High-precision lightning stroke counting monitoring device and monitoring method

Abstract

The invention relates to the field of lightning protection monitoring of a power system, and particularly discloses a high-precision lightning stroke counting monitoring device and a monitoring method. The device comprises a sensing and data acquisition module, a signal adjustment module, a data processing module, a data storage and communication module and a power supply module. A lightning stroke frequency intensity index and early warning are generated through multi-sensor synchronous acquisition triggering, signal preprocessing, feature fusion, intelligent identification and classification and data analysis. According to the invention, high-precision lightning stroke counting can be realized, interference misjudgment is reduced, and support is provided for lightning protection operation and maintenance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lightning stroke monitoring, and particularly relates to a high-precision lightning stroke counting monitoring device and a monitoring method. BACKGROUND

[0002] Lightning stroke damage to power system equipment is an important problem in the operation of the power system. Accurate monitoring of the frequency of lightning strokes is crucial for the safe operation and maintenance of the power system. Existing lightning stroke monitoring technologies mainly rely on a single electromagnetic induction technology, which can monitor lightning stroke events, but has problems such as low counting accuracy and poor anti-interference ability in complex environments. In addition, existing technologies cannot simultaneously ensure high precision and comprehensive information on frequency, limiting their application in the fine management of power systems.

[0003] Reference 1: Chinese patent document with publication number CN109557360A.

[0004] Reference 1 describes a 10KV distribution network line lightning stroke counting device, which includes a grounding module, a power module, a counting module, a display module, and a current induction module. The grounding module includes a tower and a grounding part arranged in the tower, and the lower end of the tower is provided with a window. The power module, counting module, display module, and current induction module are all located in the window. The power module is connected to the display module to provide power for the display module. The counting module is connected to the current induction module to record the number of lightning strokes. The display module is connected to the counting module to display the number of lightning strokes. The current induction module is used to induce the current after lightning stroke.

[0005] Reference 2: Chinese patent document with publication number CN106610449A.

[0006] Reference 2 describes a binary lightning stroke counting device for a lightning arrester, which includes a lightning stroke current induction circuit and a motion counter connected to the lightning arrester body. The lightning stroke current induction circuit induces the secondary current of lightning stroke on the lightning arrester body and rectifies the output to control the motion of the motion counter. The motion counter includes an outer shell, an inner shell, a first display ring, a second display ring, a third display ring, and six groups of gear sets.

[0007] Existing electromagnetic induction lightning stroke counters only count through electromagnetic induction signals, which are easily affected by external electromagnetic interference, leading to false counting or missed counting. SUMMARY

[0008] The purpose of the present application is to solve the above technical problems in the prior art and provide a high-precision lightning stroke counting monitoring device and a monitoring method. Through the cooperative work of electromagnetic sensing and acoustic sensing, the false alarm and missed alarm rates can be greatly reduced.

[0009] The application is a high-precision lightning strike counting monitoring device, which has the following technical solutions: A sensing and data acquisition module is configured to capture the physical signals of lightning strikes. A signal conditioning module is configured to convert the original signals collected by the sensing and data acquisition module into signals that can be used for analysis. A data processing module is configured to run collaborative decision-making, feature extraction, and intelligent recognition algorithms. A data storage and communication module is configured to store and report data. A power module is configured to supply power to each module in the monitoring device. The sensing and data acquisition module includes an electromagnetic induction sensor and an acoustic sensor. The electromagnetic induction sensor is used to capture the transient large current changes on the lightning rod or tower when lightning strikes. The frequency response range of the acoustic sensor needs to cover the main energy frequency band of thunder. The signal conditioning module includes a high-precision analog-to-digital converter and an analog front-end conditioning circuit. The data processing module includes a master control chip and a high-precision clock source. The data storage and communication module includes a local storage and a communication unit. The power module includes a main power supply, a backup power supply, and a power management circuit.

[0010] As a further optimization of the high-precision lightning strike counting monitoring device, the electromagnetic induction sensor has an integrator circuit that restores the differential voltage signal induced by the Rogowski coil to the true current waveform.

[0011] As a further optimization of the high-precision lightning strike counting monitoring device, the acoustic sensor is a low-frequency microphone or a sound pressure sensor, and the acoustic sensor is equipped with a wind shield.

[0012] As a further optimization of the high-precision lightning strike counting monitoring device, the analog front-end conditioning circuit includes a filter circuit and an amplifier circuit. The filter circuit includes a band-pass filter for electromagnetic signals and a high-pass / band-pass filter for acoustic signals. The amplifier circuit includes a programmable gain amplifier.

[0013] As a further optimization of the high-precision lightning strike counting monitoring device, the master control chip includes a microprocessor and a digital signal processor.

[0014] As a further optimization of the high-precision lightning strike counting monitoring device, the high-precision clock source is a GPS / Beidou time service module.

[0015] The application also provides a high-precision lightning strike counting monitoring method, which includes the following steps: S1, Multi-sensor data synchronous acquisition and triggering After the device is powered on, the electromagnetic induction sensor and the acoustic sensor continuously perform data acquisition, and the GPS / Beidou timing module ensures that each data sample of the two channels is marked with a unified microsecond-level timestamp; Real-time calculation of the current rate di / dt of the electromagnetic induction signal, when the value exceeds the preset first threshold, the system immediately determines that it is a suspected lightning strike event, and starts the event recording sequence; S2, Signal preprocessing Electromagnetic signal preprocessing: first, use an adaptive notch filter to deeply attenuate 50Hz power frequency and its main harmonics, eliminate power system steady-state magnetic field interference, and then use a digital bandpass filter to retain lightning current high-frequency characteristics and filter out power frequency residues and higher frequency radio noise; Acoustic signal preprocessing: first, use a FIR bandpass filter to filter out wind noise high-frequency components and out-of-band noise such as insect chirping, and then use the ambient noise collected by the reference microphone to estimate and subtract the relevant noise in the main microphone signal through the LMS algorithm, significantly improving the signal-to-noise ratio; S3, Feature extraction and signal fusion Perform continuous wavelet transform on the preprocessed electromagnetic and acoustic signals, use complex Morlet wavelet as the mother wavelet, and generate high-resolution time-frequency graphs; Extract peak amplitude, signal energy integral, wavelet entropy, and main frequency band energy from the electromagnetic signal; extract duration, zero-crossing rate, sound pressure level, and the first 12-dimensional mel frequency cepstral coefficients from the acoustic signal; Combine all the above features into a high-dimensional joint feature vector as the input for subsequent intelligent recognition; S4, Intelligent recognition and classification Input the joint feature vector into the pre-trained deep neural network, and the network output contains the classification results of lightning probability P 雷击 , electromagnetic interference probability P 电磁干扰 , and acoustic interference probability P 声学干扰 , and the regression results containing lightning intensity; Set a high confidence threshold, when P 雷击 is greater than the high confidence threshold, the event is finally confirmed as an effective lightning strike, and for the events confirmed as lightning strikes, record their timestamp and lightning intensity value; when P 雷击 is less than the high confidence threshold, the event is judged as an interference event and is stored in the interference log for analysis, but is not counted in the lightning count; S5, Data analysis and frequency intensity comprehensive information generation Create a record for each lightning event in the time series database, including timestamp, intensity, geographic location, etc. Count the confirmed lightning stroke events in a set statistical period, and calculate the average value, maximum value and median value of the lightning stroke intensity in the same period; Construct a frequency intensity index to reflect the overall severity of lightning activity.

[0016] As a further optimization of the high-precision lightning count monitoring method of the application: the main harmonics in step S2 are 100Hz and 150Hz, and the digital band-pass filter is a Butterworth filter with a passband of 1kHz-1MHz.

[0017] As a further optimization of the high-precision lightning count monitoring method of the application: the frequency intensity index in step S5 is log 10 (total number+1) x average intensity.

[0018] As a further optimization of the high-precision lightning count monitoring method of the application: step S5 further comprises: automatically generating periodic reports to show the trend of frequency, intensity and frequency intensity index in chart form, and automatically sending warning messages or push notifications to operation and maintenance personnel through the communication module when the key indicators exceed the threshold.

[0019] The application has the following beneficial effects: (1) By the fusion of electromagnetic induction and acoustic technology, the accuracy and reliability of lightning monitoring are significantly improved, and the counting accuracy can reach more than 99%.

[0020] (2) The signal processing algorithm using wavelet transform and neural network effectively filters out interference signals, improves the anti-interference ability of the lightning monitor, and enables it to work stably in complex electromagnetic environments.

[0021] (3) The application can record lightning frequency, provide more data support for fine management and fault analysis of power systems, and help optimize the lightning protection design and operation and maintenance strategy of power systems. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a structural schematic diagram of the monitoring device of the application; Figure 2 is a system framework schematic diagram of the monitoring device of the application; Figure 3 is a logic flowchart of the monitoring method of the application; Markings in the figure: 1, digital display; 2, L-shaped base; 3, electromagnetic induction coil; 4, insulator string. DETAILED DESCRIPTION

[0023] For better understanding of the present application, the following further illustrates the content of the present application in conjunction with examples, but the content of the present application is not limited to the following examples.

[0024] <Example 1> As shown in Figure 1 and 2 A high-precision lightning stroke counting monitoring device has a sensing and data acquisition module, a signal conditioning module, a data processing module, a data storage and communication module, and a power module.

[0025] The sensing and data acquisition module is used to capture the physical signals of lightning strokes. The sensing and data acquisition module includes an electromagnetic induction sensor and an acoustic sensor, both of which are connected to the input end of the signal conditioning module through shielded cables, for synchronous capture of physical signals when lightning strokes occur.

[0026] The electromagnetic induction sensor is used to capture the transient large current change on the lightning rod or tower when lightning strokes occur, and its output signal is a differential voltage signal proportional to the rate of change of lightning current. The sensor is integrated with an integrator circuit, which can convert the differential voltage signal into an analog voltage signal proportional to the amplitude of the lightning current.

[0027] The frequency response range of the acoustic sensor needs to cover the main energy frequency band of thunder. The acoustic sensor is a low-frequency microphone or a sound pressure sensor, and the acoustic sensor is configured with a wind shield.

[0028] The signal conditioning module is used to convert the original signals collected by the sensing and data acquisition module into signals that can be used for analysis. The signal conditioning module includes a high-precision analog-to-digital converter and an analog front-end conditioning circuit.

[0029] The high-precision analog-to-digital converter selects a 24-bit Sigma-Delta type ADC (model ADS1278), which synchronously converts the conditioned electromagnetic signals (sampling rate 1MHz) and acoustic signals (sampling rate 44.1kHz) to ensure high-precision acquisition of digital signals.

[0030] The analog front-end conditioning circuit includes a filtering circuit and an amplification circuit. The filtering circuit is designed for different types of signals: for electromagnetic signals, a band-pass filter (center frequency 500 kHz, passband range 1 kHz-1 MHz, 8th order Butterworth structure, stopband attenuation ≥60 dB) is used to retain the high-frequency characteristics of lightning current and filter out power frequency and radio frequency noise. For acoustic signals, a high-pass filter (cutoff frequency 20 Hz, 2nd order Chebyshev type I structure) and a band-pass filter (passband range 50-500 Hz, 4th order Butterworth structure) are cascaded to filter out out-of-band noise such as wind noise high-frequency components and insect chirping (usually >2 kHz). The amplification circuit uses a programmable gain amplifier (PGA, model AD8253) with a gain adjustment range of 1-64 times. The gain can be automatically adjusted according to the original signal strength (e.g., when the electromagnetic signal amplitude is <100 mV, the gain is set to 32 times; when the signal amplitude is >500 mV, the gain is set to 4 times) to avoid quantization errors caused by weak signals or saturation distortion caused by strong signals.

[0031] The data processing module is used to run collaborative decision-making, feature extraction, and intelligent recognition algorithms. The data processing module includes a master control chip and a high-precision clock source.

[0032] The master control chip uses a dual-core architecture of "microprocessor + digital signal processor (DSP)". The microprocessor is STM32H743 (32-bit ARM Cortex-M7 core), responsible for logical control, module scheduling, and data interaction; the DSP is TMS320C6748, responsible for complex algorithm operations such as continuous wavelet transform and deep neural network inference, and the two communicate through the SPI interface. This dual-core architecture can balance the flexibility of logical control and the operation efficiency of complex algorithms, avoiding processing delays caused by excessive load on a single chip.

[0033] The high-precision clock source uses a GPS / Beidou dual-mode time module (model UM220-IV). This module is connected to the master control chip through the UART interface, and adds a unified microsecond-level timestamp to each data sample of the electromagnetic signal and acoustic signal, solving the event matching error problem caused by the asynchronous data of multiple sensors, and ensuring the time accuracy of lightning strike event determination.

[0034] The data storage and communication module is used for data storage and reporting. The data storage and communication module includes a local storage and a communication unit.

[0035] The local storage uses a 16GB industrial-grade SD card (Kingston IndustrialSDHC) and 2MB of on-chip Flash (integrated into the STM32H743 chip). The SD card stores raw acquired data (7-day retention), processed lightning strike event records (permanent retention), and interference logs (30-day retention). The on-chip Flash stores device configuration parameters (such as thresholds and sampling rates) and algorithm firmware to prevent data loss in case of power failure.

[0036] The communication unit uses a 4GCat.1 communication module (model Quectel EC20) and can be optionally equipped with a LoRa module (model Semtech SX1278) for data transmission in remote areas without 4G signal. The communication unit connects to the main control chip via a UART interface, supports the TCP / IP protocol, and can report lightning strike event records, statistical data, and early warning information to the remote power monitoring platform in real time. It also supports remotely receiving configuration commands (such as threshold adjustment and sampling rate modification) issued by the platform.

[0037] The power supply module is used to supply power to the various modules in the monitoring device. The power supply module includes a main power supply, a backup power supply, and a power management circuit.

[0038] The main power supply uses 220V AC mains power via an AC-DC converter (model MeanWellRS-15-12) to power distribution boxes near power poles.

[0039] The backup power supply uses two 18650 lithium-ion batteries (each with a capacity of 2500mAh, connected in series to output 7.4V, for a total capacity of 5000mAh), which can support the device to work continuously for ≥72 hours after the main power supply fails.

[0040] The power management circuit uses the MP2359 power management chip to realize automatic switching between main and backup power supplies, and has overvoltage, overcurrent and overtemperature protection functions to prevent abnormal power supply from damaging the module.

[0041] The monitoring device has a digital display 1, which is mounted on the angle steel tower material via an L-shaped base 2. The digital display 1 contains an acoustic sensor with noise suppression and filtering methods to retain the effective frequency components of the lightning strike sound wave and stably receive the lightning strike sound signal for counting. The electromagnetic induction coil 3 is wired to the digital display 1 and fixed at the middle ring of the insulator string 4, and counts are performed by monitoring the trigger frequency of electromagnetic induction.

[0042] <Example 2> like Figure 3 As shown: A high-precision lightning strike counting and monitoring method includes the following steps: S1. Multi-sensor data synchronous acquisition and triggering After the device is powered on, the electromagnetic induction sensor and the acoustic sensor continuously collect data, and the GPS / Beidou time module ensures that each data sample of the two channels is marked with a unified microsecond timestamp. Real-time calculation of the current rate di / dt of the electromagnetic induction signal, when the value exceeds the preset first threshold, the system immediately determines a suspected lightning event and starts the event recording sequence.

[0043] Specifically, after the device is powered on, the power management circuit automatically starts the main power supply, each module completes initialization, and then the electromagnetic induction sensor and the acoustic sensor enter a continuous data collection state: The GPS / Beidou time module sends a time calibration signal to the main control chip every 1 second, and outputs a 1PPS second pulse. The main control chip marks each data sample of the electromagnetic signal (sampling rate 1MHz) and the acoustic signal (sampling rate 44.1kHz) with a unified microsecond timestamp (timestamp format: YYYY-MM-DDHH:MM:SS.ssssss) according to the pulse, ensuring the time synchronization of the two channels of data; The microprocessor reads the current signal output by the electromagnetic induction sensor in real time and calculates the current rate di / dt.

[0044] Calculation method: take the current values of the continuous 5 sampling points, calculate the current difference value of the adjacent sampling points by difference method, and then divide by the sampling interval 1μs to get the di / dt value); the preset first threshold is 10kA / μs, when the real-time calculated di / dt value exceeds the threshold for 3 consecutive sampling points, the system immediately determines a suspected lightning event and starts the event recording sequence (records the electromagnetic signal and acoustic signal 1s before and after the trigger to ensure complete capture of the event process). This trigger mechanism can effectively avoid false triggering caused by transient pulse interference and improve the reliability of event triggering.

[0045] The real-time trigger calculation of step S1 is an independent, fast and rough front-end screening mechanism. It is based on the most original signal and the simplest operation, and its core task is to capture potential events with low delay and provide a start signal and data window for the subsequent precise but time-consuming signal processing and analysis process.

[0046] S2, signal preprocessing Electromagnetic signal preprocessing: an adaptive notch filter is used to deeply attenuate the 50Hz power frequency and its 100Hz (2nd harmonic) and 150Hz (3rd harmonic) of the power system, and eliminate the steady-state magnetic field interference; a digital band-pass filter (8th order Butterworth structure, passband 1kHz-1MHz, ripple≤0.1dB) is used to retain the high-frequency characteristics of lightning current and filter out the residual noise of power frequency and >1MHz radio frequency noise.

[0047] Acoustic signal preprocessing: Out-of-band noise filtering: A 128th-order FIR bandpass filter is used to filter out high-frequency components of wind noise (usually >2kHz) and out-of-band noise such as insect chirping (>1kHz); using the noise signal collected by the reference microphone (10cm away from the main microphone, synchronously collecting ambient noise), the relevant noise in the main microphone signal is estimated and subtracted through the LMS adaptive filtering algorithm (500 iterations, convergence coefficient 0.01), so that the signal-to-noise ratio of the acoustic signal is improved by ≥20dB.

[0048] S3, Feature Extraction and Signal Fusion Continuous wavelet transforms were performed on the preprocessed electromagnetic and acoustic signals, respectively, using complex Morlet wavelets as the mother wavelet to generate high-resolution time-frequency diagrams. Peak amplitude, signal energy integral, wavelet entropy, and dominant frequency band energy were extracted from the electromagnetic signals; duration, zero-crossing rate, sound pressure level, and the first 12-dimensional Mel frequency cepstral coefficients were extracted from the acoustic signals.

[0049] Specifically, electromagnetic signal characteristics: A time-frequency graph is generated using continuous wavelet transform (mother wavelet is a complex Morlet wavelet, center frequency 200Hz, scale range 1-128), from which four key features are extracted: peak amplitude (the maximum current value at each sampling point, in kA), signal energy integral (the sum of squares of the current values ​​at all sampling points multiplied by the sampling interval of 1μs, in kA). 2 •s), wavelet entropy (the entropy value of the energy distribution of wavelet coefficients, reflecting signal complexity, ranging from 0 to 1), and dominant frequency band energy (the total energy of the 50-500kHz frequency band, accounting for the proportion of total energy). Acoustic signal characteristics: Similarly, a time-frequency diagram is generated through continuous wavelet transform, extracting four key features: duration (the total duration for which the signal sound pressure exceeds the 0.1Pa threshold, in seconds), zero-crossing rate (the number of times the signal crosses the 0V threshold per second, reflecting the signal frequency characteristics), and sound pressure level (calculated using the formula: 20 × log...). 10 (Actual sound pressure / 20μPa), unit dB), the first 12-dimensional Mel frequency cepstral coefficients (MFCC, calculated by pre-emphasis, framing (frame length 20ms, frame shift 10ms), adding Hanning window, FFT, Mel filtering (24 filters), DCT transform, reflecting the spectral envelope characteristics of the acoustic signal).

[0050] The four features of the electromagnetic signal and the four features of the acoustic signal (a total of 8 dimensions) are combined to form a high-dimensional joint feature vector, which serves as the input for subsequent intelligent recognition algorithms. Through multimodal feature fusion, the "intensity information" of the electromagnetic signal and the "time-frequency information" of the acoustic signal can be fully utilized to improve the comprehensiveness of lightning strike event determination.

[0051] S4, Intelligent Recognition and Classification The joint feature vector is input into a pre-trained deep neural network, and the network output contains the lightning strike probability P. 雷击 Electromagnetic interference probability P 电磁干扰 and acoustic interference probability P 声学干扰 The classification results and regression results containing lightning strike intensity.

[0052] The deep neural network structure is as follows: It adopts a hybrid architecture of "CNN+LSTM", where: the CNN part contains 2 convolutional layers (3×3 kernels, with 16 and 32 kernels respectively) and 2 max pooling layers (2×2 kernels) to extract local spatial features of the feature vector; the LSTM part contains 1 hidden layer (64 neurons) to extract temporal correlation features of the features; the fully connected layer contains 4 output neurons, which output the probability P of lightning strike. 雷击 Electromagnetic interference probability P 电磁干扰 Acoustic interference probability P 声学干扰 (The sum of the three is 1) and the regression value of lightning strike intensity (unit kA).

[0053] Set a high confidence threshold, when P 雷击 If the value exceeds the high confidence threshold, the event is ultimately confirmed as a valid lightning strike. For events confirmed as lightning strikes, their timestamp and lightning strike intensity value are recorded; when P... 雷击 Events below the high confidence threshold are considered interference and are stored in the interference log for analysis, but are not included in the lightning strike count.

[0054] Specifically, the model is pre-trained using 10,000 samples (5,000 real lightning strike samples, 3,000 electromagnetic interference samples (e.g., switching operations, transformer excitation), and 2,000 acoustic interference samples (e.g., heavy rain, vehicle horns)). After training, the model's classification accuracy is ≥98%, and the lightning strike intensity regression error is ≤5%. A high confidence threshold of 0.9 is set. When Plightning strike > 0.9, it is determined as a "valid lightning strike event," and the event's timestamp (accurate to μs) and lightning strike intensity value (the regression value output by the deep neural network) are recorded. When Plightning strike ≤ 0.9, it is determined as an "interference event," and the event's timestamp, interference type (maximum Pelectromagnetic interference indicates electromagnetic interference, otherwise it indicates acoustic interference), and feature vector are stored in the interference log for subsequent interference source analysis, but are not included in the lightning strike count. This intelligent recognition mechanism can significantly reduce the misclassification rate caused by interference and improve the accuracy of lightning strike counting.

[0055] S5. Data Analysis and Frequency Intensity Integrated Information Generation Create a record for each lightning strike event in the time-series database, including fields such as timestamp, intensity, and geographic location.

[0056] For example, a record is created in the local InfluxDB time series database for each valid lightning strike event, with fields including: time (time stamp), intensity (lightning strike intensity, kA), location (latitude and longitude of the device, provided by the GPS module, with an accuracy of ±10m), and device_id (unique device number, such as "LSJC-001").

[0057] Within a defined statistical period, confirmed lightning strikes are counted, and the average, maximum, and median lightning strike intensity within the same period are calculated. A frequency-intensity index is constructed to reflect the overall severity of lightning activity.

[0058] Three statistical periods were set (1 hour, 24 hours (day), and 7 days (week)), and valid lightning strike events were counted within each period: Lightning strike count: The total number of valid lightning strike events within the statistical period; Intensity statistics: Calculate the average (total intensity / total number of times) of lightning strike intensity within the period, the maximum (maximum intensity value within the period), and the median (the median value is taken after sorting the intensity values ​​to avoid the influence of extreme values); Frequency intensity index calculation: Calculated according to the formula "Frequency intensity index = log10 (total frequency + 1) × average intensity".

[0059] For example: If there are 10 lightning strikes in a 24-hour period, with an average intensity of 50 kA, then the frequency intensity index = log0 10 (11)×50≈1.04×50=52. This index can comprehensively reflect the "frequency" and "intensity" of lightning activities within a cycle, and intuitively reflect the overall severity of lightning activities. The device automatically generates a daily report for the previous 24 hours at 8:00 AM every day and a weekly report for the previous week at 8:00 AM every Monday. The reports are stored in PDF format on the local SD card and uploaded to the remote monitoring platform via a 4G module. The reports use line graphs to show the trend of lightning strike frequency, bar graphs to show the intensity distribution, and radar graphs to show the changes in the frequency intensity index.

[0060] Preset warning thresholds (such as daily lightning strike count > 50, daily average intensity > 100kA, daily frequency intensity index > 100) are set. When the statistical indicators exceed the thresholds, the main control chip controls the communication unit to send a warning SMS to the mobile phones of maintenance personnel via the SMS module (model SIM800C). The SMS message reads, "Average intensity 85kA, frequency intensity index 68, please check the lightning protection status of the transmission lines in a timely manner." Simultaneously, a warning message is pushed to the monitoring platform. This function enables proactive alerts about lightning strike risks, providing timely data support for the lightning protection maintenance of the power system.

[0061] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. A high-precision lightning strike counting monitoring device, characterized by, It has: Sensing and data acquisition module for capturing the physical signal of lightning strike; Signal conditioning module for converting the original signal collected by the sensing and data acquisition module into a signal that can be used for analysis; Data processing module for running collaborative decision, feature extraction and intelligent recognition algorithm; Data storage and communication module for data retention and reporting; Power module for supplying power to each module in the monitoring device; The sensing and data acquisition module includes an electromagnetic induction sensor and an acoustic sensor, the electromagnetic induction sensor is used to capture the transient large current change on the lightning conductor or tower when lightning occurs, and the frequency response range of the acoustic sensor needs to cover the main energy frequency band of thunder sound; The signal conditioning module includes a high-precision analog-to-digital converter and an analog front-end conditioning circuit; The data processing module includes a main control chip and a high-precision clock source; The data storage and communication module includes a local storage and a communication unit; The power module includes a main power supply, a backup power supply and a power management circuit.

2. The high-precision lightning strike counter monitoring device of claim 1, wherein: The electromagnetic induction sensor has an integrator circuit for restoring the differential voltage signal induced by the Rogowski coil to the true current waveform.

3. The high-precision lightning strike counter monitoring device of claim 1, wherein: The acoustic sensor is a low-frequency microphone or a sound pressure sensor, and the acoustic sensor is provided with a wind shield.

4. The high-precision lightning-stroke counting monitoring device according to claim 1, characterized in that: The analog front-end conditioning circuit includes a filter circuit and an amplifier circuit, the filter circuit includes a band-pass filter for electromagnetic signals and a high-pass / band-pass filter for acoustic signals, and the amplifier circuit includes a programmable gain amplifier.

5. The high precision lightning strike counter monitoring device of claim 1, wherein: The main control chip includes a microprocessor and a digital signal processor.

6. The high precision lightning strike counter monitoring device of claim 1, wherein: The high-precision clock source is a GPS / Beidou time service module.

7. A high-precision lightning strike counting monitoring method, characterized by, The method comprises the following steps: S1, multi-sensor data synchronous acquisition and triggering After the device is powered on, the electromagnetic induction sensor and the acoustic sensor continuously perform data acquisition, and the GPS / Beidou time service module ensures that each data sample of the two channels is marked with a unified microsecond-level time stamp; Real-time calculation of the current change rate di / dt of the electromagnetic induction signal, when the value exceeds a preset first threshold value, the system immediately determines it as a suspected lightning strike event and starts an event recording sequence; S2, signal preprocessing Electromagnetic signal preprocessing: first, use an adaptive notch filter to deeply attenuate 50Hz power frequency and its main harmonics to eliminate power system steady-state magnetic field interference, and then use a digital band-pass filter to retain lightning current high-frequency characteristics and filter out power frequency residues and higher frequency radio noise; Acoustic signal preprocessing: first, use a FIR band-pass filter to filter out wind noise high-frequency components and out-of-band noise such as insect chirping, and then use the environmental noise collected by the reference microphone to estimate and subtract the relevant noise in the main microphone signal through the LMS algorithm to significantly improve the signal-to-noise ratio; S3, feature extraction and signal fusion Perform continuous wavelet transform on the preprocessed electromagnetic and acoustic signals respectively, use complex Morlet wavelet as mother wavelet, and generate high-resolution time-frequency graph; Extract peak amplitude, signal energy integral, wavelet entropy and main band energy from the electromagnetic signal; extract duration, zero-crossing rate, sound pressure level and first 12-dimensional mel frequency cepstral coefficient from the acoustic signal; Combine all the above features into a high-dimensional joint feature vector as the input of subsequent intelligent recognition; S4, Intelligent identification and classification The combined feature vector is input into a pre-trained deep neural network, and the network outputs classification results containing lightning strike probability P 雷击 , electromagnetic interference probability P 电磁干扰 , and acoustic interference probability P 声学干扰 , and regression results containing lightning strike intensity; Set a high confidence threshold, when P 雷击 greater than the high confidence threshold, the event is finally confirmed as a valid lightning strike, and for the event confirmed as a lightning strike, its timestamp and lightning intensity value are recorded; when P 雷击 less than the high confidence threshold, the event is judged as an interference event, which is stored in the interference log for analysis, but not counted in the lightning count; S5, Data analysis and frequency intensity comprehensive information generation Create a record for each lightning strike event in the time series database, including fields such as timestamp, intensity, geographic location, etc. Count the confirmed lightning strike events within a set statistical period, and calculate the average, maximum, and median of the lightning strike intensity within the same period. Construct a frequency intensity index to reflect the overall severity of lightning activity.

8. The high-precision lightning strike count monitoring method of claim 7, wherein, The main harmonics in step S2 are 100 Hz and 150 Hz, and the digital band-pass filter is a Butterworth filter with a passband of 1 kHz-1 MHz.

9. The high-precision lightning strike counter monitoring method of claim 7, wherein, The step S5 frequency intensity index = log 10 (total number + 1) x average intensity.

10. The high-precision lightning strike counter monitoring method of claim 7, wherein, The step S5 further comprises: automatically generating periodic reports to show the change trend of frequency, intensity and frequency intensity index in the form of charts, and automatically sending warning messages or push notifications to operation and maintenance personnel through the communication module when the key indicators exceed the threshold.

Images