Energy consumption monitoring method and system of enterprise management platform fused with AI algorithm

Abstract

The application discloses an energy consumption monitoring method and system of an enterprise energy management platform fused with an AI algorithm, relates to the technical field of energy management, and comprises the following steps: establishing a unified time reference of an energy consumption monitoring process, time-synchronizing start and stop moments of each device in an enterprise, obtaining current change data corresponding to starting and stopping of each device, generating continuous time sequence curves, and drawing a current overlap area distribution map on a unified time scale to determine start and stop positions of current overlap. Through establishment of the unified time reference and the current overlap area distribution map, the application realizes time sequence correction and feature recognition of a multi-device starting and stopping process, ensures time consistency and signal independence of energy consumption data collection. Through adjustment of a disturbance area record table, the application executes reverse silence, delay and alternate sampling strategies, realizes active separation and error correction of current signals, and improves the precision and stability of energy consumption monitoring.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, specifically to an energy consumption monitoring method and system for an enterprise energy management platform that integrates AI algorithms. Background Technology

[0002] Enterprise energy management platforms integrating AI algorithms for energy consumption monitoring refer to embedding artificial intelligence and big data processing technologies into the entire process of energy consumption data collection, analysis, and decision-making within an enterprise's energy management system. This enables intelligent identification and dynamic optimization of energy use. The method accesses multi-source energy consumption data (electricity, water, gas, heat, etc.) within the enterprise, utilizing machine learning, deep learning, and pattern recognition algorithms, combined with big data processing, to clean, aggregate, and extract features from massive amounts of heterogeneous data. It then models and predicts energy consumption behavior under different production units, equipment, and operating conditions, thereby identifying abnormal energy consumption, uncovering energy-saving potential, and providing managers with real-time decision-making support. Unlike traditional energy consumption monitoring, the combination of AI algorithms and big data processing continuously optimizes model parameters through continuous learning and data accumulation, giving the energy management platform adaptive, predictive, and intelligent characteristics, ultimately achieving refined management and dynamic optimization of enterprise energy utilization.

[0003] The existing technology has the following shortcomings: In existing energy consumption monitoring technologies, when adjacent devices operate under high-frequency start-stop conditions, their instantaneous current waveforms tend to overlap, making it impossible to clearly distinguish the energy consumption signals obtained at the acquisition end in terms of time. Since existing AI recognition algorithms typically rely on the amplitude, frequency, and morphological characteristics of current waveforms to determine energy consumption attribution, when waveforms overlap too densely, the algorithm can easily misclassify the energy consumption of multiple devices as input from the same energy source, leading to confusion in the identification of individual energy consumption items. Such errors not only distort energy consumption data statistics but also cause deviations in energy allocation and energy-saving assessments, further affecting the accuracy of subsequent load optimization and energy scheduling in the energy management system. This problem is particularly amplified in dynamic scenarios where multiple devices operate collaboratively, making it difficult to detect and correct in a timely manner.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an energy consumption monitoring method and system for an enterprise energy management platform that integrates AI algorithms, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms, comprising the following steps: Establish a unified time benchmark for the energy consumption monitoring process, synchronize the start and stop times of various equipment in the enterprise, obtain the current change data corresponding to the start and stop of each equipment, generate continuous time series curves, and draw a distribution map of the current overlap area on a unified time scale to determine the start and end positions of the current overlap. Based on the current overlap region distribution map, the current time series curves of each time period are segmented, the boundary changes of the current waveform are analyzed, the transition characteristics of the current change process of each device are extracted, and the feature index table corresponding to each device is generated based on the transition characteristics. The feature index table is used to backtrack and compare the energy consumption sub-items, detect the misjudgment of energy consumption caused by the overlap of current waveforms, determine the time period of the misjudgment and the combination of equipment that caused the interference, and generate an interference area record table to store the interference information. Based on the interference zone record table, the energy consumption sampling rhythm and sampling order are adjusted, and cross suppression windows and sampling interval limits are set in the time dimension to form an energy consumption signal separation scheme for interference time periods, so as to reduce the phenomenon of current waveform overlap. The energy consumption sampling process is adaptively adjusted according to the energy consumption signal separation scheme. During the interference period, reverse sampling silence, short-delay sampling and alternating sampling strategies are implemented. Overlapping signals are separated by adjusting the sampling rhythm and feedback rhythm. Based on the separation results, the misjudgment of energy consumption attribution is corrected, and the corrected output of energy consumption monitoring data is completed.

[0007] Preferably, the steps for generating the current overlap region distribution map are as follows: In the monitoring platform, a reference device with stable working status is selected as the time reference source. The time signal is transmitted to the acquisition end of all monitored devices through a high-precision clock synchronization mechanism, so that the timestamps of each device are consistent at the start of data acquisition. The current changes of all devices during startup and shutdown are continuously recorded, and the collected current change data are arranged in correspondence with the established time base to generate a continuous time series curve while keeping the sampling interval constant. Using the current mutation points at the moment of start-up and shutdown of each device as time synchronization anchor points, the start-up and shutdown boundaries are marked and dynamic synchronization is completed under a unified time reference; The current change curves of each device are superimposed on a unified time scale to form a time-series overlay diagram. The start and end positions of the current overlap area are determined based on the curve change pattern, and a distribution map of the current overlap area is drawn.

[0008] Preferably, the steps for segmenting the current time series curve based on the current overlap region distribution map are as follows: Based on the current overlap region distribution map, the time periods in the overlapping region of the continuous current time series curve are calibrated, and the curve is divided into multiple time segments according to the continuity characteristics of current change. After segmentation, boundary change analysis is performed on the current waveform of each time segment to identify the transition law of the current waveform in the rising and falling regions and extract the boundary change characteristics. After obtaining the boundary change information, the transition feature points in the current change process of each device are extracted, and independent change trajectories are formed based on the time and amplitude of the transition features. Based on the extracted transition feature information, a feature index table corresponding to each device is generated, and the feature index table is matched with the time period in the current overlap area distribution map for subsequent energy consumption identification and analysis.

[0009] Preferably, when performing boundary change analysis on the current waveform, the current change curve of each time segment is locally smoothed to reduce interference caused by instantaneous fluctuations, and the starting point and ending point of the current transition feature are determined based on the time distribution of the rising edge, peak point and falling edge of the current curve, thereby improving the continuity of boundary division and the accuracy of feature extraction.

[0010] Preferably, the steps for backtracking and comparing energy consumption data using a feature index table are as follows: The transition characteristics of each device in the feature index table are arranged in chronological order and mapped one by one with the energy consumption sub-items, so that the energy consumption data is associated with the current characteristic change points in the time dimension. After the mapping is completed, the current waveform data and energy consumption sub-item results are compared retrospectively. The superposition phenomenon of current waveforms is identified by comparing and analyzing the energy consumption curves of different devices in the same time period. After obtaining the comparison results, analyze the abnormal time period, determine the time range of the misjudgment and identify the combination of interfering equipment, and form a list of interference events based on the equipment operating status and the direction of current change. The information on the combination of interfering devices and the time period of misjudgment is compiled into an interference zone record table, archived in chronological order and associated with the device feature number in the feature index table for subsequent energy consumption sampling and adjustment.

[0011] Preferably, during the generation of the interference zone record table, the start time, end time, combination of interfering devices, and energy consumption deviation value of each interference event are archived using time as the main index. The interference intensity level identifier between devices is set in the record table, so that the interference information is bidirectionally associated with the device feature number in the feature index table, so as to achieve accurate tracing of the source of energy consumption misjudgment and time-series management of interference data.

[0012] Preferably, the steps for adjusting the energy consumption sampling rhythm and sampling order based on the interference zone record table are as follows: Based on the interference zone record table, the start and end times of the interference time period and the combination of interfering equipment are extracted, and the time distribution table is rearranged in chronological order. The monitoring period is then re-divided to distinguish between the interference time period and the normal time period. After the division is completed, a cross-suppression window is set according to the degree of overlap and duration of the interfering devices. The window start point is set at the start position of the interference and the window end point is set at the end position of the interference. After establishing the cross-suppression window, the sampling interval is reset, and the interval is adjusted in stages according to the duration of the interference, and time shift compensation is performed to maintain data continuity. The sampling rhythm, sampling interval, and device sampling sequence are combined to generate an energy consumption signal separation scheme, ensuring that the sampling tasks of different devices do not overlap in time.

[0013] Preferably, in the process of forming the energy consumption signal separation scheme, the sampling order of the devices within the interference period is periodically alternated, so that the same group of devices takes turns to obtain sampling priority in adjacent interference periods. The sampling resource distribution is balanced through the time alternation mechanism, and a transition buffer is set during the alternation to maintain the continuous execution of the sampling task, thereby further reducing the phenomenon of current waveform overlap.

[0014] Preferably, the adaptive adjustment steps for the energy consumption sampling process based on the energy consumption signal separation scheme are as follows: Based on the interference time period distribution and sampling rhythm configuration recorded in the energy consumption signal separation scheme, the sampling tasks of each device are grouped and divided, and reverse sampling silence is performed during the interference time period to eliminate the initial overlapping signal. After the reverse sampling is silenced, a short-delay sampling is performed. By adjusting the sampling start time, a time-displaced sampling sequence is formed to achieve layered signal acquisition. After completing the delayed sampling, an alternating sampling strategy is executed to make the sampling order of devices within the interference group alternate periodically and to set a transition buffer at the switching boundary to maintain time continuity. The sampled data is reconstructed in time sequence and energy consumption misjudgment is corrected. The reconstructed energy consumption data is then redistributed to the devices that actually generate current changes and the corrected monitoring results are output.

[0015] The enterprise energy management platform energy consumption monitoring system integrating AI algorithms includes a time calibration module, a feature index analysis module, an interference identification and processing module, a sampling rhythm control module, and a signal separation and correction module. The time calibration module establishes a unified time reference for the energy consumption monitoring process, synchronizes the start and stop times of various devices in the enterprise, acquires the current change data corresponding to the start and stop of each device, generates continuous time-series curves, and draws a current overlap area distribution map on a unified time scale to determine the start and end positions of the current overlap. The feature index analysis module, based on the current overlap region distribution map, segments the current time series curves of each time period, analyzes the boundary changes of the current waveform, extracts the transition features in the current change process of each device, and generates the feature index table corresponding to each device based on the transition features. The interference identification and processing module uses a feature index table to backtrack and compare the energy consumption sub-items, detects misjudgments of energy consumption caused by overlapping current waveforms, determines the time period of the misjudgment and the combination of equipment that caused the interference, and generates an interference zone record table to store the interference information. The sampling rhythm control module adjusts the energy consumption sampling rhythm and sampling order according to the interference zone record table, and sets the cross suppression window and sampling interval limit in the time dimension to form an energy consumption signal separation scheme for the interference period to reduce the phenomenon of current waveform overlap. The signal separation and correction module adaptively adjusts the energy consumption sampling process according to the energy consumption signal separation scheme. During the interference period, it executes reverse sampling silence, short-delay sampling and alternating sampling strategies. By adjusting the sampling rhythm and feedback rhythm, it separates overlapping signals and corrects the misjudgment of energy consumption attribution based on the separation results, thus completing the correction output of energy consumption monitoring data.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention establishes a unified time benchmark throughout the energy consumption monitoring process and combines it with a current overlap region distribution map to achieve time-series correction of the start-up and shutdown behavior of multiple devices, ensuring consistency and continuity of energy consumption data acquisition in the time dimension. By segmentally analyzing the changes in current curve boundaries and extracting transition features, it can accurately identify the energy consumption change trajectory corresponding to the operating state of each device, thereby effectively distinguishing current aliasing caused by simultaneous start-up and shutdown of devices. This method achieves precise time synchronization of energy consumption acquisition, ensuring that the current signals of different devices have clear time boundaries at the data level, avoiding data misjudgment caused by mixed energy consumption signals, and providing an accurate foundation for subsequent energy consumption identification and energy allocation.

[0017] This invention utilizes an interference zone record table to adjust the sampling rhythm and sequence, implementing reverse sampling silence, short-delay sampling, and alternating sampling strategies during interference periods to achieve active separation of the current signal in terms of time distribution. This method reduces the probability of current waveform overlap at the sampling level, enabling the energy consumption monitoring process to dynamically adapt to changes in the operating status of multiple devices, achieving real-time separation and misjudgment correction of overlapping signals. Through coordinated control of the sampling rhythm and feedback rhythm, the energy consumption monitoring data output is more stable and accurate, thereby improving the overall monitoring accuracy and dynamic response capability of the energy management platform. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of the energy consumption monitoring method for an enterprise energy management platform that integrates AI algorithms, as described in this invention.

[0020] Figure 2 This is a schematic diagram of the module of the enterprise energy management platform energy consumption monitoring system that integrates AI algorithms according to the present invention. Detailed Implementation

[0021] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0022] This invention provides, for example Figure 1 The energy consumption monitoring method for an enterprise energy management platform that integrates AI algorithms, as shown, includes the following steps: Establish a unified time benchmark for the energy consumption monitoring process, synchronize the start and stop times of various equipment in the enterprise, obtain the current change data corresponding to the start and stop of each equipment, generate continuous time series curves, and draw a distribution map of the current overlap area on a unified time scale to determine the start and end positions of the current overlap. In order to achieve precise management of the entire process of enterprise energy consumption monitoring during implementation, the first step is to coordinate the operating time of all energy-using equipment within the enterprise and establish a complete time benchmark framework. The specific implementation steps are as follows: A stable reference device is selected as the time base source in the monitoring platform. A high-precision clock synchronization mechanism transmits the time signal from the reference device to the acquisition terminals of all monitored devices, ensuring that the timestamps of each device are completely consistent at the start of data acquisition. Secondly, the current changes of each device during startup and shutdown are continuously recorded. The collected current change data is mapped one-to-one with the established time base, ensuring that each current curve is uniformly arranged on the same time dimension, thus forming a complete set of continuous time-series curve datasets. During this process, to avoid uneven sampling intervals caused by differences in device start-up and shutdown frequencies, the time label of each sampling point is recalibrated to ensure that the distribution of data points on the time axis maintains a constant interval, ensuring consistency in subsequent time series processing.

[0023] After establishing a unified time base and establishing time correspondence with current data, the start-up and shutdown times of each device are further dynamically synchronized. In this step, the current abrupt change point at the moment of start-up and shutdown for each device is used as the time synchronization anchor point. By calculating the fluctuation relationship of current change trends between adjacent devices, the start-up and shutdown phase positions of each device on the time axis are confirmed, and the start-up and shutdown boundary points of each device are marked on the time base. This ensures that the start-up and shutdown events of all devices are accurately recorded under a unified time base, thus forming a comparable operating sequence. To ensure a closer correspondence between start-up and shutdown times and current curves, the current curves at the moment of start-up and shutdown are locally magnified for analysis to extract the changing segments of the rising and falling edges of the current, making the division of start-up and shutdown boundaries clearer. This processing provides a clear time marker for the operating states of different devices under a unified time base, providing an accurate timing basis for the subsequent identification of current overlap areas.

[0024] After establishing time synchronization and start / stop boundary markings, the current data of each device at a unified time scale are continuously mapped, and the current change curves of all devices are superimposed onto the same coordinate system in chronological order to form a complete time-series overlay diagram. To accurately reflect the current change relationships between devices, the superimposed current curves are scanned laterally with equal time steps, recording the numerical differences in current changes of each device within each time interval. When the current curves of two or more devices intersect, overlap, or approach each other within the same time interval, that time interval is identified as a current overlap zone. To ensure more precise boundary delineation of the current overlap zone, the waveform fluctuations are smoothed, making the edge regions of current changes more continuous in time. Through this mapping and scanning process, the current distribution relationships of all devices in the time dimension can be visualized, forming an overall pattern of energy consumption changes among devices in the same image.

[0025] After obtaining the current overlay map, the boundaries of the current overlap region are delineated based on the changing patterns of the overlay curves, thereby determining the start and end times of each current overlap region. In this stage, the start and end points of current overlap are determined by comparing the current change gradients of adjacent time periods in the continuous time-series curves. When the current waveforms of two devices simultaneously show an upward trend over time and persist for a certain period, this time period is identified as the start point of overlap; when one of the two waveforms returns to a stable state while the other current curve continues to change independently, this time period is identified as the end point of overlap. In this way, the distribution of the current overlap region for each device under a unified time base can be obtained, and the corresponding overlap time periods for each device are marked on the plotted current overlap region distribution map. This distribution map not only reflects the energy consumption cross-relationships of different devices on the time axis but also provides a reliable data foundation for subsequent current curve decomposition, feature extraction, and energy consumption misjudgment correction. Through precise delineation of the overlap region, the dynamic change trajectory of the current of each device on the same time scale can be clearly displayed, thus laying a complete data structure and time framework for subsequent energy consumption signal separation and optimization processing.

[0026] Based on the current overlap region distribution map, the current time series curves of each time period are segmented, the boundary changes of the current waveform are analyzed, the transition characteristics of the current change process of each device are extracted, and the feature index table corresponding to each device is generated based on the transition characteristics. To achieve in-depth analysis of the current variation process of each device, the current overlap region distribution map is used as the basic data source. The continuous current time series curves obtained under a unified time base are initially segmented according to the current variation trend. The specific implementation steps are as follows: The overlapping time periods in the current time series curves are calibrated, and the curves are divided into multiple time segments based on the continuity of current changes in each segment. Each segment corresponds to a relatively stable or significantly changing current process. During segmentation, the continuity and gradient of current changes are emphasized, separating segments with large fluctuations from stable segments to ensure that the current changes within each time segment have similar characteristic trends. To avoid segmentation errors caused by differences in current waveform characteristics between different devices, the time axis is refined during segmentation, ensuring that the start and end points of each segment are consistent with the start and end positions of the overlapping area in the current overlap distribution diagram. This guarantees that the current segmentation results for each device have uniform boundaries in the time dimension.

[0027] After segmenting the current curve, boundary change analysis is performed on the current waveform of each segment. In this stage, the segmented current curve is used as input, and edge scanning is performed on the beginning and end of each time segment to extract transition features of current changes, identifying the process of the current waveform transitioning from one stable state to another. This process focuses on analyzing boundary feature parameters such as the rate of change, peak position, and duration of change of the current curve in the rising and falling regions. Combined with the overlapping region time position determined in the previous step, the transition law of the current waveform within and outside the overlapping region is identified. This boundary analysis method can determine the transition process caused by equipment start-up and shutdown in the current waveform, as well as the waveform superposition relationship between different devices in the same time period. To make the boundary analysis results more accurate, the current change curve of each time segment is locally smoothed to reduce interference caused by instantaneous fluctuations and make the boundary changes more continuous. Through precise description of the boundaries of each segment, the start and end points of current changes for each device in the overlapping region can be obtained, thus providing clear time-domain localization for subsequent transition feature extraction.

[0028] After obtaining the boundary change information of the current waveform, transition features are extracted from the current change process of each device. This process, centered on the results of boundary change analysis, extracts transition feature points reflecting changes in the device's operating state during the current change process for each segmented interval. These transition feature points typically include the starting point of the rising segment, the peak point, the ending point of the falling segment, and the point of abrupt change in the rate of change of the current curve. By analyzing the distribution relationship of these feature points on the time axis, the dynamic process of the device from standstill to operation and from operation to shutdown can be reflected. Each device's transition feature has independent time and current amplitude characteristics, forming an independent change trajectory under a unified time reference. To ensure the completeness of the transition feature extraction, the current curve of each device is scanned throughout the entire time series, ensuring that all transition features related to the device's start-up and shutdown processes are fully captured. Simultaneously, the transition features of multiple devices within overlapping areas are distinguished, and based on the differences in the direction of current change and the peak position, the feature changes of different devices within the same time period are independently identified. In this way, feature points with clear physical meaning can be extracted from time-continuous current data, thus providing clear change indicators for the energy consumption behavior of different devices.

[0029] After extracting the transition features, a feature index table is generated for each device based on the identified feature information. This index table uses time series as its main thread, archiving and organizing the transition features of each device according to time sequence and feature category, forming a feature data structure that can be used for subsequent analysis. During the construction of the feature index table, the transition feature point information of each device is correlated with the time period in the current overlap area distribution map to accurately reflect the current change trajectory of the device within the overlap area. Each record in the feature index table includes the device identifier, feature time, feature type, and current change amplitude, thus establishing a multi-dimensional correspondence between time, device, and feature. Based on this, the current change features of the corresponding device can be quickly located within any time period, enabling comparison of current changes and identification of energy consumption behavior among different devices in energy consumption monitoring. The feature index table not only records feature data but also provides basic data support for subsequent misjudgment identification and signal separation. This structured feature organization method ensures that the current transition information of all devices is stored and associated in an orderly manner under a unified time benchmark, making the feature analysis of the entire energy consumption monitoring process continuous and traceable.

[0030] The feature index table is used to backtrack and compare the energy consumption sub-items, detect the misjudgment of energy consumption caused by the overlap of current waveforms, determine the time period of the misjudgment and the combination of equipment that caused the interference, and generate an interference area record table to store the interference information. To achieve accurate backtracking and comparison of energy consumption data, the established feature index table is used as the core reference information to match the energy consumption data collected by enterprises under a unified time benchmark on a time-period basis. The specific steps are as follows: The transition characteristics of each device in the feature index table are arranged chronologically and mapped one-to-one with the corresponding energy consumption data, thus establishing a correlation between energy consumption data and changes in current characteristics over time. This temporal correspondence clarifies the operational status changes of each device in different time periods and their corresponding energy consumption fluctuation trends. In this process, energy consumption data for each time period are continuously extracted, and based on the characteristic times recorded in the feature index table, the energy consumption change data of different devices are uniformly superimposed on the same time axis, forming a comparative curve of energy consumption changes across multiple devices. This curve visually reflects the synchronicity and differences in energy consumption among the devices during operation, providing an accurate data foundation for subsequent false positive detection.

[0031] After mapping the energy consumption data to the feature index table, a retrospective comparison is performed on the current waveform data of each device and the energy consumption results. The key to this stage is identifying potential overlapping waveforms in the energy consumption data based on the transition features identified in the feature index table. By comparing and analyzing the energy consumption curves of different devices within the same time period, if two devices exhibit overlapping transition features in the feature index table, and there are concentrated peaks of abnormal energy input in the energy consumption data, it can be inferred that current waveform superposition exists within that time period. To ensure the continuity of the comparison process, the comparison results for each time period are stored sequentially to form a complete time-chain comparison record. This not only allows observation of the energy consumption changes of a single device at different times but also tracks the interaction between devices in terms of energy input. In this way, the impact of current waveform superposition on the accuracy of energy consumption data can be traced in continuous time-series data, and the identification conditions for misjudgment features can be gradually established.

[0032] After obtaining the energy consumption data backtracking comparison results, the abnormal time periods found in the comparison are analyzed to determine the time range of misjudgments and possible combinations of interfering devices. During this process, the device transition characteristics corresponding to the feature index table are matched again with the abnormal time periods to determine if multiple devices exhibit similar transition characteristics within the same time period. If the transition characteristic times of two or more devices overlap in adjacent time periods, and the corresponding energy consumption data shows energy input exceeding the normal power range of a single device, then this time period can be identified as a high-risk area for energy consumption identification misjudgment. Furthermore, based on the operating status and current change direction of the devices in these overlapping time periods, the device combination relationship causing the misjudgment is confirmed. In this way, the specific set of devices involved in each misjudged time period can be identified, and the degree of interference generated by these devices in energy consumption acquisition can be determined. To make the results more systematic, each group of interfering device combinations and their corresponding misjudged time periods are organized into a series of interference event entries in chronological order, forming a preliminary list of interference events.

[0033] After obtaining the combinations of interfering devices and their corresponding misjudgment periods, this information is compiled into an interference zone record table for storage and management. This record table uses time as the primary index, archiving the start and end times of each interference event, the combination of interfering devices, and related energy consumption deviation values. Each interference record is bound to the corresponding device feature number in the feature index table, thus achieving a two-way association between interference information and current characteristic information. To ensure the traceability of interference information in subsequent processing, the records in the interference zone record table are arranged sequentially by occurrence time, and a level indicator of interference intensity between devices is added to distinguish the degree of energy consumption impact of different interference combinations. This interference zone record table can comprehensively reflect the identification misjudgments caused by overlapping current waveforms during energy consumption monitoring, providing clear time and device basis for subsequent sampling rhythm adjustments and signal separation. Simultaneously, by associating the interference zone record table with the feature index table, the source of misjudgment can be traced back and located; that is, the device combination causing the interference and its operating characteristics can be quickly found within any time period, thus forming a complete interference tracing chain in the energy consumption monitoring process.

[0034] Based on the interference zone record table, the energy consumption sampling rhythm and sampling order are adjusted, and cross suppression windows and sampling interval limits are set in the time dimension to form an energy consumption signal separation scheme for interference time periods, so as to reduce the phenomenon of current waveform overlap. To effectively reduce current waveform overlap during energy consumption monitoring and thus improve the separation accuracy of energy consumption data, the specific steps are as follows: Based on the interference zone record table, the overall scheduling of energy consumption sampling was adjusted. First, the start and end times of each interference period and the combination of interfering devices were extracted from the interference zone record table, and these interference events were rearranged in chronological order. This created a master table of time distribution for all interference periods, providing a time reference for adjusting the sampling rhythm. On this basis, the monitoring cycle was re-divided, distinguishing between interference periods and normal sampling periods, and the positions of the interference intervals were marked on the time axis. Subsequently, based on the density and duration of the interference intervals, the priority order of sampling times was determined, prioritizing sampling tasks during non-interference periods, while the sampling rhythm within the interference intervals was delayed or intermittent. This method initially achieved a staggered distribution of sampling tasks in the time dimension, reducing the possibility of current waveform superposition caused by simultaneous sampling by multiple devices, and providing a time framework for subsequent cross-suppression window settings.

[0035] After initial adjustment of the sampling rhythm, the information on interfering device combinations in the interference zone record table is further analyzed. Cross-suppression windows are set based on the degree of overlap and duration of each device within the interference time period. Setting the cross-suppression window is the core step in the entire energy consumption sampling adjustment process. Its function is to create a mutually exclusive sampling relationship between devices on the time axis, thereby preventing multiple devices from collecting energy consumption data at the same time. Specifically, the operating time periods of each group of interfering devices are unfolded chronologically, the start and end points of current changes for each device are determined, and the duration of their time overlap is calculated. The start point of the cross-suppression window is set at the beginning of the interference interval, and the end point is set at the end of the interference, ensuring that the sampling tasks of two or more devices are no longer triggered simultaneously within that time period. To ensure the continuity of the suppression window, a short buffer is set between windows to prevent acquisition errors caused by sudden current changes during sampling switching. This time limit for cross-suppression ensures that the current signals of each device will not be sampled superimposed within the same time period, thereby reducing waveform overlap. The duration of the cross-suppression window can be dynamically configured according to the duration of different interference combinations in the interference zone record table, ensuring that its timing is consistent with the device operating characteristics.

[0036] Based on the established cross-suppression window, the sampling interval is reset to create a segmented distribution structure for the sampling tasks over time. During this process, the sampling interval is adjusted in stages according to the duration of interference in the interference zone record table. For equipment combinations with short interference periods, a micro-interval sampling method is used, increasing the time interval between adjacent sampling points to reduce time overlap. For equipment combinations with long interference periods, an intermittent sampling method is used, pausing sampling of a particular device during the interference period and resuming normal sampling after the interference zone ends. The redistribution of sampling intervals considers not only the time sequence but also the start and end times of the cross-suppression window, ensuring precise matching between the sampling interval adjustment and the time constraints of the suppression window. To avoid introducing data gaps during the sampling interval adjustment, time shift compensation is applied to the sampling points, moving some sampling tasks to a buffer zone outside the interference zone to ensure continuous sampling data throughout the monitoring cycle. This staged interval and time compensation method allows for staggered execution of energy consumption data collection in the time distribution, effectively reducing the overlapping areas of current waveforms while ensuring data integrity.

[0037] After comprehensively adjusting the sampling rhythm and sampling interval, an energy consumption signal separation scheme is generated for the entire monitoring process to guide the execution of subsequent sampling tasks. Based on the interference zone record table, the energy consumption signal separation scheme integrates the cross-suppression window, sampling interval distribution, and device sampling order to form a time-layered sampling schedule. This schedule uniformly plans the sampling start time, sampling duration, and sampling interval for each device in each time period, ensuring that sampling tasks of different devices do not overlap within the same time range. Simultaneously, to further reduce current waveform overlap, a time alternation mechanism is introduced into the sampling task arrangement. This involves periodically alternating the device sampling order within the interference zone, allowing the same group of devices to take turns obtaining priority sampling rights in adjacent interference cycles, thereby balancing the distribution of sampling resources. During the execution of the signal separation scheme, the sampling schedule table serves as the basis for time control of the monitoring tasks. By continuously executing sampling strategies for different time periods, precise separation of energy consumption data in the time dimension is achieved. In this way, active separation of energy consumption signals at the time sampling level can be achieved in complex energy consumption environments where multiple devices are running in parallel, effectively reducing the overlap rate of current waveforms and making the collected energy consumption data more independent, clear and analyzable.

[0038] The energy consumption sampling process is adaptively adjusted according to the energy consumption signal separation scheme. During the interference period, reverse sampling silence, short-delay sampling and alternating sampling strategies are implemented. Overlapping signals are separated by adjusting the sampling rhythm and feedback rhythm. Based on the separation results, the misjudgment of energy consumption attribution is corrected, and the corrected output of energy consumption monitoring data is completed. To achieve accurate separation of current signals and dynamic correction of energy consumption data during interference periods in energy consumption monitoring, the time control parameters of the entire energy consumption sampling process are comprehensively adjusted based on the generated energy consumption signal separation scheme. The steps are as follows: Based on the interference time period distribution and sampling rhythm configuration recorded in the energy consumption signal separation scheme, the sampling tasks of each device under a unified time reference are grouped and divided into interference groups and non-interference groups according to the degree of interference correlation. For devices in the interference group, a reverse sampling silence mechanism is introduced during the sampling execution process of the interference time period. Specifically, at the beginning of the interference time period, the system first suspends the energy consumption sampling task of the corresponding interference device, keeping its sampling channel idle for a silence interval, thereby preventing the device from acquiring the mixed signal of interference in the initial stage of current waveform overlap. When the silence interval ends, the sampling task restarts. At this time, the overlapping part of the interference waveform has been separated in the time domain, and the acquired current data will be purer. Through this reverse silence method, the sampling synchronization point can be artificially cut off during the interference time period, thereby achieving the initial separation of overlapping signals.

[0039] After completing the time division for reverse sampling silence, a short-delay sampling strategy is implemented for the sampling process within the interference period. The core of this step lies in adjusting the distribution of the sampling trigger signal on the time axis by delaying the sampling start time, thus forming a time-staggered sampling sequence. Specifically, for multiple devices in the same interference group, different sampling start delays are set according to the differences in the interference duration of each device in the interference zone record table, so that the sampling start times of adjacent devices are staggered by a certain fixed interval. During this process, the sampling delay time interval is matched and configured according to the duration of the interference zone and the width of the cross-suppression window to ensure that the sampling times of two devices do not overlap throughout the entire interference cycle. When the devices start sampling in the delayed order, different stages of current waveform change can be captured in time layers, thereby achieving physical separation of overlapping signals at the sampling end. The implementation of short-delay sampling not only allows the sampling tasks of multiple devices within the same interference zone to be executed independently, but also forms a periodic sampling window structure in time distribution, providing a timing basis for subsequent alternating sampling.

[0040] After adjusting the time distribution of short-delay sampling, an alternating sampling strategy is further implemented for sampling tasks within the interference period. This strategy, based on delayed sampling, periodically alternates the sampling order of devices within the interference group, allowing different devices to take turns gaining sampling priority in adjacent sampling cycles. Specifically, the devices in the interference group are comprehensively sorted according to the duration of the interference zone, sampling frequency, and the rate of change in device power consumption to determine the cyclic sequence of alternating sampling. In the first cycle, the sampling task of the device with greater interference is prioritized, while in the second cycle, other devices perform the sampling task. This cycle repeats, ensuring that each device can acquire complete current change data in multiple sampling cycles without sampling conflicts with other devices in the same time period. To ensure the temporal continuity of the alternating sampling process, a transition buffer is set at the boundary of cycle switching, allowing sampling tasks to smoothly connect during alternation and preventing sampling interruptions or time overlaps. This alternating sampling method creates a dynamic sampling rhythm within the interference period, allowing the sampling activities of different devices to interweave in time, thereby further reducing signal interference caused by current waveform overlap.

[0041] After the reverse sampling silence, short-delay sampling, and alternating sampling strategies are executed, the sampled energy consumption data undergoes time-series reconstruction and misjudgment correction. This stage primarily relies on the sampling control strategy to time-register the collected energy consumption data, rearranging the current signals acquired in different time periods into a continuous energy consumption curve. First, the current data collected during the reverse silence phase is concatenated with the data from the delayed sampling phase to fill in the time gaps within the silence interval, ensuring the overall sampling curve remains continuous. Second, the sampling results from the alternating sampling phase are reordered according to time priority, ensuring that the sampling data from each device has a complete time-series structure on a unified time axis. After time reconstruction, energy balance comparison is performed on the energy consumption data of each device to identify energy consumption misclassifications caused by interference. When energy consumption is found to be incorrectly attributed to non-interfering devices during the interference period, the energy consumption data is redistributed to the records of devices that actually generate current changes based on the time markers of the aforementioned sampling rhythm and order. Through this correction process, the misjudged portions in the energy consumption monitoring data are corrected, and the final monitoring results accurately reflect the true energy consumption distribution of each device during the interference period. The corrected energy consumption monitoring data is output in chronological order to form a corrected energy consumption data file. This data can be used not only for subsequent energy consumption analysis and energy saving assessment, but also as a reference for system operation, realizing dynamic closed-loop adjustment of energy consumption monitoring data.

[0042] This invention establishes a unified time benchmark throughout the energy consumption monitoring process and combines it with a current overlap region distribution map to achieve time-series correction of the start-up and shutdown behavior of multiple devices, ensuring consistency and continuity of energy consumption data acquisition in the time dimension. By segmentally analyzing the changes in current curve boundaries and extracting transition features, it can accurately identify the energy consumption change trajectory corresponding to the operating state of each device, thereby effectively distinguishing current aliasing caused by simultaneous start-up and shutdown of devices. This method achieves precise time synchronization of energy consumption acquisition, ensuring that the current signals of different devices have clear time boundaries at the data level, avoiding data misjudgment caused by mixed energy consumption signals, and providing an accurate foundation for subsequent energy consumption identification and energy allocation.

[0043] This invention utilizes an interference zone record table to adjust the sampling rhythm and sequence, implementing reverse sampling silence, short-delay sampling, and alternating sampling strategies during interference periods to achieve active separation of the current signal in terms of time distribution. This method reduces the probability of current waveform overlap at the sampling level, enabling the energy consumption monitoring process to dynamically adapt to changes in the operating status of multiple devices, achieving real-time separation and misjudgment correction of overlapping signals. Through coordinated control of the sampling rhythm and feedback rhythm, the energy consumption monitoring data output is more stable and accurate, thereby improving the overall monitoring accuracy and dynamic response capability of the energy management platform.

[0044] This invention provides, for example Figure 2The energy consumption monitoring system of the enterprise energy management platform shown includes an AI algorithm-integrated module, a time calibration module, a feature index analysis module, an interference identification and processing module, a sampling rhythm control module, and a signal separation and correction module. The time calibration module establishes a unified time reference for the energy consumption monitoring process, synchronizes the start and stop times of various devices in the enterprise, acquires the current change data corresponding to the start and stop of each device, generates continuous time-series curves, and draws a current overlap area distribution map on a unified time scale to determine the start and end positions of the current overlap. The feature index analysis module, based on the current overlap region distribution map, segments the current time series curves of each time period, analyzes the boundary changes of the current waveform, extracts the transition features in the current change process of each device, and generates the feature index table corresponding to each device based on the transition features. The interference identification and processing module uses a feature index table to backtrack and compare the energy consumption sub-items, detects misjudgments of energy consumption caused by overlapping current waveforms, determines the time period of the misjudgment and the combination of equipment that caused the interference, and generates an interference zone record table to store the interference information. The sampling rhythm control module adjusts the energy consumption sampling rhythm and sampling order according to the interference zone record table, and sets the cross suppression window and sampling interval limit in the time dimension to form an energy consumption signal separation scheme for the interference period to reduce the phenomenon of current waveform overlap. The signal separation and correction module adaptively adjusts the energy consumption sampling process according to the energy consumption signal separation scheme. During the interference period, it executes reverse sampling silence, short-delay sampling and alternating sampling strategies. By adjusting the sampling rhythm and feedback rhythm, it separates overlapping signals and corrects the misjudgment of energy consumption attribution based on the separation results, thus completing the correction output of energy consumption monitoring data.

[0045] The energy consumption monitoring method for an enterprise energy management platform that integrates AI algorithms provided in this embodiment of the invention is implemented through the aforementioned energy consumption monitoring system for an enterprise energy management platform that integrates AI algorithms. For details on the specific methods and processes of the energy consumption monitoring system for an enterprise energy management platform that integrates AI algorithms, please refer to the embodiment of the aforementioned energy consumption monitoring method for an enterprise energy management platform that integrates AI algorithms, which will not be repeated here.

[0046] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An energy consumption monitoring method for enterprise energy management platforms integrating AI algorithms, characterized in that, Includes the following steps: Establish a unified time benchmark for the energy consumption monitoring process, synchronize the start and stop times of each piece of equipment in the enterprise, obtain the current change data corresponding to the start and stop of each piece of equipment, generate continuous time series curves, and draw a distribution map of the current overlap area on a unified time scale to determine the start and end positions of the current overlap. Based on the current overlap region distribution map, the current time series curves of each time period are segmented, the boundary changes of the current waveform are analyzed, the transition characteristics of the current change process of each device are extracted, and the feature index table corresponding to each device is generated based on the transition characteristics. The feature index table is used to backtrack and compare the energy consumption sub-items, detect the misjudgment of energy consumption caused by the overlap of current waveforms, determine the time period of the misjudgment and the combination of equipment that caused the interference, and generate an interference area record table to store the interference information. Based on the interference zone record table, the energy consumption sampling rhythm and sampling order are adjusted, and cross-suppression windows and sampling interval limits are set in the time dimension to form an energy consumption signal separation scheme for interference time periods. The energy consumption sampling process is adaptively adjusted according to the energy consumption signal separation scheme. During the interference period, reverse sampling silence, short-delay sampling and alternating sampling strategies are implemented. Overlapping signals are separated by adjusting the sampling rhythm and feedback rhythm, and the misjudgment of energy consumption attribution is corrected based on the separation results.

2. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 1, characterized in that, The steps for generating the current overlap region distribution map are as follows: In the monitoring platform, a reference device with stable working status is selected as the time reference source. The time signal is transmitted to the acquisition end of all monitored devices through a high-precision clock synchronization mechanism, so that the timestamps of each device are consistent at the start of data acquisition. The current changes of all devices during startup and shutdown are continuously recorded, and the collected current change data are arranged in correspondence with the established time base to generate a continuous time series curve while keeping the sampling interval constant. Using the current mutation points at the moment of start-up and shutdown of each device as time synchronization anchor points, the start-up and shutdown boundaries are marked and dynamic synchronization is completed under a unified time reference; The current change curves of each device are superimposed on a unified time scale to form a time-series overlay diagram. The start and end positions of the current overlap area are determined based on the curve change pattern, and a distribution map of the current overlap area is drawn.

3. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 2, characterized in that, The steps for segmenting the current time series curve based on the current overlap region distribution map are as follows: Based on the current overlap region distribution map, the time periods in the overlapping region of the continuous current time series curve are calibrated, and the curve is divided into multiple time segments according to the continuity characteristics of current change. After segmentation, boundary change analysis is performed on the current waveform of each time segment to identify the transition law of the current waveform in the rising and falling regions and extract the boundary change characteristics. After obtaining the boundary change information, the transition feature points in the current change process of each device are extracted, and independent change trajectories are formed based on the time and amplitude of the transition features. Based on the extracted transition feature information, a feature index table is generated for each device, and the feature index table is matched with the time period in the current overlap area distribution map.

4. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 3, characterized in that, When performing boundary change analysis on the current waveform, the current change curve of each time segment is locally smoothed to reduce the interference caused by instantaneous fluctuations, and the start and end points of the current transition characteristics are determined based on the time distribution of the rising edge, peak point and falling edge of the current curve.

5. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 3, characterized in that, The steps for backtracking and comparing energy consumption data using a feature index table are as follows: The transition characteristics of each device in the feature index table are arranged in chronological order and mapped one by one with the energy consumption sub-items, so that the energy consumption data is associated with the current characteristic change points in the time dimension. After the mapping is completed, the current waveform data and energy consumption sub-item results are compared retrospectively. The superposition phenomenon of current waveforms is identified by comparing and analyzing the energy consumption curves of different devices in the same time period. After obtaining the comparison results, analyze the abnormal time period, determine the time range of the misjudgment and identify the combination of interfering equipment, and form a list of interference events based on the equipment operating status and the direction of current change. The information on the combination of interfering devices and the time period of misjudgment is compiled into an interference zone record table, which is archived in chronological order and associated with the device feature number in the feature index table.

6. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 5, characterized in that, During the generation of the interference zone record table, the start time, end time, combination of interfering equipment, and energy consumption deviation value of each interference event are archived using time as the main index. The interference intensity level identifier between equipment is set in the record table, so that the interference information is bidirectionally associated with the equipment feature number in the feature index table.

7. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 5, characterized in that, The steps to adjust the energy consumption sampling rhythm and sampling order based on the interference zone record table are as follows: Based on the interference zone record table, the start and end times of the interference time period and the combination of interfering equipment are extracted, and the time distribution table is rearranged in chronological order. The monitoring period is then re-divided to distinguish between the interference time period and the normal time period. After the division is completed, a cross-suppression window is set according to the degree of overlap and duration of the interfering devices. The window start point is set at the start position of the interference and the window end point is set at the end position of the interference. After establishing the cross-suppression window, the sampling interval is reset, and the interval is adjusted in stages according to the duration of the interference, and time shift compensation is performed to maintain data continuity. The sampling rhythm, sampling interval, and device sampling sequence are combined to generate an energy consumption signal separation scheme, ensuring that the sampling tasks of different devices do not overlap in time.

8. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 7, characterized in that, In the process of forming the energy consumption signal separation scheme, the sampling order of the devices during the interference period is periodically alternated, so that the same group of devices takes turns to obtain sampling priority in adjacent interference periods. The sampling resource distribution is balanced through the time alternation mechanism, and a transition buffer is set during the alternation to maintain the continuous execution of the sampling task.

9. The energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms according to claim 7, characterized in that, The adaptive adjustment steps for the energy consumption sampling process based on the energy consumption signal separation scheme are as follows: Based on the interference time period distribution and sampling rhythm configuration recorded in the energy consumption signal separation scheme, the sampling tasks of each device are grouped and divided, and reverse sampling silence is performed during the interference time period to eliminate the initial overlapping signal. After the reverse sampling is silenced, a short-delay sampling is performed. By adjusting the sampling start time, a time-displaced sampling sequence is formed to achieve layered signal acquisition. After completing the delayed sampling, an alternating sampling strategy is executed to make the sampling order of devices within the interference group alternate periodically and to set a transition buffer at the switching boundary to maintain time continuity. The sampled data is reconstructed in time sequence and energy consumption misjudgment is corrected. The reconstructed energy consumption data is then redistributed to the devices that actually generate current changes and the corrected monitoring results are output.

10. An energy consumption monitoring system for an enterprise energy management platform integrating AI algorithms, used to implement the energy consumption monitoring method for an enterprise energy management platform integrating AI algorithms as described in any one of claims 1-9, characterized in that, It includes a time calibration module, a feature index analysis module, an interference identification and processing module, a sampling rhythm control module, and a signal separation and correction module. The time calibration module establishes a unified time reference for the energy consumption monitoring process, synchronizes the start and stop times of various devices in the enterprise, acquires the current change data corresponding to the start and stop of each device, generates continuous time-series curves, and draws a current overlap area distribution map on a unified time scale to determine the start and end positions of the current overlap. The feature index analysis module, based on the current overlap region distribution map, segments the current time series curves of each time period, analyzes the boundary changes of the current waveform, extracts the transition features in the current change process of each device, and generates the feature index table corresponding to each device based on the transition features. The interference identification and processing module uses a feature index table to backtrack and compare the energy consumption sub-items, detects misjudgments of energy consumption caused by overlapping current waveforms, determines the time period of the misjudgment and the combination of equipment that caused the interference, and generates an interference zone record table to store the interference information. The sampling rhythm control module adjusts the energy consumption sampling rhythm and sampling order according to the interference zone record table, and sets the cross suppression window and sampling interval limit in the time dimension to form an energy consumption signal separation scheme for the interference period. The signal separation and correction module adaptively adjusts the energy consumption sampling process according to the energy consumption signal separation scheme. During the interference period, it executes reverse sampling silence, short-delay sampling and alternating sampling strategies. By adjusting the sampling rhythm and feedback rhythm, it separates overlapping signals and corrects the misjudgment of energy consumption attribution based on the separation results.

Images