Machine learning Internet of Things equipment energy-saving control method and system

Abstract

The invention discloses an Internet of Things equipment energy-saving control method and system based on machine learning, and relates to the technical field of Internet of Things equipment, and the technical scheme is characterized in that the method comprises the following steps: classifying all-dimensional energy-saving related data of Internet of Things target equipment to obtain energy consumption feature data and scene adaptation data; on the basis of the energy consumption characteristic data, identifying same-network heterogeneous equipment of the target equipment, and evaluating the energy-saving association degree between the same-network heterogeneous equipment and other Internet of Things equipment in a historical period to obtain heterogeneous association data; processing the pre-processed energy-saving data, the historical energy-saving regulation and control data and the standard energy-saving index to obtain the energy-saving control adaptation degree of the target equipment of the Internet of Things; and matching the target equipment according to the energy-saving control adaptation degree and executing the corresponding energy-saving control operation. The method has the effect of improving the energy utilization efficiency and the overall operation benefit of the Internet of Things equipment.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) device technology, and more specifically, to a machine learning-based energy-saving control method and system for IoT devices. Background Technology

[0002] With the popularization and large-scale application of IoT technology, a large number of IoT devices are widely deployed in various scenarios. Traditional energy-saving control methods for IoT devices focus on optimizing the energy consumption of individual devices, without fully considering the network interconnection characteristics between devices and the dynamic adaptation requirements of scenario services. Existing technologies generally ignore the energy consumption linkage between heterogeneous devices on the same network. Adjusting parameters only for a single device cannot perceive and utilize this heterogeneous relationship, making it difficult to achieve optimal energy-saving effects, and may even lead to an increase in overall energy consumption due to local adjustments. Existing methods generally adopt fixed energy-saving modes or general control strategies, which cannot be adjusted according to changes in scenario conditions, resulting in a difficulty in balancing energy-saving effects and service experience. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a machine learning-based energy-saving control method and system for Internet of Things (IoT) devices.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A machine learning-based energy-saving control method for IoT devices, comprising the following steps: The energy-saving data of IoT target devices are classified into energy consumption characteristic data and scenario adaptation data. Based on energy consumption characteristic data, identify heterogeneous devices on the same network for target devices, and evaluate the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods to obtain heterogeneous correlation data. Extract energy consumption monitoring data of heterogeneous devices on the same network to other IoT devices during historical periods, and obtain a reference energy-saving feature map based on the energy consumption monitoring data and heterogeneous correlation data; The degree of energy-saving correlation between the target device and heterogeneous devices on the same network is evaluated to obtain the heterogeneous correlation value, and the degree of energy-saving correlation between the target device and scene-related devices is evaluated to obtain the scene correlation value; Preprocessed energy-saving data is obtained based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps; The energy-saving control adaptability of IoT target devices is obtained by processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators. Match the target equipment with the energy-saving control suitability and execute the corresponding energy-saving control operation.

[0005] Preferably, heterogeneous correlation data is obtained by evaluating the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods, specifically including the following steps: Collect operational status data of heterogeneous devices on the same network and other IoT devices within historical time periods; Based on the operational status data, the energy consumption change trends of heterogeneous devices on the same network and other IoT devices under different operating conditions are extracted, and the synchronization characteristics are obtained based on the energy consumption change trends of heterogeneous devices on the same network and other IoT devices under different operating conditions. The correlation level is obtained by analyzing and processing the synchronization characteristics; Heterogeneous relational data are obtained by integrating synchronization characteristics, correlation patterns, and correlation levels.

[0006] Preferably, the correlation level is obtained by analyzing and processing the synchronization characteristics, specifically including the following steps: Based on synchronization characteristics, determine the degree of impact of the operating status of heterogeneous devices on the energy consumption of other IoT devices; Based on the degree of impact, clarify the correlation pattern of energy consumption interaction between heterogeneous devices on the same network and other IoT devices; Based on the correlation rules, the energy-saving correlation between heterogeneous devices on the same network and other IoT devices is classified and determined to obtain the correlation level.

[0007] Preferably, a reference energy-saving characteristic map is obtained based on energy consumption monitoring data and heterogeneous correlation data, specifically including the following steps: A knowledge graph representing the energy-saving relationships between devices is constructed based on heterogeneous correlation data and energy consumption monitoring data. Energy-saving trend characteristics are obtained by performing time-series statistics on the knowledge graph; A reference energy-saving feature map is obtained by screening and integrating energy-saving trend features based on preset energy-saving stability benchmarks and fluctuation thresholds.

[0008] Preferably, a reference energy-saving feature map is obtained by screening and integrating energy-saving trend features based on a preset energy-saving stability benchmark and fluctuation threshold, specifically including the following steps: Energy-saving trend features that meet the preset energy-saving stability benchmark and fluctuation threshold are marked as effective energy-saving trend features; The integrated effective energy-saving trend characteristics are obtained by integrating the effective energy-saving trend characteristics; A reference energy-saving feature map is constructed based on the integrated effective energy-saving trend characteristics.

[0009] Preferably, the heterogeneity correlation value is obtained by evaluating the degree of energy-saving correlation between the target device and heterogeneous devices on the same network, specifically including the following steps: Collect data on energy consumption fluctuations of heterogeneous devices on the same network over historical periods; Based on the heterogeneous correlation data, extract the energy consumption change characteristics of energy consumption fluctuations, and clarify the energy consumption regulation rules of heterogeneous equipment in the same network based on the energy consumption change characteristics; Monitor the energy consumption operation parameters of the target equipment during the corresponding historical period, and obtain the linkage relationship between the energy consumption change of the target equipment and the energy consumption adjustment of heterogeneous equipment in the same network based on the energy consumption operation parameters and energy consumption adjustment rules. Obtain the operation parameters of heterogeneous equipment in the same network based on the linkage relationship. Based on the operating parameters of heterogeneous devices on the same network and the energy-saving correlation between heterogeneous devices on the same network and other IoT devices, determine the magnitude of the impact of energy consumption adjustment of heterogeneous devices on the energy consumption change of the target device. Heterogeneous correlation values ​​are obtained based on the magnitude of the impact of changes in the energy consumption of the target equipment and the energy consumption characteristic data of the target equipment.

[0010] Preferably, the scene correlation value is obtained by evaluating the degree of energy-saving correlation between the target device and the scene-related devices, specifically including the following steps: Based on the real-time operating status and energy consumption fluctuation information of the devices associated with the scene, as well as the environmental and load change information of the scene where the target device is located, a scene-linked energy consumption change system is constructed. Based on the scene-linked energy consumption change system, the energy consumption linkage pattern between the target device and various scene-related devices is sorted out, and effective linkage characteristics are extracted based on the energy consumption linkage pattern. Based on the effective linkage characteristics, the degree of mutual influence of energy consumption between the target device and the scene-related devices in different scene states is obtained, and the degree of mutual influence of energy consumption is quantified to form the initial scene association value. The initial scene association values ​​are modified based on the stability of collaborative operation with the equipment to obtain the scene association values.

[0011] Preferably, preprocessed energy-saving data is obtained based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps, specifically including the following steps: The comprehensive correlation coefficient is obtained by weighting and fusing the heterogeneous correlation value and the scene correlation value; Effective energy consumption characteristic information is obtained by filtering energy consumption characteristic data based on comprehensive correlation coefficients; The effective energy consumption characteristic information and the scenario adaptation data are fused to form basic energy-saving characteristic information, and the basic energy-saving characteristic information is then regularized based on the reference energy-saving characteristic map. Preprocessed energy-saving data is obtained from the comprehensive correlation coefficient and the normalized basic energy-saving characteristic information.

[0012] Preferably, the energy-saving control adaptability of the target IoT device is obtained by processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators, specifically including the following steps: The pre-processed energy-saving data and the historical energy-saving control data of the target equipment are combined into pre-processed energy-saving characteristic data; The energy-saving control response capability is evaluated based on the historical energy-saving control data of the target equipment to obtain the control response value; Obtain the standard energy-saving indicators for the application scenario of the target device, and compare the standard energy-saving indicators with the actual energy-saving indicators to obtain the degree of energy-saving compliance; The energy-saving control adaptability of target IoT devices is assessed based on pre-processed energy-saving characteristic data, control response values, and energy-saving compliance levels.

[0013] A machine learning-based energy-saving control system for IoT devices includes: Classification module: Classifies the energy-saving related data of IoT target devices from all dimensions to obtain energy consumption characteristic data and scenario adaptation data; The first evaluation module identifies heterogeneous devices on the same network based on energy consumption characteristic data, and evaluates the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods to obtain heterogeneous correlation data. First processing module: Extracts energy consumption monitoring data of heterogeneous devices on the same network to other IoT devices during historical periods, and obtains a reference energy-saving feature map based on the energy consumption monitoring data and heterogeneous correlation data; The second evaluation module evaluates the degree of energy-saving correlation between the target device and heterogeneous devices on the same network to obtain the heterogeneous correlation value, and evaluates the degree of energy-saving correlation between the target device and scene-related devices to obtain the scene correlation value. The second processing module obtains preprocessed energy-saving data based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps. The third processing module processes the pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators to obtain the energy-saving control adaptability of the target IoT device. Execution module: Matches the target device with the energy-saving control adaptation degree and executes the corresponding energy-saving control operation.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention obtains energy consumption characteristic data and scenario adaptation data by classifying and processing energy-saving related data from all dimensions. Based on the energy consumption characteristic data, it identifies heterogeneous devices within the same network. By evaluating the energy-saving correlation between heterogeneous devices within the same network and other devices, it generates heterogeneous correlation data, which can depict the energy consumption linkage pattern between devices. Based on energy consumption monitoring data and heterogeneous correlation data, it constructs a reference energy-saving characteristic map, which can intuitively present the optimal energy-saving correlation pattern between devices under different operating conditions, greatly improving the stability and scalability of energy-saving strategies. It evaluates the energy-saving correlation between the target device and heterogeneous devices within the same network and scenario-related devices, thereby generating heterogeneous correlation values ​​and scenario correlation values. Through weighted fusion, it obtains a comprehensive correlation coefficient. By jointly processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators, it obtains the energy-saving control adaptability, realizing a quantitative evaluation of the matching of the current energy-saving strategy. It can intuitively reflect the balance level between the strategy and energy-saving effect, operational stability, and business adaptability. Based on the energy-saving control adaptability, it matches and executes differentiated energy-saving control operations, significantly improving the energy utilization efficiency and overall operational benefits of IoT devices. Attached Figure Description

[0015] Fig. 1 This invention provides a schematic diagram illustrating the steps of a machine learning-based energy-saving control method for IoT devices. Fig. 2 This is a schematic diagram of a machine learning-based energy-saving control system for IoT devices, provided as an embodiment of the present invention. Detailed Implementation

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0017] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0018] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0019] Reference Figs. 1-2 As shown.

[0020] The embodiments further illustrate the energy-saving control method and system for IoT devices based on machine learning proposed in this invention.

[0021] A machine learning-based energy-saving control method for IoT devices, comprising the following steps: The energy-saving data of IoT target devices are classified into energy consumption characteristic data and scenario adaptation data. Identify heterogeneous devices on the same network based on energy consumption characteristic data.

[0022] Comprehensive energy-saving data includes direct and indirect information about the energy-saving effects of target equipment during operation. This includes the equipment's own operating parameters, such as instantaneous power, cumulative energy consumption, runtime, start-stop frequency, and power ranges for different operating modes, directly reflecting the patterns of energy consumption generation and change. It also includes environmental and business-related parameters, such as temperature, humidity, light intensity, network signal strength, data transmission volume, user usage time distribution, and business priority requirements in the equipment's environment. These parameters affect the effectiveness of energy-saving strategies. In the classification process, data directly characterizing the equipment's energy consumption generation mechanism, fluctuation patterns, and consumption levels are categorized as energy consumption characteristic data. Examples include the compressor start-stop power of a smart refrigerator, the average daily energy consumption at different cooling levels, and the correlation between door opening frequency and energy consumption increment. Furthermore, data related to the equipment's environment and business scenario is categorized as scenario adaptation data, such as the ambient temperature of the kitchen where the smart refrigerator is located, the daily door opening time distribution for users, and the bandwidth utilization rate of the IoT network.

[0023] Identifying heterogeneous devices within the same network based on energy consumption characteristic data. Heterogeneous devices within the same network refer to IoT devices that are in the same IoT network environment as the target device, but differ in device type, hardware architecture, communication protocol, or business functions, yet share network resources and some business scenarios. Examples include smart lighting devices and smart robotic vacuum cleaners in the same home IoT network, or temperature sensors and motor drive devices in the same industrial IoT network. The energy consumption characteristics of the target device are constructed, consisting of core dimensions of energy consumption characteristic data, including the average energy consumption per unit time, the variance of energy consumption fluctuations, the time period of peak energy consumption, and the slope of energy consumption changes caused by start-stop operations. The corresponding energy consumption characteristic data are extracted from all other IoT devices within the same network. The similarity between the energy consumption characteristics of the target device and the device to be identified is calculated to determine whether they are heterogeneous devices. Energy consumption characteristic similarity = 1 - (Euclidean distance between the energy consumption characteristic vectors of the target device and the device to be identified / (magnitude of the energy consumption characteristic vector of the target device + magnitude of the energy consumption characteristic vector of the device to be identified)). Euclidean distance is used to measure the degree of difference between two feature vectors in a multidimensional space, and the magnitude is used for normalization to eliminate the influence of differences in the magnitude of different feature dimensions. A reasonable similarity threshold is set, such as 0.7. When the energy consumption feature similarity between the device to be identified and the target device is lower than this threshold, the target device is determined to be a heterogeneous device on the same network; if the similarity is higher than or equal to the threshold, it is determined to be a device of the same type or homogeneous device. For example, if the target device is a smart lighting device, its energy consumption feature vector includes an average energy consumption of 8 watts per unit time, peak energy consumption periods from 7 PM to 10 PM, and a start-stop energy consumption change slope of 2 watts per second. The energy consumption feature vector of a smart air conditioner within the same network includes an average energy consumption of 800 watts per unit time, peak energy consumption periods from 12 PM to 2 PM and 8 PM to 10 PM, and a start-stop energy consumption change slope of 300 watts per second. The energy consumption feature similarity between the two is 0.2, which is below the threshold, therefore the smart air conditioner is identified as a heterogeneous device within the same network as the target device. However, the energy consumption feature vector of another smart lighting device within the same network is highly similar to that of the target device, with a similarity of 0.9, which is above the threshold, therefore it is not considered a heterogeneous device. This identification method is used to locate heterogeneous devices within the network containing the target device.

[0024] Heterogeneous correlation data is obtained by evaluating the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods. Extract historical data on energy consumption of other IoT devices from heterogeneous devices on the same network.

[0025] Extract energy consumption monitoring data of heterogeneous devices on the same network to other IoT devices within a historical period. First, clarify the selection range of the historical period. This range is the complete operating cycle of the target device and heterogeneous devices on the same network, such as continuous operating data of the past 7 days or 30 days.

[0026] The system identifies the relationships between heterogeneous devices within the same network and other IoT devices. It iterates through each historical time point of the heterogeneous device within the same network and collects energy consumption monitoring data generated by the device for all other IoT devices in the network. Specifically, this includes the instantaneous power, cumulative energy consumption, and operating state switching time of the heterogeneous device itself, as well as the instantaneous power change, cumulative energy consumption increment, and operating state response delay of other IoT devices at the corresponding time.

[0027] A reference energy-saving characteristic map is obtained based on energy consumption monitoring data and heterogeneous correlation data; The degree of energy-saving correlation between the target device and heterogeneous devices on the same network is evaluated to obtain the heterogeneous correlation value, and the degree of energy-saving correlation between the target device and scene-related devices is evaluated to obtain the scene correlation value; Preprocessed energy-saving data is obtained based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps; The energy-saving control adaptability of IoT target devices is obtained by processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators. Match the target equipment with the energy-saving control suitability and execute the corresponding energy-saving control operation.

[0028] Based on the energy-saving control adaptability, match the corresponding energy-saving control operations to the target device. A mapping relationship between adaptability ranges and energy-saving control operations needs to be established in advance, dividing the adaptability into multiple levels, each corresponding to a set of control strategies. When the energy-saving control adaptability is in the high adaptability range of 0.8 to 1, it indicates that the current control strategy has reached its optimal state, requiring no major adjustments, only fine-tuning operations, such as slightly optimizing device operating parameters, maintaining the current energy-saving mode, and continuously monitoring energy consumption and operating status to ensure stable operation. When the adaptability is in the medium adaptability range of 0.5 to 0.8, it indicates that the current strategy has room for optimization, thus requiring moderate adjustments, such as adjusting device runtime, optimizing the collaborative operating logic with heterogeneous devices on the same network or related devices in the same scenario, thereby improving energy-saving effects. When the adaptability is in the low adaptability range of 0 to 0.5, it indicates that the current strategy is severely mismatched with the scenario and device status, thus requiring deep reconstruction operations, such as switching the device's core energy-saving mode, replanning the collaborative energy-saving scheme between devices, and even temporarily adjusting the business load distribution, prioritizing energy-saving compliance and operational stability.

[0029] The heterogeneous correlation data is obtained by evaluating the energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods. This includes the following steps: Collect operational status data of heterogeneous devices on the same network and other IoT devices within historical time periods; Based on the operational status data, the energy consumption change trends of heterogeneous devices on the same network and other IoT devices under different operating conditions are extracted, and the synchronization characteristics are obtained based on the energy consumption change trends of heterogeneous devices on the same network and other IoT devices under different operating conditions. The correlation level is obtained by analyzing and processing the synchronicity characteristics, specifically including the following steps: Based on synchronization characteristics, determine the degree of impact of the operating status of heterogeneous devices on the energy consumption of other IoT devices; Based on the degree of impact, clarify the correlation pattern of energy consumption interaction between heterogeneous devices on the same network and other IoT devices; Based on the correlation rules, the energy-saving correlation between heterogeneous devices on the same network and other IoT devices is classified and determined to obtain the correlation level; Heterogeneous relational data are obtained by integrating synchronization characteristics, correlation patterns, and correlation levels.

[0030] The system collects operational status data from heterogeneous devices on the same network and other IoT devices over historical periods. For example, it selects continuous operational data from the past 30 days. This operational status data includes the device's own operating parameters and the environmental and business parameters in which the device operates. The device's own operating parameters include instantaneous power, cumulative energy consumption, start-stop time, working mode switching records, load rate, and runtime of the heterogeneous device on the same network and other IoT devices. The environmental and business parameters in which the device operates include ambient temperature, humidity, network bandwidth utilization, business task priority, and user interaction frequency. This type of data indirectly affects the device's energy consumption changes and correlations. If the heterogeneous device on the same network is a motor drive device in an industrial scenario, and the other IoT devices are temperature sensors and data acquisition terminals, the operational status data includes the motor's start-stop time, power curves under different loads, transmission rate, and energy consumption, thus forming a multi-device operational status sequence with time as the axis.

[0031] Based on operational status data, the energy consumption trends of heterogeneous devices within the same network and other IoT devices under different operating conditions are extracted to obtain synchronization characteristics. Operating conditions are categorized according to the device's business scenario and operating mode. For example, the operating conditions of motor-driven devices are divided into no-load, half-load, and full-load conditions, and smart home appliances are divided into standby, running, and enhanced running conditions. For each operating condition, the average energy consumption per unit time, energy consumption fluctuation variance, and energy consumption change slope of heterogeneous devices within the same network and other IoT devices under that condition are calculated. The energy consumption change slope is calculated as follows: (cumulative energy consumption at the end of the condition - cumulative energy consumption at the beginning of the condition) / (end time of the condition - start time of the condition), which characterizes the rate of energy consumption change of the device under that condition. Compare the energy consumption trends of heterogeneous devices on the same network with other IoT devices under the same operating conditions, and then calculate their synchronicity characteristic. The synchronicity characteristic value is calculated as follows: 1 - (Euclidean distance between the energy consumption change sequence of the heterogeneous device on the same network and the energy consumption change sequence of other IoT devices / (magnitude of the energy consumption change sequence of the heterogeneous device on the same network + magnitude of the energy consumption change sequence of other IoT devices)). The closer the synchronicity characteristic value is to 1, the more synchronized the energy consumption change trends of the two devices are, and the higher the correlation. The closer it is to 0, the greater the trend difference is, and the lower the correlation is. For example, if the energy consumption change slope of a motor drive device under full load is 50 watts / hour, and the energy consumption change slope of a temperature sensor in the same period is 0.1 watts / hour, the synchronicity characteristic value of the two devices is 0.85.

[0032] Based on synchronicity characteristics, the impact of the operating status of heterogeneous devices within the same network on the energy consumption of other IoT devices is determined. Synchronicity characteristic values ​​are divided into multiple intervals: 0 to 0.3 indicates a weak impact, 0.3 to 0.7 indicates a moderate impact, and 0.7 to 1 indicates a strong impact. The degree of impact is directly determined by the interval assignment of the synchronicity characteristic value. Based on the degree of impact, the correlation pattern of energy consumption interaction between heterogeneous devices within the same network and other IoT devices is clarified. If the impact is strong, it is determined whether changes in the operating status of heterogeneous devices within the same network cause changes in the energy consumption of other IoT devices in the same or opposite direction. For example, when the load of a motor-driven device increases, the transmission bandwidth demand of the data acquisition terminal increases, leading to a synchronous increase in its energy consumption, forming a unidirectional correlation pattern. If the impact is weak, it indicates that there is no obvious direct correlation between the energy consumption changes of the two devices, and the correlation pattern shows independent fluctuations. Based on the correlation patterns, the energy-saving correlation between heterogeneous devices on the same network and other IoT devices is classified into four levels: strong correlation, medium correlation, weak correlation, and no correlation. Strong correlation is defined as a synchronization characteristic value of 0.7 to 1, with a clear unidirectional or reverse correlation pattern; medium correlation is defined as a synchronization characteristic value of 0.3 to 0.7, with some correlation patterns; weak correlation is defined as a synchronization characteristic value of 0 to 0.3, with no obvious correlation pattern; and no correlation is defined as a synchronization characteristic value close to 0, with no correlation pattern.

[0033] Heterogeneous correlation data is obtained by integrating synchronization characteristics, correlation patterns, and correlation levels. Each pair of heterogeneous devices within the same network corresponds to heterogeneous correlation data with other IoT devices, including the synchronization characteristic value of the device pair, the correlation pattern of energy consumption interaction, the correlation level, and the corresponding operating condition information. For example, the heterogeneous correlation data between a motor drive device and a data acquisition terminal has a synchronization characteristic value of 0.85, and the correlation pattern is that the energy consumption of the data acquisition terminal increases synchronously when the motor load increases. Therefore, the correlation level is Level 1, applicable to full-load and half-load operating conditions. Heterogeneous correlation data can comprehensively quantify the degree of energy-saving correlation between heterogeneous devices within the same network and other IoT devices.

[0034] A reference energy-saving characteristic map is obtained based on energy consumption monitoring data and heterogeneous correlation data, specifically including the following steps: A knowledge graph representing the energy-saving relationships between devices is constructed based on heterogeneous correlation data and energy consumption monitoring data. Energy-saving trend characteristics are obtained by performing time-series statistics on the knowledge graph; Based on a preset energy-saving stability benchmark and fluctuation threshold, the energy-saving trend characteristics are screened and integrated to obtain a reference energy-saving feature map, which specifically includes the following steps: Energy-saving trend features that meet the preset energy-saving stability benchmark and fluctuation threshold are marked as effective energy-saving trend features; The integrated effective energy-saving trend characteristics are obtained by integrating the effective energy-saving trend characteristics; A reference energy-saving feature map is constructed based on the integrated effective energy-saving trend characteristics.

[0035] A knowledge graph representing energy-saving relationships between devices is constructed based on heterogeneous correlation data and energy consumption monitoring data. The knowledge graph uses heterogeneous devices within the same network and other IoT devices as entity nodes, with the energy-saving relationships between them as edges. The synchronization characteristics, correlation patterns, and correlation levels of the heterogeneous correlation data, as well as the energy consumption changes and trends of the energy consumption monitoring data, are used as attributes for nodes and edges. Each heterogeneous device within the same network is mapped to an independent node along with other IoT devices. Node attributes include device type, rated power, and historical average energy consumption. Connection edges are established based on device nodes with energy-saving relationships. Edge attributes include synchronization characteristics, correlation patterns, correlation levels, and the average and slope of energy consumption changes under the corresponding operating conditions. For example, if the heterogeneous device on the same network is a smart air conditioner, and other IoT devices are smart lighting and smart refrigerators, the knowledge graph will form two edges connecting the smart air conditioner to the smart lighting and smart refrigerator respectively. The edge attribute label indicates that the synchronization feature value of the smart air conditioner and the smart lighting is 0.82, the association rule is unidirectional association, and the association level is level one. At the same time, it is also noted that for every 100 watts increase in the power of the smart air conditioner, the energy consumption of the smart lighting will increase by an average of 0.3 watts.

[0036] Energy-saving trend features are obtained by performing time-series statistics on the knowledge graph. The time-series statistics are performed in preset time windows, such as 1 hour, traversing all time windows within historical periods to statistically analyze the energy consumption changes of each device node, the changes in the synchronization characteristic values ​​of each associated edge, and the distribution of association levels within each time window. Specifically, energy-saving trend features include single-device energy consumption trends, inter-device association trends, and scenario-level energy-saving trends. Single-device energy consumption trends are the average energy consumption per unit time and the variance of energy consumption fluctuation for each device within a time window; inter-device association trends are the average and magnitude of the synchronization characteristic values ​​of associated edges within a time window; scenario-level energy-saving trends are the proportion of edges with different association levels within a time window and the synergy of energy consumption changes for strongly associated devices. The average synchronization characteristic value of associated edges within a given time window = the sum of the synchronization characteristic values ​​of all associated edges within that window / the total number of associated edges within that window; energy consumption change synergy = the number of strongly associated device pairs with consistent energy consumption change directions / the total number of strongly associated device pairs. For example, the average synchronicity characteristic value of smart air conditioners and smart lighting is 0.85 between 7 pm and 8 pm, and the energy consumption change synergy is 0.9.

[0037] Based on preset energy-saving stability benchmarks and fluctuation thresholds, energy-saving trend features are screened and integrated to obtain a reference energy-saving feature map. Energy-saving trend features that meet the preset energy-saving stability benchmarks and fluctuation thresholds are marked as valid energy-saving trend features. The preset energy-saving stability benchmarks are defined as the average value of the synchronicity feature of related edges being no less than 0.7 and the energy consumption change synergy being no less than 0.8; the fluctuation thresholds are defined as the variance of energy consumption fluctuation of a single device not exceeding 10% of its historical average energy consumption and the change amplitude of the synchronicity feature value not exceeding 0.1. Energy-saving trend features are iterated through all time windows. If a feature simultaneously meets the stability benchmark and the fluctuation threshold, it is marked as a valid energy-saving trend feature; otherwise, it is discarded. For example, the average value of the synchronicity feature of smart air conditioners and smart refrigerators in a certain time window is 0.68. Since it is lower than the stability benchmark, this feature is discarded; features that meet all conditions between 7 PM and 8 PM are marked as valid.

[0038] The effective energy-saving trend characteristics are integrated to obtain the integrated effective energy-saving trend characteristics. These characteristics are then grouped according to operating conditions and scenario types. The mean of the integrated synchronicity characteristic value is calculated as the sum of the mean synchronicity characteristic values ​​of all effective characteristics within the group divided by the number of effective characteristics within the group. The synergy of energy consumption changes after integration is calculated as the sum of the synergy of energy consumption changes of all effective characteristics within the group divided by the number of effective characteristics within the group. This eliminates the influence of random fluctuations within a single time window, thus yielding energy-saving trends at the operating condition and scenario levels.

[0039] A reference energy-saving feature map is constructed based on the integrated effective energy-saving trend characteristics. Presented in a visual or structured format, the reference energy-saving feature map uses the integrated effective energy-saving trend characteristics of different operating conditions and scenarios as its core content. It includes the energy-saving correlation strength, energy consumption change patterns, and stable operating range of each device, providing a reference for the energy-saving control of target devices. For example, the reference energy-saving feature map marks the optimal energy-saving correlation range for smart air conditioners and smart lighting in a home living room scenario under cooling conditions. This is achieved when the synchronicity feature value is maintained between 0.8 and 0.9, and the energy consumption change synergy is not lower than 0.85, thus achieving the best synergistic energy-saving effect between the two and providing a reference for the energy-saving control strategy of target devices.

[0040] The heterogeneity correlation value is obtained by assessing the energy-saving correlation between the target device and heterogeneous devices on the same grid, specifically including the following steps: Collect data on energy consumption fluctuations of heterogeneous devices on the same network over historical periods; Based on the heterogeneous correlation data, extract the energy consumption change characteristics of energy consumption fluctuations, and clarify the energy consumption regulation rules of heterogeneous equipment in the same network based on the energy consumption change characteristics; Monitor the energy consumption operation parameters of the target equipment during the corresponding historical period, and obtain the linkage relationship between the energy consumption change of the target equipment and the energy consumption adjustment of heterogeneous equipment in the same network based on the energy consumption operation parameters and energy consumption adjustment rules. Obtain the operation parameters of heterogeneous equipment in the same network based on the linkage relationship. Based on the operating parameters of heterogeneous devices on the same network and the energy-saving correlation between heterogeneous devices on the same network and other IoT devices, determine the magnitude of the impact of energy consumption adjustment of heterogeneous devices on the energy consumption change of the target device. Heterogeneous correlation values ​​are obtained based on the magnitude of the impact of changes in the energy consumption of the target equipment and the energy consumption characteristic data of the target equipment.

[0041] The system collects energy consumption fluctuation data for heterogeneous devices on the same network over historical periods. This data includes the instantaneous power, cumulative energy consumption, energy consumption change, peak energy consumption fluctuation, and fluctuation period of each device at each time point. It also records information on device operating mode switching, such as the time of switching from standby mode to operating mode and the corresponding energy consumption change. If the heterogeneous device is a smart air conditioner, the energy consumption fluctuation data includes changes in cooling power, peak energy consumption at different ambient temperatures, the number of daily start-ups and shutdowns, and the energy consumption increment caused by each start-up and shutdown, thus forming a complete energy consumption fluctuation sequence along a time axis.

[0042] By extracting energy consumption fluctuation characteristics from heterogeneous correlation data, the energy consumption regulation patterns of heterogeneous devices within the same network can be clarified. The heterogeneous correlation data stores the synchronization characteristics, correlation patterns, and correlation levels of heterogeneous devices within the same network with other IoT devices. Combined with energy consumption fluctuation data, fluctuation amplitude characteristics, fluctuation period characteristics, and regulation response characteristics are extracted. The fluctuation amplitude characteristic is the difference between the maximum and minimum energy consumption changes of heterogeneous devices within the same network per unit time; fluctuation amplitude = maximum energy consumption value per unit time - minimum energy consumption value per unit time. The fluctuation period characteristic is the length of time it takes for a heterogeneous device within the same network to complete one complete rise and fall in energy consumption. The regulation response characteristic is the time required for a heterogeneous device within the same network to reach a stable energy consumption state after receiving a command indicating a change in operating conditions. Based on these characteristics, the energy consumption regulation patterns of heterogeneous devices within the same network can be clarified. For example, a smart air conditioner's cooling power increases by 50 watts for every 1 degree Celsius increase in ambient temperature, with a fluctuation period of approximately 2 hours and a regulation response time of approximately 10 minutes.

[0043] Based on energy consumption operating parameters and energy consumption adjustment patterns, the linkage between the energy consumption changes of the target device and the energy consumption adjustment of heterogeneous devices on the same network is obtained. The operating parameters of the heterogeneous devices on the same network are then derived from this linkage. The energy consumption operating parameters of the target device include its instantaneous power, cumulative energy consumption, load rate, and operating mode during the corresponding historical period. The energy consumption adjustment patterns of the heterogeneous devices on the same network are time-series aligned with the energy consumption operating parameters of the target device to determine the direction and magnitude of the energy consumption change of the target device after each energy consumption adjustment of the heterogeneous devices, thus obtaining the linkage relationship between the two. For example, if the power of a smart air conditioner increases by 50 watts, the power of the target device's smart lighting increases by 0.3 watts, thus forming a linear linkage relationship between the energy consumption adjustment of the heterogeneous devices on the same network and the energy consumption change of the target device. Based on this linear linkage relationship, the operating parameters of the heterogeneous devices on the same network in different linkage scenarios are obtained.

[0044] Based on the operating parameters of heterogeneous devices within the same network and the energy-saving correlation between these devices and other IoT devices, the impact of energy consumption adjustments by heterogeneous devices on the energy consumption changes of the target device is determined. The energy-saving correlation between heterogeneous devices and other IoT devices within the same network is derived from heterogeneous correlation data, including synchronicity characteristic values ​​and correlation levels, used to measure the overall correlation strength between the heterogeneous devices within the same network. Impact magnitude = (Target device energy consumption change / Heterogeneous device energy consumption adjustment) × Synchronicity characteristic value between the heterogeneous device and the target device. For example, if the energy consumption adjustment of the heterogeneous device within the same network is 100 watts, the energy consumption change of the target device is 0.6 watts, and the synchronicity characteristic value is 0.85, then the impact magnitude = (0.6 / 100) × 0.85 = 0.0051, meaning that for every 1 watt of energy consumption adjustment by the heterogeneous device within the same network, there is a 0.0051 watt energy consumption impact on the target device.

[0045] The heterogeneous correlation value is obtained based on the impact magnitude of the target device's energy consumption change and the target device's energy consumption characteristic data. The target device's energy consumption characteristic data includes its historical average energy consumption, rated power, and energy consumption fluctuation variance. The heterogeneous correlation value = impact magnitude × (target device rated power / target device historical average energy consumption). The final heterogeneous correlation value is obtained by combining the impact magnitude with the target device's own energy consumption characteristics. The higher the value, the stronger the energy-saving correlation between the heterogeneous devices on the same network and the target device. For example, the rated power of the target device's smart lighting is 10 watts, and the historical average energy consumption is 8 watt-hours, thus the heterogeneous correlation value is 0.0051 × (10 / 8) = 0.006375.

[0046] The energy-saving correlation between the target device and related devices in the scenario is evaluated to obtain the scenario correlation value, which includes the following steps: Based on the real-time operating status and energy consumption fluctuation information of the devices associated with the scene, as well as the environmental and load change information of the scene where the target device is located, a scene-linked energy consumption change system is constructed. Based on the scene-linked energy consumption change system, the energy consumption linkage pattern between the target device and various scene-related devices is sorted out, and effective linkage characteristics are extracted based on the energy consumption linkage pattern. Based on the effective linkage characteristics, the degree of mutual influence of energy consumption between the target device and the scene-related devices in different scene states is obtained, and the degree of mutual influence of energy consumption is quantified to form the initial scene association value. The initial scene association values ​​are modified based on the stability of collaborative operation with the equipment to obtain the scene association values.

[0047] Based on the real-time operating status and energy consumption fluctuation information of scene-related devices, as well as the environmental and load change information of the target device, a scene-linked energy consumption change system is constructed. Scene-related devices are IoT devices that are in the same application scenario as the target device and jointly participate in the scenario's business operations. For example, in a home living room scenario, smart air conditioners, smart curtains, and smart speakers operate together with the target device's smart lighting. Real-time operating status includes the start / stop status, working mode, and load rate of scene-related devices; energy consumption fluctuation information includes the instantaneous power, cumulative energy consumption, and energy consumption change per unit time of scene-related devices; environmental change information includes temperature, humidity, light intensity, and noise level; load change information includes the number of users, business task volume, and data transmission volume. A multi-dimensional linkage relationship is established between scene-related devices, target devices, environmental factors, and load factors along the scene's timeline, thus forming a scene-linked energy consumption change system. For example, the linkage system for a home living room scenario records changes in the cooling mode of the smart air conditioner, rises and falls in ambient temperature, increases and decreases in the number of users, and energy consumption changes of the smart curtains.

[0048] Based on the scene-linked energy consumption change system, the energy consumption linkage patterns between target devices and various scene-related devices are analyzed, and effective linkage characteristics are extracted from these patterns. When analyzing energy consumption linkage patterns, scene operation data is traversed in units of time windows to determine the direction and magnitude of energy consumption changes in the target device after changes in the state or energy consumption of each scene-related device, thus clarifying the linkage logic between them. For example, when smart curtains close, the light intensity in the living room decreases, forming a linkage pattern mediated by light change; when a smart speaker plays high-quality music, its own power consumption increases, occupying network bandwidth and causing an increase in the communication power consumption of smart lighting, forming a linkage pattern mediated by network resources. Based on these patterns, effective linkage characteristics are extracted, specifically including linkage response strength, linkage response latency, and linkage duration. Linkage response strength is the amount of energy consumption change in the target device caused by a unit energy consumption change in scene-related devices; linkage response strength = target device energy consumption change / scene-related device energy consumption change. Linkage response latency is the time required for the target device's energy consumption to change accordingly after a change in the state of a scene-related device; linkage duration is the length of time the target device's energy consumption change is maintained.

[0049] Based on the effective linkage characteristics, the degree of energy consumption interaction between the target device and scene-related devices in different scene states is obtained. The initial scene association value is formed by quantifying the degree of energy consumption interaction. For example, the family living room scene is divided into different states: high light and low load during the day, and low light and high load at night. The weighted sum of all effective linkage characteristics in each scene state is used as the quantified result of the degree of energy consumption interaction: Degree of energy consumption interaction = (Linkage response intensity × 0.6 + (1 / Linkage response delay) × 0.2 + (Linkage duration / Time window length) × 0.2). After normalizing this value, the initial scene association value is obtained, ranging from 0 to 1. The higher the value, the stronger the influence of scene-related devices on the energy consumption of the target device in that scene state. For example, in a low-light, high-load scenario at night, the linkage response strength of smart curtains and smart lighting is 0.5, the response delay is 5 seconds, the duration is 30 minutes, and the time window length is 60 minutes. Substituting these values ​​into the calculation, the degree of mutual influence of energy consumption is (0.5×0.6+(1 / 5)×0.2+(30 / 60)×0.2)=0.44. After normalization, the initial scene association value is 0.44.

[0050] The initial scene correlation value is corrected based on the stability of device collaborative operation to obtain the scene correlation value. Device collaborative operation stability = 1 - (sum of energy consumption fluctuation variance of all devices in the scene / sum of rated power of all devices in the scene) - (sum of operating state switching frequency of all devices in the scene / total time length in the scene). The higher the stability value, the more reliable the collaborative operation between devices. The final scene correlation value is obtained by multiplying the initial scene correlation value by the device collaborative operation stability value: Scene correlation value = initial scene correlation value × device collaborative operation stability. If the device collaborative operation stability of a low-light, high-load scene at night is 0.9, then the scene correlation value = 0.44 × 0.9 = 0.396, used to comprehensively evaluate the energy-saving correlation degree of the target device in the scene.

[0051] Preprocessed energy-saving data is obtained based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps. The specific steps include: The comprehensive correlation coefficient is obtained by weighting and fusing the heterogeneous correlation value and the scene correlation value; Effective energy consumption characteristic information is obtained by filtering energy consumption characteristic data based on comprehensive correlation coefficients; The effective energy consumption characteristic information and the scenario adaptation data are fused to form basic energy-saving characteristic information, and the basic energy-saving characteristic information is then regularized based on the reference energy-saving characteristic map. Preprocessed energy-saving data is obtained from the comprehensive correlation coefficient and the normalized basic energy-saving characteristic information.

[0052] The comprehensive correlation coefficient is obtained by weighting and fusing the heterogeneous correlation value and the scene correlation value. The heterogeneous correlation value represents the degree of energy-saving correlation between heterogeneous devices on the same network and the target device, while the scene correlation value represents the degree of energy-saving correlation between scene-related devices and the target device, reflecting the energy-saving influencing factors of the target device in the IoT environment. The comprehensive correlation coefficient = heterogeneous correlation value × heterogeneous weight + scene correlation value × scene weight, where the sum of the heterogeneous weight and the scene weight is 1. For example, if the heterogeneous weight is set to 0.4 and the scene weight is 0.6, and the heterogeneous correlation value is 0.006375 and the scene correlation value is 0.396, then the comprehensive correlation coefficient = 0.006375 × 0.4 + 0.396 × 0.6 = 0.24015. This quantifies the energy-saving correlation strength of the target device in the overall environment.

[0053] Effective energy consumption feature information is obtained by filtering energy consumption feature data based on the comprehensive correlation coefficient. The energy consumption feature data includes the instantaneous power, cumulative energy consumption, energy consumption fluctuation variance, and energy consumption change slope of the target device. Core effective features are obtained through filtering using the comprehensive correlation coefficient. If the comprehensive correlation coefficient is higher than a certain threshold, it indicates that the target device is strongly affected by external factors, and energy consumption features sensitive to external linkages should be retained. If the comprehensive correlation coefficient is lower than the threshold, it indicates that the target device is less affected by external factors, and features reflecting its own energy consumption patterns should be retained. The retained features are then normalized, mapping features of different magnitudes to the 0-1 range. The normalized feature value = (original feature value - minimum feature value) / (maximum feature value - minimum feature value). For example, if the comprehensive correlation coefficient is 0.24015, which is higher than the threshold, the energy consumption change slope and fluctuation variance are filtered out and then normalized to obtain the effective energy consumption feature information.

[0054] Effective energy consumption characteristics are fused with scenario-adaptive data to form basic energy-saving characteristics. These basic characteristics are then normalized using a reference energy-saving feature map as a benchmark. Scenario-adaptive data includes environmental parameters, load parameters, and service priorities for the target device's scenario. Normalized effective energy consumption characteristics are concatenated with the normalized characteristics of the scenario-adaptive data to form a high-dimensional basic energy-saving feature vector. The reference energy-saving feature map stores the optimal energy-saving feature distribution for different scenarios and operating conditions. The basic energy-saving feature vector is aligned with the corresponding scenario features in the reference energy-saving feature map. If a dimension of the basic energy-saving feature vector exceeds the stable range of the reference energy-saving feature map, it is corrected according to the boundary value of the reference range to ensure the basic energy-saving characteristics conform to the optimal energy-saving rules of the scenario. For example, if the stable range of the energy consumption change slope for a family living room scenario in the reference energy-saving feature map is 0.1 to 0.5, and the energy consumption change slope in the basic energy-saving features is 0.6, then the energy consumption change slope is corrected to 0.5 to ensure the feature information is consistent with the scenario's energy-saving rules.

[0055] Preprocessed energy-saving data is obtained by integrating the comprehensive correlation coefficient and the normalized basic energy-saving feature information. The comprehensive correlation coefficient is used as a weighting factor to weight the normalized basic energy-saving feature information. The preprocessed energy-saving data feature value = normalized basic energy-saving feature value × comprehensive correlation coefficient, ensuring that the importance of the energy-saving features matches the correlation strength with the target equipment. The comprehensive correlation coefficient is added as an independent dimension to the preprocessed energy-saving data, thus forming a complete dataset containing both correlation strength and normalized energy-saving features. This complete dataset retains both the energy consumption patterns and scenario adaptation information of the target equipment itself, and incorporates the correlation effects between devices.

[0056] The energy-saving control adaptability of IoT target devices is obtained by processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators. This process includes the following steps: The pre-processed energy-saving data and the historical energy-saving control data of the target equipment are combined into pre-processed energy-saving characteristic data; The energy-saving control response capability is evaluated based on the historical energy-saving control data of the target equipment to obtain the control response value; Obtain the standard energy-saving indicators for the application scenario of the target device, and compare the standard energy-saving indicators with the actual energy-saving indicators to obtain the degree of energy-saving compliance; The energy-saving control adaptability of target IoT devices is assessed based on pre-processed energy-saving characteristic data, control response values, and energy-saving compliance levels.

[0057] Preprocessed energy-saving data and historical energy-saving control data of the target equipment are combined to form preprocessed energy-saving feature data. Preprocessed energy-saving data includes comprehensive correlation coefficients and normalized basic energy-saving feature information, reflecting the current energy consumption patterns, scenario adaptability, and inter-equipment correlations of the target equipment. Historical energy-saving control data includes past energy-saving control commands, equipment operating parameters after control, and energy consumption change results, recording the target equipment's response performance under different control strategies. Features are stitched together using time windows to form preprocessed energy-saving feature data that includes current state characteristics and historical control experience.

[0058] The energy-saving control response capability is evaluated based on the historical energy-saving control data of the target equipment to obtain the control response value. The energy-saving control response capability reflects the efficiency and stability of the target equipment in executing energy-saving control commands, and is quantified through response delay, control accuracy, and fluctuation recovery capability in historical control data. Response delay refers to the time from issuing the control command to the corresponding change in equipment energy consumption; the shorter the delay, the stronger the response capability. Control accuracy refers to the degree of matching between the actual energy consumption change and the target control change; the higher the matching degree, the higher the accuracy. Fluctuation recovery capability refers to the speed at which the equipment's energy consumption returns to a stable state after control; the faster the recovery, the better the stability. The regulation response value = (1 - response delay / maximum allowable delay) × 0.3 + (actual energy consumption change / target energy consumption change) × 0.4 + (1 - fluctuation recovery time / maximum allowable recovery time) × 0.3. For example, the response delay of a target device is 5 seconds, the maximum allowable delay is 20 seconds, the actual energy consumption change is 8 watts, the target change is 10 watts, the fluctuation recovery time is 10 seconds, and the maximum allowable recovery time is 30 seconds. Substituting these values ​​into the calculation, the regulation response value is calculated as (1 - 5 / 20) × 0.3 + (8 / 10) × 0.4 + (1 - 10 / 30) × 0.3 = 0.745.

[0059] Obtain the standard energy-saving indicators for the application scenario of the target device, and compare the standard energy-saving indicators with the actual energy-saving indicators to determine the degree of energy-saving compliance. Standard energy-saving indicators are formulated by scenario business specifications or industry standards. For example, the standard energy-saving indicator for smart lighting in a home living room scenario is an average daily energy consumption of no more than 50 watt-hours, while the standard energy-saving indicator for sensors in an industrial scenario is an energy consumption fluctuation variance of no more than 5% of the rated power. The actual energy-saving indicator is calculated using the current operating data of the target device, reflecting the actual energy-saving level of the device. The single-item compliance rate = 1 - |Actual energy-saving indicator value - Standard energy-saving indicator value| / Standard energy-saving indicator value. The overall energy-saving compliance degree is obtained by averaging all single-item compliance rates, with a value ranging from 0 to 1. The closer the value is to 1, the better the energy-saving compliance.

[0060] The energy-saving control adaptability of IoT target devices is evaluated based on pre-processed energy-saving feature data, control response values, and energy-saving compliance levels. Energy-saving control adaptability characterizes the degree of matching between the current control strategy and the target device and environment; a higher adaptability indicates that the current strategy achieves optimal energy-saving performance while ensuring stable device operation. Correlation patterns are extracted from the pre-processed energy-saving feature data, and then weighted and fused based on the control response value and energy-saving compliance level. The energy-saving control adaptability is calculated as: Energy-saving control adaptability = Feature matching degree × 0.5 + Control response value × 0.2 + Energy-saving compliance level × 0.3. The feature matching degree reflects the degree of matching between the pre-processed energy-saving features and the historical best control features, with a value range of 0 to 1. For example, if the feature matching degree is 0.85, the control response value is 0.745, and the energy-saving compliance level is 0.9, the energy-saving control adaptability is calculated as: Energy-saving control adaptability = 0.85 × 0.5 + 0.745 × 0.2 + 0.9 × 0.3 = 0.844.

[0061] A machine learning-based energy-saving control system for IoT devices includes: Classification module: Classifies the energy-saving related data of IoT target devices from all dimensions to obtain energy consumption characteristic data and scenario adaptation data; The first evaluation module identifies heterogeneous devices on the same network based on energy consumption characteristic data, and evaluates the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods to obtain heterogeneous correlation data. First processing module: Extracts energy consumption monitoring data of heterogeneous devices on the same network to other IoT devices during historical periods, and obtains a reference energy-saving feature map based on the energy consumption monitoring data and heterogeneous correlation data; The second evaluation module evaluates the degree of energy-saving correlation between the target device and heterogeneous devices on the same network to obtain the heterogeneous correlation value, and evaluates the degree of energy-saving correlation between the target device and scene-related devices to obtain the scene correlation value. The second processing module obtains preprocessed energy-saving data based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps. The third processing module processes the pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators to obtain the energy-saving control adaptability of the target IoT device. Execution module: Matches the target device with the energy-saving control adaptation degree and executes the corresponding energy-saving control operation.

[0062] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0063] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0064] 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. A machine learning-based energy-saving control method for IoT devices, characterized in that, The method includes the following steps: The energy-saving data of IoT target devices are classified into energy consumption characteristic data and scenario adaptation data. Based on energy consumption characteristic data, identify heterogeneous devices on the same network for target devices, and evaluate the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods to obtain heterogeneous correlation data. Extract energy consumption monitoring data of heterogeneous devices on the same network to other IoT devices during historical periods, and obtain a reference energy-saving feature map based on the energy consumption monitoring data and heterogeneous correlation data; The degree of energy-saving correlation between the target device and heterogeneous devices on the same network is evaluated to obtain the heterogeneous correlation value, and the degree of energy-saving correlation between the target device and scene-related devices is evaluated to obtain the scene correlation value; Preprocessed energy-saving data is obtained based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps; The energy-saving control adaptability of IoT target devices is obtained by processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators. Match the target equipment with the energy-saving control suitability and execute the corresponding energy-saving control operation.

2. The machine learning-based energy-saving control method for IoT devices according to claim 1, characterized in that, The heterogeneous correlation data is obtained by evaluating the energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods. This includes the following steps: Collect operational status data of heterogeneous devices on the same network and other IoT devices within historical time periods; Based on the operational status data, the energy consumption change trends of heterogeneous devices on the same network and other IoT devices under different operating conditions are extracted, and the synchronization characteristics are obtained based on the energy consumption change trends of heterogeneous devices on the same network and other IoT devices under different operating conditions. The correlation level is obtained by analyzing and processing the synchronization characteristics; Heterogeneous relational data are obtained by integrating synchronization characteristics, correlation patterns, and correlation levels.

3. The machine learning-based energy-saving control method for IoT devices according to claim 1, characterized in that, The correlation level is obtained by analyzing and processing the synchronicity characteristics, specifically including the following steps: Based on synchronization characteristics, determine the degree of impact of the operating status of heterogeneous devices on the energy consumption of other IoT devices; Based on the degree of impact, clarify the correlation pattern of energy consumption interaction between heterogeneous devices on the same network and other IoT devices; Based on the correlation rules, the energy-saving correlation between heterogeneous devices on the same network and other IoT devices is classified and determined to obtain the correlation level.

4. The machine learning-based energy-saving control method for IoT devices according to claim 3, characterized in that, A reference energy-saving characteristic map is obtained based on energy consumption monitoring data and heterogeneous correlation data, specifically including the following steps: A knowledge graph representing the energy-saving relationships between devices is constructed based on heterogeneous correlation data and energy consumption monitoring data. Energy-saving trend characteristics are obtained by performing time-series statistics on the knowledge graph; A reference energy-saving feature map is obtained by screening and integrating energy-saving trend features based on preset energy-saving stability benchmarks and fluctuation thresholds.

5. The machine learning-based energy-saving control method for IoT devices according to claim 4, characterized in that, Based on a preset energy-saving stability benchmark and fluctuation threshold, the energy-saving trend characteristics are screened and integrated to obtain a reference energy-saving feature map, which specifically includes the following steps: Energy-saving trend features that meet the preset energy-saving stability benchmark and fluctuation threshold are marked as effective energy-saving trend features; The integrated effective energy-saving trend characteristics are obtained by integrating the effective energy-saving trend characteristics; A reference energy-saving feature map is constructed based on the integrated effective energy-saving trend characteristics.

6. The machine learning-based energy-saving control method for IoT devices according to claim 1, characterized in that, The heterogeneity correlation value is obtained by assessing the energy-saving correlation between the target device and heterogeneous devices on the same grid, specifically including the following steps: Collect data on energy consumption fluctuations of heterogeneous devices on the same network over historical periods; Based on the heterogeneous correlation data, extract the energy consumption change characteristics of energy consumption fluctuations, and clarify the energy consumption regulation rules of heterogeneous equipment in the same network based on the energy consumption change characteristics; Monitor the energy consumption operation parameters of the target equipment during the corresponding historical period, and obtain the linkage relationship between the energy consumption change of the target equipment and the energy consumption adjustment of heterogeneous equipment in the same network based on the energy consumption operation parameters and energy consumption adjustment rules. Obtain the operation parameters of heterogeneous equipment in the same network based on the linkage relationship. Based on the operating parameters of heterogeneous devices on the same network and the energy-saving correlation between heterogeneous devices on the same network and other IoT devices, determine the magnitude of the impact of energy consumption adjustment of heterogeneous devices on the energy consumption change of the target device. Heterogeneous correlation values ​​are obtained based on the magnitude of the impact of changes in the energy consumption of the target equipment and the energy consumption characteristic data of the target equipment.

7. The machine learning-based energy-saving control method for IoT devices according to claim 6, characterized in that, The energy-saving correlation between the target device and related devices in the scenario is evaluated to obtain the scenario correlation value, which includes the following steps: Based on the real-time operating status and energy consumption fluctuation information of the devices associated with the scene, as well as the environmental and load change information of the scene where the target device is located, a scene-linked energy consumption change system is constructed. Based on the scene-linked energy consumption change system, the energy consumption linkage pattern between the target device and various scene-related devices is sorted out, and effective linkage characteristics are extracted based on the energy consumption linkage pattern. Based on the effective linkage characteristics, the degree of mutual influence of energy consumption between the target device and the scene-related devices in different scene states is obtained, and the degree of mutual influence of energy consumption is quantified to form the initial scene association value. The initial scene association values ​​are modified based on the stability of collaborative operation with the equipment to obtain the scene association values.

8. The machine learning-based energy-saving control method for IoT devices according to claim 1, characterized in that, Preprocessed energy-saving data is obtained based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps. The specific steps include: The comprehensive correlation coefficient is obtained by weighting and fusing the heterogeneous correlation value and the scene correlation value; Effective energy consumption characteristic information is obtained by filtering energy consumption characteristic data based on comprehensive correlation coefficients; The effective energy consumption characteristic information and the scenario adaptation data are fused to form basic energy-saving characteristic information, and the basic energy-saving characteristic information is then regularized based on the reference energy-saving characteristic map. Preprocessed energy-saving data is obtained from the comprehensive correlation coefficient and the normalized basic energy-saving characteristic information.

9. The machine learning-based energy-saving control method for IoT devices according to claim 1, characterized in that, The energy-saving control adaptability of IoT target devices is obtained by processing pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators. This process includes the following steps: The pre-processed energy-saving data and the historical energy-saving control data of the target equipment are combined into pre-processed energy-saving characteristic data; The energy-saving control response capability is evaluated based on the historical energy-saving control data of the target equipment to obtain the control response value; Obtain the standard energy-saving indicators for the application scenario of the target device, and compare the standard energy-saving indicators with the actual energy-saving indicators to obtain the degree of energy-saving compliance; The energy-saving control adaptability of target IoT devices is assessed based on pre-processed energy-saving characteristic data, control response values, and energy-saving compliance levels.

10. A machine learning-based energy-saving control system for IoT devices, applied to the machine learning-based energy-saving control method for IoT devices as described in any one of claims 1 to 9, characterized in that, include: Classification module: Classifies the energy-saving related data of IoT target devices from all dimensions to obtain energy consumption characteristic data and scenario adaptation data; The first evaluation module identifies heterogeneous devices on the same network based on energy consumption characteristic data, and evaluates the degree of energy-saving correlation between heterogeneous devices on the same network and other IoT devices during historical periods to obtain heterogeneous correlation data. First processing module: Extracts energy consumption monitoring data of heterogeneous devices on the same network to other IoT devices during historical periods, and obtains a reference energy-saving feature map based on the energy consumption monitoring data and heterogeneous correlation data; The second evaluation module evaluates the degree of energy-saving correlation between the target device and heterogeneous devices on the same network to obtain the heterogeneous correlation value, and evaluates the degree of energy-saving correlation between the target device and scene-related devices to obtain the scene correlation value. The second processing module obtains preprocessed energy-saving data based on heterogeneous correlation values, scene correlation values, energy consumption characteristic data, scene adaptation data, and reference energy-saving feature maps. The third processing module processes the pre-processed energy-saving data, historical energy-saving control data, and standard energy-saving indicators to obtain the energy-saving control adaptability of the target IoT device. Execution module: Matches the target device with the energy-saving control adaptation degree and executes the corresponding energy-saving control operation.

Images