Energy early warning and scheduling method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses an energy early warning and scheduling method and device, electronic equipment and a storage medium, and relates to the technical field of new energy and artificial intelligence, and the method comprises the steps: collecting alarm data, BMS and energy equipment operation data to construct an original data set, and collecting multi-source energy data in real time to form an energy data pool. An original data set is cleaned, standardized and subjected to feature extraction, a training data set is constructed in combination with historical fault records, and high-quality data support is provided for model training. A fault early warning model is constructed by using a CNN-LSTM fusion algorithm, the time sequence trend of the battery temperature and voltage is accurately captured, and predictive early warning is realized; an energy scheduling model is constructed by adopting reinforcement learning and a multi-objective optimization algorithm, an optimal energy distribution strategy is generated according to real-time data and dynamic changes, the optimal strategy is converted into a control instruction, and dynamic scheduling and early warning of energy are realized. According to the invention, the problems of early warning lagging and unintelligent scheduling in the prior art are effectively solved.

Description

Technical Field

[0001] This invention relates to the fields of new energy and artificial intelligence technology, and in particular to an energy early warning and scheduling method, device, electronic device and storage medium. Background Technology

[0002] With the rapid development of the new energy industry, the application of multi-source new energy systems is becoming increasingly widespread. However, the fault warning and energy management issues they face are also becoming more complex. Traditional fault warning methods mostly rely on single-dimensional data monitoring or fixed threshold judgment, failing to fully integrate multi-dimensional core data from the BMS (such as temperature, cell voltage, SOC, SOH, etc.) for trend analysis. This results in delayed warnings, high false alarm rates, and difficulty in preventing fault risks such as battery overheating and abnormal voltage in advance, seriously affecting the safe and stable operation of the system.

[0003] Meanwhile, the current scheduling and management of multi-source renewable energy (photovoltaics, wind power, power grid, and energy storage) lacks an intelligent and collaborative mechanism. Data from the source, grid, load, and storage links are fragmented, and energy allocation relies on experience-based operations, making it difficult to dynamically adjust strategies based on real-time supply and demand changes, environmental fluctuations, and load demand. This leads to frequent problems such as low renewable energy absorption rates, high grid dependence, high operating costs, and supply-demand imbalances, hindering the sustainable development of the renewable energy industry.

[0004] Therefore, there is an urgent need for an energy early warning and scheduling method that can integrate multi-dimensional data, achieve predictive fault alerts and dynamic optimal energy allocation, and improve the efficiency of new energy utilization and the stability of system operation. Summary of the Invention

[0005] This invention provides an energy early warning and dispatching method to address the problems of delayed fault early warning and high false alarm rate in existing technologies, lack of intelligent coordination in energy dispatching, and low renewable energy absorption rate and high operating costs due to data fragmentation. The technical solution is as follows: According to one aspect of the present invention, an energy early warning and scheduling method is provided, the method comprising: collecting alarm data, BMS and energy equipment operation data to construct an original dataset; collecting energy data in real time to construct an energy data pool; the energy data includes photovoltaic power plant data, wind farm data, power grid data, energy storage system data, and load demand data; performing data cleaning, standardization processing and feature extraction on the original dataset to obtain feature variables strongly correlated with battery faults and combining them with historical fault records to construct a training dataset; the data cleaning includes removing outliers and filling in missing data; constructing a fault early warning model using a CNN-LSTM fusion deep learning algorithm; training the fault early warning model based on the training dataset to obtain a trained fault early warning model; and predicting the real-time collected operation data to obtain prediction results; the fault early warning model is used to capture the temporal trends of battery temperature and voltage; constructing an energy scheduling model using reinforcement learning and multi-objective optimization algorithms and performing iterative training; using the trained energy scheduling model based on the energy data pool and the prediction results to generate an optimal energy scheduling strategy; and converting the optimal energy scheduling strategy into control commands for energy early warning and scheduling.

[0006] In one embodiment, the construction of the original dataset by collecting alarm data, BMS, and energy equipment operation data is achieved through the following steps: collecting alarm data uploaded by the BMS; the alarm data includes abnormal temperature, individual cell voltage exceeding limits, and total voltage fluctuation; obtaining BMS operation data through a monitoring platform; the operation data includes state of charge (SOC), state of health (SOH), highest / lowest individual cell voltage and number, and highest / lowest individual cell temperature and number.

[0007] In one embodiment, the real-time acquisition of energy data to construct an energy data pool is achieved through the following steps: A data acquisition terminal is set up to collect energy data in real time; the photovoltaic power station data includes irradiance and actual output; the wind farm data includes wind speed and wind direction; the power grid data includes power supply capacity and real-time electricity price; the energy storage system data includes SOC and charging / discharging power; the load demand data includes real-time power consumption and electricity consumption periods; based on the energy data, the power generation efficiency of the photovoltaic power station, the actual output of the wind farm, the peak-valley time division of the power grid, the remaining capacity and charging / discharging efficiency of the energy storage system, and the priority distribution of load demand are obtained, and the data is aggregated through a standardized interface to construct an energy data pool.

[0008] In one embodiment, the original dataset is cleaned, standardized, and feature extracted to obtain feature variables strongly correlated with battery failures. A training dataset is then constructed by combining these feature variables with historical failure records. This is achieved through the following steps: outliers in the original dataset are removed, and missing data is filled using the mean. The original dataset is then standardized using normalization. Outliers include voltage and temperature values ​​exceeding a set range. Feature variables strongly correlated with battery failures are selected from the processed original dataset and combined with historical failure records to construct a training dataset. These feature variables include extreme values ​​of single-cell voltage and temperature change rate.

[0009] In one embodiment, a fault warning model is constructed using a CNN-LSTM fusion deep learning algorithm. The trained fault warning model is obtained by training the model on the training dataset, and prediction results are obtained from real-time collected operational data through the following steps: The fault warning model is constructed based on a CNN-LSTM fusion deep learning algorithm; the CNN is used to analyze the spatial correlation in battery operational data through local feature extraction; the LSTM is used to model the dynamic changes in temperature and voltage through temporal trend capture; using multi-dimensional features from the training dataset as input and historical fault records as labels, the fault warning model is trained through hyperparameter tuning and cross-validation; the hyperparameter tuning includes optimizing the learning rate, number of iterations, and number of network layers; the real-time collected and preprocessed operational data is input into the trained fault warning model to obtain predicted results for temperature and voltage trends. If the predicted results exceed a preset safety threshold, a warning is triggered and a predictive alert is sent; the predictive alert includes the warning type, risk level, and emergency handling suggestions.

[0010] In one embodiment, the construction and iterative training of an energy dispatch model using reinforcement learning and multi-objective optimization algorithms are achieved through the following steps: an energy dispatch model is constructed based on real-time supply and demand data, environmental forecasts, and electricity pricing policies in the energy data pool using reinforcement learning and multi-objective optimization algorithms, and the trained energy dispatch model is obtained by iteratively training the energy dispatch model by optimizing its economic efficiency, environmental friendliness, and reliability.

[0011] In one embodiment, the optimal energy dispatch strategy is generated based on the energy data pool and the prediction results using the trained energy dispatch model through the following steps: the real-time energy data pool and the prediction results of the fault early warning model are input into the trained energy dispatch model, so that the energy dispatch model dynamically calculates the optimal energy dispatch strategy according to the current state; the energy dispatch strategy includes the timing of energy storage charging and discharging, the proportion of new energy output allocation, and the grid supplementation strategy.

[0012] According to one aspect of the present invention, an energy early warning and dispatching device includes: a data acquisition module for acquiring alarm data, BMS and energy equipment operation data to construct an original dataset, and acquiring energy data in real time to construct an energy data pool; the energy data includes photovoltaic power plant data, wind farm data, power grid data, energy storage system data, and load demand data; a data processing module for cleaning, standardizing and extracting features from the original dataset to obtain feature variables strongly correlated with battery faults and constructing a training dataset in combination with historical fault records; the data cleaning includes removing outliers and filling in missing data; a fault early warning module for constructing a fault early warning model using a deep learning algorithm fused with CNN-LSTM, training the fault early warning model based on the training dataset to obtain a trained fault early warning model, and predicting the real-time acquired operation data to obtain a prediction result; the fault early warning model is used to capture the temporal trends of battery temperature and voltage; and an energy dispatching module for constructing an energy dispatching model using reinforcement learning and multi-objective optimization algorithms and iteratively training it, generating an optimal energy dispatching strategy based on the energy data pool and the prediction result using the trained energy dispatching model, and converting the optimal energy dispatching strategy into control commands for energy early warning and dispatching.

[0013] According to one aspect of the present invention, an electronic device includes at least one processor and at least one memory, wherein computer-readable instructions are stored on the memory; the computer-readable instructions are executed by one or more of the processors, causing the electronic device to implement the energy warning and scheduling method as described above.

[0014] According to one aspect of the present invention, a storage medium has computer-readable instructions stored thereon, which are executed by one or more processors to implement the energy warning and scheduling method as described above.

[0015] The beneficial effects of the technical solution provided by this invention are: In the above technical solution, this invention constructs a raw dataset by collecting alarm data, BMS, and energy equipment operation data, and forms an energy data pool by collecting multi-source energy data in real time, comprehensively integrating multi-dimensional information of the new energy system. Based on this, the raw dataset is cleaned, standardized, and its features are extracted. A training dataset is constructed by combining historical fault records, providing high-quality data support for model training. Subsequently, a fault early warning model is constructed using a CNN-LSTM fusion algorithm to accurately capture the temporal trends of battery temperature and voltage, achieving predictive early warning. Simultaneously, an energy scheduling model is constructed using reinforcement learning and multi-objective optimization algorithms to generate the optimal energy allocation strategy based on real-time data and dynamic changes. Finally, the optimal strategy is transformed into control commands to achieve dynamic energy scheduling and early warning. This invention not only improves the accuracy and timeliness of fault early warning and reduces the false alarm rate, but also significantly improves the absorption rate of new energy and the economic efficiency of system operation, effectively solving problems such as delayed early warning and unintelligent scheduling in existing technologies, providing a strong guarantee for the safe and stable operation of new energy systems. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating an energy early warning and dispatching method according to an exemplary embodiment; Figure 2 This is a flowchart illustrating the operation of a fault early warning model in an application scenario; Figure 3 yes Figure 2 Flowcharts for fault warning execution in corresponding application scenarios; Figure 4 This is a block diagram of an energy early warning and dispatching device according to an exemplary embodiment; Figure 5 This is a hardware structure diagram of an electronic device according to an exemplary embodiment; Figure 6 This is a block diagram illustrating an electronic device according to an exemplary embodiment. Detailed Implementation

[0018] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0019] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this disclosure means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0020] This invention provides an energy early warning and scheduling method. By integrating multi-source renewable energy data and multi-dimensional information from a BMS (Battery Management System), and combining it with artificial intelligence algorithms, it achieves predictive early warning of faults and dynamic optimal energy allocation. This solves the problems of delayed early warning, high false alarm rate, unintelligent energy scheduling, and low energy absorption rate associated with traditional methods, thus improving system security and economy. This energy early warning and scheduling method is applicable to energy early warning and scheduling devices, which can be electronic devices. The energy early warning and scheduling method in this invention can be applied to various scenarios, such as energy early warning and scheduling.

[0021] Please see Figure 1 This invention provides an energy early warning and scheduling method, which is applicable to electronic devices.

[0022] In the following method embodiments, for ease of description, the execution subject of each step of the method is an electronic device, but this does not constitute a specific limitation.

[0023] like Figure 1 As shown, the method may include the following steps: Step 110: Collect alarm data, BMS and energy equipment operation data to build the original dataset, and collect energy data in real time to build the energy data pool.

[0024] One possible implementation involves collecting alarm data uploaded by the BMS and obtaining BMS operation data through a monitoring platform.

[0025] The energy data includes photovoltaic power plant data, wind farm data, power grid data, energy storage system data, load demand data, etc. The alarm data includes abnormal temperature, over-limit individual voltage, total voltage fluctuation, etc. The operation data includes state of charge (SOC), state of health (SOH), highest / lowest individual voltage and number, highest / lowest individual temperature and number, etc., none of which are specified here.

[0026] In one possible implementation, a data acquisition terminal is set up to collect energy data in real time. Based on the energy data, the power generation efficiency of photovoltaic power plants, the actual output of wind farms, the peak and valley periods of the power grid, the remaining capacity and charging and discharging efficiency of energy storage systems, and the priority distribution of load demand are obtained. The data is then aggregated through standardized interfaces to build an energy data pool.

[0027] Among them, photovoltaic power plant data includes irradiance, actual output, etc.; wind farm data includes wind speed, wind direction, etc.; power grid data includes power supply capacity, real-time electricity price, etc.; energy storage system data includes SOC, charging and discharging power, etc.; and load demand data includes real-time power consumption, electricity consumption period, etc., none of which are specified here.

[0028] Specifically, alarm data uploaded by the BMS is collected through the Energy Management System (EMS), including but not limited to temperature anomaly alarms, individual cell voltage over-limit alarms, and total voltage fluctuation alarms. Simultaneously, core operational data of the BMS, such as State of Charge (SOC), State of Health (SOH), highest / lowest individual cell voltage and number, and highest / lowest individual cell temperature and number, are obtained through a monitoring platform to form a raw dataset for fault early warning. Furthermore, a multi-source data acquisition terminal is built to collect real-time data from photovoltaic power plants (irradiance, actual output, power generation efficiency), wind farms (wind speed, wind direction, actual output), power grids (supply capacity, real-time electricity price, peak-valley time division), energy storage systems (SOC, charging / discharging power, remaining capacity, charging / discharging efficiency), and load demand (real-time power consumption of various electrical devices, electricity consumption time distribution, load priority), constructing a multi-source new energy data pool.

[0029] In the above process, the embodiments of the present invention ensure the integrity of the data by comprehensively collecting multi-dimensional data, effectively solving the problem of single data in traditional methods, and improving the accuracy of fault early warning and the scientific nature of energy dispatch.

[0030] Step 120: Perform data cleaning, standardization and feature extraction on the original dataset to obtain feature variables that are strongly correlated with battery failures, and construct a training dataset by combining them with historical failure records.

[0031] One possible implementation involves removing outliers from the original dataset and filling in missing data using the mean. The original dataset is then standardized using normalization. Feature variables strongly correlated with battery failures are selected from the processed original dataset, and a training dataset is constructed by combining this with historical failure records.

[0032] Data cleaning includes removing outliers and filling in missing data. Outliers include voltage and temperature that exceed the set range, and characteristic variables include extreme values ​​of individual voltages and the rate of temperature change, etc., none of which are specified here.

[0033] Specifically, the original dataset is cleaned to remove outliers (such as voltage and temperature exceeding set ranges) and missing data is filled (using methods such as mean imputation). Then, standardization is performed to unify data units, facilitating subsequent analysis. Finally, feature extraction techniques are used to select core feature variables strongly correlated with battery faults, such as extreme values ​​of single-cell voltage and temperature change rates. Combined with historical fault records, a standardized training dataset is constructed. In the above process, this embodiment of the invention ensures high-quality data through a preprocessing workflow, providing a reliable data foundation for subsequent model training. This effectively solves the problem of insufficient preprocessing in traditional methods, improving the accuracy of fault warnings and the scientific nature of energy dispatching.

[0034] Step 130: A fault warning model is constructed using a deep learning algorithm that combines CNN and LSTM. The fault warning model is trained based on the training dataset to obtain a trained fault warning model, and the prediction result is obtained by predicting the real-time collected running data.

[0035] One possible implementation is to use a deep learning algorithm architecture based on the fusion of CNN and LSTM to build a fault warning model. The model takes multi-dimensional features from the training dataset as input and historical fault records as labels. The model is trained through hyperparameter tuning and cross-validation to obtain a trained fault warning model. The real-time collected and preprocessed running data is then input into the trained fault warning model to obtain the predicted results of temperature and voltage trends. If the predicted results exceed the preset safety threshold, an early warning is triggered and a predictive reminder is sent.

[0036] Among them, the fault warning model is used to capture the temporal trends of battery temperature and voltage; CNN is used to analyze the spatial correlation in battery operation data through local feature extraction capabilities; LSTM is used to model the dynamic changes of temperature and voltage through temporal trend capture capabilities; hyperparameter tuning includes optimizing the learning rate, number of iterations, number of network layers, etc.; predictive alerts include warning type, risk level and emergency handling suggestions, etc., none of which are limited here.

[0037] Specifically, a fault prediction model is constructed based on a deep learning algorithm architecture that integrates CNN and LSTM. The CNN module is responsible for extracting spatial correlation features from battery operation data, while the LSTM module captures the dynamic changes in time-series data such as temperature and voltage. Using multi-dimensional features from a standardized training dataset as input and historical fault records as labels, the fault prediction model is trained through hyperparameter tuning (such as optimizing the learning rate, number of iterations, and number of network layers) and cross-validation techniques to improve its prediction accuracy and robustness.

[0038] In the above process, this embodiment of the invention employs a CNN-LSTM fusion architecture, fully utilizing the local feature extraction capability of CNN and the temporal trend capture capability of LSTM to achieve accurate prediction of battery temperature and voltage trends. This step effectively solves the problems of lag and high false alarm rate in traditional early warning methods, significantly improving the advance warning and accuracy of fault warnings.

[0039] Step 140: An energy dispatching model is constructed using reinforcement learning and multi-objective optimization algorithms and iteratively trained. The trained energy dispatching model is used to generate the optimal energy dispatching strategy based on the energy data pool and prediction results. The optimal energy dispatching strategy is then converted into control commands for energy early warning and dispatching.

[0040] In one possible implementation, an energy dispatch model is constructed using reinforcement learning and multi-objective optimization algorithms based on real-time supply and demand data, environmental forecasts, and electricity pricing policies from an energy data pool. The trained energy dispatch model is then obtained by iteratively training the model to optimize its economic, environmental, and reliability aspects.

[0041] In one possible implementation, the real-time energy data pool and the prediction results of the fault early warning model are input into the trained energy dispatch model, so that the energy dispatch model can dynamically calculate the optimal energy dispatch strategy based on the current state.

[0042] The energy dispatch strategy includes the timing of energy storage charging and discharging, the proportion of new energy output allocation, and grid supplementation strategies, etc., which are not specified here.

[0043] Specifically, based on real-time supply and demand data, environmental forecast data (such as 24-hour irradiance and wind speed forecasts) and electricity pricing policies from a multi-source renewable energy data pool, an energy dispatch model is constructed using reinforcement learning and multi-objective optimization algorithms. This model aims to simultaneously optimize economic efficiency (reducing operating costs), environmental friendliness (improving clean energy consumption), and reliability (ensuring continuous power supply). By continuously inputting new real-time data, the energy dispatch model is iteratively trained, and its parameters are adjusted to adapt to the constantly changing energy supply and demand environment and electricity pricing policies, ensuring that the model always remains in an optimal state.

[0044] Furthermore, the real-time energy data pool and the prediction results from the fault early warning model are input into the trained energy dispatch model. The model dynamically calculates the optimal energy dispatch strategy based on the current state, including the timing of energy storage charging and discharging, the allocation ratio of renewable energy output, and grid supplementation strategies. Simultaneously, if the prediction results exceed a preset safety threshold, an early warning mechanism is triggered, sending predictive alerts to users via platform push notifications and app notifications. The model monitors the operational status and strategy execution effectiveness of each link in real time. If sudden changes in renewable energy output or load demand fluctuations occur, the model automatically recalculates the optimal strategy and dynamically adjusts dispatch instructions to ensure that the energy source, grid, load, and storage operate in an optimal coordinated state.

[0045] In the above process, this invention, by introducing reinforcement learning and multi-objective optimization algorithms, constructs an energy dispatch model capable of dynamically adapting to environmental changes. This effectively solves the problems of traditional dispatch methods lacking intelligent coordination mechanisms and having low renewable energy absorption rates, achieving optimal energy allocation and efficient utilization. Through a comprehensive dispatch and early warning execution mechanism, it realizes intelligent management of the energy system and proactive avoidance of fault risks. This not only improves the system's safety and stability but also significantly reduces operating and maintenance costs, providing users with a more economical and reliable energy solution.

[0046] Through the above process, this invention effectively solves the problems of delayed early warning and unintelligent scheduling in traditional methods by comprehensively collecting and preprocessing multi-dimensional data, constructing a high-precision fault early warning model, designing an intelligent energy dispatching model, and implementing a comprehensive dispatching and early warning execution mechanism. This method not only improves the safety and stability of new energy systems but also promotes the efficient utilization and sustainable development of clean energy.

[0047] In one application scenario, the energy early warning and scheduling method of the present invention is used for energy early warning and scheduling.

[0048] like Figure 2 As shown, the following steps may be included: Step S1: EMS device data acquisition and MQTT protocol transmission.

[0049] Specifically, EMS (Energy Management System) devices are deployed at the operating sites of energy equipment. These devices are equipped with high-precision sensors that can collect various energy-related data in real time and accurately. For example, for battery equipment, detailed operating parameters such as temperature, voltage, and current of individual battery cells and the entire system are collected, along with information such as the number of charge-discharge cycles and capacity changes. The collected data is initially processed in a specific format and then transmitted via the MQTT protocol. The MQTT protocol uses a publish-subscribe model, with the EMS device acting as the publisher, publishing the collected data to a specified topic, and the subsequent data processing server acting as a subscriber, subscribing to the relevant topic to receive data.

[0050] Furthermore, the lightweight nature of the MQTT protocol enables it to transmit data efficiently even with limited network bandwidth, its low power consumption makes it suitable for monitoring energy equipment that operates for extended periods, and its high reliability ensures that data is not easily lost or corrupted during transmission, making it adaptable to complex industrial network environments.

[0051] In the above process, the embodiments of the present invention accurately and comprehensively collect raw data from energy equipment through EMS equipment, and efficiently and stably transmit data using the MQTT protocol, providing a rich and reliable data source for subsequent data processing and analysis.

[0052] Step S2: Raw data rule filtering and data cleaning.

[0053] Specifically, after receiving the raw data transmitted via the MQTT protocol, the data processing server first performs rule-based filtering. Data is filtered according to a pre-defined rule system. For example, for temperature data, a reasonable temperature range is set; if the collected temperature value exceeds this range, it is considered abnormal data and discarded. For voltage data, its format is checked for compliance with standards; incorrect formats are also filtered out as invalid data. After rule-based filtering, some outliers and missing values ​​may still exist in the data. At this point, data cleaning is performed. For outliers, statistical methods are used for identification and processing, such as using box plots to determine anomaly thresholds. Data exceeding the threshold is considered outliers and corrected based on the average or median of surrounding data. For missing values, methods such as mean imputation and regression prediction imputation are used to fill in the missing values ​​based on the data's characteristics and correlations.

[0054] Furthermore, rule-based filtering can quickly remove data that clearly does not meet the requirements, reducing the amount of data to be processed and improving processing efficiency; data cleaning further optimizes data quality, ensures data integrity and accuracy, and makes the data more in line with the requirements of subsequent model analysis.

[0055] In the above process, the embodiments of the present invention effectively removed noise and errors from the original data through rule filtering and data cleaning, laying a solid foundation for subsequent accurate analysis.

[0056] Step S3: Analysis of the large early warning model fused with standardized data and CNN-LSTM.

[0057] Specifically, the cleaned data undergoes standardization. Since different types of data have different units and magnitudes—for example, temperature data is in degrees Celsius and voltage data is in volts—directly using these data for model analysis can lead to some features having an excessively large or small impact on the model. Therefore, the Z-score standardization method is used to transform the data into standard normal distribution data with a mean of 0 and a standard deviation of 1, making different features comparable. The standardized data is then input into the CNN-LSTM fusion early warning model.

[0058] Furthermore, the CNN part performs sliding convolution operations on the data using convolutional kernels to extract local features, such as identifying local patterns of voltage fluctuations in battery voltage data. The LSTM part utilizes its special gating mechanism to model the temporal features of the data, enabling it to remember long-term historical information and capture the trends of temperature, voltage, and other data over time. After the two are combined, the local features extracted by the CNN serve as input to the LSTM, which further analyzes the evolution of these features over time, thus providing a more comprehensive and in-depth analysis of the data. The model performs a comprehensive analysis of the input data and outputs temperature and voltage trend prediction results, such as predicting the rising or falling trend of battery temperature and the range of voltage fluctuations in the future.

[0059] In the above process, the embodiments of the present invention eliminate the influence of data dimensions through data standardization, and fully leverage the advantages of the two algorithms by utilizing the CNN-LSTM fusion model, thereby improving the accuracy of predicting the operating status of energy equipment.

[0060] Step S4: Compare and judge with the preset safety threshold.

[0061] Specifically, the model's output of temperature and voltage trend predictions is compared with preset safety thresholds. These safety thresholds are set based on factors such as the characteristics of the energy equipment, operating specifications, and historical experience. For example, for battery temperature, a maximum safe temperature threshold is set; if the predicted battery temperature trend indicates it will exceed this threshold, it is considered a risk of overheating. For voltage, upper and lower limits of the normal fluctuation range are set; if the predicted voltage trend exceeds these limits, it is considered an abnormal voltage. The judgment process is implemented through a specific algorithm program that compares the prediction results with the safety thresholds and outputs a judgment signal based on the comparison result.

[0062] Furthermore, by comparing with preset safety thresholds, abnormal operating conditions of energy equipment can be quickly and accurately identified, providing a basis for timely early warning measures.

[0063] In the above process, the embodiments of the present invention effectively improve the ability to identify equipment failure risks by scientifically and reasonably setting safety thresholds and using programs for accurate comparison and judgment.

[0064] Step S5: The system triggers an alert and notifies the user through multiple channels.

[0065] Specifically, when the predicted result exceeds the safety threshold, the system immediately triggers an early warning mechanism. This mechanism notifies users through multiple channels by calling relevant notification interfaces. For example, via SMS, it sends a message to the user's mobile phone containing the warning type (e.g., high temperature warning, abnormal voltage warning), risk level (classified as general, severe, emergency, etc. based on the severity of the anomaly), and emergency handling suggestions (e.g., stop using the equipment, check the equipment, contact professional repair personnel, etc.); via APP push interface, it pops up a warning notification on the relevant APP installed by the user, allowing the user to view detailed warning information and handling instructions; simultaneously, the warning information is also prominently displayed on the monitoring platform interface to remind platform operators.

[0066] Furthermore, multi-channel notification methods can ensure that users receive timely warning information regardless of the scenario, enabling them to quickly understand the abnormal status of the device and take corresponding measures according to emergency handling suggestions to reduce potential risks.

[0067] In the above process, the embodiments of the present invention improve the user's response speed to equipment failure by triggering early warnings and notifying through multiple channels, and minimize the potential losses caused by equipment failure.

[0068] Step S6: Continuous monitoring.

[0069] Specifically, after triggering an alert and notifying the user, the system does not stop operating but continues to collect and monitor data from the energy equipment via the EMS device. The EMS device continuously acquires the equipment's operating data at a preset collection frequency and transmits the data to a data processing server for analysis. The server tracks the equipment's operating status in real time and updates prediction results and judgment information based on newly collected data.

[0070] Furthermore, continuous monitoring can promptly detect changes in equipment status, such as whether the equipment has returned to normal operation after emergency measures have been taken, or whether the abnormal situation has worsened. This provides the latest data support for subsequent processing and decision-making, ensuring that the system can adjust early warning and handling strategies in a timely manner based on the actual status of the equipment.

[0071] In the above process, the embodiments of the present invention achieve full-process tracking of the operating status of energy equipment through continuous monitoring, thus ensuring the safety of the equipment throughout the entire operation process.

[0072] Through the above process, this embodiment of the invention begins with the accurate collection of data from EMS equipment and efficient transmission using the MQTT protocol. After rigorous rule filtering and data cleaning to optimize data quality, it then achieves accurate prediction through standardized processing and analysis using a CNN-LSTM fusion model. Next, it compares the data with safety thresholds to determine the equipment status, triggering early warnings and notifying users through multiple channels. Finally, continuous monitoring ensures equipment safety. Through this series of rigorous steps, the system can promptly detect potential fault risks in energy equipment and take effective measures for early warning and handling, effectively ensuring the safe and stable operation of energy equipment, improving system reliability and security, and providing strong technical support for energy management.

[0073] like Figure 3 As shown, the execution of a fault warning may include the following steps: Step S1: Data Acquisition.

[0074] Specifically, various sensors and monitoring devices deployed in the energy system collect real-time operational data of the battery and related equipment, including key parameters such as temperature, voltage, and current. This data comes from a wide range of sources, covering different parts of the battery and their operating states, ensuring a comprehensive understanding of the system's operational status.

[0075] Furthermore, the accuracy and timeliness of data acquisition are the foundation for subsequent processes. High-precision sensors and stable acquisition frequencies can provide reliable data support for subsequent analysis.

[0076] In the above process, the embodiments of the present invention acquire raw operating information of the energy system through multi-faceted data collection, providing rich data materials for fault early warning.

[0077] Step S2: Data preprocessing (cleaning, standardization, feature extraction).

[0078] Specifically, the process begins with data cleaning, where the raw data is examined to remove outliers and noisy data. For example, if temperature data exceeds the normal operating range of the battery, or if voltage data fluctuates drastically and deviates from normal patterns, these are considered outliers and removed. Next, data standardization is performed. Since different parameters have different dimensions and magnitudes, appropriate methods are used to unify the data to the same scale, facilitating subsequent analysis and processing. Finally, feature extraction is conducted, selecting features strongly correlated with battery failures from the cleaned and standardized data, such as temperature change rate and voltage extreme values. Data preprocessing improves data quality and usability, making subsequent model training more accurate and efficient.

[0079] In the above process, the embodiments of the present invention optimize the data structure through data preprocessing, laying the foundation for building an effective early warning model.

[0080] Step S3: Generate the training dataset.

[0081] Specifically, the preprocessed data is organized and labeled, and combined with historical fault records to form a training dataset. The historical fault records contain various data features of past battery failures. Combining these with the current preprocessed data makes the training dataset more representative and practical.

[0082] Furthermore, the quality of the training dataset directly affects the training effect of large models. Rich and accurate training data enables models to better learn the characteristics and patterns of battery failures.

[0083] In the above process, the embodiments of the present invention provide targeted learning samples for model training by reasonably constructing a training dataset.

[0084] Step S4: Large model construction and training.

[0085] Specifically, a large-scale model is constructed using a deep learning algorithm that combines CNN and LSTM. CNN (Convolutional Neural Network) can extract local features from data, while LSTM (Long Short-Term Memory Network) can capture temporal features of data. Combining the two fully leverages their respective advantages. The training dataset is input into the large-scale model, and by continuously adjusting the model's parameters, the model learns the features and patterns in the data, thereby achieving accurate predictions of temperature and voltage trends. The construction and training of this large-scale model is a complex process requiring substantial data and computational resources. Through multiple iterations and optimizations, the model's prediction accuracy and stability are improved. By constructing and training this large-scale model, a powerful analytical tool for fault early warning is provided.

[0086] Step S5: Output the predicted results of temperature and voltage trends.

[0087] Specifically, the trained large model analyzes and processes the new data, outputting trend predictions for temperature and voltage. These predictions reflect the battery's operating status over a future period, providing a basis for assessing the risk of battery failure.

[0088] Furthermore, the accuracy of the prediction results is crucial. The reliability of the predictions is improved by continuously optimizing the model and updating the data. By outputting prediction results through a large model, the battery's operating status can be predicted in advance.

[0089] Step S6: Compare with the preset safety threshold.

[0090] Specifically, the predicted temperature and voltage trends output by the model are compared with preset safety thresholds. These safety thresholds are determined based on factors such as the battery's rated parameters, operating specifications, and historical experience. If the predicted results exceed the safety thresholds, it indicates a potential risk of battery failure; if they do not exceed them, it indicates that the battery is operating normally. Setting reasonable safety thresholds is crucial for accurately assessing failure risk and requires adjustment and optimization based on actual conditions. By comparing the predictions with the safety thresholds, the presence of potential battery failures can be quickly determined.

[0091] Step S7: Determine whether the threshold is exceeded.

[0092] Specifically, a judgment is made based on the comparison results. If the predicted result exceeds the preset safety threshold, the process proceeds to step S8; otherwise, it returns to the data acquisition stage to continue monitoring the battery's operating status. The judgment process must be accurate to ensure timely detection of problems and appropriate measures are taken. Through this judgment stage, real-time monitoring and timely response to battery failure risks are achieved.

[0093] Step S8: Activate multi-channel early warning notification.

[0094] Specifically, when the predicted result exceeds the safety threshold, the system activates a multi-channel early warning notification mechanism. Early warning information is sent to relevant personnel through various means such as SMS, app push notifications, and platform alerts, indicating the warning type, risk level, and emergency response suggestions. For example, the warning type could be an overheating warning or an abnormal voltage warning; the risk level could be categorized as general, severe, or emergency; and the emergency response suggestions could include specific measures such as stopping the use of the equipment and conducting inspections and repairs.

[0095] Furthermore, multi-channel early warning notifications ensure that relevant personnel receive warning information in a timely manner and take effective countermeasures to reduce losses caused by the failure. Multiple channels of early warning notifications improve the timeliness and effectiveness of failure handling.

[0096] Step S9: Return to continuous monitoring (if the threshold is not exceeded).

[0097] Specifically, when the prediction result does not exceed the safety threshold, the system returns to the data acquisition stage to continue real-time monitoring of the battery's operating status. Continuous monitoring can promptly detect changes in the battery's operating status, ensuring the safe and stable operation of the system.

[0098] Furthermore, continuous monitoring is a cyclical process, constantly repeating steps such as data collection, analysis, and judgment to achieve full-process tracking of the battery's operating status. Through continuous monitoring, the long-term stable operation of the energy system is ensured.

[0099] Through the above process, this embodiment of the invention forms a complete fault early warning system, starting from data acquisition, through data preprocessing, training dataset generation, large model construction and training, prediction result output, comparison with safety thresholds, determination of whether the threshold is exceeded, and finally, early warning notification or continuous monitoring. This process enables timely detection of potential battery failure risks in energy systems and notifies relevant personnel through multiple channels to take countermeasures. Simultaneously, it continuously monitors the system's operating status, effectively ensuring the safe and stable operation of the energy system and improving its reliability and safety.

[0100] The following are embodiments of the apparatus of the present invention, which can be used to execute the energy early warning and scheduling method involved in the present invention. For details not disclosed in the embodiments of the apparatus of the present invention, please refer to the method embodiments of the energy early warning and scheduling method involved in the present invention.

[0101] Please see Figure 4 This invention provides an energy early warning and dispatching device 800.

[0102] The energy early warning and dispatching device 800 includes, but is not limited to: a data acquisition module 810, a data processing module 830, a fault early warning module 850, and an energy dispatching module 870.

[0103] Among them, the data acquisition module 810 is used to collect alarm data, BMS and energy equipment operation data to build the original dataset, and to collect energy data in real time to build the energy data pool; the energy data includes photovoltaic power plant data, wind farm data, power grid data, energy storage system data and load demand data.

[0104] The data processing module 830 is used to perform data cleaning, standardization and feature extraction on the original dataset to obtain feature variables that are strongly correlated with battery failure and to construct a training dataset by combining historical failure records; data cleaning includes removing outliers and filling in missing data.

[0105] The fault warning module 850 is used to build a fault warning model using a deep learning algorithm that combines CNN and LSTM. It trains the fault warning model based on the training dataset to obtain a trained fault warning model, and predicts the results based on the real-time collected operating data. The fault warning model is used to capture the time-series trends of battery temperature and voltage.

[0106] The energy dispatch module 870 is used to construct an energy dispatch model using reinforcement learning and multi-objective optimization algorithms and perform iterative training. The trained energy dispatch model is used to generate the optimal energy dispatch strategy based on the energy data pool and prediction results. The optimal energy dispatch strategy is then converted into control commands for energy early warning and dispatch.

[0107] It should be noted that the energy early warning and dispatch provided in the above embodiments are only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed. That is, the internal structure of the energy early warning and dispatch device will be divided into different functional modules to complete all or part of the functions described above.

[0108] Furthermore, the energy early warning and scheduling device and the energy early warning and scheduling method provided in the above embodiments belong to the same concept, and the specific way in which each module performs its operation has been described in detail in the method embodiments, and will not be repeated here.

[0109] Figure 5 A schematic diagram of the structure of an electronic device according to an exemplary embodiment is shown.

[0110] It should be noted that this electronic device is merely an example adapted to the present invention and should not be construed as providing any limitation on the scope of use of the present invention. Furthermore, this electronic device should not be interpreted as requiring or depending on having... Figure 5 One or more components of the exemplary electronic device 2000 shown.

[0111] The hardware structure of electronic devices 2000 can vary significantly due to differences in configuration or performance, such as... Figure 5 As shown, the electronic device 2000 includes: a power supply 210, an interface 230, at least one memory 250, and at least one central processing unit (CPU) 270.

[0112] Specifically, power supply 210 is used to provide operating voltage for various hardware devices on electronic device 2000.

[0113] Interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices. Of course, in other examples adapted to this invention, interface 230 may further include at least one serial-to-parallel conversion interface 233, at least one input / output interface 235, and at least one USB interface 237, etc. Figure 5 As shown, this does not constitute a specific limitation.

[0114] The memory 250 serves as a carrier for resource storage and can be a read-only memory, random access memory, disk, or optical disk, etc. The resources stored on it include the operating system 251, application programs 253, and data 255, etc., and the storage method can be temporary storage or permanent storage.

[0115] The operating system 251 is used to manage and control the various hardware devices and application programs 253 on the electronic device 2000, so as to enable the central processing unit 270 to perform calculations and processing on the massive data 255 in the memory 250. It can be Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0116] Application 253 is a computer-readable instruction based on operating system 251 that performs at least one specific task, and may include at least one module ( Figure 5 (Not shown), each module may contain computer-readable instructions for the electronic device 2000. For example, the energy early warning and dispatching device can be considered as an application program 253 deployed on the electronic device 2000.

[0117] Data 255 may be signal information, etc., and is stored in memory 250.

[0118] The central processing unit 270 may include one or more processors and is configured to communicate with the memory 250 via at least one communication bus to read computer-readable instructions stored in the memory 250, thereby enabling the computation and processing of massive amounts of data 255 in the memory 250. For example, an energy early warning and scheduling method may be implemented by the central processing unit 270 reading a series of computer-readable instructions stored in the memory 250.

[0119] Furthermore, the present invention can also be implemented through hardware circuits or a combination of hardware circuits and software. Therefore, the implementation of the present invention is not limited to any specific hardware circuit, software, or combination thereof.

[0120] Please see Figure 6 This invention provides an electronic device 2000, which may include: a desktop computer, a laptop computer, a server, etc., with sensor recognition capabilities.

[0121] exist Figure 6 In this context, the electronic device 2000 includes at least one processor 4001 and at least one memory 250.

[0122] The data interaction between the processor 4001 and the memory 250 can be achieved through at least one communication bus 4002. This communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 250. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0123] Optionally, the electronic device 2000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 2000 does not constitute a limitation on the embodiments of the present invention.

[0124] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0125] The memory 250 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program instructions or code in the form of instructions or data structures and accessible by the electronic device 2000, but not limited thereto.

[0126] The memory 250 stores computer-readable instructions, and the processor 4001 can read the computer-readable instructions stored in the memory 250 through the communication bus 4002.

[0127] The computer-readable instructions are executed by one or more processors 4001 to implement the energy warning and scheduling methods in the above embodiments.

[0128] Furthermore, this embodiment of the invention provides a storage medium storing computer-readable instructions, which are executed by one or more processors to implement the energy early warning and scheduling method described above.

[0129] This invention provides a computer program product, which includes computer-readable instructions stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, thereby enabling the electronic device to implement the energy warning and scheduling method described above.

[0130] Compared with related technologies, the beneficial effects of the present invention are: 1. This invention enables accurate early warning of energy equipment failures through multi-source data acquisition, comprehensive data preprocessing, and deep analysis using a CNN-LSTM fusion model. Multi-source data acquisition covers various operating parameters of batteries and related equipment, providing a rich information foundation for the model. Data cleaning, standardization, and feature extraction in data preprocessing optimize data quality, enabling the model to better learn data features. The CNN-LSTM fusion model combines the advantages of convolutional neural networks for extracting local features and long short-term memory networks for capturing temporal features, thereby accurately predicting temperature and voltage trends and promptly detecting potential faults.

[0131] 2. This invention possesses highly efficient fault response capabilities; through real-time comparison with preset safety thresholds and a multi-channel early warning notification mechanism, it rapidly compares model prediction results with safety thresholds. Once the threshold is exceeded, an early warning is immediately triggered, and relevant personnel are promptly notified via SMS, app push notifications, platform alerts, and other methods. Detailed warning types, risk levels, and emergency handling suggestions are provided, ensuring that relevant personnel can quickly understand the situation and take effective measures to reduce losses caused by the fault.

[0132] 3. This invention ensures the continuous and stable operation of the energy system through continuous monitoring. When the safety threshold is not exceeded, the system automatically returns to the data acquisition stage, continuously tracking and monitoring the operating status of the energy equipment, forming a cyclical monitoring system. This continuous monitoring can promptly detect changes in equipment operating status, ensuring the system remains in a safe and stable operating state and preventing system interruptions or anomalies caused by equipment failure.

[0133] 4. This invention features standardized and scientific data processing. Through rigorous operations in the data preprocessing stage, data cleaning removes outliers and noisy data, ensuring data accuracy and reliability; data standardization unifies the dimensions and magnitudes of different parameters, making the data comparable; feature extraction filters out characteristic variables strongly correlated with faults, improving data utilization efficiency. These standardized data processing steps provide high-quality data support for subsequent model training and predictive analysis, enhancing the performance of the entire early warning system.

[0134] 5. This invention can adapt to complex energy system environments; through multi-source data acquisition and flexible model analysis capabilities. Multi-source data acquisition can acquire data from different devices and operating scenarios, comprehensively reflecting the operating status of the energy system; the CNN-LSTM fusion model has strong learning and adaptability, capable of handling various complex data patterns and trends, thereby accurately predicting equipment failures under different operating conditions and environments, providing a guarantee for the stable operation of complex energy systems.

[0135] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0136] The above description is only a partial embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An energy early warning and dispatching method, characterized in that, The method includes: An initial dataset is constructed by collecting alarm data, BMS and energy equipment operation data, and an energy data pool is constructed by collecting energy data in real time; the energy data includes photovoltaic power plant data, wind farm data, power grid data, energy storage system data and load demand data; The original dataset is cleaned, standardized, and feature-extracted to obtain feature variables strongly correlated with battery failures. These features are then combined with historical failure records to construct a training dataset. The data cleaning process includes removing outliers and filling in missing data. A fault warning model is constructed using a deep learning algorithm that combines CNN and LSTM. The model is trained using the training dataset to obtain a trained fault warning model, and prediction results are obtained by predicting real-time collected operating data. The fault warning model is used to capture the time-series trends of battery temperature and voltage. An energy scheduling model is constructed using reinforcement learning and multi-objective optimization algorithms and iteratively trained. The trained energy scheduling model generates an optimal energy scheduling strategy based on the energy data pool and the prediction results. The optimal energy scheduling strategy is then converted into control commands for energy early warning and scheduling.

2. The energy early warning and dispatch method as described in claim 1, characterized in that, The collected alarm data, BMS, and energy equipment operation data form the original dataset, including: Collect alarm data uploaded by BMS; the alarm data includes abnormal temperature, over-limit individual cell voltage, and total voltage fluctuation; obtain BMS operation data through the monitoring platform; the operation data includes state of charge (SOC), state of health (SOH), highest / lowest individual cell voltage and number, and highest / lowest individual cell temperature and number.

3. The energy early warning and dispatch method as described in claim 1, characterized in that, The real-time acquisition of energy data to construct an energy data pool includes: A data acquisition terminal is set up to collect energy data in real time; the photovoltaic power station data includes irradiance and actual output; the wind farm data includes wind speed and wind direction; the power grid data includes power supply capacity and real-time electricity price; the energy storage system data includes SOC and charging / discharging power; and the load demand data includes real-time power consumption and electricity consumption period. Based on the energy data, the power generation efficiency of photovoltaic power plants, the actual output of wind farms, the peak and valley periods of the power grid, the remaining capacity and charging and discharging efficiency of energy storage systems, and the priority distribution of load demand are obtained. The data are then aggregated through standardized interfaces to build an energy data pool.

4. The energy early warning and dispatch method as described in claim 1, characterized in that, The process of cleaning, standardizing, and extracting features from the original dataset to obtain feature variables strongly correlated with battery failures, and then constructing a training dataset by combining these with historical failure records, includes: Outliers in the original dataset are removed, and missing data are filled using the mean. The original dataset is then standardized using normalization. Outliers include voltage and temperature values ​​that exceed a set range. Feature variables strongly correlated with battery failures are selected from the processed original dataset, and a training dataset is constructed by combining them with historical failure records; the feature variables include the extreme values ​​of single cell voltage and the rate of temperature change.

5. The energy early warning and dispatch method as described in claim 1, characterized in that, The fault warning model is constructed using a deep learning algorithm that integrates CNN and LSTM. The model is trained using the training dataset to obtain a trained fault warning model, and prediction results are obtained from real-time collected operational data. This includes: A fault warning model is based on a deep learning algorithm architecture that integrates CNN and LSTM. The CNN is used to analyze the spatial correlation in battery operation data through local feature extraction, and the LSTM is used to model the dynamic changes in temperature and voltage through time-series trend capture. Using the multi-dimensional features in the training dataset as input and historical fault records as labels, the fault warning model is trained through hyperparameter tuning and cross-validation to obtain a trained fault warning model; the hyperparameter tuning includes optimizing the learning rate, number of iterations, and number of network layers. The real-time collected and preprocessed operational data is input into the trained fault early warning model to obtain the predicted results of temperature and voltage trends. If the predicted results exceed the preset safety threshold, an early warning is triggered and a predictive reminder is sent. The predictive reminder includes the warning type, risk level, and emergency handling suggestions.

6. The energy early warning and dispatch method as described in claim 1, characterized in that, The process of constructing an energy dispatch model using reinforcement learning and multi-objective optimization algorithms and then iteratively training it includes: Based on real-time supply and demand data, environmental forecasts, and electricity pricing policies in the energy data pool, an energy dispatch model is constructed using reinforcement learning and multi-objective optimization algorithms. The energy dispatch model is then iteratively trained by optimizing economic efficiency, environmental friendliness, and reliability to obtain a well-trained energy dispatch model.

7. The energy early warning and dispatch method as described in claim 6, characterized in that, The step of generating an optimal energy scheduling strategy based on the trained energy scheduling model, the energy data pool, and the prediction results includes: The real-time energy data pool and the prediction results of the fault early warning model are input into the trained energy dispatch model, so that the energy dispatch model can dynamically calculate the optimal energy dispatch strategy based on the current state; the energy dispatch strategy includes the timing of energy storage charging and discharging, the proportion of new energy output allocation and the grid supplementation strategy.

8. An energy early warning and dispatching device, characterized in that, The device includes: The data acquisition module is used to collect alarm data, BMS and energy equipment operation data to build a raw dataset, and to collect energy data in real time to build an energy data pool; the energy data includes photovoltaic power plant data, wind farm data, power grid data, energy storage system data, and load demand data; The data processing module is used to perform data cleaning, standardization, and feature extraction on the original dataset to obtain feature variables that are strongly correlated with battery failures and to construct a training dataset by combining them with historical failure records; the data cleaning includes removing outliers and filling in missing data; The fault warning module is used to construct a fault warning model using a deep learning algorithm that combines CNN and LSTM. The fault warning model is trained based on the training dataset to obtain a trained fault warning model, and the prediction results are obtained by predicting the real-time collected running data. The fault warning model is used to capture the time-series trends of battery temperature and voltage. The energy scheduling module is used to construct an energy scheduling model using reinforcement learning and multi-objective optimization algorithms and perform iterative training. The trained energy scheduling model is used to generate an optimal energy scheduling strategy based on the energy data pool and the prediction results. The optimal energy scheduling strategy is then converted into control commands for energy early warning and scheduling.

9. An electronic device, characterized in that, include: At least one processor and at least one memory, wherein, The memory stores computer-readable instructions; The computer-readable instructions are executed by one or more of the processors, causing the electronic device to implement the energy warning and scheduling method as described in any one of claims 1 to 7.

10. A storage medium having computer-readable instructions stored thereon, characterized in that, The computer-readable instructions are executed by one or more processors to implement the energy early warning and scheduling method as described in any one of claims 1 to 7.

Images