A data analysis system and method based on HPLC dual-mode communication

Abstract

This invention discloses a data analysis system and method based on HPLC dual-mode communication, relating to the field of anomaly tracing technology. The invention sets a hardware timestamp in the communication terminal device to collect latency data during dual-channel data transmission in dual-mode communication, and standardizes the latency data. Anomalies in dual-mode communication are classified into four main types. Working data from dual-mode communication is collected for each of these four main anomalies to construct a four-dimensional dynamic baseline model. During dual-mode communication, characteristic indicators are collected in real time and input into the four-dimensional dynamic baseline model, outputting the real-time anomaly type and issuing warnings. Different tracing strategies are set for the four main anomalies. The tracing strategy is selected based on the real-time anomaly type to find the root cause of the anomaly. The real-time anomaly type and the root cause data from the tracing strategy are integrated and visualized. The baseline in the four-dimensional dynamic baseline model is calibrated and updated using the root cause data from the tracing strategy.

Description

Technical Field

[0001] This invention relates to the field of anomaly tracing technology, specifically to a data analysis system and method based on HPLC dual-mode communication. Background Technology

[0002] The digitalization of low-voltage power grids is crucial for ensuring reliable power supply and accommodating renewable resources. This requires the power grid to achieve more accurate electricity consumption information collection and more efficient power distribution management, necessitating reliable communication technologies to support data transmission. With the increasing prevalence of technologies such as the Internet of Things (IoT), smart homes, smart meters, and remote monitoring in daily life, these applications all require data transmission. High-speed power line carrier (HPLC) technology, as a communication technology that utilizes power lines to transmit signals, has become one of the mainstream technologies for local communication in the IoT due to its advantages such as no wiring required and ease of deployment. However, HPLC technology alone cannot achieve data transmission in wireless space, thus HPLC dual-mode communication technology has emerged.

[0003] However, new problems inevitably arise during the operation of HPLC dual-mode communication technology. Due to the increase in data transmission channels, when communication anomalies occur under dual-channel communication, it is impossible to accurately determine the type of anomaly. Furthermore, the difficulty of tracing the source of anomalies also increases with the increase in communication channels. Therefore, it is crucial to utilize the characteristics of time delay data in dual-channel communication for anomaly analysis and judgment, and to leverage the advantages of dual channels to overcome the burden brought by dual channels. Summary of the Invention

[0004] The purpose of this invention is to provide a data analysis system and method based on HPLC dual-mode communication to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A data analysis method based on HPLC dual-mode communication, the method comprising the following steps:

[0007] S100. Set a hardware timestamp in the communication terminal device to collect latency data during dual-channel data transmission in dual-mode communication and standardize the latency data. The hardware timestamp avoids software timing errors and ensures that the latency data accuracy reaches the nanosecond level, laying the foundation for subsequent anomaly judgment.

[0008] Furthermore, the specific steps for standardizing latency data are as follows:

[0009] S101. Set a hardware timestamp in the communication terminal device to obtain the data transmission time T during communication. sendWhen using a concentrator to collect data for dual-mode communication, the data arrival time during data transmission in both channels is denoted as T. plc T represents the arrival time of power channel data. rf Indicates the arrival time of the wireless channel;

[0010] S102. Calculate the delay data of the two channels using the transmission time and arrival time respectively. The formula is:

[0011] ;

[0012] In the formula, D plc D represents the time delay data of the power channel. rf Represents the latency data of the wireless channel;

[0013] Simultaneously, power and wireless channel latency are collected, and differential analysis can identify coordination anomalies between channels (such as clock synchronization failure).

[0014] S103. Standardize the collected dual-channel latency data. Eliminate the differences in latency data dimensions between different terminal devices and network environments, making the data comparable and improving the model's generalization ability.

[0015] S200. Classify the anomalies in dual-mode communication into four main anomalies, collect working data in dual-mode communication for the four main anomalies, and construct a four-dimensional dynamic baseline model.

[0016] Furthermore, the specific steps for constructing a four-dimensional dynamic baseline model are as follows:

[0017] S201. Anomalies in dual-mode communication are classified into four main types: power channel interference, wireless channel obstruction, topology link anomalies, and terminal clock synchronization failure. For the power channel interference anomaly type, the characteristic index is specified as α=D. plc The threshold parameter is calculated using historical values ​​of the characteristic indicator. The threshold parameter includes the average characteristic indicator value and the standard deviation. The formula is as follows:

[0018] ;

[0019] In the formula, μ α (t) represents the average characteristic value after applying an exponential moving average to the characteristic indicator, 0.2 represents the smoothing factor, and EMA represents the exponential moving average; StdDev represents the standard deviation of the calculated characteristic index over a historical 24 hours;

[0020] It adopts a combination of exponential moving average (EMA) and standard deviation to adapt to changes in the network environment in real time;

[0021] The threshold judgment mechanism is generated using threshold parameters, specifically as follows: When the characteristic index α in dual-mode communication is greater than the average characteristic index value plus three times the standard deviation, it is judged to be an abnormal type of power channel interference.

[0022] S202. For wireless channel obstruction types, the specified characteristic index is β=D. rf Threshold parameters are calculated using historical values ​​of the feature index, including the average value of the feature index β (μ). β (t) and standard deviation A threshold judgment mechanism is constructed using the characteristic index β-mean and standard deviation, specifically as follows: When the characteristic index β in dual-mode communication is greater than the average characteristic index value plus four times the standard deviation, it is judged as a wireless channel obstruction anomaly.

[0023] Power channel interference is assessed using a threshold of 3 times the standard deviation, while wireless obstruction is assessed using a threshold of 4 times the standard deviation. This matches the noise characteristics of different channels and reduces the false alarm rate.

[0024] S203. For topology link anomaly types, the specified characteristic index is γ = |△D_meas - △D_theory|, where γ represents the time delay difference deviation, △D_meas represents the actual measured time delay difference, and △D_theory represents the theoretical time delay difference. The actual measured time delay difference is calculated as follows: △D_meas = |D plc -D rf |;

[0025] The theoretical time delay difference is calculated as follows: In the formula, L represents the length of the conductor path, S represents the straight-line distance in space, and V... HPLC V represents the HPLC signal rate. RF Indicates the speed of the wireless signal;

[0026] Staff collect historical latency differences from the GIS topology database and calculate the average value as the latency difference threshold In_D. They then construct a threshold judgment mechanism, specifically: γ>In_D; when the characteristic index γ in dual-mode communication is greater than the latency difference threshold, it is judged as an abnormal type of topology link.

[0027] S204. For terminal clock synchronization failure type, the specified characteristic index is δ=|D plc -D rf |, δ represents the delay difference, and the calibration value in the Precision Time Protocol (PTP) is extracted as the out-of-synchronization threshold In_B; a threshold judgment mechanism is constructed, specifically: δ>In_B; when the characteristic index δ in dual-mode communication is greater than the out-of-synchronization threshold In_B, it is judged as a terminal clock out-of-synchronization anomaly.

[0028] By using the time delay difference deviation (γ) and the PTP calibration value (δ), joint monitoring of physical topology and time synchronization is achieved.

[0029] S205. Construct a four-dimensional dynamic baseline model based on the characteristic indicators and threshold judgment mechanism of the four main anomalies.

[0030] S300: During dual-mode communication, feature indicators are collected in real time, input into the four-dimensional dynamic baseline model, output the real-time anomaly type and issue an early warning.

[0031] Furthermore, the specific steps for outputting real-time anomaly types and issuing early warnings are as follows:

[0032] S301. Real-time acquisition and calculation of the characteristic indicators of four main anomalies in dual-mode communication, obtaining real-time characteristic indicator values, inputting the real-time characteristic indicator values ​​of the four main anomalies into the four-dimensional dynamic baseline model, and outputting the corresponding anomaly type when the threshold judgment mechanism in the four-dimensional dynamic baseline model is satisfied.

[0033] S302. When an anomaly is detected in dual-mode communication, calculate the deviation of the real-time characteristic index from the threshold in the threshold judgment mechanism, and calculate the anomaly confidence level of the judgment. The formula is as follows:

[0034] ;

[0035] In the formula, Con represents the anomaly confidence level, and Id... s Id represents the real-time feature index value. y The threshold value represents the characteristic index threshold in the threshold judgment mechanism; un represents the fluctuation range of the baseline in the threshold judgment mechanism. In the three main anomalies of power channel interference, wireless channel obstruction and topology link anomaly, the baseline is the average value of the corresponding characteristic index, and the fluctuation range is the standard deviation; in the main anomaly of terminal clock out of synchronization, the baseline is the PTP calibration value, and the fluctuation range is the average value of the historical PTP fluctuation amplitude.

[0036] S303. When the output exception type is not unique, determine the confidence level of different exception types, set the priority according to the confidence level from large to small, set the exception type with the highest confidence level as the highest priority, and then set the priority in order; issue warnings according to the priority from large to small.

[0037] When multiple anomalies are triggered simultaneously, priority is determined by ranking them by confidence level to avoid confusion in early warnings.

[0038] S400. Set different tracing strategies for the four main types of anomalies; select the tracing strategy according to the real-time anomaly type to find the root cause of the anomaly.

[0039] Furthermore, the specific steps for selecting a tracing strategy to find the root cause of an anomaly based on the real-time anomaly type are as follows:

[0040] S401. Set source tracing strategies for the four main anomalies respectively. The source tracing strategy for power channel interference is as follows: collect the terminal locations of all warnings in dual-mode communication, classify the collected terminal locations using a spatial clustering algorithm, collect the voltage harmonic distortion rate of all communication terminals when a warning is issued, set a harmonic distortion rate threshold, and when the voltage harmonic distortion rate of a communication terminal exceeds the harmonic distortion rate threshold, it is judged as a noise source, and the spatial location of the noise source is obtained through the classified terminal locations.

[0041] S402. The specific source tracing strategy for wireless channel obstruction is as follows: Calculate the wireless communication path loss at different transmission distances in a spatial location using a log-normal shadowing fading model, collect the initial signal strength of the wireless signal, and determine that the theoretical signal strength = initial signal strength - path loss; measure the signal strength at the corresponding location in real time, and determine that the corresponding location is an abnormal source when the real-time signal strength is less than the theoretical signal strength.

[0042] S403. The specific strategy for tracing the abnormal topology link is as follows: A sweep frequency signal is sent to the terminal through the power channel, and the input impedance Z(f) = R(f) + jX(f) is measured in real time, where R(f) represents the real part, which is the value of resistance as a function of frequency, jX(f) represents the imaginary part, which is the value of reactance as a function of frequency, and f represents the frequency of the sweep frequency signal; the theoretical impedance is extracted from the power channel topology diagram, and the correlation coefficient between the measured impedance and the theoretical impedance is calculated. When the correlation coefficient is greater than Zy, it is determined that the power channel link corresponding to the sweep frequency signal receiving terminal is abnormal; Zy represents the manually preset correlation coefficient threshold.

[0043] S404. The specific strategy for tracing the cause of terminal clock synchronization failure is as follows: Collect the lowest latency value in history, and use the lowest latency value to calculate the latency offset of all terminals under the PTP protocol concentrator. The formula is: In the formula, △t represents the time delay offset, and BaseRTT represents the time delay value from the hour.

[0044] Within the time window, the delay offset is repeatedly calculated based on the time series. When the delay offset of all terminals under the concentrator increases synchronously, the concentrator is judged to be abnormal; when the delay offset of only a single terminal increases, the corresponding single terminal is judged to be abnormal.

[0045] S405. Select the corresponding source tracing strategy according to the type of abnormality in the warning, and use the source tracing strategy to locate the root cause of each type of abnormality in the warning.

[0046] S500 integrates real-time anomaly types and root cause data from tracing strategies for visualization.

[0047] S600: Use the root source data in the tracing strategy to calibrate and update the baseline in the four-dimensional dynamic baseline model.

[0048] Furthermore, the specific steps for calibrating and updating the baseline in the four-dimensional dynamic baseline model using the root cause data from the source tracing strategy are as follows:

[0049] S601. After identifying the root cause of each type of early warning anomaly, professionals maintain the root cause, collect the normal characteristic indicators after maintenance, and use the maintained normal characteristic indicators to update all baseline values ​​in the four-dimensional dynamic baseline model. The formula is as follows:

[0050] ;

[0051] In the formula, J new J represents the updated baseline value. old x represents the old baseline value before the update. normal This represents the normal feature index after maintenance, and r represents the learning rate weight.

[0052] After each fault is resolved, a new baseline is automatically learned, and the model accuracy improves with usage time.

[0053] The four-dimensional dynamic baseline model is updated and optimized using the updated baseline values.

[0054] A data analysis system based on HPLC dual-mode communication includes a data acquisition module, a model building module, an anomaly detection module, a source tracing module, a visualization module, and an optimization module.

[0055] The data acquisition module is used to set a hardware timestamp in the communication terminal device, collect latency data during dual-channel data transmission in dual-mode communication, and standardize the latency data.

[0056] The model building module is used to classify anomalies in dual-mode communication into four main anomalies, collect working data in dual-mode communication for the four main anomalies, and build a four-dimensional dynamic baseline model.

[0057] The anomaly detection module is used to collect feature indicators in real time during dual-mode communication, input the feature indicators into the four-dimensional dynamic baseline model, output the real-time anomaly type and issue an early warning.

[0058] The source tracing module is used to set different source tracing strategies for four main types of anomalies; and to select the source tracing strategy according to the real-time anomaly type to find the root cause of the anomaly.

[0059] The visualization module is used to integrate real-time anomaly types and root cause data in the tracing strategy and display them visually.

[0060] The optimization module is used to calibrate and update the baseline in the four-dimensional dynamic baseline model using the root cause data in the tracing strategy.

[0061] The model building module includes a power channel interference unit, a wireless channel obstruction unit, a topology link anomaly unit, and a terminal clock synchronization failure unit;

[0062] The power channel interference unit is used to specify the delay data of the power channel as a feature index and calculate threshold parameters to construct a threshold judgment mechanism.

[0063] The wireless channel blocking unit is used to specify the latency data of the wireless channel as a feature indicator and calculate threshold parameters to construct a threshold judgment mechanism.

[0064] The topology link anomaly unit is used to calculate the time delay difference deviation as a feature index and to calculate the time delay difference threshold to construct a threshold judgment mechanism.

[0065] The terminal clock out-of-sync unit is used to calculate the time delay difference as a characteristic indicator, calculate the out-of-sync threshold, and construct a threshold judgment mechanism.

[0066] The anomaly detection module includes an anomaly type output unit and a priority warning unit;

[0067] The anomaly type output unit is used to input the real-time feature index values ​​of the four main anomalies into the four-dimensional dynamic baseline model. When the threshold judgment mechanism in the four-dimensional dynamic baseline model is satisfied, the corresponding anomaly type is output.

[0068] The priority warning unit is used to calculate the confidence level of different anomaly types, set the priority according to the confidence level from large to small, and issue warnings in order of priority from large to small.

[0069] The tracing module includes a power channel interference tracing unit, a wireless channel obstruction tracing unit, a topology link anomaly tracing unit, and a terminal clock synchronization failure tracing unit;

[0070] The power channel interference tracing unit is used to set a harmonic distortion rate threshold. When the voltage harmonic distortion rate of the communication terminal exceeds the harmonic distortion rate threshold, it is judged as a noise source, and the spatial location of the noise source is obtained through the classified terminal location.

[0071] The wireless channel obstruction tracing unit is used to measure the signal strength at the corresponding location in real time. When the real-time signal strength is less than the theoretical signal strength, the corresponding location is determined to be an abnormal source.

[0072] The topology link anomaly tracing unit is used to extract the theoretical impedance in the power channel topology map, calculate the correlation coefficient between the measured impedance and the theoretical impedance, and determine that the power channel link corresponding to the frequency sweep signal receiving terminal is abnormal when the correlation coefficient is greater than the correlation coefficient threshold.

[0073] The terminal clock out-of-synchronization tracing unit is used to determine that the concentrator is abnormal when the delay offset of all terminals under the concentrator increases synchronously; and to determine that the corresponding individual terminal is abnormal when only the delay offset of a single terminal increases.

[0074] Compared with the prior art, the beneficial effects of the present invention are:

[0075] 1. This invention transforms the transmission differences between the two channels into diagnostic advantages, and achieves accurate separation and localization of anomaly types by modeling the spatial distribution characteristics of delay differences.

[0076] 2. This invention constructs a four-dimensional dynamic baseline model to achieve real-time judgment of four major anomalies and accurately obtain the different types of real-time anomalies; and formulates different source tracing strategies for different anomaly types, which can achieve accurate anomaly location in dual-mode communication.

[0077] 3. This invention updates the baseline of the four-dimensional dynamic baseline model by maintaining the normal feature indicators, thus ensuring the timeliness and generalization of the model. Attached Figure Description

[0078] Figure 1 This is a module distribution diagram of a data analysis system based on HPLC dual-mode communication according to the present invention;

[0079] Figure 2 This is a schematic diagram illustrating the steps of a data analysis method based on HPLC dual-mode communication according to the present invention. Detailed Implementation

[0080] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0081] Example: Figures 1-2 As shown, the present invention provides a technical solution.

[0082] A data analysis method based on HPLC dual-mode communication, the method comprising the following steps:

[0083] S100. Set a hardware timestamp in the communication terminal device to collect latency data during dual-channel data transmission in dual-mode communication and standardize the latency data. The hardware timestamp avoids software timing errors and ensures that the latency data accuracy reaches the nanosecond level, laying the foundation for subsequent anomaly judgment.

[0084] The specific steps for standardizing latency data are as follows:

[0085] S101. Set a hardware timestamp in the communication terminal device to obtain the data transmission time T during communication. send When using a concentrator to collect data for dual-mode communication, the data arrival time during data transmission in both channels is denoted as T. plc T represents the arrival time of power channel data. rf Indicates the arrival time of the wireless channel;

[0086] S102. Calculate the delay data of the two channels using the transmission time and arrival time respectively. The formula is:

[0087] ;

[0088] In the formula, D plc D represents the time delay data of the power channel. rf Represents the latency data of the wireless channel;

[0089] Simultaneously, power and wireless channel latency are collected, and differential analysis can identify coordination anomalies between channels (such as clock synchronization failure).

[0090] S103. Standardize the collected dual-channel latency data. Eliminate the differences in latency data dimensions between different terminal devices and network environments, making the data comparable and improving the model's generalization ability.

[0091] S200. Classify the anomalies in dual-mode communication into four main anomalies, collect working data in dual-mode communication for the four main anomalies, and construct a four-dimensional dynamic baseline model.

[0092] The specific steps for constructing a four-dimensional dynamic baseline model are as follows:

[0093] S201. Anomalies in dual-mode communication are classified into four main types: power channel interference, wireless channel obstruction, topology link anomalies, and terminal clock synchronization failure. For the power channel interference anomaly type, the characteristic index is specified as α=D. plc The threshold parameter is calculated using historical values ​​of the characteristic indicator. The threshold parameter includes the average characteristic indicator value and the standard deviation. The formula is as follows:

[0094] ;

[0095] In the formula, μ α (t) represents the average characteristic value after applying an exponential moving average to the characteristic indicator, 0.2 represents the smoothing factor, and EMA represents the exponential moving average; StdDev represents the standard deviation of the calculated characteristic index over a historical 24 hours;

[0096] It adopts a combination of exponential moving average (EMA) and standard deviation to adapt to changes in the network environment in real time;

[0097] The threshold judgment mechanism is generated using threshold parameters, specifically as follows: When the characteristic index α in dual-mode communication is greater than the average characteristic index value plus three times the standard deviation, it is judged to be an abnormal type of power channel interference.

[0098] S202. For wireless channel obstruction types, the specified characteristic index is β=D. rf Threshold parameters are calculated using historical values ​​of the feature index, including the average value of the feature index β (μ). β (t) and standard deviation A threshold judgment mechanism is constructed using the characteristic index β-mean and standard deviation, specifically as follows: When the characteristic index β in dual-mode communication is greater than the average characteristic index value plus four times the standard deviation, it is judged as a wireless channel obstruction anomaly.

[0099] Power channel interference is assessed using a threshold of 3 times the standard deviation, while wireless obstruction is assessed using a threshold of 4 times the standard deviation. This matches the noise characteristics of different channels and reduces the false alarm rate.

[0100] S203. For topology link anomaly types, the specified characteristic index is γ = |△D_meas - △D_theory|, where γ represents the time delay difference deviation, △D_meas represents the actual measured time delay difference, and △D_theory represents the theoretical time delay difference. The actual measured time delay difference is calculated as follows: △D_meas = |D plc -D rf |;

[0101] The theoretical time delay difference is calculated as follows: In the formula, L represents the length of the conductor path, S represents the straight-line distance in space, and V... HPLC V represents the HPLC signal rate. RF Indicates the speed of the wireless signal;

[0102] Staff collect historical latency differences from the GIS topology database and calculate the average value as the latency difference threshold In_D. They then construct a threshold judgment mechanism, specifically: γ>In_D; when the characteristic index γ in dual-mode communication is greater than the latency difference threshold, it is judged as an abnormal type of topology link.

[0103] S204. For terminal clock synchronization failure type, the specified characteristic index is δ=|D plc -D rf|, δ represents the delay difference, and the calibration value in the Precision Time Protocol (PTP) is extracted as the out-of-synchronization threshold In_B; a threshold judgment mechanism is constructed, specifically: δ>In_B; when the characteristic index δ in dual-mode communication is greater than the out-of-synchronization threshold In_B, it is judged as a terminal clock out-of-synchronization anomaly.

[0104] By using the time delay difference deviation (γ) and the PTP calibration value (δ), joint monitoring of physical topology and time synchronization is achieved.

[0105] S205. Construct a four-dimensional dynamic baseline model based on the characteristic indicators and threshold judgment mechanism of the four main anomalies.

[0106] S300: During dual-mode communication, feature indicators are collected in real time, input into the four-dimensional dynamic baseline model, output the real-time anomaly type and issue an early warning.

[0107] The specific steps for outputting real-time anomaly types and issuing early warnings are as follows:

[0108] S301. Real-time acquisition and calculation of the characteristic indicators of four main anomalies in dual-mode communication, obtaining real-time characteristic indicator values, inputting the real-time characteristic indicator values ​​of the four main anomalies into the four-dimensional dynamic baseline model, and outputting the corresponding anomaly type when the threshold judgment mechanism in the four-dimensional dynamic baseline model is satisfied.

[0109] S302. When an anomaly is detected in dual-mode communication, calculate the deviation of the real-time characteristic index from the threshold in the threshold judgment mechanism, and calculate the anomaly confidence level of the judgment. The formula is as follows:

[0110] ;

[0111] In the formula, Con represents the anomaly confidence level, and Id... s Id represents the real-time feature index value. y The threshold value represents the characteristic index threshold in the threshold judgment mechanism; un represents the fluctuation range of the baseline in the threshold judgment mechanism. In the three main anomalies of power channel interference, wireless channel obstruction and topology link anomaly, the baseline is the average value of the corresponding characteristic index, and the fluctuation range is the standard deviation; in the main anomaly of terminal clock out of synchronization, the baseline is the PTP calibration value, and the fluctuation range is the average value of the historical PTP fluctuation amplitude.

[0112] S303. When the output exception type is not unique, determine the confidence level of different exception types, set the priority according to the confidence level from large to small, set the exception type with the highest confidence level as the highest priority, and then set the priority in order; issue warnings according to the priority from large to small.

[0113] When multiple anomalies are triggered simultaneously, priority is determined by ranking them by confidence level to avoid confusion in early warnings.

[0114] S400. Set different tracing strategies for the four main types of anomalies; select the tracing strategy according to the real-time anomaly type to find the root cause of the anomaly.

[0115] The specific steps for selecting a source tracing strategy to find the root cause of an anomaly based on the real-time anomaly type are as follows:

[0116] S401. Set source tracing strategies for the four main anomalies respectively. The source tracing strategy for power channel interference is as follows: collect the terminal locations of all warnings in dual-mode communication, classify the collected terminal locations using a spatial clustering algorithm, collect the voltage harmonic distortion rate of all communication terminals when a warning is issued, set a harmonic distortion rate threshold, and when the voltage harmonic distortion rate of a communication terminal exceeds the harmonic distortion rate threshold, it is judged as a noise source, and the spatial location of the noise source is obtained through the classified terminal locations.

[0117] S402. The specific source tracing strategy for wireless channel obstruction is as follows: Calculate the wireless communication path loss at different transmission distances in a spatial location using a log-normal shadowing fading model, collect the initial signal strength of the wireless signal, and determine that the theoretical signal strength = initial signal strength - path loss; measure the signal strength at the corresponding location in real time, and determine that the corresponding location is an abnormal source when the real-time signal strength is less than the theoretical signal strength.

[0118] S403. The specific strategy for tracing the abnormal topology link is as follows: A sweep frequency signal is sent to the terminal through the power channel, and the input impedance Z(f) = R(f) + jX(f) is measured in real time, where R(f) represents the real part, which is the value of resistance as a function of frequency, jX(f) represents the imaginary part, which is the value of reactance as a function of frequency, and f represents the frequency of the sweep frequency signal; the theoretical impedance is extracted from the power channel topology diagram, and the correlation coefficient between the measured impedance and the theoretical impedance is calculated. When the correlation coefficient is greater than Zy, it is determined that the power channel link corresponding to the sweep frequency signal receiving terminal is abnormal; Zy represents the manually preset correlation coefficient threshold.

[0119] S404. The specific strategy for tracing the cause of terminal clock synchronization failure is as follows: Collect the lowest latency value in history, and use the lowest latency value to calculate the latency offset of all terminals under the PTP protocol concentrator. The formula is: In the formula, △t represents the time delay offset, and BaseRTT represents the time delay value from the hour.

[0120] Within the time window, the delay offset is repeatedly calculated based on the time series. When the delay offset of all terminals under the concentrator increases synchronously, the concentrator is judged to be abnormal; when the delay offset of only a single terminal increases, the corresponding single terminal is judged to be abnormal.

[0121] S405. Select the corresponding source tracing strategy according to the type of abnormality in the warning, and use the source tracing strategy to locate the root cause of each type of abnormality in the warning.

[0122] S500 integrates real-time anomaly types and root cause data from tracing strategies for visualization.

[0123] S600: Use the root source data in the tracing strategy to calibrate and update the baseline in the four-dimensional dynamic baseline model.

[0124] The specific steps for calibrating and updating the baseline in the four-dimensional dynamic baseline model using the root cause data from the source tracing strategy are as follows:

[0125] S601. After identifying the root cause of each type of early warning anomaly, professionals maintain the root cause, collect the normal characteristic indicators after maintenance, and use the maintained normal characteristic indicators to update all baseline values ​​in the four-dimensional dynamic baseline model. The formula is as follows:

[0126] ;

[0127] In the formula, J new J represents the updated baseline value. old x represents the old baseline value before the update. normal This represents the normal feature index after maintenance, and r represents the learning rate weight.

[0128] After each fault is resolved, a new baseline is automatically learned, and the model accuracy improves with usage time.

[0129] The four-dimensional dynamic baseline model is updated and optimized using the updated baseline values.

[0130] A data analysis system based on HPLC dual-mode communication includes a data acquisition module, a model building module, an anomaly detection module, a source tracing module, a visualization module, and an optimization module.

[0131] The data acquisition module is used to set a hardware timestamp in the communication terminal device, collect latency data during dual-channel data transmission in dual-mode communication, and standardize the latency data.

[0132] The model building module is used to classify anomalies in dual-mode communication into four main anomalies, collect working data in dual-mode communication for the four main anomalies, and build a four-dimensional dynamic baseline model.

[0133] The anomaly detection module is used to collect feature indicators in real time during dual-mode communication, input the feature indicators into the four-dimensional dynamic baseline model, output the real-time anomaly type and issue an early warning.

[0134] The source tracing module is used to set different source tracing strategies for four main types of anomalies; and to select the source tracing strategy according to the real-time anomaly type to find the root cause of the anomaly.

[0135] The visualization module is used to integrate real-time anomaly types and root cause data in the tracing strategy and display them visually.

[0136] The optimization module is used to calibrate and update the baseline in the four-dimensional dynamic baseline model using the root cause data in the tracing strategy.

[0137] The model building module includes a power channel interference unit, a wireless channel obstruction unit, a topology link anomaly unit, and a terminal clock synchronization failure unit;

[0138] The power channel interference unit is used to specify the delay data of the power channel as a feature index and calculate threshold parameters to construct a threshold judgment mechanism.

[0139] The wireless channel blocking unit is used to specify the latency data of the wireless channel as a feature indicator and calculate threshold parameters to construct a threshold judgment mechanism.

[0140] The topology link anomaly unit is used to calculate the time delay difference deviation as a feature index and to calculate the time delay difference threshold to construct a threshold judgment mechanism.

[0141] The terminal clock out-of-sync unit is used to calculate the time delay difference as a characteristic indicator, calculate the out-of-sync threshold, and construct a threshold judgment mechanism.

[0142] The anomaly detection module includes an anomaly type output unit and a priority warning unit;

[0143] The anomaly type output unit is used to input the real-time feature index values ​​of the four main anomalies into the four-dimensional dynamic baseline model. When the threshold judgment mechanism in the four-dimensional dynamic baseline model is satisfied, the corresponding anomaly type is output.

[0144] The priority warning unit is used to calculate the confidence level of different anomaly types, set the priority according to the confidence level from large to small, and issue warnings in order of priority from large to small.

[0145] The tracing module includes a power channel interference tracing unit, a wireless channel obstruction tracing unit, a topology link anomaly tracing unit, and a terminal clock synchronization failure tracing unit;

[0146] The power channel interference tracing unit is used to set a harmonic distortion rate threshold. When the voltage harmonic distortion rate of the communication terminal exceeds the harmonic distortion rate threshold, it is judged as a noise source, and the spatial location of the noise source is obtained through the classified terminal location.

[0147] The wireless channel obstruction tracing unit is used to measure the signal strength at the corresponding location in real time. When the real-time signal strength is less than the theoretical signal strength, the corresponding location is determined to be an abnormal source.

[0148] The topology link anomaly tracing unit is used to extract the theoretical impedance in the power channel topology map, calculate the correlation coefficient between the measured impedance and the theoretical impedance, and determine that the power channel link corresponding to the frequency sweep signal receiving terminal is abnormal when the correlation coefficient is greater than the correlation coefficient threshold.

[0149] The terminal clock out-of-synchronization tracing unit is used to determine that the concentrator is abnormal when the delay offset of all terminals under the concentrator increases synchronously; and to determine that the corresponding individual terminal is abnormal when only the delay offset of a single terminal increases.

[0150] Example:

[0151] In a residential community (5000 households) in a certain city, there is one concentrator, 500 dual-mode smart meters, and 3 transformer monitoring terminals; the parameters for constructing the four-dimensional dynamic baseline model are:

[0152]

[0153] The real-time monitoring data for the characteristic indicators of meter M215 is: D plc =25.3; D rf =10.2; Inputting it into the model, it is determined that α, γ and δ all meet the threshold judgment mechanism; The anomaly types are identified as power channel interference, topology link anomaly and terminal clock out of sync, and the confidence level of power channel interference is calculated to be the highest, so it is set to the highest priority for output warning.

[0154] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A data analysis method based on HPLC dual-mode communication, characterized in that: The method includes the following steps: S100. Set a hardware timestamp in the communication terminal device, collect the latency data during dual-channel data transmission in dual-mode communication, and standardize the latency data. S200. Classify the anomalies in dual-mode communication into four main anomalies, collect working data in dual-mode communication for the four main anomalies, and construct a four-dimensional dynamic baseline model. The specific steps for constructing a four-dimensional dynamic baseline model are as follows: S201. Anomalies in dual-mode communication are classified into four main types: power channel interference, wireless channel obstruction, topology link anomalies, and terminal clock synchronization failure. For the power channel interference anomaly type, the characteristic index is specified as α=D. plc The threshold parameter is calculated using historical values ​​of the characteristic indicator. The threshold parameter includes the average characteristic indicator value and the standard deviation. The formula is as follows: ； In the formula, μ α (t) represents the average characteristic value after applying an exponential moving average to the characteristic indicator, 0.2 represents the smoothing factor, and EMA represents the exponential moving average; StdDev represents the standard deviation of the calculated characteristic index over a historical 24 hours; The threshold judgment mechanism is generated using threshold parameters, specifically as follows: When the characteristic index α in dual-mode communication is greater than the average characteristic index value plus three times the standard deviation, it is judged to be an abnormal type of power channel interference. S202. For wireless channel obstruction types, the specified characteristic index is β=D. rf Threshold parameters are calculated using historical values ​​of the feature index, including the average value of the feature index β (μ). β (t) and standard deviation A threshold judgment mechanism is constructed using the characteristic index β-mean and standard deviation, specifically as follows: When the characteristic index β in dual-mode communication is greater than the average characteristic index value plus four times the standard deviation, it is judged as a wireless channel obstruction anomaly. S203. For topology link anomaly types, the specified characteristic index is γ = |△D_meas - △D_theory|, where γ represents the time delay difference deviation, △D_meas represents the actual measured time delay difference, and △D_theory represents the theoretical time delay difference. The actual measured time delay difference is calculated as follows: △D_meas = |D plc -D rf |; The theoretical time delay difference is calculated as follows: In the formula, L represents the length of the conductor path, S represents the straight-line distance in space, and V... HPLC V represents the HPLC signal rate. RF Indicates the speed of the wireless signal; Staff collect historical latency differences from the GIS topology database and calculate the average value as the latency difference threshold In_D. They then construct a threshold judgment mechanism, specifically: γ>In_D; when the characteristic index γ in dual-mode communication is greater than the latency difference threshold, it is judged as an abnormal type of topology link. S204. For terminal clock synchronization failure type, the specified characteristic index is δ=|D plc -D rf |, δ represents the delay difference, and the calibration value in the Precision Time Protocol (PTP) is extracted as the out-of-synchronization threshold In_B; a threshold judgment mechanism is constructed, specifically: δ>In_B; when the characteristic index δ in dual-mode communication is greater than the out-of-synchronization threshold In_B, it is judged as a terminal clock out-of-synchronization anomaly. S205. Construct a four-dimensional dynamic baseline model based on the characteristic indicators and threshold judgment mechanism of four main anomalies. S300: During dual-mode communication, feature indicators are collected in real time, input into the four-dimensional dynamic baseline model, output the real-time anomaly type and issue an early warning. The specific steps for outputting real-time anomaly types and issuing early warnings are as follows: S301. Real-time acquisition and calculation of the characteristic indicators of four main anomalies in dual-mode communication, obtaining real-time characteristic indicator values, inputting the real-time characteristic indicator values ​​of the four main anomalies into the four-dimensional dynamic baseline model, and outputting the corresponding anomaly type when the threshold judgment mechanism in the four-dimensional dynamic baseline model is satisfied. S302. When an anomaly is detected in dual-mode communication, calculate the deviation of the real-time characteristic index from the threshold in the threshold judgment mechanism, and calculate the anomaly confidence level of the judgment. The formula is as follows: ； In the formula, Con represents the anomaly confidence level, and Id... s Id represents the real-time feature index value. y The threshold value represents the characteristic index threshold in the threshold judgment mechanism; un represents the fluctuation range of the baseline in the threshold judgment mechanism. In the three main anomalies of power channel interference, wireless channel obstruction and topology link anomaly, the baseline is the average value of the corresponding characteristic index, and the fluctuation range is the standard deviation; in the main anomaly of terminal clock out of synchronization, the baseline is the PTP calibration value, and the fluctuation range is the average value of the historical PTP fluctuation amplitude. S303. When the output exception type is not unique, determine the confidence level of different exception types, set the priority according to the confidence level from large to small, set the exception type with the highest confidence level as the highest priority, and then set the priority in order; issue warnings according to the priority from large to small. S400. Set different tracing strategies for the four main types of anomalies; select the tracing strategy according to the real-time anomaly type to find the root cause of the anomaly. S500 integrates real-time anomaly types and root cause data from tracing strategies for visualization. S600. Use the root source data in the tracing strategy to calibrate and update the baseline in the four-dimensional dynamic baseline model. The specific steps for calibrating and updating the baseline in the four-dimensional dynamic baseline model using the root cause data from the source tracing strategy are as follows: S601. After identifying the root cause of each type of early warning anomaly, professionals maintain the root cause, collect the normal characteristic indicators after maintenance, and use the maintained normal characteristic indicators to update all baseline values ​​in the four-dimensional dynamic baseline model. The formula is as follows: ； In the formula, J new J represents the updated baseline value. old x represents the old baseline value before the update. normal This represents the normal feature index after maintenance, and r represents the learning rate weight. The four-dimensional dynamic baseline model is updated and optimized using the updated baseline values.

2. The data analysis method based on HPLC dual-mode communication according to claim 1, characterized in that: The specific steps for standardizing the latency data in S100 are as follows: S101. Set a hardware timestamp in the communication terminal device to obtain the data transmission time T during communication. send When using a concentrator to collect data for dual-mode communication, the data arrival time during data transmission in both channels is denoted as T. plc T represents the arrival time of power channel data. rf Indicates the arrival time of the wireless channel; S102. Calculate the delay data of the two channels using the transmission time and arrival time respectively. The formula is: ； In the formula, D plc D represents the time delay data of the power channel. rf Represents the latency data of the wireless channel; S103. Standardize the collected dual-channel latency data.

3. The data analysis method based on HPLC dual-mode communication according to claim 1, characterized in that: The specific steps in S400 for selecting a tracing strategy to find the root cause of an anomaly based on the real-time anomaly type are as follows: S401. Set source tracing strategies for the four main anomalies respectively. The source tracing strategy for power channel interference is as follows: collect the terminal locations of all warnings in dual-mode communication, classify the collected terminal locations using a spatial clustering algorithm, collect the voltage harmonic distortion rate of all communication terminals when a warning is issued, set a harmonic distortion rate threshold, and when the voltage harmonic distortion rate of a communication terminal exceeds the harmonic distortion rate threshold, it is judged as a noise source, and the spatial location of the noise source is obtained through the classified terminal locations. S402. The specific source tracing strategy for wireless channel obstruction is as follows: Calculate the wireless communication path loss at different transmission distances in a spatial location using a log-normal shadowing fading model, collect the initial signal strength of the wireless signal, and determine that the theoretical signal strength = initial signal strength - path loss; measure the signal strength at the corresponding location in real time, and determine that the corresponding location is an abnormal source when the real-time signal strength is less than the theoretical signal strength. S403. The specific strategy for tracing the abnormal topology link is as follows: A sweep frequency signal is sent to the terminal through the power channel, and the input impedance Z(f) = R(f) + jX(f) is measured in real time, where R(f) represents the real part, which is the value of resistance as a function of frequency, jX(f) represents the imaginary part, which is the value of reactance as a function of frequency, and f represents the frequency of the sweep frequency signal; the theoretical impedance is extracted from the power channel topology diagram, and the correlation coefficient between the measured impedance and the theoretical impedance is calculated. When the correlation coefficient is greater than Zy, it is determined that the power channel link corresponding to the sweep frequency signal receiving terminal is abnormal; Zy represents the manually preset correlation coefficient threshold. S404. The specific strategy for tracing the cause of terminal clock synchronization failure is as follows: Collect the lowest latency value in history, and use the lowest latency value to calculate the latency offset of all terminals under the PTP protocol concentrator. The formula is: In the formula, △t represents the time delay offset, and BaseRTT represents the time delay value from the hour. Within the time window, the delay offset is repeatedly calculated based on the time series. When the delay offset of all terminals under the concentrator increases synchronously, the concentrator is judged to be abnormal; when the delay offset of only a single terminal increases, the corresponding single terminal is judged to be abnormal. S405. Select the corresponding source tracing strategy according to the type of abnormality in the warning, and use the source tracing strategy to locate the root cause of each type of abnormality in the warning.

4. A data analysis system based on HPLC dual-mode communication, employing the data analysis method based on HPLC dual-mode communication as described in any one of claims 1-3, characterized in that: The data analysis system includes a data acquisition module, a model building module, an anomaly detection module, a source tracing module, a visualization module, and an optimization module. The data acquisition module is used to set a hardware timestamp in the communication terminal device, collect latency data during dual-channel data transmission in dual-mode communication, and standardize the latency data. The model building module is used to classify anomalies in dual-mode communication into four main anomalies, collect working data in dual-mode communication for the four main anomalies, and build a four-dimensional dynamic baseline model. The anomaly detection module is used to collect feature indicators in real time during dual-mode communication, input the feature indicators into the four-dimensional dynamic baseline model, output the real-time anomaly type and issue an early warning. The source tracing module is used to set different source tracing strategies for four main types of anomalies; and to select the source tracing strategy according to the real-time anomaly type to find the root cause of the anomaly. The visualization module is used to integrate real-time anomaly types and root cause data in the tracing strategy and display them visually. The optimization module is used to calibrate and update the baseline in the four-dimensional dynamic baseline model using the root cause data in the tracing strategy.

5. A data analysis system based on HPLC dual-mode communication according to claim 4, characterized in that: The model building module includes a power channel interference unit, a wireless channel obstruction unit, a topology link anomaly unit, and a terminal clock synchronization failure unit. The power channel interference unit is used to specify the delay data of the power channel as a feature index and calculate threshold parameters to construct a threshold judgment mechanism. The wireless channel blocking unit is used to specify the latency data of the wireless channel as a feature indicator and calculate threshold parameters to construct a threshold judgment mechanism. The topology link anomaly unit is used to calculate the time delay difference deviation as a feature index and to calculate the time delay difference threshold to construct a threshold judgment mechanism. The terminal clock out-of-sync unit is used to calculate the time delay difference as a characteristic indicator, calculate the out-of-sync threshold, and construct a threshold judgment mechanism.

6. The data analysis system based on HPLC dual-mode communication according to claim 4, characterized in that: The anomaly detection module includes an anomaly type output unit and a priority warning unit; The anomaly type output unit is used to input the real-time feature index values ​​of the four main anomalies into the four-dimensional dynamic baseline model. When the threshold judgment mechanism in the four-dimensional dynamic baseline model is satisfied, the corresponding anomaly type is output. The priority warning unit is used to calculate the confidence level of different anomaly types, set the priority according to the confidence level from large to small, and issue warnings in order of priority from large to small.

7. A data analysis system based on HPLC dual-mode communication according to claim 4, characterized in that: The tracing module includes a power channel interference tracing unit, a wireless channel obstruction tracing unit, a topology link anomaly tracing unit, and a terminal clock synchronization failure tracing unit. The power channel interference tracing unit is used to set a harmonic distortion rate threshold. When the voltage harmonic distortion rate of the communication terminal exceeds the harmonic distortion rate threshold, it is judged as a noise source, and the spatial location of the noise source is obtained through the classified terminal location. The wireless channel obstruction tracing unit is used to measure the signal strength at the corresponding location in real time. When the real-time signal strength is less than the theoretical signal strength, the corresponding location is determined to be an abnormal source. The topology link anomaly tracing unit is used to extract the theoretical impedance in the power channel topology map, calculate the correlation coefficient between the measured impedance and the theoretical impedance, and determine that the power channel link corresponding to the frequency sweep signal receiving terminal is abnormal when the correlation coefficient is greater than the correlation coefficient threshold. The terminal clock out-of-synchronization tracing unit is used to determine that the concentrator is abnormal when the delay offset of all terminals under the concentrator increases synchronously; and to determine that the corresponding individual terminal is abnormal when only the delay offset of a single terminal increases.

Images