Electric energy meter automatic meter reading method and system based on Internet of Things

Abstract

The invention relates to the technical field of electric energy meters, in particular to an automatic electric energy meter reading method and system based on the Internet of Things, and the method comprises the following steps: based on an electronic electric energy meter in a distribution box, collecting instantaneous current and voltage, arranging a time stamp power sequence, comparing continuous sampling power data, and summarizing a change trend; and judging power change direction continuity to identify abnormal segments, screening uncovered sections to judge continuity, and adjusting interval identifiers to generate credible level distribution data. According to the method, data acquisition and remote real-time communication of time sequence synchronization are utilized, continuous sampling power change trend combing is combined, and through dynamic trend comparison and intra-interval direction reversal behavior judgment, short-time fluctuation and power change abnormal sections can be identified, similarity screening and classification are performed on data intervals without abnormality, and the accuracy of the abnormal data is improved. And after section classification, a credible level identifier is given in combination with trend characteristics, a periodic distribution structure is obtained, and segmented discrimination of an abnormal energy consumption interval and the dynamic perception capability of an energy consumption state are improved.

Description

Technical Field

[0001] This invention relates to the field of electricity meter technology, and in particular to an automatic meter reading method and system based on the Internet of Things (IoT). Background Technology

[0002] Electricity meters primarily involve equipment for measuring, monitoring, and recording electricity usage. This encompasses the hardware structure design of the metering device, measurement principles, electricity calculation methods, data storage and display methods, and communication technologies for remote data acquisition and management. Traditional automatic meter reading methods involve staff periodically visiting the meter installation site to manually read the displayed values, or using portable terminal devices for short-range data collection, and then returning the data to the system backend for aggregation and processing.

[0003] Existing technologies rely on manual, periodic on-site readings or close-range data collection using portable equipment. The collection cycle is limited by personnel scheduling, resulting in slow data updates from electricity meters. Long collection intervals can lead to the loss of details regarding energy fluctuations. The long information transmission chain during manual reading increases the risk of transcription errors and data omissions during data recording and transmission. Sudden power consumption anomalies cannot be captured during data generation, creating a situation where data integrity and timeliness are difficult to balance. Furthermore, timely identification of localized, short-term abnormal energy consumption sections within the distribution network is impossible. Data silos hinder subsequent management and anomaly tracing, restricting the level of intelligent and refined energy consumption monitoring. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an automatic meter reading method and system for electricity meters based on the Internet of Things.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an automatic meter reading method for electricity meters based on the Internet of Things, comprising the following steps: S1: Based on the electronic energy meter in the distribution box, analyze the energy meter output data, collect instantaneous current and voltage detection data, upload and organize the detection results according to the sampling time, and obtain the time-stamped power sequence characterization; S2: Based on the time-stamped power sequence characterization, compare two consecutive sets of active power data, determine the order of adjacent sampling times, sort out the changing trends of each set of power data, summarize the changes in sampling intervals, and obtain a continuous interval power change sequence. S3: Based on the continuous interval power change sequence, determine the continuity of the power change direction, analyze the occurrence of direction reversal in the segment, identify the active direction change interval as an abnormal detection segment, and obtain a set of abnormal direction reversal segments. S4: Based on the set of abnormal segments with reversed direction, filter out the uncovered data segments, analyze the similarity of the change amplitude of each group within the segment based on the power change trend, determine the consistency of the segment with the previous segment trend, divide it into continuous and reliable segments, and obtain stable and continuous change segments. S5: Based on the stable and continuously changing segments, adjust the data sequence interval identifiers, assign each level identifier according to the segment change trend, and combine the sampling period time order to obtain the reliable level distribution data.

[0006] The present invention improves upon this invention by including the following: the time-stamped power sequence characterization includes timestamp information, channel markers, and raw measurement data; the continuous interval power change sequence includes change direction, change rate, and sequence index; the set of direction reversal anomaly segments includes reversal time, anomaly segment number, and fluctuation judgment attribute; the stable continuous change segment includes segment number, continuity identifier, and data consistency description; and the confidence level distribution data includes level type, segment arrangement order, and confidence label.

[0007] The present invention is improved in that the step of obtaining the time-stamped power sequence characterization is specifically as follows: S111: Based on the electronic energy meter in the distribution box, analyze the collected instantaneous current data, determine the sampling order of instantaneous voltage data, optimize the synchronization of data and the time recorded by the wireless communication unit, and correlate the current data and voltage data according to time to obtain the time correlation group of electrical parameters. S112: Based on the electrical parameter time correlation group, calculate the mapping relationship between the data content generated by the upload action and the channel marker, compare the integrity of the channel marker received by the server with the sampling order, and obtain the multi-channel electrical parameter acquisition sequence; S113: Based on the multi-channel electrical parameter acquisition sequence, analyze the combination of current and voltage data in each group of data, determine the correspondence between each sampling time and channel marker, organize the dataset according to the sampling order, and obtain the time-stamped power sequence characterization.

[0008] The present invention is improved in that the step of obtaining the continuous interval power change sequence is specifically as follows: S211: Based on the time-stamped power sequence characterization, compare the active power data corresponding to adjacent sampling times, analyze the sequential relationship between each group of data, determine whether there is a continuous sequence of sampling times, identify active power data groups with temporal continuity, and obtain an adjacent power comparison set. S212: Based on the adjacent power comparison set, compare the changing trends of active power data before and after, determine the power change direction of each group of data, and combine the sampling time interval of each group to aggregate the power change performance to obtain the power change mark set of the sampling segment; S213: Based on the power change marker set of the sampling segment, filter power change segments with continuous time intervals, adjust the data arrangement order, aggregate the power change trends and time characteristics of each segment, reconstruct the data index relationship, and obtain a continuous interval power change sequence.

[0009] The present invention is improved in that the step of obtaining the set of direction-reversed abnormal segments is specifically as follows: S311: Based on the continuous interval power change sequence, analyze the power direction information of each group, determine the switching state of the direction of adjacent sampling points, compare the direction identifier of the sampling data, identify the nodes in each sampling segment where the direction changes, and obtain the power reversal node sequence. S312: Based on the power inversion node sequence, calculate the distribution characteristics of each sampling interval, and combine the node occurrence frequency with the sampling interval fluctuation of adjacent nodes, using the formula: ; By analyzing the intensity parameters of the direction reversal disturbance and selecting sampling intervals with prominent power change characteristics, a set of active direction reversal intervals is obtained. Indicates channel In the sampling interval Internal direction reversal disturbance intensity parameters, Indicates channel In the sampling interval The number of nodes with reversed direction appearing within the range. Indicates sampling interval Total number of sampling points within, Indicates channel The The sampling interval parameter corresponding to each direction-reversed node. Indicates channel In the sampling interval The average value of the sampling interval parameter for all inverted nodes in the internal direction; S313: Based on the set of active direction reversal intervals, determine the statistical attributes of each interval, organize the sampling segment numbers, fluctuation characteristics and reversal frequency, statistically classify the results of each interval, and obtain a set of abnormal direction reversal segments.

[0010] The present invention is improved in that the step of obtaining the stable continuously changing segment is specifically as follows: S411: Based on the set of abnormal segments with reversed direction, filter the continuous time segments in the data sequence that are not classified as abnormal, and mark them by the start and end sampling points of the segments to obtain the index set of uncovered segments; S412: Based on the uncovered segment index set, calculate the variation amplitude of continuous active power in each segment, compare the similarity of power variation amplitude within the same segment, identify segments with fluctuation convergence, and obtain segment power amplitude information; S413: Based on the power amplitude information of the segment, determine whether the changing trend of the current segment is consistent with that of the previous segment, optimize the segment grouping logic, and obtain a stable and continuously changing segment.

[0011] The present invention is improved in that the steps for obtaining the credibility level distribution data are specifically as follows: S511: Based on the stable and continuously changing segments, analyze the start and end indices in the original data sequence, determine whether there is overlap or repetition of the interval identifiers between each segment, optimize the arrangement order of the identifiers, adjust the label content corresponding to the conflicting identifiers, and obtain the interval identifier adjustment set. S512: Based on the interval identifier adjustment set, determine the power change direction characteristics of each segment of data, analyze the persistence of the change trend, and combine the trend and fluctuation performance to assign reliability labels for each category to obtain the segment level label group. S513: Based on the segment level labeling group, integrate the time sequence corresponding to the sampling period, compare the segment arrangement index and level label, and aggregate the segment level and time sequence data to obtain the reliable level distribution data.

[0012] The present invention is improved in that the instantaneous current refers to the instantaneous current value measured by the current sensor of the electronic energy meter at a certain moment, the voltage detection data refers to the instantaneous voltage value measured by the voltage sampling circuit of the electronic energy meter at the same moment, and the active power data refers to the instantaneous active power value calculated by multiplying the current detection result and the voltage detection result at the corresponding moment when the energy meter samples once.

[0013] An automatic meter reading system for electricity meters based on the Internet of Things (IoT), the system comprising: The time-series power acquisition module analyzes the output data of the electronic energy meter in the distribution box, calls the time recording function of the wireless unit, collects instantaneous current and voltage detection data, uploads and organizes the detection results according to the sampling time, and obtains a time-stamped power sequence characterization. The power trend analysis module, based on the time-stamped power sequence, compares two consecutive sets of active power data, determines the order of adjacent sampling times, organizes the changing trends of each set of power data, summarizes the changes in sampling intervals, and obtains a continuous interval power change sequence. Based on the continuous interval power change sequence, the anomaly identification module determines the continuity of the power change direction, analyzes the occurrence of direction reversal in the segment, identifies the active direction change interval as anomaly detection segment, and obtains a set of direction reversal anomaly segments. Based on the set of abnormal segments with reversed direction, the segment filtering module filters out uncovered data segments, analyzes the similarity of the change amplitude of each group within the segment based on the power change trend of the feeder branch, determines that the segment is consistent with the trend of the previous segment, divides it into continuous and reliable segments, and obtains stable and continuous change segments. The rating module adjusts the data sequence interval identifiers based on the stable and continuously changing segments, assigns each level identifier according to the segment change trend, and obtains reliable level distribution data by combining the sampling period time sequence.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by utilizing time-synchronized data acquisition and remote real-time communication, combined with continuous sampling of power change trends, and through dynamic trend comparison and determination of directional reversal behavior within intervals, it is possible to identify short-term fluctuations and abnormal power change segments. For data intervals without anomalies, similarity screening and classification are performed. After segment classification, a confidence level label is assigned based on trend characteristics to obtain a periodic distribution structure. This improves the segmented discrimination of abnormal energy consumption intervals and the dynamic perception of energy consumption status, thereby achieving data hierarchical aggregation, automatic anomaly focusing, and efficient energy consumption monitoring. Attached Figure Description

[0015] Figure 1 This is a flowchart of the main steps of the present invention; Figure 2 This is a flowchart illustrating the acquisition of time-stamped power sequence characterization in this invention; Figure 3 This is a flowchart illustrating the process of obtaining the continuous interval power change sequence in this invention. Figure 4 This is a flowchart illustrating the process of obtaining the set of abnormal segments with reversed direction in this invention. Figure 5 This is a flowchart illustrating the process of obtaining stable, continuously changing segments in this invention. Figure 6 This is a flowchart illustrating the process of obtaining the credibility level distribution data in this invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0017] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0018] All user-related information involved in this invention (including but not limited to biometric information, identity verification information, behavioral data, device information, and other data that can be used for identity verification and personalized services) is collected and processed with the user's full knowledge and voluntary consent. The collection, storage, and use of all information strictly comply with applicable national and regional laws and regulations, and meet relevant data protection standards and policy requirements. The use of data is limited to purposes necessary for providing the technical services of this invention, and reasonable technical and management measures will be taken to ensure the security and confidentiality of users' personal information in terms of information protection and privacy.

[0019] Example: Please refer to Figure 1 This invention provides a technical solution: an automatic meter reading method for electricity meters based on the Internet of Things, comprising the following steps: S1: Based on the electronic energy meter in the distribution box, analyze the output data of the electronic energy meter, call the time recording function of the wireless unit, collect the instantaneous current detection results and voltage detection results, pair the detected current and voltage with the sampling time through the upload action, organize the data group according to the time sequence, and obtain the time-stamped power sequence characterization. S2: Based on the time-stamped power sequence characterization, compare two sets of continuously sampled active power data, determine the order of adjacent sampling times, organize the power data change trends of each set, summarize the change performance according to the sampling interval of each set of data, and obtain the continuous interval power change sequence. S3: Based on the continuous interval power change sequence, determine the continuity of the power change direction, use trend recognition rules to analyze the behavior of direction reversal within the segment, record the distribution of reversal behavior within the sampling period, divide the detection segment according to the reversal occurrence characteristics, classify the active direction change interval as abnormal detection segment, and obtain the set of abnormal direction reversal segments. S4: Based on the set of abnormal segments with reversed direction, filter the data segments that are not covered in the data sequence. For the continuous power change trend of the feeder branch, analyze the similarity of the power change amplitude of each group within each segment, determine whether the change trend of the segment is consistent with the previous segment, and define the corresponding segment as a continuous and reliable segment to obtain a stable and continuous change segment. S5: Based on stable and continuously changing segments, adjust the interval labels of the original data sequence, assign each level label to reliable and abnormal intervals according to the segment change trend, and then combine the time sequence of each sampling period to obtain the confidence level distribution data.

[0020] The time-stamped power sequence characterization includes timestamp information, channel markers, and raw measurement data. The continuous interval power change sequence includes the change direction, change rate, and sequence index. The set of direction reversal anomaly segments includes the reversal time, anomaly segment number, and fluctuation judgment attribute. The stable continuous change segment includes the segment number, continuity identifier, and data consistency description. The confidence level distribution data includes the level type, segment arrangement order, and confidence label.

[0021] In S1, the time recording function refers to the function within the wireless communication unit (such as LoRa or NB-IoT modules) that can accurately record the moment of each data acquisition, usually relying on the module's RTC or system clock to achieve synchronization between data and time; the instantaneous current detection result refers to the instantaneous current value measured by the current sensor of the electronic energy meter at a certain moment, which reflects the current intensity flowing in the current circuit of the energy meter; the voltage detection result refers to the instantaneous voltage value measured by the voltage sampling circuit of the electronic energy meter at the same moment, which reflects the voltage level of the circuit currently monitored by the energy meter; the upload action refers to the process of transmitting the data (such as current, voltage, and sampling time) collected on-site to a remote data acquisition server or management platform through the wireless communication unit, usually in the form of data packet transmission; the sampling moment refers to the precise point in time when the data (current, voltage) is detected by the electronic energy meter and timestamped by the wireless module, to ensure the timing and synchronization of subsequent data processing.

[0022] In S2, active power data refers to the instantaneous active power value obtained by multiplying the current and voltage detection results at the corresponding moment (and considering the power factor, if applicable) during a single sampling of the energy meter; adjacent sampling time sequence refers to the temporal relationship between two consecutive samples within the same energy meter data stream, which is the time basis for analyzing power change trends; power data change trend refers to the results obtained by comparing and analyzing continuous active power data according to the time sequence to determine whether the power value increases, decreases, or remains the same over time, thus describing the direction and rate of power change; data sampling interval refers to the time difference between two consecutive data acquisitions (sampling), which is the time basis for data segmentation analysis, and is usually automatically determined by the system's sampling frequency setting; change performance refers to the specific situation of power change within the sampling interval, which can be an increase or decrease, or characteristics such as the rate of change and stability.

[0023] In S3, the continuity of power change direction refers to whether the direction of power change (e.g., continuously increasing or decreasing) remains consistent across multiple consecutive sampling periods. If the change direction frequently reverses, it is considered discontinuous; otherwise, it is considered continuous. Trend identification rules refer to the algorithms or logic used to determine the direction of power change, such as identifying the trend of power growth or decline by comparing the magnitude of adjacent power data, and whether the trend has reversed. Segments refer to continuous time slices or data segments that divide the entire data sequence according to certain standards (e.g., trends, change characteristics, etc.), with each segment having relatively uniform change characteristics. Reversal behavior refers to events where the power change direction changes from increasing to decreasing, or from decreasing to increasing, which is an important characteristic of trend change. Distribution refers to the statistical description of the occurrence time, frequency, and location of power change direction reversal events within a sampling period or segment. Detection segments refer to data segments marked as requiring further analysis or attention after trend identification and reversal behavior analysis, and are the key part of subsequent anomaly detection and credibility determination. Active direction change intervals refer to time segments within the statistical period where power change direction reversal events occur frequently, and the data in these intervals usually exhibits abnormal fluctuations.

[0024] In S4, the uncovered data segment refers to the remaining data segment that was not classified as an abnormal segment (such as an active direction change interval) during the anomaly detection process, i.e., the "normal" or "not judged as abnormal" part; the continuous power change trend of the feeder branch refers to the overall characteristics of the continuous active power change collected on the branch circuit of the industrial park or community power distribution system, reflecting the stability or change pattern of the load power consumption status of the branch; each segment refers to the data of each continuous time period after the anomaly detection segment is divided, and each segment usually contains multiple continuous sampling points; the power change amplitude refers to the change in all continuous active power values ​​within a single segment, which can be obtained by calculating the difference between each two adjacent points, and is used to measure the severity of power change within the segment; the continuous reliable segment refers to the time segment where the power change pattern is confirmed to be relatively stable and no abnormal fluctuations have occurred after trend analysis and change amplitude judgment, and the segment can be considered as a data segment with reliable metering results.

[0025] In S5, the interval identifier refers to the labeling of each time period or data segment when processing the entire power data sequence, indicating whether it belongs to a "reliable interval" or an "abnormal interval"; the segment change trend refers to the trend characteristics of power change over time within each segment, which is used for subsequent level allocation and reliability judgment; the reliable interval refers to the time interval that is identified as having a stable power data change trend and no abnormal fluctuations after multiple layers of data processing and judgment, and can be directly used for meter reading and settlement; the abnormal interval refers to the data time interval that is identified as having frequent reversals in the direction of power change or abnormal change amplitude, which usually requires subsequent verification or supplementary data collection.

[0026] Please see Figure 2 The specific steps for obtaining the time-stamped power sequence characterization are as follows: S111: Based on the electronic energy meter in the distribution box, analyze the collected instantaneous current data, determine the sampling order of the instantaneous voltage data obtained by the voltage sampling circuit, optimize the synchronization of the data with the time recorded by the wireless communication unit, and correlate the current data and voltage data according to time to obtain the electrical parameter time correlation group. The system calls the internal interface of the current sensor to read the instantaneous current value, which reflects the current intensity through the meter's measurement circuit at a specific point in time. Simultaneously, it calls the voltage sampling circuit to obtain the instantaneous voltage measurement value at the corresponding moment. Both current and voltage values ​​are sampled at a preset time period, such as once every 5 seconds. To achieve accurate time-dimensional correspondence between the two types of data, the response delays of the voltage sampling link and the current channel need to be compared separately. The processing delay deviation between the two is recorded, and the sampling time of the slower-responding channel is adjusted forward or backward through program control to form a unified time identifier. Then, the system time at that moment is obtained through the wireless communication unit, and the timestamp is obtained using the real-time clock in the wireless module. The voltage and current values ​​are time-bound. To ensure matching accuracy, during data group sampling... The sampling time difference is compared to determine if it is lower than the set error range. If it exceeds the set range (e.g., 0.5 seconds), the data set is discarded. If it is within the error range, the data set is merged into a complete electrical parameter pair. The electrical parameter pair contains a sampling time, a set of current values, and a set of voltage values, along with channel labeling information, such as data from channel A. This data set is then written to a buffer, which is sorted by timestamp. Each valid voltage and current pair is sequentially stored in the list, gradually building a complete electrical parameter time-related set. In industrial park applications, if the sampling point is during the midday peak, the current value is read as 4.2 and the voltage value as 227.3, corresponding to a sampling time of 14:03:12.256. These three values ​​are packaged into a single data item and added to the electrical parameter time series for subsequent power change analysis and trend tracking.

[0027] S112: Based on the electrical parameter time correlation group, calculate the mapping relationship between the data content generated by the upload action and the channel marker, compare the integrity of the channel marker received by the server with the sampling order, and obtain the multi-channel electrical parameter acquisition sequence; After establishing the electrical parameter time correlation group, channel labels are extracted for each data group. The channel information attached to each data entry is read sequentially. This label is automatically generated by the energy meter when the data is generated, serving as an identification identifier for the circuit to which the data belongs in the three-phase power grid system. For example, the label may be A, B, or C. The number of data records uploaded each time is compared with the required number of channels. A three-phase system should have three different channel labels. If an uploaded record only contains data groups for channels A and C, it indicates that channel B is missing. This batch of data should be marked as incomplete and removed from the analysis sequence. Next, the actual data upload order received from the server is read, and the time stamp of each data entry is compared to see if it is in ascending order. If the order is correct and the channel labels are fully covered, the upload batch is set as valid; otherwise, if the received data is not in order, the batch is set as invalid. If the channel labels are duplicated (e.g., both sets of data are labeled as channel B), or if the time labels are reversed (e.g., the first time is greater than the second time), the batch is identified as abnormal and discarded. From all uploaded data batches, datasets with complete channel information and matching sampling time order are selected. Each valid batch corresponds to one complete electrical parameter acquisition operation, forming a multi-channel electrical parameter acquisition sequence. This sequence serves as the basis for subsequent power calculations, trend identification, and other operations. For example, if a batch of uploaded data contains three records with times of 14:03:12.256, 14:03:12.310, and 14:03:12.360, labeled as channels A, B, and C respectively, and it is confirmed that the three channels are complete and the times are in sequence, then this data batch is included in the valid sequence and used for subsequent power trend analysis.

[0028] S113: Based on the multi-channel electrical parameter acquisition sequence, analyze the combination of current and voltage data in each group of data, determine the correspondence between each sampling time and channel marker, organize the dataset according to the sampling order, and obtain the time-stamped power sequence characterization; The multi-channel electrical parameter acquisition sequence constructed in the previous step is processed by data combination. Voltage and current values ​​for each channel are extracted one by one, and the three channels are merged using their timestamps as the sole sorting criterion. During the merging process, the sampling time difference is compared pairwise. For example, if the time difference between channel A and channel B is 0.1 seconds, and it is less than the set allowable time deviation threshold (e.g., 0.2 seconds), it is considered that they can be merged into the same sampling period; otherwise, the group of data is discarded. After merging the valid data, voltage and current combination items are constructed for each group of data. For example, channel A is 227.3 and 4.2, B is 226.8 and 4.1, and C is 228.0 and 4.3. The combination items are stored in the data structure in ascending time order. Next, a product calculation process is performed on each combination to calculate the instantaneous active power value of each record. The result is added to the data record in the form of a power value, while retaining the original channel information and sampling time, thus constructing a data sequence with complete power information. In power grid monitoring, by traversing this data sequence, the instantaneous active power value of each record can be obtained. The power consumption data of each channel at each time point is then combined into a complete data node through a triplet structure of channel, time, and power. The constructed time-stamped power sequence characterization data structure includes channel label, voltage value, current value, sampling time, and power value. This structure serves as the data basis input for subsequent trend judgment and anomaly identification. In field applications, for example, at 14:03:12.256, the power factor is 0.9, the power of channel A is 859, the power of channel B is 837, and the power of channel C is 882. These three form a complete set of data nodes, which are added to the sequence table for subsequent time period continuity judgment and abnormal segment extraction.

[0029] Please see Figure 3 The specific steps for obtaining the continuous interval power change sequence are as follows: S211: Based on time-stamped power sequence characterization, compare the active power data corresponding to adjacent sampling times, analyze the sequential relationship between each group of data, determine whether there is a continuous sequence of sampling times, identify active power data groups with temporal continuity, and obtain an adjacent power comparison set. The active power data in the records, arranged in ascending order of time, is scanned one by one. The time tag field of two adjacent sampling points is extracted, and it is determined whether the time value of the later data is greater than that of the earlier data. If this relationship is satisfied, the two data samples are considered to have an sequential relationship and are set as a group of analysis objects. Then, the corresponding active power values ​​are extracted from this group of data, and the magnitude of the two power values ​​is compared. If the later value is greater than the earlier value, the power of the data pair is marked as rising; if they are equal, it is marked as flat; if they are less, it is marked as falling. After completing the power value comparison, the difference in the time field is recalculated. If the time interval falls within the set allowable sampling interval range, such as between 0.9 seconds and 1.2 seconds, it is considered... The data pairs must be time-continuous; otherwise, they are considered interrupted or abnormal. For example, if the previous record time is 14:03:12.256 with a corresponding power of 865.0, and the next record time is 14:03:13.192 with a corresponding power of 872.4, the time difference is 0.936 seconds, and the power difference is 7.4. This indicates an increase in power, and the time continuity is satisfied, so it is recorded as a valid data pair. If the next record time is 14:03:15.000, the time difference is 2.744 seconds, then the current pair is discarded. After traversing all data records, all power data pairs that meet the time order and interval conditions are selected in the above manner, forming a set of data arranged in ascending time order with a clear power comparison direction, thus obtaining the adjacent power comparison set.

[0030] S212: Based on the adjacent power comparison set, compare the changing trends of active power data before and after, determine the power change direction of each group of data, and combine the sampling time interval of each group to aggregate the power change performance to obtain the power change label set of the sampling segment; The algorithm sequentially reads the two preceding and following power values ​​for each pair of power data, compares their magnitudes, and determines the direction of power change. If the subsequent value is greater than the preceding value, it is marked as an increase; otherwise, it is marked as a decrease. If the values ​​are the same, it is marked as "unchanged." After the determination, the time difference between the timestamp fields of the two data points is calculated and used as the sampling interval for that pair of power changes. The direction of power change and the time interval are then stored in a new structure in a one-to-one correspondence. To avoid abnormal data affecting trend judgment, a lower limit threshold for the power change amplitude is set to 3.0 (units are consistent). When the absolute value of the power change is less than this value, it is marked as no significant change and thus "unchanged." This threshold is for reference only. The load current fluctuation characteristics of the household side are set based on the analysis of typical residential scenarios. For example, if the electricity consumption fluctuates greatly during the day and changes little at night, it is more reasonable to take the middle level of this value. Based on this rule, during the data processing, when two consecutive power records are 873.2 and 875.1, and the difference is 1.9, which is less than 3.0, the segment is set as "flat", and the corresponding sampling interval is recorded as 1.0 seconds. The record item is constructed as: {direction=flat, interval=1.0}. This type of label item is generated one by one and merged into a power change label set for the sampling segment. The record structure includes the change direction and sampling time interval of each pair of data, forming a label sequence describing the power trend change.

[0031] S213: Based on the power change marker set of the sampling segment, filter the power change segments with continuous time intervals, adjust the data arrangement order, aggregate the power change trends and time characteristics of each segment, reconstruct the data index relationship, and obtain the continuous interval power change sequence. The process reads the time interval and direction markers for each record. First, it determines whether there is a continuous time interval between adjacent segments. If the time difference between the end time of the current segment and the start time of the next segment is less than a set breakpoint threshold, it is considered a continuous segment. The threshold is set to 1.5 seconds, with the value determined by floating 50% above the maximum sampling interval. If the difference exceeds this threshold, the segment is considered interrupted and not included in the continuous analysis area. Subsequently, multiple segments with continuous time are merged to form a power change segment. Then, the direction markers within each segment are statistically counted. If the number of upward direction markers in a segment is more than twice the number of downward direction markers, the overall trend is set to upward. Otherwise, if the number of downward direction markers is dominant, it is set to downward. If the number of equal markers is the largest and exceeds half of the total, the trend is considered to be upward. If the condition is set to stable, this criterion ensures the representativeness of the directional marking results within the segment. Then, the start time of each segment is used as the starting point of the segment index. The directional trend within the segment is bound to the time interval information, and the index structure is reconstructed to form a power change segment data structure containing index number, trend mark, start and end time. In a set of test data, segment 1 contains 6 mark items, with the directional distribution being 4 rising, 1 flat, and 1 falling, and the time interval is 1.0 second. Therefore, the trend mark of segment 1 is rising, the index number is set to 1, the start time is 14:03:12.256, and the end time is 14:03:18.256. This type of segment is constructed in sequence and sorted by the start time to generate a continuous interval power change sequence.

[0032] Please see Figure 4 The specific steps for obtaining the set of reversed anomaly fragments are as follows: S311: Based on the continuous interval power change sequence, analyze the power direction information of each group, judge the switching state of the direction of adjacent sampling points, compare the direction indicators of the sampling data, identify the nodes in each sampling segment where the direction changes, and obtain the power reversal node sequence. The power direction marker field is extracted sequentially from each sampling segment. The current segment's direction is compared one by one with the previous segment's direction to determine if a direction switch occurs. If the previous segment's direction is rising while the current segment's direction is falling, or vice versa, the starting point of the current segment is marked as a direction reversal node. During the comparison process, segments with a flat direction are skipped. Flat segments are neither considered as reversal starting points nor do they terminate the continuation of the previous trend. The direction switch determination is based on the change in the "direction" field value in the data structure. The direction value is set to three types: +1 for rising, 0 for flat, and -1 for falling. For example, if a continuous five-segment data direction sequence is: +1, +1, 0, -1, -1, then the third segment is flat, and the direction has not changed. However, the fourth segment switches from +1 to -1, which is determined to be a direction reversal event. The node position is recorded as the starting sampling point time of the fourth segment, and a unique identifier is generated for this time point. The power reversal node is numbered and includes forward and backward directional state information to form a reversal node item. The entire power change sequence is traversed, and each node whose directional state changes from rising to falling or from falling to rising is added to the result structure. Simultaneously, the segment index, sampling time, and directional switching type of the node are recorded. In an application scenario, if during a morning peak period on a certain day, the sampling times are 08:00:01, 08:00:02, 08:00:03, 08:00:04, and 08:00:05, corresponding to the directional sequence of rising, rising, falling, falling, rising, then directional switching occurs in the 3rd and 5th segments respectively. The start time of the 3rd segment (08:00:03) and the start time of the 5th segment (08:00:05) are extracted as reversal nodes respectively. The resulting power reversal node sequence consists of multiple nodes, each recording the time, forward and backward directional state, and sequence number, used for subsequent reversal frequency statistics and fluctuation segment identification.

[0033] S312: Based on the power inversion node sequence, calculate the distribution characteristics of each sampling interval, combining the node occurrence frequency and the sampling interval fluctuation of adjacent nodes, using the formula: ; By analyzing the intensity parameters of the direction reversal disturbance and selecting sampling intervals with prominent power change characteristics, a set of active direction reversal intervals is obtained. Indicates channel In the sampling interval Internal direction reversal disturbance intensity parameters, Indicates channel The number of nodes with reversed direction appearing within the sampling interval. Indicates sampling interval Total number of sampling points within, Indicates channel The The sampling interval parameter corresponding to each direction-reversed node. Indicates channel In the sampling interval The average value of the sampling interval parameter for all inverted nodes in the inner direction. Indicates the first The percentage deviation of the sampling interval of each inverted node from the average sampling interval; The direction reversal disturbance intensity parameter is a statistical parameter used to quantify the activity and fluctuation characteristics of power change direction reversal phenomena in a specific channel within a specific sampling interval. It comprehensively considers the density of direction reversals (i.e., the proportion of the number of reversal nodes to the total number of sampling points) and the unevenness of the sampling intervals between reversal nodes (i.e., the deviation of each reversal interval from the average reversal interval of the entire interval), reflecting the instability of the power change trend within that interval. If there are many direction reversal nodes or large fluctuations in the reversal interval within a sampling interval, the value of the direction reversal disturbance intensity parameter will be higher, indicating that the power change within that interval has strong discontinuity or abnormal fluctuations. Conversely, if there are few reversals and the reversal intervals are relatively uniform, the parameter value will be lower, indicating that the power change trend within the interval is relatively stable and the fluctuations are small. It is used to identify and filter time periods or data segments with abnormal fluctuation behavior in power changes. The number of times the direction-reversed node appears within the statistical sampling interval is used as a parameter. The execution method involves finding and counting all node indices that satisfy the direction sign reversal within the time interval. Assuming the interval is a 5-minute sampling window and the sampling period is 5 seconds, the total number of sampling points is... There are 3 reversal events in the node sequence, with the corresponding reversal intervals being as follows: Second, Second, The time intervals are mapped from [40, 60] to [0, 1] using a minimum-maximum normalization method, resulting in the following normalized sampling intervals: , , ;

[0034] Then calculate the average of these three items: ; By comparing each normalized time interval with the mean, calculating its offset ratio, and taking the absolute value, we obtain the following results: ; ; ; Sum the three terms and divide by the number of nodes The disturbance amplitude is obtained as follows: ; Simultaneously calculate the node density as follows: The direction reversal disturbance intensity parameter is obtained by superimposing the two values, and then substituted into the formula to calculate: ; Set disturbance intensity parameters The criteria for dividing the intervals are as follows: when When the disturbance is stable, it is determined to be a stable disturbance section, which means that the power change direction in this section is basically consistent, there are few reversal events and low fluctuations in node intervals. when When this occurs, it is determined to be a disturbance transition zone, which means that there are a certain number of direction reversals within this zone, but the overall volatility has not yet constituted a significant disturbance. when When the frequency is high, it is identified as an active disturbance segment, indicating that the direction reversal behavior is frequent in this segment and the sampling interval between nodes is unevenly distributed, showing obvious fluctuation characteristics.

[0035] The results indicate that there is a significant inconsistency between the frequency of power direction switching behavior and the sampling rhythm in the current segment. The value of this intensity parameter has crossed the judgment boundary between the stable and transitional segments, and has the conditions to be used as a key characteristic indicator for perturbation screening. Therefore, the value itself not only represents the quantitative state of the perturbation intensity in the current segment, but also serves as the admission standard for judging whether this segment should be included in the anomaly analysis process, affecting the subsequent composition and numbering distribution of the set of direction reversal anomaly segments.

[0036] S313: Based on the set of active direction reversal intervals, determine the statistical attributes of each interval, organize the sampling segment numbers, fluctuation characteristics and reversal frequency, statistically classify the results of each interval, and obtain a set of abnormal direction reversal segments. Each inversion segment data item in the set is read, and the range of sampling segment numbers covered by the segment, the number of internal inversion nodes, and the start and end time information of the segment are extracted. The continuity of the sampling segment numbers is judged to confirm whether the segment consists of multiple consecutive sampling segments. If the number interval is no more than 1, it is considered continuous. Then, the inversion frequency of the segment is statistically analyzed, that is, the number of direction reversals occurring per unit time is calculated, and this value is used as the inversion frequency value. The inversion frequency classification benchmark is set to 2 times per minute. If the inversion frequency of a segment is greater than or equal to 2 times per minute, it is marked as a high-frequency zone; if it is lower than this value, it is marked as a low-frequency zone. This benchmark value is set based on empirical data of load change rate in industrial scenarios. Combined with actual cases, in industrial production power consumption scenarios, if 12 inversion behaviors are detected in a 5-minute time period, the frequency is 2.4 times per minute. For high-frequency areas, the segment number, start and end times, number of reversals, and frequency classification are combined to form a structural record. Then, the fluctuation characteristics of the segment are evaluated. The severity of the fluctuation is determined by calculating the standard deviation of the time interval between adjacent reversal nodes. If the standard deviation is greater than 0.8 seconds, it is classified as severe fluctuation; otherwise, it is classified as gentle fluctuation. The standard deviation benchmark value is set based on a sampling period of 1 second, and 80% of it is taken as the distinction boundary. Subsequently, the fluctuation status is added to the data structure of the segment. The segment number, reversal frequency classification, and fluctuation characteristic status are classified. All segments are classified and archived into four types: high-frequency severe, high-frequency stable, low-frequency severe, and low-frequency stable, forming a set of direction reversal abnormal segments. Each record contains the number, time, frequency, and fluctuation judgment result, which are used for subsequent abnormal area removal and data confidence level assignment.

[0037] Please see Figure 5 The specific steps for obtaining stable, continuously changing paragraphs are as follows: S411: Based on the set of anomaly segments with reversed direction, filter the continuous time segments in the data sequence that have not been classified as anomalies, mark the start and end sampling points of the segments, and obtain the index set of uncovered segments; Read all segments identified as abnormal in the previous stage, extract the start and end sampling segment numbers for each segment, and construct a complete list of abnormal segment numbers. Then, in the total power change data sequence, traverse the segment numbers of each data point from beginning to end and compare them one by one with the list of abnormal segment numbers. If the current number does not belong to any abnormal segment, record it as the starting point of a normal segment. When multiple consecutive segment numbers do not appear in the list of abnormal numbers, set the first and last segment numbers as the start and end boundaries of a certain uncovered continuous segment and add it to the uncovered candidate list. In this process, the continuity of the numbers must be considered. A difference of 1 indicates continuity; if the number jumps out of bounds... If the number exceeds 1, the current segment is considered an interrupt, the current segment recording ends and the next record is started. Then, all candidate segments are deduplicated, retaining data with a start number less than the end number, and removing segments shorter than two segments to ensure that the data has a statistical basis. For example, if the set of abnormal segment numbers is [5–10], [15–17], [22–24], and the complete sequence segment numbers are 1 to 30, then the number ranges [1–4], [11–14], [18–21], [25–30] are all consecutive numbers that do not belong to abnormal segments. They are marked as uncovered segments in sequence, and a unique index value and corresponding start and end segment numbers are set for each segment to obtain the uncovered segment index set.

[0038] S412: Based on the index set of uncovered segments, calculate the variation amplitude of continuous active power in each segment, compare the similarity of power variation amplitude within the same segment, identify segments with similar fluctuation characteristics, and obtain segment power amplitude information. The sampled data within each segment is extracted one by one. First, the start and end numbers of the segment are recorded by index, and the entire power value sequence contained in the segment is extracted. Then, the continuous power data in each segment are processed in chronological order, and the power difference between two adjacent sampling points is calculated and recorded in a temporary list. This power difference is the value obtained by subtracting the previous data point from the subsequent data point, in watts, representing the increase or decrease in power change within the segment. The absolute values ​​of all power differences are then recorded, and the average, maximum, and minimum amplitudes of power change within the segment are calculated. The standard deviation is then calculated to describe the consistency of power change within the segment. The benchmark for judging similarity is set to a standard deviation not exceeding 50. If the standard deviation of the power difference within a segment is less than or equal to 50, then... If the fluctuation range is relatively consistent, it is considered that there is an abnormal fluctuation trend. The value is set with reference to the typical residential load daytime non-abrupt usage scenario. For example, if the power data in a certain segment are 890, 895, 900, 898, 902, and 906 respectively, the corresponding adjacent differences are 5, 5, -2, 4, and 4, and the absolute values ​​are 5, 5, 2, 4, and 4 respectively. The calculated average difference is 4, and the standard deviation is about 1.1, which is significantly lower than the set benchmark value of 50. Then this segment is classified as a fluctuation convergence segment. After traversing all uncovered segments, the average change range, maximum and minimum difference, standard deviation, fluctuation similarity judgment result, and index number of each segment are combined into a structural item to form a segment power amplitude information list. This list can be used for subsequent continuous trend judgment and reliable segment selection.

[0039] S413: Based on the power amplitude information of the current segment, determine whether the changing trend of the current segment is consistent with that of the previous segment, optimize the segment grouping logic, and use the formula: ; A stable, continuously changing segment is obtained, in which, Indicates the first The quantitative results of trend persistence for each segment are used to describe the similarity in the changing trends between two adjacent segments. Indicates the first The statistical measure of the power variation amplitude characteristics within a segment is obtained by summarizing and calculating based on the amplitude of continuous power variation within that segment. Indicates the first The statistical measures of the power variation amplitude characteristics within each segment are related to the variation trend of the previous segment. This represents the difference in the magnitude of change between the current segment and the previous segment. The absolute value of the difference reflects the degree of deviation in the magnitude of change between the segments. As a normalized composite quantity, it is used to proportionalize the differences in the characteristics of change between segments, so that the difference comparison has relative significance between segments. Trend persistence quantification refers to measuring the continuity and consistency of power change trends between two adjacent segments, used to assess the similarity between the current segment and the previous segment in terms of power change magnitude characteristics; when When the value is close to 1, it indicates that the power change amplitude of the current segment is very similar to that of the previous segment, the change trend remains highly consistent, and the continuity is strong, which can be judged as a continuation of the same stable change trend. When the value approaches 0, it indicates a significant difference in the power change amplitude between the current segment and the previous segment, a large deviation in the trend, and weak continuity; therefore, they should belong to different trend segments. By comparing the ratio of the absolute difference in power change amplitude between adjacent segments to the total value, the stationary continuity of power data over time can be reflected. Extract the numbered items respectively Section and the Within the segment, the power amplitude variation data is used to obtain the set of absolute values ​​of the power differences at the sampling points. Then, the mean of the samples within the segment is calculated. Assuming the first... The difference in continuous power data within the section is: [3.1W, 3.3W, 3W, 3.2W]; Corresponding mean amplitude: W; Current number The difference between the data segments is: [4.2W, 4W, 3.8W, 4.1W]; Calculated W; The mean amplitude of each segment is normalized, and the maximum value is selected for normalization. The maximum value is taken as the known maximum amplitude value of the power change difference within the analyzed period. W, after normalization: ; ; Substitute the normalized result into the calculation formula: ; The overall judgment range of the quantitative results for trend persistence can be divided into the following segments: when When a segment is identified as a "highly continuous segment," it indicates that the range of change is almost identical and can be merged unconditionally. when When the segment is judged to be a "medium-length continuous segment", it indicates that the trend of change is continuous and can be used as a candidate segment for merging. when When the boundary is identified as a "critical zone", it indicates that the trend has undergone substantial fluctuations and needs to be reassessed in conjunction with other indicators. when When this occurs, it is judged as a "trend abrupt change segment," indicating that the trend direction has deviated significantly and should be retained as an independent segment.

[0040] Therefore, the current calculation results It can be deduced that this segment meets the trend continuation condition and has the logical basis to be merged into a stable and continuously changing segment, thus serving as a component of the result of this step, "stable and continuously changing segment".

[0041] Please see Figure 6 The specific steps for obtaining the credibility level distribution data are as follows: S511: Based on stable and continuously changing segments, analyze the start and end indices in the original data sequence, determine whether there is overlap or repetition of interval identifiers between segments, optimize the arrangement order of the identifiers, adjust the label content corresponding to conflicting identifiers, and obtain the interval identifier adjustment set. The starting and ending number fields in the paragraph record are called, and the number range of all identified segments is traversed to construct the overall interval index structure. For each pair of starting and ending numbers, the interval is expanded to obtain its complete number list. All paragraph number lists are cross-referenced pairwise. If there is overlapping numbering (i.e., any number in the current segment also appears in another segment), then the current interval is considered to have overlapping or duplicate identifiers. The duplicate numbers and their corresponding multiple paragraph index numbers are recorded. In a practical implementation scenario, if the range of segment 1 is [11–18] and the range of segment 2 is [17–22], then numbers 17 and 18 are overlapping. Conflict handling is required for segments 1 and 2. In the conflict handling step, the segment identifier with the earlier starting number and longer length is retained first, and the other conflicting segment's identifier is... The identifier is adjusted to move to the next position, for example, segment 2 is changed to [19–22]. If a conflict occurs with other segments after the move, the adjustment continues until no overlapping numbers appear. If the length of the interval after the segment adjustment is less than the original length but still greater than 0, the adjusted interval is retained and its status is marked as "conflict adjusted". If the length of the interval after the adjustment is 0 or does not meet the minimum valid interval requirement, its identifier status is marked as "conflict invalid". At the same time, the identifier status field of each segment is updated to distinguish between the three types of marking status: valid, conflict adjusted, and conflict invalid. Then, all the adjusted segment identifiers are rearranged in ascending order of the starting number to form a segment index structure that is non-repeating and non-overlapping. At the same time, a structure item with segment index, status label, and start and end number as content is constructed to obtain the interval identifier adjustment set.

[0042] S512: Based on the interval identifier adjustment set, determine the power change direction characteristics of each segment of data, analyze the persistence of the change trend, and combine the trend and fluctuation performance to assign reliability labels for each category to obtain the segment level label group. The data content of each segment is retrieved, and all power data and their direction change indicators within the segment are extracted one by one. The consistency of direction within the segment is statistically analyzed. By accumulating the number of each direction indicator, the dominant direction type is determined. If the number of data points in the upward direction exceeds 80% of the total number of data points in the segment, the segment is marked as "continuously rising". If the number of data points in the downward direction meets the same percentage condition, it is marked as "continuously falling". If the proportion of the sideways direction exceeds 60%, it is marked as "stable fluctuation". If none of the types are met, it is marked as "unclear trend". Then, combined with the average fluctuation amplitude value of the segment, the trend stability is further subdivided. The fluctuation amplitude stability threshold is set as a maximum change amplitude of less than 50. If the condition is met and the direction is consistent, it is marked as "continuous and stable trend". If the direction is consistent but the fluctuation is greater than 50, it is marked as "continuous and violent trend". If the trend is unclear and the fluctuation is greater than 10, it is marked as "unclear trend". A value of 0 is marked as "unstable". The threshold settings are based on the analysis of daily load changes in residential electricity. The standard value of 50 is used to distinguish between steady state and violent state, and 100 is used to define obvious abnormal fluctuation segments. Based on the trend judgment results and fluctuation classification results, a reliability label is assigned to each segment. The label types are defined as five categories: Level 1 Reliable, Level 2 Reliable, Level 3 Reliable, Pending Review, and Unreliable. "Continuous and stable trend" matches Level 1 Reliable, "Continuous and violent trend" and "Smooth fluctuation" are set as Level 2 Reliable, "Unclear trend and moderate fluctuation" are set as Level 3 Reliable, "Unstable" is set as Unreliable, and "Consistent direction but no conclusion on fluctuation" is set as Pending Review. For example, if the segment number is 03 and the upward trend accounts for 90% of the time, and the maximum fluctuation amplitude is 36, then this segment is set as Level 1 Reliable. Using the segment number as an index, a corresponding reliability level label is added to each segment to construct a segment level label group.

[0043] S513: Based on the segment level labeling group, integrate the time sequence corresponding to the sampling period, compare the segment arrangement index and level label, and aggregate the segment level and time sequence data to obtain the reliable level distribution data; The process involves reading the start and end numbers of each segment one by one, then retrieving the sampling time field corresponding to the number in the original data sequence. This time information is combined with the grade label to generate the corresponding time interval value for each segment. For example, segment 4 has a start number of 40 and an end number of 46, corresponding to a sampling time of 10:00:40 to 10:00:46. The grade label is Level 2 Reliability. Therefore, the time combination for this segment is [10:00:40–10:00:46, Level 2 Reliability]. Next, all segments are sorted by index number, and the combined results are arranged in chronological order, constructing a structure with the fields "Segment Index, Start Time, End Time". The system generates a data table for "start and end time, level label", and simultaneously counts the occurrence frequency of each type of label and the cumulative length of its corresponding time period. It determines whether the distribution density of the same level is uniform across different time periods. If a certain level label is concentrated more than three times within a certain hour and the total cumulative time is greater than 30 minutes, then that time period is recorded as a high-confidence concentration area. An additional structure record is constructed, which includes the fields of "level type, start and end time, number of occurrences, and cumulative duration". All paragraph time sorting data is merged with the level label structure to form complete distribution data. The aggregation result is output according to the logic of "time order + level distribution" to obtain the reliable level distribution data.

[0044] An Internet of Things (IoT)-based automatic meter reading system for electricity meters, comprising: The time-series power acquisition module analyzes the output data of the electronic energy meter in the distribution box, calls the time recording function of the wireless unit, collects instantaneous current and voltage detection data, uploads and organizes the detection results according to the sampling time, and obtains a time-stamped power sequence characterization. The power trend analysis module is based on time-stamped power sequence representation. It compares two consecutive sets of active power data, determines the order of adjacent sampling times, sorts out the power data change trends of each set, summarizes the changes in sampling intervals, and obtains a continuous interval power change sequence. The anomaly identification module determines the continuity of the power change direction based on the continuous interval power change sequence, analyzes the occurrence of direction reversal in the segment, identifies the active direction change interval as anomaly detection segment, and obtains a set of direction reversal anomaly segments. The segment selection module filters out uncovered data segments based on the set of abnormal segments with reversed direction. It analyzes the similarity of the change amplitude of each group within the segment based on the power change trend of the feeder branch, judges the consistency of the segment with the trend of the previous segment, divides it into continuous and reliable segments, and obtains stable and continuous change segments. The rating module adjusts the data sequence interval labels based on stable and continuously changing segments, assigns each level label according to the segment change trend, and obtains reliable level distribution data by combining the sampling period time sequence.

[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for automatic meter reading of electricity meters based on the Internet of Things, characterized in that, Includes the following steps: S1: Based on the electronic energy meter in the distribution box, analyze the energy meter output data, collect instantaneous current and voltage detection data, upload and organize the detection results according to the sampling time, and obtain the time-stamped power sequence characterization; S2: Based on the time-stamped power sequence characterization, compare two consecutive sets of active power data, determine the order of adjacent sampling times, sort out the changing trends of each set of power data, summarize the changes in sampling intervals, and obtain a continuous interval power change sequence. S3: Based on the continuous interval power change sequence, determine the continuity of the power change direction, analyze the occurrence of direction reversal in the segment, identify the active direction change interval as an abnormal detection segment, and obtain a set of abnormal direction reversal segments. S4: Based on the set of abnormal segments with reversed direction, filter out the uncovered data segments, analyze the similarity of the change amplitude of each group within the segment based on the power change trend, determine the consistency of the segment with the previous segment trend, divide it into continuous and reliable segments, and obtain stable and continuous change segments. S5: Based on the stable and continuously changing segments, adjust the data sequence interval identifiers, assign each level identifier according to the segment change trend, and combine the sampling period time order to obtain the reliable level distribution data.

2. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The time-stamped power sequence characterization includes timestamp information, channel markers, and raw measurement data. The continuous interval power change sequence includes change direction, change rate, and sequence index. The set of direction reversal anomaly segments includes reversal time, anomaly segment number, and fluctuation judgment attribute. The stable continuous change segment includes segment number, continuity identifier, and data consistency description. The confidence level distribution data includes level type, segment arrangement order, and confidence label.

3. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The specific steps for obtaining the time-stamped power sequence characterization are as follows: S111: Based on the electronic energy meter in the distribution box, analyze the collected instantaneous current data, determine the sampling order of instantaneous voltage data, optimize the synchronization of data and the time recorded by the wireless communication unit, and correlate the current data and voltage data according to time to obtain the time correlation group of electrical parameters. S112: Based on the electrical parameter time correlation group, calculate the mapping relationship between the data content generated by the upload action and the channel marker, compare the integrity of the channel marker received by the server with the sampling order, and obtain the multi-channel electrical parameter acquisition sequence; S113: Based on the multi-channel electrical parameter acquisition sequence, analyze the combination of current and voltage data in each group of data, determine the correspondence between each sampling time and channel marker, organize the dataset according to the sampling order, and obtain the time-stamped power sequence characterization.

4. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The specific steps for obtaining the continuous interval power change sequence are as follows: S211: Based on the time-stamped power sequence characterization, compare the active power data corresponding to adjacent sampling times, analyze the sequential relationship between each group of data, determine whether there is a continuous sequence of sampling times, identify active power data groups with temporal continuity, and obtain an adjacent power comparison set. S212: Based on the adjacent power comparison set, compare the changing trends of active power data before and after, determine the power change direction of each group of data, and combine the sampling time interval of each group to aggregate the power change performance to obtain the power change mark set of the sampling segment; S213: Based on the power change marker set of the sampling segment, filter power change segments with continuous time intervals, adjust the data arrangement order, aggregate the power change trends and time characteristics of each segment, reconstruct the data index relationship, and obtain a continuous interval power change sequence.

5. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The specific steps for obtaining the set of direction-reversed abnormal segments are as follows: S311: Based on the continuous interval power change sequence, analyze the power direction information of each group, determine the switching state of the direction of adjacent sampling points, compare the direction identifier of the sampling data, identify the nodes in each sampling segment where the direction changes, and obtain the power reversal node sequence. S312: Based on the power inversion node sequence, calculate the distribution characteristics of each sampling interval, and combine the node occurrence frequency with the sampling interval fluctuation of adjacent nodes, using the formula: ； By analyzing the intensity parameters of the direction reversal disturbance and selecting sampling intervals with prominent power change characteristics, a set of active direction reversal intervals is obtained. Indicates channel In the sampling interval Internal direction reversal disturbance intensity parameters, Indicates channel In the sampling interval The number of nodes with reversed direction appearing within the range. Indicates sampling interval Total number of sampling points within, Indicates channel The The sampling interval parameter corresponding to each direction-reversed node. Indicates channel In the sampling interval The average value of the sampling interval parameter for all inverted nodes in the internal direction; S313: Based on the set of active direction reversal intervals, determine the statistical attributes of each interval, organize the sampling segment numbers, fluctuation characteristics and reversal frequency, statistically classify the results of each interval, and obtain a set of abnormal direction reversal segments.

6. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The specific steps for obtaining the stable, continuously changing segment are as follows: S411: Based on the set of abnormal segments with reversed direction, filter the continuous time segments in the data sequence that are not classified as abnormal, and mark them by the start and end sampling points of the segments to obtain the index set of uncovered segments; S412: Based on the uncovered segment index set, calculate the variation amplitude of continuous active power in each segment, compare the similarity of power variation amplitude within the same segment, identify segments with fluctuation convergence, and obtain segment power amplitude information; S413: Based on the power amplitude information of the segment, determine whether the changing trend of the current segment is consistent with that of the previous segment, optimize the segment grouping logic, and obtain a stable and continuously changing segment.

7. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The specific steps for obtaining the trust level distribution data are as follows: S511: Based on the stable and continuously changing segments, analyze the start and end indices in the original data sequence, determine whether there is overlap or repetition of the interval identifiers between each segment, optimize the arrangement order of the identifiers, adjust the label content corresponding to the conflicting identifiers, and obtain the interval identifier adjustment set. S512: Based on the interval identifier adjustment set, determine the power change direction characteristics of each segment of data, analyze the persistence of the change trend, and combine the trend and fluctuation performance to assign reliability labels for each category to obtain the segment level label group. S513: Based on the segment level labeling group, integrate the time sequence corresponding to the sampling period, compare the segment arrangement index and level label, and aggregate the segment level and time sequence data to obtain the reliable level distribution data.

8. The method for automatic meter reading of electricity meters based on the Internet of Things according to claim 1, characterized in that, The instantaneous current refers to the instantaneous current value measured by the current sensor of the electronic energy meter at a certain moment. The voltage detection data refers to the instantaneous voltage value measured by the voltage sampling circuit of the electronic energy meter at the same moment. The active power data refers to the instantaneous active power value calculated by multiplying the current detection result and voltage detection result at the corresponding moment when the energy meter samples once.

9. An automatic meter reading system for electricity meters based on the Internet of Things, characterized in that, The system is used to implement the Internet of Things-based automatic meter reading method for electricity meters as described in any one of claims 1-8, and the system includes: The time-series power acquisition module analyzes the output data of the electronic energy meter in the distribution box, calls the time recording function of the wireless unit, collects instantaneous current and voltage detection data, uploads and organizes the detection results according to the sampling time, and obtains a time-stamped power sequence characterization. The power trend analysis module, based on the time-stamped power sequence, compares two consecutive sets of active power data, determines the order of adjacent sampling times, organizes the changing trends of each set of power data, summarizes the changes in sampling intervals, and obtains a continuous interval power change sequence. Based on the continuous interval power change sequence, the anomaly identification module determines the continuity of the power change direction, analyzes the occurrence of direction reversal in the segment, identifies the active direction change interval as anomaly detection segment, and obtains a set of direction reversal anomaly segments. Based on the set of abnormal segments with reversed direction, the segment filtering module filters out uncovered data segments, analyzes the similarity of the change amplitude of each group within the segment based on the power change trend of the feeder branch, determines that the segment is consistent with the trend of the previous segment, divides it into continuous and reliable segments, and obtains stable and continuous change segments. The rating module adjusts the data sequence interval identifiers based on the stable and continuously changing segments, assigns each level identifier according to the segment change trend, and obtains reliable level distribution data by combining the sampling period time sequence.

Images