A virtual power plant operation control method and control system based on a battery swap cabinet

Abstract

The present application relates to the field of two-wheel vehicle battery replacement cabinet, and discloses a virtual power plant operation control method and control system based on a battery replacement cabinet, comprising the following steps: detecting the battery state of the battery replacement cabinet and the real-time load state of the power grid, and constructing a battery replacement demand prediction model based on the detection results and combining an LSTM algorithm. Through the battery replacement demand prediction model, the real-time state of the power grid is combined to schedule the working mode of the battery replacement cabinet, and the charging and discharging strategy of the battery replacement cabinet is optimized in combination with a distributed collaborative control algorithm. The present application can utilize the energy storage characteristics of the batteries in the battery replacement cabinet to realize power grid peak shaving and valley filling, improve energy utilization efficiency, and enhance power grid stability.

Description

Technical Field

[0001] This invention relates to the field of battery swapping cabinets for two-wheeled vehicles, and in particular to a virtual power plant operation control method and control system based on the battery swapping cabinet. Background Technology

[0002] With the increasing popularity of electric vehicles, battery swapping stations for two-wheeled vehicles have become widely used as a convenient energy replenishment facility. Traditional battery swapping stations only have battery charging capabilities and cannot effectively utilize the energy storage advantages of batteries to participate in grid regulation. Meanwhile, in areas with scheduled power outages, the demand for uninterruptible power supplies (UPS) is growing. Virtual power plants participate in grid services by aggregating distributed energy resources and unifying their scheduling and management.

[0003] The batteries in the battery swapping cabinet, as distributed energy storage units, are part of a virtual power plant. Therefore, the control method needs to coordinate these dispersed batteries to achieve demand response from the power grid.

[0004] To leverage the energy storage characteristics of batteries within the battery swapping cabinet to achieve peak shaving and valley filling in the power grid, provide uninterrupted power, improve energy efficiency, and enhance grid stability, this invention aims to provide a virtual power station system and its control method based on a two-wheeled vehicle battery swapping cabinet. Specifically, it provides a virtual power plant operation control method and control system based on the battery swapping cabinet. This application can also be used to develop electric vehicles, emergency energy storage modules, and charging stations as basic units for virtual power plants. Summary of the Invention

[0005] This invention overcomes the shortcomings of the prior art and provides a virtual power plant operation control method and control system based on a battery swapping cabinet.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] The first aspect of this invention provides a virtual power plant operation control method based on a battery swapping cabinet, comprising the following steps:

[0008] Real-time battery status detection and current grid load status detection are performed on the battery swapping cabinet. Based on the detection results, a battery swapping demand prediction model for the battery swapping cabinet is constructed using a long short-term memory network.

[0009] Based on the real-time load status of the target power grid, the PSC module is used to adjust the working mode of the target battery and generate the actual battery swapping demand curve.

[0010] By combining the battery swapping demand prediction model and the actual battery swapping demand curve, the ADMM distributed optimization algorithm is used to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0011] Furthermore, in a preferred embodiment of the present invention, the real-time battery status detection and current grid load status detection of the battery swapping cabinet, combined with the detection results, are used to construct a battery swapping demand prediction model for the battery swapping cabinet through a long short-term memory network, specifically as follows:

[0012] Identify the battery swapping cabinets that require operational control, designate them as target battery swapping cabinets, and designate the batteries in the target battery swapping cabinets as target batteries.

[0013] Inside the target battery swapping cabinet, a power control unit is acquired. Based on the power control unit, the real-time quantity of the target batteries and the real-time remaining power of the target batteries are detected in the battery swapping cabinet. At the same time, the current output power value of the target batteries is calculated based on the power control unit.

[0014] Collect the real-time temperature and state of charge of different target batteries, and combine them with the current output power value of the target battery to generate the real-time status of the target battery.

[0015] The power grid connected to the target battery swapping cabinet is identified as the target power grid, and the real-time load status of the target power grid is read. The real-time load status of the target power grid includes the power grid load power, load distribution, and load fluctuation status.

[0016] A big data network is introduced and connected to the target battery swapping cabinet. Based on the big data network, the historical battery swapping data of the target battery swapping cabinet is retrieved and determined. At the same time, the battery swapping demand period is preset. Based on the big data network, the weather information of the target battery swapping cabinet's location during the battery swapping demand period is retrieved.

[0017] A long short-term memory network algorithm is introduced. Based on the long short-term memory network algorithm, combined with the real-time status of the target battery, the real-time load status of the target power grid, the historical battery swapping data of the target battery swapping cabinet, and the weather information of the target battery swapping cabinet's location during the battery swapping demand period, a battery swapping demand prediction model is constructed.

[0018] Furthermore, in a preferred embodiment of the present invention, the construction of the battery swapping demand prediction model specifically includes:

[0019] Based on the battery swapping demand period, a timestamp is constructed and marked as the target timestamp. The time step in the target timestamp is 24 cycles, calculated in hours.

[0020] Data analysis is performed on the weather information of the target battery swapping cabinet location during the battery swapping demand period, and the numerical weather information is retained and processed, while the weather status is converted into unique thermal codes.

[0021] Based on the timestamp, the real-time status of the target battery, the real-time load status of the target power grid, the historical battery swapping data of the target battery swapping cabinet, and numerical weather information within 24 cycles are sampled and integrated so that the data in one sample corresponds to one cycle.

[0022] The Long Short-Term Memory (LSTM) network algorithm is introduced, and a LSTM blank model is constructed. All samples and timestamps are imported into the LSTM blank model, and one-hot encoding is set in the LSTM blank model to obtain the preliminary LSTM network model.

[0023] Obtain the network architecture of the initially defined long short-term memory network model, wherein the network architecture of the initially defined long short-term memory network model includes a fully connected layer and an LSTM layer;

[0024] In the initial setup of the Long Short-Term Memory (LSTM) network model, loss functions are set in different LSTM layers. The loss function is set as the mean squared error of the samples. The LSTM layers are trained by continuously calculating the mean squared error of the samples, and the mean squared error is tuned in different LST layers based on the Bayesian optimization algorithm.

[0025] A standard training step size is preset. When the training step size of the LSTM layer is equal to the standard training step size, training of all LSTM layers is stopped, and all LSTM layers are connected to the fully connected layer to obtain the battery swapping demand prediction model. At the same time, the battery swapping demand prediction curve is output in the fully connected layer.

[0026] The battery swapping demand forecast curve predicts the battery swapping demand for different target batteries in the next 24 hours. The battery swapping demand records the rate of change of the output power of the target battery and the change of the load status of the target power grid.

[0027] Furthermore, in a preferred embodiment of the present invention, the step of adjusting the operating mode of the target battery using a PSC module based on the real-time load status of the target power grid and generating an actual battery swapping demand curve specifically involves:

[0028] Based on big data networks, determine the grid load power corresponding to the target power grid when it is in a trough state, a peak state, and a power outage state;

[0029] Determine the real-time load power of the target power grid, and analyze the real-time load status of the target power grid based on the power grid load range corresponding to the target power grid being in the trough state, peak state, and power outage state respectively.

[0030] A PSC module is installed in the target battery swapping cabinet. The PSC module includes a rectifier and an inverter, which are used to convert AC power to DC power in the target battery swapping cabinet.

[0031] If the real-time load power of the target power grid is within the range of the power grid load power corresponding to the trough state, then the target battery swapping cabinet is controlled to connect to the target power grid and charge the target battery through the PSC module.

[0032] If the real-time load power of the target power grid is within the range of the power grid load power corresponding to the peak state, then the PSC module controls the target battery swapping cabinet to invert the electrical energy in the target battery into AC power and feed it into the target power grid.

[0033] If the real-time load power of the target power grid is within the power grid load power range corresponding to the power outage state, then the load connected to the target battery swapping cabinet is obtained, and through the PSC module, the target battery swapping cabinet is controlled to invert the electrical energy in the target battery into AC power and provide electrical energy to the load connected to the target battery swapping cabinet.

[0034] During the operation of the PSC module, a battery swapping demand curve is output in real time and calibrated as the actual battery swapping demand curve. The actual battery swapping demand curve records the battery swapping demand of the target battery.

[0035] Furthermore, in a preferred embodiment of the present invention, the step of combining the battery swapping demand prediction model and the actual battery swapping demand curve, and optimizing the charging and discharging strategy of the target battery swapping cabinet using the ADMM distributed optimization algorithm, specifically involves:

[0036] Based on the battery swapping demand prediction model, the battery swapping demand prediction curve is obtained, and the time point corresponding to the generation of the actual battery swapping demand curve is determined and labeled as a type of time point.

[0037] Based on the first type of time point, the curve is truncated from the battery swapping demand prediction curve to obtain a second type of battery swapping demand prediction curve, and the second type of battery swapping demand prediction curve is analyzed to overlap with the actual battery swapping demand curve.

[0038] The overlap error rate is preset, and the overlap error between the predicted curve of a type of battery swapping demand and the actual battery swapping demand curve is calculated. If the overlap error between the predicted curve of a type of battery swapping demand and the actual battery swapping demand curve is kept within the overlap error rate, the target battery swapping cabinet is calibrated as a qualified battery swapping cabinet.

[0039] If the overlap error between the predicted battery swapping demand curve and the actual battery swapping demand curve does not remain within the overlap error rate, then the numerical weather information will be updated in real time in the battery swapping demand prediction model, and the PSC module will be corrected in real time.

[0040] If the overlap error between the predicted battery swapping demand curve and the actual battery swapping demand curve still does not remain within the overlap error rate after the numerical weather information is updated in real time and the PSC module performs real-time parameter correction, then the ADMM distributed optimization algorithm is introduced to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0041] Furthermore, in a preferred embodiment of the present invention, the introduction of the ADMM distributed optimization algorithm to optimize the charging and discharging strategy of the target battery swapping cabinet specifically includes:

[0042] The actual battery swapping demand curve is transformed into actual battery swapping demand data samples and fed back into the battery swapping demand prediction model. At the same time, the constraints in the battery swapping demand prediction model are updated.

[0043] Among them, the updated constraint is to adjust the lower limit of the number of target batteries available based on battery swapping demand;

[0044] The ADMM distributed optimization algorithm is introduced. Based on the ADMM distributed optimization algorithm, the target battery swapping cabinet is treated as an independent node, and the global target of the target power grid is preset, wherein the global target of the target power grid is the peak-shaving and valley-filling demand of the target power grid.

[0045] All target battery swapping cabinets are acquired, and the planning information of adjacent target battery swapping cabinets is exchanged. Based on the ADMM distributed optimization algorithm, a global consistency iteration is performed on adjacent target battery swapping cabinets. The global consistency iteration is to alternately update the local variables of the target battery swapping cabinet and the variables of adjacent target battery swapping cabinets through the ADMM distributed optimization algorithm until the target battery swapping cabinet can still match the global target of the target power grid after meeting the local battery swapping requirements.

[0046] If the target battery swapping cabinet can still match the global target of the target power grid after meeting the local battery swapping demand, the overlap error between the battery swapping demand prediction curve and the actual battery swapping demand curve will remain within the overlap error rate, thus obtaining a qualified battery swapping cabinet.

[0047] A second aspect of the present invention also provides a virtual power plant operation control system based on a battery swapping cabinet. The virtual power plant operation control system includes a memory and a processor. The memory stores a virtual power plant operation control method. When the virtual power plant operation control method is executed by the processor, it performs the following steps:

[0048] Real-time battery status detection and current grid load status detection are performed on the battery swapping cabinet. Based on the detection results, a battery swapping demand prediction model for the battery swapping cabinet is constructed using a long short-term memory network.

[0049] Based on the real-time load status of the target power grid, the PSC module is used to adjust the working mode of the target battery and generate the actual battery swapping demand curve.

[0050] By combining the battery swapping demand prediction model and the actual battery swapping demand curve, the ADMM distributed optimization algorithm is used to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0051] This invention addresses the technical deficiencies in the prior art and offers the following advantages: By detecting the battery status of the battery swapping cabinet and the real-time load status of the power grid, a battery swapping demand prediction model is constructed based on the detection results and combined with the LSTM algorithm. Using this demand prediction model, and considering the real-time power grid status, the working mode of the battery swapping cabinet is scheduled, and the charging and discharging strategy of the cabinet is optimized using a distributed cooperative control algorithm. This invention leverages the energy storage characteristics of the batteries within the battery swapping cabinet to achieve peak shaving and valley filling in the power grid, improving energy utilization efficiency and enhancing grid stability. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0053] Figure 1 A flowchart of a virtual power plant operation control method based on a battery swapping cabinet is shown;

[0054] Figure 2 A flowchart illustrating the method for optimizing the charging and discharging strategy of the target battery swapping cabinet is shown.

[0055] Figure 3 A program view of a virtual power plant operation control system based on a battery swapping cabinet is shown. Detailed Implementation

[0056] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0058] Figure 1 A flowchart illustrating a virtual power plant operation control method based on a battery swapping cabinet is shown, including the following steps:

[0059] S102: Real-time battery status detection and current grid load status detection are performed on the battery swapping cabinet. Based on the detection results, a battery swapping demand prediction model for the battery swapping cabinet is constructed through a long short-term memory network.

[0060] S104: Based on the real-time load status of the target power grid, the PSC module is used to adjust the working mode of the target battery and generate the actual battery swapping demand curve.

[0061] S106: Combining the battery swapping demand prediction model and the actual battery swapping demand curve, the ADMM distributed optimization algorithm is used to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0062] Furthermore, in a preferred embodiment of the present invention, the real-time battery status detection and current grid load status detection of the battery swapping cabinet, combined with the detection results, are used to construct a battery swapping demand prediction model for the battery swapping cabinet through a long short-term memory network, specifically as follows:

[0063] Identify the battery swapping cabinets that require operational control, designate them as target battery swapping cabinets, and designate the batteries in the target battery swapping cabinets as target batteries.

[0064] Inside the target battery swapping cabinet, a power control unit is acquired. Based on the power control unit, the real-time quantity of the target batteries and the real-time remaining power of the target batteries are detected in the battery swapping cabinet. At the same time, the current output power value of the target batteries is calculated based on the power control unit.

[0065] Collect the real-time temperature and state of charge of different target batteries, and combine them with the current output power value of the target battery to generate the real-time status of the target battery.

[0066] The power grid connected to the target battery swapping cabinet is identified as the target power grid, and the real-time load status of the target power grid is read. The real-time load status of the target power grid includes the power grid load power, load distribution, and load fluctuation status.

[0067] A big data network is introduced and connected to the target battery swapping cabinet. Based on the big data network, the historical battery swapping data of the target battery swapping cabinet is retrieved and determined. At the same time, the battery swapping demand period is preset. Based on the big data network, the weather information of the target battery swapping cabinet's location during the battery swapping demand period is retrieved.

[0068] A long short-term memory network algorithm is introduced. Based on the long short-term memory network algorithm, combined with the real-time status of the target battery, the real-time load status of the target power grid, the historical battery swapping data of the target battery swapping cabinet, and the weather information of the target battery swapping cabinet's location during the battery swapping demand period, a battery swapping demand prediction model is constructed.

[0069] It's important to note that battery swapping cabinets are devices used for charging and swapping batteries for two-wheeled vehicles. These cabinets contain batteries with energy storage capabilities, which can be used for peak shaving and valley filling, providing uninterrupted power. Since each cabinet contains multiple batteries, the first step is to calculate the current output power of each battery. This calculation provides data support for building a battery swapping demand prediction model and generates the real-time status of the target battery. After reading the real-time load status of the target power grid, and considering that battery swapping involves exchanging power between the cabinet's batteries and the grid, the real-time load status of the target power grid and the real-time status of the target batteries are combined to construct the battery swapping demand prediction model. Historical battery swapping data can also provide support, offering a reference for predicting the timing and extent of battery swapping. Furthermore, different weather conditions, such as high or low temperatures, can alter battery swapping demand. Therefore, it's necessary to retrieve weather information for the target battery swapping cabinet's location during the peak battery swapping period. The Long Short-Term Memory (LSTM) algorithm is an algorithm that can perform time-series prediction and output the battery swapping demand curve for a future period of time. It is used to predict the battery swapping demand between the battery swapping station and the power grid, and to allocate the battery swapping station power in advance based on the predicted battery swapping demand.

[0070] Furthermore, in a preferred embodiment of the present invention, the construction of the battery swapping demand prediction model specifically includes:

[0071] Based on the battery swapping demand period, a timestamp is constructed and marked as the target timestamp. The time step in the target timestamp is 24 cycles, calculated in hours.

[0072] Data analysis is performed on the weather information of the target battery swapping cabinet location during the battery swapping demand period, and the numerical weather information is retained and processed, while the weather status is converted into unique thermal codes.

[0073] Based on the timestamp, the real-time status of the target battery, the real-time load status of the target power grid, the historical battery swapping data of the target battery swapping cabinet, and numerical weather information within 24 cycles are sampled and integrated so that the data in one sample corresponds to one cycle.

[0074] The Long Short-Term Memory (LSTM) network algorithm is introduced, and a LSTM blank model is constructed. All samples and timestamps are imported into the LSTM blank model, and one-hot encoding is set in the LSTM blank model to obtain the preliminary LSTM network model.

[0075] Obtain the network architecture of the initially defined long short-term memory network model, wherein the network architecture of the initially defined long short-term memory network model includes a fully connected layer and an LSTM layer;

[0076] In the initial setup of the Long Short-Term Memory (LSTM) network model, loss functions are set in different LSTM layers. The loss function is set as the mean squared error of the samples. The LSTM layers are trained by continuously calculating the mean squared error of the samples, and the mean squared error is tuned in different LST layers based on the Bayesian optimization algorithm.

[0077] A standard training step size is preset. When the training step size of the LSTM layer is equal to the standard training step size, training of all LSTM layers is stopped, and all LSTM layers are connected to the fully connected layer to obtain the battery swapping demand prediction model. At the same time, the battery swapping demand prediction curve is output in the fully connected layer.

[0078] The battery swapping demand forecast curve predicts the battery swapping demand for different target batteries in the next 24 hours. The battery swapping demand records the rate of change of the output power of the target battery and the change of the load status of the target power grid.

[0079] It should be noted that the battery swapping demand time period is based on a 24-hour day, retaining numerical weather data such as temperature and humidity, as these data can affect battery swapping demand. The purpose of converting weather conditions into one-hot encoding is to assist the model's prediction. Ensuring that data from a single sample corresponds to a specific period ensures spatiotemporal overlap; maintaining consistency in both time and space guarantees the accuracy and reliability of the prediction. Training the initially defined Long Short-Term Memory (LSTM) network model involves training the network architecture of the initially defined LSTM network. Training in the LSTM layer of the network architecture and merging it with the fully connected layer outputs the prediction results. The model is trained using regression training with a loss function; the mean squared error (MSE) is the loss function. Training the LSTM layer is achieved by continuously calculating the MSE of the samples. The Bayesian optimization algorithm is used to tune the hyperparameters of the MSE of the sample data. Training stops when the standard training duration is reached, thus obtaining the battery swapping demand prediction model and outputting the battery swapping demand prediction curve.

[0080] Furthermore, in a preferred embodiment of the present invention, the step of adjusting the operating mode of the target battery using a PSC module based on the real-time load status of the target power grid and generating an actual battery swapping demand curve specifically involves:

[0081] Based on big data networks, determine the grid load power corresponding to the target power grid when it is in a trough state, a peak state, and a power outage state;

[0082] Determine the real-time load power of the target power grid, and analyze the real-time load status of the target power grid based on the power grid load range corresponding to the target power grid being in the trough state, peak state, and power outage state respectively.

[0083] A PSC module is installed in the target battery swapping cabinet. The PSC module includes a rectifier and an inverter, which are used to convert AC power to DC power in the target battery swapping cabinet.

[0084] If the real-time load power of the target power grid is within the range of the power grid load power corresponding to the trough state, then the target battery swapping cabinet is controlled to connect to the target power grid and charge the target battery through the PSC module.

[0085] If the real-time load power of the target power grid is within the range of the power grid load power corresponding to the peak state, then the PSC module controls the target battery swapping cabinet to invert the electrical energy in the target battery into AC power and feed it into the target power grid.

[0086] If the real-time load power of the target power grid is within the power grid load power range corresponding to the power outage state, then the load connected to the target battery swapping cabinet is obtained, and through the PSC module, the target battery swapping cabinet is controlled to invert the electrical energy in the target battery into AC power and provide electrical energy to the load connected to the target battery swapping cabinet.

[0087] During the operation of the PSC module, a battery swapping demand curve is output in real time and calibrated as the actual battery swapping demand curve. The actual battery swapping demand curve records the battery swapping demand of the target battery.

[0088] It should be noted that the steps required by the battery swapping cabinet differ depending on the grid's state. Real-time monitoring of the grid load and the battery status within the cabinet is crucial to ensure the stability of both the grid and the batteries. This allows the battery's energy storage capabilities to be utilized for peak shaving and valley filling, providing uninterrupted power, improving energy efficiency, and enhancing grid stability. Therefore, peak shaving and valley filling are implemented. When the grid's real-time load power falls within the range corresponding to different states, the PSC module converts AC to DC power to ensure energy flow. Specifically, during grid off-peak hours, the PSC module charges the batteries within the cabinet; during peak hours, it inverts the battery power into AC and feeds it into the grid.

[0089] Figure 2 A flowchart illustrating a method for optimizing the charging and discharging strategy of a target battery swapping cabinet is shown, including the following steps:

[0090] S202: Combining the battery swapping demand prediction model and the actual battery swapping demand curve, the ADMM distributed optimization algorithm is used to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0091] S204: Introduce the ADMM distributed optimization algorithm to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0092] Furthermore, in a preferred embodiment of the present invention, the step of combining the battery swapping demand prediction model and the actual battery swapping demand curve, and optimizing the charging and discharging strategy of the target battery swapping cabinet using the ADMM distributed optimization algorithm, specifically involves:

[0093] Based on the battery swapping demand prediction model, the battery swapping demand prediction curve is obtained, and the time point corresponding to the generation of the actual battery swapping demand curve is determined and labeled as a type of time point.

[0094] Based on the first type of time point, the curve is truncated from the battery swapping demand prediction curve to obtain a second type of battery swapping demand prediction curve, and the second type of battery swapping demand prediction curve is analyzed to overlap with the actual battery swapping demand curve.

[0095] The overlap error rate is preset, and the overlap error between the predicted curve of a type of battery swapping demand and the actual battery swapping demand curve is calculated. If the overlap error between the predicted curve of a type of battery swapping demand and the actual battery swapping demand curve is kept within the overlap error rate, the target battery swapping cabinet is calibrated as a qualified battery swapping cabinet.

[0096] If the overlap error between the predicted battery swapping demand curve and the actual battery swapping demand curve does not remain within the overlap error rate, then the numerical weather information will be updated in real time in the battery swapping demand prediction model, and the PSC module will be corrected in real time.

[0097] If the overlap error between the predicted battery swapping demand curve and the actual battery swapping demand curve still does not remain within the overlap error rate after the numerical weather information is updated in real time and the PSC module performs real-time parameter correction, then the ADMM distributed optimization algorithm is introduced to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0098] It should be noted that an overlap analysis is performed on the predicted battery swapping demand curve and the actual battery swapping demand curve. If the overlap rate is high, it proves that the battery swapping demand of the power grid and the battery swapping cabinet matches the preset results. The preset results are predicted based on historical data and weather conditions, and are within a controllable range. Therefore, if the overlap error between the predicted and actual battery swapping demand curves remains within the overlap error rate, the target battery swapping cabinet is calibrated as a qualified battery swapping cabinet. Conversely, if the overlap error between the predicted and actual battery swapping demand curves does not remain within the overlap error rate, it proves that the predicted results do not match the actual results. In reality, there may be abnormalities in power consumption, battery swapping demand, etc. It is necessary to investigate the cause of the discrepancy and coordinate the control of the battery swapping cabinet to ensure its safe operation. Problems such as sensor data loss leading to communication delays may occur in the battery swapping cabinet, resulting in deviations. It could also be due to a mismatch between weather information and actual conditions.

[0099] Furthermore, in a preferred embodiment of the present invention, the introduction of the ADMM distributed optimization algorithm to optimize the charging and discharging strategy of the target battery swapping cabinet specifically includes:

[0100] The actual battery swapping demand curve is transformed into actual battery swapping demand data samples and fed back into the battery swapping demand prediction model. At the same time, the constraints in the battery swapping demand prediction model are updated.

[0101] Among them, the updated constraint is to adjust the lower limit of the number of target batteries available based on battery swapping demand;

[0102] The ADMM distributed optimization algorithm is introduced. Based on the ADMM distributed optimization algorithm, the target battery swapping cabinet is treated as an independent node, and the global target of the target power grid is preset, wherein the global target of the target power grid is the peak-shaving and valley-filling demand of the target power grid.

[0103] All target battery swapping cabinets are acquired, and the planning information of adjacent target battery swapping cabinets is exchanged. Based on the ADMM distributed optimization algorithm, a global consistency iteration is performed on adjacent target battery swapping cabinets. The global consistency iteration is to alternately update the local variables of the target battery swapping cabinet and the variables of adjacent target battery swapping cabinets through the ADMM distributed optimization algorithm until the target battery swapping cabinet can still match the global target of the target power grid after meeting the local battery swapping requirements.

[0104] If the target battery swapping cabinet can still match the global target of the target power grid after meeting the local battery swapping demand, the overlap error between the battery swapping demand prediction curve and the actual battery swapping demand curve will remain within the overlap error rate, thus obtaining a qualified battery swapping cabinet.

[0105] It should be noted that, firstly, actual demand data needs to be fed back to the prediction model in real time, and the predicted values ​​need to be updated in the short term, while simultaneously updating the constraints. Specifically, updating the constraints involves adjusting the lower limit of the number of available target batteries based on battery swapping demand to ensure that neither the battery swapping cabinet nor the power grid experiences excessive load. Secondly, the ADMM distributed optimization algorithm is introduced. ADMM is an algorithm capable of global consistency iteration, treating the battery swapping cabinet as an independent node that can maintain local optimization objectives, such as maximizing its own revenue and meeting user needs. The global objective is to meet the power grid's peak-shaving and valley-filling requirements. The global consistency iteration ensures that the battery swapping cabinet's batteries and the power grid remain stable, satisfying both AC / DC power conversion and feedback to the grid, as well as the steps of connecting the target battery swapping cabinet to the target power grid for charging, while preventing excessive load during the swapping process, thus resulting in a qualified battery swapping cabinet.

[0106] like Figure 3As shown, a second aspect of the present invention also provides a virtual power plant operation control system based on a battery swapping cabinet. The virtual power plant operation control system includes a memory 31 and a processor 32. The memory 31 stores a virtual power plant operation control method. When the virtual power plant operation control method is executed by the processor 32, it performs the following steps:

[0107] Real-time battery status detection and current grid load status detection are performed on the battery swapping cabinet. Based on the detection results, a battery swapping demand prediction model for the battery swapping cabinet is constructed using a long short-term memory network.

[0108] Based on the real-time load status of the target power grid, the PSC module is used to adjust the working mode of the target battery and generate the actual battery swapping demand curve.

[0109] By combining the battery swapping demand prediction model and the actual battery swapping demand curve, the ADMM distributed optimization algorithm is used to optimize the charging and discharging strategy of the target battery swapping cabinet.

[0110] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A virtual power plant operation control method based on a battery swapping cabinet, characterized in that, Includes the following steps: Real-time battery status detection and current grid load status detection are performed on the battery swapping cabinet. Based on the detection results, a battery swapping demand prediction model for the battery swapping cabinet is constructed using a long short-term memory network. Based on the real-time load status of the target power grid, the PCS module is used to adjust the working mode of the target battery and generate the actual battery swapping demand curve. The battery swapping demand curve is obtained based on the battery swapping demand prediction model. If the overlap error between the battery swapping demand curve and the actual battery swapping demand curve does not remain within the overlap error rate, and after real-time parameter correction of the PCS module, it still does not remain within the overlap error rate, then the ADMM distributed optimization algorithm is introduced to optimize the charging and discharging strategy of the target battery swapping cabinet. Specifically, the introduction of the ADMM distributed optimization algorithm to optimize the charging and discharging strategy of the target battery swapping cabinet involves: The actual battery swapping demand curve is transformed into actual battery swapping demand data samples and fed back into the battery swapping demand prediction model. At the same time, the constraints in the battery swapping demand prediction model are updated. Among them, the updated constraint is to adjust the lower limit of the number of target batteries available based on battery swapping demand; The ADMM distributed optimization algorithm is introduced. Based on the ADMM distributed optimization algorithm, the target battery swapping cabinet is treated as an independent node, and the global target of the target power grid is preset, wherein the global target of the target power grid is the peak-shaving and valley-filling demand of the target power grid. All target battery swapping cabinets are acquired, and the planning information of adjacent target battery swapping cabinets is exchanged. Based on the ADMM distributed optimization algorithm, a global consistency iteration is performed on adjacent target battery swapping cabinets. The global consistency iteration is to alternately update the local variables of the target battery swapping cabinet and the variables of adjacent target battery swapping cabinets through the ADMM distributed optimization algorithm until the target battery swapping cabinet can still match the global target of the target power grid after meeting the local battery swapping requirements. If the target battery swapping cabinet can still match the global target of the target power grid after meeting the local battery swapping demand, the overlap error between the battery swapping demand prediction curve and the actual battery swapping demand curve will remain within the overlap error rate, thus obtaining a qualified battery swapping cabinet.

2. The virtual power plant operation control method based on a battery swapping cabinet as described in claim 1, characterized in that, The process involves real-time battery status detection and current grid load status detection for the battery swapping cabinet. Based on the detection results, a battery swapping demand prediction model for the cabinet is constructed using a long short-term memory network. Specifically: Identify the battery swapping cabinets that require operational control, designate them as target battery swapping cabinets, and designate the batteries in the target battery swapping cabinets as target batteries. Inside the target battery swapping cabinet, a power control unit is acquired. Based on the power control unit, the real-time quantity of the target batteries and the real-time remaining power of the target batteries are detected in the battery swapping cabinet. At the same time, the current output power value of the target batteries is calculated based on the power control unit. Collect the real-time temperature and state of charge of different target batteries, and combine them with the current output power value of the target battery to generate the real-time status of the target battery. The power grid connected to the target battery swapping cabinet is identified as the target power grid, and the real-time load status of the target power grid is read. The real-time load status of the target power grid includes the power grid load power, load distribution, and load fluctuation status. A big data network is introduced and connected to the target battery swapping cabinet. Based on the big data network, the historical battery swapping data of the target battery swapping cabinet is retrieved and determined. At the same time, the battery swapping demand period is preset. Based on the big data network, the weather information of the target battery swapping cabinet's location during the battery swapping demand period is retrieved. A long short-term memory network algorithm is introduced. Based on the long short-term memory network algorithm, combined with the real-time status of the target battery, the real-time load status of the target power grid, the historical battery swapping data of the target battery swapping cabinet, and the weather information of the target battery swapping cabinet's location during the battery swapping demand period, a battery swapping demand prediction model is constructed.

3. The virtual power plant operation control method based on a battery swapping cabinet as described in claim 2, characterized in that, The construction of the battery swapping demand prediction model specifically involves: Based on the battery swapping demand period, a timestamp is constructed and marked as the target timestamp. The time step in the target timestamp is 24 cycles, calculated in hours. Data analysis is performed on the weather information of the target battery swapping cabinet location during the battery swapping demand period, and the numerical weather information is retained and processed, while the weather status is converted into unique thermal codes. Based on the timestamp, the real-time status of the target battery, the real-time load status of the target power grid, the historical battery swapping data of the target battery swapping cabinet, and numerical weather information within 24 cycles are sampled and integrated so that the data in one sample corresponds to one cycle. The Long Short-Term Memory (LSTM) network algorithm is introduced, and a LSTM blank model is constructed. All samples and timestamps are imported into the LSTM blank model, and one-hot encoding is set in the LSTM blank model to obtain the preliminary LSTM network model. Obtain the network architecture of the initially defined long short-term memory network model, wherein the network architecture of the initially defined long short-term memory network model includes a fully connected layer and an LSTM layer; In the initial setup of the Long Short-Term Memory network model, loss functions are set in different LSTM layers. The loss function is set as the mean squared error of the samples. The LSTM layers are trained by continuously calculating the mean squared error of the samples, and the mean squared error is tuned in different LSTM layers based on the Bayesian optimization algorithm. A standard training step size is preset. When the training step size of the LSTM layer is equal to the standard training step size, training of all LSTM layers is stopped, and all LSTM layers are connected to the fully connected layer to obtain the battery swapping demand prediction model. At the same time, the battery swapping demand prediction curve is output in the fully connected layer. The battery swapping demand forecast curve predicts the battery swapping demand for different target batteries in the next 24 hours. The battery swapping demand records the rate of change of the output power of the target battery and the change of the load status of the target power grid.

4. The virtual power plant operation control method based on a battery swapping cabinet as described in claim 1, characterized in that, The process involves adjusting the operating mode of the target battery using the PCS module based on the real-time load status of the target power grid, and generating an actual battery swapping demand curve. Specifically: Based on big data networks, determine the grid load power corresponding to the target power grid when it is in a trough state, a peak state, and a power outage state; Determine the real-time load power of the target power grid, and analyze the real-time load status of the target power grid based on the power grid load range corresponding to the target power grid being in the trough state, peak state, and power outage state respectively. A PCS module is installed in the target battery swapping cabinet. The PCS module includes a rectifier and an inverter, which are used to convert AC power to DC power in the target battery swapping cabinet. If the real-time load power of the target power grid is within the range of the power grid load power corresponding to the trough state, then the target battery swapping cabinet is controlled to connect to the target power grid and charge the target battery through the PCS module. If the real-time load power of the target power grid is within the range of the power grid load power corresponding to the peak state, then the PCS module controls the target battery swapping cabinet to convert the electrical energy in the target battery into AC power and feed it into the target power grid. If the real-time load power of the target power grid is within the power grid load power range corresponding to the power outage state, then the load connected to the target battery swapping cabinet is obtained, and through the PCS module, the target battery swapping cabinet is controlled to invert the electrical energy in the target battery into AC power and provide electrical energy to the load connected to the target battery swapping cabinet. During the operation of the PCS module, a battery swapping demand curve is output in real time and calibrated as the actual battery swapping demand curve, wherein the actual battery swapping demand curve records the battery swapping demand of the target battery.

5. The virtual power plant operation control method based on a battery swapping cabinet as described in claim 1, characterized in that, The process of obtaining the battery swapping demand curve based on the battery swapping demand prediction model is as follows: Based on the battery swapping demand prediction model, the battery swapping demand prediction curve is obtained, and the time point corresponding to the generation of the actual battery swapping demand curve is determined and labeled as a type of time point. Based on the first type of time point, the curve is truncated from the battery swapping demand prediction curve to obtain a second type of battery swapping demand prediction curve, and the second type of battery swapping demand prediction curve is analyzed to overlap with the actual battery swapping demand curve. The overlap error rate is preset, and the overlap error between the predicted curve of a type of battery swapping demand and the actual battery swapping demand curve is calculated. If the overlap error between the predicted curve of a type of battery swapping demand and the actual battery swapping demand curve is kept within the overlap error rate, the target battery swapping cabinet is calibrated as a qualified battery swapping cabinet. If the overlap error between the predicted battery swapping demand curve and the actual battery swapping demand curve does not remain within the overlap error rate, then the numerical weather information will be updated in real time in the battery swapping demand prediction model, and the parameters of the PCS module will be corrected in real time. If the overlap error between the predicted battery swapping demand curve and the actual battery swapping demand curve still does not remain within the overlap error rate after the numerical weather information is updated in real time and the PCS module performs real-time parameter correction, then the ADMM distributed optimization algorithm is introduced to optimize the charging and discharging strategy of the target battery swapping cabinet.

6. A virtual power plant operation control system based on a battery swapping cabinet, characterized in that, The virtual power plant operation control system includes a memory and a processor. The memory stores a virtual power plant operation control method program based on a battery swapping cabinet. When the processor executes the virtual power plant operation control method program based on a battery swapping cabinet, it implements the steps of the virtual power plant operation control method based on a battery swapping cabinet as described in any one of claims 1-5.

Images