CN-121981856-A - Energy data aggregation monitoring system and method based on edge calculation

CN121981856ACN 121981856 ACN121981856 ACN 121981856ACN-121981856-A

Abstract

The invention belongs to the technical field of energy data analysis, and discloses an energy data aggregation monitoring system and method based on edge calculation; the method comprises the steps of receiving original metering values reported by an affiliated energy acquisition terminal at a plurality of edge nodes to obtain an edge original energy aggregation sequence, carrying out sequence division on the edge original energy aggregation sequence, constructing an abnormal sensitive feature set by analyzing deviation degrees among different local working condition fragments, generating a monitoring response task queue by analyzing differences between the abnormal sensitive feature set and historical local energy data in each edge node, carrying out energy consumption evolution simulation on an adjusting effect of the monitoring response task queue in the edge nodes to determine an optimal simulation track, and carrying out power control on equipment of the edge nodes.

Inventors

CAO JINWEI
LIN CUNHUI

Assignees

瑞诺技术(深圳)有限公司

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. The energy data aggregation monitoring method based on edge calculation is characterized by comprising the following steps of: Receiving original metering values reported by the affiliated energy acquisition terminals at a plurality of edge nodes, and carrying out time sequence merging on the energy original metering values according to equipment identifiers to obtain an edge original energy aggregation sequence; performing sequence division on the edge original energy aggregation sequence to obtain a local working condition segment sequence, and constructing a local working condition label according to the energy consumption characteristics of the local working condition segment; constructing an abnormal sensitive feature set by analyzing the deviation degree between different local working condition fragments; Generating a monitoring response task queue by analyzing the difference between the abnormal sensitive feature set and the historical local energy data in each edge node; performing energy consumption evolution simulation on the adjusting effect of the monitoring response task queue in the edge node, and determining an optimal simulation track; And performing power control on the equipment of the edge node according to the optimal simulation track.
2. The method for monitoring and controlling the aggregation of energy data based on edge calculation according to claim 1, wherein the sequence division of the edge original energy aggregation sequence comprises: In the edge node, taking an original measurement value of a single device at continuous sampling time as row input, taking the identification of multiple devices at the same sampling time as column input, and constructing a local energy consumption matrix; according to the local energy consumption matrix, calculating an energy consumption variation matrix between adjacent columns along the time direction; Acquiring a device switching value state matrix and a scheduling instruction time sequence of an edge node, respectively comparing the device switching value state matrix and the scheduling instruction time sequence with an energy consumption variable quantity matrix element by element, identifying a working condition event and generating a working condition event marking matrix; and according to the time index of the working condition event occurrence in the working condition event marking matrix, segmenting the edge original energy aggregation sequence according to the time sequence to obtain a plurality of local working condition fragments.
3. The method for monitoring the aggregation of energy data based on edge calculation according to claim 2, wherein the method for constructing the local working condition label comprises the following steps: Extracting average energy consumption, energy consumption fluctuation and matching degree with a scheduling plan of each local working condition segment in the duration time of the local working condition segment, and constructing an energy consumption feature vector; clustering the different energy consumption feature vectors to obtain a plurality of stable energy consumption feature vector clusters; and constructing a local working condition descriptor for each stable energy consumption characteristic vector cluster, and using the local working condition descriptor as a local working condition label of a local working condition segment corresponding to each energy consumption characteristic vector in the energy consumption characteristic vector cluster.
4. The edge computing-based energy data aggregation monitoring method of claim 3, wherein the method for constructing the anomaly-sensitive feature set comprises the following steps: searching local working condition fragments in the edge nodes according to the local working condition labels, and integrating search results to obtain a local working condition fragment set; Calculating average energy consumption values of all local working condition fragments in the local working condition fragment set at each time point, and carrying out time sequence combination on the average energy consumption values to obtain a local energy consumption baseline curve under the local working condition label; calculating a relative deviation rate sequence of each local working condition segment according to the local energy consumption baseline curve; identifying a section exceeding a preset deviation threshold value in the relative deviation rate sequence, and marking the section as an abnormal section; Calculating the deviation duration time, the average deviation rate, the maximum deviation rate and the deviation development direction of each abnormal section in the relative deviation rate sequence, and constructing an abnormal sensitive characteristic; Combining the abnormal sensitive features of different relative deviation rate sequences to generate an abnormal sensitive feature set.
5. The edge computing-based energy data aggregation monitoring method of claim 4, wherein the generating a monitoring response task queue comprises: Carrying out weighted fusion on different feature dimension values of the abnormal sensitive features to obtain an abnormal index; Acquiring a device identification set corresponding to the local adjustable load data of the edge node and the abnormal sensitive feature set, and searching according to the device identification set to obtain a searching load record set; the searching load record set is composed of a plurality of load records, and each searching record comprises rated power, allowable load reduction power and influence indexes obtained by searching according to the equipment identifier corresponding to each abnormal sensitive characteristic; According to the reduction power of the abnormal index computing equipment, judging whether the reduction power does not exceed the allowable load reduction power corresponding to the equipment identifier; If the cut power does not exceed the allowable cut power, marking the corresponding equipment as candidate regulating equipment, and if the cut power exceeds the allowable cut power, not marking; combining the equipment identification, the influence index and the load reduction proportion of the candidate adjusting equipment to generate a response initial task; weighting and fusing the influence indexes of different response initial tasks and the load reduction proportion to obtain the priority of the response initial tasks; and sorting the response initial tasks in a descending order according to the priority, and constructing a monitoring response task queue according to the sorting order.
6. The method for monitoring the aggregate energy data based on edge computation according to claim 5, wherein the performing, in the edge node, energy consumption evolution simulation on the adjustment effect of the monitoring response task queue includes: extracting a recent energy consumption sequence from the local working condition segments, and searching historical working condition segments which are consistent with the local working condition labels of the local working condition segments and are in the same period in the edge nodes; based on the last period interval of the recent energy consumption sequence, intercepting each historical working condition segment, constructing a historical reference energy consumption sequence, and integrating to obtain an energy consumption reference sequence set; respectively extracting average energy consumption values and average change slopes of a recent energy consumption sequence and an energy consumption reference sequence set in a final period of time, and recording the average energy consumption values and the average change slopes as time attribute features and working condition position features of the corresponding sequences; splicing the energy consumption sequence, the time attribute feature and the working condition position feature to obtain a multidimensional feature vector; sequentially inputting the multidimensional feature vectors into a basic prediction model according to the time sequence of the energy consumption sequence, and performing forward operation once; Carrying out time sequence combination on the operation result of each sampling point to obtain a predicted energy consumption value sequence, and marking the predicted energy consumption value sequence as a non-regulation predicted track; and expanding the non-regulation predicted track in the predicted window by adopting a rolling mode to obtain a rolling predicted track.
7. The edge computing-based energy data aggregation monitoring method of claim 6, wherein the method of determining an optimal simulated trajectory comprises: Superposing and reducing power on the rolling predicted track, and obtaining a plurality of virtual energy consumption tracks according to different superposition window lengths; comparing each virtual energy consumption track with a local energy consumption baseline curve, reserving the virtual energy consumption tracks with the relative deviation rate smaller than a preset deviation threshold value, and marking the virtual energy consumption tracks as candidate energy consumption tracks; Sliding rightwards along the time sequence of the virtual energy consumption track, and marking the sampling points marked by sliding as target sampling points; When each target sampling point is updated, calculating the average deviation rate from the starting point to the target sampling point, and comparing the average deviation rate with a preset deviation threshold; if the average deviation rate is smaller than the preset deviation threshold, generating the minimum adjustment duration of the candidate energy consumption track according to the interval duration between the starting point and the target sampling point; Comparing the minimum adjustment time lengths of different candidate energy consumption tracks, marking the candidate energy consumption track corresponding to the minimum value of the minimum adjustment time length as a target energy consumption track, and determining the expected duration time length of the corresponding task; Extracting the starting relative time, the expected duration, the reduction power and the equipment identification of each task in the monitoring response task queue, and constructing a corresponding task disturbance template to form a task disturbance template set; Constructing a plurality of task combination schemes by performing feasibility analysis of task combination on the task disturbance template set; And (3) superposing the rolling predicted track on each task combination scheme to obtain a candidate simulated track, and evaluating the candidate simulated track to determine an optimal simulated track.
8. The edge computing-based energy data aggregation monitoring method of claim 7, wherein the method of constructing a multiple task combination scheme comprises: Enumerating task template pairs of any two task disturbance templates in the task disturbance template set, checking the task template pairs, and judging whether the same equipment identification exists in the task template pairs; If the same equipment identifier does not exist in the task template pair, marking the task template pair as a combinable template pair; If the same equipment identifier exists in the two task disturbance templates, calculating the time overlapping duration of the two task disturbance templates in the execution process, and if the time overlapping duration is greater than a preset overlapping duration threshold value, not marking; If the time overlapping duration is not greater than a preset overlapping duration threshold, judging whether the accumulated reduction power of the two task disturbance templates is greater than the allowable load reduction power or not; if the accumulated reduced power is not greater than the allowable load reduction power, the task template pair can be combined with the template pair, and if the accumulated reduced power is greater than the allowable load reduction power, marking is not carried out; Constructing a combinable template subset according to task disturbance templates in different combinable template pairs; calculating the ratio of the reduction power of each task disturbance template in the composable template subset to the influence index of the equipment to obtain the contribution degree of the reduction power; the task disturbance templates are ordered in descending order according to the power contribution degree to obtain a task template sequence; sequentially accumulating the cut power in the task template sequence to obtain combined cut power, and judging whether the combined cut power reaches the target power; And if the target power is reached, stopping accumulation, and reducing a task disturbance template corresponding to the power in the process of combining and accumulating to obtain a task combination scheme.
9. The edge computing-based energy data aggregation monitoring method of claim 8, wherein the method of evaluating candidate simulated trajectories comprises: respectively calculating peak load, deviation area with a local energy consumption baseline curve and task combination quantity for each candidate simulation track, and obtaining a simulation evaluation index through weighted fusion; and comparing the simulation evaluation indexes of different candidate simulation tracks, and marking the candidate simulation track corresponding to the minimum value of the simulation evaluation index as the optimal simulation track.
10. The energy data aggregation monitoring system based on edge calculation is applied to the energy data aggregation monitoring method based on edge calculation as set forth in any one of claims 1 to 9, and is characterized by comprising the following steps: The energy consumption data acquisition module is used for receiving original metering values reported by the affiliated energy acquisition terminals at a plurality of edge nodes, and carrying out time sequence merging on the energy original metering values according to equipment identification to obtain an edge original energy aggregation sequence; The energy consumption data dividing module is used for carrying out sequence division on the edge original energy aggregation sequence to obtain a local working condition segment sequence, and constructing a local working condition label according to the energy consumption characteristics of the local working condition segment; The abnormal energy consumption data extraction module is used for constructing an abnormal sensitive feature set by analyzing the deviation degree among different local working condition fragments; The edge node response task generating module is used for generating a monitoring response task queue by analyzing the difference between the abnormal sensitive feature set and the historical local energy data in each edge node; The energy consumption simulation optimization module is used for performing energy consumption evolution simulation on the adjustment effect of the monitoring response task queue in the edge node and determining an optimal simulation track; and the energy consumption control module is used for controlling the power of the equipment of the edge node according to the optimal simulation track.

Description

Energy data aggregation monitoring system and method based on edge calculation Technical Field The invention relates to the technical field of energy data analysis, in particular to an energy data aggregation monitoring system and method based on edge calculation. Background With the improvement of the permeability of renewable energy sources and the increase of the number of energy consumption terminals, a large number of energy source acquisition terminals are commonly deployed on a power distribution side and a park side, and metering data from various devices and a plurality of time scales are locally aggregated and monitored through edge computing nodes so as to reduce cloud communication and storage pressure and improve response speed. In the prior art, the energy data aggregation monitoring mainly relies on centralized modeling and regulation of a cloud center, threshold value alarming or coarse granularity statistical analysis based on single equipment is adopted on the edge side, real-time characterization of fine granularity energy consumption behavior on an edge node with limited resources is difficult, abnormal behavior is difficult to map to adjustable equipment in time, linkage regulation has hysteresis and an insufficient regulation target, dynamic influence evaluation on energy consumption tracks of different regulation task combinations in a future period is lacking, and insufficient reduction or excessive regulation is easy to occur in abnormal monitoring and regulation, so that the operation load of equipment is increased, and the abnormal convergence effect is weakened. Therefore, it is necessary to provide an energy data aggregation monitoring method for carrying out local working condition modeling, abnormal sensitive characteristic aggregation and multitasking combined prediction regulation on the energy consumption behavior at the edge node side, and in the background, how to carry out unified aggregation modeling on multi-equipment energy consumption data at the edge node side, and support continuous monitoring and fine management on the load state and the operation working condition become an important direction in the construction of an energy management system. Disclosure of Invention In order to overcome the defects in the prior art and achieve the purposes, the invention provides the following technical scheme that the energy data aggregation monitoring method based on edge calculation comprises the following steps: Receiving original metering values reported by the affiliated energy acquisition terminals at a plurality of edge nodes, and carrying out time sequence merging on the energy original metering values according to equipment identifiers to obtain an edge original energy aggregation sequence; performing sequence division on the edge original energy aggregation sequence to obtain a local working condition segment sequence, and constructing a local working condition label according to the energy consumption characteristics of the local working condition segment; constructing an abnormal sensitive feature set by analyzing the deviation degree between different local working condition fragments; Generating a monitoring response task queue by analyzing the difference between the abnormal sensitive feature set and the historical local energy data in each edge node; performing energy consumption evolution simulation on the adjusting effect of the monitoring response task queue in the edge node, and determining an optimal simulation track; And performing power control on the equipment of the edge node according to the optimal simulation track. Preferably, the sequence dividing the edge original energy aggregation sequence includes: In the edge node, taking an original measurement value of a single device at continuous sampling time as row input, taking the identification of multiple devices at the same sampling time as column input, and constructing a local energy consumption matrix; according to the local energy consumption matrix, calculating an energy consumption variation matrix between adjacent columns along the time direction; Acquiring a device switching value state matrix and a scheduling instruction time sequence of an edge node, respectively comparing the device switching value state matrix and the scheduling instruction time sequence with an energy consumption variable quantity matrix element by element, identifying a working condition event and generating a working condition event marking matrix; and according to the time index of the working condition event occurrence in the working condition event marking matrix, segmenting the edge original energy aggregation sequence according to the time sequence to obtain a plurality of local working condition fragments. Preferably, the method for constructing the local working condition label comprises the following steps: Extracting average energy consumption, energy consumption fluctuation and matching degree with a schedul