A voltage regulator and energy router joint planning method considering extreme working conditions

Abstract

The application discloses a voltage regulator and energy router joint planning method considering extreme working conditions, and belongs to the technical field of power system planning and regulation. The method is characterized in that: a comprehensive scene set of conventional and extreme scenes is constructed, a double-layer optimization planning model with annualized investment cost and weighted voltage deviation as double targets is established, the site selection and capacity determination scheme of VVR and EER is solved in the upper layer, the VVR tapping head gear and EER port power are optimized in the lower layer, and a multi-objective evolutionary algorithm based on quantum coding is adopted for solving. The application can improve the voltage safety under extreme working conditions while taking into account the economy, does not depend on high-precision prediction, effectively overcomes the integral saturation failure problem of energy storage, and provides a robust voltage regulation equipment planning scheme for an active distribution network.

Description

Technical Field

[0001] This invention belongs to the field of power system planning and control, and specifically relates to a joint planning method for voltage regulators and energy routers that takes into account extreme operating conditions. Background Technology

[0002] As the penetration rate of distributed photovoltaic and other renewable energy sources in active distribution networks continues to increase, the operating characteristics of distribution networks have undergone profound changes. The intermittency, volatility, and uncertainty of renewable energy output have made the problems of system voltage exceeding limits and frequent fluctuations increasingly prominent. Especially under extreme weather conditions or severe fluctuations in source and load (such as sudden increases in photovoltaic power or drastic load changes), the risk of voltage instability increases significantly, seriously threatening the safe operation of the power grid.

[0003] To suppress voltage fluctuations, various adjustment methods have been employed in existing technologies.

[0004] A common approach is to utilize battery energy storage systems to absorb or release power to mitigate voltage fluctuations. However, the effectiveness of this approach heavily relies on accurate forecasts of photovoltaic output and load demand. Under extreme conditions, source-load fluctuations often exhibit significant non-stationary, heavy-tailed distribution characteristics, making it difficult for existing forecasting models to avoid systematic biases. Since energy storage systems are essentially energy integrators, their state of charge (SOC) drifts unidirectionally with the accumulation of forecast errors, easily reaching capacity limits and falling into integral saturation during prolonged periods of extreme conditions. Once integral saturation occurs, the energy storage system will lose its effective power throughput capacity at the critical moments when voltage regulation is most needed, or even completely lose its regulatory power, leading to severe voltage exceedance accidents. Furthermore, energy storage systems have high investment costs and their lifespan is limited by the number of charge-discharge cycles, making large-scale, economical deployment in wide-area distribution networks difficult.

[0005] Another approach is to use traditional voltage / reactive power regulators (such as transformers and voltage regulators with on-load tap changers). These devices adjust the voltage reference by changing the branch turns ratio through mechanical tap switching. Their advantages include high voltage regulation capacity, high reliability, and independence from prediction and integral saturation. However, the mechanical tap's operation is discrete and lag-dependent, with inherent dead zones and resolution limitations. It cannot finely smooth high-frequency voltage residuals, resulting in a stepped voltage curve and making it difficult to precisely control the voltage near the rated value.

[0006] Another solution is to use flexible interconnection devices such as energy routers. These devices, based on power electronic conversion technology, possess millisecond-level continuous power regulation capabilities and can quickly respond to voltage fluctuations. However, their regulation capabilities are limited by converter capacity and DC-side energy constraints. When facing network-wide, large-amplitude voltage deviations, relying solely on energy routers for power routing often faces enormous power throughput demands, easily leading to equipment overload, difficulty in providing sustained and strong voltage support, and even potential shutdown due to prolonged overload.

[0007] It is evident that existing voltage regulation schemes all suffer from inherent flaws that are difficult to overcome: energy storage schemes rely on prediction and are prone to integral saturation; traditional voltage regulators exhibit discrete regulation and have dead zones; and energy routers have limited capacity and are unable to undertake network-wide voltage regulation tasks. Even when voltage regulators and energy routers are simply combined, the inherent differences in their regulation characteristics (the former is discrete, the latter is continuous) mean that how to make them work together and complement each other's strengths remains an unsolved technical challenge.

[0008] Therefore, there is an urgent need to propose a voltage regulation scheme that can overcome the above-mentioned defects and simultaneously leverage the high-throughput support capability of voltage regulators and the fine-tuning characteristics of energy routers without relying on high-precision prediction. Summary of the Invention

[0009] To address the aforementioned problems in the prior art, this invention provides a joint planning method for voltage regulators and energy routers that takes into account extreme operating conditions.

[0010] The technical problem to be solved by this invention is achieved through the following technical solution: A joint planning method for voltage regulators and energy routers considering extreme operating conditions includes the following steps: Step 1: Obtain historical operation data of the active distribution network, extract regular operation scenarios from the historical operation data and add extreme boundary scenarios to construct a comprehensive scenario set containing multiple types of scenarios; Step 2: Construct a two-layer optimization planning model comprising an upper-layer model and a lower-layer model. The upper-layer model uses the annualized investment cost and the weighted voltage deviation for all scenarios as dual objectives, and the location and capacity setting schemes of voltage regulators (VVRs) and energy routers (EERs) as decision variables. The weighted voltage deviation is the weighted sum of voltage deviations for all scenarios. The voltage deviation is the deviation between the per-unit voltage value of a node in the active distribution network and the rated per-unit voltage value. Under the location and capacity setting scheme given by the upper-layer model, the lower-layer model aims to minimize the voltage deviation of a single scenario, using the tap position of the VVR and the power of each port of the EER as decision variables. It obtains the voltage trajectory of each node through power flow simulation and calculates the voltage deviation, returning it to the upper-layer model. Step 3: Solve the bi-level optimization programming model using a quantum-encoded multi-objective evolutionary algorithm. In the algorithm, an individual is a location-capacity scheme, and the population is a set of individuals. Each individual includes a location variable and a capacity variable. The location variable uses quantum phase angle encoding, and the capacity variable uses real number encoding. In each generation of population evolution, the location encoding is updated through a quantum rotation gate, the capacity encoding is updated through a heavy-tailed random walk, and the external archive is maintained using Pareto sorting and the hypervolume contribution index. The external archive is a set of non-dominant location-capacity schemes obtained from the population during the evolutionary process. Step 4: Iterate through Step 3 until convergence, then select the optimal addressing and occupancy scheme from the current external archives and output it.

[0011] Optionally, the step of extracting regular operating scenarios from the historical operating data and adding extreme boundary scenarios to construct a comprehensive scenario set containing multiple types of scenarios includes: Regular operating scenarios are extracted from the historical operating data using the K-medoids clustering algorithm; Based on the load extreme index and photovoltaic fluctuation index in the historical operating data, extreme boundary scenarios are extracted from the historical operating data; the load extreme index is the ratio of the maximum daily total load to the average daily load, and the photovoltaic fluctuation index is the maximum daily photovoltaic power fluctuation rate; A comprehensive scenario set is constructed based on the normal operating scenarios and the extreme boundary scenarios.

[0012] Optionally, the rotation step size of the quantum rotating gate is determined in the following manner: For each location and capacity scheme in the population, the difference in weighted voltage deviation between it and the reference location and capacity scheme selected from the external archive is normalized by dividing by the difference between the maximum and minimum weighted voltage deviations in the current generation population, and then multiplied by the base step size coefficient after being mapped by the sigmoid function to obtain the rotation step size.

[0013] Optionally, the addressing variable is encoded using quantum phase angle, including: mapping the binary decision of whether to install a VVR or EER at each candidate installation location to the phase angle of a qubit, wherein the squared probability amplitude value corresponding to the phase angle represents the probability of installing the device at that location; The capacity variables are encoded as real numbers, including encoding the rated capacity of the VVR and the rated capacity of the EER port as real number variables, and the value range of each real number variable is constrained by the upper limit of the capacity.

[0014] Optionally, maintaining the external archive using Pareto ordering and hypervolume contribution metrics includes: Pareto non-dominated sorting was used to stratify the site selection and gradation schemes in the population based on two objectives: annualized investment cost and weighted voltage deviation. The schemes that do not dominate each other were stored in an external archive. The external archive size is managed using a super-volume contribution index: For each site selection and capacity determination scheme in the external archive, the super-rectangular volume enclosed by the reference point in the target space formed by the annualized investment cost and weighted voltage deviation is calculated, and the super-rectangular volume is used as the contribution of the site selection and capacity determination scheme; when the number of site selection and capacity determination schemes in the external archive exceeds a preset upper limit, the site selection and capacity determination scheme with the smallest contribution is deleted; wherein, the reference point corresponds to the largest annualized investment cost and the largest weighted voltage deviation.

[0015] Optionally, the step of selecting the optimal addressing and capacity stabilization scheme from the current external archives includes: normalizing the annualized investment cost and weighted voltage deviation of each addressing and capacity stabilization scheme in the current external archives, calculating the Euclidean distance from each addressing and capacity stabilization scheme to the ideal point, and selecting the addressing and capacity stabilization scheme with the smallest distance as the optimal addressing and capacity stabilization scheme output; the annualized investment cost of the ideal point is zero and the weighted voltage deviation is zero. Optionally, the annualized investment cost is expressed as: ; In the formula, This represents the annualized investment cost; This is a pre-defined set of VVR candidate installation branches; For branch index; Cost per unit capacity of VVR; branch road The rated capacity of the VVR on the device; Cost per unit capacity of parallel capacitor banks in a VVR; branch road The capacitor configuration capacity is used in conjunction with the VVR; This is a pre-defined set of potential EER access nodes. For EER port index; Cost per unit capacity of EER ports; For the first Rated configuration capacity of each EER port; This is the capital recovery coefficient.

[0016] Optionally, the weighted voltage deviation for all scenarios is expressed as: ; In the formula, Weighted voltage deviation for all scenarios; The total number of scenes in the comprehensive scene set; For scene indexing; For the scene The weights; This represents the total number of nodes; For node index; the node is the electrical connection point in the active distribution network, the VVR is installed between the nodes at both ends of the branch to regulate the node voltage, and the EER is connected to the node to inject or absorb power into the node; This represents the total number of scheduling periods within a single scenario. Indexed by time period; In the scene Next, node exist The voltage trajectory is composed of the per-unit voltage amplitude values ​​of each node at each time point. This is the voltage rating reference value.

[0017] Optionally, the objective function of the lower-level model is: ; In the formula, For the scene Voltage deviation; numerical value This is the per-unit value of the rated voltage; This represents the active power balance constraint of a node, which is used to ensure that the sum of active power injected into each node in each time period is equal to the sum of active power flow flowing out of that node. This represents the reactive power balance constraint at the node, which is used to constrain the sum of reactive power injected into each node in each time period to be equal to the sum of reactive power flowing out of that node. This represents a voltage state transition constraint based on linearized DistFlow, used to constrain the voltage squared difference between the nodes at both ends of a branch to be determined by the active and reactive power flow and impedance of the branch. This indicates the VVR tap position constraint, used to constrain the tap position to be a discrete integer, and the adjustment range of the tap position in adjacent time periods does not exceed the maximum value; This indicates the EER port power constraint, which is used to constrain the input power and output power of each EER port to be limited by the rated capacity, and to ensure that the same EER port cannot simultaneously input and output power, and to constrain the sum of the output power of all EER ports to be equal to the sum of the input power multiplied by the conversion efficiency; This indicates node voltage safety constraints, which are used to ensure that the per-unit value of the voltage amplitude at each node is kept within the allowable range.

[0018] In a second aspect, the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes the computer program stored in the memory, it implements the steps of the above-described joint planning method for voltage regulators and energy routers that takes into account extreme operating conditions.

[0019] Compared with the prior art, the present invention has at least the following beneficial effects: The voltage regulator and energy router joint planning method provided by this invention, which considers extreme operating conditions, can simultaneously cover normal operating scenarios and long-tail extreme operating conditions through a joint mechanism of clustering in the scenario construction stage and forced addition of extreme scenarios. Addressing the shortcomings of energy storage solutions that rely on prediction and are prone to integral saturation, a joint planning of VVR and EER is adopted. Both are instantaneous response devices and do not involve energy time-scale integration, thus physically avoiding SOC integral saturation failure caused by the accumulation of prediction errors. To address the problems of discrete adjustment and dead zones in traditional voltage regulators, the continuous adjustment capability of EER is used to fill the tap adjustment dead zone of VVR, achieving synergistic complementarity between discrete coarse adjustment and continuous fine adjustment. To address the limitations of energy router capacity and its inability to undertake network-wide voltage regulation tasks, the high-throughput electromagnetic voltage regulation capability of VVR is used to undertake the main voltage support, significantly reducing the power throughput requirements of EER and avoiding equipment overload risks. A quantum encoding and co-evolutionary mechanism was designed for the hybrid decision space of "discrete addressing-continuous volume determination," overcoming the problems of traditional algorithms easily getting trapped in local optima and premature convergence. Simultaneously, an external archive is maintained through a hypervolume contribution index, ensuring the optimal balance between economy and safety under extreme conditions in the final solution. The combined effect of these techniques allows this invention to leverage the high-throughput support capability of VVR and the fine-tuning characteristics of EER without relying on high-precision prediction, achieving an organic synergy between the two and effectively solving the problems existing in current technologies.

[0020] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0021] Figure 1 The diagram illustrates the failure mechanism of an energy storage system (ESS) under extreme conditions; Figure 2 The topology diagram of the VVR is shown below; Figure 3 The diagram shows the topology of the EER; Figure 4 This is a flowchart of a joint planning method for voltage regulators and energy routers that takes into account extreme operating conditions, provided by the present invention; Figure 5 A block diagram of the VVR-EER two-layer joint planning architecture of the present invention is shown; Figure 6 The algorithm flow of the present invention is shown; Figure 7This is a schematic diagram of the structure of an electronic device provided by the present invention. Detailed Implementation

[0022] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0023] Energy storage systems (ESS) are commonly used in voltage regulation of existing power distribution networks, achieving peak shaving and valley filling and voltage support through the storage and release of electrical energy. However, the regulation capability of ESS systems is highly dependent on the accurate prediction of the time-varying trajectories of photovoltaic power output and load demand. Under extreme weather conditions or severe source-load fluctuations, the actual source-load sequence often exhibits significant non-stationary, heavy-tailed distribution characteristics, inevitably leading to systematic biases in the prediction model. These biases accumulate over time, continuously through the integral dynamic equation of the energy storage system's state of charge (SOC), eventually causing the SOC to continuously drift unidirectionally and reach the upper and lower capacity limits, i.e., entering integral saturation. At this point, due to the hard constraint on SOC safety, the energy storage system's charging and discharging regulation rights are forcibly locked. Even if the system is in an emergency situation such as voltage exceeding limits, the energy storage cannot execute effective power commands, thus losing its voltage regulation capability. Figure 1 The mechanism by which ESS leads to integral saturation and control failure due to the accumulation of prediction errors under extreme operating conditions is illustrated.

[0024] To address the failure issues of existing energy storage solutions under extreme conditions, and to leverage the high-throughput support capability of voltage regulators and the fine-tuning characteristics of energy routers without relying on high-precision predictions, this invention provides a joint planning method for voltage regulators (VVRs) and energy routers (EERs) that considers extreme conditions. This method replaces traditional prediction-dependent energy storage systems with a combination of VVRs and EERs. The VVR undertakes the main voltage support task for the entire network, reducing the power throughput pressure on the EERs, while the EERs precisely fill the regulation dead zones caused by VVR tap changes. This physically avoids the SOC integral saturation failure problem caused by the accumulation of prediction errors. By not relying on high-precision prediction models, it significantly reduces investment costs while improving the voltage self-healing capability and operational resilience of active distribution networks under extreme uncertainty environments.

[0025] like Figure 2 As shown, the VVR adopts a structure that combines a multi-tap series voltage regulating winding with a parallel capacitor bank, and is connected in series to the distribution network branch. By operating the tap, a variable compensation voltage is injected into the branch to change the line voltage reference. Figure 2 middle, This represents the voltage phasor at the first node of the VVR access branch. This represents the voltage phasor at the end node of the VVR access branch. Indicates the connection node and nodes The equivalent resistance of the branch, This represents the equivalent reactance of the branch. This represents the equivalent resistance of the VVR series winding. This represents the equivalent reactance of the VVR series winding. - It is the switching switch for the capacitor bank on the parallel side of the VVR. For example... Figure 3 As shown, the EER uses a multi-port power electronic conversion structure to connect to the distribution network node, and achieves cross-node power mutual assistance through power routing between ports.

[0026] Based on a detailed modeling of the physical topology of VVR and EER, this invention first collects historical operating data and extracts features. Through clustering algorithms and extreme index filtering, the uncertain operating environment is distilled into a comprehensive scenario set containing both normal operating scenarios and extreme boundary scenarios. Secondly, a two-layer optimization planning model is constructed. The upper-layer model has two objectives: annualized investment cost and weighted voltage deviation across all scenarios. It optimizes long-term investment benefits and voltage safety under extreme conditions as opposing objectives. Finally, an innovative quantum-encoded multi-objective evolutionary algorithm is used to solve the problem. In this algorithm, individuals are assigned to addressing and capacity-determining schemes. The addressing variable is encoded using quantum phase angle, and the capacity variable is encoded using real numbers. In each generation of population evolution, the addressing code is updated through a quantum rotation gate, and the capacity code is updated through a heavy-tailed random walk. Pareto sorting and hypervolume contribution index are used to maintain the external archives and search for the globally optimal addressing and capacity-determining scheme. This determines the installation location and capacity of the VVR and the access location and port capacity of the EER. The VVR undertakes the main voltage support task of the entire network to reduce the power throughput pressure on the EER. The EER accurately fills the regulation dead zone caused by the VVR tap changer action. It makes full use of the complementary characteristics of the VVR and EER, and balances voltage safety and investment economy under extreme conditions without relying on high-precision prediction.

[0027] The joint planning method provided by this invention will be described in detail below. For example... Figure 4 As shown, the joint planning method provided by the present invention includes the following steps: Step 1: Obtain historical operating data of the active distribution network, extract regular operating scenarios from the historical operating data and add extreme boundary scenarios to construct a comprehensive scenario set containing multiple types of scenarios.

[0028] Here, historical operational data includes grid-wide load data and distributed photovoltaic data. By collecting historical operational data and extracting its features (such as load time-series features, photovoltaic output time-series features, and / or the combination of both), and through clustering algorithms and extreme index filtering, the uncertain operating environment is distilled into a comprehensive set of scenarios containing both typical features and extreme boundary features.

[0029] For example, in one implementation, step 1 involves extracting regular operating scenarios from historical operating data and adding extreme boundary scenarios to construct a comprehensive scenario set containing multiple types of scenarios, including: Regular operating scenarios are extracted from historical operating data using the K-medoids clustering algorithm; extreme boundary scenarios are extracted from historical operating data based on load extreme indicators and photovoltaic fluctuation indicators; the load extreme indicator is the ratio of the maximum daily total load to the daily average load, and the photovoltaic fluctuation indicator is the daily maximum photovoltaic power fluctuation rate; a comprehensive scenario set is constructed based on regular operating scenarios and extreme boundary scenarios.

[0030] For example, collect historical operating data of the target distribution network for the entire year, including active and reactive loads and distributed photovoltaic output at electrical connection points (hereinafter referred to as nodes) within the target distribution network. Reconstruct the 8760 hours of data for the year into 365 daily operating vectors. Use K-medoids clustering, K-means clustering, spectral clustering, or a Gaussian mixture model-based clustering algorithm to partition the daily operating vector set, and select... Cluster centroids are used as a typical scenario. Extreme load indices are defined. The photovoltaic fluctuation index is defined as the ratio of the maximum daily total load to the average daily load. This represents the maximum intraday photovoltaic power fluctuation rate. An extreme scenario forced inclusion logic is implemented: if an extreme indicator for a given operating day exceeds the 95th percentile and is not selected in a cluster, it is forcibly incorporated into the comprehensive scenario set as an extreme boundary scenario. Then, weights are assigned to each scenario. Satisfy weight normalization across all scenarios , The comprehensive scenario set is output as the number of scenarios in the comprehensive scenario set, including extreme boundary characteristics. In addition, the extreme indicators in this invention, besides the ratio of maximum daily total load to average daily load and the daily maximum photovoltaic power fluctuation rate, can also include indicators such as voltage collapse risk index or net load gradient.

[0031] In step 1, the long-tail distribution scenario (i.e. extreme boundary scenario) that represents extreme robustness is included in the planning set, which enables the subsequent planning scheme to take into account both the economy of normal operation and the safety under extreme conditions.

[0032] Step 2: Construct a two-layer optimization planning model consisting of an upper-layer model and a lower-layer model. The upper-layer model has two objectives: annualized investment cost and weighted voltage deviation for all scenarios. The location and capacity setting schemes of voltage regulators (VVRs) and energy routers (EERs) are used as decision variables. The weighted voltage deviation is the weighted sum of voltage deviations for all scenarios. Voltage deviation is the deviation between the per-unit voltage value of a node in the active distribution network and the rated per-unit voltage value. Under the location and capacity setting scheme given by the upper-layer model, the lower-layer model aims to minimize the voltage deviation of a single scenario. The tap position of the VVR and the power of each port of the EER are used as decision variables. The voltage trajectory of each node is obtained through power flow simulation calculation, and the voltage deviation is calculated and returned to the upper-layer model.

[0033] Specifically, the first step is to construct an active distribution network fine-tuning model that takes into account the physical characteristics of VVR and EER: (1) Constructing the generalized power injection model of VVR: The branch connected to the VVR is modeled as a function of line impedance. VVR equivalent internal resistance and controlled series voltage source The composite system constitutes a branch impedance. for: ; In the formula, For connecting nodes and nodes The original resistance of the branch, For connecting nodes and nodes The reactance of the branch circuit; and These are the equivalent resistance and reactance of the VVR series winding, respectively. , .

[0034] According to Kirchhoff's voltage law, the regulating current generated by VVR Stripped from branch current: ; In the formula, This refers to the voltage phasor injected into the VVR series winding through the tap changer operation.

[0035] Further establish VVR at the node Equivalent complex power injection generated : ; In the formula, the asterisk (*) in the upper right corner indicates conjugate.

[0036] Considering the engineering characteristics of active distribution network nodes where the voltage amplitude is maintained around 1.0 (1 times the rated voltage) and the phase angle difference is extremely small, the following is introduced: By approximating the data, a generalized power injection model for VVR at the node is derived: ; ; In the formula, Injecting nodes into VVR The equivalent active power, Injecting nodes into VVR The equivalent reactive power, Compensation reactive power provided for the VVR parallel-side capacitor bank.

[0037] according to Figure 2 The multi-tap design shown is Gear variable constraint: ; ; In the formula, For any branch road During the period Tap position; For single-range voltage adjustment step size, For branch roads during time periods of , This is the smallest tap position. This is the maximum tap position. It is a set of integers.

[0038] (2) Constructing a multi-port active routing model for EER: Suppose that EER has The first port. (For the first...) Each port during the time period Net injection power Modeling: ; ; ; In the formula, Active power absorbed by the port. The active power released by the port; This represents the maximum port capacity. This is a power direction indicator variable.

[0039] Energy exchange between EER ports must adhere to conversion efficiency. constraint: .

[0040] The above steps fully describe the mechanism by which VVR shifts the voltage reference by changing the network's physical equivalent impedance, and the mechanism by which EER compensates for residuals by real-time port power redistribution, providing physical model support for subsequent two-layer joint planning.

[0041] Then, a lower-level model is constructed. This lower-level model aims to minimize the voltage deviation in a single scenario, using the VVR tap position and the power at each port of the EER as decision variables. Voltage trajectories at each node are obtained through power flow simulation, and the voltage deviation is calculated. Its objective function is: ; In the formula, For scene indexing, For time period index, The node index is an electrical connection point in an active distribution network. The VVR is installed between the nodes at both ends of the branch to regulate the node voltage. The EER is connected to the node to inject or absorb power into the node. Representing a scene Voltage deviation; This represents the total number of scheduling periods within a single scenario. This represents the total number of nodes; In the scene Next, node exist The voltage amplitude per unit value at each moment; the voltage trajectory is composed of the voltage amplitude per unit value at each node at each moment, numerically... This is the per-unit value of the rated voltage; This represents the active power balance constraint of a node, which is used to ensure that the sum of active power injected into each node in each time period is equal to the sum of active power flow flowing out of that node. This represents the reactive power balance constraint at the node, which is used to constrain the sum of reactive power injected into each node in each time period to be equal to the sum of reactive power flowing out of that node. This represents a voltage state transition constraint based on linearized DistFlow, used to constrain the voltage squared difference between the nodes at both ends of a branch to be determined by the active and reactive power flow and impedance of the branch. This indicates the VVR tap position constraint, used to constrain the tap position to be a discrete integer, and the adjustment range of the tap position in adjacent time periods does not exceed the maximum value; This indicates the EER port power constraint, which is used to constrain the input power and output power of each EER port to be limited by the rated capacity, and to ensure that the same EER port cannot simultaneously input and output power, and to constrain the sum of the output power of all EER ports to be equal to the sum of the input power multiplied by the conversion efficiency; This indicates node voltage safety constraints, which are used to ensure that the per-unit value of the voltage amplitude at each node is kept within the allowable range.

[0042] Based on the objective function of the lower-level model, it can be understood that the lower-level model aims to minimize the scene. Voltage deviation of all nodes throughout the entire time period.

[0043] The active power balance constraints at the above nodes are specifically as follows: ; In the formula, For the grid connection point (the access point where power is fed into the distribution network from the upstream power grid) in Injecting nodes at all times The active power; For nodes exist Predicted output of distributed photovoltaic power at any given time; For EER port in Injecting nodes at all times Active power regulation, For VVR in The equivalent injected active power generated by the constant transformer ratio adjustment For nodes exist Active power load demand at any given time; For nodes The set of branch roads at the beginning; For connecting nodes and nodes The side road The ever-flowing tide of merit; For nodes For the set of branches at the end; To start from the parent node Inflow node The positive trend.

[0044] The specific reactive power balance constraints at the above nodes are as follows: ; In the formula, Reactive power injected into the grid connection point; For nodes VVR matching capacitor bank at The amount of reactive power compensation provided at all times; The equivalent reactive power injection generated by VVR regulation; For nodes The reactive load demand; For connecting nodes and nodes The unproductive current flowing out of the branch road To start from the parent node Inflow node The reactive current flows into the branch road.

[0045] The voltage state transition constraints based on the linearized DistFlow described above are specifically as follows: ; In the formula, For nodes The square of the voltage amplitude, i.e. , For nodes The square of the voltage amplitude; and These are the combined equivalent resistance and equivalent reactance of the branch circuit after VVR access, respectively.

[0046] In this invention, the relationship between node voltage and branch power is described by linearized power flow equations, thereby constructing node power balance constraints and voltage state transition constraints for the lower-level model using linearized DistFlow power flow equations.

[0047] The specific VVR tap position constraints mentioned above are as follows: ; In the formula, For the scene Lower branch road The voltage injected into the upper VVR; For the corresponding tap position integer decision variable, the range of values ​​is... ; To fix the voltage adjustment step size.

[0048] The above EER port power constraints Specifically: ; ; ; .

[0049] In the formula, For the operational conversion efficiency of EER, Indicates in the scene Next, EER Each port during the time period Net injection power .

[0050] The specific node voltage safety constraints mentioned above are as follows: ; For permitted The minimum value, Permitted The maximum value.

[0051] Then, an upper-level model is constructed, which takes the annualized investment cost and the weighted voltage deviation of all scenarios as dual objectives, and the location and capacity setting schemes of VVR and EER as decision variables. Specifically, it includes two parts: the annualized total investment cost and the weighted voltage deviation of all scenarios.

[0052] Specifically, the annualized investment cost in the upper-level model represents the annualized total investment cost of the planning scheme, and its formula can be defined as: ; In the formula, The annualized investment cost to be optimized; This is a pre-defined set of VVR candidate installation branches; For branch index; Cost per unit capacity of VVR; branch road The rated capacity of the VVR on the device; Cost per unit capacity of parallel capacitor banks in a VVR; branch road The capacitor configuration capacity is used in conjunction with the VVR; This is a pre-defined set of potential EER access nodes. For EER port index; Cost per unit capacity of EER ports; For the first Rated configuration capacity of each EER port; This is the capital recovery factor, used to discount a one-time construction investment to the average annual expenditure over the equipment's lifespan. It represents the time value of money. The formula is defined as follows: ; In the formula, The discount rate; This refers to the planned service life or financial depreciation period of the equipment.

[0053] The weighted voltage deviation of all scenes in the upper-level model can be defined as the mean square error of the weighted voltage deviation across all scenes, specifically: ; In the formula, The weighted voltage deviation for all scenarios can characterize the overall voltage deviation level of the entire network under typical and extreme operating scenarios; This represents the total number of scenes in the comprehensive scene set. For scene indexing; For the scene The weights; This represents the total number of nodes; For node indexing; This represents the total number of scheduling periods within a single scenario; for example, it can be 24. Indexed by time period; In the scene Next, node exist The per-unit value of the voltage amplitude at a given time. This is the voltage rating reference value.

[0054] Finally, define the decision variables. This represents the site selection and sizing scheme. Specifically, The location decision vector for the VVR determines which branch lines the VVR will be installed on. The VVR sizing decision vector is determined by the VVR rated capacity of each branch. composition; The access node index decision vector for the EER port; The capacity decision vector for the EER is determined by the rated configuration capacity of the EER port. composition.

[0055] At this point, both the upper-level and lower-level models have been constructed, together forming the bi-level optimization programming model of this invention. It is also necessary to clarify the coupling logic relationships between the various models included in the bi-level optimization programming model. Figure 5 A block diagram of the VVR-EER two-layer joint planning architecture of the present invention is shown. Figure 5 As shown, firstly, historical operating data of the distribution network is collected, and regular operating scenarios are extracted through clustering. Extreme boundary scenarios are then forcibly added using an extreme index filter to construct a comprehensive scenario set. Subsequently, the upper-level model is used for site selection and capacity determination decisions, and the upper-level planning scheme (decision variables) is passed to the lower-level model. For each scenario in the comprehensive scenario set, a separate lower-level model is activated. The lower-level model corresponding to each scenario aims to minimize the voltage deviation of all nodes within that scenario throughout the entire time period, by adjusting the tap position variables. and EER port power Output the optimal voltage trajectory The returned data is sent to the upper-level model. In practice, this can be achieved by calling the quadratic programming solver Gurobi to solve the lower-level model. Then, the upper-level model calculates the weighted voltage deviation based on the optimal voltage trajectory returned by the lower-level model for each scenario. Through iterative search make and At the same time, by minimizing the non-dominated solutions as much as possible, the optimal location and occupancy scheme can be obtained.

[0056] Step 3: Solve the bi-level optimization programming model using a quantum-encoded multi-objective evolutionary algorithm. In this algorithm, an individual is a location and capacity scheme, and the population is a set of individuals. Individuals include location variables and capacity variables. The location variables are encoded using quantum phase angles, and the capacity variables are encoded using real numbers. In each generation of population evolution, the location encoding is updated through a quantum rotation gate, the capacity encoding is updated through a heavy-tailed random walk, and the external archive is maintained using Pareto sorting and hypervolume contribution index. The external archive is a set of non-dominant location and capacity schemes obtained from the population during the evolution process.

[0057] Among them, the location variables include the VVR's location decision vector. Access node index decision vector of EER port Capacity variables include the capacity decision vector of the VVR. Rated configuration capacity of EER port .

[0058] Because it is necessary to simultaneously determine the addressing of VVR and EER (which are 0 / 1 discrete variables) and their corresponding capacities (which are continuous variables), and because the two are coupled and require co-optimization, this actually forms a kind of "discrete addressing". This invention addresses the hybrid decision space of "discrete addressing - continuous capacity determination." It employs a quantum two-layer chromosome encoding mechanism to implicitly map discrete addressing logic into a continuous quantum phase angle space, improving the search efficiency of large-scale combinatorial optimization. In the search operator design, a quantum rotation gate is introduced to drive the addressing decision to adaptively rotate towards the Pareto elite space, coupled with a heavy-tailed power-law perturbation of Lévy flight to drive capacity exploration. This leverages the characteristics of short-step-frequency exploration and long-step-frequency even-jumps to fundamentally overcome the local convergence limitation of Gaussian step sizes. The specific implementation of step 3 is described in detail below.

[0059] (a) Encoding method The addressing variables are encoded using quantum phase angles, including mapping the binary decision of whether to install a VVR or EER at each candidate installation location to the phase angle of a qubit (i.e., addressing encoding). The squared probability amplitude value corresponding to this phase angle represents the probability of installing the device at that location. The capacity variables are encoded using real numbers (i.e., capacity encoding), including encoding the rated capacity of the VVR and the rated capacity of the EER port as real number variables, and the value range of each real number variable is constrained by the capacity upper limit.

[0060] Therefore, the detailed mathematical expansion of an individual is defined as follows: ; In the formula, For the first The decision vector of each individual; for The phase angle of the qubits of the VVR to be installed on the candidate branch to be installed, with values ​​ranging from 1 to 2. . for The phase angle of the qubits on the candidate nodes for installing EER; For the above The undetermined rated capacity of the VVR on each branch is the real number. a side road ; For the above The undetermined rated capacity of the EER at each node (real number), corresponding to this Nodes .

[0061] Based on this, in each generation of population evolution, the aforementioned phase angles need to be encoded and decoded into actual installation decisions using quantum measurement operators for use in power flow calculations. The quantum measurement operators will convert continuous phase angles... Implicit mapping to discrete installation decision variables : ; In the formula, For the first The final discrete installation decision for each qubit (1 for installation, 0 for no installation); The state of this qubit The square of the probability amplitude, i.e., the installation probability; It is a unit step function; This is the probability threshold for determining installation; for example, it can be a fixed value of 0.5.

[0062] Building upon the aforementioned encoding method, this invention employs a three-subpopulation co-evolutionary population organization architecture. The population is divided into three subpopulations, each undertaking a different search task: the first subpopulation focuses on global exploration for site selection decisions, the second subpopulation focuses on local development of capacity allocation, and the third subpopulation balances both to maintain search equilibrium. Each subpopulation independently executes the quantum rotation gate and Lévy flight update operations and periodically exchanges external archive information to improve overall search efficiency. Here, the three-subpopulation co-evolutionary architecture and the external archive maintenance strategy based on the Hypervolume Contribution Index (HVI) achieve a balance between economic efficiency and security in search width and depth, ensuring optimal coverage and uniformity of the Pareto front solution set.

[0063] (ii) Location variable update The addressing variable is updated via a quantum rotation gate. The rotation step size of the quantum rotation gate is determined as follows: For each addressing and sizing scheme in the population, the difference in weighted voltage deviation between it and the reference addressing and sizing scheme selected from the external archive is normalized by dividing by the difference between the maximum and minimum weighted voltage deviations in the current generation of the population, and then multiplied by the base step size coefficient after being mapped by the sigmoid function to obtain the rotation step size.

[0064] Based on this, quantum rotation gates are used for the qubit level. Drive phase angle to Pareto elite solution Evolution. Updated phase angle Defined as: ; In the formula, Let this be the current evolutionary iteration number; For the first elite individuals randomly selected from external archives 3D phase angle; This is a sign function used to determine the shortest path direction for phase angle rotation; The quantum rotation step size is the amplitude.

[0065] Among them, the quantum rotation step size amplitude Fusion Difference information: ; In the formula, The preset basic quantum rotation step size scaling factor; This is a fitness difference sensitivity factor used to adjust search pressure. This represents the weighted voltage deviation corresponding to the current individual. The weighted voltage deviation corresponding to the selected elite individuals; This represents the observed maximum weighted voltage deviation in the current generation population; It represents the minimum weighted voltage deviation in the current generation population.

[0066] (iii) Capacity coding update The capacity encoding is updated via a heavy-tailed random walk. Specifically, for real-valued capacity variables, a Lévy-flying heavy-tailed random walk operator is introduced. The Lévy step size... Generated using the Mantegna algorithm, its elements The distribution is defined as: ; In the formula, The generated Lévy random walk step size component; u , v The random number variable follows a normal distribution; It is a normal distribution function; The Lévy exponent controls the thickness of the tail of the probability density function; for example, it can be set to 1.5. Normal distribution The standard deviation is determined by the Gamma function. and index The constants determined jointly satisfy the following equation: ; After obtaining the Lévy step size, perform a capacity update operation with elite guidance: ; In the formula, For the updated capacity variable of the individual; For the number of generations of evolution The reduced adaptive coefficient controls the shift of the search scope from global to local. For element-wise multiplication operators; Let be the direction vector of an individual converging towards the elite solution, where For the capacity variable of the elite, For the capacity variable of an individual.

[0067] (iv) External Archives Maintenance The external archive is maintained using Pareto ordering and a hypervolume contribution index, including: employing Pareto non-dominated ordering to stratify the site selection and sizing schemes in the population based on two objectives: annualized investment cost and weighted voltage deviation, and storing non-dominated schemes in the external archive; managing the size of the external archive using a hypervolume contribution index: for each site selection and sizing scheme in the external archive, calculating the hyperrectangular volume enclosed by it and a reference point in the target space defined by annualized investment cost and weighted voltage deviation, and using this hyperrectangular volume as the contribution of the site selection and sizing scheme; when the number of site selection and sizing schemes in the external archive exceeds a preset upper limit, deleting the site selection and sizing scheme with the smallest contribution; wherein, the reference point corresponds to the largest annualized investment cost and the largest weighted voltage deviation.

[0068] The Hypervolume Contribution Index (HVI) (also known as hyperrectangular volume) is used to measure external elite archives. Its overall performance. (Archived) The hypervolume contribution index is defined as: ; In the formula, To solve The contribution value (over-volume contribution index) of a non-dominated site selection and occupancy scheme to the overall coverage capacity of external archives. Let Lebesgue measure, and let be a rectangular surface in the biobject space; The current set of external elite archives storing non-dominated solutions; For any non-dominated solution individual in the external archive; To solve The corresponding objective function value vector ; For the solution in the target space Indicator Vectors and Reference Points The resulting hyperrectangle. When the external file size overflows, the solution with the smallest HVI value is recursively deleted, thus theoretically guaranteeing optimal coverage of the Pareto front by the solution set.

[0069] Step 4: Iterate through Step 3 until convergence, then select the optimal addressing and occupancy scheme from the current external archives and output it.

[0070] Here, there are several ways to determine whether the convergence condition is met: First, within the preset maximum number of generations, the non-dominated solution set in the external archive has not been updated, meaning no new addressing and sizing schemes have been accepted, and no existing schemes have been replaced, indicating that the Pareto front has stabilized. In this case, the convergence of the external archive can be used as the basis for determining the convergence of the algorithm. Second, the difference between the maximum and minimum values ​​of the two objective function values—annualized investment cost and weighted voltage deviation—for all individuals in the population is less than the preset tolerance, or the increment of the hypervolume index of the Pareto front between adjacent generations is less than the preset threshold, indicating that the search in the target space has made no significant progress. In this case, the magnitude of the change in the objective function value can be used to determine whether the algorithm has converged. Third, the number of iterations reaches the preset maximum number of generations, at which point the iteration is forcibly terminated. In actual implementation, the preset parameters of each criterion (such as the stagnation generation window, tolerance threshold, and maximum number of generations) can be flexibly configured according to the balance requirements of solution accuracy and computational resources. The algorithm can be considered to have converged when any of the judgment conditions are met. When the algorithm is determined to be converged, the iteration is stopped, and the optimal addressing and sizing scheme is selected from the current external archive for output.

[0071] Specifically, the optimal addressing and capacity grading scheme is selected from the current external archives and output, including: normalizing the annualized investment cost and weighted voltage deviation of each addressing and capacity grading scheme in the current external archives, calculating the Euclidean distance from each addressing and capacity grading scheme to the ideal point, and selecting the addressing and capacity grading scheme with the smallest distance as the optimal addressing and capacity grading scheme output; the annualized investment cost of the ideal point is zero and the weighted voltage deviation is zero.

[0072] After completing the evolutionary iterations in step 3, the algorithm outputs the non-dominated solution set of the final population, denoted as PF. PF contains several non-dominated site selection and sizing schemes, each corresponding to a set of objective function value vectors calculated by the upper-level model in step 2. The normalized Euclidean distance method is used to calculate each solution in PF. Distance to the ideal target point. Ideal point. This represents the theoretically optimal state where the annualized cost is zero and the weighted voltage deviation is zero. The normalized Euclidean distance formula is: ; In the formula, For the first Normalized Euclidean distances corresponding to non-dominated solutions; This is the index of the non-dominant solution in the set PF; For the first The annualized investment cost corresponding to each solution; It represents the minimum annualized investment cost of all solutions in the current non-dominated solution set PF; It represents the maximum annualized investment cost of all solutions in the current non-dominated solution set PF; For the first The weighted voltage deviation corresponding to each solution It is the minimum weighted voltage deviation of all solutions in the current non-dominated solution set PF; It is the maximum value of the weighted voltage deviation of all solutions in the current non-dominated solution set PF.

[0073] The operation of selecting the location and volume determination scheme with the shortest distance can be expressed by the formula: ; In the formula, Output the optimal addressing and occupancy scheme.

[0074] For example, Figure 6 The algorithm flow of this invention is illustrated. First, system parameters and historical operating data are input to construct a comprehensive scenario set. Then, population initialization is performed using quantum two-layer chromosome encoding (location phase angle + real capacity), and the hypervolume contribution index (HVI) is calculated. Upon entering the main loop, based on the three-subpopulation co-evolution framework, each subpopulation performs updates in parallel: the location encoding is updated through a quantum rotation gate to optimize device location, and the capacity encoding is updated through Lévy flight to optimize device capacity. For each individual, its location and capacity scheme is fixed, and the lower-level model is called to perform parallel power flow simulations for all typical scenarios (each scenario is solved independently using Gurobi). After obtaining the voltage deviation for each scenario, the annualized investment cost and weighted voltage deviation are calculated. Subsequently, Pareto non-dominated sorting and HVI archive maintenance are performed to complete one generation of evolution. The algorithm checks whether the convergence condition is met. If not, it returns to the main loop to continue iteration; if the convergence condition is met, the final non-dominated solution set is extracted from the external archive, the optimal solution is selected using normalized Euclidean distance, and the optimal location and capacity scheme is output, ending the algorithm.

[0075] In summary, this invention, through a combined mechanism of clustering during the scenario construction phase and forced addition of extreme scenarios, enables the planning model to simultaneously cover both normal operating scenarios and long-tail extreme conditions, effectively solving the problem of existing solutions lacking coverage of extreme scenarios and thus failing to guarantee disaster recovery capabilities. Addressing the shortcomings of energy storage solutions that rely on prediction and are prone to integral saturation, this invention employs joint planning of VVR and EER. Both are instantaneous response devices and do not involve energy time-scale integration, thus physically avoiding SOC integral saturation failure caused by the accumulation of prediction errors. Addressing the problem of discrete regulation and dead zones in traditional voltage regulators, this invention utilizes the continuous regulation capability of EER to fill the tap changer dead zone of VVR, achieving synergistic complementarity between discrete coarse adjustment and continuous fine adjustment. Addressing the limitation of energy router capacity and its inability to handle network-wide voltage regulation, this invention utilizes the high-throughput electromagnetic voltage regulation capability of VVR to undertake the main voltage support, significantly reducing the power throughput requirements of EER and avoiding equipment overload risks. At the solution level, this invention designs a quantum encoding and co-evolutionary mechanism for the hybrid decision space of "discrete location-continuous capacity determination," overcoming the problems of traditional algorithms easily getting trapped in local optima and premature convergence. Simultaneously, by maintaining the external archive through a hypervolume contribution index, it ensures the optimal balance between economic efficiency and safety under extreme operating conditions in the final solution. The combined effect of these technologies allows this invention to leverage both the high-throughput support capability of VVR and the fine-tuning characteristics of EER without relying on high-precision prediction, achieving organic synergy between the two and effectively realizing voltage self-healing of the active distribution network under extremely uncertain source load impacts.

[0076] Compared with the prior art, the present invention has the following beneficial effects: In terms of economy and investment efficiency, this invention, through the joint planning of VVR and EER, utilizes the large-capacity electromagnetic voltage regulation capability of VVR to handle the main voltage throughput demand, significantly reducing the capacity configuration pressure of expensive power electronic EER equipment. Compared with traditional ESS solutions, this planning method achieves the same or even better voltage regulation effect while significantly reducing the total investment cost, solving the problem that energy storage systems are difficult to popularize in large-scale distribution networks due to excessively high energy costs.

[0077] Regarding operational and predictive robustness, this invention utilizes an instantaneous response device instead of a prediction-dependent device. Addressing the vulnerability of energy storage systems to SOC integral saturation and control failure under extreme conditions due to the accumulation of prediction errors, the VVR and EER combination does not involve time-scale energy integration, enabling real-time responses to instantaneous voltage deviations. In extreme weather scenarios such as surges in photovoltaic output and sudden load changes, this invention can maintain a high voltage qualification rate across the entire grid.

[0078] Regarding the comprehensiveness of the planning scheme, this invention employs a combined mechanism of K-medoids clustering and mandatory addition on extreme days during the scenario construction phase. By introducing extreme indicator filters such as load factor and photovoltaic volatility, it ensures that the planning model not only considers the economic efficiency of normal operation throughout the year but also covers extreme operating conditions in the long-tail distribution. This gives the generated site selection and capacity determination scheme strong disaster recovery attributes, enhancing the resilience of the active distribution network in responding to extreme uncertainties.

[0079] In terms of algorithm search efficiency and practicality, this invention unifies discrete location decisions into a continuous quantum phase space through a quantum two-layer chromosome encoding mechanism, optimizing the search efficiency of the decision space for large-scale power distribution network planning from the source of encoding. Utilizing the adaptive phase guidance of the quantum rotating gate and the heavy-tailed power-law jump characteristics of Lévy flight, a dynamic balance is achieved between "refined local directional development" and "wide-area global leap-based exploration" in the search process, fundamentally overcoming the shortcomings of easily getting trapped in local optima and premature convergence in complex "discrete-continuous" mixed-integer nonlinear programming problems. Furthermore, by introducing an external archive maintenance strategy based on the hypervolume contribution index (HVI), the final Pareto front solution set is mathematically guaranteed to maximize the uniform coverage and distribution breadth of the target space, providing decision-makers with a more scientific, complete, and stable optimal compromise configuration scheme.

[0080] Furthermore, in practical implementation, the voltage regulator (VVR) and energy router (EER) in this invention can be replaced with equivalent components depending on the specific application scenario and equipment selection requirements. For example, the VVR can be replaced with a power transformer with on-load tap changing function, which can also achieve discrete voltage regulation through tap switching; the EER can be replaced with an intelligent soft switch, which also has multi-port power routing capability and power exchange between ports.

[0081] The voltage regulator and energy router joint planning method considering extreme operating conditions provided by this invention can be applied to electronic devices. Specifically, the electronic device can be a desktop computer, a portable computer, a smart mobile terminal, a server, etc. In practical applications, the electronic device can be a planning workstation in a power distribution network dispatch center, a server hosting a power system simulation platform, or a cloud computing device running power distribution network planning software, etc.

[0082] Based on the same inventive concept, the present invention also provides an electronic device, such as... Figure 7 As shown, it includes a processor 601, a communication interface 602, a memory 603, and a communication bus 604, wherein the processor 601, the communication interface 602, and the memory 603 communicate with each other through the communication bus 604. Memory 603 is used to store computer programs; When the processor 601 executes the program stored in the memory 603, it implements the steps of the above-described and extreme operating condition joint planning method for voltage regulator and energy router.

[0083] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus.

[0084] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0085] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0086] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0087] It should be noted that, for the electronic device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and relevant parts can be referred to in the description of the method embodiments.

[0088] It should be noted that the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the invention described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention.

[0089] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0090] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings and the disclosure in carrying out the claimed invention. In the description of the invention, the word "comprising" does not exclude other components or steps, "a" or "an" does not exclude a plurality, and "a plurality" means two or more, unless otherwise explicitly specified. Furthermore, while different embodiments may describe certain measures, this does not mean that these measures cannot be combined to produce good results.

[0091] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A joint planning method for voltage regulators and energy routers considering extreme operating conditions, characterized in that, Includes the following steps: Step 1: Obtain historical operation data of the active distribution network, extract regular operation scenarios from the historical operation data and add extreme boundary scenarios to construct a comprehensive scenario set containing multiple types of scenarios; Step 2: Construct a two-layer optimization planning model consisting of an upper-layer model and a lower-layer model. The upper-layer model has two objectives: annualized investment cost and weighted voltage deviation for all scenarios. The location and capacity setting schemes of the voltage regulator (VVR) and energy router (EER) are used as decision variables. The weighted voltage deviation is the weighted sum of voltage deviations for all scenarios. The voltage deviation is the deviation between the per-unit voltage value of a node in the active distribution network and the rated per-unit voltage value. Under the location and capacity setting scheme given by the upper-layer model, the lower-layer model aims to minimize the voltage deviation of a single scenario. The tap position of the VVR and the power of each port of the EER are used as decision variables. The voltage trajectory of each node is obtained through power flow simulation calculation, and the voltage deviation is calculated and returned to the upper-layer model. Step 3: Solve the bi-level optimization programming model using a quantum-encoded multi-objective evolutionary algorithm. In the algorithm, an individual is a location-capacity scheme, and the population is a set of individuals. Each individual includes a location variable and a capacity variable. The location variable uses quantum phase angle encoding, and the capacity variable uses real number encoding. In each generation of population evolution, the location encoding is updated through a quantum rotation gate, the capacity encoding is updated through a heavy-tailed random walk, and the external archive is maintained using Pareto sorting and the hypervolume contribution index. The external archive is a set of non-dominant location-capacity schemes obtained from the population during the evolutionary process. Step 4: Iterate through Step 3 until convergence, then select the optimal addressing and occupancy scheme from the current external archives and output it.

2. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The process involves extracting regular operating scenarios from the historical operating data and adding extreme boundary scenarios to construct a comprehensive scenario set containing multiple types of scenarios, including: Regular operating scenarios are extracted from the historical operating data using the K-medoids clustering algorithm; Based on the load extreme index and photovoltaic fluctuation index in the historical operating data, extreme boundary scenarios are extracted from the historical operating data; the load extreme index is the ratio of the maximum daily total load to the average daily load, and the photovoltaic fluctuation index is the maximum daily photovoltaic power fluctuation rate; A comprehensive scenario set is constructed based on the normal operating scenarios and the extreme boundary scenarios.

3. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The rotation step size of the quantum rotating gate is determined as follows: For each location and capacity scheme in the population, the difference in weighted voltage deviation between it and the reference location and capacity scheme selected from the external archive is normalized by dividing by the difference between the maximum and minimum weighted voltage deviations in the current generation population, and then multiplied by the base step size coefficient after being mapped by the sigmoid function to obtain the rotation step size.

4. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The location variable is encoded using quantum phase angle, which includes mapping the binary decision of whether to install a VVR or EER at each candidate installation location to the phase angle of a qubit, and the squared probability value corresponding to the phase angle represents the probability of installing the device at that location. The capacity variables are encoded as real numbers, including encoding the rated capacity of the VVR and the rated capacity of the EER port as real number variables, and the value range of each real number variable is constrained by the upper limit of the capacity.

5. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The method of maintaining external archives using Pareto ordering and hypervolume contribution metrics includes: Pareto non-dominated sorting was used to stratify the site selection and gradation schemes in the population based on two objectives: annualized investment cost and weighted voltage deviation. The schemes that do not dominate each other were stored in an external archive. The external archive size is managed using a super-volume contribution index: For each site selection and capacity determination scheme in the external archive, the super-rectangular volume enclosed by the reference point in the target space formed by the annualized investment cost and weighted voltage deviation is calculated, and the super-rectangular volume is used as the contribution of the site selection and capacity determination scheme; when the number of site selection and capacity determination schemes in the external archive exceeds a preset upper limit, the site selection and capacity determination scheme with the smallest contribution is deleted; wherein, the reference point corresponds to the largest annualized investment cost and the largest weighted voltage deviation.

6. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The step of selecting the optimal addressing and capacity grading scheme from the current external archives includes: normalizing the annualized investment cost and weighted voltage deviation of each addressing and capacity grading scheme in the current external archives, calculating the Euclidean distance from each addressing and capacity grading scheme to the ideal point, and selecting the addressing and capacity grading scheme with the smallest distance as the optimal addressing and capacity grading scheme output; the annualized investment cost of the ideal point is zero and the weighted voltage deviation is zero.

7. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The annualized investment cost is expressed as follows: ； In the formula, This represents the annualized investment cost. This is a pre-defined set of VVR candidate installation branches; For branch index; Cost per unit capacity of VVR; branch road The rated capacity of the VVR on the device; Cost per unit capacity of parallel capacitor banks in a VVR; branch road The capacitor configuration capacity is used in conjunction with the VVR; This is a pre-defined set of potential EER access nodes. For EER port index; Cost per unit capacity of EER ports; For the first Rated configuration capacity of each EER port; This is the capital recovery coefficient.

8. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 1, characterized in that, The weighted voltage deviation is expressed as: ； In the formula, Weighted voltage deviation for all scenarios; The total number of scenes in the comprehensive scene set; For scene indexing; For the scene The weights; This represents the total number of nodes. For node index; the node is the electrical connection point in the active distribution network, the VVR is installed between the nodes at both ends of the branch to regulate the node voltage, and the EER is connected to the node to inject or absorb power into the node; This represents the total number of scheduling periods within a single scenario. Indexed by time period; In the scene Next, node exist The voltage amplitude per unit value at each moment, wherein the voltage trajectory is composed of the voltage amplitude per unit value of each node at each moment; This is the voltage rating reference value.

9. The joint planning method for voltage regulators and energy routers considering extreme operating conditions according to claim 8, characterized in that, The objective function of the lower-level model is: ； In the formula, For the scene Voltage deviation; numerical value This is the per-unit value of the rated voltage; This represents the active power balance constraint of a node, which is used to constrain the sum of active power injected into each node in each time period to be equal to the sum of active power flow out of that node. This represents the reactive power balance constraint at the node, which is used to ensure that the sum of reactive power injected into each node in each time period is equal to the sum of reactive power flowing out of that node. This represents a voltage state transition constraint based on linearized DistFlow, used to constrain the voltage squared difference between the nodes at both ends of a branch to be determined by the active and reactive power flow and impedance of the branch. This indicates the VVR tap position constraint, used to constrain the tap position to be a discrete integer, and the adjustment range of the tap position in adjacent time periods does not exceed the maximum value; This indicates the EER port power constraint, which is used to constrain the input power and output power of each EER port to be limited by the rated capacity, and to ensure that the same EER port cannot simultaneously input and output power, and to constrain the sum of the output power of all EER ports to be equal to the sum of the input power multiplied by the conversion efficiency; This indicates node voltage safety constraints, used to ensure that the per-unit value of the voltage amplitude at each node must remain within the allowable range.

10. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a computer program stored in memory, it implements the steps of the joint planning method for voltage regulators and energy routers taking into account extreme operating conditions as described in any one of claims 1-9.

Images