An Energy Management and Control Prediction Method Based on Quantum Deep Learning

Abstract

The present invention relates to an energy management and control prediction method based on quantum deep learning, which realizes accurate prediction and efficient management and control of the energy system in a confined space. The method includes the following steps: constructing a quantum mechanics model and a virtual reality model of the energy system in the confined space; establishing a quantum deep learning platform that collaborates with the quantum mechanics model, the virtual reality model, and the knowledge graph; the quantum deep learning platform collects and integrates the Internet of Things data of the energy system in the confined space; calculating the physical structure, fluid, and particle states in the confined space according to the knowledge graph and the Internet of Things data; predicting and analyzing the safety and energy consumption of the confined space according to the calculation results; formulating a management and control strategy and establishing an updated knowledge graph according to the prediction results; integrating the management and control strategy, the updated knowledge graph, and the quantum deep learning platform into the energy system in the confined space; and optimizing the energy system in the confined space.

Description

Technical Field

[0001] The present invention relates to the field of energy management, and particularly to an energy control and prediction method based on quantum deep learning. By means of quantum deep learning, efficient and accurate prediction of energy data is carried out, providing a scientific basis and decision-making support for energy control and management. Background Art

[0002] Traditional energy prediction methods mainly rely on statistical analysis and machine learning techniques. However, when dealing with large-scale, high-dimensional, non-linear and time-series energy data, problems such as low computational efficiency and insufficient prediction accuracy exist.

[0003] The present invention proposes an energy control and prediction method based on quantum deep learning, aiming to achieve precise prediction and efficient control of the energy system in a confined space. This method combines the high efficiency of quantum computing and the automatic feature extraction ability of deep learning. By constructing a quantum neural network (QNN) and a quantum recurrent neural network (QRNN), key features are extracted from historical and real-time collected data, and long-term dependencies in the time series are captured, so as to accurately predict the safety and future energy consumption of the energy system in the confined space. In addition, the present invention also introduces a quantum optimization algorithm to optimize the model parameters, improving the accuracy and robustness of the prediction model.

[0004] The present invention constructs a quantum deep learning platform and uses structures such as a spatio-temporal hybrid model, a state-time varying equation, energy spectrum analysis, particle position, a quantum neural network and a quantum recurrent neural network to process complex non-linear relationships and time series dependencies in energy data. This method not only considers the motion and interaction of microscopic particles in the energy system, but also combines real-time operation data to achieve high-precision prediction of the operating state of the energy system. The application of the present invention will promote the innovation of energy control technology, providing new technical means and solutions for the safe, intelligent and efficient operation of energy systems in confined spaces such as buildings, railway carriages, aircraft cockpits, and clean workshops, and having important practical theoretical significance and application value. Summary of the Invention

[0005] Technical Problems to be Solved

[0006] It solves the problems that traditional energy prediction methods mainly rely on statistical analysis and machine learning techniques, but when dealing with large-scale, high-dimensional, non-linear and time-series energy data, there are problems of low computational efficiency and low prediction accuracy. An energy control and prediction method based on quantum deep learning realizes accurate prediction of energy data and provides more scientific and efficient decision-making support for energy control in confined spaces.

[0007] Technical Solutions

[0008] The present invention provides a photovoltaic power generation prediction method based on a machine vision prediction body, comprising the following steps:

[0009] Step 1: Establish a quantum mechanics model and a virtual reality model of a closed - space energy system; the energy system includes a power system, a heating system, a ventilation and air - conditioning system. The quantum mechanics model contains all key components of the energy system and their interaction relationships, as well as quantum mechanics parameters describing these components and relationships; the quantum mechanics model is established and analyzed through quantum mechanics calculation software or algorithms.

[0010] Step 2: Establish a quantum deep - learning platform integrating the quantum mechanics model, the virtual reality model and a knowledge graph; the quantum deep - learning platform includes a computer program stored therein, and when the computer program is executed by a processor, it realizes an energy management and control prediction method based on quantum deep - learning. The construction of the quantum deep - learning platform also includes designing the architecture of the quantum neural network, the number of quantum neurons and connection methods in the input layer, hidden layer and output layer; using quantum algorithms, quantum Fourier transform, quantum phase estimation, to realize the weight update and activation function calculation between quantum neurons; combining the quantum neural network with deep - learning algorithms to improve the accuracy of model prediction; the knowledge graph includes a knowledge base of a computational fluid dynamics model; construct a knowledge base of a semantic network for entities, data and inter - relationships of the closed - space energy system, and predict potential risks and energy consumption; visually display the prediction results, such as drawing an energy demand prediction curve, an energy consumption distribution map; further optimize and adjust the energy management and control strategy according to the visual results.

[0011] Step 3: The quantum deep - learning platform collects and integrates Internet of Things data of the closed - space energy system; the data of the closed - space energy system collected through the Internet of Things is input into the quantum deep - learning platform for real - time prediction and analysis, and according to the prediction results, dynamically adjust the energy management and control strategy.

[0012] Step 4: Calculate the physical structure, fluid and particle state of the closed space according to the knowledge graph and Internet of Things data.

[0013] Step 5: Predict and analyze the safety and energy consumption of the closed space according to the calculation results; use the quantum deep - learning platform to diagnose and warn of potential risks in the closed - space energy system, and improve the reliability of the system.

[0014] Step 6: According to the prediction results, formulate a management and control strategy and establish an updated knowledge graph.

[0015] Step 7: Integrate the management and control strategy, the updated knowledge graph and the quantum deep - learning platform into the closed - space energy system.

[0016] Step Eight: Optimize the energy system in the enclosed space; the optimization is carried out to optimize the operating efficiency of the energy system in the enclosed space, reduce energy consumption and improve safety. Optimizing the energy system in the enclosed space includes training the model using the quantum gradient descent algorithm or the quantum stochastic gradient descent algorithm to minimize the prediction error; leveraging the advantages of quantum parallel computing to accelerate the model training process; and evaluating the generalization ability of the model through methods such as quantum cross-validation to avoid overfitting.

[0017] Beneficial Effects: By constructing a quantum deep learning platform, the present invention utilizes structures such as a spatio-temporal hybrid model, a state-time varying equation, energy spectrum analysis, particle position, a quantum neural network, and a quantum recurrent neural network to process complex non-linear relationships and temporal dependencies in energy data. This method not only considers the motion and interaction of microscopic particles in the energy system but also combines real-time operation data to achieve intelligent control and high-precision prediction of the operating state of the energy system in the enclosed space. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The present invention will be described in detail below with reference to the accompanying drawings.

[0020] As Figure 1 shown, the present invention provides an energy management and prediction method based on quantum deep learning, including the following steps:

[0021] Step One: Establish a quantum mechanics model and a virtual reality model of the energy system in the enclosed space. The energy system includes a power system, a heating system, a ventilation and air conditioning system. The quantum mechanics model contains all the key components of the energy system and their interaction relationships, as well as the quantum mechanics parameters describing these components and relationships; the quantum mechanics model is established and analyzed through quantum mechanics calculation software and algorithms.

[0022] Construct a quantum mechanics model of the energy system in the enclosed space, system component identification and parameterization: Identify all the key components of the energy system, and according to quantum mechanics theory, assign corresponding quantum mechanics parameters to each component, such as wave functions, energy levels, transition probabilities, etc., to accurately describe its physical characteristics and interactions.

[0023] Based on the identified components and their parameters, construct a quantum mechanics model, and use the Schrödinger equation to describe the motion and energy distribution of microscopic particles in the energy system, reflecting the overall operating state of the system.

[0024] A virtual reality model of the energy system in a confined space is established. Based on the prediction results of the model and the online comparison and analysis with historical and measured data, the hidden risks of the fluid in the physical model are predicted, and fault location is carried out, realizing "fault early warning and accident predictability", thus improving the refined management of the entire life cycle of the energy system. A control and management method for the whole process application of the energy system in a confined space is established, and the standards of mapping interaction, algorithm prediction and optimization control structure are defined. Through comparison with physical entities and real-time analysis, continuous correction is carried out to become a model that is synchronized with physical entities and can be managed, with self-learning function, and real-time data updates the virtual reality model.

[0025] Step 2: Establish a quantum deep learning platform that includes a quantum mechanics model, a virtual reality model, and a knowledge graph collaboration. The quantum deep learning platform includes designing a quantum neural network architecture, an input layer, a hidden layer, and an output layer. The input layer receives preprocessed energy data. The hidden layer uses quantum neurons to utilize the superposition and entanglement characteristics of quantum states to achieve nonlinear transformation and feature extraction of data. The output layer outputs the prediction results. The hidden layer uses quantum neurons to utilize the superposition and entanglement characteristics of quantum states to achieve nonlinear transformation and feature extraction of data. It includes a quantum recurrent neural network (QRNN) or a quantum long short-term memory network (QLSTM) to better handle the time series dependence relationship in energy data, realize parallel processing of time series data, and improve the prediction ability of the model.

[0026] Use quantum algorithms such as quantum Fourier transform (QFT) and quantum phase estimation to optimize model parameters, accelerate the training process, and improve the training efficiency and prediction accuracy of the model.

[0027] The quantum deep learning platform includes a quantum neural network (QNN) as the core model, which uses the parallelism and high efficiency of quantum computing to process large-scale and high-dimensional energy data. Combining quantum machine learning algorithms, quantum support vector machine (QSVM) and quantum random forest (QRF) to enhance the prediction ability of the model.

[0028] Utilize the automatic feature extraction ability of the quantum deep learning platform to extract key features from historical data, and use the quantum approximate optimization algorithm (QAOA) to optimize model parameters, improving the learning efficiency and prediction accuracy of the model.

[0029] Time series prediction and long-term dependence capture introduce a quantum recurrent neural network (QRNN) or a quantum long short-term memory network (QLSTM) to capture the long-term dependence relationship in the time series. Utilize the characteristics such as quantum entanglement and superposition states to improve the sensitivity and prediction accuracy of the model to time series changes.

[0030] The quantum deep learning platform includes a computer program stored. When the computer program is executed by a processor, it realizes an energy management and control prediction method based on quantum deep learning. The program develops energy management and control prediction software based on quantum mechanics, realizing functions such as model construction, quantum state analysis, and prediction result output. The system central database adopts application modes such as NoSQL databases, business databases, and topic databases. Among them, the NoSQL database selects the distributed database MongoDB, and the business database and topic database select the relational database Oracle. Spark is compatible with Hadoop's distributed file system (such as HDFS), and processes a wider range of data and application scenarios by expanding Hadoop's ecosystem. Spark uses Hadoop as the underlying storage system and at the same time utilizes Hadoop's resources for more efficient parallel computing.

[0031] The construction of the quantum deep learning platform also includes designing the architecture of the quantum neural network, the number of quantum neurons and connection methods in the input layer, hidden layer, and output layer; using quantum algorithms, quantum Fourier transform, and quantum phase estimation to realize the weight update and activation function calculation between quantum neurons; combining the quantum neural network with deep learning algorithms to improve the accuracy of model prediction.

[0032] Using the quantum mechanics model, calculate the energy spectrum of the energy system, the distribution of all possible energy levels in the system. By analyzing the time-varying equation of the quantum state, predict the trend of the energy system's evolution over time, including the changes in physical quantities such as the position, velocity, and acceleration of microscopic particles in the system, as well as the interactions. Using the transition probability theory in quantum mechanics, predict the possibility of particle transitions between energy levels in the energy system, and evaluate the stability and reliability of the system under different operating conditions.

[0033] Construct a knowledge graph: Generate corresponding classes for the equipment, sensors, etc. in the heating and air-conditioning system of the enclosed space; connect them with object properties according to the equipment organizational structure in the system; map the collected real-time data to the corresponding classes through data properties; define the definition of the domain ontology, including the construction of classes and properties, to describe the equipment, sensors, and their mutual relationships in the system; define anomaly detection rules, and the specific rules are used to detect operating anomalies such as too high supply air temperature of the air handling unit; use the constructed knowledge graph to run the anomaly detection semantic rules; when the semantic rules are activated, output the corresponding conclusions, including the relevant information of the abnormal components and the type of abnormal problems; realize the prediction and diagnosis of the safety of the power, heating, ventilation, and air-conditioning systems in the enclosed space.

[0034] The knowledge graph includes a knowledge base of computational fluid dynamics models; constructs a knowledge base of semantic networks for the entities, data, and interrelationships of the energy system in the enclosed space, and predicts latent risks and energy consumption; visually displays the prediction results, such as drawing an energy demand prediction curve and an energy consumption distribution map; further optimizes and adjusts the energy management strategy according to the visual results.

[0035] Step 3: The quantum deep learning platform collects and integrates the Internet of Things data of the energy system in the enclosed space, inputs the data into the quantum deep learning platform, performs real-time prediction and analysis, and dynamically adjusts the energy management strategy according to the real-time prediction results.

[0036] Data collection and preprocessing: Clean the Internet of Things data, remove outliers and duplicate data, and perform data normalization and standardization processing to improve data quality and prediction accuracy. Convert the data into a form suitable for model training.

[0037] Step 4: Calculate the physical structure, fluid, and particle state of the enclosed space based on the knowledge graph and the Internet of Things data; use the CNN-LSTM-SVM spatio-temporal hybrid model algorithm to establish a standard framework for the model, improve the computing power of the prediction and management function of the energy system in the enclosed space on the Huawei Cloud, reduce the model training time, and improve the technical optimization level and efficiency. Through the trained CNN-LSTM-SVM network, it can automatically extract local features within the effective area and sequential features between regions, enhance the prediction effect, and improve the prediction accuracy. First, use the CNN model to extract the spatial features of the data, then the LSTM model extracts the temporal features of the data, and then update the SVM classifier according to the trained CNN-LSTM network. According to the trained CNN-LSTM network and the new SVM classifier, a trained CNN-LSTM-SVM prediction network is constructed.

[0038] Based on the device information and historical energy consumption data in the knowledge graph, construct a latent risk and energy consumption prediction model. Use graph database technology to convert the collected data and defined ontology into a knowledge graph. The nodes in the graph represent entities, and the edges represent the relationships between entities. The ontology definition includes device classes, which describe various devices in the heating and air conditioning systems; sensor classes, which describe various sensors installed in the system; location classes, which describe the physical locations of devices and sensors in the system; data point classes, which describe data collection points; safety and energy efficiency problem classes, which describe the energy efficiency problems and abnormalities in the system.

[0039] Step 5: Predict and analyze the safety and energy consumption of the enclosed space based on the calculation results. For safety, use a quantum deep learning platform to diagnose and warn of potential risks in the energy system of the enclosed space, improving the reliability of the system. In the knowledge graph, define and apply anomaly detection rules, such as detecting too high supply air temperature of the air handling unit or abnormal sensor data. Based on the anomaly detection results, conduct a risk assessment of the system's safety and predict possible failures or safety issues.

[0040] Energy consumption prediction: Draw the actual operating capacity diagram of the unit based on the basic parameters of the terminal air handling unit of the unit; Obtain the capacity diagram of the air handling unit of a fixed model in the preset database; Determine whether the deviation between the actual operating capacity diagram of the unit and the capacity diagram of the air handling unit of the fixed model meets the convergence conditions, where the convergence conditions include stable unit operation and reasonable motor efficiency; If the convergence conditions are met, output the hourly energy consumption prediction results of the air conditioning system according to the actual operating capacity diagram of the unit; Achieve accurate prediction of the energy consumption of the enclosed space and energy-saving analysis.

[0041] Step 6: Based on the prediction results, formulate control strategies and establish an updated knowledge graph. Real-time collect energy system data through the Internet of Things, transmit it to the quantum deep learning platform for correction and prediction. Use the quantum gradient descent algorithm to train the model and utilize the parallelism of quantum computing to accelerate the gradient calculation process.

[0042] Use the quantum cross-validation method to evaluate the generalization ability of the model. Divide the dataset into multiple subsets for training and validation respectively to evaluate the performance of the model on different datasets, avoid overfitting, and improve the generalization ability of the model. According to the results of quantum cross-validation, adjust the architecture and parameters of the model, such as the number of quantum neurons, connection methods, activation functions, etc., to optimize the model performance. According to the evaluation results, adjust the model architecture and parameters to optimize the model performance. According to the feedback results, dynamically adjust the energy control strategy to adapt to system changes.

[0043] Step 7: Integrate the control strategy, updated knowledge graph, and quantum deep learning platform into the energy system of the enclosed space. This includes detailed data on fluid flow in the enclosed space, the distribution of physical quantities such as velocity, temperature, humidity, and harmful substance concentration, as well as whether the air flow is uniform, whether there are local overheating or overcooling phenomena, whether the harmful substance concentration exceeds the standard, and whether there are phenomena such as air flow short circuits, vortices, and dead corners that are not conducive to air circulation and uniform temperature distribution. At the same time, evaluate whether the harmful substance concentration exceeds the safety standard to ensure that the indoor air quality meets health requirements. Integrate the prediction model based on quantum deep learning into the energy system to achieve intelligent control.

[0044] Step Eight: Optimize the operating efficiency of the enclosed space energy system, reduce energy consumption, and improve safety, including fluid momentum, heat, and mass equations, changes in fluid temperature or density, and the setting of flow velocity; fluid properties, density, viscosity, boundary conditions, inlet velocity, temperature, pressure, distribution of heat sources and cold sources, etc. At the same time, according to the actual usage of the enclosed space, set different operating conditions, such as indoor temperature set value, personnel activity intensity, outdoor climate conditions, etc.; optimize the enclosed space energy system by using quantum gradient descent algorithm or quantum stochastic gradient descent algorithm to train the model to minimize the prediction error; utilize the advantage of quantum parallel computing to accelerate the model training process; evaluate the generalization ability of the model through methods such as quantum cross-validation, avoid overfitting, adjust system parameters, and improve the air distribution design, etc.

[0045] It should be noted that the technical solutions of the present invention are only illustrated in combination with the above embodiments and not limited thereby. Therefore, the protection scope of the present invention shall be defined by the protection scope of the claims of this application.

Claims

1. An energy management and control prediction method based on quantum deep learning, characterized in that: The following steps are involved: Step 1: Construct a quantum mechanics model and a virtual reality model of the confined space energy system; Step 2: Establish a quantum deep learning platform that includes quantum mechanics models, virtual reality models and knowledge graphs; Step 3: The quantum deep learning platform collects and integrates IoT data of the confined space energy system; Step 4: Calculate the physical structure, fluid and particle state of the confined space based on the knowledge graph and IoT data; Step 5: Predict and analyze the safety and energy consumption of the confined space based on the calculation results; Step 6: Based on the prediction results, formulate management and control strategies and establish a new knowledge graph; Step 7: Integrate the management and control strategies, new knowledge graphs and quantum deep learning platforms into the confined space energy system; Step 8: Optimize the confined space energy system.

2. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The quantum mechanical model contains all the key components of the energy system and their interactions, as well as the quantum mechanical parameters that describe these components and relationships.

3. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The energy system includes an electric power system, a heating system, and a ventilation and air conditioning system.

4. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The quantum mechanics model is established and analyzed by quantum mechanics calculation software and algorithms.

5. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The aforementioned collection and integration of IoT data of confined space energy systems is to input the confined space energy system data collected through the IoT into the aforementioned quantum deep learning platform for real-time prediction and analysis, and dynamically adjust the energy management and control strategy based on the prediction results.

6. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The knowledge graph includes a knowledge base of computational fluid dynamics models, constructs a knowledge base of semantic networks for entities, data and mutual connections of confined space energy systems, and predicts hidden risks and energy consumption; visualizes the prediction results, draws energy demand prediction curves and energy consumption distribution maps; and further optimizes and adjusts energy management strategies based on the visualization results.

7. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The quantum deep learning platform includes a stored computer program, which, when executed by a processor, can implement the energy management and prediction method based on quantum deep learning described in any one of claims 1 to 6.

8. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The safety mentioned above is to use the quantum deep learning platform to diagnose and warn of hidden risks in confined space energy systems, thereby improving the reliability of the system.

9. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The quantum deep learning platform includes designing the architecture of the quantum neural network, the number of quantum neurons in the input layer, hidden layer and output layer, and the connection method; using quantum algorithms, quantum Fourier transforms, and quantum phase estimation to achieve weight updates and activation function calculations between quantum neurons; combining quantum neural networks with deep learning algorithms to improve the accuracy of model predictions.

10. The energy management and control prediction method based on quantum deep learning according to claim 1 is characterized in that: The optimization of the confined space energy system is to optimize the operating efficiency of the confined space energy system, reduce energy consumption and improve safety, including using a quantum gradient descent algorithm or a quantum stochastic gradient descent algorithm to train the model to minimize prediction errors; using the advantages of quantum parallel computing to accelerate the model training process; and evaluating the generalization ability of the model through quantum cross-validation to avoid overfitting.

Images