Multi-objective optimization autonomous regulation method and system for distribution network district feeder layer

Abstract

This invention provides a multi-objective optimization autonomous control method and system for the feeder layer of a distribution network. The method includes: for any control moment within the current control cycle, establishing a prediction window based on the current operating data of flexible power resources, modeling the state variables within the window, and obtaining a prediction result of the current operating state; constructing a multi-objective optimization model, using the prediction result as input to the multi-objective optimization model, and outputting the optimal control strategy within the prediction time domain; setting a control step size, and based on preset equipment-differentiated control logic, executing only the control strategy for the first time step to obtain the execution result; dynamically correcting the model parameters and control weights of the multi-objective optimization model for the current control cycle based on the execution result, and re-predicting the operating state and updating the control strategy in the next control cycle. This invention introduces a collaborative response mechanism for multiple types of load resources, achieving continuous improvement in dynamic adjustment capabilities in typical operating scenarios.

Description

Technical Field

[0001] This invention relates to the field of autonomous operation technology for distribution transformer areas / feeders, and more specifically, to a multi-objective optimization autonomous control method and system for the feeder layer of distribution transformer areas. Background Technology

[0002] With the rapid development of new energy, electric vehicles, and smart electrical equipment, the load patterns and operating characteristics at the end of the distribution network are undergoing profound changes. At the distribution substation / feeder level, a large number of flexible power resources, such as electric vehicle charging and discharging equipment and residential and commercial air conditioning, are being centrally connected, resulting in loads that are decentralized, highly volatile, and highly uncertain, posing new challenges to the operational safety, voltage quality, and economic operation of the substations.

[0003] On the one hand, new types of loads such as electric vehicles have characteristics such as adjustable power, flexible timing, and partial support for bidirectional energy interaction. Without proper guidance, they are prone to causing concentrated charging in certain periods, leading to problems such as overload of transformers or feeders and voltage exceeding limits. On the other hand, air conditioning loads, as typical comfort loads, directly affect user experience when they are in operation. When participating in regulation, they must simultaneously meet multiple restrictions such as power constraints, temperature comfort range, and start-stop frequency, which increases the complexity of regulation strategy design.

[0004] To address these issues, existing power distribution network operation and control technologies mostly employ centralized dispatching or single-objective optimization methods to manage loads uniformly or simply reduce them. However, these methods often focus on overall optimization at the system or regional level, making it difficult to fully characterize the differences in equipment type, control granularity, and user behavior among flexible resources at the end of the distribution area / feeder layer, and lacking rapid response and autonomous control capabilities for actual operating scenarios.

[0005] Meanwhile, with the gradual deployment of edge computing and smart terminals, the operation and control of distribution networks is evolving from traditional centralized dispatch to an autonomous operation mode of "local perception—autonomous decision-making—rapid execution". Against this backdrop, how to construct an autonomous control method at the transformer substation / feeder layer that can integrate multiple types of flexible power resources, balance operational safety, power quality, economy, and user comfort, and possess rolling optimization and closed-loop adaptive capabilities has become a pressing technical problem to be solved in the field of intelligent distribution network operation.

[0006] Currently, no descriptions or reports of technologies similar to this invention have been found, and no similar information has been collected domestically or internationally. Summary of the Invention

[0007] To address the aforementioned shortcomings in the prior art, this invention provides a multi-objective optimization autonomous control method and system for the feeder layer of a distribution network.

[0008] According to a first aspect of the present invention, a multi-objective optimization autonomous control method for the feeder layer of a distribution network is provided, comprising: For the current control period Δ t Any adjustment moment within t A forecasting window is established based on the current operational data of flexible power resources. t , t + T p The system models the state variables within the prediction window to obtain the prediction results of the current operating state; wherein, the flexible power resources include: electric vehicle charging and discharging resources and air conditioning load resources; A multi-objective optimization model is constructed, and the prediction results are used as input to the multi-objective optimization model to output the prediction time domain. T p The optimal control strategy within the system; Based on control step size Δ t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy was implemented, and the results were obtained. Based on the execution results, the model parameters and control weights of the multi-objective optimization model are dynamically corrected for the current control cycle, and the operating state is re-predicted and the control strategy is updated in the next control cycle, forming a multi-objective optimization autonomous control closed-loop control architecture.

[0009] According to a second aspect of the present invention, a multi-objective optimization autonomous control system for the feeder layer of a distribution network is provided, comprising: The state awareness and prediction module is used to detect and predict the current control cycle Δ t Any adjustment moment within t A forecasting window is established based on the current operational data of flexible power resources. t , t + T p The system models the state variables within the prediction window to obtain the prediction results of the current operating state; wherein, the flexible power resources include: electric vehicle charging and discharging resources and air conditioning load resources; The optimization decision module is used to construct a multi-objective optimization model, taking the prediction results as input to the multi-objective optimization model and outputting the prediction time domain. T p The optimal control strategy within the system; The strategy execution and feedback module is based on the control step size Δ. t According to the preset device differentiation control logic, only the first time step is executed. t, t +Δ t The control strategy is implemented to obtain the execution result; based on the execution result, the model parameters and control weights of the multi-objective optimization model are dynamically corrected in the current control cycle, and the operating state is re-predicted and the control strategy is updated in the next control cycle, forming a multi-objective optimization autonomous control closed-loop control architecture.

[0010] By adopting the above technical solution, the present invention has at least one of the following beneficial effects compared with the prior art: This invention targets the large number of dispersed flexible power resources at the end of the distribution network substation / feeder layer. It fully considers the differences in timing characteristics, control granularity, and execution constraints between electric vehicle charging and discharging resources and comfort loads such as air conditioning. It performs refined modeling and collaborative regulation of resources, which can support the collaborative optimization and executable regulation of multiple types of flexible resources in autonomous operation scenarios at the substation level.

[0011] This invention is a rolling optimization and fast closed-loop control mechanism for the local autonomous operation of distribution network substations / feeder layers. It is specifically designed to address the uncertainties in user behavior, comfort constraints, and differentiated execution logic of flexible loads such as electric vehicles and air conditioners, thereby meeting the actual needs of refined and autonomous regulation at the end of the distribution network.

[0012] This invention combines the operating characteristics of distribution network substations / feeder layers to construct a power grid operation constraint and typical flexible load model. In particular, it systematically characterizes engineering constraints such as electric vehicle charging and discharging constraints, air conditioning temperature comfort range, and start-stop frequency, and can be directly applied to the actual scenario of multi-objective autonomous control of distribution network substations. Attached Figure Description

[0013] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart illustrating the multi-objective optimization autonomous control method for the feeder layer of a distribution network in a preferred embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of the components of a multi-objective optimization autonomous control system for the feeder layer of a distribution network in a preferred embodiment of the present invention.

[0015] Figure 3 This is the complete execution flow of the multi-objective optimization autonomous control method for the feeder layer of a distribution network under the rolling optimization mechanism in a specific application example of the present invention. Detailed Implementation

[0016] The embodiments of the present invention are described in detail below: These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

[0017] One embodiment of the present invention provides a multi-objective optimization autonomous control method for the feeder layer of a distribution network. This method systematically proposes a mutually supportive local optimization model and autonomous control strategy that integrates electric vehicle and air conditioning loads to meet the autonomous operation requirements of distribution network areas / feeders. The model focuses on improving operational safety, power quality, economy, and user comfort. By introducing a collaborative response mechanism for multiple types of load resources, it constructs a multi-objective control system that conforms to the actual operating characteristics of the distribution area and designs a corresponding hierarchical strategy execution mechanism to achieve the orderly implementation of the strategy in typical operating scenarios and the continuous improvement of dynamic adjustment capabilities.

[0018] Specifically, such as Figure 1 As shown, the multi-objective optimization autonomous control method for the feeder layer of a distribution network provided in this embodiment may include: S1, for the current control cycle Δ t Any adjustment moment within t A forecasting window is established based on the current operational data of flexible power resources. t , t + T p The system models the state variables within the prediction window to obtain the prediction results of the current operating state; among which, flexible power resources include: electric vehicle charging and discharging resources and air conditioning load resources; S2, Construct a multi-objective optimization model, using the prediction results as input to the multi-objective optimization model, and outputting the prediction time domain. T p The optimal control strategy within the system; S3, based on control step size Δ t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy was implemented, and the results were obtained. S4. Based on the execution results, the model parameters and control weights of the multi-objective optimization model are dynamically corrected for the current control cycle, and the operating state is re-predicted and the control strategy is updated in the next control cycle, forming a multi-objective optimization autonomous control closed-loop control architecture.

[0019] In some preferred embodiments, S1 above, which models the state variables within the prediction window to obtain the current operating state prediction result, may further include: S11, during the control period Δ t Within the system, real-time data on the current operation of flexible power resources in distribution network areas and feeders is collected. S12, using current operating data and time series analysis methods, sets a short-term forecast period (i.e., forecast window) for the future. t , t + T p The operating state within the [ ] is predicted and modeled to obtain the current operating state prediction result.

[0020] The operational data includes: node voltage, electric vehicle access status, air conditioning operation status, electricity price curve, and outdoor temperature.

[0021] In some preferred embodiments, S12 above utilizes current operating data and combines it with time series analysis methods to set a short-term forecast period for the future. t , t + T p The system performs predictive modeling of the operating state within a given area to obtain the current operating state prediction result. This can further include: S121 adopts a combined prediction model that combines moving average and disturbance correction, smooths historical operating data and introduces correction terms to eliminate the impact of random fluctuations on prediction accuracy. S122, based on the combined prediction model, respectively, the prediction period [ t , t + T p Short-term forecasts are made for each state variable within the forecast window, forming a sequence of state variables within the forecast window; the state variables include: load, electricity price, and external temperature; S123, the sequence of state variables is used as the prediction result of the current running state, and subsequently used as the input of the multi-objective optimization model.

[0022] The prediction results may further include: electric vehicle access status prediction information, including the expected access / disconnection time, adjustable time window, and state of charge (SOC) change trend; air conditioning load operation status prediction information, including indoor temperature change trend, adjustable power range, and start / stop feasibility status; distribution area or feeder layer node voltage and transformer / feeder load rate change trend prediction results, used to characterize voltage over-limit risk and operational safety margin; adjustable resource response capability prediction information, including the time-series distribution of available charging and discharging capacity of electric vehicles and adjustable power range of air conditioning; and grid operation boundary prediction information, including the change trend of purchased power demand and power fluctuation range.

[0023] In some preferred embodiments, in S2 above, a multi-objective optimization model is constructed, the prediction results are used as input to the multi-objective optimization model, and the prediction time domain is output. T p The optimal control strategy within the system can further include: S21. Establish the optimization objective function of the multi-objective optimization model. The optimization objective function includes: operation safety objective, voltage qualification rate objective, economic operation objective and user comfort objective. By adjusting the weight coefficients of each objective through weighted summation, flexible objective preference configuration can be achieved under different operation scenarios. S22, determine the behavioral variables of various resources in time series; S23, Constructing system constraints for a multi-objective optimization model, including: power grid operation constraints, electric vehicle operation constraints, and air conditioning operation constraints; S24. Using the prediction results as input to the multi-objective optimization model and the behavioral variables as output variables, the optimization objective function is solved based on system constraints to obtain the prediction time domain. T p The optimal control strategy within the system.

[0024] In some preferred embodiments, the above-mentioned step S21, establishing the objective function of the multi-objective optimization model, may further include: Let the overall optimization objective function be... for: In the formula, , , and The weight factors for each optimization objective satisfy the following conditions: ; This indicates the load rate of the feeder or transformer, serving as a target for operational safety. This represents the penalty for node voltage deviation, which is the target for voltage compliance rate. This represents the cost of electricity purchase, serving as an economic operating target. For the purpose of user comfort; in: Operational security objectives The maximum load rate level of the system is defined as follows: In the formula, Discrete time interval after regulation t The actual power of the transformer or feeder. The rated capacity of the transformer or feeder; The goal By suppressing peak values, the risk of exceeding operational limits is reduced.

[0025] Voltage qualification rate target Used to maintain the voltage of the transformer area and feeder node within the allowable range. The inner part is defined as follows: In the formula, Indicates the voltage of the transformer substation or feeder node; This indicates the minimum permissible voltage value for a transformer substation or feeder node. This indicates the maximum permissible voltage value for a transformer substation or feeder node. The goal By penalizing voltage deviations beyond the limit at each time point, the strategy is encouraged to maintain voltage stability while ensuring the control effect.

[0026] Economic Operation Targets Used in load regulation to reduce the electricity purchase cost of distribution areas through peak shaving and valley filling, responding to price signals, etc., it is defined as follows: In the formula, Indicates electricity price, Indicates time t The amount of electricity purchased by the distribution area from the main grid; The goal It can guide electric vehicles to charge during off-peak electricity prices and discharge appropriately during peak electricity prices, thus optimizing the electricity consumption structure.

[0027] User comfort goals Introducing temperature offset and start / stop frequency This is a constraint term used to ensure that the use of air conditioning in regulation does not significantly affect the user experience, and is defined as follows: In the formula, and These represent the relative weights of different sub-objectives under the comfort dimension; The goal It constrains indoor temperature fluctuations within a set range (e.g., ±1°C) while suppressing energy consumption increases and equipment wear caused by frequent start-stop cycles.

[0028] In some preferred embodiments, the above-mentioned S22, which determines the behavioral variables of various resources in the time series in the prediction results, may further include: Let the total optimization time domain be T , T= [ t , t + T p To control the cycle. To control the step size, the total optimization time domain will be optimized. T Divided into Each time step; the current running data contains [number] time steps; Electric vehicle charging and discharging resources and For air conditioning load resources, the following behavioral variable system is established: For any electric vehicle At any control moment Its main behavioral variables are , indicating the first i electric vehicles t The charging and discharging power at any given time, in kW, is a behavior variable that is affected by the maximum and minimum charging and discharging power limits of the electric vehicle and the state of charge (SOC) constraints of the battery. like >0 indicates the first i electric vehicles t It is always in a charging state; like <0 indicates that the first i electric vehicles t Discharges continuously (supports V2G function); like =0, indicating the first i electric vehicles t They did not participate in regulation at any time; For any air conditioner At any control moment Considering key behavioral variables , and ;in: Indicates the first j air conditioner t Operating power at any given time (unit: kW); for inverter air conditioners, this value is a continuous variable, typically measured in [time range]. Adjustable within a certain range; for fixed-frequency air conditioners, this value is a fixed value during start-stop operation. Or 0; where, Indicates inverter air conditioner j Minimum operating power Indicates variable frequency air conditioner j Maximum operating power Indicates fixed-frequency air conditioner j Rated operating power; Indicates the first j The indoor temperature of the user under the control of the air conditioner (unit: °C) is used to reflect the impact of air conditioning control on user comfort. This variable is constrained by the dynamic changes in indoor temperature. A binary control variable, representing the first... j air conditioner t The start / stop status at any given time; for fixed-frequency air conditioners, if This indicates that the air conditioner is running. = ;like This indicates that the air conditioner is off. =0; This behavior variable can also be used to describe operational constraints such as start / stop frequency and minimum running time; Based on the above behavioral variables, defined at time... t Overall behavior vector for: Behavior Vector The moment-to-moment behavior characteristics of all key control resources serve as the core variable basis for subsequent solutions to the optimization problem.

[0029] For different types of air conditioning equipment, the model can switch the variables based on whether continuous power regulation is supported. or binary state variable The control mode enhances the model's adaptability.

[0030] In some preferred embodiments, the system constraints for constructing the multi-objective optimization model in step S23 above may further include: Establish power grid operation constraints, including: node voltage constraints and feeder and transformer load rate constraints; wherein: For any node n and time t The node voltage constraints are defined as follows: In the formula, For nodes n At any moment t The voltage; and The minimum and maximum voltages specified by the system (typically 0.9–1.1 pu). The load factor constraints for the feeder and transformer are defined as follows: In the formula, For a moment t The actual load, This refers to the rated capacity of the equipment.

[0031] Establish operating constraints for electric vehicles, including: discharge power boundary constraints, State of Charge (SOC) energy balance constraints, SOC limit constraints, and adjustable time window constraints; among which: For any electric vehicle i At any control moment t Within this context, the discharge power boundary constraints are defined as follows: In the formula, For electric vehicles i During the adjustment period t The charging and discharging power, For electric vehicles i Maximum discharge power, For electric vehicles i Maximum charging power; When electric vehicles support V2G Indicates the maximum discharge power. Indicates the maximum charging power; The vehicle's state of charge (SOC) changes over time, and the SOC energy balance constraint is defined as follows: In the formula, For a moment t+ 1. Battery state of charge For a moment t Battery state of charge, For charging and discharging efficiency, Rated battery capacity (unit: kWh); To ensure that the battery is not overcharged or over-discharged, the state of charge (SOC) must meet the SOC limit constraint, which is defined as follows: In the formula, The minimum allowable SOC value, This is the highest allowed SOC value; normally, it is set to... =0.2, set =0.9; Electric vehicle resources can only participate in regulation during the time when users connect their vehicles to the power grid; if the vehiclei In time period If the pause occurs, the adjustable time window constraint is defined as follows: In the formula, The time of connection to the power grid. The time for disconnecting from the power grid, For vehicles i The SOC value at the moment of disconnection from the power grid. The minimum state of charge value set for the end of the stay for electric vehicle users.

[0032] Establish air conditioning operation constraints, including: power regulation range constraints, room temperature dynamic constraints, temperature comfort range constraints, and start / stop frequency constraints for fixed-frequency air conditioners; among which: For inverter air conditioners, continuous adjustment capability is required, and their operating power is limited by minimum / maximum values. The corresponding power adjustment range constraint is defined as follows: For fixed-frequency air conditioners, their power is either a fixed value (in the operating state) or 0 (in the off state), and is determined by a 0-1 variable. To limit this, the corresponding power adjustment range constraint is defined as follows: In the formula, For the first j air conditioner t Operating power at any given time To indicate inverter air conditioner j Minimum operating power To indicate inverter air conditioner j Maximum operating power For the first j air conditioner t The running status at any given moment, =0 indicates the first j air conditioner t Always in the off state. =1 indicates the first j air conditioner t It is always running; For fixed frequency air conditioners j Rated operating power; Indoor temperature changes dynamically with the operation of the air conditioner and changes in the outside temperature. Therefore, the dynamic change of room temperature is approximately linear. The dynamic constraints of room temperature are defined as follows: In the formula, For the first j+The current room temperature for a user under the control of one air conditioner. For the first j The current room temperature for the user under the control of the air conditioner. To unify the external temperature, For building thermal inertia parameters, This is the coefficient of performance (negative value). For the first j air conditioner t Operating power at any given time; Air conditioning should be used to maintain room temperature within a comfortable range. Within this range, typically ±1°C of the set temperature, the temperature comfort zone constraint is defined as follows: In the formula, For the first j The current room temperature for the user under the control of the air conditioner. The lowest temperature set by the user. The maximum temperature set by the user; The start-stop frequency constraints for fixed-frequency air conditioners are defined as follows: Set the minimum start-stop interval for a fixed-frequency air conditioner. Introducing binary variables Indicates the start / stop status of the air conditioner and defines the start / stop change quantity. for: In the formula, For the first j air conditioner t The start / stop status at any given time. For the first j air conditioner t- Start-stop status at moment 1; The total number of start-stop operations per day shall not exceed The constraints are as follows: The interval between consecutive start-stop operations must not be less than The constraints are as follows: In some preferred embodiments, the above S3 is based on the control step size Δ t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy, upon achieving the desired results, may further include: Based on the preset device differentiation control logic, the first time step [ t , t +Δt The control strategy is transformed into actual executable control commands; among which, the equipment-differentiated control logic includes: Determine if an electric vehicle is connected to the grid; if so, execute the following control logic: if the device supports bidirectional V2G function, the control strategy is charging and discharging power, and the charging and discharging power is directly mapped to the actual power command and sent to the charging pile to control the vehicle's power behavior; if the device only supports unidirectional charging, the negative power in the control strategy will be automatically set to zero, and only the charging strategy will be retained; the control command is issued in the form of a set time period or power curve; if not, that is, the vehicle is not connected, is about to leave the station, or has insufficient SOC, the vehicle is determined to be uncontrollable and the control command is automatically blocked. Determine the type of air conditioning load; if it is an inverter model, the control strategy is continuous power quantity, which is converted into an air conditioning temperature setting command according to the set ratio; if it is a fixed frequency model, the control strategy is start-stop state, and a delay judgment is made in combination with start-stop frequency constraints before execution. Embedded basic exception handling logic: if device communication fails, control command is abnormal, or user manually withdraws from participation, the corresponding control weight will be automatically adjusted or the device will be removed in the next round of rolling optimization.

[0033] In some preferred embodiments, S4 above, which involves dynamically adjusting the model parameters and control weights of the multi-objective optimization model based on the execution results for the current control cycle, may further include: Real-time acquisition of execution results, including: actual charging and discharging behavior of electric vehicles, air conditioning operation curves, node voltage deviation, and total load response; Compare and analyze the execution results with the prediction results to identify the sources of prediction and execution deviations; The execution results are used to correct the model parameters and adjust the weights of each sub-objective in the optimization objective, thereby updating the control strategy within the current control cycle.

[0034] Based on the same inventive concept, one embodiment of the present invention also provides a multi-objective optimization autonomous control system for the feeder layer of a distribution network area.

[0035] Specifically, such as Figure 2 As shown, the multi-objective optimization autonomous control system for the feeder layer of a distribution network provided in this embodiment may include: The state awareness and prediction module is used to detect and predict the current control cycle Δ t Any adjustment moment within t A forecasting window is established based on the current operational data of flexible power resources. t , t + T pThe system models the state variables within the prediction window to obtain the prediction results of the current operating state; among which, flexible power resources include: electric vehicle charging and discharging resources and air conditioning load resources; The optimization decision module is used to construct a multi-objective optimization model. It takes the prediction results as input to the multi-objective optimization model and outputs the prediction time domain. T p The optimal control strategy within the system; The strategy execution and feedback module is based on the control step size Δ. t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy is implemented to obtain the execution results. Based on the execution results, the model parameters and control weights of the multi-objective optimization model are dynamically corrected in the current control cycle, and the operating state is re-predicted and the control strategy is updated in the next control cycle, forming a multi-objective optimization autonomous control closed-loop control architecture.

[0036] It should be noted that the steps in the method provided by the present invention can be implemented using corresponding modules, devices, units, etc. in the system. Those skilled in the art can refer to the technical solution of the method to realize the composition of the system. That is, the embodiments in the method can be understood as preferred examples for building the system, and will not be elaborated here.

[0037] The design and working principles of the technical solutions provided in the above embodiments of the present invention will be further explained in detail below with reference to a specific application example.

[0038] like Figure 3 As shown, the complete execution process of the multi-objective optimization autonomous control method for the feeder layer of the distribution network area under the rolling optimization mechanism is systematically demonstrated.

[0039] The key to achieving autonomous operation at the distribution transformer / feeder level lies in activating the numerous flexible regulation resources distributed at the end, enhancing their rapid response and regulation capabilities to the grid's operating status. Two typical terminal types—electric vehicles (EVs) and comfort loads such as variable frequency air conditioners—are characterized by wide distribution, controllable power, and high adjustability, constituting important control targets and supporting resources at the distribution transformer level. While their operating characteristics differ, they are complementary in timing and regulation direction, providing a foundation for establishing a multi-objective optimized regulation mechanism. This section, starting from the analysis of regulation characteristics, proposes a mutually supportive and coordinated response mechanism between charging / discharging resources and air conditioning loads, supporting the subsequent construction of optimization models and autonomous strategies.

[0040] Electric vehicles (EVs) are typical mobile distributed energy units. Once connected to the power distribution system, they not only act as electrical loads but also possess energy storage and regulation capabilities. With the continuous increase in EV penetration, user-side EV clusters have gradually formed virtual energy storage resource pools with considerable controllable value at the distribution transformer level. Their main characteristics are as follows: Privately owned electric vehicles typically have charging capacities ranging from 2 to 7 kW. Some models supporting V2G (Vehicle-to-Grid) functionality can also discharge back into the grid, exhibiting bidirectional charging / discharging regulation capabilities. This characteristic allows them to participate in peak shaving operations during peak load periods and to charge and store energy during off-peak periods, achieving cross-period energy transfer and local load buffering, making them a typical adjustable energy storage resource. Furthermore, the adjustable periods for electric vehicles are relatively flexible. Most users' parking activities are concentrated after the evening peak and at night, providing ample time windows for their participation in nighttime control strategies and helping to improve the controllability of the distribution system's nighttime load.

[0041] In designing scheduling strategies, it is essential to fully consider users' travel needs and ensure that their State of Charge (SOC) before scheduled departure is not lower than the minimum travel guarantee threshold to avoid impacting user experience due to control operations. Furthermore, given the volatility and randomness of electric vehicle behavior characteristics such as access time, initial SOC, and dwell time, model construction typically requires incorporating historical statistical data or scenario sets to establish probabilistic constraints or uncertainty descriptions, thereby enhancing the robustness and feasibility of the control strategy.

[0042] Despite some behavioral uncertainties, electric vehicles exhibit relatively stable power changes during actual operation and possess good load predictability. Especially after being incorporated into the edge autonomy framework, their regulation behavior is more easily visualized and managed and rolled out, enabling the construction of an integrated control loop of "prediction-optimization-execution," which is of great value in enhancing the autonomous capabilities of the distribution area.

[0043] Air conditioning load, as a typical comfort load in power distribution network areas, is widely distributed in residential buildings, commercial buildings, and office spaces. It accounts for a high proportion of the load during the high-temperature summer months and is one of the main factors causing transformer overload and voltage fluctuations. From the perspective of adjustable characteristics, air conditioners can be divided into two categories: fixed-frequency and variable-frequency. Variable-frequency air conditioners have stronger continuous power regulation capabilities and are an important resource for building flexible autonomous response mechanisms. However, fixed-frequency air conditioners still exist in some areas and also possess certain regulation potential.

[0044] In terms of responsiveness, inverter air conditioners can achieve linear power control within a range of 30% to 100% through inverter compressors, providing rapid response and making them suitable for fine-grained adjustments. While fixed-frequency air conditioners can only control start-stop, they can participate in basic response through rotary or periodic adjustments. Regarding comfort constraints, air conditioner operation directly affects the user's temperature experience; therefore, control strategies must keep indoor temperature fluctuations within ±1°C while limiting frequent start-stop cycles to ensure equipment safety and user satisfaction.

[0045] In terms of temporal distribution, air conditioning load exhibits a clear peak concentration, especially during hot and sunny summers, when the load is mainly concentrated from daytime to evening, showing a large-scale synchronous start-up and shutdown characteristic, which can easily cause sudden load surges and voltage disturbances in local transformer areas. Regarding operational constraints, air conditioning equipment has limitations such as minimum start-up and shutdown times, minimum operating times, and power change rates. Control measures must balance response effectiveness with equipment protection to ensure stable operation.

[0046] Furthermore, air conditioning loads are widely distributed and commonly accessed on the user side, exhibiting good adaptability for edge deployment. Integration with local autonomous control systems is expected to enable rapid response, voltage optimization, and load balancing at the transformer / feeder level, providing crucial support for flexible control of the distribution network.

[0047] Charging and discharging resources (represented by electric vehicles) and air conditioning loads, as two typical flexible power resources at the end of distribution substations / feeders, exhibit significant differences in their operational characteristics: the former possesses bidirectional energy regulation capabilities and certain energy storage attributes, while the latter is characterized by a high proportion of purely energy-consuming loads constrained by user comfort. However, it is precisely this difference that provides the foundation for the coordinated response of the two types of resources in autonomous regulation, demonstrating good complementarity and coupling potential. A rationally designed regulation mechanism, treating both as a unified resource pool for coordinated scheduling, can not only improve local regulation capabilities and operational efficiency but also enhance the system's resilience to operational risks, creating a resource synergy benefit of "1+1>2".

[0048] Firstly, in terms of time, the two have a clear complementary adjustment timeline. Air conditioning load in summer is mainly concentrated during the high-temperature daytime hours, typically between 10:00 AM and 6:00 PM, constituting the main component of daytime peak load. Electric vehicles, on the other hand, are mostly connected to the grid after users return home, with their charging periods usually distributed from evening to night or even the early morning of the following day. This creates a natural peak-shaving effect between the adjustable time periods of EVs and the air conditioning load. Through reasonable time period division and response mechanism design, peak shaving control can be achieved during the day using methods such as air conditioning rotation control, while at night, the charging and discharging capacity of EVs can be fully utilized to achieve load balance, effectively alleviating load pressure at different times and improving the stability of the transformer substation operation.

[0049] Secondly, from the perspective of power direction complementarity, air conditioning, as a purely energy-consuming load, is typically regulated by load management based on reducing power demand. Electric vehicles, on the other hand, possess dual "source-load" attributes. They can participate in charging management as a load and also supply power to the distribution area via V2G, acting as a "virtual power source" in regulation. During peak load periods, EVs can appropriately release electricity to support local voltage and load; during off-peak electricity prices or periods of reduced air conditioning load, they can fill the load gap through charging. This two-way adjustment mechanism enables the two types of resources to form a dynamic complementary relationship during grid supply and demand fluctuations, constructing a stronger power regulation buffer and improving the operational flexibility and adaptability of the autonomous system.

[0050] Third, at the level of control objectives, the two share complementary and synergistic characteristics. The core of air conditioning regulation lies in controlling indoor temperature fluctuations to ensure user comfort; therefore, its adjustment strategy must strictly limit the range of temperature changes and the frequency of start-stop operations, with the primary control objective being minimizing user perception. Electric vehicles, on the other hand, focus more on economy and energy efficiency, such as guiding charging behavior through electricity price signals, participating in demand response, or obtaining incentives through peak shaving and valley filling. When participating in a multi-objective optimization model, these two types of resources can respond according to their weights: EVs primarily undertake the objectives of economy and power balance, while air conditioning assists in regulation while ensuring comfort is not compromised. By flexibly allocating strategy execution weights, a reasonable trade-off between multiple objectives can be achieved, ensuring the overall autonomous system achieves comprehensive optimization across four dimensions: safety, economy, quality, and user satisfaction.

[0051] Furthermore, EVs and air conditioners possess complementary spatial response capabilities. In a typical distribution area / feeder, air conditioners and electric vehicles are distributed across different user nodes, exhibiting high decentralization and controllable access. This spatial distribution characteristic allows for differentiated regulation based on node location, voltage status, and load size when local grid problems such as voltage exceedances or feeder overloads occur. For example, if the voltage at a node is too high, EVs can be prioritized for charging at that node to absorb electricity, or the air conditioner's operating power can be reduced to alleviate voltage rise; when there is a significant local power shortage, EVs at that node can be activated for discharge, or some air conditioner load can be appropriately shut down to achieve local voltage or load balancing. By constructing a resource response mechanism based on node status awareness, the spatial accuracy and emergency response capabilities of autonomous control are significantly improved.

[0052] Furthermore, the resource sharing mechanism can be enhanced with various strategies in practical deployments to improve its robustness. For example, in scenarios with large load forecasting errors or high uncertainty in user behavior, fault-tolerant scheduling methods can be adopted to dynamically assess resource availability and automatically switch the primary adjustment resources. When user participation is unstable, the stability of user-side resource response can be improved by setting comfort constraints and economic incentive mechanisms. Ultimately, forming a four-dimensional collaborative multi-resource control network of "time-power-target-space" is a key technical path to improve the autonomous operation quality of transformer substations / feeders.

[0053] In summary, the multidimensional complementary relationship between charging and discharging resources and air conditioning loads in terms of operational characteristics lays a solid foundation for building an efficient, flexible, and reliable autonomous control mechanism for distribution substations. Through coordinated design of the two in dimensions such as adjustment timing, power direction, control objectives, and response space, not only can the potential of end-point resources be fully explored, but also refined management and dynamic adaptive control of the distribution system's operating status can be achieved, effectively supporting the intelligent transformation of future active distribution networks.

[0054] To achieve coordinated control of load resources at the transformer substation / feeder level and meet the comprehensive optimization needs of the distribution system in terms of operational safety, power quality, economy, and user experience, a multi-objective optimization model that can dynamically reflect the actual operating status, resource control characteristics, and constraint boundaries needs to be constructed. This model should possess the following functional characteristics: capable of making feasible local control decisions for multi-source heterogeneous flexible resources; possessing periodic rolling optimization capabilities for autonomous control scenarios; and supporting control trade-offs and weight configuration for multi-objective coordination. Therefore, this invention proposes a local multi-objective optimization model suitable for autonomous operation of transformer substations / feeders. The model covers key modules such as setting the control objective function, defining variables, constraints on adjustable resource characteristics, grid operation constraints, user preference constraints, and rolling mechanism design, forming a complete deployable optimization computation framework.

[0055] This invention constructs a multi-objective local optimization model aimed at solving the problem of coordinated regulation of typical flexible load resources at the distribution area / feeder level, thereby enhancing autonomous operation capabilities. The core resources addressed by the model include electric vehicle (EV) charging and discharging resources and terminal equipment with regulation potential, such as air conditioning loads. Considering typical problems in the operation of the distribution network's end points, such as large peak-valley load differences, frequent voltage fluctuations, and concentrated user behavior, the model's objective system covers four core dimensions: operational safety, power quality, economy, and user comfort. It strives to achieve a balance between local supply and demand, equipment operation boundary control, system power purchase cost optimization, and user experience assurance during the regulation process.

[0056] The time-related variables involved in the technical solution of this invention are defined as follows: tFor control moment or discrete time index, it represents any control moment (time point) on the discrete time axis. Δ t To control the step size or control period, it represents the time interval between two adjacent control moments, used for strategy execution; T The total optimization time domain represents the complete time range covered by a single solution of the multi-objective optimization model. T =[ t , t + T p ]; T p The prediction time domain length represents the prediction window. t , t + T p The length of time.

[0057] This model employs a rolling optimization mechanism for dynamic scheduling decisions. Specifically, at any given time... t In the future time domain T p To optimize the window, based on the current operating status information of the transformer area / feeder and the planned load curve, generate [ t , t + T p A flexible resource control strategy within the system; followed by a certain control step size. Periodic updates (e.g., every 15 minutes) enable continuous perception of operational status and dynamic adjustment of the control loop. This sliding window-style rolling optimization structure effectively addresses the uncertainties caused by changes in end-point load, improving the timeliness and stability of control response while ensuring strategy feasibility.

[0058] During the modeling process, typical behavioral constraints for electric vehicles and air conditioning loads were uniformly defined, such as minimum SOC requirements, start-stop restrictions, operating power boundaries, and adjustment time windows, ensuring the model's consistency and adaptability in controlling different resources. Simultaneously, the optimized model incorporates a multi-objective trade-off mechanism. Based on changes in demand under different operating scenarios, the weighting factors in the objective function can be adjusted to achieve flexible scheduling guided by either economy or comfort while ensuring safety, thus supporting the efficient implementation of autonomous strategies.

[0059] To meet the diverse operational needs of distribution network substations / feeders in autonomous scenarios, a multi-objective optimization function is constructed, comprising four categories of key performance indicators, corresponding to the system's operational safety, voltage qualification rate, economic operation, and user comfort control objectives, respectively. This objective function is expressed in a weighted summation form. By adjusting the weight coefficients of each sub-objective, flexible objective preference configuration can be achieved under different operational scenarios to meet practical engineering requirements. The overall optimization objective function is as follows: in: , , and The weighting factors for each optimization objective satisfy... ; For feeder or transformer load rate targets; Penalty for node voltage deviation; To achieve the target for electricity purchase cost; For the purpose of user comfort.

[0060] 1. Operational security objectives Overload operation of transformers or feeders not only increases the risk of equipment failure but also affects the power supply reliability of the entire distribution area. The safety objective function, primarily used to constrain the system's maximum load rate, is defined as: in, Discrete time interval after regulation t The actual power of the transformer or feeder. The rated capacity of the transformer or feeder aims to suppress peak values ​​and reduce the risk of exceeding operating limits.

[0061] 2. Voltage qualification rate target Voltage quality is a crucial indicator of the power supply capacity of a power distribution system. Control measures must ensure voltage stability at transformer substations / feeder nodes. Maintain within the allowable range Inside. The definition is as follows: This approach penalizes voltage deviations beyond the limit at each time point, encouraging the strategy to maintain voltage stability while ensuring the control effect.

[0062] 3. Economic Operation Objectives This objective focuses on reducing the electricity purchase cost for distribution transformer areas through peak shaving and valley filling, and responding to price signals during load regulation. It is defined as follows: in, For electricity price, carve tThe power output of electricity purchased by the distribution area from the main grid. This target can guide electric vehicles to charge during off-peak electricity prices and discharge appropriately during peak electricity prices, thus optimizing the electricity consumption structure.

[0063] 4. User comfort goals To ensure that the user experience is not significantly affected when the air conditioner is involved in regulation, a temperature offset is introduced. and start / stop frequency As constraints, construct the following objective: in and This represents the relative weights of different sub-objectives within the comfort dimension. This function constrains indoor temperature fluctuations within a set range (e.g., ±1°C) while suppressing energy consumption increases and equipment wear caused by frequent start-stop cycles.

[0064] The aforementioned multi-objective optimization function constructs a universal and scalable objective system by systematically modeling four key indicators: operational safety, voltage quality, economy, and user comfort. This system can cover the core control requirements of autonomous operation of distribution areas / feeders. In practical applications, the model can flexibly adjust the weighting factors to adapt to different operating scenarios and strategy preferences, supporting multiple operating modes such as safety-first, economy-first, or comfort-first. Simultaneously, this objective function provides a unified evaluation criterion and optimization direction for subsequent flexible resource modeling, rolling optimization mechanisms, and strategy execution logic, laying the foundation for achieving efficient closed-loop control of autonomous control strategies.

[0065] To achieve coordinated and optimized control of flexible loads in transformer substations / feeder layers, this multi-objective optimization model needs to define the temporal behavioral variables of various resources as inputs for decision-making during the optimization process. Let the total optimization time domain be... T , T= [ t , t + T p To control the cycle. The control step size (discrete time step) is divided into: 1 time step; contains a total of 1 time step; Electric vehicle resources and Air conditioning load resources. Considering the different physical control mechanisms and operating states of the two types of resources, this invention constructs the following optimization decision variable system: 1. Decision variables for electric vehicles For any electric vehicle At any moment Its main decision variables are , for the first i electric vehicles t The charging and discharging power at any given time, measured in kW.

[0066] like >0 indicates the first i electric vehicles t The time indicates that it is in a charging state; if <0 indicates that the first i electric vehicles t Discharge continuously (e.g., if V2G functionality is supported); if =0, indicating the first i electric vehicles t The time frame is not involved in regulation. This variable is affected by the maximum / minimum charge / discharge power limits of electric vehicles and their battery state of charge (SOC) constraints.

[0067] 2. Decision variables for air conditioning load For any air conditioner Arbitrary control time t Consider the following core regulatory variables , and .

[0068] in, For the first j air conditioner t Operating power at any given time (unit: kW). For inverter air conditioners, this value is a continuous variable, typically measured in [time range]. Adjustable within a specified range. For fixed-frequency air conditioners, this value is a fixed value during start-stop operation. Or 0; The first under air conditioning control j The indoor temperature (in °C) for each user is used to reflect the impact of air conditioning control on user comfort. This variable is constrained by a dynamic indoor temperature change model.

[0069] In addition, to represent the start-stop behavior of a fixed-frequency air conditioner, the following binary control variables need to be introduced. , indicating the first j air conditioner t The start / stop status at any given time. For fixed-frequency air conditioners, if... This indicates that the air conditioner is running. = ;like This indicates that the air conditioner is off. =0. This state variable can also be used to describe operational constraints such as start / stop frequency and minimum running time.

[0070] 3. Scheduling Vector Construction Combining the above variables, defined at time... t The overall scheduling vector is: The above scheduling vector The moment-to-moment behavioral characteristics of all key control resources form the core variable foundation for subsequent optimization problems. For different types of air conditioning equipment, the model can adjust the variables based on whether continuous power regulation is supported. or binary state variable The control mode enhances the model's adaptability.

[0071] To ensure the feasibility and effectiveness of the optimization model under realistic conditions, system constraint modeling is required for the operating boundaries of the transformer substation / feeder system, the behavioral characteristics of flexible load resources, and user experience. The constraints are mainly divided into three categories: grid operation constraints, electric vehicle operation constraints, and air conditioning operation constraints, as detailed below: 1. Power grid operation constraints a. Node voltage constraints The voltage at each node in the transformer substation / feeder system should be maintained within the allowable operating range to avoid undervoltage or overvoltage conditions. For any node... n and time t ,have: in, For nodes n At any moment t The voltage; and The minimum and maximum voltages specified by the system (typically 0.9–1.1 pu).

[0072] b. Feeder / Transformer load rate constraints To prevent equipment from operating under overload, the maximum load level of feeders or transformers must be limited. For a moment t The actual load, Given the rated capacity of the equipment, the feeder / unit variable load rate constraint is shown in equation (5-8).

[0073] 2. Electric vehicle operation constraints To ensure that the adjustment does not affect users' travel and maintains the safe operation of the battery, the following constraints need to be set: a. Discharge power boundary constraints For any electric vehicle i At any control moment t Within this range, its charging and discharging power should be within the permissible range: If the vehicle supports V2G, then Indicates the maximum discharge power; This indicates the maximum charging power.

[0074] b. SOC energy balance constraint The state of charge of a vehicle changes over time and must be described using the energy balance formula: in, For a moment t Battery state of charge; For charge and discharge efficiency; Rated battery capacity (unit: kWh); To control the step size, such as 15 minutes.

[0075] c. SOC Limit Constraints To ensure the battery is not overcharged or over-discharged, the state of charge should meet the following requirements: Generally, it can be set. =0.2, =0.9.

[0076] d. Adjustable time window constraint Electric vehicle resources can participate in regulation only during the time users connect their vehicles to the power grid. If the vehicle... i In time period If we stay, then: In addition, to meet travel needs, you must ensure the following before leaving the station: in, The minimum state of charge value set for the end of the stay for electric vehicle users.

[0077] 3. Air conditioning operation constraints As a typical comfort load, air conditioning requires constraint modeling that considers its operating mechanism, thermodynamic characteristics, and user comfort requirements. a. Power adjustment range constraints For inverter air conditioners, continuous adjustment capability is required, and their operating power is limited by minimum / maximum values: For fixed-frequency air conditioners, their power is either a fixed value (in operating state) or 0 (in off state), determined by a 0-1 variable. To limit: b. Dynamic model of room temperature The indoor temperature changes dynamically with the operation of the air conditioner and the change in the outside temperature. An approximate linear model is as follows: in, For the first j Current room temperature for each user; To unify the external temperature, These are the building's thermal inertia parameters; This is the coefficient of performance (negative value).

[0078] c. Temperature comfort range constraints Air conditioning should be used to maintain room temperature within a comfortable range. Within this range, it is generally ±1°C of the set temperature. For example, when the temperature is set to 26°C, the comfort zone is [25, 27]°C.

[0079] d. Start-stop frequency limitation (fixed-frequency air conditioner) Frequent start-stop cycles may affect equipment lifespan and energy efficiency; therefore, a minimum start-stop interval must be set. Input binary variable This indicates the air conditioner's on / off status.

[0080] Based on this, the start-stop change quantity is further defined: The total number of start-stop operations per day shall not exceed The constraints are as follows: The interval between consecutive start-stop operations must not be less than The constraints are as follows, and can be expressed through state constraints or logical judgments as follows: .

[0081] The above constraint system comprehensively covers the operational boundaries and behavioral limitations of various typical resources in the transformer / feeder layer, providing a reliable guarantee for solving subsequent optimization models and implementing practical strategies. In engineering deployment, constraint parameters (such as capacity, voltage limits, and comfort zones) can be customized based on equipment models and user preferences, thereby achieving scenario adaptation of the model and precise response control.

[0082] With the widespread access of distributed control resources and the highly diversified user behaviors, traditional centralized control models are struggling to meet the growing demand for localized autonomous operation at the distribution network / feeder level. Therefore, there is an urgent need to construct an autonomous control strategy system with hierarchical management, real-time response, and self-correction capabilities to support the efficient and coordinated scheduling of flexible resources at the end of the distribution network.

[0083] Based on the aforementioned multi-objective optimization model, this invention proposes a control strategy architecture and hierarchical execution mechanism suitable for autonomous operation scenarios of transformer substations / feeders, and further introduces a rolling optimization control mechanism to improve the dynamic adaptability and continuous effectiveness of the control strategy.

[0084] The autonomous control strategy for transformer substations / feeders revolves around a core objective: to achieve near-optimal system performance across multiple operational dimensions, including safety, voltage compliance, economy, and user comfort, through continuous optimization and real-time feedback, given limited local information. To achieve this objective, the overall strategy system is divided into three functional parts, forming a closed-loop control logic.

[0085] The first part is the state perception and prediction section. This section is responsible for collecting real-time operational data of the transformer substations / feeders, including node voltage, electric vehicle access status, air conditioning operation status, electricity price curves, and outdoor temperature. It then uses time series analysis methods to predict and model the state variables for a short-term forecast period. This prediction result will serve as the core input variable for the optimization model, determining the timeliness and foresight of the control strategy.

[0086] Secondly, there is the optimization decision-making section. This section takes the predicted state as input and calls the multi-objective optimization model established earlier to solve for the optimal control scheme for the flexible load in the current prediction time domain. The optimization objectives cover four dimensions: operational safety, voltage stability, economical operation, and user comfort. A weight adjustment mechanism supports strategy adaptation under various operating scenarios. This module is the intelligent hub of autonomous control, undertaking the core functions of resource coordination and objective balancing.

[0087] Finally, the strategy execution and feedback section transforms the optimization results into control commands recognizable by the equipment side, which are then sent to terminals such as EV charging facilities and air conditioning control units. Simultaneously, it records the state changes after strategy execution, forming an operation log and performance feedback, serving as an important reference for the next round of optimization. These three modules work together to form a continuous closed-loop control system of "prediction-optimization-execution-feedback".

[0088] To address the challenges of rapid load fluctuations, frequent electricity price changes, and high uncertainty in user behavior within the operating environment of distribution areas / feeders, this strategy introduces a rolling optimization mechanism (Model Predictive Control, MPC) as the control backbone to construct a high-frequency, short-cycle dynamic control process.

[0089] The basic process of rolling optimization is as follows: In each control cycle Δ t The system collects the current operating status and constructs a prediction window. t , t + T pThe system models variables such as load, electricity price, and temperature during the forecast period, inputs them into a multi-objective optimization model for solution, and outputs the optimal control strategy for that time domain; subsequently, only the first time step is executed. t , t +Δ t Based on the control results, the status is re-collected and the strategy is updated in the next control cycle to achieve a continuous rolling and self-correcting operating logic.

[0090] To ensure the real-time performance and convergence of the rolling mechanism, the key parameters are set as follows: [The following parameters are used to control the period Δ]. t As the rolling time step, i.e., the control step size, this control step size Δ t It is recommended to set it to 15 minutes; optimize the prediction window. T p The forecast window is typically between 1 and 2 hours; the forecast of load and external temperature uses time series modeling based on historical operating data, employs a moving average method to characterize the changing trend, and introduces a disturbance correction mechanism to correct for short-term random fluctuations in order to obtain the forecast window. t , t + T p The load and temperature prediction results within the prediction window are presented. To address the uncertainties in the arrival time, dwell time, and initial state of charge of electric vehicles, a scenario-based modeling approach is introduced when constructing the optimization model within the prediction window. This approach discretizes different possible electric vehicle arrival and departure behaviors and dynamically selects or updates representative scenarios during the rolling optimization process to improve the adaptability and robustness of the control strategy to fluctuations in user behavior.

[0091] This mechanism not only improves the real-time response capability of the strategy, but also solves the problems of strategy lag and rigidity in centralized scheduling. It is particularly suitable for transformer / feeder scenarios with frequent load changes and dense flexible resources, providing solid support for autonomous control capabilities.

[0092] The decision variables output by the optimization model need to be transformed into actual executable control commands, which will then be applied to the two flexible resources: electric vehicles and air conditioning load. Considering the diversity and functional boundaries of the resource control interfaces, the strategy execution logic is divided into the following two parts: Regarding the electric vehicle control logic, if the device supports bidirectional V2G functionality, the optimized output charging and discharging power can be directly mapped to actual power commands (in kW) and sent to the charging pile, which then controls the vehicle's power behavior. If the device only supports unidirectional charging, negative power values ​​will be automatically reset to zero, retaining only the charging strategy. Simultaneously, control commands can be issued according to set time periods or power curves, facilitating flexible scheduling configuration. If a vehicle is not connected, is about to leave the station, or has insufficient SOC, the system will determine that the vehicle is uncontrollable and automatically disable commands.

[0093] In the air conditioning load control logic, if the air conditioning unit is an inverter model, the optimization result is a continuous power quantity, which the system can convert into an air conditioning temperature setting command according to a set ratio (e.g., 70% power corresponds to 27°C). If the air conditioning unit is a fixed-frequency type, the optimization result is an on / off state (0 / 1). Before execution, a delay judgment needs to be made in conjunction with the on / off frequency constraint to avoid frequent start-stop of the equipment. To improve the fault tolerance of the strategy, the strategy execution logic also embeds a basic exception handling mechanism. For example, if the equipment communication fails, the control command is abnormal, or the user manually withdraws from participation, the system will automatically adjust its control weight or remove the equipment in the next round of rolling optimization.

[0094] The overall strategy execution process not only demonstrates a high degree of adaptability to equipment differences, but also possesses automatic fault tolerance, state awareness, and response adjustment capabilities, laying the foundation for the deployment of the strategy in actual engineering projects.

[0095] An effective control strategy lies not only in the accuracy of the one-time optimization result, but also in its ability to continuously self-evolve. To this end, a complete strategy feedback and self-correction mechanism is proposed to achieve closed-loop execution and dynamic updating of the strategy.

[0096] The first step is strategy feedback collection, including actual EV charging and discharging behavior, air conditioning operation curves, node voltage deviations, and total load response. This data is transmitted back through intelligent devices to form a complete strategy execution profile. The second step is strategy effectiveness evaluation. The system compares and analyzes the execution results with the model predictions to identify the sources of prediction and execution deviations, such as whether voltage limits are exceeded, whether peak-shaving pricing is implemented, and whether power quality has improved.

[0097] Based on this, model parameters are corrected, including correcting thermal inertia parameters for temperature change curve deviations, correcting access probability distributions for sudden changes in electric vehicle departure behavior, and updating sliding prediction weights for sudden changes in load curves. Simultaneously, the objective function weighting factors can be fine-tuned based on strategy effectiveness and user feedback. For example, the weight of the temperature offset penalty term can be appropriately increased when comfort deviations are too large, and the weight of the economic objective can be strengthened when electricity price response is not significant.

[0098] An embodiment of the present invention also provides a computer control terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can be used to perform any of the methods described in the above embodiments of the present invention.

[0099] Optionally, the memory is used to store programs; the memory may include volatile memory, such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random-access memory (DDR SDRAM), etc.; the memory may also include non-volatile memory, such as flash memory. The memory is used to store computer programs (such as application programs and functional modules that implement the above methods), computer instructions, etc., and the aforementioned computer programs and computer instructions can be partitioned and stored in one or more memories. Furthermore, the aforementioned computer programs, computer instructions, data, etc., can be accessed by the processor.

[0100] A processor is used to execute computer programs stored in memory to implement the various steps of the methods or various modules of the systems involved in the above embodiments. For details, please refer to the relevant descriptions in the preceding method and system embodiments.

[0101] The processor and memory can be separate structures or integrated structures. When the processor and memory are separate structures, they can be coupled together via a bus.

[0102] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can be used to perform the method of any of the above embodiments of the present invention.

[0103] Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transfer of computer programs from one place to another. Storage media can be any available medium accessible to a general-purpose or special-purpose computer. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. Alternatively, the ASIC can reside in a user device. Of course, the processor and storage medium can also exist as discrete components in a communication device.

[0104] The multi-objective optimization autonomous control method and system for the feeder layer of a distribution network provided in the above embodiments of the present invention starts with state perception and short-term prediction. It achieves collaborative decision-making for flexible resources such as electric vehicles and air conditioners through a multi-objective optimization model, and completes strategy issuance and execution with the support of differentiated equipment control logic. Simultaneously, relying on real-time feedback and deviation evaluation of execution results, it achieves dynamic correction of model parameters and control weights. This control framework not only effectively improves the adaptive capability of the distribution area / feeder layer to load fluctuations and user behavior uncertainties, but also enhances the comprehensive coordination level of local operation between safety, economy, and comfort, providing a clear and feasible technical path for the engineering-based autonomous control of flexible resources at the end of the distribution network.

[0105] Any matters not covered in the above embodiments of the present invention are well-known in the art.

[0106] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. A multi-objective optimization autonomous control method for the feeder layer of a distribution network, characterized in that, include: For the current control period Δ t Any adjustment moment within t A forecasting window is established based on the current operational data of flexible power resources. t , t + T p The system models the state variables within the prediction window to obtain the prediction results of the current operating state; wherein, the flexible power resources include: electric vehicle charging and discharging resources and air conditioning load resources; A multi-objective optimization model is constructed, and the prediction results are used as input to the multi-objective optimization model to output the prediction time domain. T p The optimal control strategy within the system; Based on control step size Δ t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy was implemented, and the results were obtained. Based on the execution results, the model parameters and control weights of the multi-objective optimization model are dynamically corrected for the current control cycle, and the operating state is re-predicted and the control strategy is updated in the next control cycle, forming a multi-objective optimization autonomous control closed-loop control architecture.

2. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 1, characterized in that, The step of modeling the state variables within the prediction window to obtain the current running state prediction result includes: During the control cycle, real-time data on the current operation of flexible power resources in the distribution network area and feeders is collected. Using the current operational data and time series analysis methods, a short-term forecast period is set for the future. t , t + T p The operating state within the [ ] is predicted and modeled to obtain the current operating state prediction result.

3. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 2, characterized in that, The method utilizes the current operating data and combines it with time series analysis to set a short-term forecast period for the future. t , t + T p The system performs predictive modeling of the operating state within a given area to obtain the current operating state prediction result, including: A combined prediction model that combines moving average and perturbation correction is adopted to smooth historical data and introduce correction terms to eliminate the impact of random fluctuations on prediction accuracy. Based on the aforementioned combined prediction model, the prediction period [ t , t + T p Short-term predictions are made for each state variable within the prediction window to form a sequence of state variables within the prediction window. The sequence of state variables is used as the prediction result of the current running state, and subsequently used as the input of the multi-objective optimization model.

4. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 1, characterized in that, The multi-objective optimization model is constructed by using the prediction results as input and outputting the prediction time domain. T p The optimal control strategies within the system include: An optimization objective function for a multi-objective optimization model is established, which includes: operational safety objective, voltage qualification rate objective, economic operation objective, and user comfort objective. Determine the behavioral variables of various resources over time; The system constraints for constructing the multi-objective optimization model include: power grid operation constraints, electric vehicle operation constraints, and air conditioning operation constraints. Using the prediction results as input to a multi-objective optimization model and the behavioral variables as output variables, the optimization objective function is solved based on the system constraints to obtain the prediction time domain. T p The optimal control strategy within the system.

5. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 4, characterized in that, The objective function for establishing the multi-objective optimization model includes: Let the overall optimization objective function be... for: In the formula, , , and The weight factors for each optimization objective satisfy the following conditions: ; This indicates the load rate of the feeder or transformer, serving as a target for operational safety. This represents the penalty for node voltage deviation, which is the target for voltage compliance rate. This represents the cost of electricity purchase, serving as an economic operating target. For the purpose of user comfort; in: The operational safety objective The definition is as follows: In the formula, Discrete time interval after regulation t The actual power of the transformer or feeder. The rated capacity of the transformer or feeder; The voltage qualification rate target The definition is as follows: In the formula, Indicates the voltage of the transformer substation or feeder node; This indicates the minimum permissible voltage value for a transformer substation or feeder node. This indicates the maximum permissible voltage value for a transformer substation or feeder node. The economic operation objectives The definition is as follows: In the formula, Indicates electricity price, Indicates time t The amount of electricity purchased by the distribution area from the main grid; The user comfort goal The definition is as follows: In the formula, This is the temperature offset. For start / stop frequency, and These represent the relative weights of different sub-objectives under the comfort dimension.

6. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 4, characterized in that, The determination of the behavioral variables of various resources over time includes: Let the total optimization time domain be T , T= [ t , t + T p To control the cycle. To control the step size, the total optimization time domain T Divided into Each time step; the current running data contains [number] time steps; Electric vehicle charging and discharging resources and For air conditioning load resources, the following behavioral variable system is established: For any electric vehicle At any control moment Its behavioral variables are , indicating the first i electric vehicles t The charging and discharging power at any given moment; like >0 indicates the first i electric vehicles t It is always in a charging state; like <0 indicates that the first i Electric vehicles supporting V2G functionality t Discharges continuously; like =0, indicating the first i electric vehicles t They did not participate in regulation at any time; For any air conditioner At any control moment Its behavioral variables include: , and ;in: Indicates the first j air conditioner t The operating power at any given time; for inverter air conditioners, this value is a continuous variable and... Adjustable within a certain range; for fixed-frequency air conditioners, this value is a fixed value during start-stop operation. Or 0; where, Indicates inverter air conditioner j Minimum operating power Indicates variable frequency air conditioner j Maximum operating power Indicates fixed-frequency air conditioner j Rated operating power; Indicates the first j Indoor temperature under air conditioning control; A binary control variable, representing the first... j air conditioner t The start / stop status at any given time; for fixed-frequency air conditioners, if This indicates that the air conditioner is running. = ;like This indicates that the air conditioner is off. =0; Based on the above behavioral variables, defined at time... t Overall behavior vector for: 。 7. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 4, characterized in that, The system constraints for constructing the multi-objective optimization model include: Establish power grid operation constraints, including: node voltage constraints and feeder and transformer load rate constraints; wherein: For any node n and time t The node voltage constraints are defined as follows: In the formula, For nodes n At any moment t The voltage; and The minimum and maximum voltages specified by the system; The load factor constraints for the feeder and transformer are defined as follows: In the formula, For a moment t The actual load, The rated capacity of the equipment; Establish operating constraints for electric vehicles, including: discharge power boundary constraints, State of Charge (SOC) energy balance constraints, SOC limit constraints, and adjustable time window constraints; among which: For any electric vehicle i At any control moment t Within this context, the discharge power boundary constraints are defined as follows: In the formula, For electric vehicles i During the adjustment period t The charging and discharging power, For electric vehicles i Maximum discharge power, For electric vehicles i Maximum charging power; When electric vehicles support V2G Indicates the maximum discharge power. Indicates the maximum charging power; The SOC energy balance constraint is defined as follows: In the formula, For a moment t+ 1. Battery state of charge For a moment t Battery state of charge, For charging and discharging efficiency, This refers to the battery's rated capacity. The SOC limit constraint is defined as follows: In the formula, The minimum allowable SOC value, This is the highest allowed SOC value; Set up vehicles i In time period If the pause occurs, the adjustable time window constraint is defined as follows: In the formula, The time of connection to the power grid. The time for disconnecting from the power grid, For vehicles i The SOC value at the moment of disconnection from the power grid. The minimum state of charge value set for the end of the stay for electric vehicle users; Establish air conditioning operation constraints, including: power regulation range constraints, room temperature dynamic constraints, temperature comfort range constraints, and start / stop frequency constraints for fixed-frequency air conditioners; among which: For inverter air conditioners, the corresponding power adjustment range constraints are defined as follows: For fixed-frequency air conditioners, the corresponding power adjustment range constraints are defined as follows: In the formula, For the first j air conditioner t Operating power at any given time To indicate inverter air conditioner j Minimum operating power To indicate inverter air conditioner j Maximum operating power For the first j air conditioner t The running status at any given moment, =0 indicates the first j air conditioner t Always in the off state. =1 indicates the first j air conditioner t It is always running; For fixed frequency air conditioners j Rated operating power; The dynamic constraints for room temperature are defined as follows: In the formula, For the first j+ The current room temperature for a user under the control of one air conditioner. For the first j The current room temperature for the user under the control of the air conditioner. To unify the external temperature, For building thermal inertia parameters, The coefficient of performance (COP) is the cooling capacity. For the first j air conditioner t Operating power at any given time; The temperature comfort range constraint is defined as follows: In the formula, For the first j The current room temperature for the user under the control of the air conditioner. The lowest temperature set by the user. The maximum temperature set by the user; The start-stop frequency constraints for fixed-frequency air conditioners are defined as follows: Set the minimum start-stop interval for a fixed-frequency air conditioner. Introducing binary variables Indicates the start / stop status of the air conditioner and defines the start / stop change quantity. for: In the formula, For the first j air conditioner t The start / stop status at any given time. For the first j air conditioner t- Start-stop status at moment 1; The total number of start-stop operations per day shall not exceed Establish the following constraints: The interval between consecutive start-stop operations must not be less than Introducing start-stop change Establish the following constraints: 。 8. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 1, characterized in that, The control step size Δ t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy yields the following results: Based on the preset device differentiation control logic, the first time step [ t , t +Δ t The control strategy is transformed into actual executable control commands; wherein, the device-differentiated control logic includes: Determine if an electric vehicle is connected to the power grid; if so, execute the following control logic: if the device supports bidirectional V2G function, the control strategy is charging and discharging power, and the charging and discharging power is directly mapped to the actual power command and sent to the charging pile to control the vehicle's power behavior; if the device only supports unidirectional charging, the negative power in the control strategy will be automatically set to zero, and only the charging strategy will be retained; wherein, the control command is issued in the form of a set time period or power curve; if not, that is, the vehicle is not connected, is about to leave the station, or has insufficient SOC, the vehicle is determined to be uncontrollable and the control command is automatically blocked. Determine the type of air conditioning load; if it is an inverter model, the control strategy is continuous power quantity, which is converted into an air conditioning temperature setting command according to the set ratio; if it is a fixed frequency model, the control strategy is start-stop state, and a delay judgment is made in combination with start-stop frequency constraints before execution. Embedded basic exception handling logic: if device communication fails, control command is abnormal, or user manually withdraws from participation, the corresponding control weight will be automatically adjusted or the device will be removed in the next round of rolling optimization.

9. The multi-objective optimization autonomous control method for the feeder layer of a distribution network area according to claim 1, characterized in that, The step of dynamically adjusting the model parameters and control weights of the multi-objective optimization model for the current control cycle based on the execution result includes: Real-time acquisition of execution results, including: actual charging and discharging behavior of electric vehicles, air conditioning operation curves, node voltage deviation, and total load response; The execution results are compared and analyzed with the prediction results to identify the sources of prediction and execution deviations; Using the execution results, the model parameters are corrected and the weights of each sub-objective in the optimization objective are adjusted, thereby updating the control strategy within the current control cycle.

10. A multi-objective optimization autonomous control system for the feeder layer of a distribution network, characterized in that, include: The state awareness and prediction module is used to detect and predict the current control cycle Δ t Any adjustment moment within t A forecasting window is established based on the current operational data of flexible power resources. t , t + T p The system models the state variables within the prediction window to obtain the prediction results of the current operating state; wherein, the flexible power resources include: electric vehicle charging and discharging resources and air conditioning load resources; The optimization decision module is used to construct a multi-objective optimization model, taking the prediction results as input to the multi-objective optimization model and outputting the prediction time domain. T p The optimal control strategy within the system; The strategy execution and feedback module is based on the control step size Δ. t According to the preset device differentiation control logic, only the first time step is executed. t , t +Δ t The control strategy is implemented to obtain the execution result; based on the execution result, the model parameters and control weights of the multi-objective optimization model are dynamically corrected in the current control cycle, and the operating state is re-predicted and the control strategy is updated in the next control cycle, forming a multi-objective optimization autonomous control closed-loop control architecture.

Images