Layered collaborative power distribution network light storage and charging power optimization scheduling method and system

Abstract

The invention discloses a hierarchical collaborative power distribution network optical storage and charging power optimization scheduling method, which comprises the following steps: constructing a transformer area layer-feeder layer-regional layer three-level framework according to the voltage level of a power distribution network, and defining the targets of transformer area layer operation cost minimization, feeder layer network loss minimization and regional layer power distribution optimization; each level reports a power adjustable range level by level; establishing an optimization model of each level; the area layer and the feeder layer generate and issue power control instructions step by step, and the area layer coordinates internal resource tracking execution; and after execution, each level updates the adjustable range and feeds back to the upper level to form closed-loop cooperative control. The invention further discloses a layered collaborative power distribution network light storage and charging power optimization scheduling system. According to the method, the problems of cross-level target conflict and low global efficiency caused by single-level optimization in the existing power distribution network optical storage and charging resource scheduling, and low scheduling accuracy and weak dynamic adaptability caused by no closed-loop coordination mechanism are solved.

Description

Technical Field

[0001] This invention belongs to the field of distribution network dispatching and control technology, specifically involving a hierarchical collaborative distribution network photovoltaic-storage-charging power optimization dispatching method, and also involves a hierarchical collaborative distribution network photovoltaic-storage-charging power optimization dispatching system. Background Technology

[0002] With the large-scale integration and widespread adoption of distributed photovoltaic (PV), energy storage systems, and electric vehicles, the operation of distribution networks has gradually transformed from traditional unidirectional power supply networks to deep interaction networks involving power generation, load, and energy storage. However, existing distribution network dispatching schemes generally focus on single-level optimization and lack hierarchical closed-loop coordination, resulting in low overall dispatching efficiency and poor dispatching accuracy, specifically manifested in the following two aspects:

[0003] First, existing dispatching schemes often focus on single-level optimization, neglecting the inherent coupling and conflicts between optimization objectives at different levels. For example, when the distribution area level maximizes the absorption of local photovoltaic power to reduce costs, it may inject unbalanced power into the feeder level, leading to deteriorated feeder power flow and increased network losses. Conversely, dispatching instructions issued by the feeder or regional levels to control network losses and balance power may force the distribution area level to abandon its economically optimal operating strategy, increasing its electricity purchase costs. This dispatching lacks a "distributed aggregation-centralized optimization" mechanism between levels, resulting in wasted distributed resource output and a disconnect between dispatching instructions and the actual resource adjustment capabilities, failing to achieve global optimization of the distribution network. Second, the lack of a closed-loop coordination mechanism between levels leads to low dispatching accuracy and weak dynamic adaptability. Specifically, existing hierarchical dispatching has not established a closed-loop system of "lower-level constraints - upper-level optimization - lower-level execution - feedback adjustment." The lower level only passively receives instructions, without using its own adjustable range as a constraint condition for upper-level optimization. The upper level also does not dynamically update instructions based on feedback from lower-level execution, resulting in poor feasibility of dispatching instructions and weak dynamic adaptability.

[0004] The combination of these two types of problems not only leads to increased electricity purchase costs for users and low efficiency in the utilization of distributed resources, but also triggers a series of safety operation risks such as increased power fluctuations in the distribution network, feeder voltage exceeding limits, main transformer power imbalance, and equipment overload, which seriously restricts the safe, stable, economical, and efficient operation of the distribution network. Summary of the Invention

[0005] The purpose of this invention is to provide a hierarchical collaborative method for optimizing the scheduling of photovoltaic, energy storage, and charging power in distribution networks. This method solves the problems of cross-level target conflicts and low global efficiency caused by single-level optimization in the existing scheduling of photovoltaic, energy storage, and charging resources in distribution networks, as well as the problems of low scheduling accuracy and weak dynamic adaptability caused by the lack of a closed-loop collaborative mechanism.

[0006] Another objective of this invention is to provide a hierarchical and collaborative power optimization scheduling system for photovoltaic, energy storage, and charging distribution networks.

[0007] The technical solution adopted in this invention is a hierarchical and collaborative power optimization scheduling method for photovoltaic, energy storage, and charging distribution networks, which specifically includes the following steps: Step 1: Construct a three-level architecture of distribution area layer - feeder layer - regional layer according to the voltage level of the distribution network, and clarify the objectives of minimizing operating costs at the distribution area layer, minimizing network losses at the feeder layer, and optimizing power allocation at the regional layer; Step 2: Each level reports the adjustable power range level by level; Step 3: Build optimization models for each level, and linearize the regional optimization model using the Big-M method; Step 4: The regional layer and feeder layer generate and issue power control commands level by level, and the substation layer coordinates internal resources to track and execute them; Step 5: After execution, each level updates the adjustable range and feeds back to the next level, forming a closed-loop collaborative control.

[0008] The invention is further characterized by: The specific process of each level reporting the adjustable power range in step 2 is as follows: The distribution area layer aggregates the operating status of distributed photovoltaic, energy storage systems and electric vehicles, forms the adjustable range of distribution area power and reports it to the feeder layer; The feeder layer combines the adjustable range constraints of all its subordinate transformer areas to calculate and aggregate the adjustable power range of the feeder layer, and then reports it to the regional layer. The power adjustable range is the maximum and minimum limit of power adjustment for each level.

[0009] The objective function of the region-level optimization model in step 3 is:

[0010] In the formula, N represents the total number of 10kV feeders in the area; These represent the active power and reactive power of the nth feeder, respectively. The active and reactive power reference values ​​for the nth feeder are represented as follows: ,

[0011] The constraints of the regional layer optimization model include regional power balance constraints and feeder power allocation constraints, which are expressed as follows: Regional power balance constraints:

[0012]

[0013] In the formula, , These represent the active and reactive power of the regional main transformer network provided by the upper-level power grid dispatch, which is the sum of the active and reactive power of all 10KV feeders under this 110KV main transformer; Feeder power distribution constraints:

[0014]

[0015] In the formula, Let represent the minimum and maximum active power of the nth feeder, respectively. These represent the minimum and maximum reactive power of the nth feeder, respectively.

[0016] The objective function of the feeder layer optimization model in step 3 is:

[0017] In the formula, K represents the total number of stations on the nth feeder; This represents the resistance of the nth feeder; This represents the square of the current in the k-th transformer area of ​​the n-th feeder; The constraints of the feeder layer optimization model include regional power balance constraints, dynamic power flow constraints, node voltage constraints, and power adjustable range constraints, which are expressed as follows: Regional power balance constraints:

[0018]

[0019] In the formula, , These represent the total active power and reactive power of the network below the nth feeder layer, respectively, calculated by the regional layer optimization. , These represent the active power and reactive power of the k-th transformer area, respectively. Dynamic power flow constraints:

[0020] In the formula, Indicates the nth feeder line Branch node current conjugate, Indicates from node Downstream nodes of the outflow channel Total power, Indicates from node Injection Node Total power, From node Transmit to node Branch power loss; , These represent the nodes of the nth feeder. and nodes The voltage amplitude; Indicates the connection to the nth feeder node. and the branch impedance of the node, Indicates a branch Apparent power on Represents a node The injection apparent power at the location, This represents the set of all branches in a power distribution network. These represent the sets of all nodes in the network.

[0021] Because the dynamic power flow constraints contain non-convex terms, the second-order cone relaxation principle is used to relax the non-convex constraints. The relaxed branch dynamic power flow constraints are expressed as follows:

[0022] In the formula, Represents a node The amount of active power injected into the nth feeder, Indicates from node The sum of active power flowing to all its child nodes. Indicates from node Flow to Node The active power on the branch line. Represents a node The reactive power injection amount of the nth feeder Indicates from node The sum of reactive power flowing to all its child nodes. Indicates from node Flow to Node Reactive power on the branch line, , Representing nodes respectively and nodes The square of the voltage amplitude, This indicates the reactance value of the branch circuit. Indicates a branch The resistance value; Node voltage constraints:

[0023] In the formula, This represents the voltage on the nth feeder. These represent the lower and upper limits of the permissible voltage deviation, respectively. Power adjustable range constraints:

[0024]

[0025] In the formula, These represent the active power and reactive power output of the kth transformer area of ​​the nth feeder, respectively. Let represent the minimum and maximum values ​​of the output active power of the k-th transformer area on the n-th feeder, respectively. These represent the minimum and maximum values ​​of the output reactive power of the k-th transformer area on the n-th feeder, respectively.

[0026] The objective function of the platform layer optimization model in step 3 is:

[0027] In the formula, Let T represent the operating cost of the k-th transformer area, and T represent the optimization period. These represent the charging cost and discharging cost of the energy storage system, respectively, in yuan / kWh; This represents the charging and discharging power of the energy storage system in the k-th transformer area at time t; The unit purchase cost of electricity from the superior power grid is RMB / kWh; This represents the power purchased by the k-th distribution area from the upper-level power grid at time t, i.e., when the output of distributed photovoltaic power is insufficient, it needs to purchase power from the upper-level power grid; The constraints of the substation-level optimization model include power balance constraints, distributed photovoltaic output constraints, and energy storage system-related constraints, which are expressed as follows: Power balance constraints:

[0028]

[0029] In the formula, This represents the total active load of the k-th transformer area. This represents the total reactive load of the k-th transformer area. , These represent the predicted active and reactive power outputs of the distributed photovoltaic system in the kth transformer area, respectively. Distributed photovoltaic power output constraints:

[0030]

[0031] In the formula, This represents the maximum output of distributed photovoltaic power in the k-th transformer area. These represent the minimum and maximum values ​​of reactive power output from distributed photovoltaic systems, respectively. The constraints related to energy storage systems, including upper and lower limits of charge and discharge power constraints, mutual exclusion constraints of charge and discharge, SOC constraints of energy storage systems, and reactive power constraints of energy storage systems, are expressed as follows: Upper and lower limits of charge and discharge power constraints:

[0032]

[0033] In the formula, , These represent the maximum active power during charging and discharging, respectively. , These are binary variables representing the charging and discharging states of the energy storage system in the k-th transformer area, respectively. Charge-discharge mutual exclusion constraint:

[0034]

[0035] Energy storage system SOC constraints:

[0036]

[0037] In the formula, Indicates the rated capacity of the energy storage system. Indicates time interval, These represent the charging and discharging efficiencies of the energy storage system, respectively. , , These represent the state of charge of the energy storage system at time t, and the minimum and maximum allowable states of charge of the system, respectively. Reactive power constraints of energy storage systems:

[0038] In the formula, , , These represent the reactive power, upper limit, and lower limit currently generated by the energy storage system in the k-th distribution area, respectively.

[0039] The regional optimization model is a nonlinear model. By introducing 0-1 integer variables and auxiliary continuous variables, and combining the Big-M method, the regional optimization model is linearized. The specific process is as follows: First, introduce auxiliary variables. ,make And introduce 0-1 variables Add linear constraints:

[0040] In the formula, It is a sufficiently large integer selected based on the range of active power variation; Secondly, auxiliary variables are introduced. ,make And introduce 0-1 variables Add linear constraints:

[0041] In the formula, It is a sufficiently large integer selected based on the range of reactive power variation; Subsequently, the absolute value term in the original objective function is replaced to form the objective function of the linearized region layer optimization model, as follows: .

[0042] The specific process of step 4 is as follows: The regional layer solves for the optimal power distribution of each 10kV feeder based on its own optimization model, and sends it as a control command to the corresponding feeder layer. After receiving the instructions from the area layer, the feeder layer calculates the power target value of each area based on its own optimization model and sends it to the corresponding area layer as a control instruction. After receiving instructions from the feeder layer, the distribution area layer coordinates and controls the output of internal distributed photovoltaic power, the charging and discharging power of the energy storage system, and the charging power of electric vehicles, so that the total power of the distribution area tracks the requirements of the instructions.

[0043] The specific process of closed-loop coordinated control in step 5 is as follows: After the distribution area layer executes the instruction, it updates the adjustable range of the distribution area power and reports it to the feeder layer based on the access status of the energy storage system and electric vehicles and the changes in the actual output of distributed photovoltaics. The feeder layer aggregates the updated adjustable ranges of all subordinate transformer areas, generates a new adjustable power range for the feeder layer, and reports it to the regional layer. The region layer uses the updated adjustable range of the feeder layer as a new constraint, re-solves the optimization model, generates updated control commands, and enters the next round of scheduling process.

[0044] Another technical solution adopted in this invention is a hierarchical collaborative power optimization scheduling system for photovoltaic, energy storage and charging distribution networks, which includes: a transformer substation control unit, a feeder substation control unit and a regional control unit.

[0045] The distribution area control unit is used to aggregate the operating status of distributed photovoltaic, energy storage systems and electric vehicles to generate and report the adjustable range of the distribution area, build the distribution area optimization model, execute the control commands issued by the feeder layer, update the adjustable range and provide feedback. The feeder layer control unit is used to aggregate the adjustable range of subordinate transformer areas to generate and report the adjustable range of the feeder layer, build the feeder layer optimization model, receive instructions from the regional layer and send them to the transformer area layer, update the adjustable range and provide feedback. The regional layer control unit is used to receive the adjustable range of the feeder layer, build the regional layer optimization model and linearize it using the big-M method, generate and issue feeder power allocation instructions, and receive feedback information from the feeder layer.

[0046] The beneficial effects of this invention are: This invention presents a hierarchical collaborative power optimization scheduling method for distribution networks based on photovoltaic, energy storage, and charging (PV) systems. Through a three-layer architecture, the distribution area layer controls costs to avoid excessive local costs, the feeder layer reduces network losses and minimizes energy consumption in intermediate stages, and the regional layer optimizes power allocation to ensure global power balance. Simultaneously, the adjustable range of the lower layer serves as a constraint on the upper layer, ensuring that upper-layer commands do not exceed the lower layer's adjustment capabilities and preventing command failures. A closed-loop feedback mechanism guarantees that when distributed photovoltaic (PV) output suddenly drops, the distribution area layer quickly updates its adjustable range, feeding back to the feeder layer, and then adjusting the regional layer's power allocation. This ensures that the scheduling strategy adapts to resource fluctuations in real time, fully exploring and utilizing the adjustment potential of flexible resources such as distributed PV, energy storage systems (ESS), and electric vehicles (EVs), reducing curtailment, smoothing fluctuations, and ultimately achieving the global optimal goal of "cost-network loss-power balance." Attached Figure Description

[0047] Figure 1 This is a flowchart of the hierarchical collaborative power optimization scheduling method for photovoltaic, energy storage and charging power in distribution networks according to the present invention; Figure 2 This is a diagram of the hierarchical power control framework in the hierarchical collaborative power optimization scheduling method for photovoltaic, energy storage and charging power in the distribution network of this invention; Figure 3 This is a topology diagram of the distribution network node structure in Embodiment 5 of the present invention; Figure 4 This is the voltage distribution diagram of feeder 1 in Embodiment 5 of the present invention; Figure 5 This is a graph showing the change in ESS SOC of the platform area 1 in Embodiment 5 of the present invention. Detailed Implementation

[0048] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0049] Example 1 This embodiment provides a hierarchical and collaborative method for optimizing the scheduling of photovoltaic, energy storage, and charging power in a distribution network, which specifically includes the following steps: Step 1: Construct a three-level architecture of distribution area layer - feeder layer - regional layer according to the voltage level of the distribution network, and clarify the objectives of minimizing operating costs at the distribution area layer, minimizing network losses at the feeder layer, and optimizing power allocation at the regional layer; This invention relates to a distribution network comprising distributed photovoltaic (PV), electric vehicles (EVs), and energy storage systems (ESS). A three-tier architecture is constructed based on the distribution network voltage level, corresponding to: a 220V / 380V distribution area layer, a 10kV feeder layer, and a 110kV regional layer. To meet user-side cost requirements, the distribution area layer often needs to reduce electricity purchase costs through energy storage system charging and discharging and distributed PV integration; secondly, the feeder layer needs to reduce network losses by controlling power flow to prevent node voltage exceedances and branch overloads; and the regional layer needs to allocate optimal power to each feeder based on the power output of the main transformer in the upper-level grid to ensure global power balance. Therefore, based on the three-tier architecture, each level is assigned a clear and complementary optimization objective: minimizing operating costs at the distribution area level, minimizing network losses at the feeder level, and optimizing power allocation at the regional level, forming a three-tiered coordinated dispatch system of "distribution area-feeder-region".

[0050] Step 2: Each level reports the adjustable power range level by level. The distribution area level aggregates the operating status of distributed photovoltaic (PV), energy storage system (ESS), and electric vehicle (EV) to form the adjustable power range of the distribution area and reports it to the feeder level; the feeder level combines the adjustable range constraints of its subordinate distribution areas to form the adjustable power range of the feeder level and reports it to the regional level.

[0051] Step 3: Build optimization models for each level. The regional optimization model is linearized using the Big-M method.

[0052] The regional layer optimization model uses the reference values ​​of active and reactive power of the feeder as the objective function, and includes regional power balance constraints and feeder power allocation constraints; the big-M method is used to transform the nonlinear objective function into a linear equivalent model. Feeder layer optimization model: The objective function is to minimize feeder network loss. It includes regional power balance constraints, dynamic power flow constraints, node voltage constraints and power adjustable range constraints. The second-order cone relaxation principle is used to handle the non-convex terms of the model. The transformer substation optimization model aims to minimize the operating cost of the transformer substation. The operating cost includes the charging cost and discharging cost of the energy storage system, as well as the electricity purchase cost from the upstream grid. It also includes power balance constraints, distributed photovoltaic output constraints, and energy storage system-related constraints.

[0053] Step 4: The regional layer and feeder layer generate and issue power control commands level by level, and the substation layer coordinates internal resources to track and execute them.

[0054] Step 5: After execution, each level updates the adjustable range and feeds back to the next level, forming a closed-loop collaborative control.

[0055] Through the inter-level power coordination mechanism constraints and closed-loop feedback in steps 4 and 5, such as... Figure 1As shown, the "adjustable power range" output by the lower-level unit serves as a constraint condition for the upper-level optimization model, and the result of the upper-level optimization serves as the lower-level control command. At the same time, the control command is transmitted from top to bottom in the order of "regional layer → feeder layer → transformer area layer". After the transformer area layer executes the command, the adjustable range is updated from bottom to top in the order of "transformer area layer → feeder layer → regional layer". Through the closed-loop process of "lower-level input controllable range - upper-level optimization model solution - command execution - feedback adjustment", dynamic coordination is achieved, adapting to the dispersion of distributed resources and the centralization of scheduling, and improving resource utilization.

[0056] Example 2 In the hierarchical collaborative optimization architecture of this invention, the adjustable power range is the maximum and minimum limit of power adjustment at each level, determined by the aggregation capability of the lower-level units and output as the constraint condition of the upper-level optimization model. The adjustable power ranges at each level are as follows: Distribution area level: Determined by the operating status of distributed photovoltaic, energy storage systems, and electric vehicles within the distribution area, the active power on the user side. reactive power Depending on the output of distributed photovoltaic power, the charging and discharging capacity of energy storage, and the charging demand of electric vehicles, these factors are aggregated to form the power of each transformer substation, which serves as the adjustable range for the substation level. , The output is sent to the corresponding upper-layer feeder, where .

[0057] Feeder layer: First, combining the adjustable range constraints formed by the aggregation of the substation layer, calculate the maximum power value on a feeder. Then, aggregate all the maximum power values ​​of all feeders to form the adjustable range of the feeder layer. , Passed to the region layer, where .

[0058] (3) Regional layer: Based on the power balance constraint and the adjustable range constraint of the feeder layer, the optimal power allocated to each feeder is solved by the regional layer optimization model. The top-down control command is output to provide "target guidance" for the lower layer adjustment. The lower layer unit feedback execution update is then formed, and finally a closed-loop collaborative control mechanism is formed to achieve global optimization.

[0059] Example 3 In step 3, the optimization models for each level are built, as follows: (1) The objective function of the regional layer optimization model is:

[0060] In the formula, N represents the total number of 10kV feeders in the area; These represent the active power and reactive power of the nth feeder, respectively. The active and reactive power reference values ​​for the nth feeder are represented as follows: ,

[0061] The constraints of the regional layer optimization model include regional power balance constraints and feeder power allocation constraints, which are expressed as follows: Regional power balance constraints:

[0062]

[0063] In the formula, , These represent the active and reactive power of the regional main transformer network provided by the upper-level power grid dispatch, which is the sum of the active and reactive power of all 10KV feeders under this 110KV main transformer; Feeder power distribution constraints:

[0064]

[0065] In the formula, Let represent the minimum and maximum active power of the nth feeder, respectively. These represent the minimum and maximum reactive power of the nth feeder, respectively.

[0066] Because the objective function of the above regional layer optimization model Since it is nonlinear, it needs to be linearized for easier subsequent solutions. This invention introduces 0-1 integer variables and auxiliary continuous variables, and combines them with the Big-M method for linear equivalent replacement. The specific process is as follows: First, introduce auxiliary variables. ,make And introduce 0-1 variables Add linear constraints:

[0067] In the formula, This is a sufficiently large integer selected based on the range of active power variation. It can be taken as... .

[0068] Secondly, auxiliary variables are introduced. ,make And introduce 0-1 variables Add linear constraints:

[0069] In the formula, This is a sufficiently large integer selected based on the range of reactive power variation. It can be taken as... .

[0070] Subsequently, the absolute value term in the original objective function is replaced to form the objective function of the linearized region layer optimization model, as follows: .

[0071] (2) The objective function of the feeder layer optimization model is:

[0072] In the formula, K represents the total number of stations on the nth feeder; This represents the resistance of the nth feeder; This represents the square of the current in the k-th transformer area of ​​the n-th feeder; The constraints of the feeder layer optimization model include regional power balance constraints, dynamic power flow constraints, node voltage constraints, and power adjustable range constraints, which are expressed as follows: Regional power balance constraints:

[0073]

[0074] In the formula, , These represent the total active power and reactive power of the network below the nth feeder layer, respectively, calculated by the regional layer optimization. , These represent the active power and reactive power of the k-th transformer area, respectively. Dynamic power flow constraints:

[0075] In the formula, Indicates the nth feeder line Branch node current conjugate, Indicates from node Downstream nodes of the outflow channel Total power, Indicates from node Injection Node Total power, From node Transmit to node Branch power loss; , These represent the nodes of the nth feeder. and nodes The voltage amplitude; Indicates the connection to the nth feeder node. and the branch impedance of the node, Indicates a branch Apparent power on Represents a node The injection apparent power at the location, This represents the set of all branches in a power distribution network. These represent the sets of all nodes in the network.

[0076] Because the dynamic power flow constraints contain non-convex terms, the second-order cone relaxation principle is used to relax the non-convex constraints. The relaxed branch dynamic power flow constraints are expressed as follows:

[0077] In the formula, Represents a node The amount of active power injected into the nth feeder, Indicates from node The sum of active power flowing to all its child nodes. Indicates from node Flow to Node The active power on the branch line. Represents a node The reactive power injection amount of the nth feeder Indicates from node The sum of reactive power flowing to all its child nodes. Indicates from node Flow to Node Reactive power on the branch line, , Representing nodes respectively and nodes The square of the voltage amplitude, This indicates the reactance value of the branch circuit. Indicates a branch The resistance value; Node voltage constraints:

[0078] In the formula, This represents the voltage on the nth feeder. These represent the lower and upper limits of the permissible voltage deviation, respectively. Power adjustable range constraints:

[0079]

[0080] In the formula, These represent the active power and reactive power output of the kth transformer area of ​​the nth feeder, respectively. Let represent the minimum and maximum values ​​of the output active power of the k-th transformer area on the n-th feeder, respectively. These represent the minimum and maximum values ​​of the output reactive power of the k-th transformer area on the n-th feeder, respectively.

[0081] (3) The objective function of the substation layer optimization model is:

[0082] In the formula, Let T represent the operating cost of the k-th transformer area, and T represent the optimization period. These represent the charging cost and discharging cost of the energy storage system, respectively, in yuan / kWh; This represents the charging and discharging power of the energy storage system in the k-th transformer area at time t; The unit purchase cost of electricity from the superior power grid is RMB / kWh; This represents the power purchased by the k-th distribution area from the upper-level power grid at time t, i.e., when the output of distributed photovoltaic power is insufficient, it needs to purchase power from the upper-level power grid; The constraints of the substation-level optimization model include power balance constraints, distributed photovoltaic output constraints, and energy storage system-related constraints, which are expressed as follows: Power balance constraints:

[0083]

[0084] In the formula, This represents the total active load of the k-th transformer area. This represents the total reactive load of the k-th transformer area. , These represent the predicted active and reactive power outputs of the distributed photovoltaic system in the kth transformer area, respectively. Distributed photovoltaic power output constraints:

[0085]

[0086] In the formula, This represents the maximum output of distributed photovoltaic power in the k-th transformer area. These represent the minimum and maximum values ​​of reactive power output from distributed photovoltaic systems, respectively. The constraints related to energy storage systems, including upper and lower limits of charge and discharge power constraints, mutual exclusion constraints of charge and discharge, SOC constraints of energy storage systems, and reactive power constraints of energy storage systems, are expressed as follows: Upper and lower limits of charge and discharge power constraints:

[0087]

[0088] In the formula, , These represent the maximum active power during charging and discharging, respectively. , These are binary variables representing the charging and discharging states of the energy storage system in the k-th transformer area, respectively. Charge-discharge mutual exclusion constraint:

[0089]

[0090] Energy storage system SOC constraints:

[0091]

[0092] In the formula, Indicates the rated capacity of the energy storage system. Indicates time interval, These represent the charging and discharging efficiencies of the energy storage system, respectively. , , These represent the state of charge of the energy storage system at time t, and the minimum and maximum allowable states of charge of the system, respectively. Reactive power constraints of energy storage systems:

[0093] In the formula, , , These represent the reactive power, upper limit, and lower limit currently generated by the energy storage system in the k-th distribution area, respectively.

[0094] Example 4 The three-level architecture constructed in the above embodiments is a closed-loop control system. The lower-level unit provides the controllable power range to the upper level, solves the optimal power plan as the lower-level control command, and then adjusts the controllable power range after receiving the command. Through the closed-loop process of "lower-level input controllable range - upper-level optimization model solution - command execution - feedback adjustment", the realization of the global optimization goal is ensured.

[0095] Step 4: Control command generation.

[0096] First, the optimal power allocation for each 10kV feeder within the region is determined using the regional optimization model. This is then used as a control command from the regional layer to the distribution area layer and sent to the corresponding nth feeder layer. Next, after receiving the regional layer power control command, the feeder layer performs optimization calculations with the goal of minimizing feeder network losses, and solves for the target power value for each distribution area. It then optimally allocates the power to the corresponding k-th distribution area. After receiving the control command, the distribution area layer coordinates and controls the distributed photovoltaic output, energy storage system charging and discharging power, and electric vehicle charging power within its area, so that the total power of each unit within the distribution area may track the control command allocated by the upper layer.

[0097] Step 5: Feedback Execution.

[0098] First, after the distribution center executes the instructions, the connection status of the energy storage system and electric vehicles, as well as the actual output of distributed photovoltaic power, change, and the distribution center re-reports the adjustable power range for the next cycle to the distribution center. , Secondly, the feeder layer controller aggregates the latest power adjustable range reported by all its subordinate transformer areas and updates it to the power adjustable range transmitted from the feeder to the area. , This is used as a key constraint in the new round of regional layer optimization calculations. This closed-loop system achieves dynamic matching between scheduling commands and the actual adjustment capabilities of distributed resources, effectively improving the system's ability to cope with uncertainties such as photovoltaic fluctuations and load changes, and ultimately ensuring the continuous achievement of the global optimization goal. The hierarchical power control framework diagram of the distribution network is shown in Figure 2.

[0099] Example 5 This embodiment uses MATLAB R2023b+YALMIP toolbox and Gurobi 12.0.1 solver to verify the method of the present invention. First, with the objectives of "optimal power allocation at the regional level, minimum network loss at the feeder level, and minimum cost at the distribution area level," the effectiveness of the inter-level constraints and closed-loop mechanism is verified through actual solution output. Second, based on a 14-node distribution network, divided into three layers according to voltage level, the regional layer consists of a single 110kV main bus, and the feeder layer has three feeders, each corresponding to 2-3 distribution areas. The structure of these feeders is optimized and scheduled, using a 14-node network as an example. Figure 3 As shown.

[0100] The basic parameters and core input data of the power distribution network are shown in Table 1 below:

[0101] Results analysis: Step 1: Each level reports the adjustable power range from bottom to top. 1. The information is reported from the transformer substation to the feeder substation. The adjustable power range of each transformer substation is calculated by aggregating the maximum output of distributed photovoltaic (PV) power (50kW), fixed load (85kW), and the charging and discharging capacity of the energy storage system (ESS) (rated capacity 100kWh, charging and discharging efficiency 0.92-0.94, initial SOC 0.5). The final adjustable power range of each transformer substation reported to feeder 1 is as follows: The active power adjustable range of transformer area 1 is [40.0, 60.0] kW, and the reactive power adjustable range is [18.0, 25.0] kVar; The active power adjustable range of transformer substation 2 is [60.0, 70.0] kW, and the reactive power adjustable range is [24.0, 28.0] kVar; Transformer Zone 3: Active power adjustable range [75.0, 85.0] kW}, reactive power adjustable range [30.0, 35.0] kVar.

[0102] 2. Feeder layer reports to region layer The feeder layer aggregates the adjustable range of its subordinate transformer substations and, in conjunction with constraints such as line resistance (0.018-0.031Ω) and allowable voltage range (9.5-10.5kV), calculates its own adjustable range and reports it to the regional layer. Feeder 1 has an adjustable active power range of [185, 210.0] kW and an adjustable reactive power range of [72.0, 88.0] kVar; Feeder 2 has an adjustable active power range of [178.0, 212.0] kW and an adjustable reactive power range of [72.0, 88.0] kVar; Feeder 3 has an adjustable active power range of [177.0, 214.0] kW and an adjustable reactive power range of [72.0, 88.0] kVar.

[0103] Step 2: Control commands are issued from top to bottom at each level. 1. The region layer sends instructions to the feeder layer. The regional optimization model aims to ensure that the active / reactive power of each feeder is close to the reference value. Combined with constraints on the total active power supplied to the main transformer and the adjustable range of the feeders, the optimal power allocation command is derived. Feeder 1 has an active power of 200.0 kW and a reactive power of 80.0 kVar; Feeder 2 has an active power of 200.0 kW and a reactive power of 80.0 kVar; Feeder 3 has an active power of 200.0kW and a reactive power of 80.0kVar.

[0104] 2. The feeder layer sends instructions to the transformer substation layer. After receiving the regional layer command, Feeder 1 optimizes the power allocation of its three subordinate transformer areas based on Distflow power flow constraints, node voltage constraints, and other constraints, with the goal of "minimizing feeder network loss". The power of each transformer area is as follows: Transformer 1 has an active power of 52.3 kW and a reactive power of 21.4 kVar; Transformer 2 has an active power output of 66.9 kW and a reactive power output of 26.6 kVar. Transformer 3 has an active power output of 80.9 kW and a reactive power output of 32.0 kVar. The total network loss of feeder 1 is calculated to be 12.61kW, and the voltage of all nodes is maintained within a safe range. Figure 4 The voltage distribution diagram of feeder 1 verifies the effectiveness of the feeder layer optimization.

[0105] Step 3: The transformer substation receives instructions and updates accordingly. Taking transformer substation 1 as an example, after receiving instructions from the feeder layer, the solution is obtained by coordinating constraints related to photovoltaics, energy storage, electric vehicles, and grid power purchase, with the goal of minimizing the operating cost of the substation layer. The optimized calculation results show that the total operating cost of substation 1 for 8 hours is 299.71 yuan, and the energy storage SOC decreases from the initial 0.5 to the safe lower limit of 0.2, reflecting that the energy storage dispatch strategy under optimal cost ensures full utilization of photovoltaic output, reasonable energy storage charging and discharging behavior, and effectively reduces power purchase costs. Simultaneously, the user side re-reports the adjustable power range for the next cycle to the substation layer, and this is aggregated and updated to the adjustable power range transmitted from the feeder to the area, forming a closed-loop system. The SOC change curve for substation 1 is shown below. Figure 5 As shown.

[0106] Example 6 This embodiment provides a hierarchical collaborative distribution network photovoltaic-storage-charging power optimization scheduling system, applying the hierarchical collaborative distribution network photovoltaic-storage-charging power optimization scheduling method described in the above embodiment, including: Distribution area control unit: used to aggregate the operating status of distributed photovoltaic, energy storage system and electric vehicle to generate and report the adjustable range of distribution area, build distribution area optimization model, execute control commands issued by feeder layer, update adjustable range and provide feedback; Feeder layer control unit: used to aggregate the adjustable range of subordinate transformer areas to generate and report the adjustable range of the feeder layer, build the feeder layer optimization model, receive instructions from the regional layer and send them to the transformer area layer, update the adjustable range and provide feedback. Regional layer control unit: Used to receive the adjustable range of the feeder layer, build the regional layer optimization model and linearize it through the big-M method, generate and issue feeder power allocation instructions, and receive feedback information from the feeder layer.

Claims

1. A hierarchical and collaborative method for optimizing the scheduling of photovoltaic, energy storage, and charging power in a distribution network, characterized in that: Specifically, the steps include the following: Step 1: Construct a three-level architecture of distribution area layer - feeder layer - regional layer according to the voltage level of the distribution network, and clarify the objectives of minimizing operating costs at the distribution area layer, minimizing network losses at the feeder layer, and optimizing power allocation at the regional layer; Step 2: Each level reports the adjustable power range level by level; Step 3: Build optimization models for each level, and linearize the regional optimization model using the Big-M method; Step 4: The regional layer and feeder layer generate and issue power control commands level by level, and the substation layer coordinates internal resources to track and execute them; Step 5: After execution, each level updates the adjustable range and feeds back to the next level, forming a closed-loop collaborative control.

2. The hierarchical collaborative power optimization scheduling method for photovoltaic, energy storage, and charging distribution networks according to claim 1, characterized in that, The specific process of reporting the adjustable power range level by level in step 2 is as follows: The distribution area layer aggregates the operating status of distributed photovoltaic, energy storage systems and electric vehicles, forms the adjustable range of distribution area power and reports it to the feeder layer; The feeder layer combines the adjustable range constraints of all its subordinate transformer areas to calculate and aggregate the adjustable power range of the feeder layer, and then reports it to the regional layer. The power adjustable range refers to the maximum and minimum limits of power adjustment for each level.

3. The hierarchical collaborative power optimization scheduling method for photovoltaic-storage-charging power in distribution networks according to claim 1, characterized in that, The objective function of the region-level optimization model described in step 3 is: In the formula, N represents the total number of 10kV feeders in the area; These represent the active power and reactive power of the nth feeder, respectively. The active and reactive power reference values ​​for the nth feeder are represented as follows: 、 The constraints of the regional layer optimization model include regional power balance constraints and feeder power allocation constraints, which are expressed as follows: Regional power balance constraints: In the formula, , These represent the active and reactive power of the regional main transformer network provided by the upper-level power grid dispatch, which is the sum of the active and reactive power of all 10KV feeders under this 110KV main transformer; Feeder power distribution constraints: In the formula, Let represent the minimum and maximum active power of the nth feeder, respectively. These represent the minimum and maximum reactive power of the nth feeder, respectively.

4. The hierarchical collaborative power optimization scheduling method for photovoltaic-storage-charging power in a distribution network according to claim 1, characterized in that, The objective function of the feeder layer optimization model described in step 3 is: In the formula, K represents the total number of stations on the nth feeder; This represents the resistance of the nth feeder; This represents the square of the current in the k-th transformer area of ​​the n-th feeder; The constraints of the feeder layer optimization model include regional power balance constraints, dynamic power flow constraints, node voltage constraints, and power adjustable range constraints, which are expressed as follows: Regional power balance constraints: In the formula, , These represent the total active power and reactive power of the network below the nth feeder layer, respectively, calculated by the regional layer optimization. , These represent the active power and reactive power of the k-th transformer area, respectively. Dynamic power flow constraints: In the formula, Indicates the nth feeder line Branch node current conjugate, Indicates from node Downstream nodes of the outflow channel Total power, Indicates from node Injection Node Total power, From node Transmit to node Branch power loss; , These represent the nodes of the nth feeder. and nodes The voltage amplitude; Indicates the connection to the nth feeder node. and the branch impedance of the node, Indicates a branch Apparent power on Represents a node The injected apparent power at the location, This represents the set of all branches in a power distribution network. These represent the sets of all nodes in the network; Because the dynamic power flow constraints contain non-convex terms, the second-order cone relaxation principle is used to relax the non-convex constraints. The relaxed branch dynamic power flow constraints are expressed as follows: In the formula, Represents a node The amount of active power injected into the nth feeder, Indicates from node The sum of active power flowing to all its child nodes. Indicates from node Flow to Node The active power on the branch line; Represents a node The reactive power injection amount of the nth feeder Indicates from node The sum of reactive power flowing to all its child nodes. Indicates from node Flow to Node Reactive power on the branch line, , Representing nodes respectively and nodes The square of the voltage amplitude, Indicates the reactance value of the branch circuit. Indicates a branch The resistance value; Node voltage constraints: In the formula, This represents the voltage on the nth feeder. These represent the lower and upper limits of the permissible voltage deviation, respectively. Power adjustable range constraints: In the formula, These represent the active power and reactive power output of the kth transformer area of ​​the nth feeder, respectively. Let represent the minimum and maximum values ​​of the output active power of the k-th transformer area on the n-th feeder, respectively. These represent the minimum and maximum values ​​of the output reactive power of the k-th transformer area on the n-th feeder, respectively.

5. The hierarchical collaborative power optimization scheduling method for photovoltaic-storage-charging power in a distribution network according to claim 1, characterized in that, The objective function of the station area layer optimization model described in step 3 is: In the formula, Let T represent the operating cost of the k-th transformer area, and T represent the optimization period. These represent the charging cost and discharging cost of the energy storage system, respectively, in yuan / kWh; This represents the charging and discharging power of the energy storage system in the k-th transformer area at time t; The unit purchase cost of electricity from the superior power grid is RMB / kWh; This represents the power purchased by the k-th distribution area from the upper-level power grid at time t, i.e., when the output of distributed photovoltaic power is insufficient, it needs to purchase power from the upper-level power grid; The constraints of the transformer substation optimization model include power balance constraints, distributed photovoltaic output constraints, and energy storage system-related constraints, which are expressed as follows: Power balance constraints: In the formula, This represents the total active load of the k-th transformer area. This represents the total reactive load of the k-th transformer area. , These represent the predicted active and reactive power outputs of the distributed photovoltaic system in the kth transformer area, respectively. Distributed photovoltaic power output constraints: In the formula, This represents the maximum output of distributed photovoltaic power in the k-th transformer area. These represent the minimum and maximum values ​​of reactive power output from distributed photovoltaic systems, respectively. The constraints related to energy storage systems, including upper and lower limits of charge and discharge power constraints, mutual exclusion constraints of charge and discharge, SOC constraints of energy storage systems, and reactive power constraints of energy storage systems, are expressed as follows: Upper and lower limits of charge and discharge power constraints: In the formula, , These represent the maximum active power during charging and discharging, respectively. , These are binary variables representing the charging and discharging states of the energy storage system in the k-th transformer area, respectively. Charge and discharge mutual exclusion constraint: Energy storage system SOC constraints: In the formula, Indicates the rated capacity of the energy storage system. Indicates time interval, These represent the charging and discharging efficiencies of the energy storage system, respectively. , , These represent the state of charge of the energy storage system at time t, and the minimum and maximum allowable states of charge of the system, respectively. Reactive power constraints of energy storage systems: In the formula, , , These represent the reactive power, upper limit, and lower limit currently generated by the energy storage system in the k-th distribution area, respectively.

6. The hierarchical collaborative power optimization scheduling method for photovoltaic-storage-charging distribution networks according to claim 3, characterized in that, The region-level optimization model is a nonlinear model. By introducing 0-1 integer variables and auxiliary continuous variables, and combining the Big-M method, the region-level optimization model is linearized. The specific process is as follows: First, introduce auxiliary variables. ,make And introduce 0-1 variables Add linear constraints: In the formula, It is a sufficiently large integer selected based on the range of active power variation; Secondly, auxiliary variables are introduced. ,make And introduce 0-1 variables ; Add linear constraints: In the formula, It is a sufficiently large integer selected based on the range of reactive power variation; Subsequently, the absolute value term in the original objective function is replaced to form the objective function of the linearized region layer optimization model, as follows: 。 7. The hierarchical collaborative power optimization scheduling method for photovoltaic-storage-charging distribution networks according to claim 3, characterized in that, The specific process of step 4 is as follows: The regional layer solves for the optimal power distribution of each 10kV feeder based on its own optimization model, and sends it as a control command to the corresponding feeder layer. After receiving the instructions from the area layer, the feeder layer calculates the power target value of each area based on its own optimization model and sends it to the corresponding area layer as a control instruction. After receiving instructions from the feeder layer, the distribution area layer coordinates and controls the output of internal distributed photovoltaic power, the charging and discharging power of the energy storage system, and the charging power of electric vehicles, so that the total power of the distribution area tracks the requirements of the instructions.

8. The hierarchical collaborative distribution network photovoltaic-storage-charging power optimization scheduling method according to claim 7, characterized in that, The specific process of closed-loop collaborative control described in step 5 is as follows: After the distribution area layer executes the instruction, it updates the adjustable range of the distribution area power and reports it to the feeder layer based on the access status of the energy storage system and electric vehicles and the changes in the actual output of distributed photovoltaics. The feeder layer aggregates the updated adjustable ranges of all subordinate transformer areas, generates a new adjustable power range for the feeder layer, and reports it to the regional layer. The region layer uses the updated adjustable range of the feeder layer as a new constraint, re-solves the optimization model, generates updated control commands, and enters the next round of scheduling process.

9. A hierarchical and collaborative power optimization and scheduling system for photovoltaic, energy storage, and charging distribution networks, characterized in that: The hierarchical collaborative distribution network power optimization scheduling method for photovoltaic, energy storage and charging as described in any one of claims 1-8 includes: a transformer substation control unit, a feeder substation control unit, and a regional control unit.

10. The hierarchical collaborative power optimization scheduling system for photovoltaic-storage-charging distribution networks according to claim 9, characterized in that, The transformer substation control unit is used to aggregate the operating status of distributed photovoltaic, energy storage systems and electric vehicles to generate and report the adjustable range of the transformer substation, build a transformer substation optimization model, execute control commands issued by the feeder layer, update the adjustable range and provide feedback. The feeder layer control unit is used to aggregate the adjustable range of subordinate transformer areas to generate and report the adjustable range of the feeder layer, build the feeder layer optimization model, receive instructions from the regional layer and send them to the transformer area layer, update the adjustable range and provide feedback. The regional layer control unit is used to receive the adjustable range of the feeder layer, build a regional layer optimization model and linearize it using the big-M method, generate and issue feeder power allocation instructions, and receive feedback information from the feeder layer.

Images