An industrial equipment energy consumption monitoring method and system based on the Internet of Things

Abstract

The application discloses an industrial equipment energy consumption monitoring method and system based on Internet of Things, relates to the technical field of Internet of Things monitoring, and solves the technical problems of inaccurate time sequence calibration of multi-source energy consumption monitoring data of industrial equipment and inaccurate energy consumption analysis results caused by rigid data extraction mode; comprising: acquiring energy consumption monitoring data of a plurality of monitoring devices on each industrial equipment; performing time sequence calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring data set; performing data extraction on each calibrated change curve in the energy consumption monitoring data set to obtain to-be-analyzed energy consumption data; inputting parameter values of a plurality of energy consumption parameters corresponding to each moment in the to-be-analyzed energy consumption data into an energy consumption analysis model to obtain corresponding energy consumption monitoring results; and performing time sequence synchronous calibration and adaptive parameter value extraction on each energy consumption monitoring data, so that the accuracy and reliability of the analysis data are ensured, and the accuracy of the analysis results obtained subsequently based on the data is ensured.

Description

Technical Field

[0001] This application belongs to the field of Internet of Things (IoT) monitoring technology, specifically an IoT-based method and system for monitoring the energy consumption of industrial equipment. Background Technology

[0002] Energy consumption monitoring of industrial equipment is of great significance for ensuring safe operation of equipment, optimizing energy utilization efficiency, and timely detection of energy consumption anomalies. It is a key link in the field of industrial Internet of Things.

[0003] However, the energy consumption data for monitoring industrial equipment is diverse. The same equipment often requires multiple different monitoring devices to monitor its energy consumption data. Each monitoring device has its own independent clock reference. This leads to discrepancies between the analysis results and the actual results when the data is analyzed later due to the time asynchrony between the different data. This affects other operations based on the analysis results. Existing solutions are not precise enough in their calibration methods when dealing with the time synchronization problem of the above-mentioned multi-source energy consumption monitoring data. Moreover, the method of extracting the data to be analyzed from the calibrated data is relatively rigid. This results in poor data quality when inputting into the energy consumption analysis model. It is very easy to cause misjudgment or omission of energy consumption status and cannot provide accurate early warning and handling strategies for energy consumption anomalies in industrial equipment. Summary of the Invention

[0004] This application provides an IoT-based method and system for monitoring the energy consumption of industrial equipment, which solves the technical problems of inaccurate time-series calibration of multi-source energy consumption monitoring data of industrial equipment and inaccurate energy consumption analysis results caused by rigid data extraction methods in the prior art.

[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, a method for monitoring the energy consumption of industrial equipment based on the Internet of Things is provided, including: Acquire energy consumption monitoring data from several monitoring devices on various industrial equipment; perform time-series calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; the energy consumption monitoring dataset includes calibrated change curves corresponding to several monitoring devices and several actual measurement times; Data is extracted from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; the energy consumption data to be analyzed includes parameter values ​​at several times corresponding to several energy consumption parameters; The parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed are input into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results.

[0006] Based on the above technical solution, the IoT-based industrial equipment energy consumption monitoring method and system provided in this application acquires energy consumption monitoring data from several monitoring devices on various industrial equipment; performs time-series calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; extracts data from each calibrated change curve in the energy consumption monitoring dataset to obtain energy consumption data to be analyzed; inputs the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results; performs time-series synchronous calibration on each energy consumption monitoring data, uses physical constraints to solve for the optimal time-series deviation value; and adaptively calculates the neighborhood step size, constructs a stable window, and takes the intersection to determine the selected window set, widening the window when the data is dense and shrinking the window when the data is sparse, accurately extracting the common analysis time and parameter values ​​with high reliability for each monitoring device; ensuring the accuracy and reliability of the analyzed data, thereby ensuring the accuracy of the subsequent analysis results obtained from the data.

[0007] In conjunction with the first aspect above, in one possible implementation, the step of performing time-series calibration on the energy consumption monitoring data of each monitoring device to obtain the energy consumption monitoring dataset includes: Acquire several energy consumption monitoring data belonging to the same industrial equipment; acquire several hard relationships corresponding to the industrial equipment, and several energy consumption parameters corresponding to the hard relationships; the hard relationships include energy relationships between different components or different elements of the same equipment, such as energy conservation, energy conversion and power supply efficiency. Energy consumption monitoring data corresponding to several energy consumption parameters belonging to the same hard relation are divided into the same monitoring data group; based on each hard relation, the energy consumption monitoring data in its corresponding monitoring data group are synchronized in time to obtain the energy consumption monitoring dataset corresponding to the industrial equipment.

[0008] In conjunction with the first aspect above, in one possible implementation, energy consumption monitoring data in the corresponding monitoring data groups are synchronized in time based on each hard relation to obtain the energy consumption monitoring dataset corresponding to the industrial equipment, including: S1: Obtain each energy consumption monitoring data in the energy consumption monitoring data group, and extract the monitoring parameter values ​​collected at each time in the energy consumption monitoring data; fit the several monitoring parameter values ​​into a time-series change curve according to the chronological order of their corresponding times; S2: Obtain the relationship level corresponding to each hard relationship, and sort the hard relationships from high to low in the hard relationship ranking table; the relationship level is the level of different hard relationships in the industrial equipment that is set in advance; S3: Record the energy consumption monitoring data group corresponding to the hard relationship with the highest relationship level in the hard relationship ranking table as the data group to be calibrated; S4: Obtain several time-series variation curves in the data set to be calibrated, as well as the hard relation corresponding to the data set to be calibrated; S5: Substitute each time series variation curve and hard relation into the set time series calibration function to obtain the time series deviation value corresponding to each time series variation curve; S6: Remove the hard relation from the hard relation sorting table, and offset each time-series change curve in the energy consumption monitoring data group on the time axis. The offset amount is the time-series deviation value corresponding to the time-series change curve, and the offset direction is the sign of the time-series deviation value. When the time-series deviation value is positive, the time-series change curve is offset to the right on the time axis; when the time-series deviation value is negative, the time-series change curve is offset to the left on the time axis. The offset time-series change curve is recorded as the calibrated change curve, and its corresponding time-series deviation value is recorded as the calibrated time difference. S7: Determine if there are still hard relationships in the hard relationship sorting table. If yes, proceed to S3; otherwise, generate an energy consumption monitoring dataset based on each calibrated change curve and calibrated time difference.

[0009] In conjunction with the first aspect above, in one possible implementation, the timing calibration function can be expressed in the following form: ; in, This represents the time-series deviation value corresponding to the energy consumption monitoring data numbered n in the energy consumption monitoring data group; This is the hard relation function corresponding to the energy consumption monitoring data set; The time-series variation curve corresponding to the energy consumption monitoring data numbered n in the energy consumption monitoring data group; Shift the time-series change curve corresponding to the energy consumption monitoring data with number n in the energy consumption monitoring data group. The subsequent curve; The deviation of the time-series variation curve under the hard relationship is the amount of deviation. , The set offset period; For any calibrated change curve in the energy consumption monitoring data set, The calibrated time difference corresponding to the change curve. ; The total number of other energy consumption monitoring data remaining in the energy consumption monitoring data group after removing the energy consumption monitoring data corresponding to the calibrated change curve, and n is the number of the other energy consumption monitoring data remaining in the energy consumption monitoring data group after removing the energy consumption monitoring data corresponding to the calibrated change curve.

[0010] In conjunction with the first aspect above, in one possible implementation, generating the energy consumption monitoring dataset based on each calibrated change curve and the calibrated time difference includes: Obtain each calibrated change curve and its corresponding calibrated time difference, as well as the time when each monitoring parameter value in the energy consumption monitoring data corresponding to the calibrated change curve was collected; The time corresponding to each monitoring parameter value is offset on the time axis, and the offset amount is the calibrated time difference corresponding to the calibrated change curve. The offset direction is the sign of the time series deviation value. When the time series deviation value is positive, the time series change curve is offset to the right on the time axis. When the time series deviation value is negative, the time series change curve is offset to the left on the time axis. Several offset times are recorded as the measured times corresponding to the calibrated change curve. Each calibrated change curve and its corresponding measured time points are integrated into an energy consumption monitoring dataset.

[0011] In conjunction with the first aspect above, in one possible implementation, the step of extracting data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed includes: Acquire each calibrated change curve and its corresponding measured times; and the acquisition interval corresponding to the measured times; substitute the acquisition interval into a set neighborhood step size generation function to obtain the corresponding neighborhood step size; one expression of the neighborhood step size generation function includes: ; in, The neighborhood step size; For the collection interval, The set reference acquisition interval is used to quantize each acquisition interval, thereby removing the dimensions. The set proportionality coefficient, and ; In this embodiment ; A stable window is constructed with the measured time corresponding to the calibrated change curve as the center and the neighborhood step size as the radius; several stable windows corresponding to the calibrated change curve are constructed sequentially, and a set of stable windows corresponding to the calibrated change curve is constructed; one form of the stable window set is: ; in, This is the set of stable windows for the calibrated numbered curves corresponding to number n; This indicates the start time of the j-th stable window on the calibrated curve corresponding to number n; This indicates the termination time of the j-th stable window on the calibrated curve corresponding to number n; This represents the number of stable windows corresponding to the calibrated numbered curve corresponding to number n. Obtain the set of stable windows corresponding to each calibrated change curve, and substitute each set of stable windows into the analysis point window selection function to obtain the corresponding selection window set; one representation of the analysis point window selection function includes: ; in, To select a set of windows; The set of stable windows for the calibrated numbered curves corresponding to number n; Select several time points from the selected window set as analysis time points, obtain the parameter values ​​corresponding to each calibrated change curve at the analysis time point, and integrate each energy consumption parameter, as well as several parameter values ​​corresponding to the energy consumption parameter and their corresponding analysis time points into the energy consumption data to be analyzed.

[0012] In conjunction with the first aspect above, in one possible implementation, selecting several moments from the selection window set as analysis moments includes: Extract each selection window from the selection window set and obtain the set analysis step size; set a number of time intervals in the selection window with the analysis step size as the interval, and record them as analysis time; set the analysis time corresponding to each selection window in sequence.

[0013] In conjunction with the first aspect above, in one possible implementation, one training method for the energy consumption analysis model includes: Acquire historical energy consumption monitoring data of industrial equipment, generate corresponding energy consumption data to be analyzed based on the energy consumption monitoring data, extract the parameter values ​​of each energy consumption parameter at each time in the energy consumption data to be analyzed, obtain the energy consumption status at each time, and integrate the parameter values ​​and energy consumption status of each energy consumption parameter at several times into several training data and test data. The artificial intelligence model is trained using training data and tested using validation data. The model is then used to obtain an artificial intelligence model whose input is the parameter values ​​of various energy consumption parameters at several times and whose output is the energy consumption state corresponding to the parameter values ​​at each time. The energy consumption states at each time are integrated into energy consumption monitoring results, and these energy consumption monitoring results are used as the final output to obtain an energy consumption analysis model. The artificial intelligence model includes neural network models, etc.

[0014] Secondly, this application provides an IoT-based industrial equipment energy consumption monitoring system, including: a data acquisition module, a data calibration module, and an energy consumption analysis module. The data acquisition module is used to acquire energy consumption monitoring data from several monitoring devices on various industrial equipment. The data calibration module includes a timing calibration unit and a data extraction unit; The timing calibration unit is used to perform timing calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; the energy consumption monitoring dataset includes calibrated change curves corresponding to several monitoring devices and several actual measurement times. The data extraction unit is used to extract data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; the energy consumption data to be analyzed includes parameter values ​​at several times corresponding to several energy consumption parameters. The energy consumption analysis module is used to input the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results.

[0015] This application provides an IoT-based method and system for monitoring the energy consumption of industrial equipment. It can acquire energy consumption monitoring data from several monitoring devices on various industrial equipment; perform time-series calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; extract data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; input the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into an energy consumption analysis model to obtain the corresponding energy consumption monitoring results; perform time-series synchronous calibration on each energy consumption monitoring data, use physical constraints to solve for the optimal time-series deviation value; and adaptively calculate the neighborhood step size, construct a stable window, and determine the selected window set by taking the intersection. The window is widened when the data is dense and narrowed when the data is sparse, accurately extracting the common analysis time and parameter values ​​with high reliability for each monitoring device; ensuring the accuracy and reliability of the analyzed data, thereby ensuring the accuracy of the subsequent analysis results obtained from the data.

[0016] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram illustrating the steps of the industrial equipment energy consumption monitoring method in this application; Figure 2 This is a schematic diagram of the module connections of the industrial equipment energy consumption monitoring system in this application. Detailed Implementation

[0019] The technical solutions of this application will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0020] Please see Figure 1 The first aspect of this application provides a method for monitoring the energy consumption of industrial equipment based on the Internet of Things, including: The process involves acquiring energy consumption monitoring data from multiple monitoring devices on various industrial equipment. It's understood that industrial equipment typically contains multiple components or parts, requiring the deployment of monitoring devices in different locations to achieve comprehensive energy consumption awareness. These monitoring devices can be built-in data acquisition interfaces or external smart sensors or IoT gateway modules. Energy consumption monitoring data reflects the time-varying sequence of a specific energy consumption parameter for the corresponding component. This parameter is not limited to operating power but can also include various physical quantities characterizing the equipment's energy consumption status, such as operating current, operating voltage, and instantaneous energy consumption. Due to clock drift in different monitoring devices and the fact that data is transmitted via different communication links, such as Wi-Fi, Bluetooth, 5G, and Ethernet, facing varying network congestion and latency, the acquired multi-source heterogeneous data naturally exhibits misalignment and inconsistency on the timeline. This temporal disorder of the raw data constitutes the basis for subsequent accurate analysis. The core obstacle to analysis is to perform time-series calibration on the energy consumption monitoring data of various monitoring devices to obtain an energy consumption monitoring dataset. The energy consumption monitoring dataset includes calibrated change curves corresponding to several monitoring devices and several measured times. Specifically, the purpose of time-series calibration is to eliminate the time axis misalignment caused by clock drift or communication delay in the aforementioned steps, so that the data of different monitoring devices belonging to the same industrial equipment can be aligned in physical time. The calibrated change curve is a continuous function expression obtained after time-series offset correction of the original scattered data or fluctuation curve, which reflects the smooth evolution law of parameters with real physical time. The measured times are the real data acquisition timestamps after being synchronously offset with the calibrated change curves, preserving the discrete time anchors of the original observation values. This embodiment provides a time-consistent and physically self-consistent basic data source for subsequent data extraction by dual calibration of curve shape and timestamps and integration into an energy consumption monitoring dataset. Data is extracted from the calibrated change curves in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed. The energy consumption data to be analyzed includes parameter values ​​at several times corresponding to several energy consumption parameters. Specifically, although the calibrated change curves are continuous, artificial intelligence models usually require structured numerical matrices with fixed dimensions and discrete time steps as input. Therefore, it is necessary to extract discrete parameter values ​​at specific times from the continuous curves. In addition, due to the differences in the acquisition frequency of different monitoring devices, data points are dense and reliable in some intervals of the curve, while data points are sparse and have large fitting errors in other intervals. Simple fixed-interval sampling will cause the extracted values ​​in sparse areas to deviate from the true physical state, while dense areas may introduce redundant noise. Therefore, data extraction is not a simple numerical reading, but requires adaptive screening of high-confidence time intervals and parameter values ​​based on the data distribution characteristics to ensure that the extracted energy consumption data to be analyzed has consistent reliability and representativeness across device dimensions. The energy consumption data to be analyzed is input into an energy consumption analysis model to obtain corresponding energy consumption monitoring results. Specifically, the energy consumption analysis model is a pre-trained artificial intelligence model, such as a neural network model, support vector machine, or decision tree. The model receives a structured numerical matrix corresponding to the energy consumption data to be analyzed, where each row corresponds to an analysis time and each column corresponds to the parameter value of an energy consumption parameter. The model maps the joint distribution state of multi-dimensional parameters to specific energy consumption status labels through the feature mapping relationship learned internally, thereby outputting energy consumption monitoring results. The energy consumption monitoring results not only include the current macroscopic operating status of the equipment, such as low power consumption, normal, high power consumption, and abnormal, but can also be further associated with specific abnormality types or failure probabilities. Since the data input to the model has undergone strict time-series calibration and adaptive extraction, the interference of time-series deviation and low-confidence noise is eliminated, and the model can accurately capture the real energy consumption characteristic fluctuations, avoiding misjudgments and omissions caused by chaotic input data. Subsequently, corresponding processing strategies can be formulated based on the energy consumption monitoring results, such as reducing the energy consumption of some industrial equipment and shutting down for maintenance. The system can automatically formulate and execute corresponding processing strategies, such as reducing the operating power of non-critical components to save energy, or triggering shutdown maintenance warnings when abnormally high power consumption is detected, thereby realizing a complete business closed loop from data perception to intelligent decision-making.

[0021] Based on the above technical solution, the IoT-based industrial equipment energy consumption monitoring method and system provided in this application acquires energy consumption monitoring data from several monitoring devices on various industrial equipment; performs time-series calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; extracts data from each calibrated change curve in the energy consumption monitoring dataset to obtain energy consumption data to be analyzed; inputs the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results; performs time-series synchronous calibration on each energy consumption monitoring data, uses physical constraints to solve for the optimal time-series deviation value; and adaptively calculates the neighborhood step size, constructs a stable window, and takes the intersection to determine the selected window set, widening the window when the data is dense and shrinking the window when the data is sparse, accurately extracting the common analysis time and parameter values ​​with high reliability for each monitoring device; ensuring the accuracy and reliability of the analyzed data, thereby ensuring the accuracy of the subsequent analysis results obtained from the data.

[0022] In one possible implementation, the energy consumption monitoring data of each monitoring device is time-series calibrated to obtain an energy consumption monitoring dataset, including: acquiring several energy consumption monitoring data belonging to the same industrial equipment; acquiring several hard relationships corresponding to the industrial equipment, and several energy consumption parameters corresponding to the hard relationships; the hard relationships refer to the objectively existing energy relationships between different components or between different elements of the same equipment; including energy conservation, energy conversion, and power supply efficiency, etc., between different components or between different elements of the same equipment. For example, a component includes functional sub-component one, functional sub-component two, and heat dissipation sub-component three. At this time, the monitored energy consumption data includes the total power of the component, the partial power of functional sub-component one, the partial power of functional sub-component two, and the partial power of heat dissipation sub-component three. At this time, the total power of the component and each partial power satisfy the law of energy conservation, that is, within a set short time period, the total energy consumption of the component should be equal to the sum of the energy consumption of each sub-component and the product of the efficiency. Energy consumption monitoring data corresponding to several energy consumption parameters belonging to the same hard relation are divided into the same monitoring data group; based on each hard relation, the energy consumption monitoring data in the corresponding monitoring data group are synchronized in time to obtain the energy consumption monitoring dataset corresponding to the industrial equipment; by calibrating the data by grouping according to hard relations, the calibration process is limited to a local data subset with clear physical constraints, avoiding the contradictions caused by global forced alignment across physical logic; in this embodiment, through time synchronization, the data in the same group are not only aligned in time.

[0023] In one possible implementation, energy consumption monitoring data in the corresponding monitoring data group is synchronized in time based on each hard relation to obtain the energy consumption monitoring dataset for industrial equipment, including: S1: Obtain each energy consumption monitoring data in the energy consumption monitoring data group, and extract the monitoring parameter values ​​collected at each time in the energy consumption monitoring data; fit the several monitoring parameter values ​​into a time-series change curve according to the chronological order of their corresponding times; S2: Obtain the relationship level corresponding to each hard relationship, and sort the hard relationships from high to low in the hard relationship ranking table; the relationship level is a pre-defined level of different hard relationships in industrial equipment, and the classification is mainly based on the macroscopic degree of physical coverage: if the hard relationship is the relationship between the total energy consumption of industrial equipment and the energy consumption between each component, then the relationship level corresponding to this hard relationship is the highest, because it constrains the most macroscopic energy input and output balance of the equipment; if the hard relationship is the relationship between the power, current or voltage inside a component, then the relationship level of this hard relationship is lower; if the hard relationship is the relationship between the energy consumption of a component and the energy consumption of the corresponding elements of the component, then this hard relationship is between the above two hard relationships; hard relationships can be power conservation, series and parallel connections corresponding to current, series and parallel connections corresponding to voltage, etc. S3: Record the energy consumption monitoring data group corresponding to the hard relationship with the highest relationship level in the hard relationship ranking table as the data group to be calibrated; S4: Obtain several time-series variation curves in the data set to be calibrated, as well as the hard relation corresponding to the data set to be calibrated; S5: Substitute each time series variation curve and hard relation into the set time series calibration function to obtain the time series deviation value corresponding to each time series variation curve; one expression of the time series calibration function includes: ; in, This represents the time-series deviation value corresponding to the energy consumption monitoring data numbered n in the energy consumption monitoring data group; This is the hard relation function corresponding to the energy consumption monitoring data set; The time-series variation curve corresponding to the energy consumption monitoring data numbered n in the energy consumption monitoring data group; Shift the time-series change curve corresponding to the energy consumption monitoring data with number n in the energy consumption monitoring data group. The subsequent curve; The deviation of the time-series variation curve under the hard relationship is the amount of deviation. , The set offset period; For any calibrated change curve in the energy consumption monitoring data set, The calibrated time difference corresponding to the change curve. ; The total number of other energy consumption monitoring data remaining in the energy consumption monitoring data group after removing the energy consumption monitoring data corresponding to the calibrated change curve, and n is the number of the other energy consumption monitoring data remaining in the energy consumption monitoring data group after removing the energy consumption monitoring data corresponding to the calibrated change curve. This represents the lower limit of the overlapping portion of the various time-series variation curves. This serves as an upper constraint on the overlapping portions of the various time-series variation curves. Specifically, the core of this time-series calibration function lies in the fusion of integral optimization and physical constraints; it solves for the deviation within a continuous time window, rather than relying solely on discrete isolated points, effectively smoothing out the interference of instantaneous noise on the calibration; the hard relation function K is directly integrated into the deviation solution process as a physical constraint, and its physical meaning is: when the various time-series variation curves shift... Then, substitute the offset curve into the theoretical value calculated in the hard relation function K. This value should then match any calibrated change curve within the group. The actual value difference is minimized. In other words, the goal is to find a set of optimal offsets that minimize the residual when the calibrated data is substituted into the hard relation, thus ensuring that the calibration results are not only time-aligned; regarding the offset period T, its endpoint value can be set according to the statistical characteristics of communication delay. For example, if the maximum communication delay of a certain type of sensor is usually no more than 500 milliseconds, then T can be set to 500 milliseconds. exist Searching within the bounded space ensures that the search space covers the actual latency while avoiding the computational redundancy caused by unbounded search.

[0024] S6: Remove the hard relation from the hard relation sorting table, and offset each time-series change curve in the energy consumption monitoring data group on the time axis. The offset amount is the time-series deviation value corresponding to the time-series change curve, and the offset direction is the sign of the time-series deviation value. When the time-series deviation value is positive, the time-series change curve is offset to the right on the time axis; when the time-series deviation value is negative, the time-series change curve is offset to the left on the time axis. The offset time-series change curve is recorded as the calibrated change curve, and its corresponding time-series deviation value is recorded as the calibrated time difference. S7: Determine if there are still hard relationships in the hard relationship sorting table. If yes, proceed to S3; otherwise, generate an energy consumption monitoring dataset based on each calibrated change curve and calibrated time difference.

[0025] In one possible implementation, an energy consumption monitoring dataset is generated based on each calibrated change curve and the calibrated time difference, including: acquiring each calibrated change curve and the corresponding calibrated time difference, as well as the time when the values ​​of each monitoring parameter in the energy consumption monitoring data corresponding to the calibrated change curve were collected; The time corresponding to each monitoring parameter value is offset on the time axis, and the offset amount is the calibrated time difference corresponding to the calibrated change curve. The offset direction is the sign of the time series deviation value. When the time series deviation value is positive, the time series change curve is offset to the right on the time axis, and when the time series deviation value is negative, the time series change curve is offset to the left on the time axis. Several offset times are recorded as the measured times corresponding to the calibrated change curve. It can be understood that the value corresponding to the measured time of the calibrated change curve is recorded as the value collected by its monitoring equipment. Each calibrated change curve and its corresponding measured time points are integrated into an energy consumption monitoring dataset.

[0026] Specifically, the calibrated change curve represents the overall offset of the original scatter plot fitted curve, while the measured time is a synchronous offset of the original discrete acquisition timestamps. The necessity of synchronously offsetting the measured time with the curve lies in the fact that although the fitted curve is continuous, the parameter values ​​at non-measured times are obtained through interpolation, and their reliability is slightly lower than that of the actual acquisition points. Synchronously offsetting the measured time preserves the anchor positions of the original high-confidence data points on the real physical time axis, enabling subsequent data extraction to accurately identify which parameter values ​​are the actual observation values ​​and which are the interpolated values, thus providing a unified discrete anchor point for subsequent adaptive extraction. The construction logic and intersection selection mechanism of the adaptive stable window are further elaborated. It should be understood that traditional data extraction schemes typically use fixed time windows or fixed step sizes for sampling. This rigid extraction method has significant drawbacks: when data points are dense, the fixed window may be too long, causing the fluctuation characteristics under different operating states to be averaged, missing key transient abnormal fluctuations; when data points are sparse, the fixed window may be too short, resulting in a lack of constraints from real sampling points within the window, causing the extracted interpolation parameter values ​​to deviate from the actual physical state, leading to extremely low reliability. Therefore, this embodiment introduces a mechanism for adaptively calculating the neighborhood step size based on the sampling interval and constructing a stable window. To address the aforementioned issues of extraction distortion and insufficient confidence, in one possible implementation, data extraction is performed on each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed. This includes: acquiring each calibrated change curve and its corresponding measured time points; and the acquisition interval corresponding to the measured time points; substituting the acquisition interval into a set neighborhood step size generation function to obtain the corresponding neighborhood step size; wherein, the measured time point is the actual physical time anchor point after time-series calibration, and the acquisition interval reflects the density distribution of the original data points before and after the measured time point; one expression of the neighborhood step size generation function includes: ; in, The neighborhood step size; For the collection interval, The set reference acquisition interval is used to quantize each acquisition interval, thereby removing the dimensions. The set proportionality coefficient, and ; In this embodiment It is understandable that when performing curve fitting based on data points, if the density of data points is high (i.e., the acquisition interval between different data points is short), under the same fitting conditions, the fitted points within the acquisition interval are subject to more constraints, resulting in higher reliability of each point on the fitted curve. This means that the parameters corresponding to points at non-measured times are closer to the actual values. Conversely, if the density of data points is low (i.e., the acquisition interval between different data points is long), under the same fitting conditions, the fitted points within the acquisition interval are subject to fewer constraints, resulting in lower reliability of each point on the fitted curve. Furthermore, the values ​​corresponding to fitted points on the fitted curve that are closer to the actual detection points are more reliable, and the values ​​at the actual detection points are completely reliable. Therefore, by determining the neighborhood step size using the aforementioned neighborhood step size generation function, when the density of real data points is high (i.e., the acquisition interval between real data points is short), the corresponding neighborhood step size can be appropriately extended. When the density of real data points is low (i.e., the acquisition interval between real data points is long), the corresponding neighborhood step size can be appropriately shortened to ensure that the values ​​corresponding to the parameters of each calibrated change curve within its corresponding window are more reliable. A stable window is constructed with the measured time corresponding to the calibrated change curve as the center and the neighborhood step size as the radius. Several stable windows corresponding to the calibrated change curve are constructed sequentially. Each stable window is a local interval defined on the time axis with each measured time as the center and the adaptively calculated neighborhood step size as the radius. Within this interval, the curve shape is strongly constrained by the central measured point, therefore the parameter value at any time within the interval has high confidence. A set of stable windows corresponding to the calibrated change curve is then constructed. One form of expression for the set of stable windows is: ; in, This is the set of stable windows for the calibrated numbered curves corresponding to number n; This indicates the start time of the j-th stable window on the calibrated curve corresponding to number n; This indicates the termination time of the j-th stable window on the calibrated curve corresponding to number n; This represents the number of stable windows corresponding to the calibrated numbered curves corresponding to number n; this set intuitively depicts which time periods of the monitoring device are highly reliable along the entire time axis, providing a local confidence mask for subsequent cross-device screening; Understandably, data within a stable window has higher reliability and can be used as the basis for subsequent analysis of whether industrial equipment is experiencing abnormal energy consumption, thus making the analysis results more accurate. Obtain the set of stable windows corresponding to each calibrated change curve, and substitute each set of stable windows into the analysis point window selection function to obtain the corresponding selection window set; one representation of the analysis point window selection function includes: ; in, To select a set of windows; The set of stable windows for the calibrated numbered curves corresponding to number n; specifically, since the energy consumption status of industrial equipment is determined by the multi-dimensional parameters of multiple monitoring devices, the high reliability interval of a single device cannot guarantee the reliability of cross-device joint analysis; the analysis point window selection function selects common time segments with high confidence in all dimensions by taking the intersection of the stable window sets of all monitoring devices; only when the moment falls within the intersection T, the parameter values ​​of each energy consumption parameter are strongly constrained by the real measured points on each calibrated change curve, thereby ensuring the physical consistency and high reliability of the joint parameter vector of the input model in the cross-device dimension, and completely filtering out local false features caused by the sparsity of data from a single device; Select several time points from the selected window set as analysis time points, obtain the parameter values ​​corresponding to each calibrated change curve at the analysis time point, and integrate each energy consumption parameter, as well as several parameter values ​​corresponding to the energy consumption parameter and their corresponding analysis time points into the energy consumption data to be analyzed.

[0027] In one possible implementation, several time points are selected from the selection window set as analysis time points. This includes: extracting each selection window from the selection window set and obtaining the set analysis step size; setting several time points in the selection windows at intervals of the analysis step size and recording them as analysis time points; and sequentially setting the analysis time points corresponding to each selection window. Specifically, after determining the common time segments with high confidence, instead of randomly sampling scattered points, points are evenly distributed at intervals of the set analysis step size. This even distribution method has significant advantages in temporal continuity: it ensures that the analysis time points are equidistantly distributed on the time axis, so that the extracted energy consumption data to be analyzed constitutes a discrete sequence with a uniform time step size, avoiding spectral distortion and temporal resolution fluctuations caused by random sampling or non-equidistant sampling, and providing a structured and temporally continuous standard input matrix for the subsequent temporal convolution or sequence feature extraction of the energy consumption analysis model. It should be understood that the specific value of the analysis step size can be flexibly set according to the model's requirements for temporal resolution and is not a fixed and immutable constant.

[0028] In one possible implementation, a training method for the energy consumption analysis model includes: acquiring several historical energy consumption monitoring data of industrial equipment; generating corresponding energy consumption data to be analyzed based on the energy consumption monitoring data; extracting parameter values ​​corresponding to each energy consumption parameter at each time point in the energy consumption data to be analyzed; acquiring the energy consumption status corresponding to each time point, wherein the energy consumption status is the current energy consumption status of the industrial equipment obtained by experts based on the parameter values ​​corresponding to the energy consumption parameters at each time point of the corresponding industrial equipment, and the energy consumption status includes low power consumption, normal, high power consumption, abnormal, etc.; and integrating the parameter values ​​corresponding to each energy consumption parameter at several times point and the energy consumption status into several training data and test data. The artificial intelligence model is trained using training data and tested using validation data. The model is then used to obtain an artificial intelligence model whose input is the parameter values ​​of various energy consumption parameters at several times and whose output is the energy consumption state corresponding to the parameter values ​​at each time. The energy consumption states at each time are integrated into energy consumption monitoring results, and these energy consumption monitoring results are used as the final output to obtain an energy consumption analysis model. The artificial intelligence model includes neural network models, etc.

[0029] Specifically, in traditional model training schemes, raw, heterogeneous data from multiple sources with clock drift and communication delays are often directly fed into the neural network. This causes the model to absorb timing deviations and physical logic contradictions as features when learning energy consumption feature mappings. For example, when the total power curve and the partial power curve are misaligned on the time axis, the model may incorrectly learn a false mapping relationship that violates the law of conservation of energy, such as "total power is not equal to the sum of partial power", thus causing serious misjudgment risks during the inference stage. In this embodiment, the historical energy consumption monitoring data used to generate the energy consumption data to be analyzed first undergoes hierarchical iterative calibration based on hard relationships to eliminate time-series bias and ensure physical consistency. Subsequently, it undergoes adaptive stable window extraction based on the acquisition interval to filter out low-reliability interpolation points in sparse regions and retain high-confidence parameter values ​​within the intersection. This high-quality data, free from time-series bias and highly consistent in physical logic, helps the artificial intelligence model learn a true and pure energy consumption feature mapping, avoiding the risk of model misjudgment caused by chaotic training data, and significantly improving the model's generalization ability and inference accuracy in complex industrial scenarios.

[0030] The artificial intelligence models include, but are not limited to, neural network models, such as convolutional neural networks (CNN), recurrent neural networks (RNN), or long short-term memory networks (LSTM), or algorithm models that can handle multi-dimensional time-series feature classification, such as support vector machines, decision trees, or random forests. The specific choice depends on the dimensionality and real-time requirements of the industrial equipment energy consumption data.

[0031] Regarding energy consumption status, specific classification examples can include low power consumption, normal, high power consumption, and abnormal. It should be understood that the classification of energy consumption status is not limited to the above four labels. In practical applications, it can be flexibly expanded or refined according to the operating characteristics and monitoring granularity of different industrial equipment. For example, for some precision machining equipment that requires fine differentiation of load levels, the "normal" status can be further subdivided into "light load normal" and "full load normal"; the "abnormal" status can also be subdivided into "overcurrent abnormal," "short circuit abnormal," or "idling abnormal," etc. These energy consumption statuses are usually obtained by domain experts based on the parameter values ​​corresponding to various energy consumption parameters of the industrial equipment at the corresponding time, combined with the equipment's mechanism model and historical operating experience, representing the objective operational health of the equipment under real physical time. By binding and integrating high-confidence parameter values ​​with expert-labeled objective energy consumption statuses into training and validation data, it ensures that the label ground truth and input features are absolutely aligned on the time axis during the model's supervised learning process, fundamentally eliminating model training collapse caused by misalignment of feature and label time sequences.

[0032] Please see Figure 2 Secondly, this application provides an IoT-based industrial equipment energy consumption monitoring system, including: a data acquisition module, a data calibration module, and an energy consumption analysis module. The data acquisition module is used to acquire energy consumption monitoring data from several monitoring devices on various industrial equipment. The data calibration module includes a timing calibration unit and a data extraction unit; The timing calibration unit is used to perform timing calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; the energy consumption monitoring dataset includes calibrated change curves corresponding to several monitoring devices and several actual measurement times. The data extraction unit is used to extract data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; the energy consumption data to be analyzed includes parameter values ​​at several times corresponding to several energy consumption parameters. The energy consumption analysis module is used to input the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results.

[0033] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0034] How this application works: By acquiring energy consumption monitoring data from several monitoring devices on various industrial equipment; performing time-series calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; extracting data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; inputting the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results; by performing time-series synchronous calibration on each energy consumption monitoring data, using physical constraints to solve for the optimal time-series deviation value; and by adaptively calculating the neighborhood step size, constructing a stable window and taking the intersection to determine the selected window set, widening the window when the data is dense and shrinking the window when the data is sparse, the common analysis time and parameter values ​​of each monitoring device with high reliability are accurately extracted; this ensures the accuracy and reliability of the analyzed data, thereby ensuring the accuracy of the subsequent analysis results obtained from this data.

[0035] The above embodiments are only used to illustrate the technical methods of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of this application without departing from the spirit and scope of the technical methods of this application.

Claims

1. A method for monitoring the energy consumption of industrial equipment based on the Internet of Things, characterized in that, include: Acquire energy consumption monitoring data from several monitoring devices on various industrial equipment; Energy consumption monitoring data from various monitoring devices are time-series calibrated to obtain an energy consumption monitoring dataset; the energy consumption monitoring dataset includes calibrated change curves corresponding to several monitoring devices and several actual measurement times; Data is extracted from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; the energy consumption data to be analyzed includes parameter values ​​at several times corresponding to several energy consumption parameters; The parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed are input into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results.

2. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 1, characterized in that, The energy consumption monitoring dataset is obtained by performing time-series calibration on the energy consumption monitoring data of each monitoring device, including: Acquire several energy consumption monitoring data belonging to the same industrial equipment; acquire several hard relationships corresponding to the industrial equipment, and several energy consumption parameters corresponding to the hard relationships; Energy consumption monitoring data corresponding to several energy consumption parameters belonging to the same hard relation are divided into the same monitoring data group; based on each hard relation, the energy consumption monitoring data in its corresponding monitoring data group are synchronized in time to obtain the energy consumption monitoring dataset corresponding to the industrial equipment.

3. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 2, characterized in that, Based on the time-series synchronization of energy consumption monitoring data in the corresponding monitoring data groups according to each hard relation, the energy consumption monitoring dataset corresponding to the industrial equipment is obtained, including: S1: Obtain each energy consumption monitoring data in the energy consumption monitoring data group, and extract the monitoring parameter values ​​collected at each time in the energy consumption monitoring data; fit the several monitoring parameter values ​​into a time-series change curve according to the chronological order of their corresponding times; S2: Obtain the relationship level corresponding to each hard relationship, and sort the hard relationships from high to low in the hard relationship sorting table; S3: Record the energy consumption monitoring data group corresponding to the hard relationship with the highest relationship level in the hard relationship ranking table as the data group to be calibrated; S4: Obtain several time-series variation curves in the data set to be calibrated, as well as the hard relation corresponding to the data set to be calibrated; S5: Substitute each time series variation curve and hard relation into the set time series calibration function to obtain the time series deviation value corresponding to each time series variation curve; S6: Remove the hard relation from the hard relation sorting table, and offset each time series change curve in the energy consumption monitoring data group on the time axis, with the offset amount being the time series deviation value corresponding to the time series change curve; record the offset time series change curve as the calibrated change curve, and record its corresponding time series deviation value as the calibrated time difference; S7: Determine if there are still hard relationships in the hard relationship sorting table. If yes, proceed to S3; otherwise, generate an energy consumption monitoring dataset based on each calibrated change curve and calibrated time difference.

4. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 3, characterized in that, One expression of the timing calibration function includes: ； in, This represents the time series deviation value corresponding to the energy consumption monitoring data numbered n in the energy consumption monitoring data group; This is the hard relation function corresponding to the energy consumption monitoring data set; The time-series variation curve corresponding to the energy consumption monitoring data numbered n in the energy consumption monitoring data group; Shift the time-series change curve corresponding to the energy consumption monitoring data with number n in the energy consumption monitoring data group. The subsequent curve; The deviation of the time-series variation curve under the hard relationship; , The set offset period; For any calibrated change curve in the energy consumption monitoring data set, The calibrated time difference corresponding to the change curve. ; The total number of other energy consumption monitoring data remaining in the energy consumption monitoring data group after removing the energy consumption monitoring data corresponding to the calibrated change curve, and n is the number of the other energy consumption monitoring data remaining in the energy consumption monitoring data group after removing the energy consumption monitoring data corresponding to the calibrated change curve.

5. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 3, characterized in that, The energy consumption monitoring dataset generated based on each calibrated change curve and calibrated time difference includes: Obtain each calibrated change curve and its corresponding calibrated time difference, as well as the time when each monitoring parameter value in the energy consumption monitoring data corresponding to the calibrated change curve was collected; The time corresponding to each monitoring parameter value is offset on the time axis, and the offset amount is the calibrated time difference corresponding to the calibrated change curve; several offset times are recorded as the measured times corresponding to the calibrated change curve. Each calibrated change curve and its corresponding measured time points are integrated into an energy consumption monitoring dataset.

6. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 1, characterized in that, The process involves extracting data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed, including: Acquire each calibrated change curve and its corresponding measured time; and the acquisition interval corresponding to the measured time; substitute the acquisition interval into the set neighborhood step size generation function to obtain the corresponding neighborhood step size; A stable window is constructed with the measured time corresponding to the calibrated change curve as the center and the neighborhood step size as the radius; several stable windows corresponding to the calibrated change curve are constructed in sequence, and a set of stable windows corresponding to the calibrated change curve is constructed. Obtain the set of stable windows corresponding to each calibrated change curve, and substitute each set of stable windows into the analysis point window selection function to obtain the corresponding selection window set; one representation of the analysis point window selection function includes: ； in, To select a set of windows; The set of stable windows for the calibrated numbered curves corresponding to number n; Select several time points from the selected window set as analysis time points, obtain the parameter values ​​corresponding to each calibrated change curve at the analysis time point, and integrate each energy consumption parameter, as well as several parameter values ​​corresponding to the energy consumption parameter and their corresponding analysis time points into the energy consumption data to be analyzed.

7. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 6, characterized in that, One expression of the neighborhood step size generating function includes: ； in, The neighborhood step size; The sampling interval; This is the set reference acquisition interval; The set scaling factor.

8. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 6, characterized in that, The step of selecting several times from the selection window set as analysis times includes: Extract each selection window from the selection window set and obtain the set analysis step size; set a number of time intervals in the selection window with the analysis step size as the interval, and record them as analysis time; set the analysis time corresponding to each selection window in sequence.

9. The method for monitoring the energy consumption of industrial equipment based on the Internet of Things according to claim 1, characterized in that, One training method for the energy consumption analysis model includes: Acquire historical energy consumption monitoring data of industrial equipment, generate corresponding energy consumption data to be analyzed based on the energy consumption monitoring data, extract the parameter values ​​of each energy consumption parameter at each time in the energy consumption data to be analyzed, obtain the energy consumption status at each time, and integrate the parameter values ​​and energy consumption status of each energy consumption parameter at several times into several training data and test data. The artificial intelligence model is trained using training data and tested using validation data. The model is then used to obtain an artificial intelligence model that takes the parameter values ​​of various energy consumption parameters at several times as input and outputs the energy consumption state corresponding to the parameter values ​​at each time. The energy consumption states at each time are integrated into the energy consumption monitoring results, and the energy consumption monitoring results are used as the final output to obtain the energy consumption analysis model.

10. An industrial equipment energy consumption monitoring system based on the Internet of Things (IoT), comprising an application of the Internet-based industrial equipment energy consumption monitoring method according to any one of claims 1 to 9, characterized in that, include: Data acquisition module, data calibration module, and energy consumption analysis module. The data acquisition module is used to acquire energy consumption monitoring data from several monitoring devices on various industrial equipment. The data calibration module includes a timing calibration unit and a data extraction unit; The timing calibration unit is used to perform timing calibration on the energy consumption monitoring data of each monitoring device to obtain an energy consumption monitoring dataset; the energy consumption monitoring dataset includes calibrated change curves corresponding to several monitoring devices and several actual measurement times. The data extraction unit is used to extract data from each calibrated change curve in the energy consumption monitoring dataset to obtain the energy consumption data to be analyzed; the energy consumption data to be analyzed includes parameter values ​​at several times corresponding to several energy consumption parameters. The energy consumption analysis module is used to input the parameter values ​​of several energy consumption parameters corresponding to each time point in the energy consumption data to be analyzed into the energy consumption analysis model to obtain the corresponding energy consumption monitoring results.

Images