Indoor intelligent energy-saving method and system based on internet of things and AI

Abstract

The application provides an indoor intelligent energy-saving method based on the Internet of Things and AI, and belongs to the technical field of intelligent energy-saving control, and comprises the following steps: 1, collecting indoor environment data in layers through Internet of Things equipment, including basic environment data and dynamic behavior data; 2, generating an environment state matrix based on the basic environment data; 3, generating a behavior-energy consumption correlation graph based on the dynamic behavior data and the environment state matrix; 4, generating a dynamic energy-saving strategy based on the behavior-energy consumption correlation graph and preset energy-saving targets and comfort constraints; and 5, executing the dynamic energy-saving strategy through the Internet of Things equipment and monitoring the execution effect in real time, and optimizing the environment state matrix and the behavior-energy consumption correlation graph based on the execution effect. The application improves the intelligentization and self-automation level of the energy-saving effect.

Description

Technical Field

[0001] This invention relates to the field of intelligent energy-saving control technology, and in particular to indoor intelligent energy-saving methods and systems based on the Internet of Things and AI. Background Technology

[0002] With the increasing severity of global energy shortages and environmental pollution, smart buildings and energy-saving technologies have become important research areas in the field of energy management. Existing smart energy-saving systems mainly rely on simple timer switches and temperature control devices to regulate the indoor environment. Although they can save energy to some extent, they cannot dynamically adjust energy usage strategies in real time and accurately based on environmental conditions and user behavior.

[0003] Currently, some IoT-based smart home energy-saving systems utilize devices such as temperature and humidity sensors and light sensors to monitor the indoor environment and optimize device operation through preset algorithms. However, existing systems often overlook the impact of user behavior patterns on energy consumption, lack dynamic energy-saving strategies based on the fusion of environmental and behavioral data, and fail to improve energy-saving performance through continuous optimization mechanisms.

[0004] Therefore, the present invention provides an indoor smart energy-saving method and system based on the Internet of Things and AI. Summary of the Invention

[0005] This invention provides an indoor smart energy-saving method and system based on the Internet of Things (IoT) and AI, which combines IoT devices with AI technology to achieve intelligent energy-saving control of the indoor environment. First, environmental data is collected hierarchically via IoT to establish a comprehensive environmental state matrix and behavior-energy consumption correlation map. Based on this data, dynamic energy-saving strategies can be generated to effectively balance energy consumption and comfort. Through real-time monitoring and feedback adjustments, not only is energy saving improved, but environmental comfort is also ensured, avoiding the negative impact of excessive energy saving on the living experience. Furthermore, the combination of environmental state and behavioral data makes the energy-saving strategy more precise and personalized, enhancing the intelligence and autonomy of energy-saving performance.

[0006] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, comprising:

[0007] Step 1: Collect indoor environmental data in layers through IoT devices. The environmental data includes basic environmental data and dynamic behavior data.

[0008] Step 2: Generate an environmental state matrix based on basic environmental data;

[0009] Step 3: Generate a behavior-energy consumption correlation map based on dynamic behavior data and the environmental state matrix;

[0010] Step 4: Generate dynamic energy-saving strategies based on the behavior-energy consumption correlation graph and preset energy-saving targets and comfort constraints;

[0011] Step 5: Implement dynamic energy-saving strategies through IoT devices and monitor the implementation effect in real time. Optimize the environmental state matrix and behavior-energy consumption correlation graph based on the implementation effect.

[0012] Preferably, an environmental state matrix is ​​generated based on basic environmental data and an environmental state projection model, including:

[0013] The basic environmental data is cleaned and normalized.

[0014] A spatiotemporal convolutional neural network is used as the environmental state inference model, and the processed basic environmental data is used as the model input to output the environmental state matrix.

[0015] Preferably, the generation of behavior-energy consumption correlation maps based on dynamic behavior data and environmental state matrices includes:

[0016] Feature extraction is performed on dynamic behavioral data to obtain personnel behavior characteristics and equipment power consumption characteristics;

[0017] By analyzing all environmental parameters in the environmental state matrix, the state characteristics of several spatial regions are obtained;

[0018] A behavior-energy consumption correlation map is generated based on personnel behavior characteristics, equipment power consumption characteristics, and spatial area state characteristics.

[0019] Preferably, a behavior-energy consumption correlation map is generated based on personnel behavior characteristics, equipment power consumption characteristics, and spatial area state characteristics, including:

[0020] Quantify and classify personnel behavior characteristics, and use them as personnel behavior nodes in the behavior-energy consumption correlation graph;

[0021] Define equipment power consumption nodes based on equipment power consumption characteristics;

[0022] The state characteristics of a spatial region are used as the environment-related nodes in the behavior-energy consumption correlation graph.

[0023] Obtain the association strength coefficients between all nodes;

[0024] Generate a behavior-energy consumption correlation graph based on the correlation strength coefficient between all nodes.

[0025] Preferably, the association strength coefficients between all nodes are obtained, including:

[0026] Perform feature dimension analysis on all nodes to determine several feature dimensions for each node and the feature values ​​for each feature dimension;

[0027] The association strength coefficient between any two nodes is determined based on several feature dimensions corresponding to each node and the feature values ​​corresponding to each feature dimension:

[0028] ;

[0029] in, The correlation strength coefficient between node i and node j and Let represent the standardized values ​​of the feature values ​​of node i and node j in the k-th feature dimension, respectively. The preset weights for the k-th feature dimension are: , and These represent the rates of change of the feature values ​​of nodes i and j over time in the k-th feature dimension, respectively. The number of feature dimensions determined when determining the association strength coefficient between node i and node j The sign for the maximum value.

[0030] Preferably, a dynamic energy-saving strategy is generated based on the behavior-energy consumption correlation map and preset energy-saving targets and comfort constraints, including:

[0031] Based on preset energy-saving targets and comfort constraints, the operating parameters of each indoor device are determined, and the operating parameters of each indoor device are encoded to form a multi-dimensional vector;

[0032] The solution space is constructed based on all multidimensional vectors, where each multidimensional vector corresponds to a particle in the solution space;

[0033] Historical energy consumption data of indoor devices are obtained, and an energy consumption estimation model is constructed by combining the behavior-energy consumption correlation map to obtain the total energy consumption;

[0034] The degree of energy consumption deviation is determined based on the preset energy-saving target and total energy consumption;

[0035] Several comfort indicators were obtained based on the behavior-energy consumption correlation graph;

[0036] The degree of comfort deviation is obtained based on preset comfort constraints and comfort indicators;

[0037] The degree of deviation in energy consumption and the degree of deviation in comfort are combined according to preset weights to obtain the fitness value of each particle;

[0038] The fitness value of each particle is monitored in real time, and the corresponding termination condition is determined based on the fitness value of each particle.

[0039] Based on the rules of the MOPSO algorithm, each particle updates its velocity and position according to its own historical best position and the global best position in the particle swarm corresponding to each particle in the solution space until the termination condition is met, at which point the updating stops.

[0040] The position of each particle in the solution space that has stopped updating is obtained, and then the operating parameters of each device are determined to generate a dynamic energy-saving strategy.

[0041] Preferably, the termination condition is determined based on the fitness value of each particle, including:

[0042] The corresponding termination iteration coefficient is determined based on the fitness value of each particle:

[0043] ;

[0044] in, The termination iteration coefficient corresponding to the fitness value of each particle. Let g be the fitness value of each particle in the g-th iteration. Let g be the fitness value of each particle in the (g+1)th iteration. The weight coefficients for each particle in the g-th and g+1-th iterations. This is the second-order difference influence factor, with a value range of (0, 1). For the number of iterations, This is an adjustment factor, with a value range of (0, 0.5).

[0045] The termination condition is determined based on the termination iteration coefficient of each particle.

[0046] Preferably, dynamic energy-saving strategies are executed through IoT devices and the execution effect is monitored in real time. Based on the execution effect, the environmental state matrix and behavior-energy consumption correlation map are optimized, including:

[0047] Based on the control interface of IoT devices, dynamic energy-saving strategies are sent to the corresponding device execution terminals. At the same time, the operating status data of the devices and the real-time change data of the indoor environment are collected in real time through the sensors on the devices.

[0048] The system compares and analyzes the equipment's operating status data and real-time environmental change data with the expected effects of the dynamic energy-saving strategy. If the deviation exceeds the preset range, the system uses a gradient descent-based algorithm to adjust and optimize the parameters of the environmental state inference model and the behavior-energy consumption correlation map based on the actual execution results.

[0049] Indoor smart energy-saving systems based on IoT and AI include:

[0050] Data acquisition module: Collects indoor environmental data in layers through IoT devices, including basic environmental data and dynamic behavior data;

[0051] Matrix generation module: Generates an environmental state matrix based on basic environmental data;

[0052] Map generation module: Generates behavior-energy consumption correlation maps based on dynamic behavior data and environmental state matrix;

[0053] Strategy generation module: Generates dynamic energy-saving strategies based on behavior-energy consumption correlation graphs and preset energy-saving targets and comfort constraints;

[0054] Results optimization module: Implements dynamic energy-saving strategies through IoT devices and monitors the implementation effect in real time. Based on the implementation effect, optimizes the environmental state matrix and behavior-energy consumption correlation graph.

[0055] Compared with the prior art, the beneficial effects of this application are as follows:

[0056] By combining IoT devices with AI technology, intelligent energy-saving control of the indoor environment is achieved. First, environmental data is collected hierarchically via IoT to establish a comprehensive environmental status matrix and behavior-energy consumption correlation map. Based on this data, dynamic energy-saving strategies can be generated to effectively balance energy consumption and comfort. Through real-time monitoring and feedback adjustments, not only is energy saving improved, but environmental comfort is also ensured, avoiding the negative impact of excessive energy saving on the living experience. Furthermore, the combination of environmental status and behavioral data makes energy-saving strategies more precise and personalized, enhancing the intelligence and autonomy of energy-saving effects. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0058] Figure 1 This is a flowchart illustrating the indoor smart energy-saving method based on the Internet of Things and AI provided in an embodiment of the present invention.

[0059] Figure 2 This is a schematic diagram of the structure of an indoor smart energy-saving system based on the Internet of Things and AI provided in an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0061] Example 1:

[0062] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, such as... Figure 1 As shown, it includes:

[0063] Step 1: Collect indoor environmental data in layers through IoT devices. The environmental data includes basic environmental data and dynamic behavior data.

[0064] Step 2: Generate an environmental state matrix based on basic environmental data;

[0065] Step 3: Generate a behavior-energy consumption correlation map based on dynamic behavior data and the environmental state matrix;

[0066] Step 4: Generate dynamic energy-saving strategies based on the behavior-energy consumption correlation graph and preset energy-saving targets and comfort constraints;

[0067] Step 5: Implement dynamic energy-saving strategies through IoT devices and monitor the implementation effect in real time. Optimize the environmental state matrix and behavior-energy consumption correlation graph based on the implementation effect.

[0068] In this embodiment, basic environmental data includes temperature, humidity, and light intensity, while dynamic behavioral data includes personnel activity trajectories and equipment usage patterns.

[0069] In this embodiment, the environmental state matrix includes the spatiotemporal evolution characteristics of temperature distribution, humidity distribution, and light distribution. Temperature distribution: the temperature change trend of different areas indoors; humidity distribution: the humidity change trend of different areas indoors; light distribution: the light intensity change trend of different areas indoors. The matrix is ​​stored in time series form and can predict the environmental state in the future.

[0070] In this embodiment, the dynamic energy-saving strategy includes equipment control schemes, environmental regulation schemes, and behavior guidance schemes.

[0071] The beneficial effects of the above technical solution are as follows: By combining IoT devices with AI technology, intelligent energy-saving control of the indoor environment is achieved. Firstly, by collecting environmental data in layers through the IoT, a comprehensive environmental state matrix and behavior-energy consumption correlation map are established. Based on this data, dynamic energy-saving strategies can be generated to effectively regulate the balance between energy consumption and comfort. Through real-time monitoring and feedback adjustments, not only is energy saving improved, but environmental comfort is also ensured, avoiding the negative impact of excessive energy saving on the living experience.

[0072] Example 2:

[0073] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, which generates an environmental state matrix based on basic environmental data and an environmental state inference model, including:

[0074] The basic environmental data is cleaned and normalized.

[0075] A spatiotemporal convolutional neural network is used as the environmental state inference model, and the processed basic environmental data is used as the model input to output the environmental state matrix.

[0076] In this embodiment, a spatiotemporal convolutional neural network is used as the environmental state inference model, and basic environmental data is used as the model input to obtain the environmental state matrix. This includes converting the basic environmental data into a spatiotemporal data format, i.e., a three-dimensional tensor (time × space × feature), where: the time dimension represents the time series of the data; the spatial dimension represents the spatial distribution of sensors indoors; and the feature dimension represents temperature, humidity, and light intensity. The input data is a three-dimensional tensor (time × space × feature), representing the spatiotemporal distribution of the basic environmental data. The model structure includes: a spatiotemporal convolutional layer: using a three-dimensional convolutional kernel (time × space × feature) to capture the spatiotemporal features of the data; kernel size: for example, 3×3×3 (time × space × feature); convolutional layer output: a feature map after extracting spatiotemporal features. A pooling layer: using a three-dimensional max-pooling layer to reduce the dimension of the feature map; kernel size: for example, 2×2×2 (time × space × feature). A fully connected layer: flattening the feature map output by the pooling layer into a one-dimensional vector; mapping the feature vector to the dimensions of the environmental state matrix through the fully connected layer. Output layer:

[0077] The output environmental state matrix includes the spatiotemporal evolution characteristics of temperature, humidity, and illumination distributions. Model training: Historical baseline environmental data is used as the training set, and the environmental state matrix serves as the label. Loss function: Mean Squared Error (MSE), used to measure the difference between model predictions and actual values. Optimization algorithm: Adam optimizer, used to update model parameters. Training process: The loss function is minimized through backpropagation until the model converges. Model prediction: Preprocessed baseline environmental data is input into the trained ST-CNN model; the model output is the environmental state matrix, including the spatiotemporal evolution characteristics of temperature, humidity, and illumination distributions.

[0078] In this embodiment, the environmental state matrix structure is as follows: the environmental state matrix is ​​a three-dimensional tensor (time × space × feature), where: the time dimension represents the changes in environmental state over a future period; the spatial dimension represents the environmental state of different areas indoors; and the feature dimension represents temperature, humidity, and light intensity. Matrix storage and application: the environmental state matrix is ​​stored as a time series for subsequent steps (such as multi-source data fusion and dynamic weight allocation).

[0079] The beneficial effects of the above technical solution are as follows: By employing a spatiotemporal convolutional neural network as the environmental state inference model, the spatiotemporal correlation in the data can be fully explored, improving the prediction accuracy of the environmental state matrix. Compared with existing technologies, the environmental state inference process is optimized, the system's intelligence and adaptability are enhanced, energy-saving strategies become more precise and efficient, and the level of intelligence in indoor energy management is improved.

[0080] Example 3:

[0081] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, which generates a behavior-energy consumption correlation map based on dynamic behavior data and an environmental state matrix, including:

[0082] Feature extraction is performed on dynamic behavioral data to obtain personnel behavior characteristics and equipment power consumption characteristics;

[0083] By analyzing all environmental parameters in the environmental state matrix, the state characteristics of several spatial regions are obtained;

[0084] A behavior-energy consumption correlation map is generated based on personnel behavior characteristics, equipment power consumption characteristics, and spatial area state characteristics.

[0085] In this embodiment, AI technology is used to extract features from dynamic behavioral data to obtain personnel behavior characteristics and equipment power consumption characteristics. For example, computer vision and machine learning algorithms (such as convolutional neural networks and support vector machines) are used to analyze video data collected by smart cameras to identify personnel behavior patterns (such as walking, standing, sitting at work, and meeting discussions), movement trajectories, and activity areas. For equipment power consumption data, deep learning models (such as Long Short-Term Memory networks, LSTM) are used to analyze equipment power consumption patterns, power change trends, and fault diagnosis, among other equipment power consumption characteristics.

[0086] In this embodiment, all environmental parameters in the environmental state matrix are analyzed to obtain the state characteristics of several spatial regions. This includes using clustering algorithms, such as K-means clustering, to analyze the data in the environmental state matrix and divide the indoor space into different regions. For example, based on the similarity of temperature and humidity, the office is divided into window-side areas, central areas, and wall-side areas. Each region has relatively consistent environmental state characteristics. State index calculation: For each spatial region, a series of state indices are determined. For example, the average temperature, humidity standard deviation, and light intensity variation rate within the region are calculated. These indices reflect the environmental stability and trends of the region. Simultaneously, by combining personnel behavior data and equipment power consumption data, the correlation between personnel activity density, equipment utilization rate, and environmental state in different regions is analyzed.

[0087] The beneficial effects of the above technical solution are as follows: Through feature extraction and data analysis, it accurately captures the characteristics of personnel behavior, equipment power consumption, and spatial area status, effectively constructing a behavior-energy consumption correlation map. Compared with existing technologies, it can more accurately identify and correlate the relationship between behavior and energy consumption, providing more precise data support for the optimization of dynamic energy-saving strategies and improving the intelligence and refinement of indoor energy efficiency management.

[0088] Example 4:

[0089] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, which generates a behavior-energy consumption correlation map based on human behavior characteristics, equipment power consumption characteristics, and spatial area state characteristics, including:

[0090] Quantify and classify personnel behavior characteristics, and use them as personnel behavior nodes in the behavior-energy consumption correlation graph;

[0091] Define equipment power consumption nodes based on equipment power consumption characteristics;

[0092] The state characteristics of a spatial region are used as the environment-related nodes in the behavior-energy consumption correlation graph.

[0093] Obtain the association strength coefficients between all nodes;

[0094] Generate a behavior-energy consumption correlation graph based on the correlation strength coefficient between all nodes.

[0095] In this embodiment, personnel behavioral characteristics are quantified and classified as personnel behavior nodes in the behavior-energy consumption correlation graph. For example, characteristics such as personnel activity type (standing, walking, working, etc.), activity area, and activity time are defined as different nodes. For activity area, each area can be set as a node according to the indoor space division; for activity time, nodes are set according to different time periods (such as weekday mornings and afternoons, weekends, etc.). These nodes can comprehensively reflect the behavioral status of personnel in different aspects.

[0096] In this embodiment, defining device power consumption nodes based on device power consumption characteristics involves using device type (lighting fixtures, air conditioners, office equipment, etc.), power characteristics under different power consumption modes, and harmonic characteristics as nodes. For example, for lighting fixtures, power consumption at different brightness levels is set as nodes; for air conditioners, power consumption corresponding to different temperature settings in cooling and heating modes is set as nodes, thereby accurately describing the device's power consumption behavior.

[0097] In this embodiment, the spatial region state characteristics are used as environmental related nodes in the association map. For example, the temperature state (high temperature, medium temperature, low temperature) and humidity state (high humidity, medium humidity, low humidity) of different indoor areas are used as nodes, which are combined with personnel behavior nodes and equipment power consumption nodes to construct a node system that comprehensively reflects the relationship between indoor environment, personnel behavior and equipment power consumption.

[0098] In this embodiment, all nodes include: personnel behavior nodes, equipment power consumption nodes, and environment-related nodes.

[0099] In this embodiment, a behavior-energy consumption correlation graph is generated based on the correlation strength coefficient between all nodes. This includes: creating an empty graph framework containing a set of personnel behavior nodes P, a set of equipment power consumption nodes E, and a set of environment-related nodes A. Nodes in each set are initialized according to the previously quantified and classified definitions. For example, the personnel behavior node set includes nodes representing different activity types, activity areas, and activity times; the equipment power consumption node set includes nodes representing different equipment types, power characteristics, and harmonic characteristics; and the environment-related node set includes nodes representing state characteristics such as temperature, humidity, and light intensity in different spatial areas. Establishing node connections: Iterate through all node pairs. For each pair of nodes (i, j), where i belongs to node set X (X can be P, E, or A) and j belongs to node set Y (Y can be P, E, or A, and X and Y can be the same), determine whether to establish a connection edge based on the calculated correlation strength coefficient. If the coefficient is greater than a preset threshold (e.g., 0.3, which can be adjusted according to actual conditions), then a connection edge is established between nodes i and j, indicating a certain degree of correlation between these two nodes. Set the edge weights: For node pairs (i, j) with established edges, adjust the association strength coefficient. This serves as the weight of the edge. The weight reflects the strength of the association between two nodes; the larger the weight, the stronger the association between the nodes. For example, if the association strength coefficient between the node "person stays in a certain area for more than 30 minutes" and the node "lighting equipment in that area is turned on" is 0.8, then the weight of the edge connecting these two nodes in the graph would be set to 0.8. Visualizing the graph: The constructed behavior-energy consumption association graph is visualized using graph visualization tools (such as Graphviz, NetworkX, etc.). In the visualization interface, different types of nodes can be distinguished by different shapes and colors; for example, personnel behavior nodes are represented by circles, equipment power consumption nodes by squares, and environment-related nodes by triangles.

[0100] The beneficial effects of the above technical solution are as follows: By quantifying and classifying personnel behavioral characteristics, defining equipment power consumption nodes and spatial area status characteristics, a more comprehensive and detailed behavior-energy consumption correlation map is constructed. Compared with existing technologies, by accurately acquiring the correlation strength between nodes, the implementation effect of energy-saving strategies is optimized, and the intelligence, accuracy, and real-time performance of energy consumption management are improved.

[0101] Example 5:

[0102] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, which obtains the correlation strength coefficient between all nodes, including:

[0103] Perform feature dimension analysis on all nodes to determine several feature dimensions for each node and the feature values ​​for each feature dimension;

[0104] The association strength coefficient between any two nodes is determined based on several feature dimensions corresponding to each node and the feature values ​​corresponding to each feature dimension:

[0105] ;

[0106] in, The correlation strength coefficient between node i and node j and Let represent the standardized values ​​of the feature values ​​of node i and node j in the k-th feature dimension, respectively. The preset weights for the k-th feature dimension are: , and These represent the rates of change of the feature values ​​of nodes i and j over time in the k-th feature dimension, respectively. The number of feature dimensions determined when determining the association strength coefficient between node i and node j The sign for the maximum value.

[0107] In this embodiment, the preset weights are determined based on the degree of influence of each feature dimension on the node association strength in the actual scenario, and different weight values ​​are set accordingly. For example, in an indoor smart energy-saving scenario, if temperature has a significant impact on device energy consumption, then the weight of temperature-related feature dimensions can be set relatively high; while the weight of some feature dimensions that have a smaller impact on the overall association is reduced accordingly. By introducing weights, the importance of different feature dimensions in calculating association strength can be reflected more accurately.

[0108] In this embodiment, the rate of change over time is calculated based on the corresponding time series data for human behavior nodes, such as the rate of change of activity time, the rate of change of power consumption nodes of equipment, and the rate of change of temperature of environmental related nodes.

[0109] In this embodiment, the number of feature dimensions is used to address inconsistencies in the number of dimensions. For node types with fewer dimensions, virtual dimensions can be added under specific conditions. The data values ​​of these virtual dimensions can be set to constants (such as 0), or generated based on existing dimension data using a certain algorithm. In this way, when calculating the association strength coefficient, the number of dimensions for the two types of nodes can be unified, and k can take values ​​within the same range.

[0110] The beneficial effects of the above technical solution are as follows: by performing feature dimension analysis on nodes, the correlation strength coefficient between nodes is accurately calculated, which improves the accuracy and reliability of the behavior-energy consumption correlation map. By introducing weight factors through standardization and time change rate, a more refined correlation analysis between nodes is ensured, thereby optimizing energy-saving decisions and strategies and improving energy-saving efficiency and intelligence level.

[0111] Example 6:

[0112] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, which generates a dynamic energy-saving strategy based on a behavior-energy consumption correlation graph and preset energy-saving targets and comfort constraints, including:

[0113] Based on preset energy-saving targets and comfort constraints, the operating parameters of each indoor device are determined, and the operating parameters of each indoor device are encoded to form a multi-dimensional vector;

[0114] The solution space is constructed based on all multidimensional vectors, where each multidimensional vector corresponds to a particle in the solution space;

[0115] Historical energy consumption data of indoor devices are obtained, and an energy consumption estimation model is constructed by combining the behavior-energy consumption correlation map to obtain the total energy consumption;

[0116] The degree of energy consumption deviation is determined based on the preset energy-saving target and total energy consumption;

[0117] Several comfort indicators were obtained based on the behavior-energy consumption correlation graph;

[0118] The degree of comfort deviation is obtained based on preset comfort constraints and comfort indicators;

[0119] The degree of deviation in energy consumption and the degree of deviation in comfort are combined according to preset weights to obtain the fitness value of each particle;

[0120] The fitness value of each particle is monitored in real time, and the corresponding termination condition is determined based on the fitness value of each particle.

[0121] Based on the rules of the MOPSO algorithm, each particle updates its velocity and position according to its own historical best position and the global best position in the particle swarm corresponding to each particle in the solution space until the termination condition is met, at which point the updating stops.

[0122] The position of each particle in the solution space that has stopped updating is obtained, and then the operating parameters of each device are determined to generate a dynamic energy-saving strategy.

[0123] In this embodiment, the operating parameters of each indoor device are encoded. For example, for an indoor environment containing air conditioning, lighting, and ventilation equipment, the solution vector can be represented as [air conditioning set temperature, air conditioning fan speed, air conditioning cooling / heating mode, lighting brightness, lighting on / off mode, ventilation speed level, ventilation on / off time]. Through this encoding method, each point in the solution space can accurately correspond to a specific combination of device operating parameters.

[0124] In this embodiment, an energy consumption estimation model is constructed. For example, for air conditioning equipment, a mathematical model is established between energy consumption and these parameters by analyzing energy consumption data under different set temperatures, fan speeds, and cooling / heating modes. When calculating the energy consumption of the combination of operating parameters of the equipment represented by the particle, the parameters of each device are substituted into the corresponding energy consumption model, and the total energy consumption is accumulated. This total energy consumption is then compared with the preset target of reducing the total energy consumption by more than 30% to calculate the degree to which the energy consumption deviates from the target.

[0125] In this embodiment, several comfort indicators are obtained based on a behavior-energy consumption correlation graph. Although the graph itself does not directly present the comfort indicators, they can be derived through the relationships between nodes and edges in the graph. The correlation between personnel behavior nodes and environmental state nodes can provide clues. For example, when a person is active in a certain area, the corresponding environmental state indicators such as temperature and humidity in that area can serve as basic data for comfort assessment. From the perspective of equipment power consumption nodes, the operating status of air conditioners is closely related to temperature nodes, which can reflect whether the indoor temperature regulation is within the comfortable range. Based on these correlations, a comfort indicator calculation model can be constructed. For example, if the temperature comfort range is set to 24℃-26℃, the corresponding temperature node in the graph is found, and it is determined whether it falls within this range, thus deriving the temperature comfort indicator. Similarly, humidity is set to 40%-60% as the comfort range, and the humidity node data in the graph is used for evaluation.

[0126] In this embodiment, comfort indicators include comfort indicators such as temperature, humidity, and light intensity.

[0127] The beneficial effects of the above technical solution are as follows: By combining energy-saving goals with comfort constraints, and optimizing equipment operating parameters based on the MOPSO algorithm, a dynamic energy-saving strategy is generated. Compared with existing technologies, this method can accurately balance energy saving and comfort, and achieve efficient energy consumption management and intelligent adjustment through particle swarm optimization, thereby improving the accuracy and adaptability of energy-saving effects and user experience.

[0128] Example 7:

[0129] This invention provides an indoor smart energy-saving method based on the Internet of Things and AI, which determines the corresponding termination condition based on the fitness value of each particle, including:

[0130] The corresponding termination iteration coefficient is determined based on the fitness value of each particle:

[0131] ;

[0132] in, The termination iteration coefficient corresponding to the fitness value of each particle. Let g be the fitness value of each particle in the g-th iteration. Let g be the fitness value of each particle in the (g+1)th iteration. The weight coefficients for each particle in the g-th and g+1-th iterations. This is the second-order difference influence factor, with a value range of (0, 1). For the number of iterations, This is an adjustment factor, with a value range of (0, 0.5).

[0133] The termination condition is determined based on the termination iteration coefficient of each particle.

[0134] In this embodiment, The second-order difference represents the relative proportion of the first-order difference to the change in fitness value, reflecting the trend of fitness value change. For example, if the fitness value was originally showing a stable decreasing trend, the second-order difference is close to 0; if a sudden change occurs, the absolute value of the second-order difference will increase.

[0135] In this embodiment, This represents the cumulative sum of the relative proportions of the second-order difference and the first-order difference in all iterative steps. It is used to further adjust the size of the denominator so that the calculation results can more accurately reflect the convergence state of the algorithm.

[0136] In this embodiment, the weight coefficients for each particle in the g-th and g+1-th iterations can be set according to the actual situation. For example, as the number of iterations increases, later iteration steps may be more critical, and these coefficients can be gradually increased. This is particularly relevant in earlier iterations (i≤N / 2). The value range is 0.005-0.01; in later iterations (i>N / 2), the value range is 0.01-0.02.

[0137] In this embodiment, the adjustment factor is used to adjust the degree of influence of the second-order difference on the calculation results.

[0138] The beneficial effects of the above technical solution are as follows: By introducing a termination iteration coefficient, the stopping conditions of the particle swarm optimization algorithm can be precisely controlled. Compared with existing technologies, this method, based on particle fitness changes and second-order difference influence factors, achieves more flexible and refined termination condition settings, improves the convergence speed and optimization accuracy of the algorithm, and thus enhances the intelligence and stability of the energy-saving strategy.

[0139] Example 8:

[0140] This invention provides an indoor smart energy-saving method based on the Internet of Things (IoT) and AI. It executes dynamic energy-saving strategies through IoT devices and monitors the execution results in real time. Based on the execution results, it optimizes the environmental state matrix and behavior-energy consumption correlation graph, including:

[0141] Based on the control interface of IoT devices, dynamic energy-saving strategies are sent to the corresponding device execution terminals. At the same time, the operating status data of the devices and the real-time change data of the indoor environment are collected in real time through the sensors on the devices.

[0142] The system compares and analyzes the equipment's operating status data and real-time environmental change data with the expected effects of the dynamic energy-saving strategy. If the deviation exceeds the preset range, the system uses a gradient descent-based algorithm to adjust and optimize the parameters of the environmental state inference model and the behavior-energy consumption correlation map based on the actual execution results.

[0143] In this embodiment, the device execution end includes, for example, a smart lighting controller, a smart air conditioning gateway, and a ventilation equipment driver.

[0144] In this embodiment, the parameters of the environmental state inference model and the behavior-energy consumption correlation map are adjusted and optimized. For example, if it is found that the actual energy consumption reduction does not reach the expected target, the analysis is conducted to determine whether it is due to the environmental state model's inaccurate estimation of certain parameters or the fact that certain correlations in the behavior-energy consumption correlation map do not accurately reflect the actual situation. Then, the model and map are optimized in a targeted manner to improve the accuracy and effectiveness of the dynamic energy-saving strategy.

[0145] The beneficial effects of the above technical solution are as follows: By combining IoT devices with AI technology, intelligent energy-saving control of the indoor environment is achieved. Firstly, by collecting environmental data in layers through the IoT, a comprehensive environmental state matrix and behavior-energy consumption correlation map are established. Based on this data, dynamic energy-saving strategies can be generated to effectively regulate the balance between energy consumption and comfort. Through real-time monitoring and feedback adjustments, not only is energy saving improved, but environmental comfort is also ensured, avoiding the negative impact of excessive energy saving on the living experience.

[0146] Example 9:

[0147] This invention provides an indoor smart energy-saving system based on the Internet of Things and AI, such as... Figure 2 As shown, it includes:

[0148] Data acquisition module: Collects indoor environmental data in layers through IoT devices, including basic environmental data and dynamic behavior data;

[0149] Matrix generation module: Generates an environmental state matrix based on basic environmental data;

[0150] Map generation module: Generates behavior-energy consumption correlation maps based on dynamic behavior data and environmental state matrix;

[0151] Strategy generation module: Generates dynamic energy-saving strategies based on behavior-energy consumption correlation graphs and preset energy-saving targets and comfort constraints;

[0152] Results optimization module: Implements dynamic energy-saving strategies through IoT devices and monitors the implementation effect in real time. Based on the implementation effect, optimizes the environmental state matrix and behavior-energy consumption correlation graph.

[0153] The beneficial effects of the above technical solution are as follows: By collecting indoor environmental data and generating an environmental state matrix and a behavior-energy consumption correlation map, energy-saving strategies can be dynamically generated to optimize energy consumption. Through real-time monitoring and feedback on execution results, strategies can be continuously optimized, improving energy-saving effects while ensuring comfort, significantly improving energy efficiency and intelligence levels, and reducing energy waste.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An indoor smart energy-saving method based on the Internet of Things and AI, characterized in that, include: Step 1: Collect indoor environmental data in layers through IoT devices. The environmental data includes basic environmental data and dynamic behavior data. Step 2: Generate an environmental state matrix based on basic environmental data; Step 3: Generate a behavior-energy consumption correlation map based on dynamic behavior data and the environmental state matrix; Step 4: Generate dynamic energy-saving strategies based on the behavior-energy consumption correlation graph and preset energy-saving targets and comfort constraints; Step 5: Implement dynamic energy-saving strategies through IoT devices and monitor the implementation effect in real time. Optimize the environmental state matrix and behavior-energy consumption correlation graph based on the implementation effect. A behavior-energy consumption correlation map is generated based on dynamic behavior data and an environmental state matrix, including: Feature extraction is performed on dynamic behavioral data to obtain personnel behavior characteristics and equipment power consumption characteristics; By analyzing all environmental parameters in the environmental state matrix, the state characteristics of several spatial regions are obtained; Generate a behavior-energy consumption correlation map based on personnel behavior characteristics, equipment power consumption characteristics, and spatial area state characteristics; A behavior-energy consumption correlation map is generated based on personnel behavior characteristics, equipment power consumption characteristics, and spatial area state characteristics, including: Quantify and classify personnel behavior characteristics, and use them as personnel behavior nodes in the behavior-energy consumption correlation graph; Define equipment power consumption nodes based on equipment power consumption characteristics; The state characteristics of a spatial region are used as the environment-related nodes in the behavior-energy consumption correlation graph. Obtain the association strength coefficients between all nodes; Generate a behavior-energy consumption correlation graph based on the correlation strength coefficients between all nodes; Obtain the association strength coefficients between all nodes, including: Perform feature dimension analysis on all nodes to determine several feature dimensions for each node and the feature values ​​for each feature dimension; The association strength coefficient between any two nodes is determined based on several feature dimensions corresponding to each node and the feature values ​​corresponding to each feature dimension: ； in, The correlation strength coefficient between node i and node j and Let represent the standardized values ​​of the feature values ​​of node i and node j in the k-th feature dimension, respectively. The preset weights for the k-th feature dimension are: , and These represent the rates of change of the feature values ​​of nodes i and j over time in the k-th feature dimension, respectively. The number of feature dimensions determined when determining the association strength coefficient between node i and node j To determine the sign of the maximum value; The number of feature dimensions is used to address inconsistencies in the number of dimensions. For node types with fewer dimensions, virtual dimensions can be added under specific conditions. The data values ​​of these virtual dimensions can be set to constants or generated based on existing dimension data through a certain algorithm. In this way, when calculating the association strength coefficient, the number of dimensions of the two types of nodes can be unified, and k can take values ​​within the same range. Dynamic energy-saving strategies are generated based on behavior-energy consumption correlation maps and preset energy-saving targets and comfort constraints, including: Based on preset energy-saving targets and comfort constraints, the operating parameters of each indoor device are determined, and the operating parameters of each indoor device are encoded to form a multi-dimensional vector; The solution space is constructed based on all multidimensional vectors, where each multidimensional vector corresponds to a particle in the solution space; Historical energy consumption data of indoor devices are obtained, and an energy consumption estimation model is constructed by combining the behavior-energy consumption correlation map to obtain the total energy consumption; The degree of energy consumption deviation is determined based on the preset energy-saving target and total energy consumption; Several comfort indicators were obtained based on the behavior-energy consumption correlation graph; The degree of comfort deviation is obtained based on preset comfort constraints and comfort indicators; The degree of deviation in energy consumption and the degree of deviation in comfort are combined according to preset weights to obtain the fitness value of each particle; The fitness value of each particle is monitored in real time, and the corresponding termination condition is determined based on the fitness value of each particle. Based on the rules of the MOPSO algorithm, each particle updates its velocity and position according to its own historical best position and the global best position in the particle swarm corresponding to each particle in the solution space until the termination condition is met, at which point the updating stops. The position of each particle in the solution space that has stopped updating is obtained, and then the operating parameters of each device are determined to generate a dynamic energy-saving strategy. The termination condition is determined based on the fitness value of each particle, including: The corresponding termination iteration coefficient is determined based on the fitness value of each particle: ； in, The termination iteration coefficient corresponding to the fitness value of each particle. Let g be the fitness value of each particle in the g-th iteration. Let g be the fitness value of each particle in the (g+1)th iteration. The weight coefficients for each particle in the g-th and g+1-th iterations. This is the second-order difference influence factor, with a value range of (0, 1). For the number of iterations, This is an adjustment factor, with a value range of (0, 0.5). The termination condition is determined based on the termination iteration coefficient of each particle.

2. The indoor smart energy-saving method based on the Internet of Things and AI according to claim 1, characterized in that, An environmental state matrix is ​​generated based on basic environmental data and an environmental state projection model, including: The basic environmental data is cleaned and normalized. A spatiotemporal convolutional neural network is used as the environmental state inference model, and the processed basic environmental data is used as the model input to output the environmental state matrix.

3. The indoor smart energy-saving method based on the Internet of Things and AI according to claim 1, characterized in that, Dynamic energy-saving strategies are implemented through IoT devices, and the implementation effects are monitored in real time. Based on the implementation effects, the environmental state matrix and behavior-energy consumption correlation graph are optimized, including: Based on the control interface of IoT devices, dynamic energy-saving strategies are sent to the corresponding device execution terminals. At the same time, the operating status data of the devices and the real-time change data of the indoor environment are collected in real time through the sensors on the devices. The system compares and analyzes the equipment's operating status data and real-time environmental change data with the expected effects of the dynamic energy-saving strategy. If the deviation exceeds the preset range, the system uses a gradient descent-based algorithm to adjust and optimize the parameters of the environmental state inference model and the behavior-energy consumption correlation map based on the actual execution results.

4. An indoor smart energy-saving system based on the Internet of Things and AI, applied to the indoor smart energy-saving method based on the Internet of Things and AI as described in any one of claims 1-3, characterized in that, include: Data acquisition module: Collects indoor environmental data in layers through IoT devices, including basic environmental data and dynamic behavior data; Matrix generation module: Generates an environmental state matrix based on basic environmental data; Map generation module: Generates behavior-energy consumption correlation maps based on dynamic behavior data and environmental state matrix; Strategy generation module: Generates dynamic energy-saving strategies based on behavior-energy consumption correlation graphs and preset energy-saving targets and comfort constraints; Results optimization module: Implements dynamic energy-saving strategies through IoT devices and monitors the implementation effect in real time. Based on the implementation effect, optimizes the environmental state matrix and behavior-energy consumption correlation graph.

Images