A transformer flexible expansion method and system based on historical operation data

Abstract

The present invention discloses a method and system for flexible capacity expansion of transformers based on historical operating data. The method includes the following steps: predicting the power consumption of load equipment under the distribution transformer based on historical operating data; predicting the power generation of photovoltaic power generation equipment under the distribution transformer based on historical operating data; calculating the remaining capacity based on the rated capacity of the distribution transformer, the total load, and the total photovoltaic grid-connected capacity; when the remaining capacity is insufficient, controlling the energy storage device to discharge or to provide mutual assistance through a flexible interconnection device; when the remaining capacity is sufficient, controlling the energy storage device to charge or to transfer power to an adjacent transformer. The present invention installs a low-voltage flexible interconnection device between two transformers, predicts and analyzes the peak moments of the superimposed transformer loads based on the transformer loads and the historical data of the operation of each device, and adopts an edge-side energy management strategy that takes demand into consideration. Through flexible mutual assistance technology, efficient allocation and transfer of power resources are achieved to meet the power demand of the flexible capacity expansion of the transformer area.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network optimization and control, and in particular relates to a transformer flexible expansion method and system based on historical operation data. Background Art

[0002] With the development of society and the improvement of people's living standards, the demand for electricity continues to increase. Many transformers are designed with a certain amount of spare capacity. However, with the frequent occurrence of extreme weather and the continued progress of electricity substitution, the electricity load is increasing, and the original capacity is no longer sufficient. For example, in the public power distribution sector, the accelerated growth of electricity consumption and peak load has brought about the need for additional capacity expansion of distribution transformers. The large-scale activation of seasonal load equipment and the sudden increase in power load during peak power consumption periods cause transformers to operate under overload. The production, commissioning, and load aging of factory modules occupy a large amount of distribution capacity. In addition, the rapid development of new energy vehicles has led to an increase in the number of charging stations, further increasing the demand for transformer expansion.

[0003] Existing capacity expansion technologies are mostly based on hard expansion of transformers. Based on collected transformer operating data, transformers that require expansion are screened for hardware expansion. While the expansion of some distribution transformers can maximize performance gains across the entire substation, reduce volatility, and ensure safety, it suffers from practical drawbacks such as long process cycles, high costs, and poor flexibility. This makes it impossible to quickly and cost-effectively resolve the problem of insufficient transformer capacity during short-term peak periods. Existing flexible capacity expansion relies heavily on energy storage devices. Failure of these devices paralyzes flexible capacity expansion, rendering it ineffective and unable to truly address the transformer capacity expansion issue. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a transformer flexible capacity expansion method and system based on historical operation data. The technical solutions provided by the present invention are as follows:

[0005] A method for flexible capacity expansion of transformers based on historical operating data, for two adjacent distribution transformers connected by a flexible interconnection device, wherein the flexible interconnection device is used to transmit power between different distribution transformers through a low-voltage power flow mutual assistance device, includes the following steps:

[0006] S1. Based on historical operating data, predict the power consumption of the load equipment under each distribution transformer to obtain the total energy load under each distribution transformer at the next moment;

[0007] S2. Based on historical operating data, predict the power generation of the photovoltaic power generation equipment under each distribution transformer to obtain the photovoltaic power generation power under each distribution transformer at the next moment;

[0008] S3. Calculate the remaining capacity of the distribution transformer at the next moment based on the rated capacity of the distribution transformer, the total energy load under the distribution transformer at the next moment, and the total amount of photovoltaic grid-connected power;

[0009] S4. If the remaining capacity of a distribution transformer at the next moment is less than the set threshold, the power demand of the distribution transformer area is met by controlling the discharge of the energy storage device under the distribution transformer and / or flexibly transferring power from the adjacent distribution transformer through the flexible interconnection device;

[0010] S5. If the remaining capacity of a distribution transformer at the next moment is greater than a set threshold, the distribution transformer's load capacity can be fully utilized by controlling the charging of its subordinate energy storage devices and / or by flexibly transferring power to adjacent distribution transformers through a flexible interconnection device. Energy storage charging and discharging can also be considered economically, taking electricity costs into account.

[0011] Furthermore, step S1 includes:

[0012] Data collection: Install smart meters or remote monitoring equipment under each distribution transformer to collect real-time historical electricity consumption data of various electrical devices, including power consumption, as well as corresponding weather, seasonal and holiday information;

[0013] Data preprocessing: Use cleaning and denoising techniques on the collected historical electricity consumption data to remove abnormal data points;

[0014] Establish a load forecasting model: Based on historical electricity consumption data and corresponding weather, seasonal and holiday information, build and train a machine learning model to predict future electricity load;

[0015] Load forecasting: Based on the model prediction results, the total power consumption forecast value of each distribution transformer at the next moment will be generated, and this value will be used as a reference for subsequent flexible scheduling.

[0016] Furthermore, the load forecasting model adopts the GBDT model. The mean absolute error is used to evaluate the model quality during GBDT model training. If 1-e mae / CL max ≥85%, the model meets the application requirements, e mae Indicates the calculated mean absolute error, CL max is the maximum power load value in the data set.

[0017] Furthermore, step S2 includes:

[0018] Data collection: Collect historical power generation data of the photovoltaic system and corresponding meteorological data. The historical power generation data includes daily power generation, instantaneous power, daily power generation peak, and average power generation efficiency. The meteorological data includes solar irradiance, temperature, sunshine duration, wind speed, humidity, and cloud cover.

[0019] Data preprocessing: Cleaning and denoising techniques are used on the collected historical power generation data to remove abnormal data points, and the historical power generation data and meteorological data are aligned at the same time intervals;

[0020] Establishing a photovoltaic forecasting model: Building and training a photovoltaic forecasting model based on historical power generation data and meteorological data to predict power generation at future times;

[0021] Power generation forecast: Meteorological data is input into the trained photovoltaic forecast model to generate a forecast value of future photovoltaic power generation.

[0022] Furthermore, the photovoltaic prediction model includes a physical power generation model based on the physical characteristics of the photovoltaic system and a data-driven machine learning model; the physical power generation model calculates the theoretical power generation of the photovoltaic module by inputting meteorological data; the machine learning model includes ARIMA, SARIMA, random forest and deep learning network.

[0023] A transformer flexible capacity expansion system based on historical operating data, comprising:

[0024] Data acquisition module: responsible for collecting the operating data of the distribution transformer and its subordinate equipment and transmitting it to the edge gateway module, including the power consumption data of the load equipment, the power generation data of the photovoltaic power generation equipment, the charging and discharging status of the energy storage equipment, and meteorological data;

[0025] Data preprocessing module: used to clean, denoise and detect anomalies of the collected raw data to ensure data accuracy and consistency and filter out faulty and incomplete data;

[0026] Load forecasting module: Based on historical operating data, it predicts the power load of the distribution transformer in the next period, generates a load demand curve, and provides it to the subsequent management and optimization platform for analysis and decision-making;

[0027] Photovoltaic power generation prediction module: Based on the historical power generation data of photovoltaic power generation equipment and weather forecast information, it predicts future power generation to ensure that photovoltaic resources can reasonably participate in load scheduling;

[0028] Energy storage management module: used to manage the charging and discharging process of energy storage equipment, ensuring that the energy storage equipment can be discharged in time to relieve load pressure during peak load, and charged to reserve electricity during low load;

[0029] Flexible interconnection device control module: used to achieve flexible allocation of power resources between two adjacent distribution transformers through the flexible interconnection device. When a distribution transformer is overloaded, this module will allocate power from the adjacent distribution transformer to reduce the load;

[0030] Edge Gateway Module: This module performs real-time data processing and short-term forecasting on edge devices close to the transformers, and uploads the processed data to the management platform, reducing computing and communication delays and ensuring timely and accurate load regulation.

[0031] Management and optimization platform: As the central command module of the system, it receives all data from the edge gateway module, performs global analysis and policy optimization, generates scheduling plans and sends them to each edge node.

[0032] Furthermore, the photovoltaic power generation prediction module is implemented by establishing a photovoltaic power generation prediction model based on data drive or physical characteristics of the photovoltaic system, and communicates with the meteorological data service interface in real time to obtain meteorological data including solar irradiance, temperature, sunshine duration, wind speed, humidity, and cloud cover in a timely manner.

[0033] Furthermore, the energy storage management module dynamically adjusts the charging and discharging behavior of the energy storage equipment based on load forecasts and real-time power demand; the module is equipped with an economic optimization algorithm to optimize charging timing and discharging strategies through price signals and load demand.

[0034] Furthermore, the control modes of the flexible interconnection device control module include load balancing mode, microgrid economic mode and safe supply mode:

[0035] The load balancing mode detects the load factor difference between the transformers on both sides of the flexible interconnection device. When the load factor difference between the transformers on both sides is less than the set threshold value due to the mutual assistance of the power flow, the load balancing mode control is activated.

[0036] The microgrid economic model refers to the interconnection of microgrids or substations with photovoltaic power generation equipment and energy storage equipment. It fully considers the differences in the distributed source-storage status and peak-valley electricity prices between the two substations, builds an optimized and adjusted scheduling strategy with the goal of minimizing the purchase cost of electricity, and controls the timing of power flow mutual assistance and energy storage charging and discharging by controlling flexible interconnection devices to achieve the goal of saving electricity costs.

[0037] The safe power supply mode means that when one of the substations loses power or fails, the load is transferred in real time through flexible interconnection devices to help the substation restore power.

[0038] Furthermore, the system uses a load forecasting module to predict the load conditions in future time periods and generate a load demand curve. At the same time, the photovoltaic power generation forecasting module predicts future photovoltaic power generation based on historical power generation data and weather forecast information. The system summarizes and analyzes the rated capacity, real-time total load, and total photovoltaic grid-connected capacity of each transformer to calculate the remaining capacity of the transformer. When the remaining capacity of the transformer is insufficient, the energy storage management module will dispatch the energy storage device to start discharging to fill the load gap. If the energy storage device cannot fully meet the demand, the flexible interconnection device control module will allocate electricity from adjacent transformers for transfer to ensure the dynamic balance of the power load. When the load is low, the system will control the energy storage device to charge, or transfer excess electricity to adjacent transformers through the flexible interconnection device.

[0039] Compared with the prior art, the present invention has at least the following beneficial effects:

[0040] 1. Low cost: The biggest advantage of flexible capacity expansion technology is its low cost. By tapping into the peak energy consumption that can be exploited during actual transformer operation, the existing transformer load capacity can be maximized, meeting the needs of transformer capacity expansion without changing the existing cable laying.

[0041] 2. Multi-faceted considerations: While considering the flexible expansion of transformers, we also consider demand control during transformer operation, load balancing from the perspective of transformer life, and safety guarantees for transformer overload operation, among other strategic models, to further promote the stable and reliable use of transformers in the park.

[0042] 3. Flexible and adaptable operation: The entire system does not change the installation of the original equipment. Through the edge-side collection and computing gateway, the management policy monitoring platform is established, and the joint control module is deployed on the platform and edge side to complete multiple modes of transformer flexible expansion and operation technology under different weather conditions, different holiday conditions, different equipment operating conditions, and different equipment fault status combinations. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0044] Figure 1 This is a schematic diagram of the hierarchical composition of subordinate equipment of an existing distribution transformer in a certain park provided by an embodiment of the present invention;

[0045] Figure 2 This is a flow chart of a transformer flexible capacity expansion method provided by one embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the GBDT architecture provided by one embodiment of the present invention;

[0047] Figure 4 This is a comparison chart of the average load prediction value and actual value of a charging station provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, other embodiments obtained by ordinary technicians in this field without making creative efforts are all within the scope of protection of the present invention.

[0049] Example 1

[0050] This embodiment provides a transformer flexible capacity expansion method based on historical operating data.

[0051] like Figure 1 As shown in the figure, the hierarchical composition of the existing distribution transformer subordinate equipment in a typical park is given, including two distribution transformers, and a low-voltage power flow flexible interconnection device is set between the distribution transformers. The main purpose of the transformer flexible expansion method based on historical operation data is to achieve efficient allocation and transfer of power resources between two adjacent distribution transformers through the flexible interconnection device to meet the power demand of the transformer area and make full use of the capacity of photovoltaic power generation and energy storage equipment. Figure 2 As shown, the method mainly includes the following steps:

[0052] Step S1: Predict the power consumption of the load equipment under the distribution transformer based on historical operation data

[0053] 1. Data collection:

[0054] Install smart meters or remote monitoring equipment under each distribution transformer to collect historical data of various electrical equipment in real time, including electrical parameters such as power, voltage, and current.

[0055] The collected data should cover various typical operating conditions, such as seasonal load changes, differences in electricity consumption between holidays and non-holidays, and the impact of special operating conditions (such as equipment maintenance periods or abnormal shutdowns).

[0056] 2. Data preprocessing:

[0057] The collected historical electricity consumption data is pre-processed, and cleaning and denoising techniques are used to remove abnormal data points (such as short-term voltage fluctuations or momentary power outages caused by equipment failures).

[0058] Use algorithms such as moving average to smooth the data and eliminate the interference of short-term fluctuations on load forecasting to improve the accuracy of the forecasting model.

[0059] 3. Establish a load forecasting model:

[0060] Based on historical data, machine learning algorithms such as time series models (such as ARIMA), regression models (such as GBDT), or deep learning algorithms (such as LSTM) are used to predict future electricity load.

[0061] When training the model, time factors (such as daily, weekly, and monthly periodic fluctuations), equipment operation modes (the difference between continuous production equipment and intermittent equipment), and external factors (such as the impact of temperature changes on air conditioning loads) should be considered.

[0062] 4. Load forecasting application:

[0063] Based on the model prediction results, the total power consumption forecast value of each distribution transformer at the next moment will be generated, and this value will be used as a reference for subsequent flexible scheduling.

[0064] Step S2: Predict the power generation of the photovoltaic power generation equipment under the distribution transformer based on historical operation data

[0065] 1. Data Collection

[0066] (1) Historical data of photovoltaic power generation equipment:

[0067] Collect historical power generation data for the PV system, with a recommended sampling interval of 15 minutes or less. This data includes daily power generation, instantaneous power, daily peak power generation, and average power generation efficiency. This data can be obtained from power generation data recorded by the PV inverter or collected in real time through a remote monitoring system (such as a SCADA system).

[0068] (2) Acquisition of meteorological data:

[0069] Meteorological monitoring sensors such as solar irradiance and temperature are installed at photovoltaic power generation sites to collect meteorological data in real time.

[0070] Solar irradiance: reflects the intensity of solar energy resources and is the main influencing factor of photovoltaic power generation.

[0071] Temperature: Temperature changes will affect the conversion efficiency of photovoltaic modules. The power generation efficiency of photovoltaic modules decreases at high temperatures.

[0072] Sunshine duration: reflects the effective time that photovoltaic modules are exposed to sunlight, and is another important factor in calculating power generation.

[0073] Factors such as wind speed, humidity, and cloud cover: These data will have an indirect impact on power generation, especially cloud cover will affect changes in solar irradiance.

[0074] 2. Data preprocessing

[0075] (1) Data cleaning:

[0076] Handle missing values ​​and outliers in historical data. If there are gaps or outliers in PV power generation data (such as negative power generation), interpolation or mean filling is required. Filter or correct anomalies in meteorological data (such as extreme weather and sensor failures).

[0077] (2) Data alignment:

[0078] To synchronize PV power generation data with meteorological data, the two types of data must be aligned at the same time interval. If the sampling intervals of meteorological and power generation data differ, alignment can be achieved through interpolation or averaging. This ensures that the data are continuous in time series for subsequent time series analysis.

[0079] 3. Prediction model selection and training

[0080] In order to improve the accuracy of photovoltaic power generation prediction, a suitable model can be selected according to the amount and complexity of the data. There are two main types of models to choose from:

[0081] (1) Physical model (based on the physical characteristics of the photovoltaic system):

[0082] A physical power generation model is constructed based on the photovoltaic conversion efficiency and temperature coefficient of solar cell modules. By inputting data such as solar irradiance, temperature, and sunshine duration, the theoretical power generation of the photovoltaic modules is calculated. This model accurately reflects the physical characteristics of the photovoltaic system and is suitable for smaller-scale photovoltaic systems. However, its ability to handle complex weather conditions is limited, especially when dealing with large amounts of data, as the model is relatively simple.

[0083] (2) Machine learning model (data-driven model):

[0084] ARIMA (Autoregressive Integrated Moving Average Model): Applicable to situations where photovoltaic power generation has seasonality and periodicity, and short-term prediction can be performed by adjusting model parameters.

[0085] SARIMA (Seasonal ARIMA): Taking seasonal factors into account, a more accurate forecast of the fluctuation trend of photovoltaic power generation is made.

[0086] Random Forest: This algorithm builds multiple decision trees and combines meteorological and historical power generation data for prediction, making it suitable for processing multi-dimensional features.

[0087] LSTM (Long Short-Term Memory Network): Based on the characteristics of time series data, LSTM can capture the long-term dependency between historical power generation and future power generation, making it very suitable for processing large-scale data, especially nonlinear changes.

[0088] (3) Model training:

[0089] Based on historical power generation and weather data, the model is trained and hyperparameters are adjusted. The trained model can then be fed with real-time or projected weather data to generate power generation forecasts for future periods.

[0090] 4. Power generation forecast

[0091] By inputting meteorological forecast data such as solar irradiance, temperature, and sunshine duration into a trained model, a forecast of future photovoltaic power generation is generated. The model outputs the predicted photovoltaic power generation value for a future time (e.g., the next hour or the next day) and generates a power generation curve. Hourly and daily power generation forecasts can be generated based on actual needs.

[0092] Step S3: Calculate the remaining capacity based on the rated capacity of the distribution transformer, the total load and the total amount of photovoltaic grid-connected

[0093] 1. Rated capacity and real-time calculation:

[0094] Based on the rated capacity information on the transformer nameplate, the theoretical load capacity of each distribution transformer is obtained. Then, based on the total load predicted in steps S1 and S2 and the total amount of PV grid-connected, the real-time remaining capacity of each transformer is calculated using the following formula: Remaining capacity = rated capacity - (total load - total amount of PV grid-connected). If the remaining capacity falls below the safety threshold, the transformer may be overloaded, and immediate regulatory measures are required.

[0095] 2. Real-time monitoring and analysis:

[0096] The load changes of each transformer are monitored in real time. The data can be transmitted to the centralized management platform through the edge computing gateway. The platform will analyze the load conditions and automatically generate a response strategy.

[0097] Step S4: When the remaining capacity is insufficient, control the energy storage device to discharge or transfer power through the flexible interconnection device

[0098] 1. Discharge scheduling of energy storage equipment:

[0099] When the system detects that the remaining capacity of a distribution transformer falls below a set threshold, it automatically dispatches the transformer's energy storage devices to begin discharging. Energy storage devices such as lithium batteries and supercapacitors gradually release energy based on the predicted load shortfall, ensuring the transformer's safe operation.

[0100] The discharge process is automatically adjusted by the management system to ensure that the discharge speed matches the load gap, and real-time monitoring and feedback are carried out through the edge computing platform.

[0101] 2. Power supply dispatching of flexible interconnection devices:

[0102] If energy storage equipment cannot fully compensate for the load shortfall, the system uses a flexible interconnection device to transfer excess power from adjacent distribution transformers to the more heavily loaded transformer, alleviating overload pressure. The flexible interconnection device transfers power between transformers using low-voltage power flow mutualization equipment, ensuring the real-time and stable dynamic load scheduling.

[0103] Step S5: When there is sufficient remaining capacity, control the energy storage device to charge or transfer power to the adjacent transformer

[0104] 1. Energy storage equipment charging scheduling:

[0105] When a transformer has excess capacity, the energy storage device is dispatched into charging mode. The charging process is optimized based on predicted future load demand and electricity price fluctuations to ensure maximum economic benefits. The charging rate and time are dynamically adjusted based on predicted future load curves to avoid unnecessary energy waste.

[0106] 2. Transfer power to adjacent transformers:

[0107] If adjacent transformers are in a high-load state and have insufficient energy storage, power resources will be automatically transferred through flexible interconnection devices to help relieve pressure on adjacent transformers and achieve efficient scheduling and optimized allocation of grid resources.

[0108] Example 2

[0109] This embodiment provides a transformer flexible expansion method based on historical operating data. Unlike the first embodiment, this embodiment adopts a gradient boosting decision tree (GBDT) algorithm in the load equipment power consumption prediction link. The load of the load equipment in the next few hours and days is predicted based on historical power consumption data and related weather, season, holiday and other characteristics.

[0110] Specifically, historical electricity consumption data refers to the historical electricity consumption data of the previous ten hours or the previous ten days at the prediction time point; weather refers to the weather forecast data of the predicted date obtained based on the date, including temperature, wind speed, humidity, etc.; season refers to one of the four seasons of spring, summer, autumn and winter corresponding to the predicted date calculated based on the date; whether it is a holiday refers to whether the predicted date is a holiday calculated based on the date.

[0111] Load forecasting is a typical machine learning regression prediction problem. The complete algorithm process includes four parts: data preprocessing, feature extraction, model training and model evaluation.

[0112] 1. Data preprocessing

[0113] Missing value imputation

[0114] 2. Feature extraction

[0115] According to the description of the prediction features, the corresponding features are obtained for each time point that needs to be predicted, including historical electricity consumption data, weather, season, and whether it is a holiday.

[0116] 3. Model training

[0117] Using the above data preprocessing and feature extraction methods, construct the training data set and the validation data set. Use the training data set to train the GBDT model. Figure 3 As shown in Figure 2, the core concept of the GBDT algorithm is to combine multiple decision trees to form a powerful ensemble model. During training, the GBDT algorithm iteratively constructs a series of decision trees, fitting the residuals from the previous round to reduce error. In each iteration, the algorithm constructs a new decision tree based on the gradient information of the current dataset and updates the model's predictions. The final prediction is obtained by taking the weighted sum of the predictions from all decision trees.

[0118] Specifically, the process of the GBDT algorithm is as follows:

[0119] 1) Initialize an empty decision tree set;

[0120] 2) For each iteration:

[0121] a. Calculate the gradient information of the current data set;

[0122] b. Build a new decision tree based on the gradient information;

[0123] c. Add the newly constructed decision tree to the set;

[0124] d. Use the updated decision tree set to predict the data;

[0125] e. Calculate the prediction error and update the gradient information;

[0126] 3) Repeat step 2) until the preset number of iterations or error threshold is reached.

[0127] 4. Model Evaluation

[0128] The GBDT model is verified on the validation set and evaluated using MAE (mean absolute error) and RMSE (root mean square error). Assume that the error calculated by the MAE of the model is e mae , the largest power load in the data set is CL maxThe default minimum value is 0. The prediction effect must meet the following requirements: 1-e mae / CL max ≥85%. Figure 4 As shown in Figure 1, a comparison chart of the effects of using the above prediction method to predict the load of a charging station under a distribution transformer is shown. It can be seen from the figure that the predicted load curve is basically consistent with the actual load curve, and the prediction deviation at each moment is kept within the allowable deviation range.

[0129] Example 3

[0130] Based on the methods described in the preceding embodiments, this embodiment provides a flexible transformer capacity expansion system based on historical operating data. This system aims to dynamically achieve flexible transformer capacity expansion by fully utilizing historical operating data, real-time load monitoring, energy storage management, and the rational allocation of photovoltaic power generation. The system will include multiple modules, each with specific functions, working together to ensure that the transformer can operate efficiently under different load conditions. The following is a detailed description of the system's module structure, functional description, and workflow.

[0131] 1. System modules and functions

[0132] (1) Data acquisition module:

[0133] Responsible for collecting all operational data from distribution transformers and their associated equipment, including load power consumption data, photovoltaic power generation data, energy storage device charge and discharge status, and meteorological data. This data is primarily used to support subsequent data analysis and forecasting. Specifically, this data may include smart meters and sensors installed on transformers, photovoltaic equipment, and energy storage devices, transmitting real-time data to edge computing nodes. Collected data includes operating parameters such as voltage, current, power, frequency, and temperature.

[0134] (2) Data preprocessing module:

[0135] It is used to clean, denoise, and detect anomalies in the collected raw data to ensure data accuracy and consistency and filter out faulty and incomplete data. Data smoothing algorithms and anomaly detection algorithms (such as those based on statistical analysis or machine learning) can be used to clean and preprocess the input data.

[0136] (3) Load forecasting module:

[0137] Based on historical operating data, the power load of distribution transformers in the next period is predicted, and a load demand curve is generated. This curve is then provided to the subsequent capacity management module for analysis and decision-making. Load forecasting models can be built using time series analysis (such as the ARIMA model) or deep learning algorithms (such as LSTM). The forecast is primarily based on historical load data, seasonal variations, and the impact of holidays.

[0138] (4) Photovoltaic power generation prediction module:

[0139] Based on historical PV generation data and weather forecasts, this module predicts future power generation, ensuring that PV resources can participate appropriately in load scheduling. This can be achieved by establishing a PV power generation prediction model based on data-driven or physical characteristics of the PV system, taking into account parameters such as sunshine duration, light intensity, and temperature. This module communicates in real time with a weather data service interface to ensure the accuracy of power generation predictions.

[0140] (5) Energy storage management module:

[0141] This module manages the charging and discharging of energy storage devices, ensuring timely discharge to alleviate load pressure during peak loads and charging to conserve energy during low loads. The energy storage scheduling strategy dynamically adjusts the charging and discharging behavior of energy storage devices based on load forecasts and real-time power demand. An internal economic optimization algorithm optimizes charging timing and discharging strategies based on price signals and load demand.

[0142] (6) Flexible interconnection device control module:

[0143] A flexible interconnection device enables flexible allocation of power resources between two adjacent transformers. When a transformer becomes overloaded, the module diverts power from a neighboring transformer to reduce the load. The flexible interconnection device utilizes low-voltage power flow mutual assistance technology, combined with the transformer's real-time operating status and remaining capacity, to ensure dynamic load balancing. The control module collaborates with the transformer and energy storage device through a communication protocol.

[0144] In some embodiments, the flexible interconnection device control module mainly includes three control modes: load balancing mode, microgrid economic mode and safe supply mode.

[0145] The load balancing mode detects the load factors of the transformers on both sides of the flexible interconnection device and achieves transformer load balancing through power flow mutual assistance. The goal is to make the load factor deviation of the transformers on both sides less than the set threshold value (the default is 10%).

[0146] The microgrid economic model refers to the process of interconnecting microgrids or substations with photovoltaic power generation equipment and energy storage equipment, taking into full consideration the resource endowment differences of the two substations in terms of distributed source-storage status and peak-valley electricity prices, and building an optimized and adjusted scheduling strategy with the goal of minimizing electricity purchase costs. By controlling the flow through flexible interconnection devices, the goal of saving electricity costs is achieved.

[0147] The safe power supply mode means that when one of the substations loses power or fails, the flexible interconnection device can support the real-time transfer of important loads to help the substation restore power.

[0148] (7) Edge computing module:

[0149] Real-time data processing and short-term forecasting are performed on edge devices close to the transformer, reducing computational and communication latency and ensuring timely and accurate load regulation. This module provides data acquisition, processing, analysis, storage, and short-term forecasting capabilities, interacting directly with cloud servers. It also performs data preprocessing, status monitoring, and fault detection, and uploads processed data to the cloud.

[0150] (8) Management and optimization platform (cloud):

[0151] As the system's central command module, it receives all data from edge computing nodes, performs global analysis and policy optimization, generates scheduling plans, and distributes them to each edge node. Relying on big data analysis and optimization algorithms, the platform dispatches and optimizes power resources across the entire park. Its primary tasks include load balancing, economic optimization, energy storage management, and photovoltaic scheduling. The management platform supports remote monitoring and maintenance.

[0152] 2. System workflow

[0153] (1) Data collection and preprocessing:

[0154] Smart meters and sensors on the transformer and its associated equipment continuously collect real-time operating data from various electrical devices, including load power consumption, photovoltaic power generation, and the charge and discharge status of energy storage devices. All data is transmitted to edge computing nodes via a communications network. A data preprocessing module cleans and removes noise from the raw data to ensure accuracy and consistency.

[0155] (2) Load and power generation forecast:

[0156] The system's load forecasting module uses historical load data, holidays, seasonal variations, and other factors to predict future load conditions and generate load demand curves. Simultaneously, the photovoltaic power generation forecasting module uses historical power generation data and weather forecasts to predict future photovoltaic power generation, ensuring the proper scheduling of photovoltaic power.

[0157] (3) Real-time load monitoring and remaining capacity calculation:

[0158] The system aggregates and analyzes data such as each transformer's rated capacity, total real-time load, and total PV grid-connected capacity to calculate the transformer's remaining capacity. If a transformer's remaining capacity approaches a critical value (threshold), the system enters dispatch mode.

[0159] (4) Energy storage management and flexible interconnection device scheduling:

[0160] When the transformer's remaining capacity is insufficient, the energy storage management module dispatches the energy storage device to discharge and fill the load gap. If the energy storage device cannot fully meet demand, the flexible interconnection device control module diverts power from adjacent transformers to ensure dynamic load balance. When the load is low, the system controls the energy storage device to charge or transfers excess power to adjacent transformers via the flexible interconnection device.

[0161] (5) Management and optimization:

[0162] The management platform (cloud) monitors the entire system's operating status in real time and dynamically adjusts each transformer's operating strategy and power dispatch plan based on load forecasts, photovoltaic power generation forecasts, and energy storage device status. An optimization algorithm analyzes load distribution, energy storage status, and electricity prices to develop cost-effective dispatch strategies, ensuring safe system operation while maximizing the grid's economic benefits.

[0163] In summary, the system achieves flexible transformer expansion through modular design. The modules collaborate through edge computing and a cloud-based management platform, ensuring the system can dynamically dispatch power resources under varying load conditions, optimizing grid efficiency, reducing hardware expansion costs, and extending transformer life.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Under the idea of ​​the present invention, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the present invention as described above. For the sake of simplicity, they are not provided in detail. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in this field should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for flexible capacity expansion of transformers based on historical operating data, wherein two adjacent distribution transformers are connected via a flexible interconnection device, wherein the flexible interconnection device is used to transmit power between the different distribution transformers via a low-voltage power flow mutual assistance device, and a flexible interconnection device control module realizes flexible allocation of power resources between the two adjacent distribution transformers via the flexible interconnection device, characterized in that: The following steps are involved: S1. Based on historical operating data, predict the power consumption of the load equipment under each distribution transformer to obtain the total energy load under each distribution transformer at the next moment; S2. Based on historical operating data, predict the power generation of the photovoltaic power generation equipment under each distribution transformer to obtain the photovoltaic power generation power under each distribution transformer at the next moment; S3. Calculate the remaining capacity of the distribution transformer at the next moment based on the rated capacity of the distribution transformer, the total energy load under the distribution transformer at the next moment, and the total amount of photovoltaic grid-connected power; S4. If the remaining capacity of a distribution transformer at the next moment is less than the set threshold, the power demand of the distribution transformer area is met by controlling the discharge of the energy storage device under the distribution transformer and / or flexibly transferring power from the adjacent distribution transformer through the flexible interconnection device; S5. If the remaining capacity of a distribution transformer at the next moment is greater than a set threshold, the load capacity of the distribution transformer is fully utilized by controlling the charging of the energy storage device under the distribution transformer and / or flexibly transferring power to the adjacent distribution transformer through the flexible interconnection device; The control modes of the flexible interconnection device control module include load balancing mode, microgrid economic mode and safe supply mode: The load balancing mode detects the load factor difference between the transformers on both sides of the flexible interconnection device. When the load factor difference between the transformers on both sides is less than the set threshold value due to the mutual assistance of the power flow, the load balancing mode control is activated. The microgrid economic model refers to the interconnection of microgrids or substations with photovoltaic power generation equipment and energy storage equipment. It fully considers the differences in the distributed source-storage status and peak-valley electricity prices between the two substations, builds an optimized and adjusted scheduling strategy with the goal of minimizing the purchase cost of electricity, and controls the timing of power flow mutual assistance and energy storage charging and discharging by controlling flexible interconnection devices to achieve the goal of saving electricity costs. The safe power supply mode means that when one of the substations loses power or fails, the load is transferred in real time through flexible interconnection devices to help the substation restore power.

2. The transformer flexible capacity expansion method according to claim 1, characterized in that: Step S1 includes: Data collection: Install smart meters or remote monitoring equipment under each distribution transformer to collect real-time historical electricity consumption data of various electrical devices, including power consumption, as well as corresponding weather, seasonal and holiday information; Data preprocessing: Use cleaning and denoising techniques on the collected historical electricity consumption data to remove abnormal data points; Establish a load forecasting model: Based on historical electricity consumption data and corresponding weather, seasonal and holiday information, build and train a machine learning model to predict future electricity load; Load forecasting: Based on the model prediction results, the total power consumption forecast value of each distribution transformer at the next moment will be generated, and this value will be used as a reference for subsequent flexible scheduling.

3. The transformer flexible capacity expansion method according to claim 2, characterized in that: The load forecasting model adopts the GBDT model. The mean absolute error is used to evaluate the model quality during GBDT model training. , then the model meets the application requirements, represents the calculated mean absolute error, is the maximum power load value in the data set.

4. The transformer flexible capacity expansion method according to claim 1, characterized in that: Step S2 includes: Data collection: Collect historical power generation data of the photovoltaic system and corresponding meteorological data. The historical power generation data includes daily power generation, instantaneous power, daily power generation peak, and average power generation efficiency. The meteorological data includes solar irradiance, temperature, sunshine duration, wind speed, humidity, and cloud cover. Data preprocessing: Cleaning and denoising techniques are used on the collected historical power generation data to remove abnormal data points, and the historical power generation data and meteorological data are aligned at the same time intervals; Establishing a photovoltaic forecasting model: Building and training a photovoltaic forecasting model based on historical power generation data and meteorological data to predict power generation at future times; Power generation forecast: Meteorological data is input into the trained photovoltaic forecast model to generate a forecast value of future photovoltaic power generation.

5. The transformer flexible capacity expansion method according to claim 4, characterized in that: The photovoltaic prediction model includes a physical power generation model based on the physical characteristics of the photovoltaic system and a data-driven machine learning model; the physical power generation model calculates the theoretical power generation of the photovoltaic module by inputting meteorological data; the machine learning model includes ARIMA, SARIMA, random forest and deep learning network.

6. A transformer flexible capacity expansion system based on the method of claim 1, characterized in that: include: Data acquisition module: responsible for collecting the operating data of the distribution transformer and its subordinate equipment and transmitting it to the edge gateway module, including the power consumption data of the load equipment, the power generation data of the photovoltaic power generation equipment, the charging and discharging status of the energy storage equipment, and meteorological data; Data preprocessing module: used to clean, denoise and detect anomalies of the collected raw data to ensure data accuracy and consistency and filter out faulty and incomplete data; Load forecasting module: Based on historical operating data, it predicts the power load of the distribution transformer in the next period, generates a load demand curve, and provides it to the subsequent management and optimization platform for analysis and decision-making; Photovoltaic power generation prediction module: Based on the historical power generation data of photovoltaic power generation equipment and weather forecast information, it predicts future power generation to ensure that photovoltaic resources can reasonably participate in load scheduling; Energy storage management module: used to manage the charging and discharging process of energy storage equipment, ensuring that the energy storage equipment can be discharged in time to relieve load pressure during peak load, and charged to reserve electricity during low load; Flexible interconnection device control module: used to achieve flexible allocation of power resources between two adjacent distribution transformers through the flexible interconnection device. When a distribution transformer is overloaded, this module will allocate power from the adjacent distribution transformer to reduce the load; Edge Gateway Module: This module performs real-time data processing and short-term forecasting on edge devices close to the transformers, and uploads the processed data to the management platform, reducing computing and communication delays and ensuring timely and accurate load regulation. Management and optimization platform: As the central command module of the system, it receives all data from the edge gateway module, performs global analysis and policy optimization, generates scheduling plans and sends them to each edge node.

7. The transformer flexible capacity expansion system according to claim 6, characterized in that: The photovoltaic power generation prediction module is implemented by establishing a photovoltaic power generation prediction model based on data drive or the physical characteristics of the photovoltaic system, and communicates with the meteorological data service interface in real time to timely obtain meteorological data including solar irradiance, temperature, sunshine duration, wind speed, humidity, and cloud cover.

8. The transformer flexible capacity expansion system according to claim 6, characterized in that: The energy storage management module dynamically adjusts the charging and discharging behavior of the energy storage equipment based on load forecasts and real-time power demand; the module is equipped with an economic optimization algorithm that optimizes charging timing and discharging strategies through price signals and load demand.

9. The transformer flexible capacity expansion system according to claim 6, characterized in that: The system uses a load forecasting module to predict the load conditions in future time periods and generate a load demand curve. At the same time, the photovoltaic power generation forecasting module predicts future photovoltaic power generation based on historical power generation data and weather forecast information. The system summarizes and analyzes the rated capacity, real-time total load, and total photovoltaic grid-connected capacity of each transformer to calculate the remaining capacity of the transformer. When the remaining capacity of the transformer is insufficient, the energy storage management module will dispatch the energy storage device to start discharging to fill the load gap. If the energy storage device cannot fully meet the demand, the flexible interconnection device control module will allocate electricity from adjacent transformers for transfer to ensure the dynamic balance of the power load. When the load is low, the system will control the energy storage device to charge, or transfer excess electricity to adjacent transformers through the flexible interconnection device.

Images