Substation-energy storage station-photovoltaic station cooperative power control method

Abstract

The invention discloses a transformer substation-energy storage station-photovoltaic station cooperative power control method, which belongs to the technical field of electric power, constructs a photovoltaic power dynamic prediction and bus power smooth track by fusing heterogeneous data of a transformer substation, an energy storage station, a photovoltaic station and a meteorological platform, and combines rolling optimization control and millisecond-level fluctuation suppression to realize the cooperative power control of the transformer substation, the energy storage station, the photovoltaic station and the meteorological platform. The three-station coordinated power grid power stable output is realized, the technical problems of power coordination, prediction feed-forward and rapid disturbance compensation of the transformer substation-energy storage station-photovoltaic station are solved, the photovoltaic prediction accuracy is improved, a pre-regulation instruction is output in advance, the system feed-forward control capability is enhanced, the bus power tracking capability is improved, and the power supply efficiency is improved. And the energy storage power fluctuation and the SOC safety boundary are controlled, local quick response and closed-loop dynamic compensation are realized, and the anti-disturbance capability and the stability of the system are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power, and particularly relates to a collaborative power control method for a substation - energy storage station - photovoltaic station. Background Art

[0002] With the large - scale access of new energy, intermittent power sources such as photovoltaic power generation have put forward higher requirements for the power stability and regulation ability of the power grid. To improve the new - energy consumption capacity, technical paths such as photovoltaic - energy - storage collaborative control or source - load - energy - storage joint regulation are often adopted, and the energy - storage system is used to alleviate the fluctuation of photovoltaic output.

[0003] However, existing technologies mostly focus on a single station, only realizing the simple tracking control of the energy - storage system and local photovoltaic, and unable to coordinate the power flow among the substation, photovoltaic station and energy - storage station at the system level, making it difficult to meet the high - precision and fast - dynamic - response power regulation requirements on the bus side.

[0004] In addition, existing control strategies usually adopt fixed strategies or static optimization methods, lacking the real - time response ability for prediction uncertainty and rapid disturbances. In the scenario of severe fluctuations and frequent jumps in photovoltaic power, problems such as response lag and amplified bus - power fluctuations are likely to occur in the control system, and even voltage / frequency over - limits may be caused. At the same time, although some current systems have introduced prediction algorithms, they have not fully integrated environmental information and weather factors, and there are still significant deficiencies in prediction accuracy and dynamic adaptability. Summary of the Invention

[0005] The purpose of the present invention is to provide a collaborative power control method for a substation - energy storage station - photovoltaic station, which solves the technical problems of power coordination, predictive feed - forward and rapid disturbance compensation among the substation, energy - storage station and photovoltaic station.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: A collaborative power control method for a substation - energy storage station - photovoltaic station, comprising the following steps: Step 1: Collect heterogeneous data from the substation, energy - storage station, photovoltaic station and meteorological platform, pre - process the heterogeneous data, and construct an original data set with unified time series; the heterogeneous data includes bus voltage, active power, scheduling instructions, current SOC status, predicted SOC status, energy - storage charge - discharge capacity, battery health status SOH, current photovoltaic power generation, local environmental data and weather forecast; extract features from the original data set to obtain features reflecting the bus status, energy - storage boundary capacity and photovoltaic power generation trend; Step 2: Based on historical data and local environmental data, construct a short-term photovoltaic power prediction model to generate a primary prediction curve for photovoltaic power generation in the next 10 minutes; introduce the weather factor vector extracted from weather forecasts, construct a multi-model fusion mechanism to obtain the prediction result of the dynamic trend of photovoltaic power, and generate a confidence interval; according to the prediction result of the dynamic trend of photovoltaic power, construct the active power curve of the target bus in the next 10 minutes, fuse the slope limit and the fluctuation interval annotation to form a smooth power expectation trajectory, and output the energy storage pre-regulation instruction in advance according to the predicted fluctuation trend. Step 3: With the goal of stable output of bus power, construct a collaborative optimization control model. The inputs of the collaborative optimization control model include the prediction result of the dynamic trend of photovoltaic power, the active power curve of the target bus, the current SOC state, the predicted SOC state, the charge and discharge capacity of the energy storage, the battery health state SOH, the real-time load power, the bus voltage frequency, and the dispatching instruction. Set constraint conditions for the collaborative optimization control model, including photovoltaic output constraint, energy storage power constraint, SOC boundary constraint, voltage frequency constraint, and power change rate constraint; through rolling solution of the collaborative optimization control model, output the control instruction set per minute. After the control instruction set is generated, embed a millisecond-level fluctuation suppression mechanism in the energy storage PCS. When the bus voltage or frequency undergoes a rapid disturbance, the millisecond-level fluctuation suppression mechanism triggers the droop control logic to quickly correct the power offset and perform dynamic response compensation.

[0007] Preferably, when performing Step 1, it specifically includes the following steps: When performing Step 1, it specifically includes the following steps: Step 1-1: Collect heterogeneous data from substations, energy storage stations, photovoltaic stations, and meteorological platforms, specifically: On the substation side, obtain the bus voltage, active power, and dispatching instruction through the SCADA system. On the energy storage station side, collect the energy storage SOC state, current power, maximum power capacity, and battery health state SOH through the BMS / EMS interface. On the photovoltaic station side, obtain the current photovoltaic power generation through the inverter interface, and at the same time connect to the local environmental transmitter to obtain local environmental data. On the meteorological platform side, access the third-party weather forecast platform through the API to obtain short-term weather forecasts in the future. Step 1-2: Perform formatting processing on the heterogeneous data, including cleaning the units, timestamps, and missing values; based on the bus power sampling period, align the time axis uniformly; perform noise filtering and interpolation to complete, and form a continuous time series that can be used for modeling. Step 1-3: Integrate the processed data to obtain the original dataset. The variable fields of the original dataset include: bus voltage, active power, SOC data, energy storage capacity, SOH data, photovoltaic power generation, local environmental data, and weather forecast factors; Each piece of data in the original dataset is aligned according to the timestamp; Step 1-4: Extract features from the original dataset to obtain three types of business features and a structured feature set, including: Bus status features, which are used to represent the amplitude of bus voltage fluctuations and the trend of active power changes; Energy storage capacity boundary features, which are used to represent the SOC interval, maximum or minimum power response ability, and health state estimation; Photovoltaic power generation trend features, which are used to represent the increase and decrease rate of power generation, the current irradiation trend, and the future weather factor vector.

[0008] Preferably, feature extraction is performed on each index in the original dataset. Specifically: The extraction of bus status features uses a sliding window method to calculate the statistics of voltage and active power: The change rate of active power can be expressed as: ; Among them, is the active power at time t, represents the sampling time interval, represents the change rate of active power; The calculation of the voltage fluctuation amplitude is: ; Among them, T represents the sliding window width, is the time variable, represents the voltage fluctuation amplitude, represents the bus voltage at time Extraction of energy storage capacity boundary features: The current SOC interval can be judged to be in the high, medium, or low energy area according to the SOC value provided by the BMS interface (BMS system) on the energy storage station side; the maximum power capacity is provided by the BMS interface; The health state SOH can affect the maximum charge and discharge capacity. Define the corrected energy storage boundary capacity as: ; Among them, SOH represents the battery health state, that is, the SOH data, represents the corrected maximum power capacity of the energy storage system; is the rated maximum power of the energy storage system; The deployment of the energy storage system is on the energy storage station side; Feature extraction of photovoltaic power generation trend: ; Among them, is the change rate of photovoltaic power generation at time t, representing the power change amount per unit time; represents the actual power generation of the photovoltaic system at time t.

[0009] Preferably, when performing step 2, it specifically includes the following steps: Step 2-1: Using historical operation data and local environmental data, construct a short-term photovoltaic power prediction model using the LSTM model to quickly predict the primary prediction curve of photovoltaic power generation within the next 10 minutes, that is, the primary prediction result; Step 2-2: Adopt a multi-model fusion weighted algorithm to fuse the primary prediction result and the weather factor vector to obtain a secondary prediction result, which is used to represent the dynamic trend prediction result of photovoltaic power within the next 10 minutes, that is, photovoltaic dynamic prediction; The multi-model fusion weighted algorithm is shown in the following formula: ; Among them, represents the fused secondary prediction value, represents the primary prediction result, represents the result predicted by the lightweight neural network model after introducing the weather factor; represents the weighting coefficient, represents the time step, ; Based on the secondary prediction result, generate a confidence interval to reflect the upper and lower limits of the predicted value; Step 2-3: According to the photovoltaic prediction result, construct the active power curve of the target bus for the next 10 minutes; If there is a fluctuating time section on the active power curve of the target bus, that is, the fluctuation section, then increase the upper and lower power buffer zones and identifiers to reserve adjustment space for the energy storage system; Set tolerance redundancy on the active power curve of the target bus to cope with prediction errors or sudden weather disturbances; Finally, output the active power curve of the target bus, the fluctuation section identifier, and the dynamic adjustment margin annotation; Step 2-4: When it is predicted that there will be power fluctuations in the future, generate and output the energy storage pre-regulation instruction in advance and send it to the energy storage system; The generated energy storage pre-regulation instruction is as follows: If it is predicted that there is an upward fluctuation, generate an energy storage pre-regulation instruction to enter the "pre-charge" state in advance and send it to the energy storage station; If a downward fluctuation is predicted, a pre-regulation instruction for energy storage to enter the "pre-discharge" state is generated in advance and sent to the energy storage station.

[0010] Preferably, a lightweight neural network model, i.e., the LightGBM model, is used for modeling. The weather factor vector is introduced and predicted. The model input includes the weather factor vector x obtained from weather forecasting. weather , specifically: x weather =[I solar ,C cloud ,P rain ,v wind ,T air ; Among them, I solar represents the predicted solar irradiance intensity, C cloud represents the cloud cover factor, P rain represents the precipitation probability, v wind represents the wind speed, T air represents the air temperature; The output of the LightGBM model is : ; That is the LightGBM model; The confidence interval is as follows: ; Among them, represents the upper and lower confidence band widths; this interval is used for subsequent energy storage pre-regulation space judgment; When performing step 2-3, it specifically includes introducing a slope limiter based on the photovoltaic prediction result to limit the jump of the active power curve of the target bus; According to the secondary prediction result , construct the active power sequence of the target bus for the next 10 minutes ; The slope limiter is used to constrain the change rate and prevent the target bus power from jumping too fast, specifically as follows: ; represents the allowable maximum power change slope, which is a preset value; When the first-order difference of the power of the prediction curve exceeds the threshold, it is marked as a fluctuation section , that is, the fluctuation section identifier: ; Reserve the upper and lower limit power buffers for the fluctuation section: ; ; Among them, is the power buffer margin; To cope with prediction errors and sudden disturbances, a small amount of tolerance redundancy is also added in the non-fluctuating section: ; Among them, represents the tolerance redundancy amount, is much smaller than , represents the adjusted reference power value.

[0011] Preferably, when performing step 3, it specifically includes the following steps: Step 3-1: Obtain photovoltaic dynamic prediction, bus target power curve, current SOC state, predicted SOC state, energy storage charge and discharge capacity, battery health state SOH, current real-time load active power, current bus voltage, current bus frequency, and superior dispatching instructions, and construct a control optimization input set; the control optimization input set is updated every minute; Step 3-2: Set multi-dimensional constraints, specifically including: Photovoltaic output constraint, used to control the output power of the photovoltaic inverter within the predicted upper and lower bounds; Energy storage power constraint, used to control the output power of the energy storage not to exceed the energy storage charge and discharge capacity; SOC boundary constraint, used to control the energy storage SOC to always operate within the set healthy range; Voltage and frequency constraint, used to control the bus voltage to be stable within the allowable deviation range and the frequency to be maintained within the preset fluctuation band; Power change rate constraint, used to introduce the bus power change rate as a limiting term to avoid shocks or disturbances caused by command jumps; Step 3-3: Construct a collaborative optimization control model, specifically as the following formula: ; Among them, represents the actual bus output power, represents the predicted bus output power, represents the variance of the energy storage charge and discharge power change, represents the first derivative term of the bus power, are all weight factors; Step 3-4: Perform rolling solution on the collaborative optimization control model in minutes, output the control instructions within the current time slice, generate a control instruction set, and the control instruction set is sent to the energy storage station and the photovoltaic station to drive the collaborative operation of the energy storage system and the photovoltaic inverter, so that the bus power is output smoothly; Step 3-5: After the control instruction set is issued, synchronously activate the millisecond-level fluctuation suppression mechanism, which operates inside the local controller of the energy storage PCS. When a rapid disturbance in the bus voltage or frequency is detected, the energy storage PCS automatically performs real-time power compensation according to the droop control mechanism, as follows: Step 3-5-1: Frequency droop compensation. If the frequency offset is , then the instantaneous output change of the energy storage is: ; Where, represents the locally set frequency response coefficient of the energy storage PCS, represents the instantaneous output change of the energy storage; Step 3-5-2: Voltage droop compensation. If the voltage offset is , then the output change of the energy storage is: ; Where, represents the locally set voltage response coefficient of the energy storage PCS, represents the output change of the energy storage; Step 3-5-3: The formula for the final actual output power of the energy storage is as follows: ; Where, represents the actual output power of the energy storage.

[0012] The collaborative power control method for a substation - energy storage station - photovoltaic station according to the present invention solves the technical problems of power collaboration, predictive feedforward, and fast disturbance compensation for a substation - energy storage station - photovoltaic station. The present invention uses multi-dimensional data from the substation, photovoltaic, energy storage, and meteorological platforms to construct a unified time series and structured feature set, providing reliable input for subsequent modeling and control. By combining the LSTM deep model with meteorological factors, the accuracy of photovoltaic prediction is improved, pre-adjustment instructions are output in advance, the feedforward control ability of the system is enhanced, a collaborative optimization model with multi-dimensional constraints is introduced to improve the bus power tracking ability, and the power fluctuation and SOC safety boundary of the energy storage are controlled. A millisecond-level fluctuation suppression mechanism is embedded to achieve local fast response and closed-loop dynamic compensation, significantly improving the anti-disturbance ability and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is the main flowchart of the present invention; Figure 2 is the flowchart of Step 1 of the present invention; Figure 3 is the flowchart of Step 2 of the present invention; Figure 4It is the flowchart of step 3 of the present invention; Figure 5 It is the flowchart of steps 3 - 5 of the present invention. Detailed implementation manners

[0014] From Figures 1-5 A coordinated power control method for a substation - energy storage station - photovoltaic station shown as follows includes the following steps: Step 1: Collect heterogeneous data from the substation, energy storage station, photovoltaic station, and meteorological platform, preprocess the heterogeneous data, and construct an original data set with unified time series; the heterogeneous data includes bus voltage, active power, dispatching instructions, current SOC status, predicted SOC status, energy storage charge - discharge capacity, battery health status SOH, current photovoltaic power generation, local environmental data, and weather forecast; extract features from the original data set to obtain features reflecting the bus status, energy storage boundary capacity, and photovoltaic power generation trend; When performing step 1, it specifically includes the following steps: Step 1 - 1: Collect heterogeneous data from the substation, energy storage station, photovoltaic station, and meteorological platform, specifically: On the substation side, obtain the bus voltage, active power, and dispatching instructions through the SCADA system; On the energy storage station side, collect the energy storage SOC status, current power, maximum power capacity, and battery health status SOH through the BMS / EMS interface; On the photovoltaic station side, obtain the current photovoltaic power generation through the inverter interface, and at the same time connect to the local environmental transmitter to obtain local environmental data; the local environmental data is collected by environmental sensors on the photovoltaic station side, such as temperature - humidity transmitters, illuminance transmitters, etc.

[0015] On the meteorological platform side, access a third - party weather forecast platform through the API to obtain short - term future weather forecasts; In this embodiment, a regulation center platform and a coordinated control platform are deployed on the substation side. The coordinated control platform communicates with the regulation center platform through the Internet, and realizes data interaction in ways such as the IEC104 protocol or MQTT, such as the issuance of regulation instructions, configuration of constraint parameters, and feedback of status, etc.

[0016] The coordinated control platform communicates with terminal systems or devices such as inverters, environmental sensors, energy storage station PCS systems, energy storage BMS systems, smart meters, and synchronous phasor measurement devices PMU on the photovoltaic station side or the energy storage station side through the Internet (TCP), serial port (Modbus), or wireless network, etc.

[0017] The coordinated control platform is responsible for real - time collection of heterogeneous data, photovoltaic prediction and target construction, coordinated optimization regulation, generation, issuance, and feedback of control instructions, etc.

[0018] Step 1-2: Format the heterogeneous data, including cleaning the units, timestamps, and missing values; align the time axis based on the bus power sampling period; perform noise filtering and interpolation to form a continuous time series that can be used for modeling. Step 1-3: Integrate the formatted data to obtain the original dataset. The variable fields of the original dataset include: bus voltage, active power, SOC data, energy storage capacity, SOH data, photovoltaic power generation, local environmental data, and weather forecast factors. Each piece of data in the original dataset is aligned according to the timestamp. Step 1-4: Extract features from the original dataset to obtain three types of business features and a structured feature set, including: Bus status features, used to represent the amplitude of bus voltage fluctuations and the trend of active power changes. Energy storage capacity boundary features, used to represent the SOC interval, maximum or minimum power response capacity, and health state estimation. Photovoltaic power generation trend features, used to represent the rate of increase or decrease of power generation, the current irradiation trend, and the future weather factor vector.

[0019] In this embodiment, statistical analysis and time series feature engineering methods can be used to extract features from the various indicators in the original dataset. Specifically: The extraction of bus status features uses a sliding window method to calculate the statistics of voltage and active power: The rate of change of active power can be expressed as: ; where is the active power at time t, represents the sampling time interval, represents the rate of change of active power.

[0020] The calculation of the voltage fluctuation amplitude is: ; where T represents the sliding window width, is the time variable, represents the voltage fluctuation amplitude, represents the bus voltage at time

[0021] Extraction of energy storage capacity boundary features: The current SOC interval can be determined according to the SOC value provided by the BMS interface (BMS system) on the energy storage station side to determine whether it is in the high, medium, or low energy area; the maximum power capacity is provided by the BMS interface. The State of Health (SOH) can affect the maximum charge and discharge capabilities. The revised energy storage boundary capability is defined as: ; where SOH represents the battery health state, i.e., the SOH data, represents the maximum power capability of the revised energy storage system; is the rated maximum power of the energy storage system; The deployment of the energy storage system is on the side of the energy storage station.

[0022] Photovoltaic power generation trend feature extraction: ; where, is the change rate of the photovoltaic power generation at time t, representing the power change amount per unit time; represents the actual power generation of the photovoltaic system at time t; The irradiance change trend is calculated by the moving average of the local sensor data, and the weather factor vector is obtained from the weather forecast.

[0023] Step 2: Based on historical data and local environmental data, construct a short-term photovoltaic power prediction model to generate a primary prediction curve for photovoltaic power generation in the next 10 minutes; introduce the weather factor vector extracted from the weather forecast, construct a multi-model fusion mechanism to obtain the prediction result of the dynamic trend of photovoltaic power, and generate a confidence interval; according to the prediction result of the dynamic trend of photovoltaic power, construct the active power curve of the target bus in the next 10 minutes, fuse the slope limit and the fluctuation interval annotation to form a smooth power expectation trajectory, and output the energy storage pre-regulation instruction in advance according to the predicted fluctuation trend; When performing Step 2, it specifically includes the following steps: Step 2-1: Utilize historical operation data and local environmental data to construct a short-term photovoltaic power prediction model using the LSTM model to quickly predict the primary prediction curve of photovoltaic power generation in the next 10 minutes, i.e., the primary prediction result; The LSTM model is an existing technology, so it will not be described in detail.

[0024] The input data of the LSTM model includes historical photovoltaic power, light intensity, and local temperature. Both the light intensity and local temperature are local environmental data.

[0025] The historical operation data includes the historical photovoltaic output sequence, and the local environmental data includes light intensity and temperature; the output of the LSTM model is as follows: ; The update frequency of the LSTM model is 1-minute sliding update, represents the time step, initially defaulting to 1 minute.

[0026] Step 2-2: Adopt the multi-model fusion weighted algorithm to fuse the primary prediction result and the weather factor vector to obtain the secondary prediction result, which is used to represent the dynamic trend prediction result of the photovoltaic power within the next 10 minutes, that is, the photovoltaic dynamic prediction; The multi-model fusion weighted algorithm is shown in the following formula: ; Among them, represents the fused secondary prediction value, represents the primary prediction result, represents the prediction result in the weather forecast; represents the weighting coefficient, represents the time step, ; In this embodiment, a lightweight neural network model, that is, the LightGBM model, is used for modeling to introduce the weather factor vector and make predictions. The model input includes the weather factor vector x obtained from the weather forecast weather , specifically: x weather =[I solar ,C cloud ,P rain ,v wind ,T air ; Among them, I solar represents the predicted solar irradiance intensity, C cloud represents the cloud cover factor, P rain represents the precipitation probability, v wind represents the wind speed, T air represents the air temperature; The output of the LightGBM model is : ; That is the LightGBM model.

[0027] The LightGBM model is a prior art, so it will not be described in detail.

[0028] Based on the secondary prediction result, a confidence interval is generated to reflect the upper and lower limits of the predicted value; The confidence interval is shown as follows: ; Among them, represents the width of the upper and lower confidence bands; this interval is used for subsequent judgment of the energy storage pre-regulation space.

[0029] Step 2-3: According to the photovoltaic prediction result, construct the active power curve of the target bus for the next 10 minutes, including: Based on the photovoltaic prediction results, a slope limiter is introduced to limit the jump of the active power curve of the target bus; According to the secondary prediction results , construct the active power sequence of the target bus for the next 10 minutes .

[0030] The slope limiter is used to constrain the rate of change and prevent the power of the target bus from jumping too fast, as follows: ; represents the allowable maximum power change slope, which is a preset value.

[0031] If there is a time period with fluctuations on the active power curve of the target bus, that is, a fluctuation section, upper and lower power buffer zones and identifiers are added to reserve adjustment space for the energy storage system; When the first-order difference of the power of the prediction curve exceeds the threshold, it is marked as a fluctuation section , that is, the fluctuation section identifier: ; Reserve upper and lower power buffer zones for the fluctuation section: ; ; Among them, is the power buffer margin.

[0032] Set tolerance redundancy on the active power curve of the target bus to cope with prediction errors or sudden weather disturbances; In this embodiment, to cope with prediction errors and sudden disturbances, a small amount of tolerance redundancy is also added in the non-fluctuation section: ; Among them, represents the tolerance redundancy amount, is much smaller than , represents the adjusted reference power value.

[0033] Finally, output the active power curve of the target bus, the fluctuation section identifier, and the dynamic adjustment margin annotation; Step 2-4: When it is predicted that there will be power fluctuations in the future, generate and output the energy storage pre-regulation instruction in advance, and send it to the energy storage system; The generated energy storage pre-regulation instruction is as follows: If an upward fluctuation is predicted, generate an energy storage pre-regulation instruction to enter the "pre-charge" state in advance, and send it to the energy storage station; If a downward fluctuation is predicted, generate an energy storage pre-regulation instruction to enter the "pre-discharge" state in advance, and send it to the energy storage station.

[0034] In this embodiment, for all fluctuation sections perform trend judgment to generate the energy storage pre-regulation state in advance: If the power prediction shows an upward trend (discharge reduction or charging required), generate a "pre-charge" command: ; If the power prediction shows a downward trend (power generation reduction), generate a "pre-discharge" command: ; Among them, represents the dynamic regulation margin and the control command of the energy storage, respectively representing the pre-charge setting value and the pre-discharge setting value.

[0035] Step 3: Taking the stable output of the bus power as the goal, construct a collaborative optimization control model. The inputs of the collaborative optimization control model include the dynamic trend prediction result of the photovoltaic power, the active power curve of the target bus, the current SOC state, the predicted SOC state, the charge and discharge capacity of the energy storage, the battery health state SOH, the real-time load power, the bus voltage frequency, and the dispatching instruction; Set constraint conditions for the collaborative optimization control model, including photovoltaic output constraint, energy storage power constraint, SOC boundary constraint, voltage frequency constraint, power change rate constraint; by performing rolling solution on the collaborative optimization control model, output the control instruction set per minute; After the control instruction set is generated, embed a millisecond-level fluctuation suppression mechanism in the energy storage PCS. When the bus voltage or frequency undergoes rapid disturbance, the millisecond-level fluctuation suppression mechanism triggers the droop control logic to quickly correct the power offset and perform dynamic response compensation.

[0036] When performing Step 3, it specifically includes the following steps: Step 3-1: Obtain the photovoltaic dynamic prediction, the bus target power curve, the current SOC state, the predicted SOC state, the charge and discharge capacity of the energy storage, the battery health state SOH, the current real-time load active power, the current bus voltage, the current bus frequency, and the superior dispatching instruction, and construct a control optimization input set; the control optimization input set is updated once per minute; In this embodiment, the real-time load active power can be collected by an intelligent electricity meter, and the current bus voltage and the current bus frequency can both be obtained by collecting through a synchronized phasor measurement device PMU.

[0037] In this embodiment, the obtained data is classified as follows: Photovoltaic system: instantaneous output power , light intensity , component temperature ; Energy storage system: charge and discharge power , SOC state , SOH state ; Bus electrical quantities: voltage , frequency , bus active power ; Dispatch information: superior instructions ; Prediction results: photovoltaic dynamic prediction , bus target power curve , predicted SOC curve .

[0038] Use a unified time reference for timestamp mapping to construct a standard time series index X: ; where represents the moment, with a value range of 0 to n.

[0039] The data vector corresponding to each moment after alignment is expressed as: .

[0040] Step 3-2: Set multi-dimensional constraints, specifically including: Photovoltaic output constraint, used to control the output power of the photovoltaic inverter within the predicted upper and lower bounds; ; where represent the predicted lower limit and predicted upper limit of the photovoltaic power respectively.

[0041] Energy storage power constraint, used to control the output power of the energy storage not to exceed the charge and discharge capacity of the energy storage; ; where represent the minimum charge and discharge power and the maximum charge and discharge power of the energy storage system respectively.

[0042] SOC boundary constraint, used to control the SOC of the energy storage to always operate within the set healthy range; ; where represent the minimum allowable SOC and the maximum allowable SOC of the energy storage system respectively.

[0043] Voltage and frequency constraint, used to control the bus voltage to be stable within the allowable deviation range and the frequency to be maintained within the preset fluctuation band; ; respectively represent the allowable deviation ranges of voltage and frequency; The power change rate constraint is used to introduce the bus power change rate as a limiting term to avoid impacts or disturbances caused by instruction jumps; ; represents the maximum allowable change rate of the bus power.

[0044] Step 3-3: Construct a collaborative optimization control model, specifically as the following formula: ; where, represents the actual bus output power, represents the predicted bus output power, represents the variance of the charge / discharge power change of the energy storage, represents the first derivative term of the bus power, are all weight factors; Step 3-4: Perform rolling solution on the collaborative optimization control model with a minute as the period, output the control instructions within the current time slice, generate a control instruction set, and the control instruction set is sent to the energy storage station and the photovoltaic station to drive the coordinated operation of the energy storage system and the photovoltaic inverter, so that the bus power is output smoothly; The control values in the control instruction set include: The expected power value of the bus at the current time point , which is used as the target for bus power regulation; The reference charge / discharge power of the energy storage system , which is used to drive the energy storage system to participate in regulation; The upper limit of the output of the photovoltaic inverter , which is used to prevent inverter overload caused by prediction deviation; Step 3-5: After the control instruction set is sent, synchronously activate the millisecond-level fluctuation suppression mechanism, and the millisecond-level fluctuation suppression mechanism runs inside the local controller of the energy storage PCS; When it is detected that the bus voltage or frequency undergoes rapid disturbances, the energy storage PCS automatically performs real-time power compensation according to the droop control mechanism, specifically as the following steps: Step 3-5-1: Frequency droop compensation. If the frequency offset is , then the instantaneous output change of the energy storage is: ; where, represents the frequency response coefficient set locally by the energy storage PCS, represents the instantaneous output change of the energy storage; Step 3-5-2: Voltage droop compensation. If the voltage offset is , then the change in energy storage output is: ; Among them, represents the voltage response coefficient locally set by the energy storage PCS, represents the change in energy storage output; Step 3-5-3: The formula for the final actual output power of the energy storage is as follows: ; Among them, represents the actual output power of the energy storage.

[0045] A collaborative power control method for a substation - energy storage station - photovoltaic station according to the present invention solves the technical problems of power coordination, predictive feedforward, and fast disturbance compensation for the substation - energy storage station - photovoltaic station. The present invention uses multi-dimensional data from substations, photovoltaics, energy storage, and meteorological platforms to construct a unified time-series and structured feature set, providing a reliable input for subsequent modeling and control. By combining the LSTM deep model with weather factors, the accuracy of photovoltaic prediction is improved, pre-regulation instructions are output in advance, the feedforward control ability of the system is enhanced, a collaborative optimization model with multi-dimensional constraints is introduced to improve the bus power tracking ability, and the power fluctuation and SOC safety boundary of the energy storage are controlled. A millisecond-level fluctuation suppression mechanism is embedded to achieve local fast response and closed-loop dynamic compensation, significantly improving the anti-disturbance ability and stability of the system.

Claims

1. A coordinated power control method for a substation - energy storage station - photovoltaic station, characterized in that: It includes the following steps: Step 1: Collect heterogeneous data from substations, energy storage stations, photovoltaic stations, and meteorological platforms, preprocess the heterogeneous data, and construct an original data set with unified time series; the heterogeneous data includes bus voltage, active power, dispatching instructions, current SOC status, predicted SOC status, energy storage charge and discharge capacity, battery health status SOH, current photovoltaic power generation, local environmental data, and weather forecast; extract features from the original data set to obtain features reflecting bus status, energy storage boundary capacity, and photovoltaic power generation trend; Step 2: Based on historical data and local environmental data, construct a short-term photovoltaic power prediction model to generate a primary prediction curve for photovoltaic power generation in the next 10 minutes; introduce a weather factor vector extracted from the weather forecast, construct a multi-model fusion mechanism to obtain the prediction result of the dynamic trend of photovoltaic power, and generate a confidence interval; according to the prediction result of the dynamic trend of photovoltaic power, construct the target bus active power curve for the next 10 minutes, fuse the slope limit and fluctuation interval annotation to form a smooth power expectation trajectory, and output the energy storage pre-regulation instruction in advance according to the predicted fluctuation trend; Step 3: With the goal of stable output of bus power, construct a collaborative optimization control model. The inputs of the collaborative optimization control model include the prediction result of the dynamic trend of photovoltaic power, the target bus active power curve, the current SOC status, the predicted SOC status, the energy storage charge and discharge capacity, the battery health status SOH, the real-time load power, the bus voltage frequency, and the dispatching instruction; Set constraint conditions for the collaborative optimization control model, including photovoltaic output constraint, energy storage power constraint, SOC boundary constraint, voltage frequency constraint, and power change rate constraint; output the control instruction set per minute by performing rolling solution on the collaborative optimization control model; After the control instruction set is generated, embed a millisecond-level fluctuation suppression mechanism in the energy storage PCS. When the bus voltage or frequency undergoes rapid disturbance, the millisecond-level fluctuation suppression mechanism triggers the droop control logic to quickly correct the power offset and perform dynamic response compensation.

2. The collaborative power control method for a substation - energy storage station - photovoltaic station according to claim 1, characterized in that: When performing Step 1, it specifically includes the following steps: When performing Step 1, it specifically includes the following steps: Step 1-1: Collect heterogeneous data from substations, energy storage stations, photovoltaic stations, and meteorological platforms. Specifically: On the substation side, obtain bus voltage, active power, and dispatching instructions through the SCADA system; On the energy storage station side, collect the energy storage SOC status, current power, maximum power capacity, and battery health status SOH through the BMS / EMS interface; On the photovoltaic station side, obtain the current photovoltaic power generation through the inverter interface, and at the same time connect to the local environmental transmitter to obtain local environmental data; On the meteorological platform side, access the third-party weather forecast platform through the API to obtain the short-term weather forecast in the future; Step 1-2: Perform formatting processing on the heterogeneous data, including cleaning units, timestamps, and missing values; align the time axis uniformly based on the bus power sampling period; perform noise filtering and interpolation completion to form a continuous time series for modeling; Step 1-3: Integrate the formatted data to obtain the original dataset. The variable fields of the original dataset include: bus voltage, active power, SOC data, energy storage capacity, SOH data, photovoltaic power generation, local environmental data, and weather forecast factors; Each piece of data in the original dataset is aligned according to the timestamp; Step 1-4: Extract features from the original dataset to obtain three types of business features and a structured feature set, including: Bus status features, which are used to represent the amplitude of bus voltage fluctuations and the change trend of active power; Energy storage capacity boundary features, which are used to represent the SOC interval, maximum or minimum power response capacity, and health state estimation; Photovoltaic power generation trend features, which are used to represent the increase and decrease rate of power generation, the current irradiation trend, and the future weather factor vector.

3. The collaborative power control method for a substation - energy storage station - photovoltaic station according to claim 2, characterized in that: Extract features for each index in the original dataset. Specifically: The bus status feature extraction uses a sliding window method to calculate the statistics of voltage and active power: The active power change rate is expressed as: ； wherein, is the active power at time t, represents the sampling time interval, represents the rate of change of active power; The voltage fluctuation amplitude is calculated as: ； where T represents the sliding window width, is the time variable, represents the voltage fluctuation amplitude, represents the bus voltage at Energy storage capacity boundary feature extraction: The current SOC interval is judged to be in the high, medium, or low energy area according to the SOC value provided by the BMS interface on the energy storage station side; the maximum power capacity is provided by the BMS interface; Define the modified energy storage boundary capacity as: ； Among them, SOH represents the state of health of the battery, that is, the SOH data, represents the maximum power capacity of the corrected energy storage system; is the rated maximum power of the energy storage system; The deployment of the energy storage system is on the energy storage station side; Photovoltaic power generation trend feature extraction: ； Among them, is the change rate of the photovoltaic power generation at time t, indicating the power change amount per unit time; represents the actual power generation of the photovoltaic system at time t.

4. The collaborative power control method for a substation - energy storage station - photovoltaic station according to claim 1, characterized in that: When performing Step 2, it specifically includes the following steps: Step 2-1: Use historical operation data and local environmental data to construct a short-term photovoltaic power prediction model using the LSTM model to quickly predict the primary prediction curve of photovoltaic power generation within the next 10 minutes, that is, the primary prediction result; Step 2-2: Adopt a multi-model fusion weighted algorithm to fuse the primary prediction result and the weather factor vector to obtain the secondary prediction result, which is used to represent the dynamic trend prediction result of photovoltaic power within the next 10 minutes, that is, the photovoltaic dynamic prediction; The multi-model fusion weighted algorithm is shown in the following formula: ； Among them, represents the secondary predicted value after fusion, represents the primary prediction result, represents the result of prediction through the lightweight neural network model after introducing the weather factor; represents the weighting coefficient, represents the time step, ; Based on the secondary prediction result, generate a confidence interval to reflect the upper and lower limits of the predicted value; Step 2-3: According to the photovoltaic prediction result, construct the target bus active power curve for the next 10 minutes; If there is a fluctuating time section on the target bus active power curve, that is, the fluctuating section, then add an upper and lower limit power buffer zone and an identifier to reserve adjustment space for the energy storage system; Set tolerance redundancy on the target bus active power curve to cope with prediction errors or sudden weather disturbances; Finally, output the target bus active power curve, the fluctuating section identifier, and the dynamic adjustment margin annotation; Step 2-4: When it is predicted that there will be power fluctuations in the future, generate and output the energy storage pre-regulation instruction in advance and send it to the energy storage system; The generated energy storage pre-regulation instruction is shown as follows: If an upward fluctuation is predicted, generate an energy storage pre-regulation instruction to enter the "pre-charge" state in advance and send it to the energy storage station; If a downward fluctuation is predicted, generate an energy storage pre-regulation instruction to enter the "pre-discharge" state in advance and send it to the energy storage station.

5. The collaborative power control method for a substation - energy storage station - photovoltaic station according to claim 4, characterized in that: A lightweight neural network model, namely the LightGBM model, is used for modeling. The weather factor vector is introduced and predicted. The model input includes the weather factor vector x obtained from weather forecasts weather , specifically: x weather =[I solar ,C cloud ,P rain ,v wind ,T air ; Among them, I solar represents the predicted solar irradiance intensity, C cloud represents the cloud cover factor, P rain represents the precipitation probability, v wind represents the wind speed, T air represents the air temperature; The output of the LightGBM model is : ； Namely, the LightGBM model; The confidence interval is shown as follows: ; Among them, represents the width of the upper and lower confidence bands; this interval is used for subsequent judgment of the energy storage pre-regulation space; When performing Step 2-3, it specifically includes introducing a slope limiter based on the photovoltaic prediction result to limit the jump of the active power curve of the target busbar; According to the secondary prediction result , construct the active power sequence of the target bus for the next 10 minutes ; The slope limiter is used to constrain the rate of change and prevent the target busbar power from jumping too fast, specifically as follows: ； Indicates the allowable maximum power change slope, which is a preset value; When the first-order difference of the power of the prediction curve exceeds the threshold, it is marked as a fluctuation section , that is, the fluctuation section identifier: ； Reserve upper and lower limit power buffers for the fluctuation section: ； ； Among them, is the power buffer margin; To cope with prediction errors and sudden disturbances, a small amount of tolerance redundancy is also added in the non-fluctuation section: ； Among them, represents the tolerance redundancy amount, represents the adjusted reference power value.

6. The collaborative power control method for a substation - energy storage station - photovoltaic station according to claim 1, wherein: When performing Step 3, it specifically includes the following steps: Step 3-1: Obtain photovoltaic dynamic prediction, busbar target power curve, current SOC state, predicted SOC state, energy storage charge and discharge capacity, battery health state SOH, current real-time load active power, current busbar voltage, current busbar frequency, and superior dispatching instructions, and construct a control optimization input set; the control optimization input set is updated every minute; Step 3-2: Set multi-dimensional constraints, specifically including: Photovoltaic output constraint, which is used to control the output power of the photovoltaic inverter within the predicted upper and lower bounds; Energy storage power constraint, which is used to control the output power of the energy storage not to exceed the energy storage charge and discharge capacity; SOC boundary constraint, which is used to control the energy storage SOC to always operate within the set healthy range; Voltage and frequency constraint, which is used to control the busbar voltage to be stable within the allowable offset range and the frequency to be maintained within the preset fluctuation band; Power change rate constraint, which is used to introduce the busbar power change rate as a limiting term to avoid shocks or disturbances caused by instruction jumps; Step 3-3: Construct a cooperative optimization control model, specifically as the following formula: ； Among them, represents the actual bus output power, represents the predicted bus output power, represents the variance of the charge and discharge power change of the energy storage, represents the first derivative term of the bus power, are all weight factors; Step 3-4: Perform rolling solution on the cooperative optimization control model in minutes, output the control instructions within the current time slice, generate a control instruction set, and the control instruction set is sent to the energy storage station and the photovoltaic station to drive the coordinated operation of the energy storage system and the photovoltaic inverter, so that the busbar power is output smoothly; Step 3-5: After the control instruction set is sent, synchronously activate the millisecond-level fluctuation suppression mechanism, and the millisecond-level fluctuation suppression mechanism runs inside the local controller of the energy storage PCS; When it is detected that the busbar voltage or frequency undergoes rapid disturbances, the energy storage PCS automatically performs real-time power compensation according to the droop control mechanism, specifically as the following steps: Step 3-5-1: Frequency droop compensation. If the frequency offset is , then the instantaneous output change of the energy storage is: ； Among them, represents the frequency response coefficient locally set by the energy storage PCS, represents the instantaneous output change of the energy storage; Step 3-5-2: Voltage droop compensation. If the voltage offset is , then the change in the energy storage output is: ； Among them, represents the voltage response coefficient locally set by the energy storage PCS, represents the change in energy storage output; Step 3-5-3: The final formula for the actual output power of the energy storage is as follows: ； Among them, represents the actual output power of the energy storage.

Images