Area-oriented mobile energy storage scheduling method

Abstract

The invention discloses a transformer area-oriented mobile energy storage scheduling method, which comprises the following steps of: acquiring historical and real-time measurement data of each node based on a three-phase line topological structure and line parameters of a transformer area; respectively establishing a quantile time sequence prediction model for the load and the photovoltaic output of each node in the prediction time period, and obtaining a quantile prediction set based on the load and the photovoltaic output; generating scenes / samples according to the quantile prediction set, and load and photovoltaic samples of each node at each moment contained in each scene; establishing a three-phase unbalanced power flow model satisfying a branch balance and power flow relationship, and establishing a mobile energy storage model based on node selection and energy constraint; for a three-phase unbalanced power flow model and a mobile energy storage model, joint optimization of a rolling time domain is established, and the optimization takes expected operation cost minimization as a target; and the optimized three-phase unbalanced power flow model and the mobile energy storage model are used for scheduling, so that the problems of safe operation and efficient absorption of the high-permeability photovoltaic transformer area are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of transformer area power dispatching, in particular to a mobile energy storage dispatching method for transformer areas. BACKGROUND

[0002] With the deepening of the "double carbon" strategy, the penetration rate of distributed photovoltaic (DPV) in distribution networks continues to rise. While promoting energy clean-up, it also poses unprecedented challenges to the operation of distribution networks, especially low-voltage transformer areas. The traditional "passive" operation mode of distribution networks is difficult to cope with many problems caused by high proportion of photovoltaic access, such as reverse flow of power flow, voltage out-of-limit, and aggravation of three-phase imbalance, which seriously restricts the consumption capacity of photovoltaic and threatens the safe and stable operation of the power grid.

[0003] Currently, to solve the above problems, the industry generally uses reactive power compensation devices (such as SVG) and traditional voltage regulation equipment (such as on-load voltage regulation transformer) based on local measurement for treatment. However, these methods have obvious limitations: first, they are local and passive response control, lack a global optimization perspective, and are difficult to coordinate multiple resources within the system; second, their regulation capacity is limited and cannot solve the timing power imbalance problem caused by photovoltaic volatility. In recent years, although some studies have proposed using stationary energy storage (SES) for peak shaving and valley filling, the investment cost is high, the layout is not flexible, and it is difficult to adapt to the complex scenario of dynamic load changes, with poor economic benefits.

[0004] At the same time, mobile energy storage (MES) technology shows great application potential due to its flexible space-time transfer capability. It can be used as a "mobile power point" to be flexibly dispatched among multiple transformer areas, which can theoretically greatly improve resource utilization efficiency and grid flexibility. However, integrating MES into optimal dispatching of distribution networks is a very complex system engineering problem. The core difficulties are: first, a joint optimization model needs to be built that integrates MES mobile path constraints and three-phase unbalanced power flow constraints of distribution networks. This model is essentially a large-scale, non-convex, non-linear mixed integer programming (MINLP) problem, which has high computational complexity and is difficult to solve directly; second, the strong randomness and uncertainty of photovoltaic output limit the effectiveness of deterministic optimization methods, and random optimization or robust optimization methods must be introduced, further increasing the complexity of the model; third, there is a lack of consideration of long-term operation indicators such as transformer life and voltage imbalance, and economic considerations, making it difficult to achieve truly optimal dispatching.

[0005] The patent application publication No. CN118868098A discloses a power distribution network supply guarantee scheduling method based on distributed power-supply-load-mobile energy storage. Although the mobile energy storage is considered in the scheduling scheme, the emphasis is on highlighting the cross-network coupling (road network + power grid) and robust supply guarantee. Specifically, at the coupling level, the patent optimizes the road network as an independent network together with the power distribution network, highlighting the "road-power distribution network coordination". In the mobile energy storage scheduling logic, the patent needs to consider the vehicle moving path, traffic congestion and driving energy consumption. In terms of robustness and uncertainty processing, the patent introduces the road resistance function, the Gaussian mixed traffic flow combined with the Monte Carlo scene, and uses the C&CG robust algorithm. In terms of objective function setting, the patent emphasizes the dual objectives of peak load shifting and network loss reduction, and integrates multiple types of power distribution regulation devices.

[0006] Therefore, there is an urgent need in the art for an intelligent optimization scheduling method that can effectively integrate mobile energy storage, photovoltaic, flexible load and other distributed resources, and can perform efficient and reliable calculation under uncertain environment, to solve the safe operation and efficient consumption problem of high penetration rate photovoltaic area. SUMMARY

[0007] The technical problem to be solved by the present application is to provide an intelligent optimization scheduling method that can effectively integrate mobile energy storage, photovoltaic, flexible load and other distributed resources, and can perform efficient and reliable calculation under uncertain environment.

[0008] To solve the above technical problems, the present application provides the following technical solutions:

[0009] A mobile energy storage scheduling method for area, comprising:

[0010] Based on the three-phase line topology structure and line parameters of the area, historical and real-time measurement data of each node are collected;

[0011] A quantile time series prediction model is established for the load and photovoltaic output of each node in the prediction period, and a quantile prediction set based on the load and photovoltaic output is obtained;

[0012] Scenarios / samples are generated according to the quantile prediction set, and each scenario contains the load and photovoltaic sample of each node at each time;

[0013] A three-phase unbalanced power flow model that satisfies the branch balance and power flow relationship is established, and a mobile energy storage model based on node selection and energy constraints is established;

[0014] A joint optimization in rolling time domain is established for the three-phase unbalanced power flow model and the mobile energy storage model, and the optimization aims to minimize the expected operation cost;

[0015] The improved three-phase unbalanced power flow model and the mobile energy storage model are applied for scheduling.

[0016] Technical effects: In terms of coupling level, the application focuses on the space-time combination within the power grid. In terms of mobile energy storage scheduling logic, the application is based on node selection and energy constraints. In terms of robustness and uncertainty processing, the application focuses on the uncertainty of load and photovoltaic prediction. In terms of objective function setting, the application contains voltage unbalance degree and degradation cost. In summary, the application realizes fine distribution area scheduling and prediction-driven.

[0017] In the embodiment, for any branch and phase, the expression of the three-phase unbalanced power flow model is:

[0018]

[0019] In the formula, P mn,φ,t represents the active power of the line from node m to node n of phase φ at time period t, Q mn,φ,t represents the reactive power of the line from node m to node n of phase φ at time period t, P nk,φ,t represents the active power of the line from node n to its downstream node k of phase φ at time period t, Q nk,φ,t represents the reactive power of the line from node n to its downstream node k of phase φ at time period t, D n represents the set of downstream nodes starting from node n, represents the injected active / reactive power, U m,φ,t represents the voltage amplitude of node m of phase φ at time period t, U n,φ,t represents the voltage amplitude of node n of phase φ at time period t, R mn represents the resistance of the line m to n, X mn represents the reactance of the line m to n, represents the current square of the line from node m to node n of phase φ at time period t.

[0020] In the embodiment, a mobile energy storage model based on node selection and energy constraints is established, including: setting a mobile energy storage unit set For each mobile energy storage unit A binary parking variable x v,n,t ∈{0,1} is defined at time period t, if the value is 1, it means that the energy storage unit v is parked on node n at time period t, if the value is 0, it means not, and the energy storage state evolution constraint is defined as:

[0021]

[0022] and satisfies

[0023] In the formula, Ev,t+1 E represents the energy state of energy storage unit v at time t+1. v,t η represents the energy state of energy storage unit v during time period t. c Indicates charging efficiency. η represents the charging power of energy storage unit v during time period t, where Δt represents the length of the time period. d Indicates discharge efficiency. This represents the discharge power of energy storage unit v during time period t; and These represent the minimum and maximum energy states of the energy storage unit v, respectively. This represents the maximum charging and discharging power of the energy storage unit v.

[0024] In this embodiment, the binary docking variable in the mobile energy storage model satisfies vehicle path and time window constraints:

[0025]

[0026] And when x v,i,t =1 and x v,j,t' When = 1, t' ≥ t + τ must be satisfied. ij , where τ ij Let N be the travel time from node i to node j, and N be the set of nodes.

[0027] In this embodiment, the expression aimed at minimizing the desired operating cost is:

[0028]

[0029] The objective of the joint optimization satisfies scenario-based physical constraints, device constraints, and the following opportunity constraints:

[0030]

[0031] In the formula, This represents the expected value of ω over all possible scenarios. These represent the grid cost, loss cost, degradation cost, and penalty cost for time period t under scenario ω, respectively. This represents the probability under scenario ω. and These represent the minimum and maximum voltage limits for node n at phase φ and time period t, respectively, where α is the tolerance for exceeding the limit risk, and U... n,φ,t This represents the voltage amplitude at node n with phase φ during time period t.

[0032] In this embodiment, in the joint optimization, the battery degradation cost is represented by a pricing method based on lifetime consumption:

[0033]

[0034] wherein, is the total degradation cost of the energy storage unit v at time period t, is the unit energy degradation cost of the energy storage unit v, is the rated capacity, represents the charging power of the energy storage unit v at time period t, and Δt represents the time period length, represents the discharging power of the energy storage unit v at time period t.

[0035] In the embodiment, the target of the joint optimization also includes a light abandonment penalty term, when the photovoltaic output exceeds the absorption capacity of the distribution network in the scenario, the target is calculated according to the following expression:

[0036]

[0037] wherein, π curt is the light abandonment unit price, (·) + is a positive part function, (·) + = max{0, ·}, indicates that the output is itself when the input is positive, and the output is 0 when the input is negative, is the photovoltaic output predicted value, is the photovoltaic output actual value.

[0038] In the embodiment, the joint optimization is performed in a rolling mode at the time step Δt, if the risk assessment shows that there is a threshold value situation in the current window, the emergency control action is triggered, including inverter voltage-reactive droop control, temporary reduction of flexible load and rapid injection of nearby mobile energy storage;

[0039] Wherein, the following trigger rules are used to determine whether to enter the emergency risk avoidance process:

[0040] If there is a node n and a phase φ in any scenario ω such that Or The probability estimate value of the situation exceeds the threshold value β, the emergency risk avoidance is triggered;

[0041] The emergency risk avoidance includes performing in priority: inverter voltage-reactive droop control, rapid injection of nearby mobile energy storage, temporary reduction of flexible load.

[0042] In the embodiment, in the inverter voltage-reactive droop control, the inverter reactive power support adopts the local droop control or the reactive power setting based on optimization issued, and satisfies the following droop relationship:

[0043]

[0044] wherein, is the reactive power output, K q is the droop coefficient, is the reference voltage, Un,t is a node voltage measurement value;

[0045] The flexible load temporary reduction action includes: peak shifting of the load, which is targeted at controllable interruptible load or adjustable load, energy shifting in a scheduled adjustment range and response time length, and satisfaction of load energy conservation:

[0046]

[0047] In the formula, is a load power change amount, which represents an adjustment amount of the flexible load power on the node n at time t. is a maximum adjustable power, which represents an upper limit of the power that the flexible load of the node n is allowed to adjust at time t.

[0048] In the embodiment, the joint optimization problem adopts a two-stage or decomposition solving strategy: the first stage is an integer programming solution based on vehicle routes and parking time windows, and the second stage is a continuous optimization of power distribution and power flow under fixed routes.

[0049] Compared with the prior art, the beneficial effects of the present application are:

[0050] 1. Improving renewable energy consumption capacity

[0051] The present application significantly reduces the photovoltaic curtailment rate through fine multi-scenario prediction and mobile energy storage scheduling.

[0052] 2. Improving power quality and operation safety

[0053] The three-phase power flow imbalance constraint ensures that the voltages of the phases are close to balance, and reduces the risk of motor overheating and voltage out-of-limit.

[0054] 3. Significantly improving scheduling flexibility and economy

[0055] The mobile energy storage unit can be moved to the load peak or the voltage weak node as needed to realize "one car multi-point" sharing; and the high cost of building large fixed energy storage for a single area is avoided.

[0056] 4. Fast solving speed, suitable for real-time rolling optimization

[0057] Through SOCP / SDP convexity, the large-scale area optimization problem is converted into convex optimization that can converge within minutes, meeting the actual scheduling needs.

[0058] 5. Enhancing system robustness

[0059] Quantile prediction + Copula scenario maintains the spatial-temporal correlation of load and photovoltaic, so that the scheduling scheme is still stable and feasible under various meteorological and load disturbances. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 A flow chart of a mobile energy storage scheduling method for a transformer area according to an embodiment of the present application.

[0061] Figure 2 A network architecture diagram of a scheduling method according to an embodiment of the present application. DETAILED DESCRIPTION

[0062] For the convenience of those skilled in the art to understand the technical scheme of the present application, the technical scheme of the present application will be further described in conjunction with the drawings of the specification.

[0063] The terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0064] Please refer to Figure 1 and Figure 2 The present application provides a mobile energy storage scheduling method for a transformer area, comprising:

[0065] S10, based on the three-phase line topology structure and line parameters of the transformer area, collecting historical and real-time measurement data of each node.

[0066] In this embodiment, the historical and real-time measurement data of each node includes node load Node voltage U n,φ,t , photovoltaic output Ambient temperature T t , time period identifier and electricity price information.

[0067] S20, for each node, a quantile time series prediction model is established for the load and photovoltaic output in the prediction period, and a quantile prediction set based on the load and photovoltaic output is obtained.

[0068] In this embodiment, for each node n, a quantile time series prediction model is established for the load and photovoltaic output in the prediction period t, and a quantile prediction set Where τ∈(0,1] represents the quantile level, represents the load prediction value, represents the photovoltaic output prediction value.

[0069] Wherein, the quantile time series prediction model introduces a high temperature correction term in the model input, so that the load prediction satisfies:

[0070]

[0071] In the formula, Xn,t H is a historical load feature vector, t H is a holiday and time encoding, γ is a temperature threshold of node n, n f is a temperature sensitivity coefficient, θ (X n,t , T t , H t ) represents a load prediction model, and θ represents a model parameter.

[0072] In the embodiment, the high-temperature correction term is represented by a node-related segmented function, and the mathematical form is as follows:

[0073]

[0074] In the formula, H n (T t ) represents a high-temperature correction value of node n at time t, which is used to be superimposed on the original load prediction result, γ n,1 , γ n,2 represents a temperature sensitivity coefficient (first-order and second-order terms) of node n, which is obtained through historical regression or model training, γ represents a temperature threshold of node n, which is set according to the actual situation of the transformer area.

[0075] Further, in the embodiment, the temperature threshold γ The rolling time domain sliding window length H ∈ [4, 24] (the number of time periods, in Δt units), and the out-of-limit risk tolerance α ∈ [0.01, 0.2].

[0076] Specifically, the load is the main uncertain factor, so the quantile time series prediction model corresponding to the load prediction is given above, which is a quantile long short-term memory network (QuantileLSTM), and the available real-time monitoring or simple model corresponding to the photovoltaic output prediction.

[0077] In the embodiment, the quantile time series prediction model is trained by minimizing the following quantile loss function:

[0078]

[0079] In the formula, is the actual load, is the predicted τ quantile load, τ is the quantile level, and the training target is to minimize the sum of the losses of all nodes, all time points and all quantiles. A sliding window online fine-tuning strategy based on historical observations is used to support rolling update.

[0080] S30, according to the quantile prediction set, a scenario / sample is generated, and each node contains the load and photovoltaic sample at each time point.

[0081] In the embodiment, S future scenarios {ω s |s=1} S Each scenario contains the load and photovoltaic sample of each node at each time.

[0082] Further, the quantile prediction result is modeled by Copula to model the joint distribution between nodes and photovoltaics, and scenarios are generated based on the joint distribution to maintain spatial-temporal correlation.

[0083] S40, a three-phase unbalanced power flow model satisfying branch balance and power flow relationship is established, and a mobile energy storage model based on node selection and energy constraint is established.

[0084] In the embodiment, the three-phase unbalanced power flow modeling step is: establishing a three-phase unbalanced power flow model of the transformer area, satisfying the branch balance and power flow relationship, for any branch (m→n) and phase φ, the following constraints are established:

[0085]

[0086] In the formula, P mn,φ,t represents the active power of the line from node m to node n of phase φ at time period t, Q mn,φ,t represents the reactive power of the line from node m to node n of phase φ at time period t, P nk,φ,t represents the active power of the line from node n to its downstream node k of phase φ at time period t, Q mn,φ,t represents the reactive power of the line from node n to its downstream node k of phase φ at time period t, D n represents the set of downstream nodes starting from node n, represents the injected active / reactive power (including load, photovoltaic output, and energy storage charging and discharging), U m,φ,t represents the voltage amplitude of node m of phase φ at time period t, U n,φ,t represents the voltage amplitude of node n of phase φ at time period t, R mn represents the resistance of the line m to n, X mn represents the reactance of the line m to n, represents the current square of the line from node m to node n of phase φ at time period t.

[0087] Further, in the embodiment, the three-phase unbalanced power flow model respectively establishes power balance and voltage constraints for each phase, and additionally introduces a voltage imbalance degree constraint, which is represented as:

[0088]

[0089] where, represents the maximum voltage value in all nodes, all phase types. represents the minimum voltage value in all nodes, all phase types. represents the three-phase average voltage of the whole network, ε is the upper limit of the allowable unbalance degree, and N is the node set. Specifically, the voltage unbalance tolerance ε ∈ [0.01, 0.05].

[0090] Further, in the present embodiment, the second-order term of the alternating current flow in the three-phase unbalanced flow modeling is approximated by convexization means such as second-order cone relaxation (SOCP) or semi-definite relaxation (SDP), so as to improve the solving efficiency and ensure solvability in a typical district network.

[0091] In the present embodiment, the mobile energy storage model description step is: setting a mobile energy storage unit set For each mobile energy storage unit Defining a binary parking variable x v,n,t ∈ {0, 1} at time period t. If the value is 1, it means that the energy storage unit v is parked at the node n at time period t, and if the value is 0, it means that it is not parked, and the energy storage state evolution constraint is defined as:

[0092]

[0093] and satisfies

[0094] wherein E v,t+1 represents the energy state of the energy storage unit v at time period t+1, E v,t represents the energy state of the energy storage unit v at time period t, η c represents the charging efficiency, i.e. the energy conversion efficiency of the energy storage unit in the charging process, represents the charging power of the energy storage unit v at time period t, and Δt is the time period length, η d represents the discharging efficiency, i.e. the energy conversion efficiency of the energy storage unit in the discharging process, represents the discharging power of the energy storage unit v at time period t. and respectively represent the minimum and maximum energy states of the energy storage unit v. represents the maximum charging and discharging power of the energy storage unit v.

[0095] Further, in the present embodiment, the mobile energy storage parking variable x v,n,t satisfies the vehicle path and time window constraints, including but not limited to:

[0096]

[0097] and when x v,i,t = 1 and xv,j,t' = 1 needs to satisfy t' > t + τ ij where τ ij is the travel time from node i to node j.

[0098] Further, in the embodiment, the energy storage unit also considers the ramp rate constraint of charging / discharging and grid access delay, specifically:

[0099]

[0100] where are the upper limits of charging / discharging ramp rate, respectively.

[0101] S50, for the three-phase unbalanced power flow model and the mobile energy storage model, a rolling time domain joint optimization is established, and the optimization aims to minimize the expected operation cost.

[0102] In the embodiment, a rolling time domain joint optimization problem is established, and the decision variables include the charging / discharging power of fixed energy storage and mobile energy storage, the mobile energy storage parking decision, and the flexible load peak shifting amount. The optimization aims to minimize the expected operation cost:

[0103]

[0104] wherein the joint optimization aims to satisfy the physical constraints, equipment constraints based on scenario ω, and the following opportunity constraints:

[0105]

[0106] wherein represents the expected value for all possible scenarios ω, represents the grid cost, loss cost, degradation cost and penalty cost of period t under scenario ω, respectively, represents the probability under scenario ω, and represent the minimum and maximum voltage limits of node n in phase φ and period t, respectively, and α is the over-limit risk tolerance, and U n,φ,t represents the voltage amplitude of node n in phase φ at period t.

[0107] Further, in the embodiment, the battery degradation cost in the joint optimization adopts an approximate expression based on life consumption:

[0108]

[0109] wherein is the total degradation cost of period t, is the unit energy degradation cost of energy storage unit v, is the rated capacity, Pv(t) represents the charging power of the energy storage unit v at time period t, and Δt represents the length of the time period. Pv(t) represents the charging power of the energy storage unit v at time period t, and Δt represents the length of the time period.

[0110] Further, in the present embodiment, the joint optimization objective further includes a curtailment penalty term When the photovoltaic output exceeds the absorption capacity of the distribution network in the scenario, the objective is calculated according to the following expression:

[0111]

[0112] In the formula, π curt is the curtailment unit price, (·) + is a positive function, (·) + = max{0, ·} indicates that the output is itself when the input is positive, and the output is 0 when the input is negative, is the predicted value of the photovoltaic output, is the actual value of the photovoltaic output.

[0113] Further, in the present embodiment, the chance constraint can be converted into a deterministic equivalent form or approximated by sampling, and the probability of exceeding the limit is controlled at a confidence level α in a rolling window.

[0114] Further, in the present embodiment, the three-phase unbalanced power flow model and the chance constraint can be mathematically expressed as a mixed integer nonlinear programming (MINLP) or converted into a mixed integer second-order cone programming (MISOCP) after relaxation and approximation, so as to be solved by using an existing solver. Specifically, in order to guarantee the solvability of the calculation and the implementability of the engineering, the following second-order cone relaxation form is used for the original non-convex power flow constraint (three-phase unbalanced power flow model):

[0115] S mn,φ,t = P mn,φ,t +jQ mn,φ,t ;

[0116]

[0117] and the cone constraint is introduced as |S mn,φ,t | 2 ≤ w mn,φ,t · w mn,φ,t ≥ 0, so that part of the nonlinear terms are converted into second-order cone constraints for use by the MISOCP solver.

[0118] where S mn,φ,t is a complex power, representing the apparent power on the phase φ flowing from the node m to the node n at the time period t. P mn,φ,t is the active power, Q mn,φ,t is the reactive power, U m,φ,t is the voltage amplitude, I mn,φ,t is the current amplitude, and wmn,φ,t j is the imaginary unit.

[0119] Further, in the embodiment, the application also includes a rolling execution and emergency triggering step: joint optimization is executed in a rolling mode at time step Δt, if the risk assessment shows that there is an over-threshold situation within the current window, an emergency control action is triggered, including inverter voltage-reactive power droop control, temporary reduction of flexible load, and rapid injection of nearby mobile energy storage.

[0120] Wherein, the following triggering rules are used to determine whether to enter the emergency risk avoidance process:

[0121] If there is a node n and a phase φ in any scenario ω such that Or The probability estimate value of the situation exceeds the threshold value β, the emergency risk avoidance is triggered.

[0122] The emergency risk avoidance includes priority execution: inverter voltage-reactive power droop control, rapid injection of nearby mobile energy storage, and temporary reduction of flexible load.

[0123] Further, in the embodiment, in the inverter voltage-reactive power droop control, the inverter reactive power support uses the local droop control or the reactive power setting based on optimization, which satisfies the following droop relationship:

[0124]

[0125] In the formula, Q is the reactive power output, K q is the droop coefficient, U is the reference voltage, U n,t is the node voltage measurement value;

[0126] The temporary reduction of flexible load action includes: load peak shifting (flexible load response) to controllable interruptible load or adjustable load, energy shifting is performed within the agreed adjustment range and response time length to meet the load energy conservation:

[0127]

[0128] In the formula, is the load power change, which represents the adjustment amount of the flexible load power on node n at time t. is the maximum adjustable power, which represents the upper limit of the power that the flexible load on node n is allowed to adjust at time t.

[0129] Further, in the embodiment, the optimization also approximately evaluates the hotspot temperature rise and life consumption of the transformer in the area, and the transformer life consumption term is in the form of a penalty A target function is introduced, which approximately evaluates the satisfaction of:

[0130]

[0131] where ΔL t is the equivalent life loss, κ1,η1 are empirical coefficients, is the equivalent load current, I rated is the transformer rated current.

[0132] Further, in the present embodiment, the time resolution Δt of the rolling optimization is selected as any value among 15 minutes, 30 minutes or 60 minutes, and only the optimization of the future H time periods (sliding window length H) is performed in each rolling, and then the sliding window is moved forward to realize online scheduling.

[0133] Further, in the present embodiment, the membership threshold and the weight are supervised learning or expert parameter tuning with the historical out-of-limit and accident data of the actual transformer area to realize adaptive calibration of the weight.

[0134] S60, the optimized three-phase unbalanced power flow model and the mobile energy storage model are applied for scheduling.

[0135] In the present embodiment, the optimized three-phase unbalanced power flow model and the mobile energy storage model are deployed on the edge side to realize local inverter issued reactive power control and fast emergency response, and the cloud side is deployed for joint optimization and historical data backtracking analysis, and the cloud-edge-end collaborative work ensures the real-time and robustness of the system.

[0136] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application, and any reference signs in the claims should not be regarded as limiting the claims.

[0137] The above-described embodiments only represent the implementation of the present application, and the protection scope of the present application is not limited to the above-described embodiments. For those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application.

Claims

1. A mobile energy storage scheduling method for a transformer area, characterized in that, Comprise: Based on the three-phase line topology and line parameters of the transformer area, collect the historical and real-time measurement data of each node; For each node, establish a quantile time series prediction model for the load and photovoltaic output in the prediction period, and obtain a quantile prediction set based on the load and photovoltaic output; Generate scenarios / samples according to the quantile prediction set, and each scenario contains the load and photovoltaic sample of each node at each time; Establish a three-phase unbalanced power flow model that satisfies the branch balance and power flow relationship, and a mobile energy storage model based on node selection and energy constraints; For the three-phase unbalanced power flow model and the mobile energy storage model, establish a rolling time domain joint optimization with the goal of minimizing the expected operating cost; Apply the optimized three-phase unbalanced power flow model and mobile energy storage model for scheduling.

2. The mobile energy storage scheduling method for a substation area according to claim 1, characterized in that, For any branch and phase, the expression of the three-phase unbalanced power flow model is: where P mn,φ,t represents the real power of the line from node m to node n for phase φ at time period t, Q mn,φ,t represents the reactive power of the line from node m to node n for phase φ at time period t, P nk,φ,t represents the real power of the line from node n to its downstream node k for phase φ at time period t, Q nk,φ,t represents the reactive power of the line from node n to its downstream node k for phase φ at time period t, D n represents the set of downstream nodes from node n, represents the injected real / reactive power, U m,φ,t represents the voltage magnitude of node m for phase φ at time period t, U n,φ,t represents the voltage magnitude of node n for phase φ at time period t, R mn represents the resistance of line m to n, X mn represents the reactance of line m to n, represents the current square of the line from node m to node n for phase φ at time period t. 3.The mobile energy storage scheduling method for a substation area according to claim 1, characterized in that, A mobile energy storage model based on node selection and energy constraints is established, including: setting a mobile energy storage unit set For each mobile energy storage unit A binary stop variable x is defined at time period t v,n,t ∈{0,1}, if the value is 1, it means that at time period t, the energy storage unit v stops at node n, if the value is 0, it means not, define the energy storage state evolution constraint as: and satisfies where E v,t+1 denotes the energy state of the energy storage unit v at time period t+1, E v,t denotes the energy state of the energy storage unit v at time period t, η c denotes the charging efficiency, denotes the charging power of the energy storage unit v at time period t, Δt denotes the time period length, η d denotes the discharging efficiency, denotes the discharging power of the energy storage unit v at time period t; and denote the minimum and maximum energy state of the energy storage unit v, respectively, denotes the maximum charge and discharge power of the energy storage unit v.

4. The mobile energy storage scheduling method for a substation area according to claim 3, characterized in that, For the binary stop variable in the mobile energy storage model, it satisfies the vehicle path and time window constraints: and when x v,i,t = 1 and x v,j,t' = 1, it needs to satisfy t ' ≥ t + τ ij , where τ ij is the travel time from node i to node j, and N is the set of nodes.

5. The mobile energy storage scheduling method for a substation area according to claim 1, characterized in that, The expression for minimizing the expected operating cost is: The objective of the joint optimization satisfies the scenario-based physical constraints, equipment constraints, and the following opportunity constraints: wherein denotes the expectation over all possible scenarios ω, denotes the grid cost, loss cost, degradation cost and penalty cost at time period t under scenario ω, respectively, denotes the probability under scenario ω, and denote the minimum and maximum voltage limits at node n for phase φ and time period t, respectively, and α is the overreach risk tolerance, U n,φ,t denotes the voltage magnitude at node n for phase φ and time period t.

6. The mobile energy storage scheduling method for a substation area according to claim 5, characterized in that, In the joint optimization, the battery degradation cost is represented by the life consumption pricing method: wherein is the total degradation cost for time period t, is the unit energy degradation cost for energy storage unit v, is the rated capacity, denotes the charging power of energy storage unit v at time period t, Δt denotes the time period length, denotes the discharging power of energy storage unit v at time period t.

7. The mobile energy storage scheduling method for a substation area according to claim 5, characterized in that, The objective of the joint optimization also includes the light abandonment penalty item, which is calculated according to the following expression when the photovoltaic output in the scenario exceeds the absorption capacity of the distribution network: wherein π curt is the light abandonment univariate, (·) + is the positive part function, (·) + = max{0, ·} is the positive part function, (·) is the photovoltaic power output prediction value, is the photovoltaic power output actual value. 8.The mobile energy storage scheduling method for a substation area according to claim 5, characterized in that, The joint optimization is executed in a rolling mode at time step Δt, and if the risk assessment shows that there is an over-threshold situation within the current window, an emergency control action is triggered, including inverter voltage-reactive power droop control, temporary reduction of flexible load, and rapid injection of nearby mobile energy storage; The following trigger rules are used to determine whether to enter the emergency risk avoidance process: If the probability estimate of the case that there exists a node n and a phase φ in any scenario ω such that or exceeds a threshold β, an emergency escape is triggered. The emergency risk avoidance includes the following priority execution: inverter voltage-reactive power droop control, nearby mobile energy storage rapid injection, and temporary reduction of flexible load.

9. The mobile energy storage scheduling method for a substation area according to claim 8, characterized in that, In the inverter voltage-reactive power droop control, the inverter reactive power support uses the local droop control or the reactive power setting based on optimization, which satisfies the following droop relationship: wherein is the reactive power output, K q is the droop coefficient, is the reference voltage, U n,t is the node voltage measurement; The temporary reduction of flexible load action includes: load peak shifting, which targets controllable interruptible load or adjustable load, and moves energy within the agreed adjustment range and response time, satisfying the load energy conservation: In the formula, is the load power variation, indicating the adjustment amount of the flexible load power on the node n at time t. is the maximum adjustable power, indicating the upper limit of the power that the flexible load of the node n is allowed to adjust at time t.

10. The mobile energy storage scheduling method for a substation area according to claim 1, characterized in that, The joint optimization problem uses a two-stage or decomposition solving strategy: the first stage is an integer programming solution based on vehicle route and stop time window, and the second stage is a continuous optimization of power distribution and power flow under fixed route.

Images