Smart meter low internal resistance lead type supercapacitor system and application method thereof

The smart meter low internal resistance lead-type supercapacitor system solves the problem of unbalanced power supply and demand by monitoring and predicting the peak and valley characteristics of power, combined with the dynamic charging and discharge of supercapacitors, and improves energy utilization efficiency and power supply stability.

CN118646004BActive Publication Date: 2024-12-03NANTONG JIANGHAI NEW ENERGY CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411112583.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-12-03
Estimated Expiration
2044-08-14

AI Technical Summary

Technical Problem

Traditional power scheduling methods are difficult to cope with the imbalance of supply and demand during peak and valley periods of electricity, resulting in waste of energy and instability of power supply. The traditional energy storage methods are inefficient and cannot fully utilize the surplus electricity during the valley period.

Method used

The smart meter low internal resistance lead type supercapacitor system is adopted. Through the smart meter, the peak and valley characteristics of the power are monitored and predicted, combined with the supercapacitor, the excess electricity is stored during the power valley period, and the electricity is released during the power peak period, thereby achieving a balance between power supply and demand. The system includes a smart meter module, a peak-to-valley characteristic prediction module, a supercapacitor module and a capacitor control module. It uses real-time electricity consumption and power supply data flow to predict the peak-to-valley characteristic of the power, and dynamically charge and discharge the supercapacitor based on an adaptive control algorithm.

Benefits of technology

It has achieved a balance between power supply and demand, improved energy utilization efficiency, ensured the stability of power supply, and solved the problem of inflexible power peak and valley scheduling.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118646004B_ABST
    Figure CN118646004B_ABST
Patent Text Reader

Abstract

The present invention discloses a low-resistance lead-type supercapacitor system for smart meters and an application method thereof, which belongs to the field of capacitor control technology, wherein the system includes: a smart meter module for obtaining a real-time power consumption data stream; a power supply side interaction module for obtaining a real-time power supply data stream; a peak-valley characteristic prediction module for obtaining auxiliary power dispatching tasks; a supercapacitor module including a plurality of supercapacitors; and a capacitor control module for controlling the supercapacitor module. The present application solves the technical problems of unbalanced power supply and demand in the power system, especially the inflexible power dispatching during power peak and valley periods, resulting in energy waste and unstable power supply, and achieves the effect of monitoring and predicting power peak and valley characteristics through smart meters, combining supercapacitors to store excess power during power valley periods and release power during power peak periods, thereby achieving a balance in power supply and demand, improving energy utilization efficiency, and ensuring the stability of power supply.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of capacitor control, and in particular to a low internal resistance lead-type supercapacitor system for a smart electric meter and an application method thereof. Background Art

[0002] With the continuous development of modern power systems, the balance of power supply and demand has become an increasingly important issue. Especially during power peak and valley periods, due to the rapid changes in power demand, traditional power dispatching methods are often not flexible enough and difficult to effectively cope with the challenges of imbalanced supply and demand. This not only leads to energy waste, but also seriously affects the stability and reliability of power supply. Traditional power dispatching methods usually rely on historical data, with low control accuracy and slow response speed, making it difficult to accurately predict and quickly respond to power peak and valley characteristics. In addition, during power valley periods, there is often a large amount of surplus electricity in the power grid, and traditional energy storage methods are often unable to fully utilize these power resources due to their low efficiency and high cost. Summary of the invention

[0003] The present application provides a low-resistance lead-type supercapacitor system for a smart meter and an application method thereof, aiming to solve the technical problems of imbalance in power supply and demand in the power system, especially the inflexible power dispatching during power peak and valley periods, which leads to energy waste and unstable power supply. The application achieves the effect of balancing power supply and demand, improving energy utilization efficiency, and ensuring the stability of power supply by monitoring and predicting power peak and valley characteristics through smart meters, combining supercapacitors to store excess power during power valley periods and release power during power peak periods.

[0004] In view of the above problems, the present application provides a low internal resistance lead-type supercapacitor system for a smart meter and an application method thereof.

[0005] The first aspect disclosed in the present application provides a smart meter low-resistance lead-type supercapacitor system, the system comprising: a smart meter module, the smart meter module is used to obtain a real-time power consumption data stream on the power consumption side of the power grid; a power supply side interaction module, the power supply side interaction module is used to obtain a real-time power supply data stream on the power supply side of the power grid; a peak-valley feature prediction module, the peak-valley feature prediction module is used to predict power peak-valley features based on the real-time power consumption data stream and the real-time power supply data stream, and obtain auxiliary power dispatching tasks, wherein the auxiliary power dispatching tasks include power valley period auxiliary dispatching tasks and power peak period auxiliary dispatching tasks; a supercapacitor module, the supercapacitor module comprises a plurality of supercapacitors, wherein the supercapacitor module is connected to the power consumption side of the power grid, the power supply side of the power grid, and the smart meter module through a low-resistance lead module; a capacitor control module, the capacitor control module is used to control the supercapacitor module based on a capacitor adaptive control algorithm in combination with the low-resistance lead module and the auxiliary power dispatching task.

[0006] Another aspect disclosed in the present application provides a method for a low-resistance lead-type supercapacitor for a smart meter, the method comprising: obtaining a real-time power consumption data stream on the power consumption side of the power grid; obtaining a real-time power supply data stream on the power supply side of the power grid; predicting power peak and valley characteristics based on the real-time power consumption data stream and the real-time power supply data stream to obtain auxiliary power dispatching tasks, wherein the auxiliary power dispatching tasks include power valley period auxiliary dispatching tasks and power peak period auxiliary dispatching tasks; based on a capacitor adaptive control algorithm, the supercapacitor module is controlled in combination with a low-resistance lead module and the auxiliary power dispatching tasks.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] By adopting the method of obtaining the real-time power consumption data stream of the power consumption side of the power grid; obtaining the real-time power supply data stream of the power supply side of the power grid; predicting the peak and valley characteristics of electricity based on the real-time power consumption data stream and the real-time power supply data stream, obtaining the auxiliary power dispatching task, wherein the auxiliary power dispatching task includes the auxiliary dispatching task of the power valley period and the auxiliary dispatching task of the power peak period; and controlling the supercapacitor module based on the capacitor adaptive control algorithm, in combination with the low internal resistance lead module and the auxiliary power dispatching task. The technical problem of the imbalance of power supply and demand in the power system, especially the inflexible power dispatching during the power peak and valley periods, resulting in energy waste and unstable power supply is solved, and the effect of monitoring and predicting the peak and valley characteristics of electricity through smart meters, combining supercapacitors to store excess electric energy during the power valley period and release electric energy during the power peak period, achieving the effect of balancing power supply and demand, improving energy utilization efficiency, and ensuring the stability of power supply is achieved.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A structural schematic diagram of a low internal resistance lead-type supercapacitor system for a smart meter is provided for the embodiment of the present application;

[0011] Figure 2 A flow chart of a low internal resistance lead-type supercapacitor method for a smart meter is provided for an embodiment of the present application.

[0012] Explanation of the accompanying drawings: smart meter module 1, power supply side interaction module 2, peak-valley characteristic prediction module 3, supercapacitor module 4, capacitor control module 5. DETAILED DESCRIPTION

[0013] The present application solves the technical problems of imbalance in power supply and demand in the power system, especially the inflexible power dispatching during power peak and valley periods, which leads to energy waste and unstable power supply, by providing a low-resistance lead-type supercapacitor system for a smart meter and an application method thereof. The application achieves the effect of balancing power supply and demand, improving energy utilization efficiency, and ensuring the stability of power supply by monitoring and predicting power peak and valley characteristics through smart meters, combining supercapacitors to store excess power during power valley periods and release power during power peak periods.

[0014] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0015] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a low internal resistance lead-type supercapacitor system for a smart meter, the system comprising:

[0016] A smart meter module 1 is used to obtain a real-time electricity consumption data stream from the electricity consumption side of the power grid.

[0017] Specifically, the smart meter module 1 is a key component in the power system, and its core function is to capture and process the power consumption data stream on the power consumption side of the power grid in real time, such as power consumption, power factor, etc. This module integrates high-precision metering sensors, stable communication interfaces, etc., and can continuously monitor and record the user's power consumption, generate real-time and accurate real-time power consumption data streams, and provide strong support for subsequent data analysis, prediction and scheduling.

[0018] The power supply side interaction module 2 is used to obtain the real-time power supply data flow of the power supply side of the power grid.

[0019] Specifically, the power supply side interaction module 2 is another crucial link in the power system. It undertakes the important task of real-time data interaction with the power supply side of the power grid. Through efficient communication interfaces and protocols, this module establishes a stable data connection with various equipment on the power supply side (such as power plants, substations, etc.), and obtains the real-time power supply data flow of the power grid in real time, such as power generation, transmission voltage, current intensity and other key parameters. These data not only reflect the real-time operating status of the power grid, but also provide an important basis for the dispatch and management of the power system.

[0020] The peak-valley characteristic prediction module 3 is used to predict the peak-valley characteristics of electricity based on the real-time electricity consumption data stream and the real-time power supply data stream to obtain auxiliary power dispatching tasks, wherein the auxiliary power dispatching tasks include power valley period auxiliary dispatching tasks and power peak period auxiliary dispatching tasks.

[0021] Specifically, in the peak-valley feature prediction module 3, the system terminal combines the obtained real-time power consumption data stream and the real-time power supply data stream with the corresponding time period to be predicted, and inputs them into the power peak feature prediction model and the power valley feature prediction model. Through the prediction of the power peak feature prediction model and the power valley feature prediction model, the system terminal can obtain two auxiliary power dispatching tasks, namely, the power peak auxiliary dispatching task and the power valley auxiliary dispatching task. During the power peak period, that is, the period when the power demand surges, the predicted power peak auxiliary dispatching task includes predicting the power peak time node and predicting the peak gap power. Through this power peak auxiliary dispatching, the system terminal performs dynamic discharge of the supercapacitor module 4 in the subsequent control process of the supercapacitor module 4 to relieve the power supply pressure and avoid overload of the power system. During the power valley period, that is, the period when the power demand is relatively low, the predicted power valley auxiliary dispatching task includes predicting the power valley time node and predicting the surplus power in the valley. Through this power valley auxiliary dispatching task, the system terminal performs dynamic charging of the supercapacitor module 4 in the subsequent control process of the supercapacitor module 4 to improve the overall energy efficiency.

[0022] Furthermore, the peak-valley characteristic prediction module 3 is used to predict the peak-valley characteristics of electricity based on the real-time electricity consumption data stream and the real-time power supply data stream to obtain an auxiliary power dispatching task, including:

[0023] Obtain the prediction feature time zone; construct the power peak feature prediction model and the power valley feature prediction model based on the ARIMA-LSTM combined model.

[0024] In a preferred embodiment, in the power system, in order to more accurately predict the peak and valley characteristics of power demand, the system terminal uses a combination model of ARIMA (autoregressive integrated moving average model) and LSTM (long short-term memory network) to construct a power peak characteristic prediction model and a power valley characteristic prediction model. This combination model combines the advantages of ARIMA in dealing with linear trends and seasonal changes, and the ability of LSTM in capturing complex nonlinear relationships in time series, so that it can more comprehensively capture the dynamic changes in power demand. First, the system terminal obtains the prediction feature time zone, which is an important step in the prediction process. This involves an in-depth analysis of historical power data to identify the periodic changes in power demand in different time periods (such as days, weeks, months, and seasons). By determining these characteristic time zones, a prediction model can be constructed more specifically to improve the accuracy of the prediction. Subsequently, based on the ARIMA-LSTM combination model, a power peak characteristic prediction model and a power valley characteristic prediction model are constructed respectively. For the power peak feature prediction model, we will focus on those time periods with historical surges in power demand, use the ARIMA model to analyze the linear trends and seasonal factors of these time periods, and use the LSTM model to capture the nonlinear patterns of power demand changes and potential impacts of emergencies during these time periods. By combining the two, the model can more accurately predict the demand characteristics of future power peaks. Similarly, the power valley feature prediction model will focus on the time periods with historically low power demand. These time periods may be affected by a variety of factors, such as weather conditions, holiday arrangements, etc. The ARIMA model will help identify the basic trends and seasonal fluctuations during these time periods, while the LSTM model can capture the subtle differences and potential laws of power demand changes during these time periods. The power peak feature prediction model and the power valley feature prediction model constructed by the ARIMA-LSTM combined model can make full use of the information in the historical power data, combine linear and nonlinear analysis methods, and realize accurate prediction of the peak and valley characteristics of future power demand. This not only helps to make power dispatch more accurate and reasonable, but also improves the safety of power system operation.

[0025] The predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream are input into the power peak characteristic prediction model to generate the power peak auxiliary scheduling task; based on the predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream, according to the power valley characteristic prediction model, the power valley auxiliary scheduling task is obtained; the power peak auxiliary scheduling task and the power valley auxiliary scheduling task are added to the auxiliary power scheduling task.

[0026] In a preferred embodiment, in the process of power dispatch optimization, the system terminal uses the predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream as key inputs, and applies them to the power peak characteristic prediction model and the power valley characteristic prediction model respectively to generate targeted auxiliary dispatch tasks. This process aims to guide the rational allocation of power resources through accurate predictions to ensure the stable operation and efficient utilization of the power grid. For the power peak characteristic prediction model, its core task is to predict the time node of the power peak and the power gap that may exist during the peak period. By comprehensively analyzing the surge trend of real-time power demand and the real-time status of the power supply side, the power peak characteristic prediction model can identify the arrival of the peak power demand in advance, estimate the amount of electricity required to be supplemented, and output these prediction results as power peak auxiliary dispatch tasks to effectively fill the power gap and ensure the reliability of peak power supply. On the other hand, the power valley characteristic prediction model focuses on predicting the time node of the power valley and the surplus power that may be generated during the valley. By analyzing the downward trend of real-time power demand and the redundant capacity of the power supply side, the power valley feature prediction model can determine the period of low power demand and estimate the surplus power resources, and output these prediction results as power valley auxiliary scheduling tasks to optimize the utilization rate of power resources. Finally, the system terminal adds the power peak auxiliary scheduling tasks and the power valley auxiliary scheduling tasks to the auxiliary power scheduling tasks, providing strong support for the subsequent control of the supercapacitor module 4.

[0027] Furthermore, based on the ARIMA-LSTM combined model, a power peak period feature prediction model and a power valley period feature prediction model are constructed, including:

[0028] The ARIMA-LSTM combined model includes an ARIMA model and an LSTM model; a power peak feature record library and a power valley feature record library are retrieved; linear feature learning is performed on the power peak feature record library according to the ARIMA model to generate an initial power peak prediction model; nonlinear feature compensation learning is performed on the initial power peak prediction model according to the LSTM model to generate the power peak feature prediction model; based on the ARIMA model and the LSTM model, feature joint learning is performed on the power valley feature record library to generate the power valley feature prediction model.

[0029] In an optional implementation, the system terminal uses the ARIMA-LSTM combined model to construct a power peak feature prediction model and a power valley feature prediction model. Among them, the ARIMA-LSTM combined model integrates the ARIMA model and the LSTM model. The ARIMA model is good at capturing linear trends and seasonal changes in time series data, while the LSTM model is good at processing long-term dependencies and nonlinear relationships in the sequence. In order to complete the training of the power peak feature prediction model and the power valley feature prediction model, the system terminal retrieves two key period record libraries from the power data system, namely the power peak feature record library and the power valley feature record library. These two record libraries contain historical data of power demand during peak hours and valley hours, respectively. Subsequently, the ARIMA model is used to perform linear feature learning on the power peak feature record library. The ARIMA model selects appropriate ARIMA model parameters (p, d, q) by analyzing statistical characteristics such as autocorrelation, difference and moving average in the power peak feature data. Among them, p represents the order of the autoregressive term, d represents the order of the difference (used to convert the non-stationary series into a stationary series), and q represents the order of the moving average term. Then use one of the methods such as maximum likelihood estimation (MLE) and least squares (OLS) to estimate the parameter values ​​of the ARIMA model. These parameters include autoregressive coefficients, moving average coefficients, etc. After that, the fitting effect of the model is evaluated by residual analysis (such as residual autocorrelation test). If the residual sequence shows significant autocorrelation, it means that the model may be insufficient and the parameters need to be further adjusted. Otherwise, the fitted ARIMA model is output as the initial power peak forecasting model. After obtaining the initial power peak forecasting model, the system terminal determines the architecture of the LSTM model according to the complexity of the problem and the characteristics of the data, including the number of LSTM layers, the number of neurons in each LSTM layer, whether to use dropout to prevent overfitting, etc. Then randomly initialize the weights and biases of the LSTM model. These parameters will be updated by the back propagation algorithm during the training process. Then, the system terminal divides the power peak characteristic data into training set, validation set and test set, and sets the hyperparameters in the training process, such as learning rate, batch size, training rounds, etc. The selection of these parameters has an important impact on the training effect and performance of the model. The LSTM model is then trained using the training set. During the training process, the predicted value is calculated through forward propagation, and the gradient is calculated and the model parameters are updated through the back propagation algorithm. This process will be repeated for multiple rounds until the preset training rounds are reached. After each round of training, the performance of the model is evaluated using the validation set. The model parameters are adjusted according to the evaluation results to improve the prediction accuracy of the model. The LSTM model can capture nonlinear characteristics and long-term dependencies in time series data through its internal memory units and gating mechanisms.During the training process, the LSTM model will learn and compensate for the nonlinear parts ignored in the initial power peak prediction model. Through the training of the LSTM model, the prediction results of the initial power peak prediction model will be optimized. The LSTM model will output a more accurate power peak prediction value containing more nonlinear features. Finally, the output of the LSTM model is fused with the output of the initial power peak prediction model to generate the final power peak feature prediction model. This fusion can be based on weighted average, stacking model and other methods. The test set is then used to evaluate the final generated power peak feature prediction model. The evaluation indicators may include mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), etc. The system terminal outputs the power peak feature prediction model that has passed the evaluation.

[0030] Similarly, the system terminal uses the same method as above to perform joint feature learning on the power valley period feature record library based on the ARIMA model and the LSTM model. The ARIMA model is responsible for capturing the linear change trend of the power load during the valley period, while the LSTM model is responsible for supplementing and correcting the nonlinear deviations in these linear predictions. Through the synergy of the two, a power valley period feature prediction model is generated.

[0031] In summary, the ARIMA-LSTM combined model combines the advantages of two different models. It not only utilizes the ability of the ARIMA model in dealing with linear trends and seasonal changes, but also takes advantage of the LSTM model's expertise in capturing complex nonlinear relationships, thereby improving the accuracy and reliability of power load forecasting.

[0032] The supercapacitor module 4 includes a plurality of supercapacitors, wherein the supercapacitor module 4 is connected to the power consumption side of the power grid, the power supply side of the power grid, and the smart meter module 1 through a low internal resistance lead module.

[0033] Specifically, the supercapacitor module 4 is a device that integrates multiple supercapacitors, which uses the characteristics of supercapacitors' fast charging and discharging, high power density, and long cycle life to store and release electrical energy. This supercapacitor module 4 is closely connected to the power consumption side, power supply side, and smart meter module 1 of the power grid through a low internal resistance lead module. This design aims to improve the flexibility, response speed, and stability of the power system.

[0034] The capacitor control module 5 is used to control the supercapacitor module 4 based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power scheduling task.

[0035] Specifically, after obtaining the auxiliary power dispatch task, the system terminal extracts the predicted time nodes of the power peak and valley periods and the corresponding power gap or surplus. Subsequently, the state information of each supercapacitor is collected in real time, including the remaining power, charging / discharging rate, temperature, etc., to construct the charging constraint domain and discharge constraint domain of the capacitor. This information is crucial to ensure that the capacitor operates within a safe and efficient range. Afterwards, based on the predicted power gap or surplus and the state constraints of the capacitor, the system terminal uses the capacitor adaptive control algorithm to make intelligent charging and discharging decisions. During the power peak period, the capacitor adaptive control algorithm will find the optimal discharge plan to supplement the power shortage of the power grid; during the power valley period, it will plan the best charging strategy to store excess power. In this process, the capacitor adaptive control algorithm will use the discharge expectation optimization and charging expectation optimization methods to screen and optimize multiple possible charging and discharging schemes to find a solution that can both meet the power grid demand and maximize the efficiency of capacitor use. In order to further improve the quality of the charging and discharging scheme, the capacitor adaptive control algorithm will introduce a fitness evaluation mechanism. By predicting and evaluating the discharge efficiency, stability, safety and other characteristics of different charging and discharging schemes, and performing weighted calculations based on weight conditions, the fitness value of each scheme is obtained. Then, the iterative optimization method is used to continuously search and compare in the initial optimization space until the winning scheme with the highest fitness is found as the final auxiliary power supply scheme or auxiliary charging control scheme. Finally, according to the predicted time node and the efficient transmission characteristics of the low internal resistance lead module, the supercapacitor module 4 is controlled in real time to assist in power supply or charging of the power grid. At the same time, the state changes of the power grid and the capacitor will be continuously monitored, and the control strategy will be adjusted according to the actual situation to ensure the stable operation and high-efficiency output of the power system.

[0036] Furthermore, the capacitor control module 5 is used to control the supercapacitor module 4 based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power dispatching task, including:

[0037] According to the power peak auxiliary scheduling task, the predicted power peak time node and the predicted peak power gap are extracted; according to the supercapacitor module 4, the discharge constraint information of each supercapacitor is collected to obtain a capacitor discharge constraint domain; based on the multiple supercapacitors, a discharge decision is made according to the capacitor discharge constraint domain to obtain multiple capacitor discharge schemes, wherein each capacitor discharge scheme includes the discharge control parameters corresponding to each supercapacitor.

[0038] In a preferred embodiment, the system terminal first extracts two key information from the power peak auxiliary scheduling task: one is the expected power peak time node, that is, when the peak of power demand will occur; the other is the predicted power gap during the peak period, that is, the difference between the actual power demand and the current supply capacity. Subsequently, in order to effectively use the supercapacitor module 4 to fill the peak power gap, the system terminal collects the discharge limit information of each supercapacitor, including the maximum discharge power that can be provided, the discharge time limit, and any other constraints that may affect its discharge capacity. Based on this information, the system terminal constructs a capacitor discharge constraint domain, which limits the range in which each capacitor can work safely and effectively when participating in the scheduling. After obtaining the capacitor discharge constraint domain, the system terminal begins to make discharge decisions based on historical experience. This decision-making process considers all available supercapacitors and generates multiple capacitor discharge schemes based on their discharge constraints. Each scheme specifies in detail how each supercapacitor should be controlled during the power peak period, including their key parameters such as discharge start time, discharge rate, and duration. These schemes are designed to maximize the use of the energy storage capacity of supercapacitors, fill the gap in the power peak period with minimal cost or risk, and ensure the stable operation of the power system.

[0039] According to the predicted peak power gap, the discharge expectation optimization is performed on the multiple capacitor discharge schemes to obtain the capacitor discharge initial optimization space; based on the discharge fitness analytical weight condition, the discharge fitness optimization is performed on the capacitor discharge initial optimization space to obtain the peak auxiliary power supply scheme; based on the predicted power peak time node and the low internal resistance lead module, the supercapacitor module 4 is controlled according to the peak auxiliary power supply scheme to provide auxiliary power supply to the power consumption side of the power grid.

[0040] In a preferred embodiment, when facing the predicted peak power gap, the system terminal develops a series of possible capacitor discharge schemes based on the discharge capacity of multiple supercapacitors. These schemes are intended to fill the power demand gap during the peak power period by discharging supercapacitors. Subsequently, in order to find the optimal discharge scheme, the system terminal performs expected optimization on these initial discharge schemes. The purpose of this step is to evaluate the potential of each scheme in filling the power gap through the peak capacitor discharge total mapping network, and compare it with the expected capacitor discharge total, narrow the selection range, and form an initial optimization space containing discharge schemes that are more likely to become the optimal solution, that is, the capacitor discharge initial optimization space. Afterwards, in order to further refine the selection, the system terminal optimizes the fitness of the schemes in this initial optimization space according to the analytical weight conditions of the discharge fitness. Discharge fitness involves multiple factors, such as discharge efficiency, cost, impact on grid stability, etc., and these factors will be assigned different weights. Through this step, the auxiliary power supply scheme with the greatest fitness is found as the peak auxiliary power supply scheme. Then, when the peak power time node arrives, the system terminal uses the low internal resistance lead module to quickly and efficiently control the supercapacitor module 4 to provide auxiliary power supply to the power consumption side of the power grid according to the determined peak auxiliary power supply plan. This can not only effectively fill the power gap, but also ensure the stable operation of the power system and meet the power needs of users.

[0041] Further, according to the predicted peak-period shortfall power, the plurality of capacitor discharge schemes are optimized for discharge expectation to obtain an initial optimization space for capacitor discharge, including:

[0042] Loading a peak gap power sample set and a capacitor discharge total amount sample set; constructing a peak capacitor discharge total amount mapping network based on the peak gap power sample set and the capacitor discharge total amount sample set; inputting the predicted peak gap power into the peak capacitor discharge total amount mapping network to obtain an expected capacitor discharge total amount.

[0043] In a feasible implementation, the system terminal loads the peak gap power sample set and the capacitor discharge total sample set from the power data system. These two sample sets contain the power gap records during the peak power period in the historical data and the total discharge records of the supercapacitors in the corresponding time period, respectively. Subsequently, the system terminal determines the number of network layers of the feedforward neural network, which usually includes an input layer, several hidden layers and an output layer. The number of neurons in the input layer should match the number of features of the peak gap power. If there is only one feature, there is one neuron; the output layer should have one neuron, representing the predicted total capacitor discharge. The peak gap power sample set and the capacitor discharge total sample set are then divided to obtain a training set, a validation set and a test set. After that, the weights and biases of the feedforward neural network are initialized with random values, and the feedforward neural network is iteratively trained using the training set data. In each iteration, forward propagation is performed to calculate the predicted value of the output layer, and compared with the corresponding sample capacitor discharge total amount, and then the loss value is calculated by the mean square error, and back propagation is performed to calculate the gradient of the loss function with respect to the network parameters. The parameters of the network are updated using the optimization algorithm SGD to minimize the loss value. During the training process, hyperparameters such as learning rate, batch size, training rounds, number of hidden layers, number of neurons, etc. are adjusted according to the performance of the validation set, and early stopping is used to prevent overfitting, that is, when the loss value of the validation set does not decrease significantly in multiple consecutive rounds, the training is stopped. Then, the performance of the feedforward neural network is evaluated using the test set data, the RMSE value and other related indicators on the test set are calculated, and the difference between the prediction results of the feedforward neural network and the total amount of sample capacitor discharge is analyzed to find possible sources of error and improvement directions, such as changing the architecture of the neural network, such as adding or reducing hidden layers, changing activation functions, etc. After the improvement is completed, and the RMSE value and other related indicators on the test set meet the expected requirements, the system terminal outputs the current feedforward neural network and constructs a peak capacitor discharge mapping network. After obtaining the peak capacitor discharge mapping network, the system terminal inputs the predicted peak gap power value into the already constructed peak capacitor discharge mapping network. The peak capacitor discharge mapping network will quickly calculate the expected capacitor discharge amount to fill the power gap based on the mapping relationship it has learned. In this way, the system terminal can more accurately predict and plan the discharge behavior of supercapacitors during power peak periods to ensure the stable operation of the power system and the reliability of power supply.

[0044] Traversing the multiple capacitor discharge schemes, extracting a first capacitor discharge scheme; predicting the total discharge amount according to the first capacitor discharge scheme to obtain a first predicted capacitor discharge amount; judging whether the first predicted capacitor discharge amount is greater than / equal to the expected capacitor discharge amount; if the first predicted capacitor discharge amount is greater than / equal to the expected capacitor discharge amount, setting the first capacitor discharge scheme as a first optimal capacitor discharge scheme, and adding the first optimal capacitor discharge scheme to the capacitor discharge initial optimization space.

[0045] In a feasible implementation, the system terminal traverses all available capacitor discharge schemes, and takes the first capacitor discharge scheme traversed as the object of current processing, that is, the first capacitor discharge scheme. Based on the selected first capacitor discharge scheme, the system terminal will input the key parameters of the first capacitor discharge scheme, such as initial voltage, capacity, discharge resistance, etc., into the discharge prediction model to predict the total discharge amount, so as to estimate the total amount of electricity that the capacitor can discharge under the scheme, that is, the first predicted capacitor discharge amount. Among them, the discharge prediction model is trained based on the capacitor discharge scheme sample set and the capacitor discharge total sample set using a method similar to the aforementioned training peak capacitor discharge total mapping network. Subsequently, the system terminal determines whether the first predicted capacitor discharge total amount is greater than or equal to the expected capacitor discharge total amount. If the first predicted capacitor discharge total amount is greater than or equal to the expected capacitor discharge total amount, it means that this scheme is feasible or at least meets the basic requirements. At this time, the system terminal will mark the first capacitor discharge scheme as the first optimal capacitor discharge scheme and add it to the capacitor discharge initial optimization space. This space is used to store all discharge schemes that meet the conditions and may be optimal for further analysis or selection.

[0046] Furthermore, based on the discharge fitness analytical weight condition, the discharge fitness optimization is performed on the initial optimization space of the capacitor discharge to obtain a peak auxiliary power supply scheme, including:

[0047] According to the initial optimization space of capacitor discharge, a Pth optimal capacitor discharge scheme is extracted, wherein P is a positive integer; according to the Pth optimal capacitor discharge scheme, feature prediction is performed to obtain a Pth discharge feature prediction result, wherein the Pth discharge feature prediction result includes a Pth predicted discharge efficiency coefficient, a Pth predicted discharge stability coefficient and a Pth predicted discharge safety factor; according to the Pth discharge feature prediction result, normalization is performed to obtain a Pth discharge feature prediction result.

[0048] In an optional embodiment, in the constructed capacitor discharge initial optimization space, the system terminal extracts the Pth capacitor discharge scheme as the current processing scheme according to the storage order. Wherein, P is a positive integer used to identify the specific scheme of current concern. Subsequently, the system terminal selects a gradient boosting tree model such as random forest regression or XGBoost as a feature prediction model. These models are good at processing complex relationships and nonlinear features, and can automatically process the interaction between features, and then extract sample optimization capacitor discharge schemes from the power data system, and feature label each sample optimization capacitor discharge scheme based on historical experience and expert advice to obtain sample discharge feature prediction results. Afterwards, the sample optimization capacitor discharge scheme and the sample discharge feature prediction results are divided to obtain a training set, a validation set, and a test set, and the basic parameters of the feature prediction model are set, such as the number of trees (n_estimators), the maximum depth of the tree (max_depth), the minimum number of samples required for splitting (min_samples_split), etc. Then, the feature prediction model is trained using the training set data. The model will build multiple decision trees, and each tree will randomly select a part of features and samples from the training set for training, and learn the mapping relationship between input features (discharge scheme parameters) and output features (discharge feature values). The prediction results of each tree will be summarized (usually taking the average) to obtain the final prediction result. Then, according to the performance on the validation set, grid search, random search or Bayesian optimization methods are used to adjust the model parameters, and the training process is repeated until the optimal model parameter combination is found. When the training reaches the maximum number of iterations or stops training midway, the system terminal test set data evaluates the trained model and calculates the prediction error to measure the accuracy and generalization ability of the model. If the expected requirements are met, the current feature prediction model is output. Otherwise, measures such as reducing the number of trees, reducing the maximum depth of the tree, and increasing the minimum number of samples required for leaf nodes are tried. After obtaining the feature prediction model, the system terminal inputs the Pth optimal capacitor discharge scheme into the feature prediction model, and the feature prediction model obtains the Pth discharge feature prediction result based on the learned mapping relationship. Among them, the Qth discharge characteristic prediction results include the Pth predicted discharge efficiency coefficient, the Pth predicted discharge stability coefficient and the Pth predicted discharge safety factor. The discharge efficiency coefficient reflects the efficiency of energy conversion of the capacitor during the discharge process. A high value indicates high energy conversion efficiency, while a low value indicates energy loss. The discharge stability coefficient evaluates the stability and controllability of the discharge process. A high value indicates that the discharge process is stable and easy to control, while a low value may indicate the presence of fluctuations or instability. The discharge safety factor considers the safety risks that the discharge process may cause to equipment, personnel or the environment. A high value indicates that the discharge process is safe and reliable, while a low value may indicate the presence of safety hazards.Since the prediction results of different features may have different dimensions and ranges, the system terminal normalizes these prediction results for the convenience of subsequent comparison and analysis. Normalization is a method of scaling data so that it falls into a small specific interval (usually [0,1] or [-1,1]). The system terminal uses the difference between the current feature coefficient and the minimum feature coefficient and the difference between the maximum feature coefficient and the minimum feature coefficient to calculate the ratio and obtain the discharge feature prediction summary result of the Pth discharge scheme. This summary result converts all feature prediction values ​​into dimensionless and comparable values, which facilitates subsequent optimization decisions.

[0049] The P-th discharge feature prediction and sorting results are weightedly calculated according to the discharge fitness analytical weight condition to generate the P-th discharge fitness; based on the capacitor discharge initial optimization space, the P+1-th optimal capacitor discharge scheme is extracted, and according to the discharge fitness analytical weight condition, the P+1-th discharge fitness is calculated; based on the P-th discharge fitness, the P+1-th discharge fitness, the P-th optimal capacitor discharge scheme and the P+1-th optimal capacitor discharge scheme, current fitness optimization is performed to obtain the current winning discharge fitness and the current winning capacitor discharge scheme; based on the current winning discharge fitness and the discharge fitness analytical weight condition, the current winning capacitor discharge scheme is iteratively optimized according to the capacitor discharge initial optimization space until the number of fitness optimizations meets the fitness optimization number threshold, and the peak auxiliary power supply scheme is generated.

[0050] In an optional embodiment, the system terminal first obtains a preset discharge fitness analysis weight condition, which includes a discharge efficiency weight, a discharge stability weight, and a discharge safety weight, which are used to evaluate the relative importance of different features in the capacitor discharge process. The system terminal uses these weights to weighted sum the Pth discharge feature prediction and sorting results to obtain the Pth discharge fitness. Subsequently, the P+1th optimal capacitor discharge scheme is extracted from the initial optimal space of capacitor discharge, and the P+1th discharge fitness is calculated through the same feature prediction, normalization sorting and weighted calculation steps mentioned above. If the P+1th discharge fitness is greater than the Pth discharge fitness, the P+1th discharge fitness is used as the current winning discharge fitness, and the P+1th optimal capacitor discharge scheme is used as the current winning capacitor discharge scheme. On the contrary, the Pth discharge fitness is used as the current winning discharge fitness, and the Pth optimal capacitor discharge scheme is used as the current winning capacitor discharge scheme. This step is to retain the discharge scheme with better performance in the continuous iteration process. Afterwards, based on the current winning discharge fitness and discharge fitness parsing weight conditions, the current winning capacitor discharge scheme is iteratively optimized for fitness. This process continuously extracts new discharge schemes from the initial optimization space of capacitor discharge and calculates their discharge fitness to find a discharge scheme with better fitness. The iteration will continue until the number of fitness optimizations reaches the preset fitness optimization threshold, or the discharge fitness is no longer significantly improved. When the number of fitness optimizations meets the fitness optimization threshold or reaches the convergence condition, the iteration is stopped and the current winning capacitor discharge scheme is output as the peak auxiliary power supply scheme. This scheme performs well in discharge efficiency, stability and safety, and is suitable for providing auxiliary power supply during the peak period of the power grid.

[0051] Furthermore, the capacitor control module 5 is used to control the supercapacitor module 4 based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power dispatching task, including:

[0052] According to the power valley auxiliary scheduling task, the predicted power valley time node and the predicted valley surplus power are extracted; according to the supercapacitor module 4, the charging constraint information of each supercapacitor is collected to obtain the capacitor charging constraint domain; based on the multiple supercapacitors, charging decisions are made according to the capacitor charging constraint domain to obtain multiple capacitor charging schemes, wherein each capacitor charging scheme includes the charging control parameters corresponding to each supercapacitor; according to the predicted valley surplus power, the multiple capacitor charging schemes are optimized according to charging expectations to obtain the capacitor charging initial optimization space; based on the charging fitness analytical weight condition, the capacitor charging initial optimization space is optimized for charging fitness to obtain the valley auxiliary charging control scheme; based on the predicted power valley time node and the low internal resistance lead module, the supercapacitor module 4 is controlled to charge according to the valley auxiliary charging control scheme.

[0053] In an optional implementation, for the power valley auxiliary scheduling task, the system terminal extracts the predicted power valley time node and the predicted valley surplus power. This information is the basis for the subsequent formulation of the supercapacitor charging strategy. Then collect the charging constraint information of each supercapacitor, such as the maximum charging rate, the minimum charging interval, the temperature limit, etc., to construct the capacitor charging constraint domain. This constraint domain defines the conditions that each supercapacitor must comply with during the charging process. Subsequently, based on multiple supercapacitors and capacitor charging constraint domains, the system makes charging decisions through algorithms and generates multiple capacitor charging schemes. Each charging scheme includes a set of charging control parameters for each supercapacitor, such as the charging start time, the charging rate, etc. After that, using a similar method as described above, the information of the predicted valley surplus power is used to optimize the charging expectation of multiple capacitor charging schemes. The purpose of this step is to select a charging scheme that can maximize the use of valley surplus power and meet the capacitor charging constraints from the initial charging scheme set, and form the initial optimization space for capacitor charging. Then, based on the charging fitness analysis weight conditions (charging efficiency, charging stability, charging cost, etc.), the initial optimization space for capacitor charging is further optimized for charging fitness. This step is carried out through similar iterative optimization as mentioned above. By constantly comparing and evaluating the adaptability of different charging schemes, the valley auxiliary charging control scheme is finally obtained. This scheme performs well in multiple dimensions and can efficiently use the surplus power in the valley period to charge the supercapacitor. Finally, the system terminal controls the supercapacitor module 4 to charge according to the valley auxiliary charging control scheme based on the predicted power valley time node and the low internal resistance lead module. This process ensures that the supercapacitor can be effectively charged during the power valley period and provide reserve energy for subsequent power supply.

[0054] In summary, the low internal resistance lead-type supercapacitor system for smart meters provided in the embodiments of the present application has the following technical effects:

[0055] The smart meter module 1 is used to obtain the real-time power consumption data stream of the power consumption side of the power grid; the power supply side interaction module 2 is used to obtain the real-time power supply data stream of the power supply side of the power grid; the peak-valley feature prediction module 3 is used to predict the peak-valley features of electricity based on the real-time power consumption data stream and the real-time power supply data stream, and obtain auxiliary power scheduling tasks, wherein the auxiliary power scheduling tasks include power valley period auxiliary scheduling tasks and power peak period auxiliary scheduling tasks; the supercapacitor module 4 includes multiple supercapacitors, wherein the supercapacitor module 4 is connected to the power consumption side of the power grid, the power supply side of the power grid, and the smart meter module 1 through a low internal resistance lead module; the capacitor control module 5 is used to control the supercapacitor module 4 based on the capacitor adaptive control algorithm, in combination with the low internal resistance lead module and the auxiliary power scheduling task. Through the above steps, the imbalance of power supply and demand in the power system, especially the inflexible power dispatching during power peak and valley periods, which leads to energy waste and unstable power supply, is solved. The smart meter is used to monitor and predict the peak and valley characteristics of power, combined with supercapacitors to store excess power during the valley period and release power during the peak period, to achieve the effect of balancing power supply and demand, improving energy utilization efficiency, and ensuring the stability of power supply.

[0056] Embodiment 2 is based on the same inventive concept as the low internal resistance lead-type supercapacitor system for smart meters in the aforementioned embodiment. Figure 2 As shown, the embodiment of the present application provides a low internal resistance lead-type supercapacitor method for a smart meter, the method comprising:

[0057] Obtain a real-time power consumption data stream on the power consumption side of the power grid; obtain a real-time power supply data stream on the power supply side of the power grid; predict power peak and valley characteristics based on the real-time power consumption data stream and the real-time power supply data stream, and obtain auxiliary power dispatching tasks, wherein the auxiliary power dispatching tasks include power valley auxiliary dispatching tasks and power peak auxiliary dispatching tasks; based on the capacitor adaptive control algorithm, the supercapacitor module 4 is controlled in combination with the low internal resistance lead module and the auxiliary power dispatching tasks.

[0058] Furthermore, the peak-valley characteristic prediction module 3 is used to predict the peak-valley characteristics of electricity based on the real-time electricity consumption data stream and the real-time power supply data stream to obtain an auxiliary power dispatching task, including:

[0059] Obtain the predicted characteristic time zone; construct a power peak characteristic prediction model and a power valley characteristic prediction model according to the ARIMA-LSTM combined model; input the predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream into the power peak characteristic prediction model to generate the power peak auxiliary scheduling task; based on the predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream, according to the power valley characteristic prediction model, obtain the power valley auxiliary scheduling task; add the power peak auxiliary scheduling task and the power valley auxiliary scheduling task to the auxiliary power scheduling task.

[0060] Furthermore, based on the ARIMA-LSTM combined model, a power peak period feature prediction model and a power valley period feature prediction model are constructed, including:

[0061] The ARIMA-LSTM combined model includes an ARIMA model and an LSTM model; a power peak feature record library and a power valley feature record library are retrieved; linear feature learning is performed on the power peak feature record library according to the ARIMA model to generate an initial power peak prediction model; nonlinear feature compensation learning is performed on the initial power peak prediction model according to the LSTM model to generate the power peak feature prediction model; based on the ARIMA model and the LSTM model, feature joint learning is performed on the power valley feature record library to generate the power valley feature prediction model.

[0062] Furthermore, the capacitor control module 5 is used to control the supercapacitor module 4 based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power dispatching task, including:

[0063] According to the power peak auxiliary scheduling task, the predicted power peak time node and the predicted peak power gap are extracted; according to the supercapacitor module 4, the discharge constraint information of each supercapacitor is collected to obtain the capacitor discharge constraint domain; based on the multiple supercapacitors, a discharge decision is made according to the capacitor discharge constraint domain to obtain multiple capacitor discharge schemes, wherein each capacitor discharge scheme includes the discharge control parameters corresponding to each supercapacitor; according to the predicted peak power gap, the multiple capacitor discharge schemes are optimized according to the discharge expectation to obtain the capacitor discharge initial optimization space; based on the discharge fitness analytical weight condition, the capacitor discharge initial optimization space is optimized according to the discharge fitness to obtain the peak auxiliary power supply scheme; based on the predicted power peak time node and the low internal resistance lead module, the supercapacitor module 4 is controlled according to the peak auxiliary power supply scheme to provide auxiliary power supply to the power consumption side of the power grid.

[0064] Further, according to the predicted peak-period shortfall power, the plurality of capacitor discharge schemes are optimized for discharge expectation to obtain an initial optimization space for capacitor discharge, including:

[0065] Load a peak gap power sample set and a capacitor discharge total sample set; construct a peak capacitor discharge total mapping network based on the peak gap power sample set and the capacitor discharge total sample set; input the predicted peak gap power into the peak capacitor discharge total mapping network to obtain an expected capacitor discharge total; traverse the multiple capacitor discharge schemes and extract the first capacitor discharge scheme; predict the discharge total according to the first capacitor discharge scheme to obtain a first predicted capacitor discharge total; determine whether the first predicted capacitor discharge total is greater than / equal to the expected capacitor discharge total; if the first predicted capacitor discharge total is greater than / equal to the expected capacitor discharge total, set the first capacitor discharge scheme as a first optimal capacitor discharge scheme, and add the first optimal capacitor discharge scheme to the capacitor discharge initial optimization space.

[0066] Furthermore, based on the discharge fitness analytical weight condition, the discharge fitness optimization is performed on the initial optimization space of the capacitor discharge to obtain a peak auxiliary power supply scheme, including:

[0067] According to the capacitor discharge initial optimization space, extract the Pth optimal capacitor discharge scheme, wherein P is a positive integer; perform feature prediction according to the Pth optimal capacitor discharge scheme to obtain a Pth discharge feature prediction result, wherein the Pth discharge feature prediction result includes a Pth predicted discharge efficiency coefficient, a Pth predicted discharge stability coefficient and a Pth predicted discharge safety factor; perform normalization and sorting according to the Pth discharge feature prediction result to obtain a Pth discharge feature prediction sorting result; perform weighted calculation on the Pth discharge feature prediction sorting result according to the discharge fitness analytical weight condition to generate a Pth discharge fitness; based on the capacitor discharge initial optimization space, extract the P+ 1. searching for an optimal capacitor discharge scheme, and calculating the P+1th discharge fitness according to the discharge fitness parsing weight condition; performing current fitness optimization based on the Pth discharge fitness, the P+1th discharge fitness, the Pth optimal capacitor discharge scheme and the P+1th optimal capacitor discharge scheme to obtain a current winning discharge fitness and a current winning capacitor discharge scheme; performing iterative fitness optimization on the current winning capacitor discharge scheme according to the capacitor discharge initial optimization space based on the current winning discharge fitness and the discharge fitness parsing weight condition, until the number of fitness optimizations meets the fitness optimization number threshold, generating the peak auxiliary power supply scheme.

[0068] Furthermore, the capacitor control module 5 is used to control the supercapacitor module 4 based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power dispatching task, including:

[0069] According to the power valley auxiliary scheduling task, the predicted power valley time node and the predicted valley surplus power are extracted; according to the supercapacitor module 4, the charging constraint information of each supercapacitor is collected to obtain the capacitor charging constraint domain; based on the multiple supercapacitors, charging decisions are made according to the capacitor charging constraint domain to obtain multiple capacitor charging schemes, wherein each capacitor charging scheme includes the charging control parameters corresponding to each supercapacitor; according to the predicted valley surplus power, the multiple capacitor charging schemes are optimized according to charging expectations to obtain the capacitor charging initial optimization space; based on the charging fitness analytical weight condition, the capacitor charging initial optimization space is optimized for charging fitness to obtain the valley auxiliary charging control scheme; based on the predicted power valley time node and the low internal resistance lead module, the supercapacitor module 4 is controlled to charge according to the valley auxiliary charging control scheme.

[0070] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0071] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to the selection of multiple elements individually or in whole. Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application intends to include these modifications and variations.

Claims

1. Low internal resistance lead type supercapacitor system for smart meter, It is characterized in that The system comprises: A smart meter module, which is used to obtain real-time power consumption data stream from the power consumption side of the power grid; A power supply side interaction module, the power supply side interaction module is used to obtain a real-time power supply data stream on the power supply side of the power grid; A peak-valley feature prediction module, the peak-valley feature prediction module is used to predict the peak-valley features of electricity based on the real-time electricity consumption data stream and the real-time power supply data stream, and obtain auxiliary power dispatching tasks, wherein the auxiliary power dispatching tasks include power valley period auxiliary dispatching tasks and power peak period auxiliary dispatching tasks; A supercapacitor module, wherein the supercapacitor module comprises a plurality of supercapacitors, wherein the supercapacitor module is connected to the power consumption side of the power grid, the power supply side of the power grid, and the smart meter module via a low internal resistance lead module; A capacitor control module, the capacitor control module is used to control the supercapacitor module based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power dispatching task; The peak-valley feature prediction module is used to predict the peak-valley features of electric power based on the real-time electric power consumption data stream and the real-time power supply data stream to obtain an auxiliary electric power dispatching task, including: Get the predicted feature time zone; According to the ARIMA-LSTM combined model, the power peak period characteristic prediction model and the power valley period characteristic prediction model are constructed; Inputting the predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream into the power peak characteristic prediction model to generate the power peak auxiliary scheduling task; Based on the predicted characteristic time zone, the real-time power consumption data stream and the real-time power supply data stream, and according to the power valley characteristic prediction model, the power valley auxiliary scheduling task is obtained; Adding the power peak period auxiliary dispatching task and the power valley period auxiliary dispatching task to the auxiliary power dispatching task; According to the ARIMA-LSTM combined model, the power peak period feature prediction model and the power valley period feature prediction model are constructed, including: The ARIMA-LSTM combined model includes an ARIMA model and an LSTM model; Retrieve the power peak period characteristic record library and the power valley period characteristic record library; Performing linear feature learning on the power peak period feature record library according to the ARIMA model to generate an initial power peak period prediction model; Performing nonlinear feature compensation learning on the initial power peak period prediction model according to the LSTM model to generate the power peak period feature prediction model; Based on the ARIMA model and the LSTM model, feature joint learning is performed on the power valley period feature record library to generate the power valley period feature prediction model; The capacitor control module is used to control the supercapacitor module based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power scheduling task, including: According to the power peak auxiliary dispatching task, extract the predicted power peak time node and the predicted peak shortage power; According to the supercapacitor module, collecting discharge constraint information of each supercapacitor to obtain a capacitor discharge constraint domain; Based on the multiple supercapacitors, a discharge decision is made according to the capacitor discharge constraint domain to obtain multiple capacitor discharge schemes, wherein each capacitor discharge scheme includes a discharge control parameter corresponding to each supercapacitor; According to the predicted peak-period shortfall power, the plurality of capacitor discharge schemes are optimized for discharge expectation to obtain an initial optimization space for capacitor discharge; Based on the discharge fitness analytical weight condition, the discharge fitness optimization is performed on the initial optimization space of the capacitor discharge to obtain a peak auxiliary power supply solution; Based on the predicted power peak time node and the low internal resistance lead module, controlling the supercapacitor module to provide auxiliary power supply to the power consumption side of the power grid according to the peak auxiliary power supply scheme; According to the predicted peak-period shortfall power, the plurality of capacitor discharge schemes are optimized for discharge expectation to obtain an initial optimization space for capacitor discharge, including: Load peak period gap power sample set and capacitor discharge total amount sample set; Constructing a peak period capacitor discharge total amount mapping network according to the peak period gap power sample set and the capacitor discharge total amount sample set; Inputting the predicted peak-period shortfall power into the peak-period capacitor discharge total amount mapping network to obtain the expected capacitor discharge total amount; Traversing the plurality of capacitor discharge schemes and extracting a first capacitor discharge scheme; Predicting the total discharge amount according to the first capacitor discharge scheme to obtain a first predicted total capacitor discharge amount; Determining whether the first predicted total capacitor discharge amount is greater than / equal to the expected total capacitor discharge amount; If the first predicted capacitor discharge total amount is greater than / equal to the expected capacitor discharge total amount, setting the first capacitor discharge scheme as a first optimal capacitor discharge scheme, and adding the first optimal capacitor discharge scheme to the capacitor discharge initial optimal space; Based on the discharge fitness analytical weight condition, the discharge fitness optimization is performed on the initial optimization space of the capacitor discharge to obtain a peak auxiliary power supply solution, including: Extracting the Pth optimal capacitor discharge scheme according to the capacitor discharge initial optimization space, where P is a positive integer; Performing feature prediction according to the Pth optimal capacitor discharge scheme to obtain a Pth discharge feature prediction result, wherein the Pth discharge feature prediction result includes a Pth predicted discharge efficiency coefficient, a Pth predicted discharge stability coefficient, and a Pth predicted discharge safety coefficient; Normalizing and arranging the P-th discharge characteristic prediction result to obtain a P-th discharge characteristic prediction and arrangement result; Performing weighted calculation on the P-th discharge feature prediction and sorting result according to the discharge fitness analysis weight condition to generate a P-th discharge fitness; Based on the capacitor discharge initial optimization space, extract the P+1th optimal capacitor discharge scheme, and calculate the P+1th discharge fitness according to the discharge fitness analytical weight condition; Perform current fitness optimization based on the Pth discharge fitness, the P+1th discharge fitness, the Pth optimal capacitor discharge scheme, and the P+1th optimal capacitor discharge scheme to obtain a current winning discharge fitness and a current winning capacitor discharge scheme; Based on the current winning discharge fitness and the discharge fitness analytical weight condition, the current winning capacitor discharge scheme is iteratively optimized for fitness according to the capacitor discharge initial optimization space until the number of fitness optimizations meets the fitness optimization number threshold, thereby generating the peak auxiliary power supply scheme.

2. The low internal resistance lead type supercapacitor system for smart electric meter as claimed in claim 1, It is characterized in that The capacitor control module is used to control the supercapacitor module based on a capacitor adaptive control algorithm in combination with the low internal resistance lead module and the auxiliary power scheduling task, including: Extracting predicted power valley time nodes and predicted valley surplus power according to the power valley auxiliary dispatching task; According to the supercapacitor module, charging constraint information of each supercapacitor is collected to obtain a capacitor charging constraint domain; Based on the multiple supercapacitors, charging decisions are made according to the capacitor charging constraint domain to obtain multiple capacitor charging schemes, wherein each capacitor charging scheme includes charging control parameters corresponding to each supercapacitor; According to the predicted valley-period surplus power, the plurality of capacitor charging schemes are optimized for charging expectation to obtain an initial optimization space for capacitor charging; Based on the charging fitness analytical weight condition, the charging fitness is optimized for the initial optimization space of the capacitor charging to obtain a valley period auxiliary charging control scheme; Based on the predicted power valley time node and the low internal resistance lead module, the supercapacitor module is controlled to charge according to the valley auxiliary charging control scheme.

3. Application method of low internal resistance lead type supercapacitor system for smart meter, It is characterized in that The method is applied to the smart meter low internal resistance lead type supercapacitor system according to any one of claims 1 to 2, and the method comprises: Obtain real-time power consumption data stream from the power consumption side of the power grid; Obtain real-time power supply data flow from the power supply side of the power grid; Based on the real-time power consumption data stream and the real-time power supply data stream, power peak and valley characteristics are predicted to obtain auxiliary power dispatching tasks, wherein the auxiliary power dispatching tasks include power valley period auxiliary dispatching tasks and power peak period auxiliary dispatching tasks; Based on the capacitor adaptive control algorithm, the supercapacitor module is controlled in combination with the low internal resistance lead module and the auxiliary power scheduling task.