A photovoltaic curtain wall energy storage management system

By using data acquisition, temperature prediction and optimization algorithms in the photovoltaic curtain wall energy storage management system, voltage and thermal management problems in the photovoltaic curtain wall energy storage process are solved, and the efficient and stable operation of the system and the stability of the power grid are achieved.

CN118798444BActive Publication Date: 2024-12-06TIANJIN FEIYU CURTAIN WALL DECORATION ENGCO
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411275277.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2024-12-06
Estimated Expiration
2044-09-12

AI Technical Summary

Technical Problem

There are voltage management and thermal management problems in the energy storage process of photovoltaic curtain walls, which leads to overheating of the equipment, affecting performance and life, and thus causing voltage fluctuations and instability in the power grid.

Method used

Design an energy storage management system for photovoltaic curtain walls. The curtain wall is divided into the care area through the data acquisition module, combining autoregressive integral sliding average, random forest and XGBoost model to predict temperature, use the management solution output module to perform thermal management and control, and find the best algorithm Get the best thermal management strategy.

Benefits of technology

It realizes efficient thermal management during photovoltaic curtain wall energy storage, improves system stability and efficiency, and reduces equipment failures and grid instability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118798444B_ABST
    Figure CN118798444B_ABST
Patent Text Reader

Abstract

The present invention discloses an energy storage management system for a photovoltaic curtain wall, and relates to the field of thermal management of photovoltaic curtain wall energy storage. The technical solution thereof includes: a data acquisition module divides the curtain wall into individual attention areas according to environmental factors, and collects photovoltaic component data, energy storage component data and thermal management data in each attention area; a time prediction module is used to obtain an original data set at a current time point, and predict the temperature of the energy storage component and the photovoltaic component at the next time point according to a temperature prediction model, and classify the predicted temperature; a management solution output module is used to obtain a strategy enrichment set, and obtain the best management solution in the strategy enrichment set through an optimization algorithm and output it, and thermally manage the energy storage of the photovoltaic curtain wall according to the output best management solution, so as to realize fast and accurate output of the thermal management solution during photovoltaic curtain wall energy storage.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of thermal management of photovoltaic curtain wall energy storage, and more specifically, to an energy storage management system for a photovoltaic curtain wall. Background Art

[0002] Photovoltaic curtain walls need to convert and store light energy during the energy storage process. The growth of distributed photovoltaics may cause voltage management and thermal management problems in the distribution network, such as overvoltage, undervoltage, line and transformer reverse overload, etc. Photovoltaic modules will generate additional heat when converting solar energy for energy storage. An effective thermal management system is needed to maintain the appropriate operating temperature of the photovoltaic modules to ensure efficient operation of the system.

[0003] Improper thermal management may cause overheating of photovoltaic modules and other electrical equipment, affecting their performance and lifespan. Degraded equipment performance may cause voltage fluctuations and instability in the power grid. Photovoltaic modules may produce hot spots when partially shaded or with poor heat dissipation, which will cause local temperature rise, affect the electrical performance of the module, and even damage the battery cells, causing voltage instability. If the thermal problems of the energy storage system are not properly managed, it may affect its charging and discharging efficiency, and thus affect the stability of the power grid. Therefore, how to obtain a thermal management solution for photovoltaic curtain wall energy storage is now urgently needed. Summary of the invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide an energy storage management system for a photovoltaic curtain wall, so as to achieve rapid and accurate output of thermal management solutions when storing energy in the photovoltaic curtain wall.

[0005] To achieve the above object, the present invention provides the following technical solution: an energy storage management system for a photovoltaic curtain wall, the energy storage management system for a photovoltaic curtain wall comprising:

[0006] Data acquisition module: divide the curtain wall into The photovoltaic module data, energy storage module data and thermal management data in each care area are collected, and the photovoltaic module data and energy storage module data are combined into an original data set;

[0007] Time prediction module: obtains the original data set at the current time point, predicts the temperature of the energy storage component and the photovoltaic component at the next time point through the temperature prediction model, divides the predicted temperature into the first level and the second level, and performs thermal management and control on the first level scene through the management solution output module;

[0008] Management solution output module: obtains a rich set of strategies, obtains the best management solution in the rich set of strategies through an optimization algorithm and outputs it, and performs thermal management on the energy storage of the photovoltaic curtain wall according to the outputted best management solution;

[0009] The modules are connected to each other via wired and / or wireless means.

[0010] Preferably, the photovoltaic component data include the voltage, temperature, power, irradiance, working status and current on the photovoltaic component, the energy storage component data include the battery voltage, battery current, battery temperature, state of charge, health status and number of charge and discharge cycles on the energy storage component, and the thermal management data include ambient temperature, radiator temperature, wind speed, coolant temperature and cooling system operating status.

[0011] Preferably, the curtain wall is divided into The methods for each region include:

[0012] A base point is selected on the curtain wall, and the initial area is determined by the range function with the base point as the center. According to the initial area, the position of the base point is adaptively adjusted until the distance between two adjacent initial areas is less than the set value; the position of the current base point is fixed, and the attention area is divided according to the initial area, and the adjacent attention areas are seamlessly connected;

[0013] Defining Range Functions ;

[0014] in, represents the radius of the initial region, is the working status value of the photovoltaic module. If it works normally, , if there is a fault or shutdown, then ; is the unit irradiance, To take care of the effective photovoltaic area of ​​the curtain wall in the area, To take care of the output power of the curtain wall in the area, is the area of ​​the care area, and Indicates the adjustment parameter.

[0015] Preferably, predicting the temperature of the energy storage component and the photovoltaic component at the next time point by using the temperature prediction model specifically includes:

[0016] collect Group original data sets, perform clustering on the original data sets to obtain sample sets;

[0017] Defining the temperature prediction model ;

[0018] in, , and represents the weight coefficient, represents the predicted value of the first weak model, represents the predicted value of the second weak model, represents the predicted value of the third weak model;

[0019] Set weight coefficient ;

[0020] in, represents the index of the weight coefficient, and 1,2,3, Indicates that the qth weak model at the previous time point The predicted value at For the previous time point The actual measured value of the temperature at , q is the index of the weak model, and 1,2,3, is the rated weight value;

[0021] Define the first weak model using autoregressive integrated moving average, and ;

[0022] and To adjust the parameters, is the noise term, is the initial noise term, Indicates time point Energy storage component data or photovoltaic component data at Indicates the number of items, Represents a temporal index;

[0023] Define the second weak model using random forest construction decision trees and get the average value of the decision trees ;

[0024] in, Indicates time point Previous The predicted value of a decision tree, is the index of the decision tree, is the number of decision trees;

[0025] Define the third weak model to use XGBoost for gradient boosting, and reduce the prediction error through continuous iteration. ;

[0026] in, represents the number of iterations, Indicates time point Previous The weak learner of the iteration, is the index of the iteration number;

[0027] The sample set is divided into a training set and a test set. The temperature prediction model is trained using the training set and evaluated using the test set data. The evaluation standard is to stop when the mean square error is less than a preset threshold, thereby obtaining a trained temperature prediction model.

[0028] Input the original data set at the current time point t into the trained temperature prediction model to predict the temperature at the next time point The temperature of the upper energy storage components and photovoltaic components.

[0029] Preferably, the method of performing clustering processing on the original data set to obtain a sample set includes:

[0030] collect The original data set at consecutive historical time points is standardized and normalized;

[0031] Use a sliding window to divide the original data set into h_k segments, and then obtain h_k sub-series sets, and set the length of the sliding window , set the standard deviation as the statistical feature within the sliding window;

[0032] Use the improved DBSCAN method to perform cluster analysis on h_k sub-sequence sets and set the neighborhood radius And the range of points , if the number of sub-sequence sets in the neighborhood is within the range of point values, then the sub-sequence set is judged to be a core point, otherwise it is a free point;

[0033] in, Indicates the minimum number of sub-sequence sets that constitute a cluster, Indicates the maximum number of sub-sequence sets that constitute a cluster;

[0034] Calculate the distance between any two core points. If the distance is less than the threshold L_k, the two core points and the sub-sequence sets in the neighborhood are classified into the same cluster, and these clusters are recursively expanded until the number of sub-sequence sets in the cluster reaches the maximum number of sub-sequence sets. Stop when

[0035] Delete the sub-sequence set marked as a free point;

[0036] Neighborhood radius Represents the neighborhood range of each sub-timing set, that is, the radius of the spherical area centered on the sub-timing set, draws a visual distance graph, calculates the distance from each sub-timing set to its k_op nearest neighbors in the distance graph, and selects the mutation position in the distance graph as value;

[0037] Point value range , the threshold L_k is selected by empirical rules;

[0038] It is assumed that h_k sub-time series sets are divided into H_df clusters, each cluster is used as a sub-sample group, and all sub-sample groups are collected to obtain a sample set.

[0039] Preferably, the length of the sliding window , the acquisition methods include:

[0040] Set a candidate data set, the candidate data set consists of different candidate values, each candidate value is pre-set as the length of the sliding window, and the original data set is divided into h_k sub-time series sets using the pre-set sliding window length;

[0041] Calculate the data density within each segment ,in, Indicates The data of the photovoltaic module or energy storage module is consistent with the data of the The distance between the PV module data or energy storage module data, Indicates the number of PV module data or energy storage module data in the sliding window. represents a constant,

[0042] Calculate the volatility of data within each segment ,in, Indicates PV module data or energy storage module data, Represents the mean value of the PV module data or energy storage module data in the sliding window, Indicates the index of data read;

[0043] Collect the data density and data volatility of all candidate values ​​corresponding to the sliding window length, extract the candidate value with the smallest sum of data density and data volatility, and set it as the length of the sliding window .

[0044] Preferably, the method for classifying the predicted temperature comprises:

[0045] If the next time point is predicted If the temperature on the upper part is greater than the threshold value A_F, it is judged as the first level, which is marked as a danger warning and is in an abnormal working state. Thermal management regulation is required, and the management solution output module is started;

[0046] If the next time point is predicted If the temperature on the upper side is less than or equal to the threshold value A_F, it is judged as the second level, and the second level is marked as safe and controllable, in normal working state, and no thermal management regulation is required.

[0047] Preferably, the method for obtaining a policy enrichment set comprises:

[0048] Construct a solution explosion model based on neural network learning technology;

[0049] The original data set is set as the thermal management data at the current time point and the predicted temperature of the energy storage component and the photovoltaic component at the next time point. The original data set is the input of the scheme explosion model, and the output of the scheme explosion model is the strategy enrichment set, which includes the energy storage thermal management scheme of the photovoltaic curtain wall of group g_e;

[0050] The domestication dataset is set to include A_Q group domestication subsets. Each group of domestication subsets includes the original dataset at a historical time point and the corresponding strategy enrichment set. The domestication dataset is used to train the solution explosion model. The accuracy is used as the evaluation criterion. When the accuracy reaches 98%, the training is stopped to obtain the trained solution explosion model.

[0051] The original data set at time point t is input into the trained solution explosion model to obtain the strategy enrichment set.

[0052] Preferably, the method of obtaining the best management solution from a rich set of strategies through an optimization algorithm specifically includes:

[0053] The optimization goal is defined as maximum heat dissipation efficiency and minimum energy use;

[0054] Use the particle swarm optimization algorithm to find the best management solution, and initialize the number of particles to R_o;

[0055] Setting Particles The speed update formula is: ;

[0056] in, Indicates Particles during round iteration speed, Indicates Particles during round iteration location, It is a particle The local optimal position of is the global best position, represents the inertia weight, , is the learning factor, , is a random factor between 0 and 1, , , and , All are set through experience or experimental data analysis. Indicates the maximum speed limit;

[0057] Set the position update formula to: ;

[0058] in, For the Particles during round iteration location;

[0059] The constraints are set as follows: the heat dissipation efficiency is greater than the threshold FG_o1, and the energy usage is less than the threshold FG_o2;

[0060] The position and velocity of the particle are updated through the formula, and the state of the particle is optimized to find the optimal solution, which is recorded as the best management solution.

[0061] Preferably, the energy storage thermal management solution includes wind speed parameter settings, cooling system operating state settings and coolant temperature setting parameters.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] In the present invention, the advantages of multiple weak models are combined by integrating autoregressive integrated moving average, random forest and XGBoost, and the weights are dynamically adjusted according to the accuracy of each weak model through adaptive weight coefficient setting, so as to improve the accuracy of time series prediction. The purpose of clustering in time series data is to cluster time series data segments with similar characteristics together, so as to improve the training of the model by using these similarities without destroying the order of the time series, so that the model converges quickly, thereby improving the training efficiency and prediction effect, and realizing high-quality input of subsequent data depth analysis, and creating multiple feasible energy storage thermal management schemes through the scheme explosion model, improving the selectivity of the scheme, and automatically producing rich control schemes according to specific constraints, enhancing the multi-possibility and optimality of scheme selection, and the best management scheme with rich and concentrated strategies realized by the present invention is obtained and output through the optimization algorithm, and the energy storage of the photovoltaic curtain wall is thermally managed according to the output best management scheme; the adaptive control optimization mechanism continuously improves the overall efficiency of intelligent analysis, and multiple modules cooperate with each other to realize comprehensive data collection, processing and analysis, and provide real-time and accurate analysis of detection data. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 The present invention provides a schematic structural diagram of an energy storage management system for a photovoltaic curtain wall;

[0065] Figure 2 Schematic diagram of the steps of the energy storage management system applied to the photovoltaic curtain wall in the present invention. DETAILED DESCRIPTION

[0066] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0067] Embodiment 1

[0068] Reference Figure 1 and Figure 2 Embodiment 1 further illustrates a photovoltaic curtain wall energy storage management system proposed by the present invention.

[0069] Photovoltaic curtain walls need to convert light energy into energy and store it during energy storage. The growth of distributed photovoltaics may cause voltage management and thermal management problems for distribution networks, such as overvoltage, undervoltage, reverse overload of lines and transformers, etc. Photovoltaic modules will generate additional heat when converting solar energy for energy storage. An effective thermal management system is required to maintain the appropriate operating temperature of photovoltaic modules and ensure efficient operation of the system. Thermal management does have an impact on voltage management and grid stability. Although they are two different issues, there is a certain connection and mutual influence between them, mainly including the following:

[0070] Improper thermal management may cause PV panels and other electrical equipment to overheat, affecting their performance and lifespan. Degraded equipment performance may cause voltage fluctuations and instability in the power grid. PV panels may produce hot spots when partially shaded or with poor heat dissipation, which can cause local temperature rise, affect the electrical performance of the components, and even damage the cells, causing voltage instability.

[0071] Rising temperatures usually cause the open-circuit voltage of photovoltaic cells to decrease, affecting the overall output voltage level and, in turn, the voltage stability of the power grid. Continuous high temperatures may affect the stability of electrical connections, such as poor contact or wire aging, which may cause voltage fluctuations in the power grid. Extreme temperature changes may cause thermal expansion and contraction of photovoltaic modules and mounting structures, affecting the mechanical stability of the modules and the reliability of electrical connections, which may indirectly affect the stability of the power grid.

[0072] Transformers, switchgear, etc. in the power grid may require additional cooling measures in high temperature environments to ensure their normal operation and avoid equipment failure or performance degradation due to overheating. Good thermal management requires considering appropriate heat dissipation measures during system design, such as heat sinks, ventilation slots, etc. These designs may affect the voltage characteristics and grid compatibility of the system. Heat is also generated during the energy storage management process, and effective thermal management is required to maintain its performance and safety. If the thermal problems of the energy storage system are not properly managed, it may affect its charging and discharging efficiency, and then affect the stability of the power grid. Therefore, thermal management is an important part of ensuring the stable operation of photovoltaic systems and power grid equipment. Good thermal management can not only improve the efficiency and life of photovoltaic modules and other power equipment, but also reduce voltage fluctuations and grid instability caused by equipment failure or performance degradation.

[0073] The intelligent management and control system is the core of the photovoltaic curtain wall energy storage management system. Through real-time data collection, analysis and processing, and automatic control, it realizes all-round monitoring and optimization of the system, ensuring the efficient operation of photovoltaic power generation, energy storage management, thermal management, and voltage management.

[0074] Data acquisition module: divide the curtain wall into Collect the photovoltaic module data, energy storage module data and thermal management data in each care area;

[0075] PV module data includes the voltage, temperature, power, irradiance, working status and current of the PV module; energy storage module data includes the battery voltage, battery current, battery temperature, state of charge, health status and number of charge and discharge cycles of the energy storage module; thermal management data includes ambient temperature, radiator temperature, wind speed, coolant temperature and cooling system operating status.

[0076] The curtain wall is divided into The methods for each region include:

[0077] A base point is selected on the curtain wall, and the initial area is determined by the range function with the base point as the center. According to the initial area, the position of the base point is adaptively adjusted until the distance between two adjacent initial areas is less than the set value; the position of the current base point is fixed, and the attention area is divided according to the initial area, and the adjacent attention areas are seamlessly connected;

[0078] The specific method of dividing the attention area according to the initial area can be: determine the radius of the initial area through the range function, draw a circle according to the radius to obtain a circle, find the circumscribed square of the circle, record the obtained square as the initial area, divide the distance between two adjacent initial areas equally into the two initial areas, extend the square to obtain the attention area.

[0079] Defining Range Functions ;in, is the working status value of the photovoltaic module. If it works normally, , if there is a fault or shutdown, then ; is the unit irradiance, To take care of the effective photovoltaic area of ​​the curtain wall in the area, To take care of the output power of the curtain wall in the area, is the area of ​​the care area, and Represents adjustment parameters, which can be set based on experimental data analysis or experience;

[0080] The data acquisition module of the intelligent management and control system needs to collect the following data to achieve comprehensive monitoring and optimization of the photovoltaic curtain wall energy storage system. Data acquisition can be achieved by setting sensors at the required locations and reading and collecting data through sensors:

[0081] 1. Photovoltaic module data:

[0082] Current: The current value generated by the photovoltaic module, used to calculate power output and evaluate system performance. It is collected using a Hall effect current sensor or shunt. Commonly used equipment includes current clamp meters and current transmitters.

[0083] Voltage: The output voltage of a PV panel, used to evaluate system performance and diagnose potential problems. It is collected using a voltage sensor or voltage divider. Commonly used equipment includes a voltmeter and a voltage transmitter.

[0084] Power: The power generated by photovoltaic modules. It is used to calculate energy yield and evaluate system efficiency. It is calculated from current and voltage. It can also be measured directly using a power meter. Commonly used equipment includes power meters and power analyzers.

[0085] Irradiance: The intensity of solar radiation received by photovoltaic modules. It is used to evaluate photovoltaic power generation efficiency and predict power generation. It is collected using irradiance sensors. Commonly used equipment includes silicon photovoltaic cell irradiance meters and global pyranometers.

[0086] Temperature: The temperature of photovoltaic modules is used to evaluate module performance and thermal management effects. It is collected using temperature sensors such as PT100 and thermocouples. Commonly used equipment includes temperature sensors and temperature transmitters.

[0087] Working status: The operating status of the PV module, such as normal operation, failure, shutdown, etc., is used to monitor the system operation status by analyzing data such as current, voltage, power, etc., or using switch sensors for collection. Commonly used devices include relays and switch sensors.

[0088] 2. Energy storage component data:

[0089] Battery voltage: The voltage of the energy storage battery pack, used to evaluate the battery status and diagnose potential problems. It is collected using a voltage sensor or voltage divider. Commonly used equipment includes voltmeters and voltage transmitters.

[0090] Battery current: The charge and discharge current of the energy storage battery pack, which is used to calculate the charge and discharge power and evaluate the battery performance. It is collected using a Hall effect current sensor or shunt. Commonly used equipment includes current clamp meters and current transmitters.

[0091] Battery temperature: The temperature of the energy storage battery pack is used to evaluate battery performance and thermal management effects. It is collected using temperature sensors such as PT100 and thermocouples. Commonly used equipment: temperature sensors and temperature transmitters.

[0092] State of charge: The percentage of remaining power in an energy storage battery pack. It is used to evaluate the available capacity of the energy storage system and is calculated by a battery management system (BMS). Commonly used equipment includes BMS.

[0093] Health status: The health status of the energy storage battery pack is used to evaluate the battery life and performance degradation. It is calculated by BMS based on parameters such as battery internal resistance and capacity degradation. Commonly used equipment includes BMS.

[0094] Number of charge and discharge cycles: The number of charge and discharge cycles of the energy storage battery pack is used to evaluate the battery life. It is recorded and counted by BMS. Commonly used equipment includes BMS.

[0095] 3. Thermal management data:

[0096] Ambient temperature: The ambient temperature around the photovoltaic system is used to evaluate the efficiency of the thermal management system and predict system performance. It is collected using temperature sensors such as PT100 and thermocouples. Commonly used equipment includes temperature sensors and temperature transmitters.

[0097] Radiator temperature: The radiator temperature of a photovoltaic module or energy storage system is used to evaluate the heat dissipation effect and diagnose potential problems. It is collected using temperature sensors such as PT100 and thermocouples. Commonly used equipment includes temperature sensors and temperature transmitters.

[0098] Wind speed: The wind speed around the photovoltaic system is used to evaluate the natural heat dissipation effect. Wind speed sensors are used for collection. Commonly used equipment include ultrasonic anemometers and cup anemometers.

[0099] Coolant temperature: If a liquid cooling system is used, the coolant temperature needs to be collected to evaluate the cooling effect and diagnose potential problems. Use temperature sensors such as PT100 and thermocouples to collect the temperature. Commonly used equipment includes temperature sensors and temperature transmitters.

[0100] Cooling system operating status: The operating status of the cooling system, such as normal operation, failure, shutdown, etc., is used to monitor the system operating status by analyzing data such as coolant temperature and flow, or using switch sensors for collection. Commonly used equipment includes flow meters and switch sensors.

[0101] For analog signal acquisition, the analog signal output by the sensor is converted into a digital signal, and then collected and processed by a data acquisition card or PLC. For digital signal acquisition, the digital signal output by the sensor is directly collected, such as RS485, Modbus and other communication protocols. For wireless transmission, wireless technologies such as WiFi, Zigbee, and LoRa are used to transmit data to a server or cloud platform.

[0102] By collecting the above data, the intelligent management and control system can monitor the operating status of the photovoltaic curtain wall energy storage system in real time, analyze system performance, optimize system control strategies, and promptly discover and handle potential problems, thereby ensuring efficient, safe and stable operation of the system.

[0103] The data is collected at a fixed frequency, for example, every q seconds, to ensure real-time monitoring. The sensor data is aggregated and transmitted to a centralized processing center using a wireless sensor network (WSN) or a wired network. The data processing center (platform) of the present invention can use Hadoop HDFS for big data storage, use a NoSQL database (such as MongoDB) for structured data storage, use Apache Kafka for data stream processing, and combine with Apache Spark for real-time data analysis.

[0104] Time prediction module: obtains the original data set at the current time point, predicts the temperature of the energy storage component and photovoltaic component at the next time point through the temperature prediction model, and classifies the predicted temperature;

[0105] The temperature prediction model is used to predict the temperature of the energy storage component and the photovoltaic component at the next time point, including:

[0106] collect Group original data sets, perform clustering on the original data sets to obtain sample sets;

[0107] Defining the temperature prediction model ;

[0108] in, , and represents the weight coefficient, represents the predicted value of the first weak model, represents the predicted value of the second weak model, represents the predicted value of the third weak model;

[0109] Set weight coefficient ,in, represents the index of the weight coefficient, 1,2,3, that is to say Can be used to express , and , Indicates that the qth weak model at the previous time point The predicted value at , q is the index of the weak model, 1,2,3, that is to say Can be used to express , and , which can be used to represent the 1st, 2nd and 3rd weak models at the previous time point The predicted value at For the previous time point The actual measured value of the temperature at is the rated weight value;

[0110] Define the first weak model using autoregressive integrated moving average, and ;

[0111] and To adjust the parameters, is the noise term, is the initial noise term, Indicates time point Energy storage component data or photovoltaic component data at Indicates the number of items;

[0112] Define the second weak model using random forest construction decision trees and get the average value of the decision trees ;

[0113] in, Indicates time point Previous The predicted value of a decision tree, is the index of the decision tree, is the number of decision trees; the parameters of the random forest model include the number of decision trees and the hyperparameters of each decision tree, which need to be tuned through cross-validation.

[0114] Define the third weak model to use XGBoost for gradient boosting, and reduce the prediction error through continuous iteration. ;

[0115] in, represents the number of iterations, Indicates time point Previous The weak learner of the iteration, is the index of the number of iterations, where the weak learner can be a decision tree;

[0116] The sample set is divided into a training set and a test set. The temperature prediction model is trained using the training set and evaluated using the test set data. The evaluation standard is to stop when the mean square error (MSE) is less than a preset threshold, thereby obtaining a trained temperature prediction model.

[0117] Input the original data set at the current time point t into the trained temperature prediction model to predict the temperature at the next time point The temperature of the upper energy storage components and photovoltaic components.

[0118] By integrating autoregressive integrated moving average, random forest and XGBoost (ExtremeGradient Boosting), and then combining the advantages of multiple weak models, the weights are dynamically adjusted according to the accuracy of each weak model through adaptive weight coefficient setting, thereby improving the accuracy of time series prediction.

[0119] The method of clustering the original data set to obtain the sample set includes:

[0120] Identify the working status of the PV modules and the charge state and health status of the energy storage modules. When the working status of the PV modules is normal operation, and the charge state and health status of the energy storage modules are both in normal working state, that is, the remaining power of the energy storage battery pack is not fully charged or the health status of the energy storage module is within the preset normal working range, collect data on the PV modules and the energy storage modules.

[0121] The purpose of clustering time series data is to cluster time series data segments with similar characteristics together, so as to improve the training of the model by using these similarities without destroying the order of the time series, so that the model can converge quickly, thereby improving the training efficiency and prediction effect.

[0122] collect The original data set at consecutive historical time points is standardized and normalized;

[0123] Use a sliding window to divide the original data set into h_k segments, and then obtain h_k sub-series sets, and set the length of the sliding window , set the standard deviation as the statistical feature within the sliding window;

[0124] Use the improved DBSCAN method to perform cluster analysis on h_k sub-sequence sets and set the neighborhood radius And the range of points , if the number of sub-sequence sets in the neighborhood is within the range of point values, then the sub-sequence set is judged to be a core point, otherwise it is a free point;

[0125] in, Indicates the minimum number of sub-sequence sets that constitute a cluster, Indicates the maximum number of sub-sequence sets that constitute a cluster;

[0126] Calculate the distance between any two core points. If the distance is less than the threshold L_k, the two core points and the sub-sequence sets in the neighborhood are classified into the same cluster, and these clusters are recursively expanded until the number of sub-sequence sets in the cluster reaches the maximum number of sub-sequence sets. Stop when

[0127] Delete the sub-sequence set marked as a free point;

[0128] Neighborhood radius Represents the neighborhood range of each sub-timing set, that is, the radius of the spherical area centered on the sub-timing set, draws a visual distance graph, calculates the distance from each sub-timing set to its k_op nearest neighbors in the distance graph, and selects the mutation position in the distance graph as value.

[0129] Point value range , the threshold L_k are selected by empirical rules.

[0130] It is assumed that h_k sub-time series sets are divided into H_df clusters, each cluster is used as a sub-sample group, and all sub-sample groups are collected to obtain a sample set.

[0131] The length of the sliding window , the acquisition methods include:

[0132] Set a candidate data set, the candidate data set consists of different candidate values, each candidate value is pre-set as the length of the sliding window, and the original data set is divided into h_k sub-time series sets using the pre-set sliding window length;

[0133] Calculate the data density within each segment ,in, Indicates The data of the photovoltaic module or energy storage module is consistent with the data of the The distance between the PV module data or energy storage module data, Indicates the number of PV module data or energy storage module data in the sliding window. represents a constant,

[0134] Calculate the volatility of data within each segment ,in, Indicates PV module data or energy storage module data, Represents the mean value of the PV module data or energy storage module data in the sliding window, Indicates the index of data read;

[0135] Collect the data density and data volatility of all candidate values ​​corresponding to the sliding window length, extract the candidate value with the smallest sum of data density and data volatility, and set it as the length of the sliding window .

[0136] Methods for ranking predicted temperatures include:

[0137] If the next time point is predicted If the temperature on the upper part is greater than the threshold value A_F, it is judged as the first level, which is marked as a danger warning and is in an abnormal working state. Thermal management regulation is required, and the management solution output module is started;

[0138] If the next time point is predicted If the temperature on the upper side is less than or equal to the threshold value A_F, it is judged as the second level, and the second level is marked as safe and controllable, in normal working state, and no thermal management regulation is required.

[0139] Management solution output module: obtains a rich set of strategies, obtains the best management solution in the rich set of strategies through an optimization algorithm and outputs it, and performs thermal management on the energy storage of the photovoltaic curtain wall according to the outputted best management solution.

[0140] Methods for obtaining a strategy enrichment set include:

[0141] Construct a solution explosion model based on neural network learning technology;

[0142] The original data set is set as the thermal management data at the current time point and the predicted temperature of the energy storage component and the photovoltaic component at the next time point. The original data set is the input of the scheme explosion model, and the output of the scheme explosion model is the strategy enrichment set, which includes the energy storage thermal management scheme of the photovoltaic curtain wall of group g_e;

[0143] The domestication dataset is set to include A_Q group domestication subsets. Each group of domestication subsets includes the original dataset at a historical time point and the corresponding strategy enrichment set. The solution explosion model is trained using the domestication dataset, and the accuracy is used as the evaluation criterion. When the accuracy reaches 98%, the training is stopped to obtain the trained solution explosion model.

[0144] The original data set at time point t is input into the trained solution explosion model to obtain the strategy enrichment set.

[0145] The optimal management solution with rich strategy set is obtained through optimization algorithm, including:

[0146] The optimization objectives are defined as the highest heat dissipation efficiency (the fastest temperature drop rate, evaluated by the heat sink temperature) and the lowest energy usage (the lowest energy consumption during the cooling process).

[0147] Use the particle swarm optimization algorithm to find the best management solution, and initialize the number of particles to R_o;

[0148] Setting Particles The speed update formula is: ;

[0149] in, Indicates Particles during round iteration speed, Indicates Particles during round iteration location, It is a particle The local optimal position of is the global best position, represents the inertia weight, , is the learning factor, , is a random factor between 0 and 1, , , and , All are set through experience or experimental data analysis. Represents the maximum speed limit, which is used to constrain the range of particle speed to prevent particles from moving too fast;

[0150] Set the position update formula to: ;

[0151] in, For the Particles during round iteration location;

[0152] The constraints are set as follows: the heat dissipation efficiency is greater than the threshold FG_o1, and the energy usage is less than the threshold FG_o2;

[0153] The position and velocity of the particle are updated through the formula, and the state of the particle is optimized to find the optimal solution, which is recorded as the best management solution.

[0154] The energy storage thermal management solution includes wind speed parameter settings, cooling system operating status settings, and coolant temperature setting parameters.

[0155] In this embodiment, the present invention integrates autoregressive integrated moving average, random forest and XGBoost (Extreme Gradient Boosting), and then combines the advantages of multiple weak models, and dynamically adjusts the weights according to the accuracy of each weak model through adaptive weight coefficient setting, thereby improving the accuracy of time series prediction. The purpose of clustering in time series data is to cluster time series data segments with similar characteristics together, so that these similarities can be used to improve the training of the model without destroying the order of the time series, so that the model converges quickly, thereby improving the training efficiency and prediction effect, and realizing high-quality input for subsequent data depth analysis. Through the scheme explosion model, multiple feasible energy storage thermal management schemes are created to improve the selectivity of the scheme. According to specific constraints, rich control schemes are automatically produced to enhance the multi-possibility and optimality of scheme selection. The present invention obtains the best management scheme with rich and concentrated strategies through the optimization algorithm and outputs it, and performs thermal management on the energy storage of the photovoltaic curtain wall according to the output best management scheme; the adaptive control optimization mechanism continuously improves the overall efficiency of intelligent analysis, and multiple modules cooperate with each other to realize comprehensive data collection, processing and analysis, and provide real-time and accurate analysis of detection data.

[0156] Embodiment 2

[0157] Reference Figure 1 and Figure 2 , Example 2 further illustrates a photovoltaic curtain wall energy storage management system proposed by the present invention.

[0158] The energy storage management method of a photovoltaic curtain wall comprises the following steps:

[0159] Step S1: Divide the curtain wall into Collect the photovoltaic module data, energy storage module data and thermal management data in each care area;

[0160] Step S2: construct and train a temperature prediction model, import the original data set at the current time point into the trained temperature prediction model, and predict the temperature of the energy storage component and the photovoltaic component at the next time point;

[0161] Step S3: classify the temperatures of the energy storage components and the photovoltaic components at the next time point, and use the management solution output module for the temperature marked as the first level;

[0162] Step S4: construct and train a solution explosion model, and obtain a strategy enrichment set at the current time point according to the solution explosion model;

[0163] Step S5: Obtain and output the best management solution with rich strategies through an optimization algorithm, and perform thermal management on the energy storage of the photovoltaic curtain wall according to the output best management solution.

[0164] The energy storage management method of a photovoltaic curtain wall is applied to an information processing system for overall engineering quality detection and includes:

[0165] Data acquisition module: divide the curtain wall into Collect the photovoltaic module data, energy storage module data and thermal management data in each care area;

[0166] Time prediction module: obtains the original data set at the current time point, predicts the temperature of the energy storage component and photovoltaic component at the next time point through the temperature prediction model, and classifies the predicted temperature;

[0167] Management solution output module: obtains a rich set of strategies, obtains the best management solution in the rich set of strategies through an optimization algorithm and outputs it, and performs thermal management on the energy storage of the photovoltaic curtain wall according to the outputted best management solution;

[0168] The modules are connected to each other via wired and / or wireless means.

[0169] In addition, according to the implementation of the present application, the process described in the attached figure of a photovoltaic curtain wall energy storage management system can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application. Of course, the architecture shown in the attached figure of a photovoltaic curtain wall energy storage management system is only exemplary. When implementing different devices, adaptive selection or adjustment can be performed according to actual needs.

[0170] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0171] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A photovoltaic curtain wall energy storage management system, It is characterized in that The energy storage management system of the photovoltaic curtain wall includes: Data acquisition module: divides the curtain wall into q_w attention areas according to environmental factors, collects photovoltaic module data, energy storage module data and thermal management data in each attention area, and combines the photovoltaic module data and energy storage module data into the original data set; Time prediction module: obtains the original data set at the current time point, predicts the temperature of the energy storage component and the photovoltaic component at the next time point through the temperature prediction model, divides the predicted temperature into the first level and the second level, and performs thermal management and control on the first level scene through the management solution output module; The method of predicting the temperature of the energy storage component and the photovoltaic component at the next time point by using the temperature prediction model specifically includes: Collect D_z groups of original data sets, perform clustering on the original data sets to obtain sample sets; Define the temperature prediction model Y_mx=ω1×Y_m1+ω2×Y_m2+ω3×Y_m3; Among them, ω1, ω2 and ω3 represent weight coefficients, Y_m1 represents the predicted value of the first weak model, Y_m2 represents the predicted value of the second weak model, and Y_m3 represents the predicted value of the third weak model; Set weight coefficient Where o represents the index of the weight coefficient, and o = 1, 2, 3, Y_mq t-1 represents the predicted value of the qth weak model at the previous time point t-1, s_mx t-1 is the actual measured value of the temperature at the previous time point t-1, q is the index of the weak model, and q=1,2,3, is the rated weight value; Define the first weak model using autoregressive integrated moving average, and and θi are the training parameters; ε t-i is the noise term, ε t is the initial noise term, y t-i represents the energy storage component data or photovoltaic component data at time point ti, and p represents the number of items; Define the second weak model to build r_n decision trees using random forest and get the average value of the decision trees Among them, Tree r_i (X_t) represents the predicted value of the r_i-th decision tree at time point t, where r_i is the index of the decision tree and r_n is the number of decision trees; Define the third weak model to use XGBoost for gradient boosting, and reduce the prediction error through continuous iteration. Among them, K_a represents the number of iterations, f k_u (X_t) represents the weak learner of the k_uth iteration at time point t, where k_u is the index of the number of iterations; The sample set is divided into a training set and a test set. The temperature prediction model is trained using the training set and evaluated using the test set data. The evaluation standard is to stop when the mean square error is less than a preset threshold, thereby obtaining a trained temperature prediction model. Input the original data set at the current time point t into the trained temperature prediction model to predict the temperature of the energy storage component and the photovoltaic component at the next time point t+1; The method for classifying the predicted temperature comprises: If the predicted temperature at the next time point t+1 is greater than the threshold A_F, it is judged as the first level, which is marked as a danger warning and is in an abnormal working state. Thermal management control is required, and the management solution output module is started; If the predicted temperature at the next time point t+1 is less than or equal to the threshold A_F, it is determined to be the second level, which is marked as safe and controllable, in normal working state, and does not require thermal management control; Management solution output module: obtains a rich set of strategies, obtains the best management solution in the rich set of strategies through an optimization algorithm and outputs it, and performs thermal management on the energy storage of the photovoltaic curtain wall according to the outputted best management solution; The modules are connected to each other via wired and / or wireless means.

2. The energy storage management system for photovoltaic curtain wall according to claim 1, It is characterized in that The photovoltaic module data includes the voltage, temperature, power, irradiance, working status and current on the photovoltaic module; the energy storage module data includes the battery voltage, battery current, battery temperature, state of charge, health status and number of charge and discharge cycles on the energy storage module; the thermal management data includes the ambient temperature, radiator temperature, wind speed, coolant temperature and cooling system operating status.

3. The energy storage management system for the photovoltaic curtain wall according to claim 2, It is characterized in that The method of dividing the curtain wall into q_w areas by environmental factors includes: A base point is selected on the curtain wall, and the initial area is determined by the range function with the base point as the center. According to the initial area, the position of the base point is adaptively adjusted until the distance between two adjacent initial areas is less than the set value; the position of the current base point is fixed, and the attention area is divided according to the initial area, and the adjacent attention areas are seamlessly connected; Defining Range Functions Wherein, r_fw represents the radius of the initial area, Z_tz is the working state value of the photovoltaic module, if it works normally, Z_tz=1, if it fails or shuts down, Z_tz=0; F_zd is the unit irradiance, S_gf is the effective photovoltaic area of ​​the curtain wall in the cared area, P_out is the output power of the curtain wall in the cared area, S_qy is the area of ​​the cared area, and α1 and α2 represent the adjustment parameters.

4. The energy storage management system for the photovoltaic curtain wall according to claim 3, It is characterized in that The method for performing clustering processing on the original data set to obtain a sample set includes: Collect the original data sets at D_z consecutive historical time points and perform standardization and normalization; Use a sliding window to divide the original data set into h_k segments, and then obtain h_k sub-series sets. Set the length of the sliding window d_g, and set the standard deviation as the statistical feature within the sliding window; Use the improved DBSCAN method to perform cluster analysis on h_k sub-sequence sets, set the neighborhood radius σ and the point value range [min_Pts, max_Pts], if the number of sub-sequence sets in the neighborhood is within the point value range, then the sub-sequence set is judged as a core point, otherwise it is a free point; Among them, min_Pts represents the minimum number of sub-timing sets that constitute a cluster, and maxc_Pts represents the maximum number of sub-timing sets that constitute a cluster; Calculate the distance between any two core points. If the distance is less than the threshold L_k, the two core points and the sub-sequence sets in the neighborhood are classified into the same cluster, and these clusters are recursively expanded until the number of sub-sequence sets in the cluster reaches the maximum number of sub-sequence sets max_Pts. Delete the sub-sequence set marked as a free point; The neighborhood radius σ represents the neighborhood range of each sub-timing set, that is, the radius of the spherical area centered on the sub-timing set. Draw a visual distance graph. In the distance graph, calculate the distance from each sub-timing set to its k_op nearest neighbors, and select the mutation position in the distance graph as the σ value; The point value range [min_Pts, max_Pts] and the threshold L_k are selected by empirical rules; It is assumed that h_k sub-time series sets are divided into H_df clusters, each cluster is used as a sub-sample group, and all sub-sample groups are collected to obtain a sample set.

5. The energy storage management system for the photovoltaic curtain wall according to claim 4, It is characterized in that The length d_g of the sliding window is obtained by: Set a candidate data set, the candidate data set consists of different candidate values, each candidate value is pre-set as the length of the sliding window, and the original data set is divided into h_k sub-time series sets using the pre-set sliding window length; Calculate the data density within each segment Wherein, d_o(mj,mi) represents the distance between the mj-th photovoltaic module data or energy storage module data and the mi-th photovoltaic module data or energy storage module data, d_g1 represents the number of photovoltaic module data or energy storage module data in the sliding window, and zx represents a constant; Calculate the volatility of data within each segment Among them, xg_s represents the s-th photovoltaic component data or energy storage component data, represents the mean value of the PV module data or energy storage module data in the sliding window, and s represents the index of data reading; Collect the data density and data volatility in the sliding window length corresponding to all candidate values, extract the candidate value with the smallest sum of data density and data volatility, and set it as the length of the sliding window d_g.

6. The energy storage management system for the photovoltaic curtain wall according to claim 5, It is characterized in that The method for obtaining a strategy enrichment set comprises: Construct a solution explosion model based on neural network learning technology; The original data set is set as the thermal management data at the current time point and the predicted temperature of the energy storage component and the photovoltaic component at the next time point. The original data set is the input of the scheme explosion model, and the output of the scheme explosion model is the strategy enrichment set, which includes the energy storage thermal management scheme of the photovoltaic curtain wall of group g_e; The domestication data set is set to include A_Q group domestication subsets. Each group of domestication subsets includes the original data set at a historical time point and the corresponding strategy enrichment set. The domestication data set is used to train the solution explosion model. The accuracy rate is used as the evaluation standard. When the accuracy rate reaches 98%, the training is stopped to obtain the trained solution explosion model. The original data set at time point t is input into the trained solution explosion model to obtain the strategy enrichment set.

7. The energy storage management system for the photovoltaic curtain wall according to claim 6, It is characterized in that The method of obtaining the best management solution with rich strategies through the optimization algorithm specifically includes: The optimization goal is defined as maximum heat dissipation efficiency and minimum energy use; Use the particle swarm optimization algorithm to find the best management solution, and initialize the number of particles to R_o; Set the velocity update formula of particle v: in, represents the velocity of particle v at the bn+1th iteration, represents the position of particle v at the bnth iteration, lz_g v is the local optimal position of particle v, zj is the global optimal position, τ represents the inertia weight, β1 and β2 are learning factors, μ1 and μ2 are random factors between 0 and 1, τ, β1, β2 and μ1 and μ2 are all set through experience or experimental data analysis, and V_sd represents the maximum speed limit; Set the position update formula to: in, is the position of particle v at the bn+1th iteration; The constraints are set as follows: the heat dissipation efficiency is greater than the threshold FG_o1, and the energy usage is less than the threshold FG_o2; The position and velocity of the particle are updated through the formula, and the state of the particle is optimized to find the optimal solution, which is recorded as the best management solution.

8. The energy storage management system for the photovoltaic curtain wall according to claim 7, It is characterized in that The energy storage thermal management solution includes wind speed parameter settings, cooling system operating state settings and coolant temperature setting parameters.