Multi-modal large model-based energy big data intelligent system and design method

Abstract

The invention discloses an energy big data intelligent system based on a multi-modal large model and a design method, and relates to the technical field of large model application, the energy big data intelligent system comprises a visual interface for receiving user input and displaying an energy data analysis result; the data analysis module is configured to execute data analysis, energy efficiency optimization, prediction modeling and anomaly detection operation; configuration is used for constructing and training a multi-modal large model; the server is configured to perform user identity verification, authority management and data encryption; the management module is configured to manage data transmission; the multi-modal data acquisition module is configured to acquire multi-modal data from energy equipment and sensors. According to the energy big data intelligent system based on the multi-modal large model, through the multi-modal large model technology, energy data from different sources and various types can be comprehensively processed, and comprehensive analysis of the energy use condition is achieved. The system applies deep learning and machine learning algorithms to improve the accuracy and timeliness of data analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of large model application, and specifically to an energy big data intelligent system and design method based on a multimodal large model. Background Art

[0002] In today's digital age, energy big data intelligent systems are in a stage of rapid development. With the continuous maturity and popularization of artificial intelligence, machine learning and big data technologies, energy big data intelligent systems can better utilize these advanced technologies to achieve efficient data analysis and decision support. The emphasis of governments on energy management and the policy measures to promote the development of clean energy have prompted enterprises and organizations to adopt intelligent systems to improve energy efficiency and reduce carbon emissions. The proposal of global sustainable development goals requires the energy industry to manage resources more intelligently and efficiently, thus promoting the demand and development of energy big data intelligent systems. Energy management systems are gradually developing towards cross-domain integration, combining technologies such as the Internet of Things, cloud computing, and blockchain to build a smarter and more integrated energy big data management platform. The improvement of users' awareness of energy conservation and environmental protection and the increase in demand for data-driven decision-making have prompted energy management systems to become more intelligent and personalized to meet the needs of different users. Energy big data intelligent management systems are currently in a stage of vigorous development, driven by multiple factors such as technological progress, policy support, and sustainable development needs, bringing more possibilities and opportunities to the field of energy management.

[0003] However, there are still some problems in data fusion in the energy big data intelligent management system. The first is the data island problem. Energy big data comes from different sources and may be stored in different systems, resulting in data isolation and isolation, making it difficult to integrate and analyze. Second, the data quality is inconsistent. The quality of different data sources may vary, including missing data, noise interference, etc., which affects the accuracy and reliability of the analysis results. Third, cross-domain information integration is difficult. Energy big data involves a variety of data types, such as sensors, images, text, etc. These heterogeneous data need to be integrated into a unified platform for comprehensive analysis, but this is difficult. Fourth, complex correlation mining. There are complex correlations and interactions in the energy system. Traditional methods are difficult to capture these complexities, resulting in limitations in the analysis results.

[0004] In the above context, by applying multimodal big models, the energy big data intelligent management system can better overcome the current limitations, improve the ability of data integration, analysis and decision support, and promote the development of the energy management field towards a more intelligent and efficient direction. Summary of the invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is: when facing multimodal data, the existing energy big data intelligent management methods have the problems of difficulty in effectively associating data between different modes, inconsistent data standards, high data collection and cleaning costs, and untimely response to real-time data processing, as well as how to integrate multimodal data models to achieve accurate data prediction, energy efficiency optimization and anomaly detection optimization.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: an energy big data intelligent system based on a multimodal big model, comprising:

[0008] User interface layer, application layer, model layer, security layer, network layer and data collection layer;

[0009] The user interface layer is configured to receive user input and display a visual interface of energy data analysis results;

[0010] The application layer is configured to perform data analysis, energy efficiency optimization, predictive modeling and anomaly detection operations, and to achieve automated control of energy equipment through preset rules;

[0011] The model layer configuration is used to build and train a large multimodal model to handle tasks including multimodal data association, sentiment analysis, data mining, and energy consumption prediction;

[0012] The security layer is configured to perform user identity authentication, rights management, and data encryption to control data access rights and protect data security;

[0013] The network layer configuration is used to manage data transmission and support data communication between different devices and modules in the system;

[0014] The data acquisition layer is configured to collect multimodal data from energy devices and sensors, and to format, standardize and clean the collected data.

[0015] As a preferred solution of the energy big data intelligent system based on the multimodal big model described in the present invention, wherein: the user interface layer collects data from energy devices, sensors and third-party systems in real time through an application programming interface and a software development kit, and the data includes structured data and unstructured data;

[0016] Format and standardize the collected data and convert it into a unified structured form for subsequent processing;

[0017] Use interpolation methods to fill missing values ​​in the data, and use anomaly detection algorithms based on statistical analysis to remove invalid data to generate a cleaned data set;

[0018] The cleaned data set is transferred to the model layer to trigger data update and parameter adjustment of the model layer.

[0019] As a preferred solution of the energy big data intelligent system based on the multimodal big model described in the present invention, wherein: the application layer uses a multimodal data association module to match and associate sensor data, device log data and image data in time and space dimensions to generate a comprehensive energy data view for analysis;

[0020] Load and select an applicable deep learning model architecture according to business requirements, and process the comprehensive data view based on the model of the Transformer architecture;

[0021] During the model training process, the model weights are adjusted through the adaptive learning rate algorithm, overfitting is suppressed through the Dropout method, and the early stopping strategy is used to improve the training stability of the model;

[0022] The analysis results of the trained energy efficiency prediction model and anomaly detection model are output to the application layer for subsequent optimization and control operations.

[0023] As a preferred solution of the energy big data intelligent system based on the multimodal big model described in the present invention, wherein: the model layer analyzes the energy efficiency prediction results, generates optimization control instructions according to the predicted energy demand changes, and adjusts the operating parameters of the energy equipment;

[0024] Generate abnormal warning information based on abnormal detection results and convert the information into equipment adjustment instructions;

[0025] The generated control instructions are transmitted to the target energy device through the network layer to adjust the operating status of the device.

[0026] As a preferred solution of the energy big data intelligent system based on multimodal big model described in the present invention, wherein: the application layer sends a user identity authentication request to the security layer, and the security layer verifies the user identity by means of multi-factor authentication;

[0027] After verification, the security layer performs AES-256 encryption on the control instructions and transmits the encrypted control instructions to the target device through the network layer to ensure that the data is not tampered with during transmission.

[0028] As a preferred solution of the energy big data intelligent system based on the multimodal big model described in the present invention, wherein: the network layer establishes a data transmission channel based on the TCP / IP or MQTT protocol between the data acquisition layer, the model layer, the application layer and the user interface layer to realize data distribution between the modules;

[0029] When the data flow increases to the set threshold, the network layer automatically starts the load balancing function and dynamically adjusts the data transmission bandwidth of each module;

[0030] The data transmission path, time and content summary are recorded in the log to facilitate tracing analysis when a system failure occurs.

[0031] As a preferred solution of the energy big data intelligent system based on the multimodal big model described in the present invention, wherein: the data acquisition layer receives and displays the energy efficiency optimization plan and abnormal warning information transmitted by the application layer at the user interface layer, and the display method includes charts and dashboards, which is convenient for users to view energy usage;

[0032] The user interface layer receives feedback information input by the user, allowing the user to manually adjust the control instructions generated by the system or provide operation suggestions;

[0033] The user feedback information is transmitted to the application layer through the network layer. The application layer adjusts the control parameters according to the user feedback, generates updated control instructions and sends them to the energy device.

[0034] Another object of the present invention is to provide a design method for an energy big data intelligent system based on a multimodal large model, which solves the problems in the existing energy data intelligent management system that different modal data are difficult to process uniformly, data quality differences are high, it is difficult to accurately predict energy consumption needs and respond to abnormal situations in a timely manner.

[0035] To solve the above technical problems, the present invention provides the following technical solutions: A design method for an energy big data intelligent system based on a multimodal big model, comprising: a data acquisition layer collects multimodal data from energy devices and sensors in real time, performs unified formatting and cleaning operations, and converts them into standardized comprehensive energy data views;

[0036] The model layer receives comprehensive data views, integrates multimodal data based on a deep learning framework to build an analysis model, and performs training to enable the model to have energy efficiency prediction and anomaly detection capabilities;

[0037] The application layer receives the output of the analysis model, converts energy efficiency prediction and abnormal information into control instructions, and transmits them to the target device through the security layer to achieve intelligent regulation of the energy system;

[0038] The collection layer dynamically obtains new data and, based on user feedback, updates the parameters of the model layer and the control logic of the application layer.

[0039] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the energy big data intelligent method based on a multimodal large model as described above are implemented.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the energy big data intelligent method based on a multimodal big model as described above.

[0041] Beneficial effects of the present invention: The energy big data intelligent system based on multimodal big models provided by the present invention can comprehensively process energy data from different sources and various types through multimodal big model technology to achieve a comprehensive analysis of energy usage. The system applies deep learning and machine learning algorithms, and has the ability to perform energy demand forecasts, identify energy waste, and detect abnormal usage patterns, thereby improving the accuracy and timeliness of data analysis. At the same time, the model has the characteristics of self-learning and continuous optimization, allowing the system to dynamically adapt and improve performance in practical applications. Through comprehensive analysis of historical data and real-time data, the system effectively improves the accuracy of energy consumption forecasts, optimizes energy procurement and allocation strategies, and thus achieves enterprise cost savings. In addition, while improving energy efficiency, the present invention also

[0042] Reducing carbon emissions supports sustainable development goals; the system can predict equipment failures and maintenance needs, reduce downtime risks, extend equipment life, and provide reliable support for the company's energy management decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0044] Figure 1 An overall structural diagram of an energy big data intelligent system based on a multimodal big model provided for one embodiment of the present invention.

[0045] Figure 2 A module function diagram of an energy big data intelligent system based on a multimodal big model provided for one embodiment of the present invention.

[0046] Figure 3 An overall flow chart of an energy big data intelligent method based on a multimodal large model is provided for the second embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0049] Example 1

[0050] Reference Figure 1 , is an embodiment of the present invention, and provides an energy big data intelligent system based on a multimodal big model, comprising:

[0051] User interface layer 100, application layer 200, model layer 300, security layer 400, network layer 500 and data collection layer 600;

[0052] The user interface layer 100 is configured to receive user input and display a visual interface of energy data analysis results;

[0053] The application layer 200 is configured to perform data analysis, energy efficiency optimization, predictive modeling and anomaly detection operations, and realize automatic control of energy equipment through preset rules;

[0054] The model layer 300 is configured to build and train a large multimodal model to handle tasks including multimodal data association, sentiment analysis, data mining, and energy consumption prediction;

[0055] The security layer 400 is configured to perform user identity authentication, rights management, and data encryption to control data access rights and protect data security;

[0056] The network layer 500 is configured to manage data transmission and support data communication between different devices and modules within the system;

[0057] The data collection layer 600 is configured to collect multimodal data from energy devices and sensors, and to format, standardize and clean the collected data.

[0058] The user interface layer 100 collects data from energy devices, sensors and third-party systems in real time through application programming interfaces and software development kits, wherein the data includes structured data and unstructured data;

[0059] Format and standardize the collected data and convert it into a unified structured form for subsequent processing;

[0060] Use interpolation methods to fill missing values ​​in the data, and use anomaly detection algorithms based on statistical analysis to remove invalid data to generate a cleaned data set;

[0061] The cleaned data set is transmitted to the model layer 300 to trigger data update and parameter adjustment of the model layer 300.

[0062] The application layer 200 uses a multimodal data association module to match and associate sensor data, device log data, and image data in time and space dimensions to generate a comprehensive energy data view for analysis;

[0063] Load and select an applicable deep learning model architecture according to business requirements, and process the comprehensive data view based on the model of the Transformer architecture;

[0064] During the model training process, the model weights are adjusted through the adaptive learning rate algorithm, overfitting is suppressed through the Dropout method, and the early stopping strategy is used to improve the training stability of the model;

[0065] The analysis results of the trained energy efficiency prediction model and anomaly detection model are output to the application layer 200 for subsequent optimization and control operations.

[0066] The model layer 300 analyzes the energy efficiency prediction results, generates optimization control instructions according to the predicted energy demand changes, and adjusts the operating parameters of the energy equipment;

[0067] Generate abnormal warning information based on abnormal detection results and convert the information into equipment adjustment instructions;

[0068] The generated control instructions are transmitted to the target energy device through the network layer 500 to adjust the operating status of the device.

[0069] The application layer 200 sends a user identity authentication request to the security layer 400, and the security layer 400 verifies the user identity through multi-factor authentication;

[0070] After verification, the security layer 400 performs AES-256 encryption on the control instruction and transmits the encrypted control instruction to the target device through the network layer 500 to ensure that the data is not tampered with during the transmission process.

[0071] The network layer 500 establishes a data transmission channel based on TCP / IP or MQTT protocol between the data acquisition layer 600, the model layer 300, the application layer 200 and the user interface layer 100 to achieve data distribution between the modules;

[0072] When the data flow increases to a set threshold, the network layer 500 automatically starts the load balancing function and dynamically adjusts the data transmission bandwidth of each module;

[0073] The data transmission path, time and content summary are recorded in the log to facilitate tracing analysis when a system failure occurs.

[0074] The data collection layer 600 receives and displays the energy efficiency optimization scheme and abnormal warning information transmitted by the application layer 200 at the user interface layer 100. The display method includes charts and dashboards to facilitate users to view energy usage;

[0075] The user interface layer 100 receives feedback information input by the user, allowing the user to manually adjust the control instructions generated by the system or provide operation suggestions;

[0076] The user feedback information is transmitted to the application layer 200 through the network layer 500. The application layer 200 adjusts the control parameters according to the user feedback, generates updated control instructions and sends them to the energy device.

[0077] Collect multimodal data from energy devices, sensors, and third-party systems in real time through APIs, SDKs, or other integration tools, and use big data storage solutions (such as Hadoop HDFS, NoSQL databases) to store structured and unstructured data. Remove invalid, erroneous, or incomplete data records, convert data into a unified format and unit for easy analysis, handle missing data through interpolation, regression analysis, and use statistical methods or machine learning algorithms to identify and process abnormal data. Calculate key performance indicators such as load factor, loss rate, etc. from raw data; identify patterns and trends in data, such as periodic changes, abnormal fluctuations, etc.; annotate data, such as adding information such as fault type and equipment status to the data set. Associate data from different modalities, such as associating sensor data with equipment logs, and integrate data from different sources to build a comprehensive view of energy data. Audit data regularly to ensure data accuracy and reliability, and establish a data governance framework, including data standards, data security, and privacy protection.

[0078] According to the business needs of energy big data scenarios, a deep learning model transformer suitable for processing multimodal data is designed. Model training and optimization, select a suitable loss function. Depending on the task (such as classification, regression, segmentation, etc.), select a suitable loss function to measure the difference between the model prediction and the actual value; select the gradient descent optimization algorithm, and update the parameters by calculating the gradient of the loss function relative to the model parameters to minimize the loss function; select a suitable learning rate. A learning rate that is too high may cause the model to fail to converge, and a learning rate that is too low may cause the training process to be too slow. You can use a learning rate decay strategy or an adaptive learning rate algorithm; weight decay (L2 regularization) limits the size of the model weight by adding a term proportional to the square of the weight to the loss function to prevent overfitting; Dropout randomly "discards" the output of some neurons during training, which can increase the generalization ability of the model; early stopping stops training when the performance on the validation set no longer improves to avoid overfitting; batch size, choosing an appropriate batch size can speed up Speed ​​up training and improve model stability. Small batch gradient descent uses only part of the data to update model parameters in each iteration. Parameter initialization. Reasonable parameter initialization can accelerate model convergence. Common initialization methods include random initialization, Xavier initialization, He initialization, etc. Use validation sets to evaluate model performance and adjust model structure or hyperparameters based on the evaluation results. Use grid search, random search, or Bayesian optimization to find the optimal hyperparameter combination. Improve the performance of the final model by combining the prediction results of multiple models. Common methods include voting, weighted averaging, and Stacking. Use pre-trained models as a starting point and fine-tune them to adapt to new tasks. Increase data diversity and improve the generalization ability of the model through rotation, scaling, and cropping. Use hardware such as GPU and TPU to accelerate the training process, and accelerate model training through parallel computing and distributed systems.

[0079] Use the trained model to analyze the data and extract valuable information, including descriptive statistical analysis, exploratory data analysis, predictive analysis, inferential analysis, association rule learning, distributed computing, etc. Realize real-time monitoring of the energy system, set warning thresholds according to business needs and historical data, the system identifies abnormal patterns in the data, and triggers warnings when the threshold is exceeded, and sends warning information through various means such as email and text messages.

[0080] The design of the energy big data intelligent management system needs to comprehensively consider multiple aspects such as data flow, processing requirements, user interaction, and security. The data acquisition layer is the foundation of the system and is responsible for collecting relevant data on energy use. Automatically collect operating parameters of energy equipment, such as voltage, current, power, temperature, etc. Allow manual data entry when real-time monitoring is not possible, and process the collected raw data to ensure the accuracy and availability of the data. Data standardization, converting data from different sources into a unified format for subsequent processing. Ensure that data collected from different devices and systems can be synchronized to the central database in real time. The network layer is responsible for data transmission and communication. Manage connections with different devices and systems to ensure stable and reliable data flow. Ensure that data is transmitted safely and efficiently from the collection point to the data center. Support various network protocols, such as TCP / IP, MQTT, etc., to meet different data transmission requirements. Load balancing, distribute requests when network traffic is high to maintain system stability and responsiveness. Encrypt data in transit to protect data security. The security layer is key to protecting the system from unauthorized access and network attacks. Implement strong authentication mechanisms such as multi-factor authentication. Role-based access control ensures that users can only access authorized data and functions. Deploy firewalls, intrusion detection systems, etc. to prevent and respond to security threats. Encrypt stored and transmitted data to enhance data security. Record and review system activities to track and analyze security events. The application layer is the center of system logic and data processing. Data analysis uses statistical analysis, machine learning and other technologies to conduct in-depth analysis of energy usage data. Energy efficiency optimization provides energy usage optimization suggestions to help users reduce energy consumption. Predictive modeling establishes energy demand prediction models to predict future energy usage trends. Automated control automatically adjusts the operation of energy equipment according to preset rules and real-time data. Anomaly detection monitors energy usage anomalies in real time, issues timely alarms and provides solutions. The user interface layer is the interface for users to interact with the system. Energy usage data is intuitively displayed through graphics, charts and dashboards. Design intuitive and easy-to-use user interfaces to improve user experience. Allow users to customize the interface and the information they receive according to their personal preferences.

[0081] Example 2

[0082] Reference Figure 2 , is an embodiment of the present invention, and provides a design method for an energy big data intelligent system based on a multimodal big model, comprising:

[0083] The data collection layer collects multimodal data from energy devices and sensors in real time, performs unified formatting and cleaning operations, and converts them into standardized comprehensive energy data views;

[0084] The model layer receives comprehensive data views, integrates multimodal data based on a deep learning framework to build an analysis model, and performs training to enable the model to have energy efficiency prediction and anomaly detection capabilities;

[0085] The application layer receives the output of the analysis model, converts energy efficiency prediction and abnormal information into control instructions, and transmits them to the target device through the security layer to achieve intelligent regulation of the energy system;

[0086] The collection layer dynamically obtains new data and, based on user feedback, updates the parameters of the model layer and the control logic of the application layer.

[0087] Phase 1: Define system application requirements, mainly including intelligent energy monitoring and prediction, using multimodal large models

[0088] Process photos or videos of energy equipment to achieve equipment status monitoring and anomaly detection, and combine text and sensor data to achieve real-time monitoring and prediction of energy consumption; Energy efficiency optimization, integrate information from different sensors, images and text data to improve the understanding of the operating status of the energy system, thereby optimizing energy utilization efficiency, and use multimodal large models to quickly analyze complex data of the energy system to help discover potential energy-saving improvements; Fault diagnosis and maintenance, by processing text data such as equipment manuals and maintenance records, combined with image and sensor data, to achieve automatic diagnosis and preventive maintenance of equipment faults, and combined with multimodal data to identify potential factors that cause energy system failures or performance degradation; Energy market analysis and forecasting, analyze text data such as social media and news reports, combine image data of market price curves, predict future trends in the energy market, and use multimodal large models to mine hidden laws and trends in historical energy market data to assist decision-making; User behavior analysis and management, combine users' text communication, behavior trajectory and image data to build user portraits, provide users with personalized energy-saving suggestions and services, predict users' energy consumption based on multimodal data, and help users arrange energy usage plans reasonably.

[0089] The second stage: building a large multimodal model, mainly including model selection, choosing Transformer-base

[0090] Models as infrastructure, such as BERT (Bidirectional Encoder Representations from

[0091] Transformers) or their variants, which are excellent in processing sequence data such as text and can be easily integrated with other modal data; model design, fusion of multimodal data, design of model structure to integrate multimodal data such as text, images, audio, etc., can use multi-input model (Multi-Input Model) or multi-output model (Multi-Output Model) Model) to handle different types of data input; create an end-to-end multimodal model architecture to ensure effective interaction and information transfer between modal data, consider how to share some low-level features and benefit from them while retaining the unique characteristics of each modal data, determine the output form of the model, whether it is classification, regression, clustering and other tasks, and for energy big data, design a complex output layer to make intelligent decisions; model training, merge and prepare multimodal data such as text, images, audio, etc., ensure that the data format and quality meet the model requirements, divide the data set into training set, validation set and test set, use multimodal data to train the model, learn the association between modal data through joint training or alternating training, adjust hyperparameters such as learning rate and batch size to optimize model performance, evaluate model performance on the validation set, and tune the model according to indicators (such as accuracy, precision, recall, etc.) to prevent overfitting. Regularization methods or early stopping strategies can be used.

[0092] The third stage: system integration and deployment, mainly including model integration, integrating the trained multimodal model

[0093] Integration, ensure that each modal data is correctly input into the corresponding model, design data flow pipelines, pass different types of data to appropriate models for processing, consider using model fusion technology (such as voting, weighted average, etc.) to combine the outputs of each model to improve the overall prediction performance, and ensure that the output of the integrated model has a better effect on solving problems in the energy field; system testing, perform unit testing on each model and component to ensure that it functions normally and as expected, test the performance of the model on each modal data and the integration between models, perform integration testing of the overall system, verify the overall performance of the multimodal model, simulate data scenarios in actual production environments, check the stability and accuracy of the system when processing various data types, evaluate the system's performance indicators, including response time, throughput, etc., analyze the system's performance when processing large-scale energy data, and optimize the system to improve performance. System deployment, set up appropriate hardware and software environment in the production environment, ensure sufficient resources required by the system, deploy necessary components such as database, server, model server, etc., strengthen the system's security measures, including data encryption, access control, etc., to prevent data leakage and malicious intrusion, ensure that the system meets relevant laws and regulations and standards, protect user data privacy, deploy monitoring system, monitor system operation status and performance indicators in real time, discover and solve problems in time, establish continuous integration and continuous deployment (CI / CD) process, and ensure the efficiency of system updates and maintenance.

[0094] The fourth stage: system optimization and maintenance, mainly including performance optimization, optimization of the algorithms used in the system,

[0095] Improve computing efficiency and accuracy, consider using more effective algorithms or optimizing existing algorithms to speed up data processing, adjust system resource allocation, including memory, CPU, etc., to maximize the use of hardware resources, use caching technology, parallel computing and other methods to improve the system's processing capabilities, optimize database query statements and index design, improve data retrieval speed, regularly clean up useless data, compress data storage, reduce database burden and improve performance; continuous monitoring, real-time monitoring of system performance indicators such as response time, throughput, etc., timely discover performance problems, set up alarm mechanisms, and immediately notify relevant personnel in case of abnormal situations, establish a data quality monitoring system, regularly check data integrity, accuracy and consistency, and detect Data outliers and missing values, ensure that data updates are timely and reliable, monitor system security events, including potential security vulnerabilities, network attacks, etc., strengthen control over system access rights, and regularly review system security policies; data update, design data update processes to ensure that the system can receive the latest energy data in a timely manner, integrate data acquisition systems, automatically obtain real-time data and update it to the system, establish data cleaning processes, regularly clean up outdated or invalid data, maintain data accuracy, perform data deduplication, formatting, standardization and other operations to improve data quality, regularly back up data to ensure that data is not lost due to accidents, and establish a reliable data recovery mechanism to quickly restore system functions when data is damaged or lost.

[0096] Through the steps of energy multimodal big data collection and preprocessing; energy multimodal big model construction; energy multimodal big model intelligent analysis; energy big data intelligent management system design, it can process data from different sources and types, including structured and unstructured data, and improve the availability and value of data. Through real-time data stream processing, the system can quickly respond to changes in the energy market and operations and provide immediate decision support. Using advanced deep learning technology, the system can understand and analyze multimodal data to improve the accuracy of prediction and analysis. The system can continuously optimize the model through self-supervised learning to adapt to new data patterns and business needs. The system can provide specific optimization suggestions based on the data analysis structure to help improve energy efficiency and reduce costs. It can bring significant technical advantages and commercial value to the energy industry and promote the development of energy management towards intelligence, automation and efficiency.

[0097] Multimodal big models are able to process and integrate data from different sources and types (such as sensor data, video, audio, text, etc.) to provide a more comprehensive analysis of energy usage. Using deep learning and machine learning algorithms, the system is able to perform complex data analysis tasks such as predicting energy demand, identifying energy waste and abnormal usage patterns. Multimodal big models are usually capable of self-learning and optimization, and can continuously improve performance over time. By analyzing historical and real-time data, multimodal models can improve the accuracy of energy consumption forecasts and help companies optimize energy procurement and allocation. By optimizing energy use and reducing waste, companies can achieve significant cost savings. Improving energy efficiency helps reduce carbon emissions and environmental impact and supports sustainable development. The system can predict equipment failures and maintenance needs, helping companies reduce downtime and increase equipment life. Companies can make more data-driven decisions based on the in-depth analysis and insights provided by the system.

[0098] Example 3

[0099] An embodiment of the present invention is different from the first two embodiments in that:

[0100] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0101] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0102] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0103] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0104] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An energy big data intelligent system based on a multimodal big model, characterized in that: include: User interface layer (100), application layer (200), model layer (300), security layer (400), network layer (500) and data collection layer (600); The user interface layer (100) is configured to receive user input and display a visual interface of energy data analysis results; The application layer (200) is configured to perform data analysis, energy efficiency optimization, predictive modeling and anomaly detection operations, and to achieve automated control of energy equipment through preset rules; The model layer (300) is configured to build and train a large multimodal model to handle tasks including multimodal data association, sentiment analysis, data mining, and energy consumption prediction; The security layer (400) is configured to perform user identity authentication, rights management and data encryption to control data access rights and protect data security; The network layer (500) is configured to manage data transmission and support data communication between different devices and modules in the system; The data collection layer (600) is configured to collect multimodal data from energy devices and sensors, and to format, standardize and clean the collected data.

2. The energy big data intelligent system based on multimodal big model as claimed in claim 1, characterized in that: The user interface layer (100) collects data from energy devices, sensors and third-party systems in real time through an application program interface and a software development kit, wherein the data includes structured data and unstructured data; Format and standardize the collected data and convert it into a unified structured form for subsequent processing; Use interpolation methods to fill missing values ​​in the data, and use anomaly detection algorithms based on statistical analysis to remove invalid data to generate a cleaned data set; The cleaned data set is transmitted to the model layer (300) to trigger data update and parameter adjustment of the model layer (300).

3. The energy big data intelligent system based on multimodal big model as claimed in claim 2, characterized in that: The application layer (200) uses a multimodal data association module to match and associate sensor data, device log data and image data in time and space dimensions to generate a comprehensive energy data view for analysis; Load and select an applicable deep learning model architecture according to business requirements, and process the comprehensive data view based on the model of the Transformer architecture; During the model training process, the model weights are adjusted through the adaptive learning rate algorithm, overfitting is suppressed through the Dropout method, and the early stopping strategy is used to improve the training stability of the model; The analysis results of the trained energy efficiency prediction model and anomaly detection model are output to the application layer (200) for subsequent optimization and control operations.

4. The energy big data intelligent system based on multimodal big model as claimed in claim 3 is characterized by: The model layer (300) analyzes the energy efficiency prediction results, generates optimization control instructions according to the predicted energy demand changes, and adjusts the operating parameters of the energy equipment; Generate abnormal warning information based on abnormal detection results and convert the information into equipment adjustment instructions; The generated control instructions are transmitted to the target energy device through the network layer (500) to adjust the operating state of the device.

5. The energy big data intelligent system based on multimodal big model as claimed in claim 4, characterized in that: The application layer (200) sends a user identity authentication request to the security layer (400), and the security layer (400) verifies the user identity through a multi-factor authentication method; After verification, the security layer (400) performs AES-256 encryption on the control instruction and transmits the encrypted control instruction to the target device through the network layer (500) to ensure that the data is not tampered with during the transmission process.

6. The energy big data intelligent system based on multimodal big model as claimed in claim 5, characterized in that: The network layer (500) establishes a data transmission channel based on TCP / IP or MQTT protocol between the data collection layer (600), the model layer (300), the application layer (200) and the user interface layer (100) to achieve data distribution between the modules; When the data flow increases to a set threshold, the network layer (500) automatically starts the load balancing function and dynamically adjusts the data transmission bandwidth of each module; The data transmission path, time and content summary are recorded in the log to facilitate tracing analysis when a system failure occurs.

7. The energy big data intelligent system based on multimodal big model as claimed in claim 6, characterized in that: The data collection layer (600) receives and displays the energy efficiency optimization plan and abnormal warning information transmitted by the application layer (200) at the user interface layer (100), wherein the display method includes charts and dashboards, so as to facilitate users to view energy usage; The user interface layer (100) receives feedback information input by the user, allowing the user to manually adjust the control instructions generated by the system or provide operation suggestions; The user feedback information is transmitted to the application layer (200) through the network layer (500); the application layer (200) adjusts the control parameters according to the user feedback, generates updated control instructions and sends them to the energy device.

8. A design method for an energy big data intelligent system based on a multimodal big model according to any one of claims 1 to 7, characterized in that: include: The data collection layer collects multimodal data from energy devices and sensors in real time, performs unified formatting and cleaning operations, and converts them into standardized comprehensive energy data views; The model layer receives comprehensive data views, integrates multimodal data based on a deep learning framework to build an analysis model, and performs training to enable the model to have energy efficiency prediction and anomaly detection capabilities; The application layer receives the output of the analysis model, converts energy efficiency prediction and abnormal information into control instructions, and transmits them to the target device through the security layer to achieve intelligent regulation of the energy system; The collection layer dynamically obtains new data and, based on user feedback, updates the parameters of the model layer and the control logic of the application layer.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for designing an energy big data intelligent system based on a multimodal large model as described in claim 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for designing an energy big data intelligent system based on a multimodal large model as described in claim 8 are implemented.

Images