A cloud computing-based energy storage device management system

Abstract

The present invention relates to the technical field of energy storage device management, and in particular to an energy storage device management system based on cloud computing. The present invention proposes the following scheme: firstly, relevant data of the area where the energy storage device is located is obtained, the relevant data is processed through a cloud platform, and the power consumption prediction result of the area is output; secondly, the regional power consumption period is determined according to the power consumption prediction result; finally, the charging and discharging work of the energy storage device is determined according to the charge state of the battery of the energy storage device and the regional power consumption period, so as to effectively improve the energy utilization rate.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage device management, and in particular to an energy storage device management system based on cloud computing. Background Art

[0002] The energy storage device management system is a system used to control and optimize the operation of energy storage devices. Traditional energy storage device management systems usually adopt local control methods, that is, data collection, processing and control are performed locally on the energy storage device. However, with the development of energy Internet, cloud computing technology has gradually been applied to energy storage device management, realizing remote monitoring and management. The cloud computing-based energy storage device management system can obtain the status information of the energy storage device in real time, and provide optimized control strategies for the energy storage device through cloud data analysis and processing, thereby improving energy utilization efficiency and system operation stability.

[0003] For example, a Chinese patent with publication number CN116826901A relates to a charge and discharge management device, which collects the working voltage and working current of the energy storage power supply through an output acquisition circuit, and then the power management circuit determines the charge and discharge state of each energy storage power supply according to each working current, determines the first energy storage power supply and the second energy storage power supply according to the working voltage and the charge and discharge state of each energy storage power supply, and controls the first energy storage power supply and the second energy storage power supply to charge and discharge respectively. In this way, when there is a voltage difference between two energy storage power supplies that is too large in each energy storage power supply, the first energy storage power supply and the second energy storage power supply are controlled separately to prevent the first energy storage power supply and the second energy storage power supply from being directly charged and discharged in parallel. When there is no voltage difference between the two energy storage power supplies that is too large in each energy storage power supply, each energy storage power supply is directly charged and discharged, and ultimately the endurance of the energy storage power supply (in discharge mode) and the charging efficiency (in charging mode) are guaranteed, while the service life of the energy storage power supply can be extended.

[0004] The above patents have the problems raised by this background technology: traditional energy storage device management systems are difficult to achieve real-time acquisition and efficient processing of large-scale data. When faced with complex and changeable power loads, the prediction accuracy is low and it is difficult to meet actual needs. The use of fixed energy management strategies cannot be dynamically adjusted according to the actual power load and battery status, resulting in low energy utilization efficiency. In order to solve the above problems, this application designs a cloud computing-based energy storage device management system. Summary of the invention

[0005] The technical problem to be solved by the present invention is to address the deficiencies in the prior art and provide a cloud computing-based energy storage device management system, which first obtains relevant data of the area where the energy storage device is located, processes the relevant data through a cloud platform, and outputs the regional power consumption forecast results, then determines the regional power consumption period based on the power consumption forecast results, and finally determines the charging and discharging work of the energy storage device based on the charge state of the energy storage device battery and the regional power consumption period.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A cloud computing-based energy storage device management system, the energy storage device management system comprising a data acquisition module, a cloud platform processing module, a time period allocation module, and an energy management module;

[0008] The data acquisition module is used to acquire relevant data of the area where the energy storage device is located;

[0009] The cloud platform processing module is configured with a data preprocessing strategy and a power load forecasting strategy, and the power load forecasting strategy analyzes the preprocessed relevant data and outputs the power consumption forecast result of the area;

[0010] The time period allocation module is used to determine the regional power consumption period according to the regional power consumption prediction result, wherein the regional power consumption period includes a peak period, a valley period and a normal period;

[0011] The energy management module is configured with a charge and discharge scheduling strategy, which is used to determine the charging and discharging work of the energy storage device according to the charge state of the battery of the energy storage device and the regional power consumption period.

[0012] The cloud platform processing module is used to receive, store and analyze information from the data acquisition module in the cloud. The cloud platform processing module includes a data transmission unit, a data diversion unit, a data processing unit and a load prediction unit. The data preprocessing strategy is configured in the data processing unit and is used to preprocess relevant data of the area where the energy storage device is located. The preprocessing includes data cleaning, data format conversion and data normalization. The power load prediction strategy is configured in the load prediction unit and includes feature extraction logic and power consumption prediction logic.

[0013] The specific steps of the feature extraction logic are as follows:

[0014] S1.1: The relevant data of the area where the energy storage device is located includes the load value sequence of the current day and the load value sequence of the reference historical day. The load value of the current day is updated according to the load value sequence of the current day and the load value sequence of the reference historical day to obtain the load characteristics of the current day. The calculation formula for the load value update is:

[0015] ;

[0016] in, Indicates the updated load value of the current day. represents the proportional weight, Indicates the nth load value in the current day's load value sequence. represents the load variation coefficient, It refers to the average value of the nth load value in the historical daily load value sequence. The calculation formula of the load change coefficient is:

[0017] ,

[0018] Where i represents the i-th reference time in the current daily load value sequence, N represents the total number of reference times in the current daily load value sequence, Indicates the maximum load value in the current day's load value sequence. Indicates the minimum load value in the current day's load value sequence. Indicates the average load value in the current day's load value sequence;

[0019] S1.2: The relevant data of the area where the energy storage device is located also includes the weather data of the current day and the weather data of the historical days. A similar day group is obtained based on the weather data of the current day and the weather data of the historical days. The load characteristics of the similar days are calculated by the weighted average method based on the similar day group.

[0020] S1.3: Determine the date of the forecast day according to the date of the current day, and obtain the date load characteristics according to the date of the forecast day.

[0021] The specific steps of the power consumption prediction logic are as follows:

[0022] S2.1: Establishing a characteristic sequence for the current day load characteristics, similar day load characteristics and date load characteristics, and calculating the fitting value of the characteristic sequence;

[0023] S2.2: learning the fitting value of the feature sequence through a bidirectional long short-term memory neural network to obtain time series features, wherein the bidirectional long short-term memory network is composed of a first long short-term memory network layer and a second long short-term memory network layer, wherein the first long short-term memory network layer learns the features of the feature sequence from the beginning of the feature sequence, and the second long short-term memory network layer learns the features of the feature sequence from the end of the feature sequence;

[0024] S2.3: Constructing a power consumption prediction network, taking the time series features as input parameters of the power consumption prediction network, training the input parameters through the power consumption prediction network, and outputting power consumption prediction results for the region.

[0025] The power consumption prediction network comprises an input layer, a hidden layer and an output layer;

[0026] The input layer is used to learn through a self-attention mechanism based on the time series characteristics as input parameters to obtain the load time series characteristics;

[0027] The hidden layer is used to activate the load time series characteristics, calculate the weighted prediction value through the initial weight, and output the internal training result through the node threshold;

[0028] The output layer is used to convert the internal training results output by the hidden layer into a load forecast value, calculate the average error of the load forecast value through the loss function, and determine whether the average error meets the convergence requirement of the output layer. If the convergence requirement is not met, the initial weight and the node threshold are updated. If the convergence requirement is met, the load forecast value is output;

[0029] The calculation formula of the load forecast value is:

[0030] ,

[0031] in, represents the load forecast value, represents the activation function, represents the initial weight, j represents a single sequence number of the time series feature, f represents a single sub-feature in the j-th time series feature, M represents the total number of sub-features of the j-th time series feature, Score(•) represents the correlation score calculation function in the self-attention mechanism, represents the f-th sub-feature in the j-th time series feature, b represents the node threshold, and V(•) represents the value vector.

[0032] The energy management module is used for energy management of charging and discharging the battery of the energy storage device, including a battery SOC detection unit and an energy scheduling unit. The charge and discharge scheduling strategy includes SOC detection logic and energy scheduling logic. The SOC detection logic is configured in the battery SOC detection unit, and the energy scheduling logic is configured in the energy scheduling unit.

[0033] The specific steps of the SOC detection logic are as follows:

[0034] S3.1: Acquire the battery operation signal of the energy storage device and convert it into an operation state characteristic parameter, wherein the conversion includes Fourier transform and normalization;

[0035] S3.2: Optimize the boundary parameters of the operating state characteristic parameters according to the simulated annealing algorithm to determine the boundary of the characteristic parameters;

[0036] S3.3: Perform curve fitting based on the characteristic parameter boundary and the operating state characteristic parameter, convert the operating state characteristic parameter into a capacity signal through wavelet transform, translate the signal, change the position of the discontinuity point, display it in the curve, and extract the operating characteristic variable;

[0037] S3.4: Calculate the battery state of charge (SOC) based on the operating characteristic variables and the battery rated capacity. The calculation formula for the battery state of charge (SOC) is:

[0038] ,

[0039] Among them, k represents the battery power consumption decay coefficient, S represents the battery rated capacity, Indicates the peak-to-peak operating coefficient of the characteristic signal of the battery operating status, Indicates the gradient operation coefficient of the characteristic signal of the battery operation status, Represents the peak-to-peak value of the normal component of the battery operation signal, Represents the peak-to-peak value of the tangential component signal of the battery operation signal, represents the signal gradient of the tangential component of the battery operation signal, Represents the normal component signal gradient of the battery operation signal.

[0040] The specific steps of energy scheduling logic configuration are as follows:

[0041] S4.1: Obtain the regional power consumption period and battery state of charge SOC of this dispatch;

[0042] S4.2: Determine whether the current time is in the valley period or the normal period. If it is in the valley period or the normal period, determine whether the battery state of charge SOC meets the requirements. If it is not satisfied, the energy storage device is charged and discharged by optimizing sub-logic A. If it is satisfied, the energy storage device is charged and discharged by optimizing sub-logic B. Indicates the maximum charge capacity of the battery under normal operation;

[0043] S4.3: Determine whether the current time is during the peak period. If it is during the peak period, determine whether the battery state of charge SOC meets the requirements. If it is not satisfied, the energy storage device is charged and discharged through the optimization sub-logic C. If it is satisfied, the energy storage device is charged and discharged through the optimization sub-logic B, where Indicates the minimum charge capacity of the battery under normal operation.

[0044] The optimization sub-logic A refers to the energy storage device performing a discharge operation, which is used to optimize the discharge power of the power battery of the energy storage device under the constraints of the system operation, so as to minimize the total operation cost;

[0045] The optimization sub-logic B means that the energy storage device performs charging and discharging operations according to its own state of charge, and optimizes the discharge power and charging power of the power battery of the energy storage device under the constraints of system operation to minimize the total operating cost;

[0046] The optimization sub-logic C refers to the energy storage device performing charging operation, optimizing the charging power of the power battery of the energy storage device under the constraints of system operation to minimize the total operating cost;

[0047] The constraints include cost constraints, power capacity constraints, battery charging and discharging active power constraints, battery charging and discharging reactive power constraints, energy storage device operating time constraints and interactive capacity constraints.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. The present invention combines the data acquisition module with the cloud platform processing module to achieve real-time collection, storage and high-speed processing of large amounts of data. The powerful computing resources in the cloud can make more accurate predictions of regional electricity demand and better learn and predict complex and changeable electricity consumption patterns;

[0050] 2. The present invention intelligently divides peak, valley and normal periods according to the electricity consumption forecast results, and formulates corresponding charging and discharging strategies accordingly, which is helpful for load balancing and stable power supply of the power grid. Taking into account the battery charge state and regional power consumption period, the charging and discharging of the energy storage device is arranged in an intelligent manner, thereby improving energy utilization efficiency and reducing the operating cost of the energy storage device. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0052] Figure 1 This is a module diagram of a cloud computing-based energy storage device management system according to Embodiment 1 of the present invention;

[0053] Figure 2 This is a schematic diagram of load value update in Embodiment 1 of the present invention;

[0054] Figure 3 This is a schematic diagram of the power consumption prediction network principle in Example 1 of the present invention. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0056] Example 1

[0057] See also Figure 1 , an embodiment provided by the present invention: an energy storage device management system based on cloud computing, the energy storage device management system comprises a data acquisition module, a cloud platform processing module, a time period allocation module and an energy management module;

[0058] The data acquisition module is used to acquire relevant data of the area where the energy storage device is located;

[0059] The cloud platform processing module is configured with a data preprocessing strategy and a power load forecasting strategy, and the power load forecasting strategy analyzes the preprocessed relevant data and outputs the power consumption forecast result of the area;

[0060] The time period allocation module is used to determine the regional power consumption period according to the regional power consumption prediction result, wherein the regional power consumption period includes a peak period, a valley period and a normal period;

[0061] The energy management module is configured with a charge and discharge scheduling strategy, which is used to determine the charge and discharge work of the energy storage device according to the charge state of the battery of the energy storage device and the regional power consumption period;

[0062] The cloud platform processing module is used to receive, store and analyze information from the data acquisition module in the cloud. The cloud platform processing module includes a data transmission unit, a data diversion unit, a data processing unit and a load prediction unit. The data preprocessing strategy is configured in the data processing unit and is used to preprocess the relevant data of the area where the energy storage device is located. The preprocessing includes data cleaning, data format conversion and data normalization. The power load prediction strategy is configured in the load prediction unit and includes feature extraction logic and power consumption prediction logic.

[0063] The data transmission unit is used to transmit relevant data of the area where the energy storage device is located to the cloud platform based on the raw data collected from various data sources, wherein the data sources include Internet of Things devices, user inputs and sensors;

[0064] The data distribution unit is used to receive relevant data from the data transmission unit, archive and sort the relevant data according to the area number of the energy storage device, and store them in the cloud corresponding to the area number;

[0065] The data processing unit is used to perform a pre-read operation on the relevant data to be processed, and the pre-reading includes loading the data into the memory or cache in advance to improve the reading speed and optimize the system performance, and processing the pre-read data, including data cleaning, data format conversion and data normalization to ensure the quality and consistency of the data, and also includes structured processing of the data, converting unstructured data into structured data, to facilitate subsequent analysis and application;

[0066] The specific steps of the feature extraction logic are as follows:

[0067] S1.1: The relevant data of the area where the energy storage device is located includes the load value sequence of the current day and the load value sequence of the reference historical day. The load value of the current day is updated according to the load value sequence of the current day and the load value sequence of the reference historical day, and the load characteristics of the current day are obtained. The load value is updated by the load values ​​of the previous few moments of the current day and the load average value of the previous few days. The load value update includes four parameters: the load value to be updated on the current day, the average value of the load value of the reference historical day, the load change coefficient and the proportional weight. The load change coefficient is used to characterize the degree of change of the load in the power system over time. The calculation formula for the load value update is:

[0068] ;

[0069] in, Indicates the updated load value of the current day. represents the proportional weight, Indicates the nth load value in the current day's load value sequence. represents the load variation coefficient, It refers to the average value of the nth load value in the historical daily load value sequence. The calculation formula of the load change coefficient is:

[0070] ,

[0071] Where i represents the i-th reference time in the current daily load value sequence, N represents the total number of reference times in the current daily load value sequence, Indicates the number of the current day's load value sequence A load value, Indicates the number of reference historical daily load values ​​in the sequence The average value of the load value, Indicates the maximum load value in the current day's load value sequence. Indicates the minimum load value in the current day's load value sequence. It represents the average load value in the current day load value sequence. The specific day number, proportional weight and reference time of the reference historical day can be calculated by those skilled in the art through a large number of experiments;

[0072] See also Figure 2 , a schematic diagram of load value updating in an embodiment of the present invention, using data from a commercial building area, updating the fourth load value in the load value sequence to be updated, using the load value data of the previous five days as a reference day, and using the previous two moments as reference moments;

[0073] S1.2: Divide the weather data of historical days into groups of individual days in date order, calculate the similarity between the weather data of historical days and the weather data of the current day in turn, update the groups in descending order according to the similarity, extract the top five historical days in the group as similar day groups, and calculate the load characteristics of similar days by weighted average method based on the similar day groups, wherein the weather data includes temperature, wind speed and relative humidity, and the historical day refers to the day one year before the current day;

[0074] In actual situations, the closer the reference day is to the current day, the higher the weighting factor will be, so that the prediction accuracy will be higher when the dates are consecutive. However, due to the uncertainty of climate, the weather conditions may change drastically, resulting in load fluctuations, which is not conducive to the prediction of the algorithm. Therefore, the similar day selection method is used to select historical days with weather data similar to the current day.

[0075] The similar day method divides the weather data of historical days into groups of individual days according to the similarity of the data, and arranges them in order according to the similarity with the weather data of the current day. The core idea of ​​the similar day method is: first, select the current weather, and then calculate the similarity of historical weather according to the current weather. Each time the similarity is calculated, the arrangement of similar days is updated, and the calculation and update steps are repeated until the arrangement of similar days is completed. The weather similarity is measured using the Euclidean distance, and the training set of the model is optimized: the data position is adjusted according to the similarity between the data of the current day and several reference days. The reference day with the greater similarity to the current weather is closer to the current day, thereby obtaining a similar day group;

[0076] S1.3: determining the date of the forecast day according to the date of the current day, determining the date characteristics of the forecast day according to the date of the forecast day, the date characteristics including working days, rest days and holidays, and obtaining the date load characteristics according to the date characteristics of the forecast day;

[0077] First, the date is converted through the datatime function to identify the year, month and day of the current day, so as to determine the year, month and day of the predicted day. The year, month and day of the predicted day are used to determine whether it is a holiday. If it is a holiday, the date feature of the predicted day is marked as a holiday. If it is not a holiday, the specific day of the week corresponding to the specific date is determined. If it is Monday to Friday, it is marked as a working day. If it is Saturday and Sunday, it is marked as a rest day. Different types of dates are often associated with different electricity consumption patterns. The load characteristics of working days are usually manifested as obvious peaks in the morning and evening, which are related to the increase in commercial and office electricity consumption. The load characteristics of rest days are usually more dispersed, and the peak is not as obvious as on working days, which is related to the increase in residential electricity consumption. The load characteristics of holidays are usually manifested as an overall decrease in load, which is related to the specific days of the holiday. The load characteristics of the date are obtained by calculating the load mean of the same date in historical data;

[0078] The core code of the datatime function to convert the date is as follows:

[0079] %Call time function;

[0080] from datetime import datetime

[0081] %Create a time object and determine the current date through the now function;

[0082] date_object = datetime.now();

[0083] %Calculate and obtain the year, month and day of the current time object;

[0084] date_string = date_object.strftime("%Y-%m-%d");

[0085] %Year, month, and day conversion;

[0086] date_object = datetime.strptime(date_string, "%Y-%m-%d");

[0087] The specific steps of the power consumption prediction logic are as follows:

[0088] S2.1: Establish a characteristic sequence for the current day load characteristics, similar day load characteristics and date load characteristics, accumulate the characteristic sequences, establish a grey differential equation, calculate the least squares parameter of the grey differential equation, and calculate the fitting value of the characteristic sequence by discretizing the least squares parameter solution;

[0089] S2.2: learning the fitting value of the feature sequence through a bidirectional long short-term memory neural network to obtain time series features, wherein the bidirectional long short-term memory network is composed of a first long short-term memory network layer and a second long short-term memory network layer, wherein the first long short-term memory network layer learns the features of the feature sequence from the beginning of the feature sequence, and the second long short-term memory network layer learns the features of the feature sequence from the end of the feature sequence;

[0090] S2.3: constructing a power consumption prediction network, taking the time series features as input parameters of the power consumption prediction network, training the input parameters through the power consumption prediction network, and outputting power consumption prediction results for the region;

[0091] The embodiment of the present invention extracts the time series characteristics of the feature sequence through a bidirectional long short-term memory network. The bidirectional long short-term memory network is an improved long short-term memory network. Compared with the unidirectional structure of the long short-term memory network, the bidirectional long short-term memory network is composed of a first long short-term memory network layer and a second long short-term memory network layer, and can simultaneously consider forward and reverse information transmission. Such a design helps to more accurately capture the time series characteristics of the feature sequence, especially the long-term dependencies in the historical load data. By using a bidirectional long short-term memory network, the modeling ability of the sequence data is improved, thereby providing more reliable results for load forecasting. At the beginning of each time series, the activation function of the input gate and the forgetting gate is calculated, that is, the activation function of the input gate and the output gate at the current moment is calculated according to the feature sequence and time data input at the current moment and the hidden layer output at the previous moment, and then the cell state of the memory unit is updated from the previous moment to the latest moment, and the new load data and time data are stored after nonlinear transformation. Finally, the activation function and output vector of the output gate are calculated to enter the next time series;

[0092] See also Figure 3 , a diagram of a power consumption prediction network structure according to an embodiment of the present invention, wherein the power consumption prediction network comprises an input layer, a hidden layer and an output layer;

[0093] The input layer is used to learn through a self-attention mechanism using the time series feature as an input parameter to obtain the load time series feature, wherein the load time series feature is a time series feature that is not affected by the external environment;

[0094] The self-attention mechanism is a mechanism for processing sequence data, which is composed of a query vector, a key vector and a value vector. This mechanism can dynamically allocate attention weights according to the information provided by the time series features to more effectively capture important information in the power load and reduce the interference of external factors on the predicted value. The query vector is used to guide the self-attention mechanism to pay attention to the time series features of the data, the key vector is used to measure the similarity between each position in the time series features and the query vector, and the value vector contains the actual time series feature information;

[0095] The hidden layer is used to activate the load time series characteristics, calculate the weighted prediction value through the initial weight, and output the internal training result through the node threshold;

[0096] The output layer is used to convert the internal training results output by the hidden layer into a load forecast value, calculate the average error of the load forecast value through the loss function, and determine whether the average error meets the convergence requirement of the output layer. If the convergence requirement is not met, the initial weight and the node threshold are updated. If the convergence requirement is met, the load forecast value is output;

[0097] The calculation formula of the load forecast value is:

[0098] ,

[0099] in, represents the load forecast value, represents the activation function, represents the initial weight, j represents a single sequence number of the time series feature, f represents a single sub-feature in the j-th time series feature, M represents the total number of sub-features of the j-th time series feature, Score(•) represents the correlation score calculation function in the self-attention mechanism, represents the fth sub-feature in the jth time series feature, b represents the node threshold, and V(•) represents the value vector;

[0100] The time period allocation module is used to determine the regional power consumption period according to the regional power consumption prediction result, wherein the regional power consumption period includes a peak period, a valley period and a normal period;

[0101] The main purpose of optimizing and adjusting the charging and discharging period module of the energy storage device according to the power consumption forecast results of different regions is to balance the load of the power grid, reduce electricity costs, and improve energy efficiency through intelligent scheduling. The peak period is usually the period with the highest power consumption and the most expensive electricity price. During these periods, the power demand is close to or exceeds the power supply capacity of the power grid, and the energy storage device needs to be discharged. The valley period is the period with low power consumption and low electricity price. During these periods, encouraging the use of electricity or energy storage can reduce the overall electricity bill and balance the load of the power grid. The power consumption and electricity price of the normal period are between the peak and valley. During these periods, the main goal is to maintain the stability and reliability of power consumption;

[0102] The energy management module is used to manage the energy of charging and discharging the batteries of the energy storage device. When the grid load is too high or the renewable energy output is excessive, the energy storage device may be passively charged to absorb excess electricity. During grid failure or peak demand periods, the energy storage device can be quickly discharged to provide necessary power support to ensure the stable operation of the grid.

[0103] The energy management module includes a battery SOC detection unit and an energy scheduling unit, wherein the battery SOC detection unit is used to detect the remaining energy of the battery, and the energy scheduling unit is used to perform charge and discharge scheduling according to the battery SOC detection result, and the charge and discharge scheduling strategy includes SOC detection logic and energy scheduling logic, wherein the SOC detection logic is configured in the battery SOC detection unit, and the energy scheduling logic is configured in the energy scheduling unit;

[0104] The specific steps of the SOC detection logic are as follows:

[0105] S3.1: Acquire the battery operation signal of the energy storage device and convert it into an operation state characteristic parameter, wherein the conversion includes Fourier transform and normalization;

[0106] S3.2: Optimize the boundary parameters of the operating state characteristic parameters according to the simulated annealing algorithm to determine the boundary of the characteristic parameters;

[0107] S3.3: Perform curve fitting according to the characteristic parameter boundary and the operating state characteristic parameter, convert the operating state characteristic parameter into a capacity signal through wavelet transform, translate the signal, change the position of the discontinuity point, display it in the curve, and extract the operating characteristic variable, wherein the operating state characteristic parameter includes the peak-to-peak value of the normal component signal of the battery operating signal, the peak-to-peak value of the tangential component signal of the battery operating signal, the gradient of the tangential component signal of the battery operating signal, and the gradient of the normal component signal of the battery operating signal;

[0108] S3.4: Calculate the battery SOC based on the operating characteristic variables and the battery rated capacity. The calculation formula for the battery SOC is:

[0109] ,

[0110] Among them, k represents the battery power consumption decay coefficient, S represents the battery rated capacity, Indicates the peak-to-peak operating coefficient of the characteristic signal of the battery operating status, Indicates the gradient operation coefficient of the characteristic signal of the battery operation status, Represents the peak-to-peak value of the normal component of the battery operation signal, Represents the peak-to-peak value of the tangential component signal of the battery operation signal, represents the signal gradient of the tangential component of the battery operation signal, Represents the signal gradient of the normal component of the battery operation signal;

[0111] The specific steps of energy scheduling logic configuration are as follows:

[0112] S4.1: Obtain the time period of this scheduling and the battery state of charge SOC;

[0113] S4.2: Determine whether the current time is in the valley period or the normal period. If it is in the valley period or the normal period, determine whether the battery state of charge SOC meets the requirements. If it is not satisfied, the energy storage device is charged and discharged by optimizing sub-logic A. If it is satisfied, the energy storage device is charged and discharged by optimizing sub-logic B. Indicates the maximum charge capacity of the battery under normal operation;

[0114] S4.3: Determine whether the current time is during the peak period. If it is during the peak period, determine whether the battery state of charge SOC meets the requirements. If it is not satisfied, the energy storage device is charged and discharged through the optimization sub-logic C. If it is satisfied, the energy storage device is charged and discharged through the optimization sub-logic B, where Indicates the minimum charge capacity of the battery under normal operation;

[0115] The optimization sub-logic A refers to the energy storage device performing a discharge operation, which is used to optimize the discharge power of the power battery of the energy storage device under the constraints of the system operation, so as to minimize the total operation cost;

[0116] The optimization sub-logic B means that the energy storage device performs charging and discharging operations according to its own state of charge, and optimizes the discharge power and charging power of the power battery of the energy storage device under the constraints of system operation to minimize the total operating cost;

[0117] The optimization sub-logic C refers to the energy storage device performing charging operation, optimizing the charging power of the power battery of the energy storage device under the constraints of system operation to minimize the total operating cost;

[0118] The constraints include cost constraints, power capacity constraints, battery charging and discharging active power constraints, battery charging and discharging reactive power constraints, energy storage device operating time constraints and interactive capacity constraints;

[0119] In the energy management system, optimization sub-logics A, B, and C are specifically designed to manage the charging and discharging operations of energy storage devices to optimize power output, reduce costs, and ensure stable operation of the system. These logics take into account multiple constraints to ensure the effectiveness and safety of the operation. The following is a detailed analysis of these three optimization sub-logics:

[0120] The optimization sub-logic A is used for discharging operation optimization. The optimization sub-logic A focuses on the discharging operation of the energy storage device. Its goal is to optimize the discharge power of the power battery of the energy storage device under various constraints of system operation to minimize the total operating cost. When performing the discharging operation, the logic considers the current power of the battery, the maximum and minimum discharge power, the health status of the battery, and the demand constraints of the power grid.

[0121] The total operating cost includes the market price of electricity, the cost of battery aging and the marginal cost of system operation. Through intelligent algorithms, Logic A calculates when and where it is most economical to discharge the battery while ensuring that the safe discharge rate and depth of the battery are not exceeded;

[0122] Optimization sub-logic B is used for charging and discharging operation optimization. Optimization sub-logic B considers charging and discharging operations at the same time. Under the conditions of meeting all relevant constraints, it optimizes the charging and discharging power of the energy storage device to reduce the total operating cost, including the battery's state of charge, the limit of charging and discharging power, the battery's cycle life and the system's load requirements. In addition, the frequency and voltage stability requirements of the power grid also need to be considered.

[0123] Logic B uses real-time data analysis to determine changes in electricity prices and demand, thereby determining the best charging and discharging strategy, for example, charging when electricity prices are low and discharging when electricity prices are high, while avoiding energy loss and battery depletion caused by frequent switching;

[0124] The optimization sub-logic C is used for charging operation optimization. The optimization sub-logic C focuses on the charging operation of the energy storage device. Its purpose is to optimize the charging power of the power battery of the energy storage device without violating the system operation constraints to reduce the total operating cost. When executing this logic, the maximum charging capacity, charging rate, battery temperature and grid load during charging are considered. Logic C monitors the power supply status and electricity price dynamics of the grid and selects the time window with the lowest cost for charging, while ensuring that the charging process does not cause excessive burden on the grid.

[0125] The cost constraint considers the cost of electric energy, including the market electricity price and the cost of using the battery;

[0126] The power capacity constraint means that the power output of the energy storage device cannot exceed its designed maximum power to ensure the stability and safety of the equipment;

[0127] The battery charging and discharging active / reactive power constraints must ensure that the battery operates within a safe power range to avoid overcharging and discharging;

[0128] The energy storage device operation time constraint, the operation time of the energy storage device is limited to ensure the long-term stability of the equipment;

[0129] Due to interaction capacity constraints, energy exchange between energy storage devices and the grid is subject to physical and contractual limitations.

[0130] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A cloud computing-based energy storage device management system, characterized in that: The energy storage device management system includes a data acquisition module, a cloud platform processing module, a time period allocation module, and an energy management module; The data acquisition module is used to acquire relevant data of the area where the energy storage device is located; The cloud platform processing module is configured with a data preprocessing strategy and a power load forecasting strategy, and the power load forecasting strategy analyzes the preprocessed relevant data and outputs the power consumption forecast result of the area; The time period allocation module is used to determine the regional power consumption period according to the regional power consumption prediction result, wherein the regional power consumption period includes a peak period, a valley period and a normal period; The energy management module is configured with a charge and discharge scheduling strategy, which is used to determine the charging and discharging work of the energy storage device according to the charge state of the battery of the energy storage device and the regional power consumption period; The energy management module is used for energy management of charging and discharging the battery of the energy storage device, including a battery SOC detection unit and an energy scheduling unit. The charge and discharge scheduling strategy includes SOC detection logic and energy scheduling logic. The SOC detection logic is used to detect the SOC state of the battery according to the battery operation signal of the energy storage device to obtain the battery state of charge SOC. The SOC detection logic is configured in the battery SOC detection unit, and the energy scheduling logic is configured in the energy scheduling unit. The calculation formula for battery state of charge SOC is: ， Among them, k represents the battery power consumption decay coefficient, S represents the battery rated capacity, Indicates the peak-to-peak operating coefficient of the characteristic signal of the battery operating status, Indicates the gradient operation coefficient of the characteristic signal of the battery operation status, Represents the peak-to-peak value of the normal component of the battery operation signal, Represents the peak-to-peak value of the tangential component signal of the battery operation signal, represents the signal gradient of the tangential component of the battery operation signal, Represents the normal component signal gradient of the battery operation signal.

2. The cloud computing-based energy storage device management system according to claim 1, characterized in that: The cloud platform processing module is used to receive, store and analyze information from the data acquisition module in the cloud. The cloud platform processing module includes a data transmission unit, a data diversion unit, a data processing unit and a load prediction unit. The data preprocessing strategy is configured in the data processing unit and is used to preprocess relevant data of the area where the energy storage device is located. The preprocessing includes data cleaning, data format conversion and data normalization. The power load prediction strategy is configured in the load prediction unit and includes feature extraction logic and power consumption prediction logic.

3. The cloud computing-based energy storage device management system according to claim 2, characterized in that: The specific steps of the feature extraction logic are as follows: S1.1: The relevant data of the area where the energy storage device is located includes the load value sequence of the current day and the load value sequence of the reference historical day. The load value of the current day is updated according to the load value sequence of the current day and the load value sequence of the reference historical day to obtain the load characteristics of the current day. The calculation formula for the load value update is: ； in, Indicates the updated load value of the current day. represents the proportional weight, Indicates the nth load value in the current day's load value sequence. represents the load variation coefficient, It represents the average value of the nth load value in the reference historical daily load value sequence; S1.2: The relevant data of the area where the energy storage device is located also includes the weather data of the current day and the weather data of the historical days. A similar day group is obtained based on the weather data of the current day and the weather data of the historical days. The load characteristics of the similar days are calculated by the weighted average method based on the similar day group. S1.3: Determine the date of the forecast day according to the date of the current day, and obtain the date load characteristics according to the date of the forecast day.

4. The cloud computing-based energy storage device management system according to claim 3, characterized in that: The specific steps of the power consumption prediction logic are as follows: S2.1: Establishing a characteristic sequence for the current day load characteristics, similar day load characteristics and date load characteristics, and calculating the fitting value of the characteristic sequence; S2.2: learning the fitting value of the feature sequence through a bidirectional long short-term memory neural network to obtain time series features, wherein the bidirectional long short-term memory network is composed of a first long short-term memory network layer and a second long short-term memory network layer, wherein the first long short-term memory network layer learns the features of the feature sequence from the beginning of the feature sequence, and the second long short-term memory network layer learns the features of the feature sequence from the end of the feature sequence; S2.3: Constructing a power consumption prediction network, taking the time series features as input parameters of the power consumption prediction network, training the input parameters through the power consumption prediction network, and outputting power consumption prediction results for the region.

5. The cloud computing-based energy storage device management system according to claim 4, characterized in that: The power consumption prediction network includes an input layer, a hidden layer and an output layer; The input layer is used to learn through a self-attention mechanism based on the time series characteristics as input parameters to obtain the load time series characteristics; The hidden layer is used to activate the load time series characteristics, calculate the weighted prediction value through the initial weight, and output the internal training result through the node threshold; The output layer is used to convert the internal training results output by the hidden layer into a load forecast value, calculate the average error of the load forecast value through the loss function, and determine whether the average error meets the convergence requirement of the output layer. If the convergence requirement is not met, the initial weight and the node threshold are updated. If the convergence requirement is met, the load forecast value is output; The calculation formula of the load forecast value is: ， in, represents the load forecast value, represents the activation function, represents the initial weight, j represents a single sequence number of the time series feature, f represents a single sub-feature in the j-th time series feature, M represents the total number of sub-features of the j-th time series feature, Score(•) represents the correlation score calculation function in the self-attention mechanism, represents the f-th sub-feature in the j-th time series feature, b represents the node threshold, and V(•) represents the value vector.

6. The cloud computing-based energy storage device management system according to claim 5, characterized in that: The specific steps of the SOC detection logic are as follows: S3.1: Acquire the battery operation signal of the energy storage device and convert it into an operation state characteristic parameter, wherein the conversion includes Fourier transform and normalization; S3.2: Optimize the boundary parameters of the operating state characteristic parameters according to the simulated annealing algorithm to determine the boundary of the characteristic parameters; S3.3: Perform curve fitting based on the characteristic parameter boundary and the operating state characteristic parameter, convert the operating state characteristic parameter into a capacity signal through wavelet transform, translate the signal, change the position of the discontinuity point, display it in the curve, and extract the operating characteristic variable; S3.4: Calculate the battery state of charge (SOC) based on the operating characteristic variables and the battery rated capacity.

7. The cloud computing-based energy storage device management system according to claim 6, characterized in that: The specific steps of energy scheduling logic configuration are as follows: S4.1: Obtain the regional power consumption period and battery state of charge SOC of this dispatch; S4.2: Determine whether the current time is in the valley period or the normal period. If it is in the valley period or the normal period, determine whether the battery state of charge SOC meets the requirements. If it is not satisfied, the energy storage device is charged and discharged by optimizing sub-logic A. If it is satisfied, the energy storage device is charged and discharged by optimizing sub-logic B. Indicates the maximum charge capacity of the battery under normal operation; S4.3: Determine whether the current time is during the peak period. If it is during the peak period, determine whether the battery state of charge SOC meets the requirements. If it is not satisfied, the energy storage device is charged and discharged through the optimization sub-logic C. If it is satisfied, the energy storage device is charged and discharged through the optimization sub-logic B, where Indicates the minimum charge capacity of the battery under normal operation; The optimization sub-logic A refers to the energy storage device performing a discharge operation, which is used to optimize the discharge power of the power battery of the energy storage device under the constraints of the system operation, so as to minimize the total operation cost; The optimization sub-logic B means that the energy storage device performs charging and discharging operations according to its own state of charge, and optimizes the discharge power and charging power of the power battery of the energy storage device under the constraints of system operation to minimize the total operating cost; The optimization sub-logic C refers to the energy storage device performing charging operation, optimizing the charging power of the power battery of the energy storage device under the constraints of system operation to minimize the total operating cost; The constraints include cost constraints, power capacity constraints, battery charging and discharging active power constraints, battery charging and discharging reactive power constraints, energy storage device operating time constraints and interactive capacity constraints.

Images