Multi-terminal optimal coordination voltage regulation method for distribution network with fusion of power interval prediction

Abstract

The application discloses a power distribution network multi-terminal optimization collaborative voltage regulation method fusing power interval prediction, and belongs to the technical field of power system operation and control. The method comprises the following steps: a convolutional neural network model is built to predict photovoltaic output and load power of the power distribution network in a future period; power flow calculation is carried out based on power prediction, key buses are identified according to the maximum value of the estimated voltage of each bus in an interval, and the required voltage regulation amount of the key buses is calculated to determine the voltage regulation target; based on dynamic voltage sensitivity analysis, the reactive power compensation of photovoltaic inverters and the active power compensation of battery energy storage systems are coordinated in an optimal proportion in sequence to minimize power compensation, and finally, interval regulation of all bus voltages is realized. The application can effectively suppress voltage over-limit in a high photovoltaic penetration power distribution network, significantly reduce the demand for reactive power and active power compensation, enhance the safety of system operation, and prolong the service life of equipment.

Description

TECHNICAL FIELD

[0001] The application relates to a power interval prediction-fused multi-terminal optimal collaborative voltage regulation method for a power distribution network, and belongs to the technical field of power system operation and control. BACKGROUND

[0002] In recent years, global energy shortage and environmental deterioration have become increasingly prominent, accelerating the large-scale application of renewable energy represented by photovoltaics in power distribution networks. However, the high proportion of photovoltaic access leads to bidirectional power flow in the power distribution network, causing voltage out-of-limit problems, which seriously restricts the safe operation of the power grid and the photovoltaic consumption capacity. Especially in low-voltage power distribution networks, photovoltaic output and load fluctuations jointly act on the system, easily causing voltage to be too high or too low, exceeding the voltage range allowed by ANSI C84.1-2020 standard.

[0003] Traditional voltage regulation methods, such as load tap changers and step-type voltage regulators, are widely used but are difficult to respond to rapid voltage fluctuations due to mechanical action delay. Therefore, reactive and active power compensation based on power electronic devices has gradually become a research hotspot. Photovoltaic inverters can provide reactive power support, but in the case of serious overvoltage, they are easy to reach the current constraint limit, resulting in saturated compensation capacity. Although existing research has proposed methods based on model predictive control or multi-agent collaboration to introduce energy storage systems for active power compensation, forming a hybrid compensation strategy. However, existing research rarely dynamically adjusts the compensation strategy according to the real-time sensitivity of the system, and rarely effectively manages the state of charge of the energy storage, resulting in frequent action of the energy storage and shortening of the battery life. The application proposes a power interval prediction-fused multi-terminal optimal collaborative voltage regulation method for a power distribution network. SUMMARY

[0004] The purpose of the application is to overcome the deficiencies in the prior art and provide a power interval prediction-fused multi-terminal optimal collaborative voltage regulation method for a power distribution network. The improved convolutional neural network fusion prediction model is used to realize power prediction, combined with dynamic sensitivity analysis and proportional collaborative power compensation strategy, and the photovoltaic inverter reactive power compensation and energy storage system active power compensation are called in turn to realize voltage regulation within a certain interval, which significantly reduces the power compensation demand while ensuring voltage safety.

[0005] To achieve the above purpose, the application adopts the following technical scheme:

[0006] The application provides a power interval prediction-fused multi-terminal optimal collaborative voltage regulation method for a power distribution network, which comprises the following steps:

[0007] Step 1: Build a convolutional neural network model to predict the photovoltaic output and load power of the power distribution network in the future period;

[0008] Step 2: Perform power flow calculation based on power prediction, identify critical buses based on the estimated maximum voltage of each bus within the interval, calculate the required voltage regulation amount, and determine the voltage regulation target;

[0009] Step 3: Based on dynamic voltage sensitivity analysis, with the goal of minimizing power compensation, the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system are coordinated in sequence according to the optimal ratio, so as to achieve range regulation of all bus voltages.

[0010] In step 1, the constructed convolutional neural network model includes a convolutional neural network, an attention-based long short-term memory network, and a particle swarm optimization algorithm. The convolutional neural network is used to extract spatial features of power data from multiple nodes in the distribution network; the attention-based long short-term memory network is used to capture long-term dependencies in the power time series; and the particle swarm optimization algorithm is used to globally optimize the network hyperparameters to minimize the prediction error. The power prediction output of the model is represented as follows:

[0011] (1);

[0012] in, , This is the weight matrix. , For bias terms, , The context feature vector output after the attention mechanism. , For the predicted photovoltaic and load power.

[0013] Step 2 includes the following steps:

[0014] Step 2.1: Identification of critical busbars and calculation of voltage regulation, including:

[0015] Based on the predicted voltage results, determine the extreme voltage values ​​of each bus during the control period, including:

[0016] (2);

[0017] in, It is a function that calculates voltage based on power prediction. It is a busbar The per-unit value of the real-time voltage is estimated based on photovoltaic and load power prediction. It is the first Adjustment time interval The maximum value of the internal voltage.

[0018] Step 2.2: Identify key buses based on voltage extreme values ​​and calculate regulation amounts, including:

[0019] (3);

[0020] in, It is a key busbar exist Voltage regulation range during the period It is the allowable voltage deviation ratio. When the voltage deviation of the critical bus exceeds the allowable range, the photovoltaic inverter is used first to perform reactive power compensation. If the reactive power compensation is insufficient to eliminate the voltage violation, energy storage is used to further perform active power compensation.

[0021] Step 2.3: Voltage regulation objectives include:

[0022] (4);

[0023] The constraints are:

[0024] (5);

[0025] in, , It is a busbar exist Actual reactive and active power compensation during the period, , It is a busbar exist The reactive and active power compensation capacity available during this period, , It is a busbar The voltage at During the period of busbar The sensitivity of reactive and active power compensation. , It is a busbar exist Local photovoltaic inverters used for reactive power compensation and energy storage used for active power compensation during this period. Control coefficients within.

[0026] Step 3, minimizing power compensation, includes minimizing reactive and active power compensation:

[0027] Step 3.1: The proportional allocation of reactive power compensation is achieved by solving the following optimization problem:

[0028] (6);

[0029] in, It is an optimization objective function used to minimize the total reactive power compensation demand. It is a busbar Actual reactive power compensation These are constraints. Using the Lagrangian function related to reactive power compensation, the above objective function and constraints are transformed into...

[0030] (7);

[0031] in, , It concerns the Lagrange function and multiplier of reactive power.

[0032] Solving the Lagrange equation yields the optimal reactive power compensation requirement:

[0033] (8);

[0034] all The voltage of the busbar is During the period of busbar The sensitivity ratio of reactive power compensation remains unchanged:

[0035] (9);

[0036] The optimal reactive power compensation is obtained as follows:

[0037] (10);

[0038] Step 3.2: When reactive power compensation is insufficient, active power compensation is initiated. Its allocation must satisfy the state of charge constraints of the energy storage system, including:

[0039] (11);

[0040] in, It is a busbar exist The state of charge of the local energy storage system during the period, , yes The maximum and minimum limits are respectively and .

[0041] The impact of active power compensation on the state of charge (SOC) changes of the bus energy storage system includes:

[0042] (12);

[0043] The constraints are:

[0044] (13);

[0045] in, It is a busbar The capacity of local energy storage, , These are the busbars The state of charge of the energy storage is Actual changes and maximum permissible changes during the period

[0046] The maximum active power compensation that each energy storage unit can provide cannot exceed its remaining controllable charging and discharging power. The voltage regulation target for active power compensation using energy storage is:

[0047] (14);

[0048] The constraints are:

[0049] (15);

[0050] Among them, any busbar The available active power compensation capacity of local energy storage is:

[0051] (16);

[0052] Determine the control objective function for active power compensation:

[0053] (17);

[0054] The constraints are

[0055] (18);

[0056] in, It is the objective function for optimization, used to minimize the total active power compensation demand; These are constraints.

[0057] Transform the objective function and constraints using the Lagrange function:

[0058] (19);

[0059] in, , It relates to the Lagrange function and multiplier of reactive power.

[0060] Solving the Lagrange equation yields the optimal total active power compensation requirement:

[0061] (20);

[0062] all The voltage of the busbar is During the period of busbar The sensitivity ratio of active power compensation remains unchanged:

[0063] (twenty one);

[0064] The constraints are:

[0065] (twenty two);

[0066] The optimal reactive power compensation is obtained as follows:

[0067] (twenty three).

[0068] In step 4, the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system are coordinated sequentially according to the optimal ratio, ultimately achieving range regulation of all bus voltages, including:

[0069] Step 4.1: When the photovoltaic inverters on the bus are sufficient to provide the required reactive power compensation, the voltage regulation requirements of the critical bus can be met. It will then be updated to 0.

[0070] (twenty four);

[0071] in, It is based on (10) in Updated after using proportional reactive power compensation during the period ,

[0072] Step 4.2: When reactive power compensation alone is insufficient to... When the voltage is reduced to 0 and the voltage violation is completely eliminated, any bus The reactive power compensation performed will all reach their respective maximum values. The key bus voltage based on reactive power compensation is updated as follows:

[0073] (25);

[0074] Step 4.3: Further utilize the active power compensation of energy storage to eliminate the remaining voltage violation, considering that there is sufficient available energy storage capacity to regulate the bus. The voltage at any bus Using active power compensation below capacity can eliminate critical busbars. The residual voltage violation, the critical bus voltage update, and the required active power compensation are...

[0075] (26);

[0076] in, It is based on (23)(24) in Further updates were made using proportional active power compensation during the period. .

[0077] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is capable of executing the program to implement the aforementioned method for multi-terminal optimal coordinated voltage regulation of a distribution network based on fused power range prediction.

[0078] A computer-readable storage medium storing computer instructions that, when executed by a processor, implement the aforementioned method for multi-terminal optimal coordinated voltage regulation of a distribution network based on fused power range prediction.

[0079] Compared with the prior art, the advantages of the present invention are as follows:

[0080] 1. At the prediction level, this invention constructs a hybrid prediction model that integrates convolutional neural networks, attention mechanism long short-term memory networks, and particle swarm optimization algorithms, thereby achieving high-precision range prediction of photovoltaic output and load power in distribution networks and improving the foresight and robustness of subsequent voltage regulation.

[0081] Existing technologies mostly use fully connected neural networks for deterministic point prediction, which cannot quantify the uncertainty of power fluctuations, resulting in a lack of ability of regulation strategies to cope with sudden fluctuations. The present invention outputs power range prediction results, which provides a basis for subsequent safety assessment and regulation calculation based on voltage extreme values ​​(as shown in formula (2)), predicts the voltage regulation demand within a certain period of time, and enhances the safety of the entire voltage regulation system.

[0082] 2. At the level of voltage safety assessment and target setting, this invention performs forward-looking power flow calculations based on power range prediction, which enables accurate identification of key buses and precise calculation of voltage regulation.

[0083] Traditional methods often rely on real-time monitoring or decision models based on historical data, resulting in delayed responses and difficulty in handling rapid fluctuations in photovoltaic and load power. In contrast, this invention calculates the voltage extreme values ​​of each bus in the future regulation period using power range prediction results (Equation (12)) and dynamically identifies key buses with voltage deviations exceeding the allowable range and their required regulation amounts based on Equation (13). In real-world scenarios, such as the Australian real low-voltage network used in this study, bus 10 at the end of the feeder experiences the most severe voltage deviations in multiple periods. By identifying it as a key bus and using its maximum voltage deviation as the overall regulation target, the system only needs to compensate for this node to simultaneously eliminate voltage exceedance issues of other buses. While ensuring overall voltage safety, this significantly reduces the frequency of operation of regulation equipment, improving the overall control efficiency and economy.

[0084] 3. This method makes the setting of voltage regulation target (Equation (14)) more scientific and accurate, laying a solid foundation for subsequent optimization and compensation, and effectively avoiding under-regulation or over-regulation caused by inaccurate identification. It is a key link in realizing true preventive voltage control. At the level of optimization and compensation, this invention creatively introduces dynamic voltage sensitivity analysis and constructs an optimization model with the goal of minimizing the total compensation demand. It proposes a collaborative compensation mechanism based on the dynamic sensitivity proportional coefficient, realizes the optimal ratio allocation of reactive and active power, significantly improves compensation efficiency and extends equipment life.

[0085] Existing technologies either use equal allocation or fail to fully consider the differences in real-time system operating states when allocating compensation amounts, resulting in low compensation efficiency and frequent equipment operation. Compared with traditional equal allocation or fixed ratio allocation, this invention constructs an optimization objective function as shown in formulas (6) and (17), and solves it using the Lagrangian function (formulas (7)-(10) and (19)-(23)). This ensures that under any operating state, the compensation amounts of reactive power and active power are allocated according to the dynamic ratio that has the highest efficiency in affecting the critical bus voltage, ensuring that each unit of reactive / active power is used for the most effective voltage regulation. Simulation results (see Figure 2 , Figure 3 The results show that, compared with the equal allocation strategy, this method can significantly reduce the total reactive and active power compensation demand (41.8% and 34.5%, respectively), thereby greatly reducing the operating losses of the power compensation equipment. At the same time, by effectively constraining and managing the state of charge of the energy storage system through formulas (11)-(13), the frequent operation of the energy storage is significantly reduced, which is beneficial to extending its cycle life.

[0086] 4. At the level of coordinated control, this invention adopts a sequential coordination strategy of "reactive power priority and active power supplementation" and realizes safe range adjustment of all bus voltages based on the optimal ratio. While ensuring the effect, it optimizes resource utilization to the maximum extent and reduces power compensation costs.

[0087] Existing hybrid compensation schemes often utilize multiple resources simultaneously or in an unordered manner, potentially leading to resource waste, especially for lifespan-sensitive energy storage systems. This invention strictly follows the sequential adjustment logic described in steps 4.1 to 4.3: prioritizing the full utilization of the reactive power compensation capability of the photovoltaic inverter, and only when it is insufficient to eliminate voltage violations, then utilizing the active power compensation of energy storage in an optimal proportion. This strategy ensures that all bus voltages are successfully limited within a safe range (see...). Figure 4 While effectively reducing system operating costs, it also significantly extends the cycle life of BESS by reducing the frequency of energy storage operations (Formulas (24)-(25)).

[0088] In summary, the overall solution of this invention organically integrates four core elements: high-precision power range prediction, forward-looking voltage safety assessment, dynamic sensitivity-based optimization compensation, and multi-terminal resource sequential coordination, thus forming an efficient, economical, and highly adaptable voltage control system.

[0089] This solution effectively addresses the voltage exceedance problem caused by power fluctuations in distribution networks with high photovoltaic penetration, significantly improving system operational safety and voltage quality. By minimizing power compensation requirements and optimizing equipment operation strategies, it not only achieves efficient regulation technically but also reduces system compensation costs economically and extends the service life of key power electronic equipment, generating significant technical and economic benefits. It possesses outstanding substantive features and represents a significant advancement. Attached Figure Description

[0090] Figure 1 This is a flowchart of a multi-terminal optimal coordinated voltage regulation method for distribution networks based on integrated power range prediction, provided by an embodiment of the present invention.

[0091] Figure 2 This is a comparison chart of reactive power compensation requirements provided in an embodiment of the present invention.

[0092] Figure 3 This is a comparison chart of active power compensation requirements provided in an embodiment of the present invention.

[0093] Figure 4 Schematic diagram of the initial voltage curves of each busbar.

[0094] Figure 5 A schematic diagram of the voltage curve based on reactive power compensation.

[0095] Figure 6 A schematic diagram of voltage curves based on reactive and active power compensation. Detailed Implementation

[0096] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0097] Example 1: This example introduces a multi-terminal optimized coordinated voltage regulation method for distribution networks based on integrated power range prediction, including:

[0098] A convolutional neural network model was built to predict the photovoltaic output and load power of the distribution network in future time periods;

[0099] Power flow calculations are performed based on power prediction. Critical buses are identified and their required voltage regulation is calculated based on the estimated maximum voltage of each bus within the range, thus determining the voltage regulation target.

[0100] Based on dynamic voltage sensitivity analysis, with the goal of minimizing power compensation, the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system are coordinated in the optimal proportion in sequence, so as to achieve range regulation of all bus voltages.

[0101] like Figure 1 As shown in this embodiment, the method for multi-terminal optimal coordinated voltage regulation of distribution networks based on integrated power range prediction involves the following steps in its application:

[0102] (1) Convolutional neural network model is built to predict the photovoltaic output and load power of the distribution network in the future period;

[0103] (2) Based on power prediction, perform power flow calculations, estimate the maximum voltage of each bus in the range, identify key buses and calculate the required voltage regulation;

[0104] (3) Based on dynamic voltage sensitivity analysis, with the goal of minimizing power compensation, the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system are coordinated in turn according to the optimal ratio, so as to achieve range regulation of all bus voltages.

[0105] The specific steps include:

[0106] Step 1: Build a convolutional neural network model to predict the photovoltaic output and load power of the distribution network in the future.

[0107] The optimized convolutional neural network model involved includes a convolutional neural network, an attention-based long short-term memory network, and a particle swarm optimization algorithm. The convolutional neural network is used to extract spatial features of power data from multiple nodes in the distribution network; the attention-based long short-term memory network is used to capture long-term dependencies in the power time series; and the particle swarm optimization algorithm is used to globally optimize the network hyperparameters to minimize prediction error. The power prediction output of the model is expressed as:

[0108] (1);

[0109] in, , This is the weight matrix. , For bias terms, , The context feature vector output after the attention mechanism. , For the predicted photovoltaic and load power.

[0110] Step 2: Perform power flow calculation based on power prediction, identify critical buses based on the estimated maximum voltage of each bus within the interval, calculate the required voltage regulation amount, and determine the voltage regulation target;

[0111] Step 2.1: The identification of the critical busbar and the calculation of voltage regulation include:

[0112] Based on the predicted voltage results, determine the extreme voltage values ​​of each bus during the control period, including:

[0113] (2);

[0114] in, It is a function that calculates voltage based on power prediction. It is a busbar The per-unit value of the real-time voltage is estimated based on photovoltaic and load power prediction. It is the first Adjustment time interval The maximum value of the internal voltage.

[0115] Step 2.2: Identify key buses based on voltage extreme values ​​and calculate regulation amounts, including:

[0116] (3);

[0117] in, It is a key busbar exist Voltage regulation range during the period This is the permissible voltage deviation ratio. When the critical bus voltage deviation exceeds the permissible range, reactive power compensation is first performed using photovoltaic inverters. If reactive power compensation is insufficient to eliminate the voltage violation, active power compensation is further performed using energy storage.

[0118] Step 2.3: Voltage regulation objectives include:

[0119] (4);

[0120] The constraints are:

[0121] (5);

[0122] in, , It is a busbar exist Actual reactive and active power compensation during the period, , It is a busbar exist The reactive and active power compensation capacity available during this period, , It is a busbar The voltage at During the period of busbar The sensitivity of reactive and active power compensation. , It is a busbar exist Local photovoltaic inverters used for reactive power compensation and energy storage used for active power compensation during this period. Control coefficients within.

[0123] Step 3: Minimize power compensation, including minimizing reactive and active power compensation:

[0124] Step 3.1: The proportional allocation of reactive power compensation is achieved by solving the following optimization problem:

[0125] (6);

[0126] in, It is an optimization objective function used to minimize the total reactive power compensation demand. These are constraints. Using the Lagrangian function related to reactive power compensation, the above objective function and constraints are transformed into...

[0127] (7);

[0128] in, , It relates to the Lagrange function and multiplier of reactive power.

[0129] Solving the Lagrange equation yields the optimal reactive power compensation requirement:

[0130] (8);

[0131] all The voltage of the busbar is During the period of busbar The sensitivity ratio of reactive power compensation remains unchanged:

[0132] (9);

[0133] The optimal reactive power compensation is obtained as follows:

[0134] (10);

[0135] Step 3.2: When reactive power compensation is insufficient, active power compensation is initiated. Its allocation must satisfy the state of charge constraints of the energy storage system, including:

[0136] (11);

[0137] in, It is a busbar exist The state of charge of the local energy storage system during the period, , yes The maximum and minimum limits are respectively and .

[0138] The impact of active power compensation on the state of charge (SOC) of the bus energy storage system includes:

[0139] (12);

[0140] The constraints are:

[0141] (13);

[0142] in, It is a busbar The capacity of local energy storage, , These are the busbars The state of charge of the energy storage is Actual changes and maximum permissible changes during the period.

[0143] The maximum active power compensation provided by each energy storage unit cannot exceed its remaining controllable charge / discharge power. The voltage regulation target for active power compensation using energy storage is:

[0144] (14);

[0145] The constraints are:

[0146] (15);

[0147] Among them, any busbar The available active power compensation capacity of local energy storage is:

[0148] (16);

[0149] Step 3.3: Determine the control objective function for active power compensation:

[0150] (17);

[0151] The constraints are

[0152] (18);

[0153] in, It is the objective function for optimization, used to minimize the total active power compensation demand; These are constraints.

[0154] Transform the objective function and constraints using the Lagrange function:

[0155] (19);

[0156] in, , It relates to the Lagrange function and multiplier of reactive power.

[0157] Solving the Lagrange equation yields the optimal total active power compensation requirement:

[0158] (20);

[0159] all The voltage of the busbar is During the period of busbar The sensitivity ratio of active power compensation remains unchanged:

[0160] (twenty one);

[0161] The constraints are:

[0162] (twenty two);

[0163] The optimal reactive power compensation is obtained as follows:

[0164] (twenty three);

[0165] In this embodiment, the code for the cooperative power control strategy was written and executed using the software MATLAB. To test the effectiveness of the proposed strategy, Figure 2 and Figure 3 Reactive and active power compensation demand diagrams are presented for different methods. Figure 2 It can be seen that, compared with the equally distributed reactive power compensation method, the proposed proportional compensation method can significantly reduce the reactive power compensation requirement. From... Figure 3 It can be seen that, compared with the equally distributed active power compensation method, the proposed proportional compensation method can significantly reduce the active power compensation requirement. Simulation results show that the proposed method, through a proportional coordination mechanism based on dynamic voltage sensitivity, can effectively reduce compensation requirements and extend equipment life.

[0166] Step 4: Sequentially coordinate the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system according to the optimal ratio, ultimately achieving range regulation of all bus voltages, including:

[0167] Step 4.1: When the photovoltaic inverters on the bus are sufficient to provide the required reactive power compensation, the voltage regulation requirements of the critical bus can be met. It will then be updated to 0.

[0168] (twenty four);

[0169] in, It is based on (9) in Updated after using proportional reactive power compensation during the period .

[0170] Step 4.2: When reactive power compensation alone is insufficient to... When the voltage is reduced to 0 and the voltage violation is completely eliminated, any bus The reactive power compensation performed will all reach their respective maximum values. The key bus voltage based on reactive power compensation is updated as follows:

[0171] (25);

[0172] Step 4.3: Further utilize the active power compensation of energy storage to eliminate remaining voltage violations. This is based on the consideration that there is sufficient available energy storage capacity to regulate the bus. The voltage at any bus Using active power compensation below capacity can eliminate critical busbars. The residual voltage is in violation. The update of the critical bus voltage and the required active power compensation are...

[0173] (26);

[0174] in, It is based on (23)(24) in Further updates were made using proportional active power compensation during the period. .

[0175] In this embodiment, by Figures 4-6 The voltage regulation curve results verify the effectiveness of the proposed range regulation of all bus voltages based on optimal reactive and active power compensation, where the optimal ratio is determined based on dynamic sensitivity analysis. Figures 4-6 As shown, reactive power compensation alone cannot regulate the voltage of all buses to a safe range in this scenario. Therefore, active power compensation is applied to all buses. Simulation results demonstrate that the proposed method can effectively achieve range-based regulation of the voltage of all buses, thereby solving the voltage overshoot problem.

[0176] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for multi-terminal optimal coordinated voltage regulation of distribution networks based on integrated power range prediction, characterized in that, The method includes the following steps: Step 1: Build a convolutional neural network model to predict the photovoltaic output and load power of the distribution network in the future period; Step 2: Perform power flow calculation based on power prediction, identify critical buses based on the estimated maximum voltage of each bus within the interval, calculate the required voltage regulation amount, and determine the voltage regulation target; Step 3: Based on dynamic voltage sensitivity analysis, with the goal of minimizing power compensation, the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system are coordinated in turn according to the optimal ratio, so as to achieve range regulation of all bus voltages. Step 3, minimizing power compensation, includes minimizing reactive and active power compensation: Step 3.1: The proportional allocation of reactive power compensation is achieved by solving the following optimization problem: （6）； in, It is an optimization objective function used to minimize the total reactive power compensation demand. It is a busbar Actual reactive power compensation These are constraints. Using the Lagrangian function related to reactive power compensation, the above objective function and constraints are transformed into... （7）； in, , It concerns the Lagrange function and multiplier of reactive power. Solving the Lagrange equation yields the optimal reactive power compensation requirement: （8）； all The voltage of the busbar is During the period of busbar The sensitivity ratio of reactive power compensation remains unchanged: （9）； The optimal reactive power compensation is obtained as follows: （10）； Step 3.2: When reactive power compensation is insufficient, active power compensation is initiated. Its allocation must satisfy the state of charge constraints of the energy storage system, including: （11）； in, It is a busbar exist The state of charge of the local energy storage system during the period, , yes The maximum and minimum limits are respectively and , The impact of active power compensation on the state of charge (SOC) changes of the bus energy storage system includes: （12）； The constraints are: （13）； in, It is a busbar Up to the capacity of local energy storage, , These are the busbars The state of charge of the energy storage is Actual changes and maximum permissible changes during the period The maximum active power compensation provided by each energy storage unit cannot exceed its remaining controllable charging and discharging power. The voltage regulation target for active power compensation using energy storage is: （14）； The constraints are: （15）； Among them, any busbar The available active power compensation capacity of local energy storage is: （16）； Determine the control objective function for active power compensation: （17）； The constraints are （18）； in, It is the objective function for optimization, used to minimize the total active power compensation demand; These are constraints. Transform the objective function and constraints using the Lagrange function: （19）； in, , It concerns the Lagrange function and multiplier of reactive power. Solving the Lagrange equation yields the optimal total active power compensation requirement: （20）； all The voltage of the busbar is During the period of busbar The sensitivity ratio of active power compensation remains unchanged: （21）； The constraints are: （22）； The optimal reactive power compensation is obtained as follows: （23）。 2. The method for multi-terminal optimized coordinated voltage regulation of distribution networks based on integrated power range prediction according to claim 1, characterized in that, In step 1, the constructed convolutional neural network model includes a convolutional neural network, an attention-based long short-term memory network, and a particle swarm optimization algorithm. The convolutional neural network is used to extract spatial features of power data from multiple nodes in the distribution network; the attention-based long short-term memory network is used to capture long-term dependencies in the power time series; and the particle swarm optimization algorithm is used to globally optimize the network hyperparameters to minimize the prediction error. The model's power prediction output is represented as follows: （1）； in, , This is the weight matrix. , For bias terms, , The context feature vector output after the attention mechanism. , For the predicted photovoltaic and load power.

3. The method for multi-terminal optimized coordinated voltage regulation of distribution networks based on integrated power range prediction according to claim 2, characterized in that, Step 2 includes the following steps: Step 2.1: Identification of critical busbars and calculation of voltage regulation, including: Based on the predicted voltage results, determine the voltage extreme values ​​of each bus during the control period, including: （2）； in, It is a function that calculates voltage based on power prediction. It is a busbar The per-unit value of the real-time voltage is estimated based on photovoltaic and load power prediction. It is the first Adjustment time interval The maximum value of the internal voltage. Step 2.2: Identify key buses based on voltage extreme values ​​and calculate regulation amounts, including: （3）； in, It is a key busbar exist Voltage regulation range during the period This refers to the permissible voltage deviation ratio. When the critical bus voltage deviation exceeds the permissible range, reactive power compensation is first performed using photovoltaic inverters. If reactive power compensation is insufficient to eliminate the voltage violation, energy storage is used to further perform active power compensation. Step 2.3: Voltage regulation objectives include: （4）； The constraints are: （5）； in, , It is a busbar exist Actual reactive and active power compensation during the period, , It is a busbar exist The reactive and active power compensation capacity that can be provided during this period, , It is a busbar The voltage at During the period of busbar The sensitivity of reactive and active power compensation. , It is a busbar exist Local photovoltaic inverters used for reactive power compensation and energy storage used for active power compensation during this period. Control coefficients within.

4. The method for multi-terminal optimized coordinated voltage regulation of distribution networks based on integrated power range prediction according to claim 3, characterized in that, In step 4, the reactive power compensation of the photovoltaic inverter and the active power compensation of the battery energy storage system are coordinated sequentially according to the optimal ratio, ultimately achieving range regulation of all bus voltages, including: Step 4.1: When the photovoltaic inverters on the bus are sufficient to provide the required reactive power compensation, the voltage regulation requirements of the critical bus can be met. It will then be updated to 0. （24）； in, It is based on (10) in Updated after using proportional reactive power compensation during the period , Step 4.2: When reactive power compensation alone is insufficient to... When the voltage is reduced to 0 and the voltage violation is completely eliminated, any bus The reactive power compensation performed will all reach their respective maximum values. The key bus voltage based on reactive power compensation is updated as follows: （25）； Step 4.3: Further utilize the active power compensation of energy storage to eliminate the remaining voltage violation; there is sufficient available energy storage capacity to regulate the bus. The voltage at any bus Use below-capacity active power compensation to eliminate critical busbars The residual voltage violation, the critical bus voltage update, and the required active power compensation are... （26）； in, It is based on (23)(24) in Further updates were made using proportional active power compensation during the period. .

5. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the multi-terminal optimal coordinated voltage regulation method for distribution networks based on fusion power range prediction as described in any one of claims 1 to 4.

6. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, the computer instructions implement the multi-terminal optimal coordinated voltage regulation method for distribution networks based on the fusion power range prediction as described in any one of claims 1-4.

Images