Data acquisition energy management system and energy prediction analysis method

Abstract

The invention belongs to the technical field of energy management, and particularly relates to a data acquisition energy management system and an energy prediction analysis method, and the system is characterized in that a data acquisition gateway acquires data measured by a sensor or acquires real-time operation data of equipment, and transmits the data to a data layer according to a set protocol; the data layer receives the data transmitted by the data acquisition gateway, normalizes and standardizes the transmitted data through a data processing engine in the data layer, and stores the data in a database; meanwhile, the data layer trains an energy prediction model, wherein the energy prediction model is used for converting the obtained data into data for predicting the energy use trend, anomaly detection and energy-saving suggestion; the application layer monitors energy consumption in real time and performs load balance and energy optimization control by controlling operation of energy equipment; the display layer provides a user interface and a visual report to display energy prediction data and an analysis result. And the data processing efficiency and the energy prediction precision are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy management, and in particular relates to a data acquisition energy management system and an energy prediction and analysis method. Background Art

[0002] With the acceleration of industrialization and urbanization, energy consumption has increased dramatically, and energy management has become an urgent problem to be solved. Traditional energy management methods often rely on manual meter reading, statistics and analysis, which is not only inefficient, but also difficult to achieve real-time monitoring and refined management of energy consumption. In addition, with the rapid development of the Internet of Things, big data and artificial intelligence technologies, new technical means and solutions have been provided for energy management.

[0003] However, most existing energy management systems have problems such as insufficient data processing capabilities, low energy forecasting accuracy, and insufficient intelligence in energy management strategies. Data collection in many systems is difficult to fully reflect the true situation of energy consumption. In terms of data processing and energy forecasting, many systems still use traditional statistical methods or simple machine learning models, which cannot effectively mine the potential laws and trends in the data, resulting in low accuracy and reliability of energy forecasting. Summary of the invention

[0004] In view of the problems that most existing energy management systems have, such as insufficient data processing capability, low energy prediction accuracy, and insufficient intelligence of energy management strategies, the present invention provides a data acquisition energy management system and an energy prediction and analysis method.

[0005] In a first aspect, the technical solution of the present invention provides a data collection energy management system, including a perception layer, a data layer, an application layer and a presentation layer; The perception layer includes sensors and data acquisition gateways that collect energy consumption information and environmental parameter information; the data acquisition gateway collects data measured by sensors or real-time operation data of acquisition equipment and transmits it to the data layer according to the set protocol; The data layer receives data transmitted by the data acquisition gateway, normalizes and standardizes the transmitted data through the data processing engine inside the data layer and stores it in the database; at the same time, the data layer trains the energy prediction model, which is used to convert the acquired data into data for predicting energy usage trends, anomaly detection and energy-saving recommendations; The application layer monitors energy consumption in real time and performs load balancing and energy optimization control by controlling the operation of energy devices to achieve optimal energy utilization and reduce energy consumption; The presentation layer provides a user interface and visual reports to display energy forecast data and analysis results.

[0006] As a preferred embodiment of the technical solution of the present invention, the perception layer also includes one or more smart meters for directly measuring and transmitting energy consumption data to a data acquisition gateway.

[0007] As a preferred embodiment of the technical solution of the present invention, the process of training the energy prediction model at the data layer includes: Obtain historical energy data, including energy consumption data, energy production data, energy price data, and environmental parameter data; Clean historical energy data to remove outliers and missing values; Extract trend, seasonality, and cyclical features from time series data; Standardize the data; Use LSTM networks to learn long-term dependencies in time series data; Introduce CNN to extract local features from time series data; Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal and random components; Design a hybrid loss function that combines mean square error and mean absolute error and balances the model's sensitivity to different errors by adjusting weights; Use the preprocessed data to train the deep learning model; monitor the loss function value during the training process; The LSTM network model and the CNN network model are fused through the weighted average algorithm to obtain a deep time series energy prediction model.

[0008] As a preferred embodiment of the technical solution of the present invention, the specific process of energy optimization control performed by the application layer is as follows: Generate energy consumption reports based on energy data received by the data layer to identify anomalies and waste in energy use; Monitor the operating status of energy equipment; Dynamically adjust the load distribution of energy equipment based on real-time energy consumption data and equipment operating status.

[0009] Intelligent algorithms are used to optimize the operation strategies of energy equipment to achieve load balancing and energy optimization control.

[0010] As a preferred embodiment of the technical solution of the present invention, the specific process of energy optimization control performed by the application layer also includes: Dynamically adjust the equipment's operating status and load distribution based on real-time load demand and equipment capabilities; The target formula for load balancing is as follows: Load balancing rate = (actual load / total equipment capacity) × 100%.

[0011] As a preferred embodiment of the technical solution of the present invention, the application layer uses a particle swarm optimization algorithm to minimize energy consumption as the optimization goal and the balance of energy supply and demand of energy equipment as the constraint condition to optimize the operation strategy of energy equipment and achieve load balancing and energy optimization control; according to the output results of the particle swarm optimization algorithm, the operation strategy of the energy equipment is formulated, and the strategy is applied to the energy equipment and executed; the execution effect of the strategy is monitored in real time, and the strategy parameters are adjusted according to feedback.

[0012] As a preferred embodiment of the technical solution of the present invention, the specific process of optimizing the operation strategy of energy equipment by using the particle swarm optimization algorithm at the application layer is as follows: Initialize the particle swarm to generate a set of random potential solutions as particles; Calculate the fitness value of each particle according to the fitness function; Compare the current fitness value with the individual historical optimal solution and update the individual optimal solution; Compare the current fitness value with the global historical optimal solution and update the global optimal solution; Adjust the speed and position of the particles according to the update rule to approach the individual optimal solution and the global optimal solution; Repeatedly calculate fitness, update individual and global optimal solutions, and update particle speed and position until the stopping condition is reached.

[0013] As a preferred embodiment of the technical solution of the present invention, the objective function describing the energy optimization target is: min(energy consumption cost) = min(∑(energy consumption × energy price)).

[0014] As a preferred embodiment of the technical solution of the present invention, the data acquisition gateway encrypts the data and transmits the data to the data layer; The data layer is provided with a server, and the server has several initial keys built in. Each initial key includes a key serial number and a key value corresponding to the key serial number. When the data acquisition gateway needs to send data to the data layer, it sends a key request to the server. The server extracts several initial keys based on the received request and returns them to the data acquisition gateway. When the data acquisition gateway transmits data, it extracts an initial key from the several initial keys received, encrypts the data with the key value of the initial key, and encrypts the key serial number of the initial key to generate encrypted data. The encrypted data and the extracted key serial number of the initial key used to encrypt the data are sent to the data layer together.

[0015] As a preferred embodiment of the technical solution of the present invention, after receiving the data, the data layer obtains the corresponding key value for decryption based on the received key serial number, verifies the encrypted key serial number with the key value, and after the verification is passed, decrypts the data based on the decrypted key value to obtain the corresponding data.

[0016] In a second aspect, the technical solution of the present invention also provides an energy forecasting and analysis method, comprising the following steps: Receive data transmitted by the data acquisition gateway, normalize and standardize the transmitted data through the data processing engine inside the data layer, and store it in the database; Determine whether the energy prediction model has been trained; If so, the acquired data is processed and input into an energy forecasting model to convert the acquired data into data for predicting energy usage trends, anomaly detection, and energy-saving recommendations; If not, historical energy data is obtained and an energy prediction model is trained using an LSTM network and a CNN network based on the obtained historical data; execution steps: the obtained data is processed and input into the energy prediction model to convert the obtained data into data for predicting energy usage trends, anomaly detection, and energy-saving recommendations.

[0017] The steps of acquiring historical energy data and training an energy prediction model using an LSTM network and a CNN network based on the acquired historical data include: Clean historical energy data to remove outliers and missing values; Extract trend, seasonality, and cyclical features from time series data; Standardize the data; Use LSTM networks to learn long-term dependencies in time series data; Introduce CNN to extract local features from time series data; Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal and random components; Design a hybrid loss function that combines mean square error and mean absolute error and balances the model's sensitivity to different errors by adjusting weights; Use the preprocessed data to train the deep learning model; monitor the loss function value during the training process; The LSTM network model and the CNN network model are fused through the weighted average algorithm to obtain a deep time series energy prediction model, namely the energy prediction model.

[0018] It can be seen from the above technical solutions that the present invention has the following advantages: through the sensors, data acquisition gateways and intelligent meters in the perception layer, it is possible to comprehensively collect energy consumption information and environmental parameter information, providing a rich data source for subsequent data processing and energy forecasting. The data layer normalizes and standardizes the collected data through the internal data processing engine and stores it in the database. At the same time, the data layer can also train the energy forecasting model, converting the acquired data into data for predicting energy usage trends, anomaly detection and energy-saving recommendations, thereby improving data processing efficiency and the accuracy of energy forecasting.

[0019] The application layer can monitor energy consumption in real time, and perform load balancing and energy optimization control by controlling the operation of energy equipment, achieving the goal of optimal energy utilization and reducing energy consumption. At the same time, the presentation layer provides a user interface and visual reports to display energy forecast data and analysis results, making energy management more intuitive and convenient. In the process of training the energy forecast model at the data layer, deep learning technologies such as LSTM network and CNN are introduced to effectively mine the long-term dependencies and local features in time series data, improving the accuracy and reliability of energy forecasting. At the same time, the prediction performance of the model is further improved by designing a hybrid loss function and using a weighted average algorithm to fuse the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 is a schematic block diagram of a system according to an embodiment of the present invention.

[0022] Figure 2 is a schematic flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0024] like Figure 1As shown, the technical solution of the present invention provides a data acquisition energy management system, including a perception layer, a data layer, an application layer and a presentation layer; The perception layer includes sensors and data acquisition gateways that collect energy consumption information and environmental parameter information; the data acquisition gateway collects data measured by sensors or real-time operation data of acquisition equipment and transmits it to the data layer according to the set protocol; The data layer receives data transmitted by the data acquisition gateway, normalizes and standardizes the transmitted data through the data processing engine inside the data layer and stores it in the database; at the same time, the data layer trains the energy prediction model, which is used to convert the acquired data into data for predicting energy usage trends, anomaly detection and energy-saving recommendations; The application layer monitors energy consumption in real time and performs load balancing and energy optimization control by controlling the operation of energy devices to achieve optimal energy utilization and reduce energy consumption; The presentation layer provides a user interface and visual reports to display energy forecast data and analysis results.

[0025] In the embodiment of the present invention, sensors that can accurately measure energy consumption (such as electricity, gas, water, etc.) and environmental parameters (such as temperature, humidity, light intensity, etc.) are selected according to the monitoring requirements. These sensors should have high precision, low power consumption and long-term stability. Data acquisition gateways are installed in each monitoring area. These gateways are responsible for collecting data from sensors and converting them into a format that can be transmitted over the network. Ensure that the gateway supports the set communication protocol (such as Modbus, Zigbee, LoRa, etc.) so that it can be seamlessly connected with the data layer. Set the transmission frequency and protocol of the data acquisition gateway to ensure that the data is sent to the data layer in real time and accurately.

[0026] A data receiving module is built at the data layer to receive data from the data collection gateway and clean, format and standardize the data through the data processing engine. The processed data is stored in a relational database (such as MySQL, PostgreSQL) or a non-relational database (such as MongoDB) for subsequent analysis and query.

[0027] The data layer also has a model training module that uses machine learning algorithms (such as support vector machines, random forests, neural networks, etc.) to train energy prediction models. The energy prediction model learns the relationship between energy consumption and environmental parameters based on historical data and predicts future energy usage trends. At the same time, the model can also be used for anomaly detection and generation of energy-saving suggestions.

[0028] The data layer is also equipped with a data update module to regularly update historical data in the database and optimize the performance of the energy forecasting model. At the same time, the data layer is maintained to ensure high availability and security of data transmission and processing.

[0029] The application layer is equipped with a real-time monitoring module to display energy consumption data and equipment operation status in real time, and to intuitively present energy usage through charts and dashboards.

[0030] The application layer is also equipped with an energy optimization control module, which performs load balancing and energy optimization control on energy equipment based on real-time monitoring data and the results of the energy prediction model and intelligent control algorithms. For example, it automatically adjusts air conditioning temperature and lighting brightness during peak power consumption to reduce energy consumption.

[0031] The application layer also has an early warning module that sets alarm thresholds. When energy consumption is abnormal or equipment fails, alarm information is sent to relevant personnel in a timely manner.

[0032] The presentation layer provides a user interface to facilitate users to view energy forecast data, analysis results and real-time monitoring information. The interface should support multiple devices and browsers to ensure a good user experience.

[0033] The user interface generates visual reports based on energy-related data, showing energy consumption trends, energy-saving effects, equipment operating status, etc. The reports support export functions, which facilitates users to conduct further analysis and reporting.

[0034] Through real-time monitoring and intelligent control, load balancing and energy optimization control of energy equipment can be achieved, energy consumption can be effectively reduced, and energy utilization efficiency can be improved. Through automated monitoring and alarm functions, energy equipment failures can be discovered and resolved in a timely manner, reducing manual inspection and maintenance costs. Energy-saving suggestions based on energy forecasting models help companies formulate scientific energy-saving plans and achieve energy-saving and emission-reduction goals. Through visual reports and data analysis results, intuitive and accurate energy usage information is provided to corporate decision makers to support scientific decision-making and strategic planning. The intuitive user interface and visual reports improve the user experience and make energy management more convenient and efficient.

[0035] Here, the perception layer also includes one or more smart meters for directly measuring and transmitting energy consumption data to a data acquisition gateway.

[0036] In some embodiments, the process of training the energy prediction model at the data layer includes: Obtain historical energy data, including energy consumption data, energy production data, energy price data, and environmental parameter data; Clean historical energy data to remove outliers and missing values; Extract trend, seasonality, and cyclical features from time series data; Standardize the data; Use LSTM networks to learn long-term dependencies in time series data; Introduce CNN to extract local features from time series data; Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal and random components; Design a hybrid loss function that combines mean square error and mean absolute error and balances the model's sensitivity to different errors by adjusting weights; Use the preprocessed data to train the deep learning model; monitor the loss function value during the training process; The LSTM network model and the CNN network model are fused through the weighted average algorithm to obtain a deep time series energy prediction model.

[0037] In the embodiment of the present invention, historical energy data including energy consumption data, energy production data, energy price data and environmental parameter data (such as temperature, humidity, wind speed, etc.) are obtained from multiple data sources (such as energy management systems, smart meters, weather forecast systems, etc.). The integrity, accuracy and consistency of the data are ensured, and duplicate, invalid or abnormal data records are removed through data cleaning technology.

[0038] Clean historical energy data to remove obvious outliers and missing values. For missing values, interpolation, mean filling or machine learning-based methods can be used to fill them. Extract trend, seasonality and cyclical characteristics in time series data. This can be achieved through methods such as time series decomposition, seasonal difference or Fourier transform. Standardize the data to eliminate dimensional differences between different variables and improve the training efficiency and prediction accuracy of the model.

[0039] Use LSTM (Long Short-Term Memory) networks to learn long-term dependencies in time series data. LSTM networks are able to capture long-term memory effects in time series and are particularly effective for long-term trend prediction in energy forecasting. CNN (Convolutional Neural Network) is introduced to extract local features in time series data. CNN is good at processing data with local correlations and can capture short-term fluctuations and cyclical changes in time series. Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal, and random components. This helps to explain the forecast results and improve the interpretability and credibility of the model.

[0040] Design a hybrid loss function that combines the mean square error (MSE) and the mean absolute error (MAE), and balance the model's sensitivity to different errors by adjusting the weights. This helps improve the robustness of the model and the accuracy of predictions. Use preprocessed data to train the deep learning model, monitor the loss function value during training, and ensure that the model gradually converges during training. Use the weighted average algorithm to fuse the LSTM network model and the CNN network model to obtain a deep time series energy prediction model. The fusion model can combine the advantages of the two models and improve the accuracy and stability of predictions.

[0041] By combining the advantages of LSTM and CNN, the deep time series energy prediction model can capture long-term dependencies and local features in the time series, thereby improving the accuracy and stability of the prediction. Adding a time series decomposition module before the model output layer to decompose the prediction results into trend, seasonal and random components helps to explain the prediction results and improve the interpretability and credibility of the model. By designing a hybrid loss function, using early stopping, learning rate adjustment, regularization and other technologies, the robustness and generalization ability of the model are improved, so that the model can better adapt to different scenarios and data sets. The deep time series energy prediction model can provide accurate and reliable prediction results for energy management, support load balancing and energy optimization control of energy equipment, and promote intelligent and efficient energy management.

[0042] In some embodiments, the specific process of energy optimization control performed by the application layer is as follows: Generate energy consumption reports based on energy data received by the data layer to identify anomalies and waste in energy use; Monitor the operating status of energy equipment; Dynamically adjust the load distribution of energy equipment based on real-time energy consumption data and equipment operating status.

[0043] Intelligent algorithms are used to optimize the operation strategies of energy equipment to achieve load balancing and energy optimization control.

[0044] In the embodiment of the present invention, after the data layer receives and processes the energy data from the perception layer, the application layer first generates a detailed energy consumption report based on the data. The report should include the energy consumption of each energy device, the trend analysis of energy use, and the comparison with historical data.

[0045] Identify abnormal patterns and potential waste points in energy use through data analysis techniques (such as statistical analysis, machine learning algorithms, etc.). Abnormalities may manifest as sudden increases in energy consumption, long-term inefficient operation of equipment, etc.

[0046] The application layer monitors the operating status of energy equipment in real time, including the equipment's on / off status, operating efficiency, fault alarm, etc. Through sensor data and network communication, the operating parameters of the equipment are obtained in real time. An equipment operating status database is established to record the equipment's historical operating data and fault records, providing data support for subsequent energy optimization control.

[0047] Based on real-time energy consumption data and equipment operating status, the application layer dynamically adjusts the load distribution of energy devices. This includes adjusting the operating power, working hours, and startup sequence of the equipment according to changes in energy demand. The introduction of a load balancing algorithm ensures that each energy device can evenly share the load while maintaining efficient operation, avoiding situations where some devices are overloaded while others are idle.

[0048] The application layer uses advanced intelligent algorithms (such as reinforcement learning, deep learning, etc.) to optimize the operation strategy of energy equipment. These algorithms can learn the energy consumption characteristics and operation rules of equipment based on historical data and real-time data, so as to develop more efficient operation strategies. Optimization strategies may include adjusting the operating parameters of the equipment, optimizing the working mode of the equipment (such as intermittent operation, timed start, etc.), and adjusting the energy consumption of the equipment according to energy prices and environmental parameters.

[0049] By dynamically adjusting the load distribution of energy equipment and optimizing equipment operation strategies, the application layer can ensure that energy equipment maintains efficient operation while minimizing energy waste and improving energy utilization efficiency. Real-time monitoring of the operating status of energy equipment and timely identification of potential faults can help take measures in advance for maintenance and repair, avoid energy interruptions and production losses caused by equipment failures, and enhance the reliability and stability of equipment. The application layer has achieved intelligent and automated energy management by introducing advanced intelligent algorithms and data analysis technologies. This helps reduce reliance on manual intervention and improve the efficiency and accuracy of energy management. By improving energy utilization efficiency and reducing energy costs, energy optimization control at the application layer helps enterprises achieve energy conservation and emission reduction goals and promote sustainable development. At the same time, intelligent energy management also provides strong support for the green transformation and sustainable development of enterprises.

[0050] Correspondingly, the specific process of energy optimization control at the application layer also includes: Dynamically adjust the equipment's operating status and load distribution based on real-time load demand and equipment capabilities; The target formula for load balancing is as follows: Load balancing rate = (actual load / total equipment capacity) × 100%.

[0051] Specifically, the application layer uses the particle swarm optimization algorithm to minimize energy consumption as the optimization goal and the balance of energy supply and demand of energy equipment as the constraint condition to optimize the operation strategy of energy equipment and achieve load balancing and energy optimization control; according to the output results of the particle swarm optimization algorithm, the operation strategy of energy equipment is formulated, and the strategy is applied to the energy equipment and executed; the execution effect of the strategy is monitored in real time, and the strategy parameters are adjusted according to the feedback.

[0052] The specific process of using particle swarm optimization algorithm to optimize the operation strategy of energy equipment at the application layer is as follows: Initialize the particle swarm to generate a set of random potential solutions as particles; Calculate the fitness value of each particle according to the fitness function; Compare the current fitness value with the individual historical optimal solution and update the individual optimal solution; Compare the current fitness value with the global historical optimal solution and update the global optimal solution; Adjust the speed and position of the particles according to the update rule to approach the individual optimal solution and the global optimal solution; Repeatedly calculate fitness, update individual and global optimal solutions, and update particle speed and position until the stopping condition is reached.

[0053] Objective function describing the energy optimization goal: min(energy consumption cost) = min(∑(energy consumption × energy price)).

[0054] In some embodiments, the data collection gateway encrypts the data and transmits the data to the data layer; The data layer is provided with a server, and the server has several initial keys built in. Each initial key includes a key serial number and a key value corresponding to the key serial number. When the data acquisition gateway needs to send data to the data layer, it sends a key request to the server. The server extracts several initial keys based on the received request and returns them to the data acquisition gateway. When the data acquisition gateway transmits data, it extracts an initial key from the several initial keys received, encrypts the data with the key value of the initial key, and encrypts the key serial number of the initial key to generate encrypted data. The encrypted data and the extracted key serial number of the initial key used to encrypt the data are sent to the data layer together.

[0055] Correspondingly, after receiving the data, the data layer obtains the corresponding key value for decryption based on the received key serial number, verifies the encrypted key serial number with the key value, and after the verification is passed, decrypts the data based on the decrypted key value to obtain the corresponding data.

[0056] When the data acquisition gateway is preparing to send data to the data layer, it first generates a data request packet, which contains the original data to be transmitted. At the same time, the data acquisition gateway sends a key request to the server of the data layer to request the initial key used for data encryption.

[0057] After receiving the key request, the server at the data layer randomly extracts several (for example, 3 or 5) initial keys from its built-in initial keys, and packages the key serial numbers and corresponding key values ​​of these initial keys into a key response package. The server sends the key response package back to the data collection gateway.

[0058] After receiving the key response packet, the data acquisition gateway randomly extracts an initial key for data encryption. The original data is encrypted using the key value of the extracted initial key to generate encrypted data. At the same time, the data acquisition gateway uses an encryption algorithm (such as a symmetric encryption algorithm) to encrypt the key sequence number of the extracted initial key to generate an encrypted key sequence number. The data acquisition gateway packages the encrypted data and the encrypted key sequence number into a data transmission package and sends it to the data layer.

[0059] After the data layer receives the data transmission package, it first extracts the encrypted key serial number and the encrypted data. Use the initial key stored in the server to find the corresponding key value according to the encrypted key serial number (because the key serial number is unique, the corresponding key value can be directly located). Use the found key value to decrypt the encrypted key serial number to obtain the original key serial number. Compare the decrypted key serial number with the pre-encrypted key serial number extracted from the data transmission package (although the pre-encrypted key serial number is not actually transmitted directly during the data transmission process, in order to illustrate the verification process here, it is assumed that there is a "pre-encrypted key serial number" used in the verification step. In fact, this step is verified by other means, such as predefined rules or protocols). If the two are consistent, the verification is passed, indicating that the data transmission process has not been tampered with. Use the verified key value to decrypt the encrypted data to obtain the original data.

[0060] By implementing data encryption transmission between the data acquisition gateway and the data layer, the data can be effectively prevented from being stolen or tampered with during transmission, thereby enhancing the security of data transmission. The server has multiple built-in initial keys, and randomly extracts several keys for the data acquisition gateway to choose each time data is transmitted. This mechanism improves the flexibility of key management and reduces the risk of key cracking. Since different keys may be used for each data transmission, dynamic update of keys is achieved. Even if a key is leaked, it will not affect the security of previous or subsequent data transmission. By managing keys using the correspondence between key serial numbers and key values, the key management process is simplified, making the allocation, update, and revocation of keys easier and more efficient. This data encryption transmission mechanism has good scalability and reliability, and can adapt to energy management systems of different scales and complexities, providing a strong guarantee for the stable operation of the system.

[0061] like Figure 2 As shown, the embodiment of the present invention also provides an energy forecasting and analysis method, comprising the following steps: Step 1: Receive data transmitted by the data acquisition gateway, normalize and standardize the transmitted data through the data processing engine inside the data layer, and store it in the database; Step 2: Determine whether there is a trained energy prediction model; If yes, go to step 4; If not, proceed to step 3; Step 3: Obtain historical energy data and train the energy prediction model using the LSTM network and CNN network based on the acquired historical data; execute step 4.

[0062] Step 4: Process the acquired data and input it into the energy prediction model to convert the acquired data into data for predicting energy usage trends, anomaly detection and energy saving recommendations; Specifically, the steps of obtaining historical energy data and training an energy prediction model using an LSTM network and a CNN network based on the obtained historical data include: Clean historical energy data to remove outliers and missing values; Extract trend, seasonality, and cyclical features from time series data; Standardize the data; Use LSTM networks to learn long-term dependencies in time series data; Introduce CNN to extract local features from time series data; Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal and random components; Design a hybrid loss function that combines mean square error and mean absolute error and balances the model's sensitivity to different errors by adjusting weights; Use the preprocessed data to train the deep learning model; monitor the loss function value during the training process; The LSTM network model and the CNN network model are fused through the weighted average algorithm to obtain a deep time series energy prediction model, namely, an energy prediction model; It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0063] The embodiment of the energy prediction and analysis method provided in the embodiment of the present invention belongs to the same inventive concept as the data acquisition energy management system of the above-mentioned embodiments. For details not described in detail in the embodiment of the energy prediction and analysis method, reference can be made to the embodiment of the above-mentioned data acquisition energy management system.

[0064] Although the present invention has been described in detail with reference to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, a person of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions shall be within the scope of the present invention. Any person of ordinary skill in the art may easily think of changes or substitutions within the technical scope disclosed by the present invention, and these shall be within the scope of protection of the present invention.

Claims

1. A data collection energy management system, characterized in that: Includes perception layer, data layer, application layer and presentation layer; The perception layer includes sensors and data acquisition gateways that collect energy consumption information and environmental parameter information; the data acquisition gateway collects data measured by sensors or real-time operation data of acquisition equipment and transmits it to the data layer according to the set protocol; The data layer receives data transmitted by the data acquisition gateway, normalizes and standardizes the transmitted data through the data processing engine inside the data layer and stores it in the database; at the same time, the data layer trains the energy prediction model, which is used to convert the acquired data into data for predicting energy usage trends, anomaly detection and energy-saving recommendations; The application layer monitors energy consumption in real time and performs load balancing and energy optimization control by controlling the operation of energy devices to achieve optimal energy utilization and reduce energy consumption; The presentation layer provides a user interface and visual reports to display energy forecast data and analysis results.

2. The data collection energy management system according to claim 1, characterized in that: The process of training the energy prediction model at the data layer includes: Obtain historical energy data, including energy consumption data, energy production data, energy price data, and environmental parameter data; Clean historical energy data to remove outliers and missing values; Extract trend, seasonality, and cyclical features from time series data; Standardize the data; Use LSTM networks to learn long-term dependencies in time series data; Introduce CNN to extract local features from time series data; Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal and random components; Design a hybrid loss function that combines mean square error and mean absolute error and balances the model's sensitivity to different errors by adjusting weights; Use the preprocessed data to train the deep learning model; monitor the loss function value during the training process; The LSTM network model and the CNN network model are fused through the weighted average algorithm to obtain a deep time series energy prediction model.

3. The data collection energy management system according to claim 2, characterized in that: The specific process of energy optimization control at the application layer is as follows: Generate energy consumption reports based on energy data received by the data layer to identify anomalies and waste in energy use; Monitor the operating status of energy equipment; Dynamically adjust the load distribution of energy equipment based on real-time energy consumption data and equipment operating status. Intelligent algorithms are used to optimize the operation strategies of energy equipment to achieve load balancing and energy optimization control.

4. The data collection energy management system according to claim 3 is characterized in that: The specific process of energy optimization control at the application layer also includes: Dynamically adjust the equipment's operating status and load distribution based on real-time load demand and equipment capabilities; The target formula for load balancing is as follows: Load balancing rate = (actual load / total equipment capacity) × 100%.

5. The data collection energy management system according to claim 4, characterized in that: The application layer uses the particle swarm optimization algorithm to minimize energy consumption as the optimization goal and the balance of energy supply and demand of energy equipment as the constraint condition to optimize the operation strategy of energy equipment and realize load balancing and energy optimization control; according to the output results of the particle swarm optimization algorithm, the operation strategy of energy equipment is formulated, and the strategy is applied to the energy equipment and executed; Monitor the execution effect of the strategy in real time and adjust the strategy parameters based on the feedback.

6. The data collection energy management system according to claim 5, characterized in that: The specific process of using particle swarm optimization algorithm to optimize the operation strategy of energy equipment at the application layer is as follows: Initialize the particle swarm to generate a set of random potential solutions as particles; Calculate the fitness value of each particle according to the fitness function; Compare the current fitness value with the individual historical optimal solution and update the individual optimal solution; Compare the current fitness value with the global historical optimal solution and update the global optimal solution; Adjust the speed and position of the particles according to the update rule to approach the individual optimal solution and the global optimal solution; Repeatedly calculate fitness, update individual and global optimal solutions, and update particle speed and position until the stopping condition is reached.

7. The data collection energy management system according to claim 6, characterized in that: Objective function describing the energy optimization goal: min(energy consumption cost) = min(∑(energy consumption × energy price)).

8. The data collection energy management system according to claim 1, characterized in that: The data acquisition gateway encrypts the data and transmits the data to the data layer; The data layer is provided with a server, and the server has several initial keys built in. Each initial key includes a key serial number and a key value corresponding to the key serial number. When the data acquisition gateway needs to send data to the data layer, it sends a key request to the server. The server extracts several initial keys based on the received request and returns them to the data acquisition gateway. When the data acquisition gateway transmits data, it extracts an initial key from the several initial keys received, encrypts the data with the key value of the initial key, and encrypts the key serial number of the initial key to generate encrypted data. The encrypted data and the extracted key serial number of the initial key used to encrypt the data are sent to the data layer together.

9. The data collection energy management system according to claim 1, characterized in that: After receiving the data, the data layer obtains the corresponding key value for decryption based on the received key serial number, verifies the encrypted key serial number with the key value, and decrypts the data based on the decrypted key value to obtain the corresponding data after the verification.

10. An energy forecasting and analysis method, characterized in that: The steps include: Receive data transmitted by the data acquisition gateway, normalize and standardize the transmitted data through the data processing engine inside the data layer, and store it in the database; Determine whether the energy prediction model has been trained; If so, the acquired data is processed and input into an energy forecasting model to convert the acquired data into data for predicting energy usage trends, anomaly detection, and energy-saving recommendations; If not, if not, obtain historical energy data and train an energy prediction model using LSTM network and CNN network based on the acquired historical data; Execution steps: process the acquired data and input it into the energy prediction model to convert the acquired data into data for predicting energy usage trends, anomaly detection and energy-saving recommendations; The steps of acquiring historical energy data and training an energy prediction model using an LSTM network and a CNN network based on the acquired historical data include: Clean historical energy data to remove outliers and missing values; Extract trend, seasonality, and cyclical features from time series data; Standardize the data; Use LSTM networks to learn long-term dependencies in time series data; Introduce CNN to extract local features from time series data; Before the output layer of the model, a time series decomposition module is added to decompose the forecast results into trend, seasonal and random components; Design a hybrid loss function that combines mean square error and mean absolute error and balances the model's sensitivity to different errors by adjusting weights; Use the preprocessed data to train the deep learning model; monitor the loss function value during the training process; The LSTM network model and the CNN network model are fused through the weighted average algorithm to obtain a deep time series energy prediction model, namely the energy prediction model.

Images