized in that: In the intraday scheduling stage, the multi-time-scale active distribution network optimization scheduling model aims to minimize the operating cost and backup cost of the active distribution network, while considering the backup capacity provided by controllable resources including controllable distributed generators, energy storage systems, static VAR compensators, on-load tap-changing transformers and electricity purchased from the transmission network.

9. An active distribution network optimization dispatching system based on stochastic model predictive control as described in claim 8, characterized in that: In the real-time scheduling stage, the multi-time-scale active distribution network optimization scheduling model updates the prediction information of the active distribution network scenario in each stage based on the prediction information of the active distribution network scenario, with the goal of minimizing power fluctuation and operating cost in the active distribution network scenario, combined with the measured values ​​of renewable energy, and performs closed-loop control on the scheduling of the active distribution network.

10. The active distribution network optimization dispatching system based on stochastic model predictive control as claimed in claim 6, characterized in that: The constraints of the multi-time-scale active distribution network optimization scheduling model also include power flow constraints, branch capacity constraints, node voltage constraints, renewable energy generation operation constraints, controllable distributed generation operation constraints, on-load tap-changing transformer operation constraints, static VAR compensation device operation constraints, compensation capacitor group operation constraints and energy storage device operation constraints.

11. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps in the active distribution network optimization scheduling method based on stochastic model predictive control as described in any one of claims 1 to 5 are implemented.

12. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the active distribution network optimization scheduling method based on stochastic model predictive control as described in any one of claims 1 to 5 are implemented.

Images