Intelligent energy optimization management method and system based on Internet of Things

Abstract

The invention discloses an intelligent energy optimization management method and system based on the Internet of Things, and the method comprises the steps: synchronously collecting multi-dimensional physical quantities and environment parameters through perception units disposed in equipment and an environment, and forming multi-source heterogeneous data; carrying out normalization processing on the data, constructing an equipment operation feature vector, and realizing fine division of operation states in combination with a preset state classification system; load demand prediction is carried out based on the state identification result and the environmental parameters, and load interval estimation is generated; and calculating an optimal energy distribution scheme in combination with system constraints by taking the minimum energy cost and the minimum equipment loss as targets, and continuously optimizing a scheduling strategy through closed-loop monitoring and a feedback mechanism. According to the method, full-link automatic management from data acquisition, state identification and load prediction to optimization decision is realized. Through constructing the equipment operation feature vector and combining with the state classification system, the accuracy of equipment operation state identification is significantly improved, a reliable basis is provided for load prediction, and the prediction deviation is effectively reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of energy optimization, and particularly relates to an intelligent energy optimization management method and system based on the Internet of Things. BACKGROUND

[0002] In the current industrial and building energy management system, a control mode based on threshold or empirical rules is generally adopted, equipment operation parameters are collected by independent sensors, and then start-stop or adjustment instructions are issued after simple analysis by a central controller. Such a system generally only monitors basic electrical parameters such as voltage and current, lacks coordinated perception of multi-dimensional physical quantities such as temperature, pressure, flow, and environmental temperature and humidity, and data processing mostly stays at the local alarm or post-event statistics level, failing to achieve fine identification of equipment operation state and dynamic prediction of future load. Due to the lack of deep extraction and state classification of actual operation characteristics of the equipment, the system is difficult to accurately determine that the equipment is in an efficient, inefficient or idle operation state, thereby causing the energy distribution strategy to lag, and often resulting in overrunning of the equipment or energy waste.

[0003] The main problem of the prior art is that the accurate identification of the equipment operation state cannot be achieved based on multi-source heterogeneous data, resulting in large load prediction deviation, lack of dynamic optimization basis for energy distribution, slow response of overall management strategy, and limited energy efficiency improvement. SUMMARY

[0004] The purpose of the present application is to provide an intelligent energy optimization management method and system based on the Internet of Things, which realizes full-link automated management from data collection, state identification, load prediction to optimization decision-making, to solve the problems raised in the background technology.

[0005] To achieve the above purpose, the application adopts the following technical scheme: an intelligent energy optimization management method based on the Internet of Things, comprising the following steps: A sensing unit deployed in the equipment and the environment collects physical quantities and environmental parameters to generate multi-source heterogeneous data; the multi-source heterogeneous data is normalized to generate structured records and construct an equipment operation feature vector; the equipment operation state is determined according to the feature vector and a preset state classification system, the load demand in the future period is predicted based on the equipment operation state and the environmental parameters, and a load interval estimate is generated; the optimal energy distribution scheme is calculated by minimizing the energy cost and equipment loss, combining the load prediction result and system constraints; the optimal energy distribution scheme is converted into a dispatching instruction, which is issued to the equipment controller and the execution effect is monitored, and a feedback report is generated according to the actual operation data.

[0006] Preferably, the collection of the multi-source heterogeneous data comprises the following steps: Install edge computing-enabled sensing nodes on the device to simultaneously read analog and digital signals; Access the time synchronization system wirelessly or via wired connection to ensure data timestamp consistency; Interpolate and align environmental parameters and equipment status data to generate a unified time series.

[0007] Preferably, the construction of the device operation feature vector includes the following steps: The average value and standard deviation of the parameters of the multi-source heterogeneous data within the time window are calculated; The runtime percentage of the multi-source heterogeneous data is statistically analyzed to generate a runtime intensity index. The parameters and intensity indices are combined into a feature vector, which is used as the input to the state classification system.

[0008] Preferably, the state classification system includes the following criteria: High-efficiency operation: load rate between 0.75 and 1.0; Inefficient operation: Load rate between 0.1 and 0.4 and output efficiency less than 80% of the rated value; No-load operation: The load rate is less than 0.1 and the current fluctuation is small.

[0009] Preferably, the load demand forecasting includes the following steps: By combining historical load data with the output of the state classification system, a preliminary load forecast for future periods is generated. Based on the equipment state transition probabilities of the state classification system, the preliminary load forecast value is adjusted to obtain the corrected load forecast value; The revised load forecast is extended by range based on environmental parameters and price signals to generate a load range estimate, which provides input for optimization calculations.

[0010] Preferably, the optimization calculation includes the following steps: Define an objective function, which is the sum of energy costs and equipment losses, and use it as the object to be minimized; Set constraints on power balance, equipment capacity, environmental comfort, and minimum operating time, wherein the power balance is based on the load interval estimate, and the constraints are used to limit the range of feasible solutions; Using the objective function as the optimization objective and combining it with the constraints, the optimal control input is solved through numerical methods to generate device-level scheduling instructions.

[0011] Preferably, the monitoring of the execution effect includes the following steps: The system sends out scheduling instructions through the communication network and receives confirmation signals from the equipment to confirm the status of instruction reception. Upon receiving the confirmation signal, the system collects the actual operating data of the equipment in real time and compares it with the target parameters in the scheduling instructions to generate parameter deviations. If the parameter deviation exceeds the allowable range, an exception handling mechanism is triggered, which performs compensation operations based on scheduling instructions and actual operating data.

[0012] Preferably, generating the feedback report includes the following steps: Based on the actual operation data of the execution monitoring, the number of equipment response scheduling instructions and the total number of instructions are counted, the scheduling execution rate is calculated, and the energy saving amount is calculated in combination with the benchmark energy consumption model. Extract abnormal events from execution monitoring and generate abnormal event logs; The actual operating cost is compared with the target cost in the optimization calculation to generate a difference value; The system summarizes scheduling execution rate, energy savings, abnormal event records, and discrepancies, and generates a feedback report.

[0013] Preferably, the normalization processing of the multi-source heterogeneous data includes the following steps: Convert the output formats of different devices into a unified engineering unit to generate standardized data; Interpolate or downsample the standardized data to align the time series and generate structured data; The structured data is uploaded to the data center via an encrypted channel to provide input for the construction of the feature vector.

[0014] On the other hand, this invention proposes an intelligent energy optimization management system based on the Internet of Things, comprising: Sensing layer: Sensing units deployed in devices and the environment to collect physical quantities and environmental parameters; Transport layer: Enables data uploading through time synchronization systems and communication protocols; Processing layer: performs data normalization, feature extraction, state assessment, load forecasting, and optimization calculations; Execution layer: issues scheduling instructions, monitors execution results, and generates feedback reports.

[0015] Technical effects and advantages of the present invention: The intelligent energy optimization management method and system based on the Internet of Things proposed in this invention have the following advantages compared with the prior art: This method achieves fully automated management across the entire chain, from data acquisition and status identification to load forecasting and optimization decision-making. By deploying sensing units on equipment and in the environment to synchronously collect multi-dimensional physical quantities and environmental parameters, multi-source heterogeneous data is generated. This data is then normalized to construct equipment operation feature vectors, which, combined with a pre-defined status classification system, enable fine-grained classification of operating states. Based on the status identification results and environmental parameters, load demand is predicted, generating load interval estimates. Furthermore, with the goal of minimizing energy costs and equipment losses, the optimal energy allocation scheme is calculated based on system constraints, and the scheduling strategy is continuously optimized through a closed-loop monitoring and feedback mechanism. By constructing equipment operation feature vectors and combining them with a status classification system, the accuracy of equipment operation status identification is significantly improved, providing a reliable basis for load forecasting and effectively reducing forecast deviations. The optimal energy allocation scheme generated on this basis is more dynamic and adaptable, significantly reducing energy costs and equipment losses, and improving energy utilization efficiency and system response speed. Attached Figure Description

[0016] Figure 1 This is a flowchart of an intelligent energy optimization management method based on the Internet of Things according to the present invention; Figure 2 This is a block diagram of an IoT-based intelligent energy optimization management system according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention provides, for example Figure 1 This paper presents an IoT-based intelligent energy optimization management method that utilizes sensing units widely deployed across various energy-consuming devices and energy supply nodes to achieve dynamic, continuous, and high-density data collection of energy usage processes. Based on this, and combining multi-dimensional information such as environmental parameters, equipment load, operating cycles, and external energy market prices, a real-time responsive energy dispatch and allocation framework is constructed. This approach minimizes energy intensity per unit of output, reduces carbon emissions, and improves the overall system's operational economy and eco-friendliness while meeting functional requirements. This method is applicable to various energy-consuming scenarios, including industrial manufacturing, commercial buildings, data centers, and urban public facilities.

[0019] In this embodiment, an IoT-based smart energy optimization management method includes the following steps: Physical quantities and environmental parameters are collected by sensing units deployed on devices and in the environment to generate multi-source heterogeneous data. Specifically, the following steps are included: installing sensing nodes with edge computing capabilities on the devices to synchronously read analog and digital signals; connecting to a time synchronization system via wireless or wired means to ensure data timestamp consistency; and interpolating and aligning environmental parameters and device status data to generate a unified time series.

[0020] By installing edge computing-enabled sensing nodes on the equipment, analog and digital signals are simultaneously read to achieve real-time acquisition of multi-dimensional physical quantities and environmental parameters such as voltage, current, power, temperature, pressure, flow rate, rotation speed, start / stop status, and ambient temperature, humidity, illuminance, air quality, and personnel density. A time synchronization system is connected wirelessly or via wired connection to ensure that the timestamps of data collected by sensors at different locations and of different types remain consistent. Based on this, environmental parameters and equipment status data are interpolated and aligned to eliminate sampling timing discrepancies and generate a unified time series. This achieves precise alignment and structured integration of multi-source heterogeneous data in the time dimension, avoiding misjudgments of status due to asynchronous sampling, and providing a highly consistent and complete data foundation for subsequent feature vector construction and status recognition.

[0021] This method normalizes multi-source heterogeneous data to generate structured records and constructs equipment operation feature vectors. Specifically, it includes the following steps: calculating the average and standard deviation of parameters for multi-source heterogeneous data within a time window; statistically analyzing the runtime proportion of multi-source heterogeneous data to generate operational status intensity indicators; and combining the parameters and intensity indicators into a feature vector, which serves as input to the status classification system. This method transforms multi-source heterogeneous data from raw signals into quantifiable and comparable feature representations, effectively extracting key behavioral patterns of equipment operation and enhancing the consistency and discriminative power of status identification across different equipment and operating conditions.

[0022] Furthermore, the normalization process for multi-source heterogeneous data includes the following steps: converting the output formats of different devices into unified engineering units to generate standardized data; interpolating or downsampling the standardized data to achieve time series alignment and generate structured data; and uploading the structured data to the data center via an encrypted channel to provide input for feature vector construction. This achieves a unified representation of multi-source heterogeneous data at the spatial, temporal, and semantic levels, significantly improving data usability and consistency, and providing a high-quality, time-aligned, and format-uniform input data foundation for the accurate construction of subsequent feature vectors.

[0023] The system determines equipment operating status based on feature vectors and a pre-defined state classification system. It then predicts future load demand based on equipment operating status and environmental parameters, generating load range estimates. The specific state classification system includes the following criteria: high-efficiency operation: load rate between 0.75 and 1.0; low-efficiency operation: load rate between 0.1 and 0.4 and output efficiency below 80% of the rated value; no-load operation: load rate below 0.1 and small current fluctuations. This system achieves accurate and quantifiable classification of equipment operating status, effectively identifying energy-wasting states such as inefficiency and no-load, providing direct evidence for energy efficiency diagnosis. The load range prediction based on the coupling relationship between state evolution and the environment overcomes the shortcomings of traditional point prediction, which is easily affected by fluctuations.

[0024] With the goal of minimizing energy costs and equipment losses, the optimal energy allocation scheme is calculated by combining load forecasting results and system constraints. Specifically, the process includes the following steps: generating preliminary load forecasts for future periods by combining historical load data and the output of the state classification system; adjusting the preliminary load forecasts based on the equipment state transition probabilities of the state classification system to obtain revised load forecasts; and expanding the revised load forecasts into intervals based on environmental parameters and price signals to generate load interval estimates, providing input for optimization calculations.

[0025] The optimization calculation includes the following steps: defining an objective function, which is the sum of energy costs and equipment losses, used as the minimization target; setting constraints on power balance, equipment capacity, environmental comfort, and minimum operating time, where power balance is based on load interval estimation, and constraints are used to limit the range of feasible solutions; using the objective function as the optimization objective, combined with the constraints, solving for the optimal control input through numerical methods to generate equipment-level scheduling instructions. This achieves coordinated optimization of energy costs and equipment losses, avoiding the problems of excessive equipment start-up and shutdown or excessive energy consumption caused by traditional single-objective optimization; by introducing power balance constraints based on load interval estimation, the robustness of the optimization scheme to prediction deviations is enhanced, ensuring the reliability of energy supply.

[0026] The optimal energy allocation scheme is transformed into scheduling instructions, which are then sent to the equipment controller and the execution effect is monitored. Feedback reports are generated based on actual operating data.

[0027] The monitoring of execution results includes the following steps: issuing scheduling instructions through the communication network and receiving confirmation signals from the device to confirm the instruction reception status; after receiving the confirmation signal, collecting the actual operating data of the device in real time and comparing it with the target parameters in the scheduling instructions to generate parameter deviations; if the parameter deviations exceed the allowable range, triggering the anomaly handling mechanism, which performs compensation operations based on the scheduling instructions and the actual operating data.

[0028] Furthermore, generating a feedback report includes the following steps: based on the actual operating data of the execution monitoring, count the number of device responses to scheduling instructions and the total number of instructions, calculate the scheduling execution rate, and calculate the energy savings in combination with the baseline energy consumption model; extract abnormal events in the execution monitoring and generate abnormal event records; compare the actual operating cost with the target cost in the optimization calculation and generate a difference value; summarize the scheduling execution rate, energy savings, abnormal event records and difference values, and generate a feedback report.

[0029] A complete closed-loop feedback mechanism was constructed from instruction issuance to execution evaluation, realizing measurable scheduling effects, traceable abnormal processes, and verifiable optimization goals. By quantifying execution rate and energy savings, the transparency and credibility of the energy management process were improved. Based on cost difference analysis, model bias or external interference can be dynamically identified, providing data support for continuous iteration of optimization algorithms and strategy tuning, and significantly enhancing the system's adaptability and long-term operational stability.

[0030] On the other hand, this invention proposes an IoT-based intelligent energy optimization management system. The system consists of a sensing layer, a transmission layer, a processing layer, and an execution layer. These layers interact and transmit commands via a standardized communication protocol, ensuring the integrity and timeliness of information flow. The entire process operates in a cyclical manner, with each cycle comprising six main stages: data acquisition, status assessment, load forecasting, optimization calculation, scheduling execution, and feedback correction. Clear input-output dependencies exist between each stage, forming a closed-loop control structure.

[0031] like Figure 2 As shown, it includes: Sensing layer: Sensing units deployed in devices and the environment to collect physical quantities and environmental parameters; Transport layer: Enables data uploading through time synchronization systems and communication protocols; Processing layer: performs data normalization, feature extraction, state assessment, load forecasting, and optimization calculations; Execution layer: issues scheduling instructions, monitors execution results, and generates feedback reports.

[0032] In addition, the aforementioned sensing layer, transmission layer, processing layer, and execution layer, during execution, are also used to implement other steps of the aforementioned IoT-based smart energy optimization management method, as follows: Step 1: Comprehensive Perception and Fusion of Multi-Source Heterogeneous Data In the process of energy management, the operating state of equipment does not exist in isolation but is jointly affected by internal working conditions and external environments. Therefore, the prerequisite for achieving efficient management is to obtain comprehensive, accurate, and synchronous operation data. In this step, sensing units with self-identification and self-calibration capabilities are deployed on various energy terminal devices (such as motors, air-conditioning units, lighting systems, boilers, frequency converters, etc.) and energy input nodes (such as electricity meters, water meters, gas meters, heat metering devices) to continuously collect physical quantities such as voltage, current, power, temperature, pressure, flow rate, rotational speed, start / stop status, etc. At the same time, environmental monitoring nodes are arranged in the spatial dimension to obtain auxiliary information such as indoor and outdoor temperature and humidity, light intensity, air quality, and personnel density, providing context support for subsequent analysis, as follows: Establish a device-level data acquisition network: First, install intelligent sensing nodes with edge computing capabilities on each energy-consuming device to be monitored. The node is built with a multi-channel signal acquisition circuit that can simultaneously access analog quantities (such as 4 - 20mA current signals) and digital quantities (such as RS485 communication interfaces) to achieve synchronous reading of device operation parameters. For example, for a centrifugal chiller, the node will obtain the three-phase current of the compressor in real time , operating frequency , condenser outlet temperature , and evaporator inlet / outlet water temperature , . After preliminary filtering and unit conversion of these raw data locally, they are timestamped and encapsulated into standard data packets.

[0033] To ensure the time consistency of data, all nodes are connected to a unified time synchronization system via wireless or wired means and use the Network Time Protocol (NTP) or Precision Time Protocol (PTP) for clock alignment to ensure the comparability of data collected at different locations. The time synchronization error is controlled within milliseconds, which is crucial for subsequent cross-device linkage analysis. For example, when analyzing the collaborative energy-saving effect of the air-conditioning system and the lighting system, it is necessary to ensure that the data records of both are under the same time reference.

[0034] Build a spatial environment perception subsystem: Deploy an environmental perception network inside the building or factory area. This network consists of several fixed and mobile sensing units for capturing dynamic changes in the spatial dimension. Fixed nodes are installed in key areas (such as main entrances and exits, equipment rooms, and the center of the office area) to continuously monitor environmental temperature and humidity , , illuminance , and carbon dioxide concentration . Mobile nodes can be integrated into inspection robots or handheld terminals to conduct supplementary sampling during specific periods to make up for the spatial blind spots of fixed nodes.

[0035] The introduction of environmental data expands energy management beyond the efficiency of individual equipment to encompass the collaborative optimization of "human-machine-environment." For example, when a decrease in population density and sufficient lighting are detected in an area, the system can automatically lower lighting levels and reduce fresh air supply, thereby preventing energy waste. Such decisions rely on the joint analysis of environmental parameters and equipment status; therefore, the synchronization and integrity of these two types of information must be ensured during the data acquisition phase.

[0036] Achieving normalization and fusion of multi-source data: Because different devices and sensors have different output formats, units, and update frequencies, directly using raw data can lead to difficulties in subsequent processing. Therefore, before the data is uploaded to the central processing unit, preliminary data cleaning and format conversion need to be completed at the edge or gateway layer. Specifically, all physical quantities are mapped to a unified engineering unit system, and the time series is aligned using interpolation or downsampling methods.

[0037] Suppose the historical power sequence of a certain device Sampling was performed at 1-second intervals, while the ambient temperature... Updated at 5-second intervals. To achieve alignment between the two on the timeline, a linear interpolation method is used to generate... Estimated values ​​at 1-second intervals: ; in At a time point of 1 second, and The most recent 5-second sampling point is used. This formula ensures that environmental variables still have reasonable estimates at finer-grained time scales, facilitating correlation analysis with high-frequency equipment data.

[0038] After normalization, the data is organized into structured records. Each record contains fields such as device ID, timestamp, parameter name, value, and unit, and is uploaded to the data center via a secure, encrypted channel. This process not only improves data quality but also lays the foundation for subsequent status identification and trend prediction.

[0039] Establish equipment operation feature vectors: After data collection and fusion are completed, in order to further refine the information, an operational feature vector is generated for each device within each time window. This vector comprehensively reflects the overall operating status of the equipment during that time period. Let the equipment... In the time window Internal collection The parameters are as follows: Then its eigenvector is defined as: ; in For the first The average value of each parameter within the window. Its standard deviation, This represents the percentage of time the device was in operation during that period (i.e., the start-stop ratio). This vector not only contains static level information but also reflects dynamic fluctuation characteristics and usage intensity, providing rich input for subsequent condition assessment.

[0040] For example, if the standard deviation of the current of a certain motor A significant increase may indicate uneven load distribution or mechanical wear; a consistently low percentage of runtime may indicate over-configuration or improper scheduling. By constructing such feature vectors, the system can extract more discriminative information from the raw data, supporting deeper analysis.

[0041] This step achieves a comprehensive mapping from the physical world to the digital space, solving the problems of "invisibility, inaccuracy, and inaccessibility" in traditional energy management. By establishing a sensing network covering devices and the environment, the system gains unprecedented breadth and depth of data. Data normalization and fusion technologies eliminate barriers caused by heterogeneity, ensuring that information from different sources can be processed within the same framework. The construction of feature vectors completes the initial abstraction from "data" to "state," providing structured input for subsequent analysis. The completion of this stage enables the system to possess "sensing" capabilities, laying a solid foundation for intelligent decision-making.

[0042] Step Two: Dynamic Assessment and Classification of Equipment Operating Status Based on obtaining the equipment's operational feature vector, the next step is to determine the equipment's current operational status. Operational status includes not only basic modes such as "running" or "shutdown," but also multiple dimensions such as efficiency range, load level, and health status. Accurate status assessment helps identify potential energy-saving opportunities and operational risks, and is a prerequisite for developing optimization strategies, as detailed below: Define a state classification system: To achieve systematic evaluation, a universal status classification system is first established. This system divides equipment operating status into several mutually exclusive and complete categories, such as: high-efficiency operation, normal operation, inefficient operation, no-load operation, standby, fault warning, and maintenance requirements. Each status corresponds to a set of quantifiable criteria, which are derived from the equipment's design parameters, historical operating data, and industry experience.

[0043] Taking motor-type equipment as an example, its efficiency typically varies with the load rate, and there exists an optimal efficiency range. Let the rated power of the motor be... The measured average power is The load factor is then defined as: ; According to the general law of motor efficiency curves, when When defined as "efficient operation"; when When it is "normal operation"; when At that time, it was considered "inefficient operation"; when Furthermore, the current fluctuation is relatively small, which is considered "no-load operation". This classification standard can be adjusted according to the type of equipment to form differentiated evaluation rules.

[0044] State matching based on feature vectors: Using the feature vector generated in step one The vector is compared with preset state criteria to determine the current state of the device. This process can be implemented through a rule engine, which performs conditional judgments on each item in the vector and arrives at a final conclusion based on logical combinations.

[0045] For example, to determine whether a chiller unit is in an "inefficient operating" state, the following conditions must be met simultaneously: load rate ; Condensation temperature (Design value plus 5 degrees); Energy efficiency ratio ; in It can be calculated using the temperature difference between the inlet and outlet water and the flow rate: ; in, The specific heat capacity of water, For density, This is the volumetric flow rate. This formula converts temperature and flow rate data into actual cooling capacity, and is a key basis for evaluating the output efficiency of the equipment.

[0046] When all conditions are met, the system determines that the device is operating inefficiently and triggers subsequent optimization actions. The transparency of the rule engine makes the evaluation process traceable and verifiable, avoiding the uncertainty caused by black-box decision-making.

[0047] Introducing time series analysis to identify anomalous patterns: In addition to assessing static parameters, it is also necessary to pay attention to the dynamic evolution of equipment operating status. Some problems (such as progressive wear and control misalignment) may not be obvious in a single period, but will show specific trends over time. Therefore, sliding window analysis is performed on the historical series of key parameters to identify abnormal fluctuation patterns.

[0048] Suppose the power sequence of a certain device in the past The mean within each time window is Calculate its moving average With moving standard deviation : ; in This is the window width. If the current... If this continues for more than 3 cycles, it may indicate a decrease in the device's output capability; if A continuous increase indicates a decline in operational stability.

[0049] This type of trend analysis compensates for the limitations of instantaneous judgment and enhances the system's predictive capabilities. For example, a gradual increase in the current fluctuation of an air conditioner compressor may indicate wear on the internal valve plates. Although it may not yet affect the cooling effect, it provides an early warning signal for subsequent maintenance.

[0050] Generate a device status distribution map: After completing the status identification of all devices, the system generates a global status distribution map, showing the quantity and proportion of each type of device in different states at the current moment. This map not only reflects the overall operational health but can also be used to identify systemic problems.

[0051] For example, if more than 30% of lighting circuits are found to be in "standby" mode (i.e., off but still consuming a small amount of power), it indicates a power management deficiency; if multiple motors are simultaneously in "inefficient operation" mode, it may suggest unreasonable production scheduling or uneven load distribution. The map is sliced ​​and analyzed by region, system, equipment type, and other dimensions to help managers quickly locate the source of the problem.

[0052] This step represents a leap from "data" to "cognition," enabling the system to understand equipment operating conditions. By establishing a structured state classification system, complex operational behaviors are transformed into actionable category labels. Rule-driven state matching ensures the objectivity and consistency of judgments, while time series analysis enhances sensitivity to slow degradation issues. The resulting state map provides managers with an intuitive global view, becoming a crucial basis for formulating optimization strategies. The output of this stage—equipment state information—will become an important input for the next stage of load forecasting, forming a logically progressive relationship.

[0053] Step 3: Load demand forecasting based on conditions and environment Having grasped the current operational status, the next step is to predict changes in energy demand over a future period. Accurate load forecasting is a prerequisite for proactive dispatching, helping to avoid energy supply delays or surpluses, thereby improving system responsiveness and resource utilization.

[0054] Construct a set of load impact factors: Load demand is influenced by a variety of factors, including the current state of equipment, production plans, environmental conditions, and personnel activity patterns. Therefore, it is essential to first identify the main factors affecting load changes and quantify them as variables that can be input into the forecasting process.

[0055] Set future time period Total load Influenced by the following factors: Number of currently operating devices ; Percentage of equipment in "inefficient operation" ; outdoor temperature ; Expected population density ; Time-of-use electricity price signal ; These factors come from the state assessment results in step two, environmental monitoring data, and external information systems, forming the basic input set for prediction.

[0056] Establish a load change trend model: Based on historical data and current factor values, a trend extrapolation method is used to predict the direction of load changes. Let the past... The load sequence for each time period is as follows Calculate the slope of its linear trend. : ; in This is the sequence mean. If... This indicates that the load is trending upward; if If so, it indicates a downward trend.

[0057] This trend value will serve as the baseline direction for forecasting. For example, if the current trend is upward and the outdoor temperature continues to rise, the air conditioning load is likely to increase further; conversely, if the trend is downward and the population density decreases, the lighting and electrical outlet loads may decrease.

[0058] Introducing state transition probability to correct predicted values: Equipment state changes follow certain patterns, and future state distribution can be predicted by statistically analyzing historical state transition frequencies. Let the equipment start from state... Transition to state The probability is This value was derived from long-term observation data.

[0059] For example, statistics show that the probability of a lighting circuit in "standby" mode switching to "running" mode in the next cycle is... The probability of "Running" switching to "Shutting down" is... Based on the current number of devices in each state, the total number of devices in operation in the future can be predicted: ; This forecast will be used to revise the total load estimate. If the number of operating devices is expected to increase, the load forecast will be revised upwards; otherwise, it will be revised downwards.

[0060] Output future load range estimates: By combining trend analysis and state transition forecasting, load range estimates for future periods are generated. Let the baseline forecast value be... The adjustment factor considering environmental and price factors is: The final prediction interval is: ; in This is an uncertainty factor, dynamically adjusted based on the prediction period length and factor volatility. Short-term (e.g., 15-minute) predictions... Smaller values ​​are associated with longer periods (such as 24 hours), while larger values ​​are associated with longer periods (such as 24 hours).

[0061] This range estimate reflects the confidence level of the forecast, providing flexibility for subsequent optimization. For example, during peak electricity price periods, if the predicted load is in the high range, energy-saving measures should be prioritized; if it is in the low range, controls can be appropriately relaxed.

[0062] This step extends the understanding from "current situation" to "future prediction," giving the system a forward-looking capability. By integrating equipment status, environmental parameters, and behavioral patterns, a multi-factor-driven load forecasting framework is constructed. Trend analysis captures the inertial characteristics of the load, while state transition probabilities reflect the regularity of equipment behavior. The introduction of interval estimation enhances the robustness of the forecast and avoids the decision-making risks associated with single-value predictions. The output of this stage—future load forecasts—will become the core input for the fourth step of optimization calculations, used to balance supply and demand.

[0063] Step 4: Optimization calculation of energy allocation scheme Based on future load forecasts, the next step is to develop an optimal energy allocation plan. This plan must minimize total energy consumption or operating costs while meeting functional requirements, taking into account equipment capacity and system constraints, as follows: Establish the optimization objective function: The optimization objective is defined as minimizing the future time period. Comprehensive operating costs within This cost includes both energy expenses and equipment depreciation. ; in For all controllable devices, For its power, For control inputs (such as set temperature, speed). Let be the loss function. This is a weighting coefficient used to balance economic efficiency and equipment lifespan.

[0064] The construction of the objective function reflects the idea of ​​multi-objective trade-offs: on the one hand, it pursues the lowest energy cost, and on the other hand, it avoids premature aging of equipment due to frequent start-stop or extreme operating conditions.

[0065] Set system constraints: The optimization process must satisfy a series of physical and operational constraints, including: Power balance: That is, the total output equals the predicted load; Equipment capabilities: To prevent overload or ineffective operation; Environmental comfort: Indoor temperature ; Minimum runtime: Once the device is started, it must run continuously for at least [duration]. Time constraints. These constraints ensure that the optimization results are feasible in reality, avoiding a disconnect between theoretical optimality and practical feasibility.

[0066] Solving for the optimal control input: Under the objective function and constraints, the optimal control input for each device is solved numerically. For linearly separable problems, the simplex method can be used for fast solutions; for nonlinear problems, iterative approximation methods are used.

[0067] Assume the air conditioning system has Several adjustable parameters (such as supply air temperature, fan speed, and water valve opening) are combined to form a control vector. Update the control variable using gradient descent: ; in Step size, The gradient of the cost function is used. Iteration continues until convergence, yielding the optimal solution. .

[0068] Generate device-level scheduling instructions: The optimization results are translated into control commands for specific devices, such as temperature settings, start / stop signals, and power limits. These commands include execution time, target values, and priority information, and are sent to the respective device controllers via a secure channel.

[0069] For example, for a group of pumps operating in parallel, optimization might require shutting down one inefficient pump while adjusting the speed of the remaining pumps to their high-efficiency range. Such instructions directly guide equipment operation, enabling precise energy allocation.

[0070] This step represents the transformation from "prediction" to "decision-making" and is the core of the entire process. By constructing an objective function that incorporates economic and lifespan factors, the system can find a balance point among multiple objectives. The introduction of constraints ensures the feasibility of the solution, while numerical solution methods provide the implementation path. The final generated scheduling instructions provide clear operational guidelines for the execution layer. The output of this stage—the optimized scheduling scheme—directly drives the execution actions in step five.

[0071] Step 5: Execution and Feedback of Scheduling Instructions The optimized scheduling scheme is implemented, and the execution effect is monitored in real time to form a closed-loop control, as follows: Command issuance and device response: Dispatch commands are sent to the local controllers of each device via wired or wireless communication networks. The controllers parse the command content, adjust the device operating parameters, and send back confirmation information.

[0072] To ensure reliable delivery of instructions, an acknowledgment and retransmission mechanism is adopted: if the sender does not receive an acknowledgment signal from the receiver within a specified time, the instruction is retransmitted until successful or the maximum number of retries is reached.

[0073] Execution process monitoring: During command execution, actual equipment operating data is continuously collected to verify whether the adjustments are made as expected. For example, if the command requires adjusting an air conditioner's set temperature from 24°C to 26°C, the actual set value and the trend of indoor temperature changes are monitored. If a deviation exceeding the allowable range (e.g., ±0.5°C) is detected, an anomaly handling process is triggered, attempting to reissue the command or activate backup equipment.

[0074] Effect evaluation and data feedback: After the scheduling cycle ends, the actual energy consumption, environmental parameters, and equipment status data are summarized to calculate the actual operating cost. and optimization objectives Comparison: ; like A persistently large value indicates that the prediction or optimization model needs adjustment; if This indicates that the system is running well.

[0075] Generate feedback report: The execution results are compiled into a feedback report, including scheduling execution rate, energy savings, and abnormal event records, for management review. Simultaneously, key data is fed back into the historical database for future learning and improvement.

[0076] This step completes the transition from "decision-making" to "action," achieving a closed-loop management system. The instruction issuance mechanism ensures the timeliness and reliability of control, while execution monitoring guarantees the effective implementation of the plan. Effectiveness evaluation provides quantitative verification methods, and feedback reports support continuous improvement.

[0077] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart energy optimization management method based on the Internet of Things, characterized in that, Includes the following steps: Physical quantities and environmental parameters are collected by sensing units deployed in devices and the environment to generate multi-source heterogeneous data; The multi-source heterogeneous data is normalized to generate structured records and a device operation feature vector is constructed. The equipment operating status is determined based on the feature vector and the preset state classification system. Based on the equipment operating status and environmental parameters, the load demand for future periods is predicted, and a load range estimate is generated. With the goal of minimizing energy costs and equipment losses, the optimal energy allocation scheme is calculated by combining load forecasting results and system constraints. The optimal energy allocation scheme is converted into scheduling instructions, which are then sent to the equipment controller and the execution effect is monitored. A feedback report is generated based on the actual operating data.

2. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The acquisition of the multi-source heterogeneous data includes the following steps: Install edge computing-enabled sensing nodes on the device to simultaneously read analog and digital signals; Access the time synchronization system wirelessly or via wired connection to ensure data timestamp consistency; Interpolate and align environmental parameters and equipment status data to generate a unified time series.

3. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The construction of the device operation feature vector includes the following steps: The average value and standard deviation of the parameters of the multi-source heterogeneous data within the time window are calculated; The runtime percentage of the multi-source heterogeneous data is statistically analyzed to generate a runtime intensity index. The parameters and intensity indices are combined into a feature vector, which is used as the input to the state classification system.

4. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The state classification system includes the following criteria: High-efficiency operation: load rate between 0.75 and 1.0; Inefficient operation: Load rate between 0.1 and 0.4 and output efficiency less than 80% of the rated value; No-load operation: The load rate is less than 0.1 and the current fluctuation is small.

5. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The load demand forecasting includes the following steps: By combining historical load data with the output of the state classification system, a preliminary load forecast for future periods is generated. Based on the equipment state transition probabilities of the state classification system, the preliminary load forecast value is adjusted to obtain the corrected load forecast value; The revised load forecast is extended by range based on environmental parameters and price signals to generate a load range estimate, which provides input for optimization calculations.

6. The intelligent energy optimization management method based on the Internet of Things according to claim 5, characterized in that, The optimization calculation includes the following steps: Define an objective function, which is the sum of energy costs and equipment losses, and use it as the object to be minimized; Set constraints on power balance, equipment capacity, environmental comfort, and minimum operating time, wherein the power balance is based on the load interval estimate, and the constraints are used to limit the range of feasible solutions; Using the objective function as the optimization objective and combining it with the constraints, the optimal control input is solved through numerical methods to generate device-level scheduling instructions.

7. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The monitoring execution effect includes the following steps: The system sends out scheduling instructions through the communication network and receives confirmation signals from the equipment to confirm the status of instruction reception. Upon receiving the confirmation signal, the system collects the actual operating data of the equipment in real time and compares it with the target parameters in the scheduling instructions to generate parameter deviations. If the parameter deviation exceeds the allowable range, an exception handling mechanism is triggered, which performs compensation operations based on scheduling instructions and actual operating data.

8. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The generation of the feedback report includes the following steps: Based on the actual operation data of the execution monitoring, the number of equipment response scheduling instructions and the total number of instructions are counted, the scheduling execution rate is calculated, and the energy saving amount is calculated in combination with the benchmark energy consumption model. Extract abnormal events from execution monitoring and generate abnormal event logs; The actual operating cost is compared with the target cost in the optimization calculation to generate a difference value; The system summarizes scheduling execution rate, energy savings, abnormal event records, and discrepancies, and generates a feedback report.

9. The intelligent energy optimization management method based on the Internet of Things according to claim 1, characterized in that, The normalization process for the multi-source heterogeneous data includes the following steps: Convert the output formats of different devices into a unified engineering unit to generate standardized data; Interpolate or downsample the standardized data to align the time series and generate structured data; The structured data is uploaded to the data center via an encrypted channel to provide input for the construction of the feature vector.

10. An IoT-based intelligent energy optimization management system for implementing the method as described in any one of claims 1-9, characterized in that, include: Sensing layer: Sensing units deployed in devices and the environment to collect physical quantities and environmental parameters; Transport layer: Enables data uploading through time synchronization systems and communication protocols; Processing layer: performs data normalization, feature extraction, state assessment, load forecasting, and optimization calculations; Execution layer: issues scheduling instructions, monitors execution results, and generates feedback reports.

Images