Electric energy metering box remote monitoring system based on Internet of Things

Abstract

The invention relates to the technical field of remote monitoring, in particular to an electric energy metering box remote monitoring system based on the Internet of Things, which comprises an electric parameter arrangement module, a disturbance feature recognition module, an abnormal partition adjustment module, an event level judgment module and a link response detection module. According to the method, the polarity change points in the current slope derivative are extracted to construct the inflection point density sequence, the disturbance section can be positioned in a segmented mode, the change trend is marked, the disturbance section is recognized by introducing the disturbance intensity factor and combining the standardization result of the voltage change rate, the current harmonic distortion and the power factor, a misjudgment area is effectively eliminated, and the accuracy of recognition is improved. According to the method, a plurality of electrical parameter change characteristics in a section are extracted, risk grade division is implemented, fine early warning of electrical disturbance is realized, and link fluctuation identification capability and overall early warning precision are improved by introducing a multi-node response delay difference vector and evaluating node link response stability.

Description

Technical Field

[0001] This invention relates to the field of remote monitoring technology, and in particular to a remote monitoring system for electricity metering boxes based on the Internet of Things. Background Technology

[0002] The field of remote monitoring technology involves the real-time perception and remote management of the status information of target equipment or systems. It mainly includes core components such as information acquisition, communication transmission, remote control, and data processing, and is widely used in scenarios such as industrial equipment monitoring, environmental monitoring, and power system operation supervision. Among these, network control of front-end acquisition devices via communication modules, combined with a back-end platform for visualizing monitoring data and issuing anomaly alarms, is a common basic approach in this field. This technology has specific requirements for the accuracy of remote perception of equipment status, the stability of communication links, the efficiency of data interaction, and fault response time, and systematically integrates sensing technologies, communication protocols, and data management solutions to construct a monitoring system.

[0003] Traditional remote monitoring systems for electricity metering boxes refer to remote monitoring systems built for tasks such as electricity data acquisition, power anomaly monitoring, and electrical safety inspection within electricity metering boxes. They typically rely on voltage and current transformers, temperature and humidity probes, and switch quantity acquisition units to acquire data, which is then uploaded to a central server via GPRS, CDMA, or 5G communication. In these systems, fixed transmission terminals are generally deployed to achieve timed data acquisition and transmission. The backend uses set rules to manually or rule-basedly compare the received data to determine conditions such as line faults, meter malfunctions, or unauthorized opening of the box. Traditional systems have a relatively fixed structure, and their functionality usually relies on underlying hard-coded settings and periodic data upload mechanisms. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a remote monitoring system for electricity metering boxes based on the Internet of Things.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a remote monitoring system for electricity metering boxes based on the Internet of Things, the system comprising: The electrical parameter processing module acquires real-time monitoring data from the power metering box, sends the real-time monitoring data to the central monitoring equipment via the Internet of Things, performs truncation and rearrangement on the sequence data, deletes erroneous records, and generates synchronized cleaned data results. Based on the synchronous cleaning data results, the disturbance feature recognition module extracts the current inflection point. When the inflection point density increases continuously and exceeds the density change threshold, the corresponding segment is recorded as a disturbance candidate segment. If the inflection point density of subsequent segments shows a decreasing trend, it is marked as a segmentation node, and an inflection point segmentation structure is generated. The abnormal partitioning adjustment module calculates the disturbance intensity factor of each segment based on the inflection point segmentation structure, compares it with the disturbance identification reference threshold, and if it is less than the disturbance identification reference threshold, the corresponding segment is relabeled as a stable segment; if it is greater than the threshold, the disturbance segment attribute is maintained, and a disturbance segment update label is generated. The event level determination module calls the updated label of the disturbance segment, calculates the risk level assessment value of the disturbance segment, and performs interval matching with the event level classification standard to classify the risk level of each segment, implements risk warning, and generates graded disturbance warning records.

[0006] The present invention is improved in that the synchronous cleaning data results include time series synchronization information, abnormal data removal records, and effective range of electrical parameter amplitude; the inflection point segmentation structure includes a segment index table, an inflection point density distribution matrix, and disturbance candidate segment labels; the disturbance segment update label includes a disturbance segment status indicator code, intensity factor comparison results, and a segment adjustment mapping table; and the graded disturbance early warning record includes an event level coding table, an indicator interval matching sequence, and risk level mapping data.

[0007] The present invention is improved in that the electrical parameter processing module includes: The data stream receiving submodule acquires real-time monitoring data from the power metering box. The real-time monitoring data includes three-phase voltage RMS value sequence, three-phase current RMS value sequence, power factor sequence, and voltage and current phase angle sequence. The real-time monitoring data is sent to the central monitoring equipment in real time through the Internet of Things transmission path. The timestamp, sampling number, and measurement point identifier of each sequence are recorded to establish a unified structure of the original receiving dataset of electrical parameters. The time structure synchronization submodule calls the timestamp field in the original electrical parameter received dataset, performs time interval continuity judgment on the sampled sequence, deletes data rows with abrupt changes or sample number jumps, performs difference comparison on the time axis of each sequence in the retained data, extracts the minimum sampling period as the overall time alignment benchmark, constructs a unified time series index table, performs synchronous rearrangement on the original sequence according to the index table, and generates a time synchronization alignment data table. The amplitude cleaning and filtering submodule calls the three-phase voltage effective value column, three-phase current effective value column and power factor column in the time synchronization alignment data table, compares them with the set upper limit value of voltage, lower limit value of current and normal range of power factor, deletes data rows that exceed the boundary value or are zero, retains the valid sequence and reconstructs the index in a unified manner, and generates the synchronization cleaning data result.

[0008] The present invention is improved in that the disturbance feature recognition module includes: Based on the synchronous cleaning data results, the slope derivative calculation submodule selects the current sequence of any one phase of the three-phase current, performs numerical difference on the amplitude of adjacent sampling points in the current sequence, calculates the current change between the current point and the previous time point and divides it by the sampling interval to construct the current slope sequence, performs numerical difference on the current slope sequence once, extracts the difference between two consecutive slope values ​​and records the derivative direction to generate the first-order current change derivative sequence. The inflection point density extraction submodule calls the current first-order change derivative sequence, determines whether the derivative direction between two adjacent points has reversed sign, marks the position as a polarity inflection point, divides the complete sampling period into equal time intervals, counts the number of polarity inflection points per second and arranges them in time order, constructs the change frequency value corresponding to the interval per second, and generates an inflection point density sequence per unit time. The paragraph index construction submodule performs time series trend judgment on the inflection point density value based on the inflection point density sequence within the unit time period, extracts the trend sequence of density value changes in multiple consecutive time segments, and for continuous upward segments in the trend, it judges whether the density value is greater than the set inflection point density mutation threshold. If it is satisfied, the corresponding segment is marked as a disturbance candidate segment. If a downward trend appears later, the inflection point is recorded as the paragraph boundary, and a paragraph separation index is constructed to generate an inflection point segmentation structure.

[0009] The present invention is improved in that the abnormal partition adjustment module includes: The disturbance index extraction submodule filters and marks the segment numbers as disturbance candidate segments according to the inflection point segmentation structure, extracts the three-phase voltage change rate, total harmonic distortion of current and power factor data corresponding to each segment, establishes three segment index sequences respectively, and generates a multi-segment disturbance feature sequence set. The normalization quantity construction submodule calls the multi-segment perturbation feature sequence set, performs Z-score standardization on the three indicators respectively, extracts the average value of the samples in the standardized sequence, calculates the perturbation intensity factor, compares the perturbation intensity factor of each segment with the set perturbation identification reference threshold and records the comparison status, and generates a perturbation threshold response result set. The segment status correction submodule, based on the comparison status information of each segment in the disturbance threshold response result set, marks the corresponding segment as stable if the disturbance intensity factor is less than the disturbance identification reference threshold, and if it is greater than the threshold, maintains the disturbance label, reassigns status labels to all segments, and generates a disturbance segment update label.

[0010] The present invention is improved in that the formula for obtaining the disturbance intensity factor is specifically as follows: ; Among them, Y i Z represents the perturbation intensity factor of the i-th segment. i,UZ represents the average value of the voltage change rate of the i-th segment after Z-score normalization. i,H Z represents the average value of the total harmonic distortion rate of the current in the i-th segment after Z-score normalization. i,P θ represents the average power factor of the i-th segment after Z-score normalization. i This represents the power factor volatility of the i-th segment.

[0011] The present invention is improved in that the event level determination module includes: The disturbance feature extraction submodule calls the disturbance segment index retained in the disturbance segment update label, extracts the voltage peak-to-average power ratio sequence, current total harmonic distortion rate sequence, current slope change value sequence, instantaneous active power sequence and power factor sequence corresponding to each segment, extracts the maximum value, difference or range of the five indicators respectively, establishes the five-dimensional disturbance feature vector corresponding to each segment, and generates a disturbance feature parameter group; The risk assessment calculation submodule calls the five disturbance index quantities of each segment in the disturbance characteristic parameter group, extracts the values ​​of voltage peak-to-average power ratio, current total harmonic distortion rate, current change rate, power change amplitude and power factor fluctuation range, calculates and obtains the risk level assessment value of the disturbance segment, and generates a disturbance risk assessment sequence. The risk level mapping submodule retrieves the level division threshold range based on the risk value corresponding to each segment in the disturbance risk assessment sequence, performs range position judgment, maps the risk value to the corresponding risk level range number, records the mapping result between the event level and the disturbance segment, and generates a graded disturbance early warning record.

[0012] The present invention has an improvement, wherein the system further includes: The link response detection module calls the hierarchical disturbance warning record, obtains the alarm issuance command time and node status reporting time from the log, retrieves the delay values ​​of two physically adjacent nodes, constructs three node response delay vectors and calculates the maximum time difference, performs normalization processing, obtains the standardized response difference index, compares the standardized response difference index with the response stability threshold, if it exceeds the threshold, it determines that the node has a potential response degradation trend and generates a node link instability identifier. The node link instability identifier includes a response delay time group, an adjacent node difference matrix, and a stability determination flag.

[0013] The present invention is improved in that the link response detection module includes: The response delay extraction submodule calls the node identifier of each node with a risk level in the graded disturbance early warning record, obtains the alarm issuance command time and the first reporting time of the node status, calculates the time difference between the two as the target node response delay value, and records the time difference sequence and the corresponding node number information to generate a node response delay record table. The adjacency difference construction submodule retrieves the response delay values ​​of the immediate preceding and following nodes in the physical topology for each target node based on the node response delay record table, constructs a vector including the response delays of the three nodes, performs the maximum time difference calculation between the pairwise combinations, and obtains the standardized ratio by dividing the maximum time difference by the average delay value of the three nodes, thus establishing a standardized response difference sequence. The link stability determination submodule calls each segment of the ratio data in the standardized response difference sequence and compares it with the set link stability determination threshold. If the ratio of the node exceeds the link stability determination threshold, it is marked as a link node with a response offset trend. The node identifier and corresponding response anomaly information are recorded, and a node link instability identifier is generated.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by establishing a unified time series index table and implementing truncation, rearrangement, and amplitude filtering mechanisms, data rows with time misalignment and amplitude anomalies in the records are effectively eliminated, ensuring the synchronization and accuracy of monitoring data. By extracting polarity change points from the current slope derivative to construct an inflection point density sequence, disturbance sections can be segmented and located, and change trends can be marked. By introducing a disturbance intensity factor combined with the standardized results of voltage change rate, current harmonic distortion, and power factor, disturbance sections are identified, effectively eliminating misjudged areas. Multiple electrical parameter change characteristics within the section are extracted and risk level classification is implemented, achieving precise early warning of electrical disturbances. By introducing a multi-node response delay difference vector and evaluating the stability of node link response, the link fluctuation identification capability and overall early warning accuracy are improved. Attached Figure Description

[0015] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the electrical parameter processing module of the present invention; Figure 3 This is a flowchart of the disturbance feature recognition module of the present invention; Figure 4 This is a flowchart of the abnormal partition adjustment module of the present invention; Figure 5 This is a flowchart of the event level determination module of the present invention; Figure 6 This is a flowchart of the link response detection module of the present 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] Please see Figure 1 The present invention provides a technical solution, a remote monitoring system for electricity metering boxes based on the Internet of Things, the system including an electrical parameter processing module, a disturbance feature identification module, an abnormal partition adjustment module, an event level determination module, and a link response detection module; The electrical parameter processing module acquires real-time monitoring data from the power metering box. The real-time monitoring data includes the three-phase voltage RMS value sequence, the three-phase current RMS value sequence, the power factor sequence, and the voltage and current phase angle sequence. The real-time monitoring data is sent to the central monitoring equipment via the Internet of Things. The timestamp field in the sequence is called for continuous comparison. Data rows with abrupt time interval changes or misaligned sampling periods are deleted. The time axis of the data recorded by the sampling device is compared for difference. A unified time series index table is constructed based on the shortest sampling period. The sequence data is truncated and rearranged. The upper and lower limit boundary values ​​in each column of data are called for amplitude filtering. Erroneous records are deleted, and synchronous cleaned data results are generated. Based on the synchronous cleaning data results, the disturbance feature identification module selects the current slope value sequence of a certain phase in the three-phase current, performs numerical difference calculation between adjacent points to obtain the derivative sequence, extracts the points in the first derivative where the polarity direction changes as current inflection points, counts the number of inflection points in the time interval per second, constructs an inflection point density sequence, segments the inflection point density in time order and marks the trend of change. When the density continuously increases and exceeds the density change threshold, the corresponding segment is recorded as a disturbance candidate segment. If the inflection point density of subsequent segments shows a decreasing trend, the corresponding point is marked as a segment node and a segment index is generated to generate an inflection point segment structure. Inflection point density sequence is a sequence that measures the number of extreme points of the slope derivative of the current change per unit time, and is used to measure the complexity of fluctuations. The abnormal partitioning adjustment module filters the segment indexes continuously marked as disturbance candidate segments based on the inflection point segmentation structure, extracts the three-phase voltage change rate, total harmonic distortion of current and power factor sequence corresponding to each segment, performs Z-score standardization on the three indicators respectively, calculates the disturbance intensity factor of each segment, and compares it with the disturbance identification reference threshold. If it is less than the disturbance identification reference threshold, the corresponding segment is re-labeled as a stable segment. If it is greater than the threshold, the disturbance segment attribute is maintained, and a disturbance segment update label is generated. The total harmonic distortion rate of current refers to the ratio of the root mean square of each harmonic in the current to the root mean square of the fundamental frequency. It is often used to assess the intensity of harmonic distortion and represents the ratio of active power to apparent power. The event level determination module calls the retained disturbance sections in the disturbance section update tag, extracts the voltage peak-to-average power ratio, the maximum value of the total harmonic distortion rate of the current, the maximum current change rate, the instantaneous active power change amplitude and the power factor change range within the section, calculates the risk level assessment value of the disturbance section, and performs interval matching with the event level classification standard to classify the risk level of each section, implements risk warning, and generates graded disturbance warning records; The peak-to-average power ratio (PAPR) is the ratio of the peak value of a signal to its effective value. The power factor variation range is the difference between the maximum and minimum power factor within that range, and is often used to reflect power state fluctuations. The link response detection module calls the hierarchical disturbance warning record, obtains the alarm issuance command time and node status reporting time from the log, calculates the difference between the two to obtain the node's first response delay, retrieves the delay values ​​of two physically adjacent nodes, constructs three node response delay vectors and calculates the maximum time difference, performs normalization processing, obtains the standardized response difference index, compares the standardized response difference index with the response stability threshold, if it exceeds, it is determined that the node has a potential response degradation trend, and generates a node link instability identifier; The results of the synchronized cleaning data include time series synchronization information, abnormal data removal records, and effective ranges of electrical parameter amplitudes. The inflection point segmentation structure includes a segment index table, an inflection point density distribution matrix, and disturbance candidate segment labels. The disturbance segment update labels include disturbance segment status indicator codes, intensity factor comparison results, and segment adjustment mapping tables. The graded disturbance early warning records include an event level coding table, indicator interval matching sequences, and risk level mapping data. The node link instability identifiers include response delay time groups, adjacent node difference matrices, and stability judgment markers.

[0018] Please see Figure 2 The electrical parameter processing module includes: The data stream receiving submodule acquires real-time monitoring data from the power metering box. The real-time monitoring data includes three-phase voltage RMS value sequence, three-phase current RMS value sequence, power factor sequence, and voltage and current phase angle sequence. The real-time monitoring data is sent to the central monitoring equipment in real time through the Internet of Things transmission path. The timestamp, sampling number, and measurement point identifier of each sequence are recorded to establish a unified structure of the original receiving dataset of electrical parameters. Four sets of sampling sequences were continuously collected in the electricity metering box marked A01 from 08:00:00.000 to 08:00:00.060 on November 6, 2025. The effective values ​​of the three-phase voltages were [220.1V, 220.3V, 220.2V], [220.0V, 220.2V, 220.1V], and [219.9V, 220.1V, 220.0V], and the effective values ​​of the three-phase currents were [10.1A, 10.2A, 10.0A], [10.2A, 10.3A, 10.0A], and [10.2A, 10.3A, 10.0A]. The voltage and current phase angle sequences are [10.1A, 10.2A, 10.0A] and [10.1A, 10.2A, 10.0A], the power factor sequence is [0.95, 0.96, 0.95], and the voltage and current phase angle sequence is [18.19°, 16.26°, 18.19°]. The IoT transmission path sends the real-time monitoring data to the central monitoring device every 20ms. The central monitoring device records the timestamp of each data point, such as 2025-11-06 08:00:00.020, assigns a unique sampling number, such as SN1002, and marks the source measurement point identifier A01. Finally, a unified structure of the original received dataset of electrical parameters is established.

[0019] The time structure synchronization submodule calls the timestamp field in the original received data of electrical parameters, performs time interval continuity judgment on the sampled sequence, deletes data rows with abrupt changes or sample number jumps, compares the difference of each sequence time axis in the retained data, extracts the minimum sampling period as the time alignment benchmark for the whole time, constructs a unified time series index table, performs synchronous rearrangement on the original sequence according to the index table, and generates a time synchronization alignment data table. The time series [..., 2025-11-06 08:00:00.020, 2025-11-06 08:00:00.040, 2025-11-06 08:00:080,...] and the corresponding sampling number sequence [..., SN1002, SN1003, SN1005,...] are retrieved. A time interval continuity check is performed on this sampling sequence. By calculating the difference between adjacent timestamps, it is found that the time interval between 08:00:00.040 and 08:00:00.080 is 40ms. Simultaneously, the corresponding sampling number jumps from SN1003 to SN1005, indicating missing data. Therefore, the data row with sampling number SN1005 and timestamp 08:00:00.080 is deleted. For the remaining data, for example, the data row with timestamp 08:00:00.02... For the data rows at 0 and 08:00:00.040, the time axis of each sequence is compared again to calculate (08:00:00.040-08:00:00.020)=20ms. After traversing all consecutive data points, 20ms is confirmed as the minimum sampling period and used as the overall time alignment benchmark. A unified time series index table with a step size of 20ms is constructed, for example, [..., T0, T0+20ms, T0+40ms, ...]. Finally, based on this index table, all data rows in the original sequence are synchronously rearranged, placing each data point at the nearest index time point, generating a time synchronization aligned data table.

[0020] The amplitude cleaning and filtering submodule calls the three-phase voltage effective value column, three-phase current effective value column and power factor column in the time synchronization alignment data table, compares them with the set upper limit value of voltage, lower limit value of current and normal range of power factor, deletes data rows that exceed the boundary value or are zero, retains the valid sequence and uniformly reconstructs the index to generate the synchronous cleaning data result; The upper voltage limit is set at 253V. This value is based on the standard 220V single-phase voltage allowable deviation of -10% to +7%, taking into account instantaneous overvoltage and adding a 5% safety margin. The lower current limit is 0.5A, which is determined by statistical analysis of the current values ​​of equipment in standby or no-load conditions in historical data, taking the 99th percentile of its distribution. The normal power factor range is set at [0.85, 1.0], with the lower limit of 0.85 set according to the power factor requirements for high-voltage power supply users in the power supply and consumption business rules. The data table will then be used... The values ​​are calculated as follows: for example, if the effective value of phase A voltage in a certain row is 255.2V, it is compared with the upper limit of voltage 253V. Since 255.2V > 253V, the data row is deleted. Similarly, if the effective value of phase B current in a certain row is 0.2A, it is compared with the lower limit of current 0.5A. Since 0.2A < 0.5A, the data row is also deleted. The same operation is performed on records with values ​​below 0.85, such as 0.82, in the power factor column. Finally, all valid sequences within the boundary values ​​and not zero are retained, and their time indexes are reconstructed uniformly to generate the synchronous cleaned data results.

[0021] Please see Figure 3 The disturbance feature recognition module includes: The slope derivative calculation submodule, based on the synchronous cleaning data results, selects the current sequence of any one phase of the three-phase current, performs numerical difference on the amplitude of adjacent sampling points in the current sequence, calculates the current change between the current point and the previous time point and divides it by the sampling interval to construct the current slope sequence, performs numerical difference on the current slope sequence once, extracts the difference between two consecutive slope values ​​and records the derivative direction, and generates the first-order current change derivative sequence. The A-phase current sequence from the three-phase current is selected for processing. This sequence contains a current amplitude of I1 = 10.5A at time point t1 = 0.02s, and a current amplitude of I2 = 10.8A at the next adjacent sampling point t2 = 0.02s. First, numerical difference is performed on the amplitudes of these two adjacent sampling points to calculate the current change between the current point and the previous time point, ΔI = I2 - I1 = 10.8A - 10.5A = 0.3A. Then, this change is divided by the sampling interval Δt = t2 - t1 = 0.04s - 0.02s = 0.02s to calculate the current. At time t2, the current slope K2 = ΔI / Δt = 0.3A / 0.02s = 15A / s. This method is used to calculate the current slope point by point, constructing a complete current slope sequence, such as [..., 12A / s, 15A / s, 11A / s, ...]. Then, numerical difference is performed on this current slope sequence to extract the difference between two consecutive slope values. For example, calculating the difference between the slope K3 = 11A / s at time t2 and the next time t3 = 0.06s yields a derivative of 11A / s - 15A / s = -4A / s. 2And record that the direction of its derivative is negative, and so on, traversing the entire slope sequence to generate the first-order variation derivative sequence of the current.

[0022] The inflection point density extraction submodule calls the first-order derivative sequence of the current to determine whether the direction of the derivative between two adjacent points is reversed. The marked position is the polarity inflection point. The complete sampling period is divided into equal time intervals. The number of polarity inflection points per second is counted and arranged in chronological order. The change frequency value corresponding to the interval per second is constructed to generate the inflection point density sequence per unit time. The sequence is, for example, [..., 5, 8, -4, -7, 2, ...], where the unit is A / s. 2 The module checks whether the direction of the derivative between two adjacent points has reversed sign. Specifically, it checks whether the product of the current point and the previous point is negative. For example, between the sequence points 8 and -4, 8 × (-4) = 32, which is negative, indicating that the sign has reversed. This position is then marked as a polarity inflection point. Similarly, between -7 and 2, -7 × 2 = -14 is also marked as a polarity inflection point. Then, the complete sampling period, such as a monitoring period lasting 60 seconds, is divided into 60 equal time intervals, each interval lasting 1 second. The number of polarity inflection points marked in the first 1-second interval (from 0.00s to 1.00s) is counted. For example, if 25 are found, the change frequency value of this interval is 25. The same statistics are performed on the subsequent intervals per second to obtain a sequence arranged in chronological order, such as [25, 28, 30, 95, 150, 110, ...]. This sequence is the change frequency value corresponding to the interval per second, and finally, an inflection point density sequence per unit time is generated.

[0023] The paragraph index construction submodule performs time series trend judgment on the inflection point density value based on the inflection point density sequence within a unit time period, extracts the trend sequence of density value changes in multiple consecutive time segments, and for continuous upward segments in the trend, it judges whether the density value is greater than the set inflection point density change threshold. If it is satisfied, the corresponding segment is marked as a disturbance candidate segment. If a downward trend appears later, the inflection point is recorded as the paragraph boundary, and a paragraph separation index is constructed to generate an inflection point segmentation structure. A time series trend assessment is performed on the inflection point density values. This involves extracting the trend sequence of density value changes across multiple consecutive time intervals by comparing adjacent values. For example, a continuous upward trend occurs from 30 to 95 and then to 150, while a downward trend occurs from 150 to 110. For this continuous upward trend, it is determined whether the density value exceeds a set inflection point density mutation threshold. This threshold is based on statistical analysis of the inflection point density sequence under historical normal system operation, using its 90th percentile as the set value. The specific calculation process involves collecting 1000... The inflection point density value of the normal operating seconds is sorted and the 900th value is set to 80. Since 95 and 150 are both greater than the threshold of 80, the entire time period from the start of the density value to the peak value (e.g., from the 4th second to the 5th second) is marked as a disturbance candidate segment. When a downward trend occurs, that is, when the density value drops from 150 to 110, the time corresponding to the peak point of 150 (the end of the 5th second) is recorded as the boundary of the segment. A segment separation index containing the start time, end time and status identifier is constructed to generate the inflection point segment structure.

[0024] Please see Figure 4 The abnormal partition adjustment module includes: The disturbance index extraction submodule filters and marks the segment numbers as disturbance candidate segments based on the inflection point segment structure, extracts the three-phase voltage change rate, total harmonic distortion of current and power factor data corresponding to each segment, establishes the index sequence within three segments, and generates a multi-segment disturbance feature sequence set. The segment numbers marked as disturbance candidate segments are filtered out. For example, segment number 2 is selected, with a time range of 4 to 5 seconds. Based on this time range, the three-phase voltage change rate data, total harmonic distortion (THD) value data, and power factor data for the corresponding time period are extracted from the synchronously cleaned data results. Specifically, all effective voltage value data points between 4 and 5 seconds are extracted, and the difference between adjacent points is calculated and divided by the sampling interval to obtain the voltage change rate sequence, such as [0.5V / s, 1.2V / s, ..., 0.8V / s]. At the same time, all total harmonic distortion (THD) values ​​within this time period are extracted, forming a sequence [3.1%, 3.5%, ..., 4.2%], and the corresponding power factor sequence [0.92, 0.90, ..., 0.88]. The same extraction operation is performed on each segment marked as a disturbance candidate segment. Finally, the three index sequences extracted from all segments are combined to generate a multi-segment disturbance feature sequence set.

[0025] The normalization quantity construction submodule calls a multi-segment perturbation feature sequence set, performs Z-score standardization on the three indicators respectively, and extracts the sample average value in the standardized sequence using the formula: ; The perturbation intensity factor is obtained through calculation. Each segment of the perturbation intensity factor is compared with the set perturbation identification reference threshold and the comparison status is recorded to generate a perturbation threshold response result set. Among them, Y i Z represents the perturbation intensity factor of the i-th segment. i,U Z represents the average value of the voltage change rate of the i-th segment after Z-score normalization. i,H Z represents the average value of the total harmonic distortion rate of the current in the i-th segment after Z-score normalization. i,P θ represents the average power factor of the i-th segment after Z-score normalization. i The power factor volatility of the i-th segment is calculated by dividing the difference between the maximum and minimum values ​​in the power factor sequence of that segment by its average value. The disturbance identification reference threshold is set based on the 95th percentile of the disturbance intensity factor of the historical normal operation segment in the system. Using the global average and standard deviation of various indicators under normal operating conditions stored in the historical database, for example, the global average of the voltage change rate is 0.3V / s and the standard deviation is 0.2V / s, the value x in the i-th voltage change rate sequence is calculated using (x-0.3) / 0.2. Then, the sample mean of this standardized sequence is extracted to obtain Z. i,U Similarly, Z is obtained. i,H With Z i,P The disturbance intensity factor is then obtained through formula calculation. The formula constructs a comprehensive norm to fuse disturbance information from three different dimensions—voltage stability, current quality, and power factor quality—into a single scalar Y. i The square operation unifies the contribution direction of each indicator and amplifies the effect of deviations from the normal state. Taking the logarithm of the harmonic indicator is to smooth the impact of extreme harmonic shocks. Dividing the power factor indicator by its volatility plus one is to give higher weight to disturbances with drastic power factor fluctuations. Finally, the square root operation restores the dimensions. The disturbance identification reference threshold is set based on the disturbance intensity factor Y during the historical normal operation period of the system. i The 95th percentile, for example, Y-axis data collected from 10,000 normal segments. i The values ​​are sorted in ascending order, and the 9500th value is taken as 1.65. Therefore, the reference threshold is set to 1.65. In a specific example, a certain perturbation candidate segment... The normalized mean of the voltage change rate sequence is Z. i,U =1.8, the normalized mean of the total harmonic distortion (THD) sequence of the current is Z. i,H =2.5, the mean of the power factor series after standardization is Z i,P =2.2, the maximum value of the power factor sequence is 0.92, the minimum value is 0.88, and the average value is 0.90. θ is calculated as follows. i=(0.92-0.88) / 0.90≈0.044, substitute the value into the formula: ; The calculated disturbance intensity factor of 1.756 is compared with the set disturbance identification reference threshold of 1.65. Since 1.756 > 1.65, the comparison status of this segment is recorded as "disturbance," ultimately generating a disturbance threshold response result set. The advantage of this formula lies in eliminating the influence of dimensions by standardizing the three key power quality indicators—voltage, harmonics, and power factor—using Z-score. Furthermore, it utilizes a nonlinear function (logarithm) and dynamic weights (power factor volatility) to more accurately capture the intrinsic characteristics of different types of disturbances, enabling a single disturbance intensity factor to more comprehensively and robustly characterize the disturbance level of the power grid than any single indicator. This result indicates that the calculated disturbance intensity factor exceeds the upper limit of normal fluctuations, meaning that a noteworthy anomaly has indeed occurred in the power grid state during this period, requiring further detailed analysis.

[0026] The segment status correction submodule uses the comparison status information of each segment in the disturbance threshold response result set. If the disturbance intensity factor is less than the disturbance identification reference threshold, the corresponding segment is marked as a stable state. If it is greater than the threshold, the disturbance label is maintained. The status labels of all segments are reassigned to generate disturbance segment update labels. For a segment with a calculated disturbance intensity factor of 1.756, its disturbance label is maintained because its comparison status is "disturbance," which is greater than the disturbance identification reference threshold of 1.65. For another segment with a calculated disturbance intensity factor of 1.23, since this value is less than 1.65, its original "disturbance candidate" label is corrected to "stationary state." By traversing all segments and reallocating status labels according to their respective comparison results, the disturbance segment update label is generated.

[0027] Please see Figure 5 The event level determination module includes: The disturbance feature extraction submodule calls the disturbance segment index retained in the disturbance segment update label, extracts the voltage peak-to-average power ratio sequence, current total harmonic distortion rate sequence, current slope change value sequence, instantaneous active power sequence and power factor sequence corresponding to each segment, extracts the maximum value, difference or range of the five indicators respectively, establishes the five-dimensional disturbance feature vector corresponding to each segment, and generates a disturbance feature parameter set. The preserved disturbance segment index is invoked, for example, disturbance segment index 2. Within this segment (time from second 4 to second 5), the corresponding voltage peak-to-average power ratio (PAPR) sequence, current total harmonic distortion (THD) rate sequence, current slope change value sequence, instantaneous active power sequence, and power factor sequence are extracted. Then, key indicators are extracted from these five sequences. Specifically, the maximum value in the voltage PAPR sequence is calculated to be 1.48; the maximum value in the current THD rate sequence is extracted to be 5.2%; and the maximum absolute value in the current slope change value sequence (i.e., the current first-order derivative sequence) is extracted to be 85 A / s. 2 The difference between the maximum and minimum values ​​of the instantaneous active power sequence was calculated to obtain a fluctuation range of 3.1kW. The difference between the maximum and minimum values ​​of the power factor sequence was calculated to obtain a variation range of 0.04. Finally, these five index values ​​were combined to establish a five-dimensional disturbance feature vector [1.48,5.2,85,3.1,0.04] corresponding to the disturbance segment, generating a disturbance feature parameter set.

[0028] The risk assessment calculation submodule calls upon five disturbance indicators from each segment of the disturbance characteristic parameter group to extract the values ​​of voltage peak-to-average power ratio, total harmonic distortion rate of current, rate of change of current, power change amplitude, and power factor fluctuation range, using the following formula: ; The risk level assessment value of the disturbance segment is obtained through calculation, and a disturbance risk assessment sequence is generated; Among them, R j A represents the disturbance risk level assessment value for segment j. j The peak-to-average power ratio (PAPR) is calculated by dividing the maximum voltage amplitude of this segment by the effective value, μ. A B represents the mean of the peak-to-average power ratio (PAPR) of voltage across the global sample. j B is the maximum value of the total harmonic distortion rate of the current. ref The permissible standard value for current distortion set for the system, μ B C is the mean of the total harmonic distortion rate of the current. j The maximum rate of change of the slope of the current, Δt j This is the normalized value of the sampling duration of the j-th segment, obtained by dividing the original sampling duration by the sampling duration of the longest disturbance segment within the monitoring period, μ. C D is the mean of the rate of change of the current slope. j The instantaneous active power fluctuation amplitude is derived from the difference between the maximum and minimum instantaneous power within this segment, μ. D E is the global average of the instantaneous active power fluctuation amplitude. j The power factor range is calculated as the difference between the maximum and minimum power factor within that range, in μ. ETo stabilize the average power factor variation during operation, the threshold range for event level classification is set based on the median of event distribution collected by the system monitoring platform in historical operation data. The disturbance intensity scores corresponding to known safe, attention, warning, severe, and critical states are divided into five level ranges according to the quantile method, and a continuously segmented level mapping reference structure is constructed. For the aforementioned vector [1.48, 5.2, 85, 3.1, 0.04], extract the peak-to-average power ratio (PAPR) A. j =1.48, the maximum value of the total harmonic distortion rate of the current B j =5.2, maximum slope change rate C of current j =85, instantaneous active power fluctuation amplitude D j =3.1, and the power factor variation range E j =0.04, and the risk level assessment value of the disturbance segment is obtained by formula calculation. The logic of this formula is to normalize, quantify and aggregate five risk-related indicators with different physical meanings, assess the degree of deviation by comparing them with their respective means or reference values, and adjust the sensitivity of each indicator using different function forms such as square, absolute value, and logarithm. Finally, a comprehensive risk score is obtained through a root mean square-like form. The threshold range for event level classification is set based on the R corresponding to events (safe, attention, warning, serious, critical) that have been marked with levels in the historical database. j Statistical analysis is performed on the values, and intervals are divided using the percentile method. For example, [0, 1.5) is safe, [1.5, 2.5) is of concern, [2.5, 4.0) is a warning, [4.0, 6.0) is serious, and ≥6.0 is critical. A continuously segmented level mapping reference structure is constructed. In a specific example, the values ​​of each parameter are as follows: A j =1.48, μ A The historical average is 1.42; B j =5.2, B ref Based on the national standard, it is set to 5.0, μ B The historical average is 3.0; C j =85, △t j Dividing the duration of this disturbance segment (1s) by the longest disturbance segment duration (5s) within the monitoring period yields 0.2, μ. C The historical average is 50; D j =3.1, μ D The historical average is 1.5; E j =0.04, μ E Assuming the mean value of the stable segment is 0.02, substitute it into the formula to calculate: ; Generate a disturbance risk assessment sequence containing this risk value. The advantage of this formula is that it considers not only the instantaneous amplitude of each disturbance characteristic, but also the disturbance duration Δt.j and industry standard B ref By comparing the results with historical averages, a dynamic and multi-dimensional assessment of the severity of the disturbance event was achieved, making the risk assessment results more accurate and practically instructive. The results indicate that the comprehensive risk score for this disturbance event is 1.509. According to the preset risk level classification, this value falls into the "concern" level, suggesting that the root cause of the event needs to be traced, but it does not yet pose a serious threat.

[0029] Table 1: Classification of Risk Assessment Levels for Historical Events ; As shown in Table 1, this table lists the five risk levels defined by the system and their corresponding risk assessment values ​​(R). j The intervals are used to map the calculated risk values ​​to specific risk levels.

[0030] The risk level mapping submodule retrieves the level division threshold range based on the risk value corresponding to each segment in the disturbance risk assessment sequence, performs range position judgment, maps the risk value to the corresponding risk level range number, records the mapping result between the event level and the disturbance segment, and generates a graded disturbance early warning record. Based on the calculated risk value of 1.509, the threshold range for classifying risk levels is retrieved, as shown in Table 1. The range position is determined by comparing the risk value of 1.509 with the ranges [0, 1.5), [1.5, 2.5), [2.5, 4.0), [4.0, 6.0), and ≥6.0. Since 1.5≤1.509<2.5, the value is mapped to the corresponding risk level range number 2, i.e., the "attention" level. The event level is recorded as "attention" and the unique identifier of the disturbance segment is recorded, such as its start time and measurement point identifier. Finally, a graded disturbance warning record is generated.

[0031] Please see Figure 6 The link response detection module includes: The response delay extraction submodule calls the node identifier of each node with a risk level in the graded disturbance early warning record, obtains the alarm issuance command time and the first reporting time of the node status, calculates the time difference between the two as the target node response delay value, records the time difference sequence and the corresponding node number information, and generates a node response delay record table. The module calls upon the node identifier with a risk level. For example, after the node with the monitoring point identifier A01 is marked as "attention level", the central monitoring equipment issues an alarm command at time T1=2025-11-06 08:05:10.200. The first time the node's status (e.g., the status after confirming receipt of the command or executing the corresponding action) is reported to the central monitoring equipment is at time T2=2025-11-06 08:05:10.550. The module calculates the time difference between these two as the response delay value of the target node, i.e., △T=T2-T1=350ms, and records the time difference sequence value of 350ms and the corresponding node number information A01, generating a node response delay record table.

[0032] The adjacency difference construction submodule retrieves the response delay values ​​of the immediate preceding and following nodes in the physical topology for each target node based on the node response delay record table, constructs a vector including the response delays of the three nodes, performs the maximum time difference calculation between the pairwise combinations, and obtains the standardized ratio by dividing the maximum time difference by the average delay value of the three nodes, thus establishing a standardized response difference sequence. For target node A01, retrieve the response delay values ​​of its immediate preceding node A00 and immediately following node A02 in the physical topology. Assume the response delay of node A00 is 330ms and the response delay of node A02 is 360ms, as found in the record table. Construct a vector () containing the response delays of these three nodes, in milliseconds. Then, calculate the maximum time difference between each pair of elements in the vector, specifically calculating |350-330|=20. ms, |360-350|=10ms, |360-330|=30ms, take the maximum value of 30ms, then calculate the average delay of the three nodes, i.e. (330+350+360) / 3=346.67ms. Finally, divide the maximum time difference by the average delay of the three nodes to obtain the standardized ratio, i.e. 30ms / 346.67ms≈0.0865, and finally establish a standardized response difference sequence containing this ratio.

[0033] The link stability determination submodule calls each segment of the ratio data in the standardized response difference sequence and compares it with the set link stability determination threshold. If the ratio of the node exceeds the link stability determination threshold, it is marked as a link node with a response offset trend. The node identifier and corresponding response anomaly information are recorded, and a node link instability identifier is generated. The ratio 0.0865 calculated above is compared with the set link stability judgment threshold. This threshold is set based on statistical analysis of a large number of standardized response difference sequences under normal historical communication conditions, removing extreme outliers and calculating its 98th percentile, for example, set to 0.20. Since the ratio 0.0865 corresponding to node A01 does not exceed the link stability judgment threshold of 0.20, it is not marked. If the ratio calculated for another node B05 is 0.28, because it exceeds 0.20, this node is marked as a link node with a response offset trend, and the node identifier B05 and the corresponding response anomaly information are recorded, such as its standardized response difference of 0.28, ultimately generating a node link instability identifier.

[0034] 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 remote monitoring system for electricity metering boxes based on the Internet of Things, characterized in that, The system includes: The electrical parameter processing module acquires real-time monitoring data from the power metering box, sends the real-time monitoring data to the central monitoring equipment via the Internet of Things, performs truncation and rearrangement on the sequence data, deletes erroneous records, and generates synchronized cleaned data results. Based on the synchronous cleaning data results, the disturbance feature recognition module extracts the current inflection point. When the inflection point density increases continuously and exceeds the density change threshold, the corresponding segment is recorded as a disturbance candidate segment. If the inflection point density of subsequent segments shows a decreasing trend, it is marked as a segmentation node, and an inflection point segmentation structure is generated. The abnormal partitioning adjustment module calculates the disturbance intensity factor of each segment based on the inflection point segmentation structure, compares it with the disturbance identification reference threshold, and if it is less than the disturbance identification reference threshold, the corresponding segment is relabeled as a stable segment; if it is greater than the threshold, the disturbance segment attribute is maintained, and a disturbance segment update label is generated. The event level determination module calls the updated label of the disturbance segment, calculates the risk level assessment value of the disturbance segment, and performs interval matching with the event level classification standard to classify the risk level of each segment, implements risk warning, and generates graded disturbance warning records.

2. The IoT-based remote monitoring system for electricity metering boxes according to claim 1, characterized in that, The synchronous cleaning data results include time series synchronization information, abnormal data removal records, and valid ranges of electrical parameter amplitudes. The inflection point segmentation structure includes a segment index table, an inflection point density distribution matrix, and disturbance candidate segment labels. The disturbance segment update label includes a disturbance segment status indicator code, intensity factor comparison results, and a segment adjustment mapping table. The graded disturbance early warning record includes an event level coding table, an indicator interval matching sequence, and risk level mapping data.

3. The remote monitoring system for electricity metering boxes based on the Internet of Things according to claim 2, characterized in that, The electrical parameter processing module includes: The data stream receiving submodule acquires real-time monitoring data from the power metering box. The real-time monitoring data includes three-phase voltage RMS value sequence, three-phase current RMS value sequence, power factor sequence, and voltage and current phase angle sequence. The real-time monitoring data is sent to the central monitoring equipment in real time through the Internet of Things transmission path. The timestamp, sampling number, and measurement point identifier of each sequence are recorded to establish a unified structure of the original receiving dataset of electrical parameters. The time structure synchronization submodule calls the timestamp field in the original electrical parameter received dataset, performs time interval continuity judgment on the sampled sequence, deletes data rows with abrupt changes or sample number jumps, performs difference comparison on the time axis of each sequence in the retained data, extracts the minimum sampling period as the overall time alignment benchmark, constructs a unified time series index table, performs synchronous rearrangement on the original sequence according to the index table, and generates a time synchronization alignment data table. The amplitude cleaning and filtering submodule calls the three-phase voltage effective value column, three-phase current effective value column and power factor column in the time synchronization alignment data table, compares them with the set upper limit value of voltage, lower limit value of current and normal range of power factor, deletes data rows that exceed the boundary value or are zero, retains the valid sequence and reconstructs the index in a unified manner, and generates the synchronization cleaning data result.

4. The IoT-based remote monitoring system for electricity metering boxes according to claim 3, characterized in that, The disturbance feature recognition module includes: Based on the synchronous cleaning data results, the slope derivative calculation submodule selects the current sequence of any one phase of the three-phase current, performs numerical difference on the amplitude of adjacent sampling points in the current sequence, calculates the current change between the current point and the previous time point and divides it by the sampling interval to construct the current slope sequence, performs numerical difference on the current slope sequence once, extracts the difference between two consecutive slope values ​​and records the derivative direction to generate the first-order current change derivative sequence. The inflection point density extraction submodule calls the current first-order change derivative sequence, determines whether the derivative direction between two adjacent points has reversed sign, marks the position as a polarity inflection point, divides the complete sampling period into equal time intervals, counts the number of polarity inflection points per second and arranges them in time order, constructs the change frequency value corresponding to the interval per second, and generates an inflection point density sequence per unit time. The paragraph index construction submodule performs time series trend judgment on the inflection point density value based on the inflection point density sequence within the unit time period, extracts the trend sequence of density value changes in multiple consecutive time segments, and for continuous upward segments in the trend, it judges whether the density value is greater than the set inflection point density mutation threshold. If it is satisfied, the corresponding segment is marked as a disturbance candidate segment. If a downward trend appears later, the inflection point is recorded as the paragraph boundary, and a paragraph separation index is constructed to generate an inflection point segmentation structure.

5. The IoT-based remote monitoring system for electricity metering boxes according to claim 4, characterized in that, The abnormal partition adjustment module includes: The disturbance index extraction submodule filters and marks the segment numbers as disturbance candidate segments according to the inflection point segmentation structure, extracts the three-phase voltage change rate, total harmonic distortion of current and power factor data corresponding to each segment, establishes three segment index sequences respectively, and generates a multi-segment disturbance feature sequence set. The normalization quantity construction submodule calls the multi-segment perturbation feature sequence set, performs Z-score standardization on the three indicators respectively, extracts the average value of the samples in the standardized sequence, calculates the perturbation intensity factor, compares the perturbation intensity factor of each segment with the set perturbation identification reference threshold and records the comparison status, and generates a perturbation threshold response result set. The segment status correction submodule, based on the comparison status information of each segment in the disturbance threshold response result set, marks the corresponding segment as stable if the disturbance intensity factor is less than the disturbance identification reference threshold, and if it is greater than the threshold, maintains the disturbance label, reassigns status labels to all segments, and generates a disturbance segment update label.

6. The IoT-based remote monitoring system for electricity metering boxes according to claim 5, characterized in that, The specific formula for obtaining the disturbance intensity factor is as follows: ； Among them, Y i Z represents the perturbation intensity factor of the i-th segment. i,U Z represents the average value of the voltage change rate of the i-th segment after Z-score normalization. i,H Z represents the average value of the total harmonic distortion rate of the current in the i-th segment after Z-score normalization. i,P θ represents the average power factor of the i-th segment after Z-score normalization. i This represents the power factor volatility of the i-th segment.

7. The IoT-based remote monitoring system for electricity metering boxes according to claim 6, characterized in that, The event level determination module includes: The disturbance feature extraction submodule calls the disturbance segment index retained in the disturbance segment update label, extracts the voltage peak-to-average power ratio sequence, current total harmonic distortion rate sequence, current slope change value sequence, instantaneous active power sequence and power factor sequence corresponding to each segment, extracts the maximum value, difference or range of the five indicators respectively, establishes the five-dimensional disturbance feature vector corresponding to each segment, and generates a disturbance feature parameter group; The risk assessment calculation submodule calls the five disturbance index quantities of each segment in the disturbance characteristic parameter group, extracts the values ​​of voltage peak-to-average power ratio, current total harmonic distortion rate, current change rate, power change amplitude and power factor fluctuation range, calculates and obtains the risk level assessment value of the disturbance segment, and generates a disturbance risk assessment sequence. The risk level mapping submodule retrieves the level division threshold range based on the risk value corresponding to each segment in the disturbance risk assessment sequence, performs range position judgment, maps the risk value to the corresponding risk level range number, records the mapping result between the event level and the disturbance segment, and generates a graded disturbance early warning record.

8. The IoT-based remote monitoring system for electricity metering boxes according to claim 7, characterized in that, The system also includes: The link response detection module calls the hierarchical disturbance warning record, obtains the alarm issuance command time and node status reporting time from the log, retrieves the delay values ​​of two physically adjacent nodes, constructs three node response delay vectors and calculates the maximum time difference, performs normalization processing, obtains the standardized response difference index, compares the standardized response difference index with the response stability threshold, if it exceeds the threshold, it determines that the node has a potential response degradation trend and generates a node link instability identifier. The node link instability identifier includes a response delay time group, an adjacent node difference matrix, and a stability determination flag.

9. The IoT-based remote monitoring system for electricity metering boxes according to claim 8, characterized in that, The link response detection module includes: The response delay extraction submodule calls the node identifier of each node with a risk level in the graded disturbance early warning record, obtains the alarm issuance command time and the first reporting time of the node status, calculates the time difference between the two as the target node response delay value, and records the time difference sequence and the corresponding node number information to generate a node response delay record table. The adjacency difference construction submodule retrieves the response delay values ​​of the immediate preceding and following nodes in the physical topology for each target node based on the node response delay record table, constructs a vector including the response delays of the three nodes, performs the maximum time difference calculation between the pairwise combinations, and obtains the standardized ratio by dividing the maximum time difference by the average delay value of the three nodes, thus establishing a standardized response difference sequence. The link stability determination submodule calls each segment of the ratio data in the standardized response difference sequence and compares it with the set link stability determination threshold. If the ratio of the node exceeds the link stability determination threshold, it is marked as a link node with a response offset trend. The node identifier and corresponding response anomaly information are recorded, and a node link instability identifier is generated.

Images