Virtual power plant cluster scheduling optical storage cooperative power supply system

Abstract

The invention relates to the field of power system operation and control, and discloses a virtual power plant cluster scheduling optical storage cooperative power supply system, which comprises an optical storage coupling physical unit, an edge data acquisition terminal and a cooperative scheduling server. The server is provided with a whole-network data synchronization module, a dual-channel parameter online identification module, an impedance thermal life nonlinear coupling module and an impedance adaptive cooperative scheduling module. The dual-channel parameter online identification module calculates a voltage sensitivity coefficient and equivalent heat conduction thermal resistance based on the synchronous data; the impedance thermal life nonlinear coupling module determines a forced reactive power support quantity according to the sensitivity coefficient, predicts a steady-state progressive temperature by using thermal resistance and quantifies the battery life loss cost; and the impedance adaptive collaborative scheduling module constructs a multi-objective optimization model in combination with the life cost and the capacity constraint, and generates an optimal power instruction. According to the invention, by establishing a cross-domain coupling mechanism of power grid impedance, equipment thermal characteristics and battery life, optimal control of voltage stability and equipment life in a weak power grid environment is realized.

Description

Technical Field

[0001] This invention relates to the field of power system operation and control, specifically to a virtual power plant cluster dispatching photovoltaic-storage collaborative power supply system. Background Technology

[0002] With the increasing penetration of distributed energy resources, virtual power plant technology is widely used in the aggregation and coordinated control of photovoltaic and energy storage resources to participate in grid ancillary services and energy market transactions. Existing virtual power plant dispatch strategies typically treat voltage stability control on the distribution network side and energy storage battery life management on the user side as independent control links. In actual operation, especially for weak grid nodes located at long electrical distances or with high line impedance, bidirectional converters must undertake a large amount of reactive power support to maintain voltage stability at the point of common coupling. However, existing technologies often neglect the cascading effects of this forced reactive power output demand, determined by grid impedance characteristics, on the internal thermal environment of the equipment, and lack quantitative modeling of the nonlinear coupling relationship across physical domains between grid impedance, converter heat loss, and battery electrochemical aging. This makes it difficult for dispatch systems to accurately assess the additional thermal stress and lifespan reduction caused by power commands in weak grid environments, hindering precise cost quantification.

[0003] Furthermore, in terms of thermal management of integrated photovoltaic and energy storage systems, due to the large heat capacity and thermal inertia of energy storage battery packs, their internal temperature changes exhibit a time lag relative to the step changes in electrical power commands. Existing thermal safety control methods mainly rely on passive threshold judgments based on feedback values ​​from real-time temperature sensors. This protection logic based on instantaneous observations cannot anticipate the final temperature that the current power command is likely to cause when the system reaches thermal equilibrium during the scheduling decision-making stage. This results in a lack of proactive prevention capabilities against overheating risks caused by long-term heat accumulation, easily leading to the equipment operating unknowingly under accelerated aging thermal boundary conditions for extended periods.

[0004] Meanwhile, existing multi-objective optimization scheduling algorithms typically use fixed battery loss models or simplified linear depreciation parameters when calculating operating costs, failing to dynamically adjust based on real-time heat dissipation conditions and the varying strengths of different grid nodes. This homogenized scheduling mode cannot distinguish the actual operating costs of equipment under different conditions, easily leading to over-utilization of photovoltaic and energy storage units located in areas with obstructed heat dissipation channels or extremely weak grids. This accelerates the aging process of equipment at local nodes and reduces the overall utilization efficiency and economy of virtual power plant cluster assets. Summary of the Invention

[0005] This invention provides a virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system, which aims to solve the technical problem in existing virtual power plant scheduling technology that fails to fully consider the cross-physical domain nonlinear coupling relationship between the impedance characteristics of the distribution network, the thermal characteristics of power electronic equipment and the life cost of batteries, resulting in increased thermal risk of equipment and abnormal degradation of battery life in weak grid environments.

[0006] This invention provides a virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system. The system includes multiple distributed photovoltaic-storage coupled physical units, edge data acquisition terminals, and a coordination and scheduling server. The photovoltaic-storage coupled physical units are deployed at different physical nodes in the distribution network to execute power commands and provide operational status feedback. The edge data acquisition terminals are communicatively connected to the photovoltaic-storage coupled physical units to collect operational status data. The coordination and scheduling server is connected to the edge data acquisition terminals via a communication network. The coordination and scheduling server is internally configured with a network-wide data synchronization module, a dual-channel parameter online identification module, an impedance thermal lifetime nonlinear coupling module, and an impedance adaptive coordinated scheduling module.

[0007] The network-wide data synchronization module is used to construct a time sliding window and output a standardized data sequence. The dual-channel parameter online identification module is used to calculate in parallel the voltage sensitivity coefficient characterizing the grid characteristics and the equivalent thermal resistance characterizing the equipment thermal characteristics based on the standardized data sequence. The impedance thermal lifetime nonlinear coupling module is used to receive the voltage sensitivity coefficient and the equivalent thermal resistance, calculate the forced reactive power support according to the grid voltage stability requirements, and then predict the steady-state asymptotic temperature of the photovoltaic-storage coupling physical unit based on the power loss principle, ultimately quantifying the battery's lifetime loss cost. The impedance adaptive collaborative scheduling module is used to construct a multi-objective optimization model based on the lifetime loss cost, generate the optimal active power command and the optimal reactive power command, and send them to the photovoltaic-storage coupling physical unit.

[0008] Furthermore, in the photovoltaic-storage coupling physical unit, the photovoltaic power generation modules and the energy storage battery pack are respectively connected to the DC side of the bidirectional converter, and the AC side of the bidirectional converter is connected to the point of common coupling (PCC) of the distribution network. The bidirectional converter is equipped with sensors to detect the temperature of the power device heat sinks, and the battery management system is equipped with sensors to detect the temperature of the energy storage battery pack modules. The operating status data includes the PCC voltage, instantaneous photovoltaic active power, energy storage active power, energy storage reactive power, bidirectional converter heat sink temperature, and energy storage battery pack module temperature.

[0009] Furthermore, the network-wide data synchronization module performs time-series alignment processing on the collected operating status data based on the network time protocol, and calculates the voltage change, active power change, and reactive power change at adjacent moments within the time sliding window. The generated differential sequence is then transmitted to the dual-channel parameter online identification module to eliminate data analysis errors caused by clock asynchrony.

[0010] Furthermore, the dual-channel parameter online identification module utilizes the random fluctuation characteristics of instantaneous photovoltaic active power as a non-intrusive excitation source to establish a linear regression relationship between the voltage change at the point of common coupling and the power injection change within a time sliding window. By introducing a weighted least squares method to solve the linear regression relationship, the active voltage sensitivity coefficient characterizing the voltage fluctuation amplitude caused by a unit change in active power, and the reactive voltage sensitivity coefficient characterizing the voltage fluctuation amplitude caused by a unit change in reactive power are obtained, thereby enabling real-time sensing of the electrical strength of distribution network nodes.

[0011] Furthermore, the dual-channel parameter online identification module utilizes a lumped-parameter thermal network model and employs a recursive least squares method with a forgetting factor to estimate the equivalent thermal conduction resistance online. The lumped-parameter thermal network model describes the physical process by which the heat generated by the bidirectional converter power devices affects the temperature of the energy storage battery module through the thermal conduction path. By iteratively calculating and minimizing the error between the observed temperature value and the model's predicted value, the model updates the equivalent thermal conduction resistance value in real time to reflect the time-varying characteristics of the device's heat dissipation performance.

[0012] Furthermore, the impedance thermal lifetime nonlinear coupling module includes a forced reactive power support calculation unit. This unit is configured to: when the voltage at the point of common coupling exceeds a preset voltage fluctuation dead zone, use the reactive power voltage sensitivity coefficient as a scaling factor to calculate the reactive power value that must be injected or absorbed to pull the voltage back to the safety boundary, and determine this value as the forced reactive power support amount. This process reflects the rigid constraint of the grid impedance characteristics on the reactive power output of the equipment.

[0013] Furthermore, the impedance thermal lifetime nonlinear coupling module includes a steady-state thermal boundary prediction unit. This unit establishes a cross-physical domain coupling model of active power, reactive power, and temperature. First, it determines the apparent power based on the vector sum of the active power command to be dispatched from the energy storage and the forced reactive power support. Then, it calculates the expected heat loss power of the bidirectional converter based on the nonlinear relationship between apparent power and losses. Subsequently, according to Ohm's thermal law, it superimposes the product of the expected heat loss power and the equivalent thermal conduction resistance onto the currently measured temperature of the bidirectional converter radiator to obtain the steady-state asymptotic temperature. This steady-state asymptotic temperature characterizes the limiting temperature state when the equipment reaches thermal equilibrium under current grid constraints and heat dissipation conditions.

[0014] Furthermore, the impedance thermal lifetime nonlinear coupling module includes a lifetime cost dynamic quantification unit. This unit utilizes a modified Arrhenius model to establish a mapping relationship between temperature and aging rate, mapping the steady-state asymptotic temperature to a lifetime degradation acceleration factor, wherein the lifetime degradation acceleration factor exhibits an exponential nonlinear growth relationship with the steady-state asymptotic temperature. This unit obtains the monetized lifetime loss cost by calculating the product of the battery unit capacity replacement price, the lifetime degradation acceleration factor, the absolute value of the energy storage active power command, and the scheduling cycle duration.

[0015] Furthermore, the impedance adaptive collaborative scheduling module constructs a multi-objective optimization model aimed at minimizing the total operating cost of the entire system. This total operating cost is comprised of the cost of purchasing electricity from the distribution network and the lifetime depreciation cost of all photovoltaic-storage coupled physical units. During the solution process, a bidirectional converter capacity coupling constraint is configured to limit the sum of the square of the active power command of energy storage and the square of the forced reactive power support to not exceeding the square of the maximum apparent power capacity of the bidirectional converter, ensuring that the grid voltage support requirements are prioritized.

[0016] Furthermore, the multi-objective optimization model is also configured with a forward-looking thermal safety constraint, limiting the predicted steady-state asymptotic temperature to not exceed a preset temperature safety threshold, thereby achieving preventative control against thermal faults. The coordinated scheduling server employs a multi-rate timing control strategy, running the impedance adaptive coordinated scheduling module according to the power coordinated scheduling cycle, and running the dual-channel online parameter identification module according to the lower-frequency parameter identification and update cycle.

[0017] This invention provides a virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system. It has the following beneficial effects: 1. This invention utilizes a dual-channel online parameter identification module to acquire in real-time the voltage sensitivity coefficient, which characterizes grid strength, and the equivalent thermal resistance, which characterizes equipment heat dissipation capacity. Based on these parameters, an impedance-thermal-life nonlinear coupling module establishes a quantitative mapping relationship between grid impedance, equipment heat loss, and battery life degradation. This technical solution enables the system to accurately assess the additional thermal stress and lifespan reduction caused by the forced reactive power support required to maintain voltage stability in weak grid environments, achieving refined management of the equipment's full lifecycle cost.

[0018] 2. This invention utilizes a steady-state thermal boundary prediction unit to solve the temperature response lag problem caused by the thermal capacity inertia of energy storage batteries. By establishing a cross-domain physical model from electrical power commands to steady-state asymptotic temperatures, the system can predict the extreme thermal risks that current commands are likely to trigger in the future during the scheduling decision-making stage, rather than passively relying on real-time temperature feedback. The forward-looking thermal safety constraints in the impedance adaptive collaborative scheduling module directly limit this predicted temperature, thereby achieving preventive control of equipment thermal failures.

[0019] 3. This invention constructs a multi-objective optimization model that incorporates nonlinear lifetime loss costs. This model incorporates lifetime loss, determined by grid impedance and equipment heat dissipation conditions, into the objective function as a monetized cost. When making power allocation decisions, the impedance-adaptive collaborative scheduling module automatically balances the cost of electricity purchase with the lifetime loss costs of each node. At nodes with extremely weak grids or poor heat dissipation conditions, the system will proactively reduce their power allocation due to their higher lifetime loss costs. Thus, driven by the goal of economic optimization, differentiated protection is achieved for energy storage assets in different health states and under different external environments within the cluster. Attached Figure Description

[0020] Figure 1 This is the overall system architecture and physical topology diagram of the present invention; Figure 2 This is a block diagram illustrating the core control logic and data flow of the present invention. Figure 3 This is a schematic diagram of the impedance thermal lifetime nonlinear coupling mechanism of the present invention; Figure 4 This is a flowchart of the system collaborative scheduling process of the present invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see the appendix Figure 1 This invention provides a virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system, which includes multiple distributed photovoltaic-storage coupled physical units, edge data acquisition terminals, and a coordination and scheduling server.

[0023] Photovoltaic-storage coupled physical units (PV-SSDUs) are deployed at different physical nodes in the distribution network. Each PV-SSDU includes a photovoltaic (PV) module, a battery storage array, a battery management system (BMS), and a bidirectional converter. The PV module is connected to the DC side of the bidirectional converter, and the battery storage array is connected to the DC side of the bidirectional converter via the BMS. The AC side of the bidirectional converter is connected to the point of common coupling (PCC) of the distribution network. The bidirectional converter is equipped with a radiator temperature sensor to detect the real-time temperature of the radiators of its power devices. The BMS is equipped with a cell temperature sensor to detect the real-time average temperature of the modules inside the battery storage array.

[0024] The edge data acquisition terminal communicates with the battery management system, the bidirectional converter, and the power meters deployed at the common connection point to collect operational status data of the photovoltaic-storage coupling physical unit. The coordination and scheduling server is connected to the edge data acquisition terminal through a communication network to receive operational status data and issue power scheduling commands.

[0025] The virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system is logically divided into a network-wide data synchronization module, a dual-channel parameter online identification module, an impedance thermal lifetime nonlinear coupling module, and an impedance adaptive collaborative scheduling module.

[0026] The network-wide data synchronization module is configured to collect operational status data of each photovoltaic-storage coupled physical unit in real time according to a set sampling period. The operational status data includes the point of common coupling voltage, instantaneous active power of the photovoltaic system, active power of the energy storage system, reactive power of the energy storage system, temperature of the bidirectional converter radiator, and temperature of the energy storage battery module. The network-wide data synchronization module performs time-series alignment processing on the collected operational status data and constructs a time sliding window of a preset length. The network-wide data synchronization module transmits the processed data to the dual-channel parameter online identification module. Here, the first... Each optical-storage coupling physical unit in The state data at any given time includes: point of common coupling voltage. Instantaneous active power of photovoltaic Energy storage active power Energy storage reactive power Bidirectional converter radiator temperature and the temperature of the energy storage battery module .

[0027] The dual-channel online parameter identification module is configured to calculate, in parallel, voltage sensitivity parameters characterizing grid characteristics and thermal resistance parameters characterizing equipment thermal characteristics based on operating status data within a time sliding window. The dual-channel online parameter identification module includes a grid characteristic identification unit and a thermal parameter adaptive unit.

[0028] The grid characteristic identification unit is configured to utilize the fluctuation characteristics of instantaneous photovoltaic active power. Within a time sliding window, it solves the linear regression equation between the change in point of common coupling voltage and the change in power injection using the weighted least squares method, thereby obtaining the first... Active voltage sensitivity coefficient of each photovoltaic-storage coupling physical unit and reactive voltage sensitivity coefficient Active voltage sensitivity coefficient The reactive voltage sensitivity coefficient characterizes the voltage fluctuation amplitude caused by a unit change in active power. It characterizes the voltage fluctuation amplitude caused by a unit change in reactive power.

[0029] The thermal parameter adaptive unit is configured based on a lumped parameter thermal network model, and the recursive least squares method with a forgetting factor is used to estimate the first parameter online. Equivalent thermal resistance of each photoelectric storage coupling physical unit The lumped-parameter thermal network model describes the physical process by which the heat generated by the bidirectional converter affects the temperature of the energy storage battery module through heat conduction. The thermal parameter adaptive unit minimizes the error between observed and predicted temperatures through iterative calculations, and updates the equivalent thermal resistance in real time. The value.

[0030] The impedance-thermal-lifetime nonlinear coupling module is configured to receive the active voltage sensitivity coefficient, reactive voltage sensitivity coefficient, and equivalent thermal conduction resistance, and predict the steady-state asymptotic temperature of the photovoltaic-storage coupling physical unit based on the scheduling instructions to be executed, thereby quantifying the battery's lifespan cost. The impedance-thermal-lifetime nonlinear coupling module includes a forced reactive power support calculation unit, a steady-state thermal boundary prediction unit, and a lifespan cost dynamic quantification unit.

[0031] The forced reactive power support calculation unit is configured to be based on the reactive power voltage sensitivity coefficient. and the preset grid voltage reference value Calculate the amount of forced reactive power required to maintain voltage stability at the point of common coupling. Forced reactive power support The calculation follows the following logic: when the reactive voltage sensitivity coefficient As the value increases, the amount of forced reactive power required to maintain the same voltage deviation. It increases accordingly.

[0032] The steady-state thermal boundary prediction unit is configured to establish a cross-physical domain coupled model of active power, reactive power, and temperature, and calculate the steady-state asymptotic temperature that the photovoltaic-storage coupled physical unit will reach under the current scheduling command. The steady-state thermal boundary prediction unit first determines the active power command of the energy storage to be dispatched. and forced reactive power support Based on the efficiency characteristics of the bidirectional converter, the expected heat loss power of the bidirectional converter is calculated. Expected heat loss power It is positively correlated with the square of apparent power, which is determined by the active power command of energy storage. and forced reactive power support The vector sum is determined. Subsequently, the steady-state thermal boundary prediction unit is based on the currently measured bidirectional converter radiator temperature. The calculated expected heat loss power And the equivalent thermal resistance output by the thermal parameter adaptive unit. Calculate the steady-state asymptotic temperature Steady-state asymptotic temperature The calculation formula is expressed as follows: ; in, express The temperature of the bidirectional converter heatsink is measured at all times. This indicates the expected heat loss power. express The equivalent thermal resistance obtained at any time.

[0033] The lifetime cost dynamic quantization unit is configured to utilize a modified Arrhenius model to quantize the steady-state asymptotic temperature. This is mapped to the battery's lifespan degradation acceleration factor, and the lifespan cost per unit time is calculated. The lifespan cost dynamic quantification unit first calculates the lifespan degradation acceleration factor. The calculation formula is expressed as follows: ; in, Indicates the exponential factor. Indicates activation energy. Represents the ideal gas constant. This indicates the steady-state asymptotic temperature.

[0034] Subsequently, the lifetime cost dynamic quantization unit calculates the lifetime depreciation cost of the i-th optical-storage coupling physical unit. The calculation formula is expressed as follows: ; in, This indicates the replacement unit price per unit capacity of the battery. Represents the absolute value of the active power command for energy storage. Indicates the duration of the scheduling cycle.

[0035] The impedance-adaptive collaborative scheduling module is configured to construct and solve a multi-objective optimization model to generate the final active power dispatching command and reactive power dispatching command. The objective function of the multi-objective optimization model is set to minimize the total operating cost of the entire system, which includes the cost of purchasing electricity from the distribution network and the lifetime depreciation cost of all photovoltaic-storage coupled physical units. During the solution process, the impedance-adaptive collaborative scheduling module must satisfy power balance constraints, bidirectional converter capacity coupling constraints, and steady-state thermal security constraints.

[0036] The bidirectional converter capacity coupling constraint is configured to limit the active power command of energy storage. and forced reactive power support The combination shall not exceed the maximum apparent power capacity of the bidirectional converter. The constraints are expressed as follows: ; Steady-state thermal safety constraints are configured to limit steady-state asymptotic temperatures. Not exceeding the preset temperature safety threshold The constraints are expressed as follows: ; The impedance adaptive collaborative scheduling module uses a nonlinear programming algorithm to solve the multi-objective optimization model, obtains the optimal active power command and optimal reactive power command for each photovoltaic-storage coupled physical unit, and sends them to the corresponding bidirectional converter and battery management system for execution through the edge data acquisition terminal.

[0037] Please see the appendix Figure 2 The whole network status perception and preprocessing module serves as the data entry point for the virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system. It is mainly responsible for the acquisition, cleaning, timing alignment, and feature window construction of the underlying physical signals.

[0038] The network-wide status awareness and preprocessing module establishes communication connections with the battery management system, bidirectional converter controller, and smart power meters at common connection points through edge data acquisition terminals deployed at each photovoltaic-storage coupling physical unit. The module uses the Modbus transmission control protocol or the IEC61850 standard communication protocol as its underlying data transmission standard. To capture minute-level or even second-level power fluctuations in photovoltaic modules caused by cloud cover or irradiance variations, the module sets its data sampling period to seconds or sub-seconds. This high-frequency sampling mechanism ensures sufficient excitation signal bandwidth for subsequent sensitivity identification algorithms.

[0039] The raw physical data collected by the network-wide status perception and preprocessing module includes electrical quantity data and thermal quantity data. Specifically, the electrical quantity data includes: the effective value of the three-phase voltage at the point of common coupling, the real-time output power of the photovoltaic power generation modules on the DC side, the active power output of the bidirectional converter on the AC side, the reactive power output of the bidirectional converter on the AC side, and the state of charge of the energy storage battery pack. The thermal quantity data includes: the surface temperature of the heat dissipation substrate of the bidirectional converter's internal insulated gate bipolar transistor module, the average temperature of each battery module inside the energy storage battery pack, and the ambient temperature outside the integrated photovoltaic and energy storage cabinet. Among these, the surface temperature data of the bidirectional converter's heat dissipation substrate directly reflects the real-time losses and heat generation of the power devices, while the average temperature data of each battery module inside the energy storage battery pack serves as the basic physical boundary condition for assessing the battery aging rate.

[0040] To address the data timing misalignment issue caused by clock asynchrony among devices in a distributed network environment, the network-wide status awareness and preprocessing module performs time synchronization and data alignment operations based on the network time protocol. Using the standard system time of the coordination and scheduling server as a benchmark, the module re-timestamps each uploaded data packet. When the module detects data packet loss due to communication link jitter, it does not discard that moment directly. Instead, it uses a linear interpolation algorithm to calculate the average rate of change using valid data from the previous and next moments, thus filling in the missing values ​​and ensuring the continuity of the data sequence in the time dimension. Furthermore, the module is equipped with an amplitude limiting filter to remove spike noise data exceeding physically reasonable ranges caused by sensor malfunctions or electromagnetic interference.

[0041] To accommodate the batch data requirements of the subsequent least squares algorithm, the network-wide state perception and preprocessing module constructs an independent time-sliding window queue for each optical-storage coupling physical unit in memory space. The module maintains the time-sliding window queue using a first-in, first-out (FIFO) strategy. Each time the module completes a new sampling and processing cycle, it pushes the latest state vector into the head of the queue and simultaneously removes the oldest set of state vectors from the tail of the queue, thus maintaining a fixed number of historical data frames in the queue. The module calculates the voltage, active power, and reactive power changes between adjacent moments within the time-sliding window and packages these difference sequences along with the original state sequences as a standardized input matrix, which is then passed to the dual-channel parameter online identification module.

[0042] The power grid characteristic identification unit is configured to analyze the power grid impedance characteristics at the point of common coupling in real time using a non-intrusive passive observation method.

[0043] The grid characteristic identification unit receives the time-sliding window data sequence output by the network-wide state perception and preprocessing module. To avoid injecting additional artificial disturbance signals into the distribution network and thus affecting power quality, the grid characteristic identification unit directly utilizes the random fluctuation characteristics of the photovoltaic power generation module's output power as the excitation source for system identification. Due to the influence of meteorological factors such as cloud movement and atmospheric turbulence, the output power of the photovoltaic power generation module exhibits random changes on a second-level time scale. This change causes corresponding small fluctuations in the voltage at the point of common coupling. The grid characteristic identification unit captures this dynamic response relationship between power change and voltage change, thereby inferring the electrical strength parameters of the grid.

[0044] Based on the physical model of the distribution network lines, the power grid characteristic identification unit establishes a linearized sensitivity equation between voltage and power changes. At each sampling moment covered by the time sliding window, the unit assumes that the change in voltage at the point of common coupling (PCC) is approximately equal to the change in active power multiplied by the active voltage sensitivity coefficient, plus the change in reactive power multiplied by the reactive voltage sensitivity coefficient. The active voltage sensitivity coefficient physically corresponds to the equivalent resistance component between the PCC and the main grid power source, while the reactive voltage sensitivity coefficient physically corresponds to the equivalent reactance component between the PCC and the main grid power source.

[0045] Since the time-sliding window contains multiple consecutive sampling moments, the power grid characteristic identification unit extends the above linearized sensitivity equation into an overdetermined set of matrix equations. In the overdetermined set of equations, the observation vector consists of the voltage changes at each moment within the time-sliding window, the data matrix consists of the active power changes and reactive power changes at each moment within the time-sliding window, and the parameter vector to be identified consists of the active voltage sensitivity coefficient and the reactive voltage sensitivity coefficient.

[0046] Considering the susceptibility of distribution network topology to reconfiguration and load level drift, the impedance characteristics of the power grid exhibit time-varying characteristics. To improve the identification algorithm's ability to track the current power grid state, the power grid characteristic identification unit introduces a weighted least squares method with exponentially decaying weights for solution. The unit constructs a diagonal weight matrix, in which data samples corresponding to newer times on the time axis are assigned larger weight values, while data samples corresponding to older times are assigned smaller weight values. Through this weighting mechanism, the power grid characteristic identification unit can quickly respond to abrupt changes in power grid parameters while suppressing the interference of outdated data on the current estimation results.

[0047] The grid characteristic identification unit performs weighted least squares optimization to minimize the sum of squared residuals between the observed and predicted voltage changes. After obtaining the optimal solution through matrix operations, the unit outputs the active and reactive voltage sensitivity coefficients for the current moment. Larger sensitivity coefficients indicate a weak grid region with a long electrical distance from the point of common coupling (PCC), where even a small power injection can cause significant voltage fluctuations. Conversely, smaller sensitivity coefficients indicate a strong grid region with strong electrical support. The unit uses these two sensitivity coefficients as key grid constraint parameters and passes them to the subsequent impedance thermal lifetime nonlinear coupling module.

[0048] The thermal parameter adaptive unit is configured to track and calibrate the thermodynamic characteristic parameters inside the photovoltaic storage unit in real time to address thermal resistance drift caused by equipment aging, dust accumulation in the air duct, or performance degradation of the cooling fan.

[0049] The adaptive thermal parameter unit establishes a physical model of the thermal coupling between the bidirectional converter and the energy storage battery pack based on the lumped-parameter thermal network theory. In compact photovoltaic-energy storage devices, since the bidirectional converter and the energy storage battery pack share the same physical cavity or are tightly connected through metal structural components, the high heat generated by the power devices of the bidirectional converter is transferred to the energy storage battery pack through heat conduction, heat convection, and heat radiation, thereby changing the thermal boundary conditions of the energy storage battery pack. The adaptive thermal parameter unit treats the energy storage battery pack as a uniformly heated hot node and ignores the temperature gradient distribution inside the battery pack. The adaptive thermal parameter unit establishes a thermal balance equation based on the law of conservation of energy, which describes that the rate of change of thermal energy of the energy storage battery pack is equal to the total thermal power input to the energy storage battery pack minus the thermal power dissipated by the energy storage battery pack to the external environment. Among them, the total input thermal power mainly consists of the portion of the heat loss power of the bidirectional converter transferred through thermal resistance and the Joule heat of the battery pack's internal resistance.

[0050] To enable numerical computation, the adaptive thermal parameter unit transforms the continuous-time differential equations into discrete-time difference equations. The unit defines the ratio of the temperature difference between the current and previous battery module temperatures to the time sampling interval as approximately representing the rate of temperature change. The unit treats the bidirectional converter radiator temperature as the heat source boundary temperature and the equivalent thermal resistance as the thermal path parameter connecting the heat source and the battery node. In the difference equations, the equivalent thermal resistance is an unknown state variable that changes slowly over time, while the battery heat capacity is treated as a known, fixed physical constant in the calculations.

[0051] The thermal parameter adaptive unit employs a recursive least squares algorithm with a forgetting factor to estimate the equivalent thermal resistance online. The unit constructs a linear regression observation model, using the currently observed battery module temperature as the system output and a vector composed of the previous battery module temperature, bidirectional converter radiator temperature, and total power loss as the regression vector. In each sampling period, the thermal parameter adaptive unit first calculates the prior predicted value of the battery module temperature based on the previous thermal resistance estimate, and then calculates the prediction error between the measured battery module temperature and the prior predicted value.

[0052] The adaptive thermal parameter unit corrects the previous thermal resistance estimate using prediction error and the gain matrix to obtain the latest equivalent thermal conduction resistance estimate for the current moment. To adapt to the time-varying characteristics of heat dissipation conditions in actual operation, such as the gradual increase in thermal resistance due to filter blockage, the adaptive thermal parameter unit introduces an exponential forgetting factor into the algorithm. The exponential forgetting factor assigns higher weight to the latest temperature observation data while geometrically decaying the weight of historical data, thus enabling the identification results to quickly track the real trend of thermal resistance parameter changes and preventing the estimate from stagnating due to data saturation. The adaptive thermal parameter unit outputs the converged equivalent thermal conduction resistance to the steady-state thermal boundary prediction unit in real time, ensuring that subsequent thermal prediction models are always calculated based on the current actual heat dissipation performance of the equipment.

[0053] Please see the appendix Figure 3 The forced reactive power support calculation unit is configured to convert the voltage stability constraint of the common coupling point into the reactive power output demand of the bidirectional converter, thereby determining the reactive power component that must be pre-occupied in the capacity of the bidirectional converter.

[0054] The forced reactive power support calculation unit receives the real-time reactive voltage sensitivity coefficient output by the grid characteristic identification unit and the real-time voltage data of the point of common coupling collected by the network-wide status perception and preprocessing module. The forced reactive power support calculation unit internally stores preset grid operating voltage reference values ​​and allowable voltage fluctuation dead zone ranges. The voltage fluctuation dead zone range defines the upper and lower limits of voltage safety during steady-state operation of the distribution network.

[0055] The forced reactive power support calculation unit first calculates the voltage deviation between the real-time voltage at the point of common coupling and the voltage reference value. When the voltage deviation is within the voltage fluctuation dead zone, the forced reactive power support calculation unit sets the forced reactive power support amount to zero, and the bidirectional converter does not need to perform the forced voltage support task. When the voltage deviation exceeds the voltage fluctuation dead zone, the forced reactive power support calculation unit starts the reverse calculation logic.

[0056] In the reverse calculation logic, the forced reactive power support calculation unit uses the reactive voltage sensitivity coefficient as a scaling factor to calculate the amount of reactive power that must be injected or absorbed to pull the point of common coupling voltage back from its current over-limit value to the voltage safety boundary. Specifically, the forced reactive power support calculation unit divides the voltage deviation to be eliminated by the current reactive voltage sensitivity coefficient, and the quotient is the minimum forced reactive power support required to maintain voltage safety.

[0057] This calculation process reflects the rigid constraint of grid strength on equipment output: at weak grid nodes with long electrical distances, due to the high line impedance, the injection of photovoltaic active power can easily cause voltage rise. To suppress this voltage exceedance, the bidirectional converter must undertake specific inductive reactive power absorption or capacitive reactive power generation tasks. The forced reactive power support calculated by the forced reactive power support calculation unit does not participate in subsequent economic optimization selection, but is directly input to the steady-state thermal boundary prediction unit as a hard boundary condition that must be met. This means that the portion of the apparent power capacity of the bidirectional converter corresponding to the forced reactive power support will be locked for grid voltage stabilization, and only the remaining capacity and thermal capacity space can be used for active power dispatch.

[0058] To address the time-scale mismatch between the thermal response of energy storage batteries and electrical dispatch commands, the steady-state thermal boundary prediction unit is configured to establish a steady-state mapping model from the electrical domain to the thermal domain, thereby predicting the extreme thermal risks that current dispatch commands may trigger in the future.

[0059] The steady-state thermal boundary prediction unit receives the forced reactive power support from the forced reactive power support calculation unit, the equivalent thermal conduction resistance from the thermal parameter adaptive unit, and the real-time temperature of the bidirectional converter heat sink from the whole-network state perception and preprocessing module. Simultaneously, the steady-state thermal boundary prediction unit inputs the energy storage active power command to be optimized as an independent variable into the model.

[0060] The steady-state thermal boundary prediction unit first calculates the expected heat loss power of the bidirectional converter when executing a specific power command, based on the physical loss characteristics of power semiconductor devices. Then, it calculates the total apparent power of the bidirectional converter through vector synthesis, based on the active power command value to be optimized for energy storage and the forced reactive power support value. Since the conduction and switching losses of the insulated-gate bipolar transistors inside the bidirectional converter are positively correlated with the current amplitude, and the current amplitude is determined by the apparent power, the steady-state thermal boundary prediction unit uses a quadratic polynomial function to fit the nonlinear relationship between the apparent power and the expected heat loss power. Specifically, the expected heat loss power value equals the no-load base loss coefficient, plus the linear loss coefficient multiplied by the apparent power value, plus the square loss coefficient multiplied by the square of the apparent power value. This calculation step physically couples the rigid reactive power demand on the grid side with the active power dispatch demand on the user side at the heat loss level.

[0061] Subsequently, the steady-state thermal boundary prediction unit performs asymptotic temperature prediction calculations based on steady-state assumptions. Considering the large thermal inertia of the energy storage battery pack, its real-time temperature cannot instantaneously reflect the thermal shock caused by power jumps. Therefore, the steady-state thermal boundary prediction unit does not calculate the instantaneous temperature at the next moment, but instead calculates the final stable temperature when the system reaches thermal equilibrium, i.e., the steady-state asymptotic temperature. Based on Ohm's law of thermal motion, the steady-state thermal boundary prediction unit multiplies the expected heat loss power by the equivalent thermal resistance value identified in real-time by the thermal parameter adaptive unit to obtain the expected temperature rise value conducted from the bidirectional converter to the energy storage battery pack. Finally, the steady-state thermal boundary prediction unit adds the real-time temperature value of the bidirectional converter heat sink to the expected temperature rise value to obtain the steady-state asymptotic temperature value.

[0062] The steady-state asymptotic temperature value characterizes the worst thermal boundary environment that the energy storage battery pack will eventually face under current grid strength constraints and heat dissipation conditions, if the current active power command for energy storage continues to be executed. In weak grid scenarios, the larger forced reactive power support leads to an increase in expected heat loss power, thus increasing the calculated steady-state asymptotic temperature value. Similarly, in scenarios with obstructed heat dissipation channels, a larger equivalent thermal conduction resistance will also lead to an increase in the steady-state asymptotic temperature value. The steady-state thermal boundary prediction unit outputs this steady-state asymptotic temperature value, which contains future risk information, to the lifetime cost dynamic quantification unit as a physical basis for assessing the degree of battery aging acceleration.

[0063] The dynamic quantification unit configuration for lifetime cost is designed to construct an economic mapping relationship between the lifetime loss of electrochemical energy storage batteries and operating temperature and power throughput, transforming the implicit physical aging process into an explicit monetized cost function.

[0064] The lifetime cost dynamic quantification unit receives the steady-state asymptotic temperature value output by the steady-state thermal boundary prediction unit and performs calculations in conjunction with the active power command value of the energy storage to be scheduled. The lifetime cost dynamic quantification unit has built-in Arrhenius model parameters describing the electrochemical aging mechanism of lithium iron phosphate batteries or ternary lithium batteries, including the forward exponent factor, activation energy constant, and ideal gas constant.

[0065] The lifetime cost dynamic quantification unit first calculates the lifetime degradation acceleration factor, which characterizes the battery aging rate. Based on the Arrhenius equation, the unit determines that the lifetime degradation acceleration factor exhibits an exponential nonlinear growth relationship with the steady-state asymptotic temperature. Specifically, the lifetime degradation acceleration factor is equal to the forward exponential factor multiplied by the output value of the natural exponential function, where the exponent is the negative activation energy constant divided by the product of the ideal gas constant and the steady-state asymptotic temperature. This calculation logic indicates that when the steady-state asymptotic temperature increases due to the high heat loss of the bidirectional converter, the rate of side reactions inside the battery will accelerate dramatically in an exponential manner, leading to thickening of the solid electrolyte interface film and increased loss of active lithium ions.

[0066] After obtaining the lifetime degradation acceleration factor, the lifetime cost dynamic quantification unit calculates the lifetime loss cost per unit scheduling cycle. The lifetime loss cost is equal to the product of the battery unit capacity replacement price, the lifetime degradation acceleration factor, the absolute value of the energy storage active power command, and the scheduling cycle duration. In this calculation process, since the input steady-state asymptotic temperature value is itself a function of the energy storage active power command value (determined by the steady-state thermal boundary prediction unit), the lifetime loss cost and the energy storage active power command no longer follow a simple linear relationship, but instead exhibit a convex function characteristic where the cost increases sharply with increasing power.

[0067] In weak grid nodes with long electrical distances, the increased heat generation of bidirectional converters due to forced reactive power support demands leads to higher steady-state asymptotic temperatures. Consequently, the lifetime cost calculated by the lifetime cost dynamic quantification unit will be higher than that of strong grid nodes. The lifetime cost dynamic quantification unit outputs this lifetime cost function, which includes nonlinear thermal feedback characteristics, to the impedance adaptive collaborative scheduling module as a penalty term in the multi-objective optimization algorithm. This forces the scheduling system to automatically reduce the frequency of active power allocation to nodes with poor heat dissipation conditions or extremely weak grids under the drive of economic optimization, thus achieving passive protection of the photovoltaic and energy storage equipment's lifetime.

[0068] The multi-objective impedance adaptive scheduling module is configured to calculate the optimal power allocation scheme for each photovoltaic-storage coupled physical unit in the virtual power plant cluster under the dual physical constraints of grid voltage stability and equipment thermal safety.

[0069] The multi-objective impedance adaptive scheduling module receives the nonlinear lifetime loss cost function output from the lifetime cost dynamic quantification unit, the forced reactive power support value output from the forced reactive power support calculation unit, and obtains time-of-use electricity price information and load demand forecast data from the external power grid trading platform. The module constructs a comprehensive objective function that includes economic and physical loss indicators. The mathematical optimization direction of this comprehensive objective function is set to minimize the total operating cost of the entire system. The total operating cost of the entire system consists of the sum of two independent parts: the first part is the electricity purchase cost incurred by the virtual power plant in purchasing electricity from the main grid, which is equal to the product of the total exchange power at the point of common coupling and the current time-of-use electricity price; the second part is the sum of the expected lifetime loss costs of all photovoltaic-storage coupled physical units within the cluster. This part is calculated directly using the function calculation result generated by the preceding module, which includes the thermal-lifetime nonlinear mapping relationship.

[0070] To ensure the executability and security of the scheduling results at the physical level, the multi-objective impedance adaptive scheduling module sets multi-dimensional boundary constraints, including power balance constraints, bidirectional converter capacity coupling constraints, and forward-looking thermal safety constraints.

[0071] The power balance constraint set by the multi-objective impedance adaptive scheduling module requires that the algebraic sum of the output power of all photovoltaic power generation modules in the cluster, the charging and discharging power of all energy storage systems, and the switching power of the main grid interconnection line must always be equal to the total load demand power in the area in order to maintain real-time energy conservation within the microgrid.

[0072] The multi-objective impedance adaptive scheduling module is configured with bidirectional converter capacity coupling constraints to handle the competition between active and reactive power in apparent power capacity. Given that the power transmission capacity of the bidirectional converter is limited by its nominal maximum apparent power capacity, and that some capacity space is rigidly occupied by forced reactive power support, the multi-objective impedance adaptive scheduling module strictly limits the value of the energy storage active power command to be solved. Specifically, the constraint logic is: the sum of the square of the energy storage active power command value and the square of the forced reactive power support value must not exceed the square of the nominal maximum apparent power capacity of the bidirectional converter. This constraint mechanism ensures that at weak grid nodes with long electrical distances, the multi-objective impedance adaptive scheduling module automatically reduces the active power scheduling quota to prioritize reserving sufficient capacity space to meet the reactive power requirements for voltage support.

[0073] The forward-looking thermal safety constraint configuration set by the multi-objective impedance adaptive scheduling module is based on predictive results for preventative control. Unlike traditional methods that only restrict the current real-time temperature, the multi-objective impedance adaptive scheduling module directly constrains the steady-state asymptotic temperature value output by the steady-state thermal boundary prediction unit. The multi-objective impedance adaptive scheduling module requires that the steady-state asymptotic temperature value corresponding to each opto-storage coupling physical unit must be strictly less than or equal to a preset device temperature safety threshold. This means that if a high-power scheduling command has not yet triggered an over-temperature alarm at the current moment, but the model predicts that the command will cause the system temperature to exceed the limit when it reaches thermal equilibrium in the future, the multi-objective impedance adaptive scheduling module will determine that the command is invalid and discard it, thereby achieving preventative interception of thermal faults.

[0074] The multi-objective impedance adaptive scheduling module uses a nonlinear programming mathematical algorithm to iteratively solve the constructed multivariate optimization model. Since the objective function includes an exponential lifetime cost term and the constraints include a quadratic capacity constraint term, the module searches for the global optimum using sequential quadratic programming or the interior-point method. Upon convergence, the module outputs the optimal active power command and optimal reactive power command values ​​for each optical-storage coupling physical unit at the current moment, and sends these commands to the edge data acquisition terminal via the communication network for execution, completing one closed-loop scheduling control cycle.

[0075] The signal interaction process described in this embodiment of the invention constructs a closed-loop control loop that extends from the physical perception layer upwards to the cloud decision layer, and then back down from the cloud decision layer to the physical execution layer.

[0076] The signal flow originates in the underlying sensing network of the photovoltaic-storage coupling physical unit. Heat sink temperature sensors deployed on the surface of the power device substrate inside the bidirectional converter sense changes in heat flux of the power semiconductor devices in real time and convert analog temperature signals into digital temperature signals, which are then transmitted to the main controller of the bidirectional converter. A network of cell temperature sensors deployed inside the energy storage battery pack collects temperature distribution data from each battery module in real time and aggregates the data to the battery management system. Simultaneously, smart power meters installed at the point of common connection monitor the three-phase voltage amplitude, frequency, and phase information on the distribution network side in real time.

[0077] As a field-level data aggregation node, the edge data acquisition terminal periodically polls and reads register data from the battery management system, bidirectional converter main controller, and smart power meters via the fieldbus interface. The edge data acquisition terminal encapsulates the read voltage, active power, reactive power, radiator temperature, and battery module temperature data into operating status messages with a unified time stamp, and uploads them to the coordination and scheduling server via an encrypted communication channel.

[0078] After the signal enters the coordination and scheduling server, it first flows into the network-wide status awareness and preprocessing module. This module unpacks the original message, removes outliers, and aligns the timeline to generate a standardized historical status time series. This historical status time series is then routed to the input port of the dual-channel parameter online identification module.

[0079] The dual-channel online parameter identification module processes the input data in parallel. On one hand, the power grid characteristic identification unit performs regression analysis using voltage and power sequences to output active voltage sensitivity coefficients and reactive voltage sensitivity coefficients that characterize the power grid strength. On the other hand, the thermal parameter adaptive unit performs iterative thermal network calculations using temperature and power sequences to output the equivalent thermal resistance that characterizes the heat dissipation performance of the equipment.

[0080] The identified physical characteristic parameters serve as intermediate control signals, flowing into the impedance thermal lifetime nonlinear coupling module. The forced reactive power support calculation unit combines the reactive power voltage sensitivity coefficient with real-time voltage data to generate a forced reactive power support signal. The steady-state thermal boundary prediction unit receives the forced reactive power support signal, the equivalent thermal conduction resistance signal, and the real-time radiator temperature signal, and, combined with the power command to be evaluated, outputs a steady-state asymptotic temperature prediction signal. The lifetime cost dynamic quantization unit further converts the steady-state asymptotic temperature prediction signal into a lifetime loss cost function signal.

[0081] The lifetime depreciation cost function signal, along with the forced reactive power support signal, is ultimately fed into the impedance adaptive collaborative scheduling module. This module combines externally input electricity price signals and load forecast signals to perform multi-objective optimization calculations, generating a final scheduling command message containing optimal active power setpoints and optimal reactive power setpoints.

[0082] The coordination and scheduling server sends the final scheduling instruction message back to the edge data acquisition terminal. After parsing the message, the edge data acquisition terminal writes the control instructions to the bidirectional converter main controller and the battery management system via the fieldbus. Based on the received active power setpoint and reactive power setpoint, the bidirectional converter main controller adjusts the duty cycle of its internal pulse width modulation signal, thereby changing the actual current vector injected into the distribution network by the photovoltaic-storage coupling physical unit, completing a full signal closed-loop interaction.

[0083] Please see the appendix Figure 4 The virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system adopts a multi-timescale nested rolling timing control strategy to coordinate the rate difference between the high-frequency fluctuation characteristics of the underlying physical signal and the quasi-steady-state characteristics of the upper-level economic scheduling optimization.

[0084] The virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system establishes a three-tiered time base: data sampling period, parameter identification and update period, and power collaborative scheduling period. The data sampling period is set to milliseconds or seconds, driven by the hardware clock of the edge data acquisition terminal. The parameter identification and update period is set to minutes, matching the convergence speed of the dual-channel online parameter identification module. The power collaborative scheduling period is set to minutes to quarter-hours, adapting to the trading and settlement periods of the electricity market or the time constant of distribution network voltage regulation.

[0085] At the arrival of each data sampling cycle, the edge data acquisition terminal synchronously triggers the reading action of all sensors across the network to obtain a snapshot of the current instantaneous voltage, current, and temperature. The network-wide status perception and preprocessing module receives this snapshot data and stores it in a first-in-first-out time sliding window buffer queue. At this time, the system does not immediately trigger complex optimization scheduling calculations, but maintains the current active power command and reactive power command unchanged until the trigger time of the next power coordinated scheduling cycle is reached.

[0086] The dual-channel online parameter identification module runs in parallel in the background, with the parameter identification update cycle as the tick. Whenever the accumulation of new data frames in the time-sliding window buffer queue reaches a preset number or a specific variance excitation condition is met, the dual-channel online parameter identification module initiates a weighted least squares operation and a recursive least squares operation. After the operation is completed, the dual-channel online parameter identification module updates the values ​​of active power voltage sensitivity coefficient, reactive power voltage sensitivity coefficient, and equivalent thermal resistance stored in the system's internal registers. Because changes in grid topology and equipment thermal resistance drift are slow dynamic processes, the frequency of the parameter identification update cycle is lower than or equal to the frequency of the power coordination dispatch cycle, ensuring that the physical parameters stored in the registers always reflect the latest environmental characteristics of the system.

[0087] The impedance-adaptive collaborative scheduling module strictly executes optimization tasks according to the power collaborative scheduling cycle. When the trigger signal of the power collaborative scheduling cycle arrives, the impedance-adaptive collaborative scheduling module first reads the latest active voltage sensitivity coefficient, reactive voltage sensitivity coefficient, and equivalent thermal conduction resistance from the system register. It is worth noting that these parameters were calculated in the previous parameter identification and update cycle, but are considered to be the optimal estimates at the current moment.

[0088] Subsequently, the impedance adaptive collaborative scheduling module calls the impedance thermal lifetime nonlinear coupling module to construct an optimization model that includes steady-state thermal prediction and lifetime cost assessment based on the read parameters and current real-time status data. The impedance adaptive collaborative scheduling module completes the nonlinear programming solution within a limited computation time, generating a new round of optimal active power commands and optimal reactive power commands.

[0089] The coordination and scheduling server sends newly generated instructions to the photovoltaic-storage coupled physical units. Upon receiving a new instruction, the photovoltaic-storage coupled physical unit continuously tracks the new instruction throughout the next power coordination scheduling cycle until a new instruction is sent again. Through this timing design, the virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system achieves asynchronous coordination of fast sampling monitoring, slow parameter calibration, and periodic optimization control. This ensures both the ability to detect sudden voltage fluctuations in the power grid and avoids the waste of computing resources and mechanical oscillations of equipment caused by excessively frequent adjustments.

[0090] This invention provides an electronic device that serves as a physical carrier for a coordination and scheduling server or an edge computing gateway, used to perform data processing and logical operation tasks involved in a virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system.

[0091] At the physical hardware level, electronic devices mainly consist of a central processing unit (CPU), system memory, large-capacity non-volatile memory, communication interfaces, and an internal bus architecture. The internal bus architecture, serving as a high-speed data transmission channel, interconnects the CPU, system memory, large-capacity non-volatile memory, and communication interfaces, enabling the exchange of electrical signals and the transmission of control commands between these hardware components.

[0092] The central processing unit (CPU) is the core of an electronic device, configured to parse and execute computer program instructions. CPUs are implemented using general-purpose microprocessors, digital signal processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other forms of programmable logic devices. CPUs possess multi-threaded parallel computing capabilities, sufficient to meet the computational demands of the dual-channel parameter online identification module for large-scale matrix operations and the impedance adaptive collaborative scheduling module for nonlinear programming solutions.

[0093] The system memory mainly consists of random access memory and read-only memory, providing high-speed temporary data read / write space and firmware storage space for the central processing unit. The system memory contains the operating system kernel and application instances running the virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system.

[0094] The large-capacity non-volatile memory is configured to store the operating system image, database files, and application software packages. Specifically, in this embodiment, the large-capacity non-volatile memory stores computer program code that implements the functions of the network-wide status awareness and preprocessing module, the dual-channel parameter online identification module, the impedance thermal lifetime nonlinear coupling module, and the impedance adaptive collaborative scheduling module. When the electronic device starts up, the central processing unit reads the aforementioned computer program code from the large-capacity non-volatile memory and loads it into the system memory for execution. This instantiates the aforementioned functional modules at the logical level and completes data acquisition, parameter identification, thermal risk prediction, and collaborative scheduling operations according to a predetermined timing logic.

[0095] The communication interface is configured to enable physical connection between electronic devices and the external network environment. The communication interface includes an Ethernet interface, fiber optic communication interface, wireless LAN interface, or mobile communication module. Through the communication interface, the electronic devices establish bidirectional data transmission links with edge data acquisition terminals distributed across the power distribution network, optical-storage coupling physical units, and external power trading platforms, receiving real-time operating status data and sending power dispatch commands.

[0096] This invention also provides a computer-readable storage medium. The computer-readable storage medium is a non-transient storage medium, including but not limited to hard disk drives, solid-state drives, optical disks, Universal Serial Bus flash drives, or secure digital storage cards. The computer-readable storage medium stores computer program instructions. When the computer program instructions are read and executed by the central processing unit of an electronic device, the electronic device performs operations including the following steps: controlling the network-wide status perception and preprocessing module to collect and clean the operating status data of the optical-storage coupling physical units; controlling the dual-channel parameter online identification module to identify the active power voltage sensitivity coefficient, reactive power voltage sensitivity coefficient, and equivalent thermal conduction resistance based on the operating status data; controlling the impedance thermal lifetime nonlinear coupling module to calculate the forced reactive power support, predict the steady-state asymptotic temperature, and quantify the lifetime loss cost; and controlling the impedance adaptive collaborative scheduling module to generate the optimal active power command and the optimal reactive power command based on a multi-objective optimization model, and send them to the optical-storage coupling physical units through a communication interface.

Claims

1. A virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system, characterized in that, include: Multiple photovoltaic-storage coupled physical units are deployed at distribution network nodes and equipped with bidirectional converters and battery management systems. Edge data acquisition terminal, used to collect operational status data; The coordination and scheduling server is communicatively connected to the edge data acquisition terminal and is equipped with a full network data synchronization module, a dual-channel parameter online identification module, an impedance thermal lifetime nonlinear coupling module, and an impedance adaptive collaborative scheduling module. The network-wide data synchronization module is used to construct a time sliding window and output a standardized data sequence; The dual-channel parameter online identification module is used to identify the voltage sensitivity coefficient and equivalent thermal resistance based on the standardized data sequence. The impedance adaptive collaborative scheduling module is used to generate optimal active and reactive power commands based on the lifetime loss cost, and send them to the optical-storage coupling physical unit through the terminal to control the bidirectional converter to perform power regulation; The photovoltaic-storage coupling physical unit is used to control the bidirectional converter to transmit power to the distribution network point of common coupling in response to the optimal active and reactive power commands.

2. The virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system according to claim 1, characterized in that, In the photovoltaic-storage coupling physical unit, the photovoltaic power generation module and the energy storage battery pack are respectively connected to the DC side of the bidirectional converter, and the AC side of the bidirectional converter is connected to the common connection point of the distribution network. The bidirectional converter is equipped with a sensor to detect the temperature of the heat sink of the power device, and the battery management system is equipped with a sensor to detect the temperature of the energy storage battery module. The operating status data includes the voltage at the point of common coupling, instantaneous active power of the photovoltaic system, active power of the energy storage system, reactive power of the energy storage system, temperature of the bidirectional converter heat sink, and temperature of the energy storage battery module.

3. The virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system according to claim 1, characterized in that, The network-wide data synchronization module is configured to perform time-series alignment on the operating status data, and calculate the voltage change, active power change, and reactive power change at adjacent moments within the time sliding window, generating a differential sequence that is transmitted to the dual-channel parameter online identification module.

4. The virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system according to claim 1, characterized in that, The dual-channel parameter online identification module uses the fluctuation characteristics of photovoltaic instantaneous active power as excitation to establish a linear regression relationship between the change in voltage at the point of common coupling and the change in power injection within the time sliding window. By solving the linear regression relationship using the weighted least squares method, we obtain the active voltage sensitivity coefficient, which characterizes the voltage fluctuation caused by changes in active power, and the reactive voltage sensitivity coefficient, which characterizes the voltage fluctuation caused by changes in reactive power.

5. The virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system according to claim 1, characterized in that, The dual-channel parameter online identification module uses a lumped parameter thermal network model and employs a recursive least squares method with a forgetting factor to estimate the equivalent thermal resistance online. The lumped-parameter thermal network model describes the physical process by which the heating of the bidirectional converter affects the temperature of the energy storage battery module through thermal conduction, and updates the equivalent thermal resistance by minimizing the error between the observed temperature value and the model prediction value.

6. The virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system according to claim 1, characterized in that, The impedance thermal lifetime nonlinear coupling module includes a forced reactive power support calculation unit. The forced reactive power support calculation unit is configured to: when the voltage at the point of common coupling exceeds the voltage fluctuation dead zone, use the reactive voltage sensitivity coefficient as a scaling factor to calculate the reactive power value required to pull the voltage back to the safety boundary, and determine the reactive power value as the forced reactive power support amount.

7. A virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system according to claim 6, characterized in that, The impedance thermal lifetime nonlinear coupling module includes a steady-state thermal boundary prediction unit. The steady-state thermal boundary prediction unit is configured to: determine the apparent power based on the vector sum of the active power command to be scheduled and the forced reactive power support, and calculate the expected heat loss power of the bidirectional converter based on the apparent power; The steady-state asymptotic temperature is obtained by superimposing the product of the expected heat loss power and the equivalent thermal resistance onto the currently measured bidirectional converter radiator temperature.

8. A virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system according to claim 7, characterized in that, The impedance thermal lifetime nonlinear coupling module includes a lifetime cost dynamic quantization unit. The lifetime cost dynamic quantization unit is configured to use a modified Arrhenius model to map the steady-state asymptotic temperature into a lifetime decay acceleration factor that has an exponential nonlinear relationship with temperature. The lifespan loss cost is obtained by multiplying the unit replacement price of battery capacity, the lifespan degradation acceleration factor, the absolute value of the energy storage active power command, and the scheduling cycle duration.

9. A virtual power plant cluster scheduling photovoltaic-storage collaborative power supply system according to claim 8, characterized in that, The impedance adaptive collaborative scheduling module constructs a multi-objective optimization model with the goal of minimizing the sum of the power purchase cost and the lifetime depreciation cost; The multi-objective optimization model is configured with bidirectional converter capacity coupling constraints. The bidirectional converter capacity coupling constraints limit the sum of the square of the energy storage active power command and the square of the forced reactive power support to not exceed the square of the maximum apparent power capacity of the bidirectional converter.

10. A virtual power plant cluster scheduling photovoltaic-storage coordinated power supply system according to claim 9, characterized in that, The multi-objective optimization model is also configured with a forward-looking thermal safety constraint, which limits the steady-state asymptotic temperature from not exceeding a preset temperature safety threshold. The coordination and scheduling server is configured to run the impedance adaptive coordination and scheduling module according to the power coordination and scheduling cycle, and to run the dual-channel parameter online identification module according to the parameter identification and update cycle with a lower frequency.

Images