Energy consumption prediction method and system based on energy management system

Abstract

The invention discloses an energy consumption prediction method and system based on an energy management system, and relates to the technical field of energy management, and the method comprises the specific steps: S100, eye movement data collection: collecting multi-modal eye movement data through an eye movement tracking device during the course learning period of students. The comfort index and energy consumption data can be comprehensively collected, a rich data basis is provided for energy consumption prediction, the facial expression recognition camera and the voice collection microphone are utilized, the emotion calculation technology is combined, user emotion feedback data are obtained, the data dimension is enriched, energy consumption prediction is made to be closer to the actual demand of a user, and meanwhile the energy consumption prediction efficiency is improved. And subjective evaluation and adjustment behavior information of a user on the comfort level are collected through a mobile application program and an indoor control panel, and quantum state coding is performed on all data, so that deep integration and efficient processing of the data are realized.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, specifically to an energy consumption prediction method and system based on an energy management system. Background Technology

[0002] With the continuous growth of global energy demand and the increasing scarcity of energy resources, energy management has become an important issue in modern social development. In various fields, from industrial production to daily life, the effective use and conservation of energy are urgent problems to be solved. To meet this challenge, energy management systems have emerged. They help users optimize energy use strategies, reduce energy consumption, and improve energy efficiency by collecting, analyzing, and processing energy usage data. However, traditional energy management systems often rely on a single data source or simple data processing methods, making it difficult to comprehensively and accurately reflect the actual situation of energy use, and also failing to fully consider users' subjective needs and emotional feedback. Therefore, it is particularly important to develop a new type of energy management system that can comprehensively consider multi-source data, accurately predict energy consumption, and adjust management strategies in real time.

[0003] Traditional energy management systems have several shortcomings in energy consumption prediction and management. First, they often rely solely on limited sensor data, neglecting the importance of multi-source data such as user behavior, environmental factors, and user emotional feedback. This single-data-source approach limits the accuracy and comprehensiveness of predictions. Second, traditional systems typically employ classic algorithms for data processing, which are inefficient when dealing with large-scale, high-dimensional data and struggle to capture complex relationships between data points. Furthermore, traditional systems often lack real-time adjustment mechanisms, failing to adapt management strategies to changes in the external environment and user needs, resulting in low energy utilization efficiency. Finally, in terms of decision support, traditional systems often only provide simple energy consumption predictions without offering probability distributions for different energy consumption values, thus limiting the scientific rigor and flexibility of energy management decisions.

[0004] Therefore, developing energy consumption prediction methods and systems based on energy management systems provides a scientific and reasonable basis for decision-making in energy management systems, effectively promoting the efficient use and conservation of energy. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an energy consumption prediction method and system based on an energy management system. It acquires multi-dimensional data on environment, user behavior, and emotional feedback through multi-source data acquisition technology, and uses quantum computing technology for efficient data fusion and preprocessing. It uses quantum neural networks to construct a comfort and energy consumption correlation model to achieve accurate energy consumption prediction and establishes a real-time adjustment mechanism to adapt to changes in the external environment and user needs. Finally, it provides a scientific basis for energy management decisions based on quantum probability calculation, effectively improving energy utilization efficiency.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, an energy consumption prediction method based on an energy management system, the specific steps of which are as follows: S100, multi-source data acquisition: Sensors are deployed in the interior areas of the building to collect comfort indicators and energy consumption data; cameras and microphones are used to collect user facial images and voice data, and affective computing technology is used to obtain user emotional feedback; at the same time, through mobile applications and indoor control panels, subjective evaluations of user comfort and adjustment behavior information are collected, and the collected data is quantum-encoded. S200, Data Fusion and Preprocessing: Integrate multi-source data encoded by quantum state, remove abnormal data with the help of quantum entanglement detection mechanism, normalize the remaining data, and map it to the unit quantum state space; S300, Constructing a correlation model: A model relating comfort and energy consumption is built using a quantum neural network. The number of nodes in the input layer, hidden layer, and output layer is determined. The hidden layer uses the ReLU activation function. Information is transmitted between layers through quantum entanglement channels. The preprocessed dataset is divided into training set, validation set, and test set to train the model. S400, Real-time Model Adjustment: Real-time collection of user emotional feedback and comfort adjustment behavior data and encoding them into quantum states, calculation of the similarity between the quantum states of the data and the original data, and when the similarity is lower than the threshold, obtaining the parameter adjustment scheme with the smallest prediction error, and updating the model parameters in real time according to the adjustment scheme. S500, Energy Consumption Prediction and Decision Making: Based on the adjusted model, the energy consumption prediction results of the quantum state are obtained, the probability distribution of different energy consumption values ​​is calculated, and the optimal energy supply plan and equipment operation strategy are selected according to the probability distribution, taking into account the energy supply situation and cost factors.

[0007] Furthermore, in S100, the sensors used in the multi-source data acquisition are: a temperature sensor to collect real-time temperature data of different areas in the building, a humidity sensor to obtain indoor air humidity information, a light sensor to sense indoor light intensity, a smart meter to record the power consumption of different electrical devices in different time periods of the building, and a gas meter to monitor the gas usage in real time.

[0008] Furthermore, in S100, the steps of acquiring user facial images and voice data using a camera and microphone in multi-source data acquisition, and obtaining user emotional feedback using emotion computing technology are as follows: acquiring user facial images through a camera, acquiring user voice signals in real time through a microphone array, performing preprocessing and standardization, extracting relevant features, and calculating facial expression scores, voice feature scores, and cross-modal interaction scores respectively; calculating the probability distribution of the user in various emotional states through an emotion state probability distribution algorithm; selecting the emotion category with the highest probability as the final recognition result; and outputting the corresponding emotion category.

[0009] Furthermore, in step S200, during data fusion and preprocessing, the probability distribution of the user's various emotional states is calculated using an emotional state probability distribution algorithm, with the following formula: ,in It is the first The probability of a certain emotional state. It is the softmax function, which converts the score into a probability distribution. The calculation formula is: , The raw score entered. It is the total number of emotional states. Represents the natural constant. These are learnable fusion weight parameters, and , It is a facial expression feature score. It is a speech feature score. It is a score that captures the interaction between facial and voice features.

[0010] Furthermore, in step S200, a chaotic fusion algorithm is used in the data fusion and preprocessing to integrate multi-source data, as shown in the formula: ,in This is the output of the chaos fusion algorithm. It is the total number of data sources participating in the integration. Indicates the first Data encoded by a quantum state These are the weighting coefficients. It is a hyperbolic tangent activation function used to perform nonlinear transformations on the fused data to enhance feature representation capabilities.

[0011] Furthermore, in S300, the quantum neural network model in the correlation model is constructed, and the input layer receives the fused quantum state data. The hidden layer uses quantum neurons, and the calculation formula for each quantum neuron is: ,in No. The output quantum state form of a quantum neuron To connect the first The input and the first The quantum weights of neurons in a hidden layer. Represents the quantum tensor product. This is a quantum rotation gate operation used to transform the input. Is it input to the number The number of data points per quantum neuron No. The quantum state data is input to the quantum neuron. Information is transmitted between hidden layers through quantum entanglement channels. The output layer uses quantum measurement operations to convert the quantum state output into a classical energy consumption prediction value. The formula is: , This is a predicted energy consumption value. The quantum state results of the output layer, quantum measurement operations, This means converting a quantum state into classical data.

[0012] Furthermore, in S300, the training of the comfort-energy consumption correlation model in the correlation model is constructed: (1) Divide the preprocessed comprehensive dataset into training set, validation set and test set according to the proportion, divide the training set into batches and prepare the label data; (2) Randomly initialize the weight matrix and bias vector of the neural network, set the learning rate and momentum parameters, and initialize the quantum entanglement channel parameters at the same time; (3) The input layer receives preprocessed data, the hidden layer processes information through the ReLU activation function, information is transmitted between layers through the quantum entanglement channel, and the output layer calculates the predicted value using the linear activation function. (4) Calculate the mean square error between the predicted value and the true value as the loss function, and add a regularization term; (5) Update the neural network weights and quantum entanglement channel parameters by calculating the gradient of the loss function with respect to each parameter; (6) Evaluate the model performance on the validation set every 100 training iterations, record the loss and accuracy and plot the learning curve, adjust the strategy according to the validation results, and finally evaluate the model's generalization ability on the test set.

[0013] Furthermore, in S400, the step of real-time adaptive updating of the parameters of the comfort and energy consumption correlation model in real-time model adjustment is as follows: when new user emotional feedback and comfort adjustment behavior data are obtained, they are encoded into quantum states. The formula for calculating the quantum state similarity between real-time data and original data is: ,in Quantum state similarity value measures the degree of similarity between real-time data and historical data. It is the quantum state of historical data. It is the quantum state of the newly acquired data. The similarity is the inner product of two quantum states, and the square of the modulus yields the similarity, which ranges from [0, 1]. When the similarity is below a similarity threshold... When the time is right, the model update is triggered, and the prediction error of each set of parameters is calculated to obtain the parameter scheme with the minimum error. The quantum weights of the model are then updated through a quantum state superposition update operation, as shown in the formula: ,in It is the updated quantum weight. It is an update coefficient, with a value range of [0, 1], which controls the fusion ratio of historical weights and new weights. It is the historical quantum weight before the update. These are new quantum weight candidate values ​​obtained through quantum parallel computing.

[0014] Furthermore, in S500, the probability distribution of different energy consumption values ​​is calculated using quantum probability in energy consumption prediction and decision-making, with the following formula: ,in Energy consumption value The probability of occurrence These are quantum states corresponding to different energy consumption values, generated by discretization within a preset energy consumption range. It is the energy consumption prediction result of the quantum state output by the model.

[0015] On the other hand, an energy consumption prediction system based on an energy management system includes: a multi-source data acquisition module, a data fusion and preprocessing module, a quantum neural network modeling module, a real-time model adjustment module, and an energy consumption prediction and decision-making module. The multi-source data acquisition module: deploys temperature, humidity, and light sensors, as well as smart meters and gas meters to collect environmental and energy consumption data; captures user emotional feedback through facial expression recognition cameras and voice microphone arrays; and collects user comfort evaluations and equipment adjustment behaviors in conjunction with mobile applications and indoor control panels. All data is transmitted to the data fusion and preprocessing module after being quantum-encoded. The data fusion and preprocessing module integrates multi-source data encoded in quantum states, performs feature fusion using a chaotic fusion algorithm, identifies and removes outliers through a quantum entanglement detection mechanism, normalizes the remaining data, and maps it to a unit quantum state space. The quantum neural network modeling module constructs a quantum neural network containing an input layer, a hidden layer, and an output layer. The input layer receives preprocessed quantum state data, the hidden layer processes information through quantum neurons and quantum entanglement channels, and the output layer converts the quantum state into an energy consumption prediction value. The dataset is divided into a training set, a validation set, and a test set, and the model parameters are iteratively optimized using a stochastic gradient descent algorithm. The real-time model adjustment module: collects new user feedback and behavior data in real time and encodes them into quantum states, calculates the similarity between the quantum states of the data and historical data, triggers model updates when the similarity is lower than a threshold, evaluates multiple parameter adjustment schemes using quantum parallel computing, selects the optimal scheme and updates the model weights through quantum state superposition operations; The energy consumption prediction and decision-making module generates energy consumption prediction results for quantum states based on the adjusted model, calculates the probability distribution of different energy consumption values, and the energy management system combines real-time energy prices and renewable energy power generation factors to generate the optimal energy supply plan and equipment operation strategy.

[0016] Compared with existing technologies, this energy consumption prediction method and system based on an energy management system has the following advantages: I. This invention, by deploying multiple sensors within building interiors, comprehensively collects comfort indicators and energy consumption data, providing a rich data foundation for energy consumption prediction. Utilizing facial expression recognition cameras and voice acquisition microphones, combined with affective computing technology, it acquires user emotional feedback data, enriching the data dimensions and making energy consumption predictions more closely aligned with users' actual needs. Simultaneously, it collects users' subjective evaluations of comfort and adjustment behaviors through mobile applications and indoor control panels, and encodes all data into quantum states, achieving deep data integration and efficient processing. By employing a quantum entanglement detection mechanism to eliminate outlier data, normalizing the remaining data, and mapping it to a unit quantum state space, it improves the accuracy and reliability of the data, providing a data foundation for subsequent model training and prediction.

[0017] II. This invention utilizes a quantum neural network to build a model linking comfort and energy consumption, achieving accurate modeling of the complex relationship between comfort and energy consumption. It establishes a real-time model adjustment mechanism that collects user emotional feedback and comfort adjustment behavior data in real time, encodes them as quantum states, calculates their similarity to the original data's quantum states, and when the similarity is below a threshold, uses quantum parallel computing to obtain the parameter adjustment scheme with the minimum prediction error. Through quantum state superposition operations, the model's quantum weights are updated adaptively in real time, enabling the model to continuously adapt to changes in the external environment and user needs, thus maintaining high accuracy and reliability. Finally, based on the adjusted model, this invention can generate quantum state energy consumption prediction results and calculate the probability distribution of different energy consumption values ​​using quantum probability, providing a scientific and reasonable decision-making basis for energy management systems and achieving efficient energy utilization and conservation.

[0018] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1 This is a framework diagram of an energy consumption prediction method based on an energy management system. Figure 2 This is a flowchart of an energy consumption prediction system based on an energy management system. Detailed Implementation

[0021] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0022] Example 1: Energy consumption prediction and optimization for smart office buildings.

[0023] A smart office building encompasses multiple functional areas including offices, meeting rooms, and server rooms. It needs to balance energy consumption and comfort under dynamic user behaviors (such as office hours and meeting frequency) and environmental variables (such as seasonal changes and light intensity). The system constructs an adaptive energy consumption prediction system through multi-source sensor networks, quantum state data processing, and quantum neural network models. Figure 2 As shown.

[0024] Multi-source data acquisition and quantum state encoding: Deploy temperature, humidity, and light sensors and smart meters in various areas to collect real-time environmental parameters (such as temperature). ,humidity Light intensity ) and equipment energy consumption data (such as electricity consumption) Gas consumption It uses facial expression recognition cameras and microphone arrays to capture users' facial images and voice signals, and extracts facial expression scores through emotion computing technology. Speech feature score and cross-modal interaction score And calculate the probability distribution of users' emotional states using a formula. The formula is: ,in It is the first The probability of a certain emotional state. It is the softmax function, which converts the score into a probability distribution. The calculation formula is: , The raw score entered. It is the total number of emotional states. Represents the natural constant. These are learnable fusion weight parameters, and All collected data is converted into qubit form through quantum state encoding. For example, mapping continuous environmental parameters to quantum superposition states and encoding user behavior data to quantum entangled states, such as... Figure 1 As shown.

[0025] Data fusion and preprocessing: Abnormal data (such as sudden energy consumption changes caused by equipment failure) is identified and removed through a quantum entanglement detection mechanism. Valid data is retained and normalized, mapped to a unit quantum state space, and then integrated using a chaotic fusion algorithm. The formula is as follows: ,in This is the output of the chaos fusion algorithm. It is the total number of data sources participating in the integration. Indicates the first Data encoded by a quantum state These are the weighting coefficients. It is a hyperbolic tangent activation function used to perform nonlinear transformations on the fused data, enhancing the feature representation ability. By enhancing the nonlinear correlation between features through the hyperbolic tangent activation function, the fused quantum state is generated. .

[0026] Quantum Neural Network Model Construction and Training: Constructing a three-layer quantum neural network, with the input layer receiving the fused quantum state data. It includes multi-dimensional features such as environment, energy consumption, and user feedback. The hidden layer uses quantum neurons, and the calculation formula for each quantum neuron is: ,in No. The output quantum state form of a quantum neuron To connect the first The input and the first The quantum weights of neurons in a hidden layer. Represents the quantum tensor product. This is a quantum rotation gate operation used to transform the input. Is it input to the number The number of data points per quantum neuron No. Quantum state data input to a quantum neuron is manipulated through a quantum rotation gate. The input is transformed, and information is transmitted between layers through quantum entanglement channels to enhance feature interaction. The output layer converts the quantum state into a classical energy consumption prediction value through a quantum measurement operation. The formula is: , This is a predicted energy consumption value. The quantum state results of the output layer, quantum measurement operations, This means converting a quantum state into classical data.

[0027] The preprocessed dataset is divided into training, validation, and test sets. The model parameters are optimized using the stochastic gradient descent algorithm. The loss function is the mean squared error between the predicted and true values. A regularization term is introduced to prevent overfitting. The performance is evaluated on the validation set after several training batches, and the learning curve is plotted to adjust the training strategy.

[0028] The model adaptively adjusts in real time: New user feedback and behavioral data are collected in real time and encoded as quantum states |Xnew>. The similarity S between this quantum state and historical data quantum states |X> is calculated using the following formula: ,in Quantum state similarity value measures the degree of similarity between real-time data and historical data. It is the quantum state of historical data. It is the quantum state of the newly acquired data. The similarity is the inner product of two quantum states, squared by taking the modulus, and its value ranges from [0, 1]. When S is below a preset threshold... When the model is updated, multiple sets of candidate parameter values ​​are generated through quantum parallel computing. The prediction error of each set of parameters is calculated, the scheme with the smallest error is selected, and the model weights are updated using quantum state superposition operations. The formula is as follows: ,in It is the updated quantum weight. It is an update coefficient, with a value range of [0, 1], which controls the fusion ratio of historical weights and new weights. It is the historical quantum weight before the update. The new quantum weight candidate values ​​are obtained through quantum parallel computing, which control the fusion ratio of historical weights and new weights.

[0029] Energy consumption prediction and decision optimization: Based on the adjusted model output of quantum state energy consumption prediction result |Pq>, the probability distribution of different energy consumption values ​​Pk is calculated by formula. Combined with energy supply conditions (such as grid peak and valley periods, renewable energy output) and cost factors, the optimal energy supply plan under the probability distribution is selected. For example, when a high probability of peak energy consumption is predicted, the air conditioning set temperature is adjusted in advance, non-critical equipment is operated off-peak, and the peak load is reduced.

[0030] In summary, in the context of intelligent office buildings, the energy consumption prediction method based on an energy management system integrates multi-source data acquisition and quantum state encoding, combining environmental, energy consumption, and user feedback information. It utilizes chaotic fusion algorithms and quantum neural networks to construct a dynamic correlation model. Through quantum state similarity calculation and real-time parameter update mechanisms, the model can adapt to dynamic factors such as personnel flow and seasonal changes in the office environment. Ultimately, it optimizes energy scheduling based on energy consumption probability distribution, achieving a balance between comfort and energy costs. This solution leverages the parallelism and entanglement characteristics of quantum computing to improve prediction accuracy and response speed, providing a scientifically feasible technical path for the low-carbon and intelligent management of office buildings.

[0031] Example 2: Energy consumption prediction and emergency management in smart hospitals.

[0032] A top-tier hospital, including operating rooms, ICUs, and laboratories, has areas with extremely high requirements for environmental stability. It needs to ensure the continuous operation of medical equipment while responding to sudden energy fluctuations (such as power grid failures and extreme weather). The system integrates medical environmental parameters, equipment operating status, and emergency energy data to build an energy consumption prediction and emergency decision-making system with real-time response capabilities.

[0033] Multi-source data acquisition and quantum state encoding: Deploying high-precision sensors in clean areas (such as operating rooms and ICUs) to collect temperature and humidity data. air quality Data; real-time energy consumption of medical equipment (such as ventilators and imaging equipment) monitored through smart meters. And connect to backup power (such as diesel generator fuel level). UPS battery power Medical staff can adjust environmental parameters (such as operating room temperature) through the control panel. The system records adjustment behavior data and uses speech recognition technology to capture medical staff's feedback on the environment (such as "ICU humidity needs adjustment"). Feedback features are generated through emotion computing, preprocessed and standardized, relevant features are extracted, and facial expression scores, speech feature scores, and cross-modal interaction scores are calculated. The probability distribution of the user's various emotional states is calculated using an emotion state probability distribution algorithm, with the formula: The system selects the emotion category with the highest probability as the final recognition result and outputs the corresponding emotion category. The collected data is then quantum-encoded. After quantum-state encoding, all data are represented by high-dimensional quantum states for medical sensitive parameters (such as operating room temperature and humidity) to ensure that minute fluctuations can be accurately captured. The backup power supply state is encoded as a quantum entangled state, which is associated with the grid load and emergency response requirements.

[0034] Data fusion and preprocessing: A quantum entanglement detection mechanism is used to remove outliers in medical environment parameters (such as temperature and humidity data exceeding medical standards) and trigger real-time alarms. A chaotic fusion algorithm is used to fuse environmental data, energy consumption data, backup power status, and user feedback. The formula is as follows: Generate a fused quantum state containing spatiotemporal correlation features. .

[0035] Quantum Neural Network Model Construction and Training: Designing a multi-layer quantum neural network, with the input layer receiving data including medical environment data. Equipment energy consumption Backup power status Multidimensional quantum state data is used. The hidden layer extracts long-term temporal features (such as the impact of circadian rhythms on energy consumption) through quantum neurons and entangled channels, and introduces the ReLU activation function to enhance nonlinear expressive power. The output layer is divided into two branches: one predicts the total energy consumption in the future period. The probability of another predicted backup power supply starting up. The quantum measurement operation converts the data into classical values. During the training process, historical emergency scenario data (such as sudden changes in energy consumption during power grid failures) are added. The mean square error loss function is used to optimize the model's ability to identify abnormal operating conditions, ensuring that the prediction error is controllable under extreme conditions.

[0036] Real-time model adjustment and emergency response: Real-time acquisition of new data and calculation of quantum state similarity formula: When an abnormal similarity is detected that is lower than the similarity threshold, When this happens, a model update is triggered, along with a fast parameter update mechanism: quantum parallel computing is used to evaluate multiple parameter adjustment schemes in a short time, and real-time data features are preferentially fused through quantum state superposition operations, as shown in the formula. .

[0037] Quantum state prediction results based on quantum neural network output Calculate the probability distribution of different energy consumption values. The formula is: It identifies high-probability energy consumption intervals (such as peak energy consumption during peak surgical periods) and combines meteorological early warning data (such as high temperature warnings) with the status of backup power to generate a dynamic energy dispatching scheme: when the grid load is predicted to be tight, it selects the combination of equipment with the lowest energy consumption through probability distribution optimization, and at the same time starts the backup power preheating to ensure seamless emergency switching.

[0038] In summary, in the context of smart hospitals, this energy consumption prediction method addresses the high stability requirements of the medical environment. Through high-precision sensors and quantum state encoding technology, it captures key parameters such as temperature, humidity, and energy consumption of medical equipment in real time. It also integrates feedback from medical staff and the status of backup power supplies. The quantum neural network model, through entangled channels and quantum measurement operations, achieves accurate prediction of energy consumption in the medical area and assessment of emergency start-up probability. When an energy anomaly is detected, a real-time update mechanism based on quantum state superposition and a probability-driven emergency strategy can quickly adjust equipment operating priorities, ensuring the continuity of power supply to critical medical equipment. This solution deeply integrates the characteristics of quantum computing with medical emergency management, significantly enhancing the hospital's ability to respond to sudden energy crises while improving energy efficiency, thus providing a solid energy guarantee for medical safety.

[0039] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An energy consumption prediction method based on an energy management system, characterized in that, The specific steps of this prediction method are as follows: S100, multi-source data acquisition: Sensors are deployed in the interior areas of the building to collect comfort indicators and energy consumption data; cameras and microphones are used to collect user facial images and voice data, and affective computing technology is used to obtain user emotional feedback; at the same time, through mobile applications and indoor control panels, subjective evaluations of user comfort and adjustment behavior information are collected, and the collected data is quantum-encoded. S200, Data Fusion and Preprocessing: Integrate multi-source data encoded by quantum state, remove abnormal data with the help of quantum entanglement detection mechanism, normalize the remaining data, and map it to the unit quantum state space; S300, Constructing a correlation model: A model relating comfort and energy consumption is built using a quantum neural network. The number of nodes in the input layer, hidden layer, and output layer is determined. The hidden layer uses the ReLU activation function. Information is transmitted between layers through quantum entanglement channels. The preprocessed dataset is divided into training set, validation set, and test set to train the model. S400, Real-time Model Adjustment: Real-time collection of user emotional feedback and comfort adjustment behavior data and encoding them into quantum states, calculation of the similarity between the quantum states of the data and the original data, and when the similarity is lower than the threshold, obtaining the parameter adjustment scheme with the smallest prediction error, and updating the model parameters in real time according to the adjustment scheme. S500, Energy Consumption Prediction and Decision Making: Based on the adjusted model, the energy consumption prediction results of the quantum state are obtained, the probability distribution of different energy consumption values ​​is calculated, and the optimal energy supply plan and equipment operation strategy are selected according to the probability distribution, taking into account the energy supply situation and cost factors.

2. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In S100, the sensors used in the multi-source data acquisition are: a temperature sensor to collect real-time temperature data of different areas in the building, a humidity sensor to obtain indoor air humidity information, a light sensor to sense indoor light intensity, a smart meter to record the power consumption of different electrical devices in different time periods of the building, and a gas meter to monitor the gas usage in real time.

3. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In step S100, the steps of acquiring user facial images and voice data using a camera and microphone in multi-source data acquisition and obtaining user emotional feedback using emotion computing technology are as follows: acquiring user facial images through a camera, acquiring user voice signals in real time through a microphone array, performing preprocessing and standardization, extracting relevant features, and calculating facial expression scores, voice feature scores, and cross-modal interaction scores respectively; calculating the probability distribution of the user in various emotional states through an emotion state probability distribution algorithm; selecting the emotion category with the highest probability as the final recognition result; and outputting the corresponding emotion category.

4. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In step S200, during data fusion and preprocessing, the probability distribution of a user's various emotional states is calculated using an emotional state probability distribution algorithm. The formula is as follows: ,in It is the first The probability of a certain emotional state. It is the softmax function, which converts the score into a probability distribution. The calculation formula is: , The raw score entered. It is the total number of emotional states. Represents the natural constant. These are learnable fusion weight parameters, and , It is a facial expression feature score. It is a speech feature score. It is a score that captures the interaction between facial and voice features.

5. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In step S200, the data fusion and preprocessing process utilizes a chaotic fusion algorithm to integrate multi-source data. The formula is as follows: ,in This is the output of the chaos fusion algorithm. It is the total number of data sources participating in the integration. Indicates the first Data encoded by a quantum state These are the weighting coefficients. It is a hyperbolic tangent activation function used to perform nonlinear transformations on the fused data to enhance feature representation capabilities.

6. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In step S300, the quantum neural network model in the correlation model is constructed, and the input layer receives the fused quantum state data. The hidden layer uses quantum neurons, and the calculation formula for each quantum neuron is: ,in No. The output quantum state form of a quantum neuron To connect the first The input and the first The quantum weights of neurons in a hidden layer. Represents the quantum tensor product. This is a quantum rotation gate operation used to transform the input. Is it input to the number The number of data points per quantum neuron No. The quantum state data is input to the quantum neuron. Information is transmitted between hidden layers through quantum entanglement channels. The output layer uses quantum measurement operations to convert the quantum state output into a classical energy consumption prediction value. The formula is: , This is a predicted energy consumption value. The quantum state results of the output layer, quantum measurement operations, This means converting a quantum state into classical data.

7. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, S300 involves training the comfort-energy consumption correlation model within the correlation model: (1) Divide the preprocessed comprehensive dataset into training set, validation set and test set according to the proportion, divide the training set into batches and prepare the label data; (2) Randomly initialize the weight matrix and bias vector of the neural network, set the learning rate and momentum parameters, and initialize the quantum entanglement channel parameters at the same time; (3) The input layer receives preprocessed data, the hidden layer processes information through the ReLU activation function, information is transmitted between layers through the quantum entanglement channel, and the output layer calculates the predicted value using the linear activation function. (4) Calculate the mean square error between the predicted value and the true value as the loss function, and add a regularization term; (5) Update the neural network weights and quantum entanglement channel parameters by calculating the gradient of the loss function with respect to each parameter; (6) Evaluate the model performance on the validation set every 100 training iterations, record the loss and accuracy and plot the learning curve, adjust the strategy according to the validation results, and finally evaluate the model's generalization ability on the test set.

8. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In step S400, the real-time adaptive update of the parameters of the comfort and energy consumption correlation model during real-time model adjustment is as follows: when new user emotional feedback and comfort adjustment behavior data are obtained, they are encoded into quantum states. The formula for calculating the quantum state similarity between real-time data and original data is: ,in Quantum state similarity value measures the degree of similarity between real-time data and historical data. It is the quantum state of historical data. It is the quantum state of the newly acquired data. The similarity is the inner product of two quantum states, and the square of the modulus yields the similarity, which ranges from [0, 1]. When the similarity is below a similarity threshold... When the time is right, the model update is triggered, and the prediction error of each set of parameters is calculated to obtain the parameter scheme with the minimum error. The quantum weights of the model are then updated through a quantum state superposition update operation, as shown in the formula: ,in It is the updated quantum weight. It is an update coefficient, with a value range of [0, 1], which controls the fusion ratio of historical weights and new weights. It is the historical quantum weight before the update. These are new quantum weight candidate values ​​obtained through quantum parallel computing.

9. The energy consumption prediction method based on an energy management system according to claim 1, characterized in that, In the S500, energy consumption prediction and decision-making process calculates the probability distribution of different energy consumption values ​​using quantum probability, with the following formula: ,in Energy consumption value The probability of occurrence These are quantum states corresponding to different energy consumption values, generated by discretization within a preset energy consumption range. It is the energy consumption prediction result of the quantum state output by the model.

10. An energy consumption prediction system based on an energy management system, characterized in that, The system is applicable to the energy consumption prediction method based on an energy management system as described in any one of claims 1-9, and the system includes: a multi-source data acquisition module, a data fusion and preprocessing module, a quantum neural network modeling module, a real-time model adjustment module, and an energy consumption prediction and decision-making module; The multi-source data acquisition module: deploys temperature, humidity, and light sensors, as well as smart meters and gas meters to collect environmental and energy consumption data; captures user emotional feedback through facial expression recognition cameras and voice microphone arrays; and collects user comfort evaluations and equipment adjustment behaviors in conjunction with mobile applications and indoor control panels. All data is transmitted to the data fusion and preprocessing module after being quantum-encoded. The data fusion and preprocessing module integrates multi-source data encoded in quantum states, performs feature fusion using a chaotic fusion algorithm, identifies and removes outliers through a quantum entanglement detection mechanism, normalizes the remaining data, and maps it to a unit quantum state space. The quantum neural network modeling module constructs a quantum neural network containing an input layer, a hidden layer, and an output layer. The input layer receives preprocessed quantum state data, the hidden layer processes information through quantum neurons and quantum entanglement channels, and the output layer converts the quantum state into an energy consumption prediction value. The dataset is divided into a training set, a validation set, and a test set, and the model parameters are iteratively optimized using a stochastic gradient descent algorithm. The real-time model adjustment module: collects new user feedback and behavior data in real time and encodes them into quantum states, calculates the similarity between the quantum states of the data and historical data, triggers model updates when the similarity is lower than a threshold, evaluates multiple parameter adjustment schemes using quantum parallel computing, selects the optimal scheme and updates the model weights through quantum state superposition operations; The energy consumption prediction and decision-making module generates energy consumption prediction results for quantum states based on the adjusted model, calculates the probability distribution of different energy consumption values, and the energy management system combines real-time energy prices and renewable energy power generation factors to generate the optimal energy supply plan and equipment operation strategy.

Images