A method and system for detecting laser multi-channel sequential power-up failure

Abstract

The application provides a laser multi-channel sequential power-on failure detection method and system. In the laser multi-channel sequential power-on process, the driving voltage data, voltage abnormal duration and acoustic wave signal of each channel are collected. According to the driving voltage data, the voltage abnormal channel is determined. Each acoustic wave signal is subjected to short-time Fourier transform to obtain a time-frequency spectrum. The acoustic characteristic waveform is matched with a preset short-circuit acoustic characteristic template or a preset arc acoustic characteristic template. The channel whose matching degree exceeds a set threshold is marked as an acoustic abnormal channel. According to the channel which is simultaneously marked as the voltage abnormal channel and the acoustic abnormal channel, the abnormal correlation matrix is constructed in combination with the physical position relationship of the channel to determine a failure channel set. According to a historical failure characteristic database, an identification report is obtained. The technical scheme provided by the application realizes accurate positioning, type classification and risk warning of the short-circuit or arc failure channel in the laser power-on process.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optoelectronic technology, in particular to a laser multi-channel sequential power-on failure detection method and system. BACKGROUND

[0002] In a high-power, multi-channel integrated laser system, local short circuit, arc discharge and other faults may be caused by factors such as aging of components, degradation of insulation performance, or connection abnormalities during power-on, resulting in abnormal fluctuations in drive voltage, and even causing a chain of damage. Therefore, a detection mechanism that can automatically identify potential electrical faults in real time is urgently needed, which not only requires accurate positioning of the abnormal channel, but also needs to distinguish the fault type and assess its risk level.

[0003] Currently, the existing mainstream scheme attempts to deploy a distributed voltage sensor network to continuously collect voltage drop data during the power-on process of each channel, and to calculate the voltage change rate and the degree of deviation from the reference value in real time based on a sliding window algorithm. When the preset threshold is exceeded, it is determined that there is an electrical abnormality in the channel. To further improve the credibility of the judgment basis, some systems introduce a simple acoustic monitoring module to capture the sudden noise generated during the power-on process using a microphone array. Through energy threshold detection, it is identified whether there is a discharge explosion sound or mechanical dithering sound. If the voltage abnormality and the sudden increase in acoustic signals occur at the same time, the alarm level is enhanced. However, the existing scheme has significant defects, for example, it is difficult to distinguish between transient interference and real short circuit, resulting in a high false positive rate; the acoustic signal processing is limited to the energy level and lacks detailed analysis of the sound spectrum characteristics, making it impossible to distinguish the essential differences between arc sound, poor contact sound and environmental noise, etc. SUMMARY

[0004] The present application provides a laser multi-channel sequential power-on failure detection method and system to solve the problems in the prior art, such as the difficulty in distinguishing between transient interference and real short circuit, the high false positive rate, and the inability to distinguish the essential differences between arc sound, poor contact sound and environmental noise.

[0005] In a first aspect, the present application provides a laser multi-channel sequential power-on failure detection method, comprising:

[0006] During the laser multi-channel sequential power-on process, the drive voltage data, voltage abnormality duration and acoustic signal of each channel are collected;

[0007] According to the drive voltage data, the voltage change amplitude and voltage abnormality duration of each channel are calculated, and the channels whose voltage change amplitude exceeds the preset voltage change threshold or whose voltage abnormality duration exceeds the preset time threshold are marked as voltage abnormality channels;

[0008] performing short-time Fourier transform on each acoustic signal to obtain a time-frequency spectrum, and extracting an acoustic feature waveform from the time-frequency spectrum, and marking a channel as an acoustic abnormal channel if the acoustic feature waveform matches a preset short-circuit acoustic feature template or a preset arc acoustic feature template by more than a set threshold;

[0009] constructing an abnormal correlation matrix according to the channels that are simultaneously marked as voltage abnormal channels and acoustic abnormal channels, and combining the physical location relationship of the channels, and selecting a correlation group in which the abnormal correlation strength of the channels exceeds a preset strong correlation threshold from the abnormal correlation matrix, and combining to determine a failure channel set;

[0010] According to the historical failure feature database, the spatial distribution characteristics of the failure channel set are matched with historical failure cases to obtain an identification report containing failure positioning location, abnormal type classification result and short-circuit risk warning level.

[0011] Optionally, according to the driving voltage data, the voltage variation amplitude and the voltage abnormal duration of each channel are calculated, and a channel is marked as a voltage abnormal channel if the voltage variation amplitude exceeds a preset voltage variation threshold or the voltage abnormal duration exceeds a preset time threshold, including:

[0012] The driving voltage data is subjected to sliding window smoothing processing to obtain a smoothed voltage sequence, and the change rate of the smoothed voltage sequence is calculated;

[0013] When the change rate is within a preset low fluctuation range, the voltage value corresponding to the change rate is taken as a reference voltage value, and the real-time voltage deviation value between each voltage value in the smoothed voltage sequence and the reference voltage value is calculated;

[0014] The time point at which the real-time voltage deviation value exceeds a preset deviation trigger threshold is marked as a voltage fluctuation starting point, and the time point at which the real-time voltage deviation value first falls from the exceeding state to not more than the preset deviation trigger threshold is marked as a voltage fluctuation termination point, and the voltage variation interval is determined in combination with the voltage fluctuation starting point;

[0015] The maximum voltage value and the minimum voltage value are extracted from the voltage variation interval, and the difference between the maximum voltage value and the minimum voltage value is calculated, and the difference is taken as the voltage variation amplitude;

[0016] The voltage abnormal duration of the voltage variation interval is calculated, and a channel corresponding to a voltage variation amplitude exceeding a preset voltage variation threshold or a voltage abnormal duration exceeding a preset time threshold is marked as a voltage abnormal channel.

[0017] Optionally, the driving voltage data is subjected to sliding window smoothing processing to obtain a smoothed voltage sequence, including:

[0018] extracting a voltage collection time and a corresponding voltage value from the driving voltage data, and arranging all the voltage values to obtain a time sequence voltage sequence;

[0019] calculating a time interval average of adjacent voltage collection times, taking the time interval average as a sampling period, and determining a sliding window length in combination with a preset window time length;

[0020] based on the sliding window length, dividing the time sequence voltage sequence into multiple data segments, and calculating an average voltage value of all voltage values in each data segment, taking the average voltage value of each data segment as a smoothed voltage value, and splicing all the smoothed voltage values to form a smoothed voltage sequence.

[0021] Optionally, performing short-time Fourier transform on each acoustic signal to obtain a time-frequency spectrum, and extracting an acoustic feature waveform from the time-frequency spectrum, marking a channel as an acoustic abnormal channel if the matching degree of the acoustic feature waveform with a preset short-circuit acoustic feature template or a preset arc acoustic feature template exceeds a set threshold, including:

[0022] frame processing on the acoustic signal to obtain multiple acoustic frames, and adding a preset window function to each acoustic frame to obtain a windowed acoustic frame;

[0023] decomposing the windowed acoustic frame to obtain a time-frequency spectrum composed of multiple frequency components and corresponding energy values;

[0024] extracting a key waveform feature representing an acoustic event from the time-frequency spectrum to form an acoustic feature waveform;

[0025] calculating a first matching degree of the acoustic feature waveform with a preset short-circuit acoustic feature template and a second matching degree of the acoustic feature waveform with a preset arc acoustic feature template, and marking a channel as an acoustic abnormal channel if the first matching degree or the second matching degree exceeds a preset acoustic matching threshold.

[0026] Optionally, frame processing on the acoustic signal to obtain multiple acoustic frames, and adding a preset window function to each acoustic frame to obtain a windowed acoustic frame, including:

[0027] determining a single-frame duration based on a sampling period of the acoustic signal, frame processing on the acoustic signal based on the single-frame duration to obtain multiple initial acoustic frames, performing zero padding at the end of an initial acoustic frame with a duration less than the single-frame duration to obtain an acoustic frame, arranging all the acoustic frames in a frame processing order to obtain an effective acoustic frame sequence;

[0028] adding a preset window function to each acoustic frame in the effective acoustic frame sequence to obtain a windowed acoustic frame.

[0029] Optionally, according to the channels labeled as voltage abnormal channels and acoustic abnormal channels at the same time, combined with the channel physical position relationship, an abnormal correlation matrix is constructed, and a correlation group in which the abnormal correlation strength of the channels in the group exceeds a preset strong correlation threshold is selected from the abnormal correlation matrix to determine a failure channel set, including:

[0030] All channels labeled as voltage abnormal channels and acoustic abnormal channels at the same time are summarized to obtain a double abnormal channel set;

[0031] The preset channel physical position relationship is read, and the physical distance and abnormal trigger time window of any two channels in the double abnormal channel set are extracted;

[0032] If there is an overlapping period between the two abnormal trigger time windows, the corresponding channels are labeled as time overlapping state, and if there is no overlapping period, the corresponding channels are labeled as no overlapping state;

[0033] The corresponding basic correlation degree of the physical distance is queried from the preset correlation degree mapping table, and when the two channels are in the time overlapping state, the final correlation degree is determined as the basic correlation degree, and when the two channels are in the no overlapping state, the final correlation degree is determined as zero;

[0034] Based on the final correlation degree of all channel pairs, an abnormal correlation matrix is constructed;

[0035] The channel pairs corresponding to the elements in the abnormal correlation matrix whose element values are greater than a preset strong correlation threshold are labeled as correlation pairs, and the correlation pairs containing at least one same channel are merged to form a correlation group;

[0036] The channels of each correlation group are extracted to constitute a failure channel set.

[0037] Optionally, according to the historical failure feature database, the spatial distribution features of the failure channel set are matched with historical failure cases to obtain an identification report containing failure positioning position, abnormal type classification result, and short circuit risk warning level, including:

[0038] Based on the physical positions of all channels in the failure channel set, a first spatial distribution feature is formed;

[0039] The similarity between the spatial distribution feature and the first spatial distribution feature of the historical failure case is calculated, the target failure case with the highest similarity is selected, and the failure position of the target failure case is taken as the failure positioning position;

[0040] According to the abnormal type label of each channel in the failure channel set, the channel quantity proportion of each abnormal type is counted to form an abnormal type classification result;

[0041] selecting, from a historical failure feature database, historical risk propagation data matching an abnormal level and a corresponding abnormal type of an associated group to which each channel in the failure channel set belongs, and calculating a risk probability of the affected channel according to the historical risk propagation data to generate a short-circuit risk warning level;

[0042] integrating the failure positioning position, the abnormal type classification result, and the short-circuit risk warning level to generate an identification report.

[0043] In a second aspect, the present application provides a laser multi-channel sequential power-on failure detection system, comprising:

[0044] a collection module configured to collect driving voltage data, voltage abnormal duration, and acoustic wave signals of each channel during a laser multi-channel sequential power-on process;

[0045] a calculation module configured to calculate voltage variation amplitudes and voltage abnormal durations of the channels according to the driving voltage data, and mark a channel as a voltage abnormal channel if the voltage variation amplitude of the channel exceeds a preset voltage variation threshold or the voltage abnormal duration of the channel exceeds a preset time threshold;

[0046] a transformation module configured to perform short-time Fourier transform on each acoustic wave signal to obtain a time-frequency spectrum, extract acoustic feature waveforms from the time-frequency spectrum, and mark a channel as an acoustic abnormal channel if the matching degree between the acoustic feature waveform of the channel and a preset short-circuit acoustic feature template or a preset electric arc acoustic feature template exceeds a set threshold;

[0047] a construction module configured to construct an abnormal association matrix according to channels that are simultaneously marked as voltage abnormal channels and acoustic abnormal channels, select an associated group in which the abnormal association strength of the channels exceeds a preset strong correlation threshold from the abnormal association matrix, and determine a failure channel set in combination;

[0048] a matching module configured to match spatial distribution features of the failure channel set with historical failure cases according to a historical failure feature database to obtain an identification report containing a failure positioning position, an abnormal type classification result, and a short-circuit risk warning level.

[0049] In a third aspect, the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a laser multi-channel sequential power-on failure detection method as described in the first aspect above.

[0050] In a fourth aspect, the application provides a computer storage medium storing a computer program, which, when executed by a computer, implements the laser multi-channel sequential power-on failure detection method according to the first aspect.

[0051] In the application, during the laser multi-channel sequential power-on process, the driving voltage data, voltage abnormal duration and acoustic wave signal of each channel are collected; according to the driving voltage data, the voltage variation amplitude and voltage abnormal duration of each channel are calculated, and the channel whose voltage variation amplitude exceeds a preset voltage variation threshold or whose voltage abnormal duration exceeds a preset time threshold is marked as a voltage abnormal channel; each acoustic wave signal is subjected to short-time Fourier transform to obtain a time-frequency spectrum, and an acoustic characteristic waveform is extracted from the time-frequency spectrum, and the channel whose matching degree with a preset short-circuit acoustic feature template or a preset electric arc acoustic feature template exceeds a set threshold is marked as an acoustic abnormal channel; according to the channels that are simultaneously marked as voltage abnormal channels and acoustic abnormal channels, an abnormal correlation matrix is constructed in combination with the physical position relationship of the channels, and a correlation group whose channel abnormal correlation strength in the group exceeds a preset strong correlation threshold is selected from the abnormal correlation matrix to determine a failure channel set; according to a historical failure feature database, the spatial distribution features of the failure channel set are matched with historical failure cases to obtain an identification report containing a failure positioning position, an abnormal type classification result and a short-circuit risk warning level. The technical solution provided by the application realizes multi-dimensional data fusion of electrical states and physical acoustic responses, provides a comprehensive data basis for subsequent composite criterion analysis, improves the detection sensitivity of voltage disturbance, realizes accurate identification of specific fault acoustic fingerprints, overcomes the limitations of relying only on acoustic energy intensity judgment, improves the specificity of acoustic detection, can distinguish isolated interference from local faults with propagation characteristics, suppresses false positives and reveals potential chain failure risks.

[0052] Further, the application performs sliding window smoothing processing on the driving voltage data, extracts the change rate of the smoothed voltage sequence, dynamically determines the reference voltage value in the low fluctuation range, and then calculates the real-time voltage deviation; based on a preset deviation trigger threshold, the start and end points of voltage fluctuation are marked to define the voltage variation interval, and the difference between the maximum and minimum voltage values in the interval is extracted as the voltage variation amplitude, and the voltage abnormal duration of the interval is calculated; finally, the voltage variation amplitude and the voltage abnormal duration are combined to make a comprehensive judgment on whether the channel is abnormal. The problem of high false positive rate and weak anti-interference ability caused by single threshold judgment in the existing scheme is solved, and a more accurate electrical fault identification mechanism is provided for reliable power-on monitoring of multi-channel lasers in complex electromagnetic environments.

[0053] These and other aspects of the application will be more apparent in the following description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are only some embodiments of the present application, and do not represent all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of the present application.

[0055] Figure 1 A flow chart of a laser multi-channel sequential power-on failure detection method provided by the present application;

[0056] Figure 2 A structural schematic diagram of a laser multi-channel sequential power-on failure detection system provided by the present application;

[0057] Figure 3 A structural schematic diagram of a computing device provided by the present application. DETAILED DESCRIPTION

[0058] In order to make the person skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application.

[0059] In some of the descriptions in the specification and claims of the present application and the above-mentioned drawings, a plurality of operations appearing in a specific order are included, but it should be clearly understood that these operations can be executed or in parallel without the order appearing in the text, and the serial numbers of the operations such as 101, 102, etc. are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel. It should be noted that the "first", "second", etc. in the text are used to distinguish different messages, devices, modules, etc., and do not represent the order, nor limit that the "first" and "second" are different types.

[0060] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of the present application.

[0061] Aiming at the problem of local short circuit or arc discharge fault caused by aging of components, insulation failure and other reasons in the process of sequential power-on of high-power multi-channel laser, the application synchronously collects the driving voltage time sequence data and acoustic signal of each channel, uses the voltage change amplitude and abnormal duration double criteria to improve the stability of electrical abnormality detection, introduces short-time Fourier transform for time-frequency analysis of acoustic signal; based on the voltage and acoustic double-mode abnormality marking results, introduces the channel physical position relationship to construct an abnormal correlation matrix, identifies the abnormal channel group with spatial aggregation and synchronization, and suppresses the misjudgment caused by isolated noise or transient interference; outputs a comprehensive diagnostic report containing fault location, type classification and risk level, improves the accuracy, robustness and intelligent level of laser power-on fault detection under complex working conditions.

[0062] Figure 1 A flow chart of a laser multi-channel sequential power-on failure detection method is provided for the embodiments of the application, as shown in Figure 1 The method comprises:

[0063] Step 101: In the process of sequential power-on of laser multi-channel, collect the driving voltage data, voltage abnormal duration and acoustic signal of each channel.

[0064] In this step, the driving voltage data refers to the voltage value sequence generated by each channel in the driving process in the process of sequential power-on of laser multi-channel, which is used to calculate the voltage change amplitude to judge whether the voltage is abnormal. The voltage abnormal duration refers to the length of time that the voltage of the channel is in an abnormal state, which is used to assist in judging whether the channel is a voltage abnormal channel. The acoustic signal refers to the waveform data of the sound signal generated by each channel when working in the process of sequential power-on of laser multi-channel, which is used to extract acoustic characteristic waveform to judge whether the sound is abnormal.

[0065] In the embodiments of the application, in the process of sequential power-on of laser multi-channel, the driving voltage data, voltage abnormal duration and acoustic signal of each channel are collected in real time by a data acquisition device.

[0066] Step 102: According to the driving voltage data, calculate the voltage change amplitude and voltage abnormal duration of each channel, and mark the channel whose voltage change amplitude exceeds the preset voltage change threshold or voltage abnormal duration exceeds the preset time threshold as a voltage abnormal channel.

[0067] In this step, the voltage change amplitude refers to the difference between the maximum voltage value and the minimum voltage value in the voltage change interval in the channel driving voltage data, which is used for comparison with the preset voltage change threshold to determine whether the voltage is abnormal. The preset voltage change threshold refers to a critical value set based on the voltage fluctuation range when the laser is working normally, which is used to determine whether the voltage change amplitude exceeds the normal range. The preset time threshold refers to a critical value set based on the allowed duration of voltage abnormality when the laser is working normally, which is used to determine whether the voltage abnormality duration exceeds the normal range. The voltage abnormal channel refers to the channel in the driving voltage data whose voltage change amplitude exceeds the preset voltage change threshold or whose voltage abnormality duration exceeds the preset time threshold, which is used for subsequent cross-validation with the acoustic abnormal channel.

[0068] Step 103: Perform short-time Fourier transform on each acoustic signal to obtain a time-frequency spectrum, and extract acoustic feature waveforms from the time-frequency spectrum. Channels whose matching degree with the preset short-circuit acoustic feature template or the preset arc acoustic feature template exceeds a set threshold are marked as acoustic abnormal channels.

[0069] In this step, the time-frequency spectrum refers to the spectrum diagram obtained by performing short-time Fourier transform on the acoustic signal, which can reflect the change of signal frequency and energy over time, and is used to extract acoustic feature waveforms. The acoustic feature waveform refers to the waveform extracted from the time-frequency spectrum, which can represent the key characteristics of the acoustic signal, and is used for matching with the preset acoustic feature template. The preset short-circuit acoustic feature template refers to the acoustic feature waveform template extracted from the acoustic signal collected in historical short-circuit failure cases, which is used for matching with the current acoustic feature waveform to determine whether it is a short circuit. The preset arc acoustic feature template refers to the acoustic feature waveform template extracted from the acoustic signal collected in historical arc failure cases, which is used for matching with the current acoustic feature waveform to determine whether it is an arc. The set threshold refers to a critical value set based on historical acoustic matching data, which is used to determine the matching degree of the acoustic feature waveform with the preset acoustic feature template, and is used to mark the acoustic abnormal channel. The acoustic abnormal channel refers to the channel whose matching degree with the preset short-circuit acoustic feature template or the preset arc acoustic feature template exceeds the set threshold, which is used for subsequent cross-validation with the voltage abnormal channel.

[0070] Step 104: According to the channels that are simultaneously marked as voltage abnormal channels and acoustic abnormal channels, construct an abnormal association matrix based on the physical location relationship of the channels, and select from the abnormal association matrix an association group whose intra-group channel abnormal association strength exceeds a preset strong correlation threshold, to determine the failure channel set.

[0071] In this step, the abnormal association matrix refers to a matrix constructed according to the channels labeled as voltage abnormal channels and acoustic abnormal channels at the same time, and the correlation strength between channels is calculated in combination with the physical position relationship, which is used for screening the association group. The preset strong correlation threshold refers to a critical value set based on historical channel association data, which is used to judge whether the correlation strength between channels is strong enough, and is used to select the association group from the abnormal association matrix. The association group refers to a combination of channels in the abnormal association matrix, the abnormal association strength of which in the group exceeds the preset strong correlation threshold, which is used to form the failed channel set. The failed channel set refers to a set obtained by aggregating the association groups, which is used for matching analysis with historical failure cases.

[0072] Step 105: According to the historical failure feature database, the spatial distribution features of the failed channel set are matched with the historical failure cases to obtain an identification report containing the failure positioning location, abnormal type classification result and short circuit risk warning level.

[0073] In this step, the historical failure feature database refers to a database storing historical failure related data of the laser, which is used to provide historical reference for current failure analysis. The historical failure case refers to a specific case recorded in the historical failure feature database when the laser failed in the past, which is used to match the spatial distribution features of the current failed channel set. The failure positioning location refers to the specific location where the current failure occurs, which is determined by matching the spatial distribution features of the failed channel set with the historical failure cases, and is used for failure positioning in the identification report. The abnormal type classification result refers to the result obtained by statistically classifying the abnormal types of each association group in the failed channel set, which is used for abnormal classification in the identification report. The short circuit risk warning level refers to the level reflecting the risk degree of the possible spread of the current failure, which is calculated based on historical risk spread data, and is used for risk warning in the identification report. The identification report refers to the report generated by integrating the failure positioning location, abnormal type classification result and short circuit risk warning level, which is used to provide complete information of the failure detection.

[0074] The embodiments of the present application realize accurate detection, positioning and warning of composite defects by collecting multi-dimensional data during the power-on process of the laser multi-channel sequence, combining voltage and acoustic dual abnormality judgment, constructing an abnormal association matrix to screen the association group, and generating an identification report by matching with historical data, thereby improving the accuracy and systematicness of the laser multi-channel failure detection.

[0075] The present application provides a specific embodiment, step 102, according to the driving voltage data, the voltage change amplitude and the voltage abnormal duration of each channel are calculated, the channels with voltage change amplitude exceeding the preset voltage change threshold or voltage abnormal duration exceeding the preset time threshold are marked as voltage abnormal channels, which specifically includes the following steps:

[0076] Step 201: performing sliding window smoothing processing on the driving voltage data to obtain a smoothed voltage sequence, and calculating a change rate of the smoothed voltage sequence.

[0077] In this step, the smoothed voltage sequence refers to a voltage sequence obtained by performing sliding window smoothing processing on the driving voltage data, which is used for subsequent calculation of the change rate and the real-time voltage deviation value. The change rate refers to a ratio of a difference between two adjacent smoothed voltage values in the smoothed voltage sequence to a time interval corresponding to the collection time, which is used for judging whether the voltage is in a stable state.

[0078] In the embodiment of the present application, the driving voltage data is processed by the sliding window smoothing processing technology to obtain the smoothed voltage sequence, and then the change rate of the smoothed voltage sequence is obtained by calculating the ratio of the difference between two adjacent smoothed voltage values to the corresponding time interval, and the calculation formula is: change rate = (next smoothed voltage value - previous smoothed voltage value) / time interval.

[0079] Step 202: when the change rate is in a preset low fluctuation range, taking the voltage value corresponding to the change rate as a reference voltage value, and calculating a real-time voltage deviation value between each voltage value in the smoothed voltage sequence and the reference voltage value.

[0080] In this step, the preset low fluctuation range refers to an interval set based on the fluctuation range of the voltage change rate when the laser works normally, which is used for determining the reference voltage value. The reference voltage value refers to the voltage value corresponding to the change rate when the change rate is in the preset low fluctuation range, which is used for calculating the real-time voltage deviation value. The real-time voltage deviation value refers to the difference between each voltage value in the smoothed voltage sequence and the reference voltage value, which is used for judging whether the voltage fluctuates.

[0081] In the embodiment of the present application, when the change rate of the smoothed voltage sequence is in the preset low fluctuation range, the voltage value corresponding to the change rate is taken as the reference voltage value, and the real-time voltage deviation value is obtained by calculating the difference between each voltage value in the smoothed voltage sequence and the reference voltage value, and the calculation formula is: real-time voltage deviation value = smoothed voltage value - reference voltage value.

[0082] Step 203: marking a time point when the real-time voltage deviation value exceeds a preset deviation triggering threshold as a voltage fluctuation starting point, marking a time point when the real-time voltage deviation value falls from the exceeding state to a state of not exceeding the preset deviation triggering threshold for the first time as a voltage fluctuation termination point, and determining a voltage change interval in combination with the voltage fluctuation starting point.

[0083] In this step, the preset deviation trigger threshold refers to a critical value set based on the maximum deviation allowed by the voltage when the laser is normally working, which is used to mark the voltage fluctuation starting point. The voltage fluctuation starting point refers to the time when the real-time voltage deviation value exceeds the preset deviation trigger threshold, which is used to determine the starting point of the voltage change interval. The voltage fluctuation termination point refers to the time when the real-time voltage deviation value falls from the exceeding state to not more than the preset deviation trigger threshold for the first time, which is used to determine the end point of the voltage change interval. The voltage change interval refers to the time interval delimited by the voltage fluctuation starting point and the voltage fluctuation termination point, which is used to extract the maximum voltage value and the minimum voltage value.

[0084] In the embodiments of the present application, the real-time voltage deviation value of each voltage value in the smoothed voltage sequence and the reference voltage value is monitored in real time. When the real-time voltage deviation value exceeds the preset deviation trigger threshold for the first time, the time is marked as the voltage fluctuation starting point. The change of the real-time voltage deviation value is continuously monitored. When the real-time voltage deviation value falls from the state of exceeding the preset deviation trigger threshold to not more than the preset deviation trigger threshold for the first time, the time is marked as the voltage fluctuation termination point. The time interval between the voltage fluctuation starting point and the voltage fluctuation termination point is determined as the voltage change interval.

[0085] Step 204: Extract the maximum voltage value and the minimum voltage value from the voltage change interval, and calculate the difference between the maximum voltage value and the minimum voltage value, which is taken as the voltage change amplitude.

[0086] In this step, the maximum voltage value refers to the maximum voltage value extracted from the voltage change interval, which is used to calculate the voltage change amplitude. The minimum voltage value refers to the minimum voltage value extracted from the voltage change interval, which is used to calculate the voltage change amplitude.

[0087] In the embodiments of the present application, the maximum voltage value and the minimum voltage value are extracted from the voltage change interval, and the voltage change amplitude is obtained by calculating the difference between the maximum voltage value and the minimum voltage value.

[0088] Step 205: Calculate the voltage abnormality duration of the voltage change interval, and mark the channel corresponding to the voltage change amplitude exceeding the preset voltage change threshold or the voltage abnormality duration exceeding the preset time threshold as the voltage abnormality channel.

[0089] In the embodiments of the present application, the time length of the voltage change interval is calculated to obtain the voltage abnormality duration, and the voltage change amplitude is compared with the preset voltage change threshold, and the voltage abnormality duration is compared with the preset time threshold. If any of them exceeds, the corresponding channel is marked as the voltage abnormality channel.

[0090] The embodiment of the application realizes accurate marking of the voltage abnormal channel by performing smoothing processing on the driving voltage data, determining the reference voltage and the voltage variation interval, and accurately calculating the voltage variation amplitude and the voltage abnormal duration.

[0091] The application provides an embodiment, step 201, performing sliding window smoothing processing on the driving voltage data to obtain a smoothed voltage sequence, specifically including the following steps:

[0092] Step 211: extracting the voltage collection time and the corresponding voltage value from the driving voltage data, and arranging all the voltage values to obtain a time series voltage sequence.

[0093] In this step, the voltage collection time refers to the corresponding time point when the driving voltage data is collected, which is used to construct the time series voltage sequence. The time series voltage sequence refers to the sequence of voltage values arranged in the order of voltage collection time, which is used for subsequent division of data segments.

[0094] In the embodiment of the application, the voltage collection time and the voltage value corresponding to each voltage are extracted from the driving voltage data, and all the voltage values are arranged in the order of voltage collection time to obtain the time series voltage sequence.

[0095] Step 212: calculating the average value of the time interval of adjacent voltage collection times, taking the average value of the time interval as the sampling period, and determining the sliding window length in combination with the preset window time length.

[0096] In this step, the average value of the time interval refers to the value obtained by calculating the arithmetic mean of the time interval of adjacent two voltage collection times in the time series voltage sequence, which is used to determine the sampling period. The sampling period refers to the average time interval of adjacent two voltage collections determined based on the average value of the time interval, which is used to calculate the sliding window length. The preset window time length refers to the time length covered by each sliding window, which is used to determine the sliding window length. The sliding window length refers to the number of voltage values contained in each sliding window, which is used to divide the data segments.

[0097] In the embodiment of the application, the time interval of adjacent two voltage collection times in the time series voltage sequence is calculated, the average value of all time intervals is calculated by calculating the arithmetic mean, the calculation formula is: time interval average value=(sum of all adjacent time intervals) ÷ number of adjacent time intervals, the average value of the time interval is taken as the sampling period, and the sliding window length is obtained by dividing the window time length by the sampling period in combination with the preset window time length, the calculation formula is: sliding window length=preset window time length ÷ sampling period.

[0098] Step 213: based on the sliding window length, the time sequence of voltages is divided into multiple data segments, and the average voltage value of all voltage values in each data segment is calculated, the average voltage value of each data segment is taken as a smoothed voltage value, and all smoothed voltage values are spliced to form a smoothed voltage sequence.

[0099] In this step, the data segment refers to a combination of continuous voltage values obtained from the time sequence of voltages based on the sliding window length, which is used to calculate the average voltage value. The average voltage value refers to the value obtained by calculating the arithmetic mean of all voltage values in the data segment, which is used as the smoothed voltage value. The smoothed voltage value refers to the average voltage value of the data segment as the smoothed value corresponding to the window, which is used to construct the smoothed voltage sequence.

[0100] In the embodiment of the present application, based on the determined sliding window length, voltage values equal to the sliding window length are sequentially intercepted from the starting position of the time sequence of voltages as data segments, the sum of all voltage values in each data segment is divided by the number of voltage values to obtain the average voltage value, and the calculation formula is: average voltage value = (sum of all voltage values in the data segment) / number of voltage values in the data segment. The average voltage value of each data segment is taken as a smoothed voltage value, and all smoothed voltage values are spliced in the order of data segment interception to form a smoothed voltage sequence.

[0101] The embodiment of the present application arranges voltage values in time sequence to form a time sequence of voltages, determines the sampling period and the sliding window length based on the collection interval, and then calculates the average voltage value by the sliding window to construct a smoothed voltage sequence, thereby achieving smoothing processing of the driving voltage data, reducing noise interference in the original data, and providing a stable and reliable data basis for subsequent voltage change analysis.

[0102] The present application provides a specific embodiment, step 103, performing short-time Fourier transform on each acoustic signal to obtain a time-frequency spectrum, and extracting an acoustic characteristic waveform from the time-frequency spectrum. The acoustic characteristic waveform is marked as an acoustic abnormal channel if the matching degree between the acoustic characteristic waveform and the preset short-circuit acoustic feature template or the preset arc acoustic feature template exceeds a set threshold, and the specific steps include the following steps:

[0103] Step 301: frame processing is performed on the acoustic signal to obtain a plurality of acoustic frames, and a preset window function is added to each acoustic frame to obtain a windowed acoustic frame.

[0104] In this step, the acoustic frame refers to a short-time signal segment obtained by frame processing the acoustic signal, which is used for subsequent window function processing. The preset window function refers to a function that is preset for smoothing processing of the acoustic frame, and the amplitude thereof changes smoothly from both ends to the middle, thereby improving the accuracy of subsequent decomposition. The windowed acoustic frame refers to a signal segment obtained by multiplying the amplitude of the acoustic frame point by point with the preset window function, which is used for time-frequency spectrum decomposition.

[0105] In the embodiment of the present application, the continuous acoustic wave signal collected is segmented by a frame processing technology according to a set single frame length to obtain a plurality of continuous acoustic wave frames that may have overlaps; a preset window function is called to multiply the amplitude of each time point of each acoustic wave frame with the amplitude of the corresponding time point of the window function point by point to obtain a windowed acoustic wave frame.

[0106] Step 302: The windowed acoustic wave frame is decomposed to obtain a time-frequency spectrum composed of a plurality of frequency components and corresponding energy values.

[0107] In the embodiment of the present application, the windowed acoustic wave frame is decomposed by a short-time Fourier transform technology to obtain a time-frequency spectrum composed of a plurality of frequency components and corresponding energy values of each frequency component.

[0108] Step 303: Key waveform features representing acoustic events are extracted from the time-frequency spectrum to form an acoustic feature waveform.

[0109] In this step, the key waveform features refer to feature parameters extracted from the time-frequency spectrum that can represent the nature of acoustic events, which are used to form an acoustic feature waveform.

[0110] In the embodiment of the present application, key waveform features representing acoustic events are extracted from the time-frequency spectrum, and these features are integrated to form an acoustic feature waveform.

[0111] Step 304: The first matching degree of the acoustic feature waveform and a preset short-circuit acoustic feature template and the second matching degree of the acoustic feature waveform and a preset arc acoustic feature template are calculated, and a channel corresponding to a first matching degree or a second matching degree exceeding a preset acoustic matching threshold is marked as an acoustic abnormal channel.

[0112] In this step, the first matching degree refers to the degree of similarity between the acoustic feature waveform and the preset short-circuit acoustic feature template, which is used to determine whether it is a short-circuit related acoustic anomaly. The second matching degree refers to the degree of similarity between the acoustic feature waveform and the preset arc acoustic feature template, which is used to determine whether it is an arc related acoustic anomaly. The preset acoustic matching threshold refers to a critical value preset for determining whether the matching degree meets the standard, which is used to mark the acoustic abnormal channel.

[0113] In the embodiment of the present application, the first matching degree of the acoustic feature waveform and the preset short-circuit acoustic feature template is calculated by a matching algorithm, and the calculation formula is: first matching degree=(number of coincident features of the acoustic feature waveform and the preset short-circuit acoustic feature template) / total number of features, and the second matching degree of the acoustic feature waveform and the preset arc acoustic feature template is calculated, and the calculation formula is: second matching degree=(number of coincident features of the acoustic feature waveform and the preset arc acoustic feature template) / total number of features, and a channel whose first matching degree or second matching degree exceeds the preset acoustic matching threshold is marked as an acoustic abnormal channel.

[0114] The embodiment of the application realizes accurate marking of the acoustic abnormal channel by frame dividing, windowing and time-frequency spectrum decomposition of the sound wave signal, extracting features and matching with a preset template, forms double verification in combination with voltage abnormality detection, and improves the accuracy of multi-channel failure detection of the laser.

[0115] The application provides an embodiment, step 301, frame dividing processing is performed on the sound wave signal to obtain a plurality of sound wave frames, a preset window function is added to each sound wave frame to obtain a windowed sound wave frame, and the specific steps include the following steps:

[0116] Step 311: based on the sampling period of the sound wave signal, the single-frame time length is determined, the sound wave signal is frame divided according to the single-frame time length to obtain a plurality of initial sound wave frames, the initial sound wave frame with a time length less than the single-frame time length is zero-padded at the end to obtain a sound wave frame, and all sound wave frames are arranged in a frame dividing order to obtain an effective sound wave frame sequence.

[0117] In this step, the single-frame time length is determined based on the sampling period of the sound wave signal, the time length of each sound wave frame, and is used to unify the time length standard of the sound wave frame. The initial sound wave frame refers to the original signal segment obtained by frame dividing the sound wave signal, and the time length thereof may be less than the single-frame time length and is used for subsequent zero padding. The effective sound wave frame sequence refers to a sequence formed by arranging all sound wave frames after the end zero padding in a frame dividing order, and is used for subsequent window function addition processing.

[0118] In the embodiment of the application, based on the sampling period of the sound wave signal, the single-frame time length is determined by setting the single-frame time length as a fixed multiple of the sampling period, the calculation formula is: single-frame time length = sampling period x fixed multiple, the sound wave signal is frame divided according to the single-frame time length to obtain a plurality of initial sound wave frames, the initial sound wave frame with a time length less than the single-frame time length is zero-padded at the end to supplement the time length to obtain a sound wave frame, and all sound wave frames are arranged in a frame dividing order to obtain an effective sound wave frame sequence.

[0119] Step 312: a preset window function is added to each sound wave frame in the effective sound wave frame sequence to obtain a windowed sound wave frame.

[0120] In the embodiment of the application, each sound wave frame in the effective sound wave frame sequence is obtained, a preset window function is called, the amplitude of each time point in the sound wave frame is multiplied point by point with the amplitude of the corresponding time point of the window function, and a windowed sound wave frame after the window function processing of each sound wave frame is obtained.

[0121] The embodiment of the application realizes the standardization segmentation and smoothing processing of the sound wave signal by determining the single frame length based on the sampling period, zero padding the sound wave signal after segmentation, and adding window function processing, reduces the signal edge distortion, provides stable and consistent basic data for subsequent time-frequency spectrum analysis, and improves the accuracy of acoustic feature extraction.

[0122] The application provides an embodiment, step 104, constructing an abnormal correlation matrix according to the channels labeled as voltage abnormal channels and acoustic abnormal channels at the same time, combining the channel physical position relationship, and selecting a correlation group with an abnormal correlation strength in a group exceeding a preset strong correlation threshold from the abnormal correlation matrix to determine a failure channel set, specifically including the following steps:

[0123] Step 401: All channels labeled as voltage abnormal channels and acoustic abnormal channels at the same time are summarized to obtain a double abnormal channel set.

[0124] In this step, the double abnormal channel set refers to a channel set labeled as voltage abnormal channels and acoustic abnormal channels at the same time, which is used for subsequent analysis of the correlation between channels.

[0125] In the embodiment of the application, the identification information of all channels labeled as voltage abnormal channels and acoustic abnormal channels at the same time is summarized to obtain a double abnormal channel set containing these channels.

[0126] Step 402: The preset channel physical position relationship is read, and the physical distance and abnormal trigger time window of any two channels in the double abnormal channel set are extracted.

[0127] In this step, the preset channel physical position relationship refers to the position information of each channel of the laser in the physical space, which is used to extract the physical distance between channels. The physical distance refers to the straight-line distance between two channels in the physical space, which is used to determine the basic correlation degree. The abnormal trigger time window refers to the time interval from the starting time to the ending time when the channel is labeled as abnormal, which is used to determine the time correlation between channels.

[0128] In the embodiment of the application, the physical distance between any two channels in the double abnormal channel set and the abnormal trigger time window of each channel are extracted from the preset channel physical position relationship.

[0129] Step 403: If there is an overlapping period between the two abnormal trigger time windows, the corresponding channel is labeled as time overlap state, and if there is no overlapping period, the corresponding channel is labeled as no overlap state.

[0130] In the embodiment of the present application, the abnormal triggering time windows of the two channels are compared, if there is an overlapping period, the two channels are marked as time overlapping state, if there is no overlapping period, it is marked as no overlapping state.

[0131] Step 404: querying the basic correlation degree corresponding to the physical distance from the preset correlation degree mapping table, when the two channels are in the time overlapping state, the final correlation degree is determined as the basic correlation degree, and when the two channels are in the no overlapping state, the final correlation degree is determined as zero.

[0132] In this step, the preset correlation degree mapping table refers to the correspondence table of the physical distance and the basic correlation degree, wherein the smaller the physical distance, the greater the corresponding basic correlation degree, which is used to query the basic correlation degree. The basic correlation degree refers to the correlation degree value corresponding to the physical distance queried from the preset correlation degree mapping table, which is used to calculate the final correlation degree. The final correlation degree refers to the correlation degree obtained by adjusting the basic correlation degree in combination with the time overlapping state, which is equal to the basic correlation degree when the time is overlapping, and is zero otherwise, including the correlation strength reflecting the space-time coupling between channels, which is used to construct the abnormal correlation matrix.

[0133] In the embodiment of the present application, the basic correlation degree corresponding to the physical distance of the two channels is queried from the preset correlation degree mapping table, when the two channels are in the time overlapping state, the basic correlation degree is determined as the final correlation degree, and when the two channels are in the no overlapping state, the final correlation degree is determined as zero.

[0134] Step 405: constructing an abnormal correlation matrix based on the final correlation degree of all channel pairs.

[0135] In the embodiment of the present application, a two-dimensional abnormal correlation matrix is constructed in the order of channel identification based on the final correlation degree of all channel pairs, and the matrix element value is the final correlation degree of the corresponding channel pair.

[0136] Step 406: marking the channel pair corresponding to the element with an element value greater than a preset strong correlation threshold in the abnormal correlation matrix as a correlation pair, and merging the correlation pairs containing at least one same channel to form a correlation group.

[0137] In this step, the correlation pair refers to the combination of the two channels corresponding to the element with an element value greater than a preset strong correlation threshold in the abnormal correlation matrix, which is used to form a correlation group.

[0138] In the embodiment of the present application, the channel pair corresponding to the element with an element value greater than a preset strong correlation threshold in the abnormal correlation matrix is marked as a correlation pair, and the correlation pairs containing at least one same channel are merged to form a correlation group.

[0139] Step 407: extracting the channels of each correlation group to constitute a failure channel set.

[0140] In the embodiment of the present application, the channels contained in each correlation group are extracted, and the channels are integrated to form a failure channel set.

[0141] In the embodiment of the present application, the final correlation degree between channels is calculated by summarizing double abnormal channels, combining physical positions and time windows, constructing an abnormal correlation matrix and determining a correlation group, and finally forming a failure channel set, which realizes the analysis of the spatial and temporal coupling relationship between multiple channels and accurately identifies the joint failure channels with strong correlation.

[0142] The present application provides a specific embodiment, step 105, according to the historical failure feature database, matching the spatial distribution characteristics of the failure channel set with the historical failure cases to obtain an identification report containing the failure positioning position, abnormal type classification result and short circuit risk warning level, which specifically includes the following steps:

[0143] Step 501: Forming a first spatial distribution feature based on the physical positions of all channels in the failure channel set.

[0144] In this step, the first spatial distribution feature refers to the spatial distribution feature formed based on the physical positions of all channels in the failure channel set, which is used to match the spatial distribution feature of the historical failure case.

[0145] In the embodiment of the present application, based on the physical position information of all channels in the failure channel set, the distribution characteristics of these positions, such as coordinate distribution and density, are integrated to form the first spatial distribution feature.

[0146] Step 502: Calculate the similarity of the spatial distribution feature and the first spatial distribution feature of the historical failure case, and select the target failure case with the highest similarity, and take the failure position of the target failure case as the failure positioning position.

[0147] In this step, the target failure case refers to the case with the highest similarity to the first spatial distribution feature selected from the historical failure cases, which is used to determine the positioning position of the current failure.

[0148] In the embodiment of the present application, the first spatial distribution feature is extracted, and the spatial distribution feature of each case in the historical failure case is read; through the feature matching algorithm, the similarity of the first spatial distribution feature and the spatial distribution feature of each historical failure case is calculated from the position overlap ratio, the density deviation value, the coverage range matching degree and other dimensions; all similarities are compared, and the historical failure case with the largest similarity value is selected as the target failure case, and the failure position recorded in the target failure case is determined as the current failure positioning position.

[0149] Step 503: According to the abnormal type label of each channel in the failure channel set, the proportion of the number of channels of each abnormal type to the total number of channels is counted to form an abnormal type classification result.

[0150] In this step, the abnormal type label refers to the labeling of the abnormal type of each channel in the failure channel set, which is used to count the abnormal type classification result.

[0151] In the embodiments of the present application, according to the abnormal type label of each channel in the failure channel set, the proportion of the number of channels of each abnormal type to the total number of channels is counted to form an abnormal type classification result.

[0152] Step 504: From the historical failure feature database, historical risk diffusion data matching the abnormal level and corresponding abnormal type of the associated group to which each channel in the failure channel set belongs is selected, and according to the historical risk diffusion data, the risk probability of the affected channel is calculated to generate a short-circuit risk warning level.

[0153] In this step, the abnormal level refers to the abnormal severity level of the associated group in the failure channel set, which is determined based on the comparison of the number of channels in the associated group, the average spatial correlation degree and the preset level division standard, and is used to match the historical risk diffusion data. The historical risk diffusion data refers to the abnormal diffusion related data corresponding to a specific abnormal level and abnormal type stored in the historical failure feature database, which is used to calculate the current risk probability. The risk probability refers to the possibility of abnormal diffusion to other channels calculated based on the historical risk diffusion data, which is used to generate a short-circuit risk warning level.

[0154] In the embodiments of the present application, from the historical failure feature database, historical risk diffusion data matching the abnormal level and corresponding abnormal type of the associated group to which each channel in the failure channel set belongs is selected, and based on the historical risk diffusion data, the risk probability of abnormal diffusion to other channels is calculated, and the calculation formula is: risk probability=(number of times of the same type of abnormality diffusing to the target channel in history) / total number of times of the same type of abnormality, to obtain the risk probability of the affected channel, and then generate a short-circuit risk warning level.

[0155] Step 505: The failure positioning position, the abnormal type classification result and the short-circuit risk warning level are integrated to generate an identification report.

[0156] In the embodiments of the present application, the failure positioning position, the abnormal type classification result and the short-circuit risk warning level are integrated according to a preset format to generate an identification report containing the failure positioning position, the abnormal type classification result and the short-circuit risk warning level.

[0157] The embodiment of the application realizes accurate positioning, classification and early warning of failure by matching the spatial distribution characteristics of the failure channel set with historical data, combining abnormal type statistics and risk probability calculation, and generating an identification report containing multi-dimensional information, thereby improving the systematicness and forwardness of laser multi-channel failure detection.

[0158] Figure 2 A structural schematic diagram of a laser multi-channel sequential power-up failure detection system is provided for the embodiment of the application, as shown in the figure, the system comprises: Figure 2

[0159] The acquisition module 21 is configured to acquire driving voltage data, voltage abnormal duration and acoustic wave signals of each channel during the laser multi-channel sequential power-up process.

[0160] The calculation module 22 is configured to calculate voltage variation amplitude and voltage abnormal duration of each channel according to the driving voltage data, and mark a channel as a voltage abnormal channel if the voltage variation amplitude exceeds a preset voltage variation threshold or the voltage abnormal duration exceeds a preset time threshold.

[0161] The transformation module 23 is configured to perform short-time Fourier transform on each acoustic wave signal to obtain a time-frequency spectrum, extract an acoustic characteristic waveform from the time-frequency spectrum, and mark a channel as an acoustic abnormal channel if the matching degree between the acoustic characteristic waveform and a preset short-circuit acoustic characteristic template or a preset electric arc acoustic characteristic template exceeds a set threshold.

[0162] The construction module 24 is configured to construct an abnormal correlation matrix according to the channels that are marked as voltage abnormal channels and acoustic abnormal channels at the same time, and select a correlation group in which the abnormal correlation strength of the channels exceeds a preset strong correlation threshold from the abnormal correlation matrix, and determine a failure channel set in combination.

[0163] The matching module 25 is configured to match the spatial distribution characteristics of the failure channel set with historical failure cases according to a historical failure feature database to obtain an identification report containing failure positioning position, abnormal type classification result and short-circuit risk early warning level.

[0164] Figure 2 The laser multi-channel sequential power-up failure detection system can perform Figure 1 The implementation principle and technical effects of the laser multi-channel sequential power-up failure detection method described in the embodiment shown in the figure will not be described again. The specific operation of each module and unit of the laser multi-channel sequential power-up failure detection system described in the above embodiment has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0165] In one possible design,​Figure 2 A laser multi-channel sequential power-up failure detection system of the embodiments can be implemented as a computing device, such as Figure 3 As shown, the computing device can include a storage component 31 and a processing component 32.

[0166] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called for execution by the processing component 32.

[0167] The processing component 32 is configured to perform the above Figure 1 A laser multi-channel sequential power-up failure detection method of the embodiments.

[0168] The processing component 32 can include one or more processors to execute the computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more Application-Specific Integrated Circuit (ASIC), Digital Signal Process (DSP), Digital Signal Process Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor or other electronic elements, for executing the above method.

[0169] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or their combination, such as Random Access Memory (RAM), Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0170] Of course, the computing device can also include other components, such as input / output interfaces, display components, communication components, etc.

[0171] The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, etc.

[0172] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0173] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can be a cloud server, and the processing component, the storage component, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0174] The embodiment of the application further provides a computer storage medium storing a computer program, and the computer program can implement the above-mentioned Figure 1 A laser multi-channel sequential power-on failure detection method is provided.

[0175] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-mentioned system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0176] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0177] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and the necessary general hardware platform, and of course, it can also be realized by hardware. Based on this understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium such as ROM / RAM, magnetic disk, optical disk, etc., and includes a plurality of instructions for making a computer device (which can be a personal computer, server, or network device, etc.) execute the methods described in each embodiment or some parts of the embodiment.

[0178] Finally, it should be noted that the above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the same; although the present application has been described in detail with reference to the foregoing examples, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for laser multi-channel sequential power-up failure detection, the method comprising: The method comprises the following steps: During the multi-channel sequential power-on process of the laser, the driving voltage data, voltage abnormal duration and acoustic wave signal of each channel are collected; According to the driving voltage data, the voltage variation amplitude and voltage abnormal duration of each channel are calculated, and the channel whose voltage variation amplitude exceeds the preset voltage variation threshold or whose voltage abnormal duration exceeds the preset time threshold is marked as a voltage abnormal channel; Each acoustic wave signal is subjected to short-time Fourier transform to obtain a time-frequency spectrum, and an acoustic characteristic waveform is extracted from the time-frequency spectrum, and the channel whose matching degree with the preset short-circuit acoustic characteristic template or the preset arc acoustic characteristic template exceeds the set threshold is marked as an acoustic abnormal channel; According to the channels which are marked as voltage abnormal channels and acoustic abnormal channels at the same time, the abnormal correlation matrix is constructed in combination with the physical position relationship of the channels, and the correlation group whose abnormal correlation strength in the group exceeds the preset strong correlation threshold is selected from the abnormal correlation matrix to determine the failure channel set; According to the historical failure feature database, the spatial distribution features of the failure channel set are matched with the historical failure cases to obtain an identification report containing the failure positioning position, abnormal type classification result and short-circuit risk warning level.

2. The method of claim 1, wherein, According to the driving voltage data, the voltage variation amplitude and voltage abnormal duration of each channel are calculated, and the channel whose voltage variation amplitude exceeds the preset voltage variation threshold or whose voltage abnormal duration exceeds the preset time threshold is marked as a voltage abnormal channel, comprising: The driving voltage data is subjected to sliding window smoothing processing to obtain a smoothed voltage sequence, and the change rate of the smoothed voltage sequence is calculated; When the change rate is within the preset low fluctuation range, the voltage value corresponding to the change rate is taken as a reference voltage value, and the real-time voltage deviation value between each voltage value in the smoothed voltage sequence and the reference voltage value is calculated; The time point when the real-time voltage deviation value exceeds the preset deviation trigger threshold is marked as a voltage fluctuation starting point, the time point when the real-time voltage deviation value falls from the exceeding state to not more than the preset deviation trigger threshold for the first time is marked as a voltage fluctuation termination point, and the voltage variation interval is determined in combination with the voltage fluctuation starting point; The maximum voltage value and the minimum voltage value are extracted from the voltage variation interval, and the difference between the maximum voltage value and the minimum voltage value is calculated as the voltage variation amplitude; The voltage abnormal duration of the voltage variation interval is calculated, and the channel whose voltage variation amplitude exceeds the preset voltage variation threshold or whose voltage abnormal duration exceeds the preset time threshold is marked as a voltage abnormal channel.

3. The method of claim 2, wherein, The driving voltage data is subjected to sliding window smoothing processing to obtain a smoothed voltage sequence, comprising: The voltage collection time and the corresponding voltage value are extracted from the driving voltage data, and all the voltage values are arranged to obtain a time sequence voltage sequence; The time interval average value of adjacent voltage collection times is calculated, the time interval average value is taken as a sampling period, and the sliding window length is determined in combination with the preset window time length; Based on the sliding window length, the time sequence voltage sequence is divided into multiple data segments, and the average voltage value of all voltage values in each data segment is calculated, the average voltage value of each data segment is taken as the smoothed voltage value, and all smoothed voltage values are spliced to form a smoothed voltage sequence.

4. The method of claim 1, wherein, Each acoustic signal is subjected to short-time Fourier transform to obtain a time-frequency spectrum, and an acoustic feature waveform is extracted from the time-frequency spectrum, and a channel whose matching degree with a preset short-circuit acoustic feature template or a preset arc acoustic feature template exceeds a set threshold is marked as an acoustic abnormal channel, including: Frame processing is performed on the acoustic signal to obtain a plurality of acoustic frames, and a preset window function is added to each acoustic frame to obtain a windowed acoustic frame; Decompose the windowed acoustic frame to obtain a time-frequency spectrum composed of a plurality of frequency components and corresponding energy values; Key waveform features representing acoustic events are extracted from the time-frequency spectrum to form an acoustic feature waveform; The first matching degree of the acoustic feature waveform and the preset short-circuit acoustic feature template, and the second matching degree of the acoustic feature waveform and the preset arc acoustic feature template are calculated, and a channel whose first matching degree or second matching degree exceeds a preset acoustic matching threshold is marked as an acoustic abnormal channel.

5. The method of claim 3, wherein, Frame processing is performed on the acoustic signal to obtain a plurality of acoustic frames, and a preset window function is added to each acoustic frame to obtain a windowed acoustic frame, including: Based on the sampling period of the acoustic signal, a single frame duration is determined, and the acoustic signal is frame processed according to the single frame duration to obtain a plurality of initial acoustic frames, the initial acoustic frames with a duration less than the single frame duration are subjected to tail zero padding processing to obtain acoustic frames, and all acoustic frames are arranged in frame processing order to obtain an effective acoustic frame sequence; A preset window function is added to each acoustic frame in the effective acoustic frame sequence to obtain a windowed acoustic frame.

6. The method of claim 1, wherein, According to the channels marked as voltage abnormal channels and acoustic abnormal channels at the same time, an abnormal correlation matrix is constructed in combination with the physical position relationship of the channels, and a correlation group whose intra-group channel abnormal correlation strength exceeds a preset strong correlation threshold is selected from the abnormal correlation matrix to determine a failed channel set, including: All channels marked as voltage abnormal channels and acoustic abnormal channels at the same time are summarized to obtain a dual abnormal channel set; A preset physical position relationship of the channels is read, and the physical distance and abnormal trigger time window of any two channels in the dual abnormal channel set are extracted; If there is an overlapping period between the two abnormal trigger time windows, the corresponding channels are marked as time overlapping state, and if there is no overlapping period, the corresponding channels are marked as no overlapping state; The corresponding basic correlation degree of the physical distance is queried from a preset correlation degree mapping table, and when the two channels are in the time overlapping state, the final correlation degree is determined as the basic correlation degree, and when the two channels are in the no overlapping state, the final correlation degree is determined as zero; Based on the final correlation degree of all channel pairs, an abnormal correlation matrix is constructed; The channel pairs corresponding to the elements in the abnormal correlation matrix whose element values are greater than a preset strong correlation threshold are marked as correlation pairs, and the correlation pairs containing at least one same channel are merged to form a correlation group; Extract the channels of each correlation group to form a failure channel set.

7. The method of claim 1, wherein, According to the historical failure feature database, the spatial distribution characteristics of the failure channel set are matched with historical failure cases to obtain an identification report containing a failure positioning location, an abnormal type classification result, and a short-circuit risk warning level, including: Based on the physical positions of all channels in the failure channel set, a first spatial distribution feature is formed; The similarity of the spatial distribution feature and the first spatial distribution feature of the historical failure cases is calculated, and the failure location of the target failure case with the highest similarity is selected as the failure positioning location; According to the abnormal type label of each channel in the failure channel set, the channel quantity proportion of each abnormal type is counted to form an abnormal type classification result; From the historical failure feature database, historical risk diffusion data matching the abnormal level and corresponding abnormal type of the correlation group to which each channel in the failure channel set belongs are selected, and the risk probability of the affected channel is calculated according to the historical risk diffusion data to generate a short-circuit risk warning level; The failure positioning location, the abnormal type classification result, and the short-circuit risk warning level are integrated to generate an identification report.

8. A laser multi-channel sequential power-up failure detection system, characterized by, including: The acquisition module is configured to acquire driving voltage data, voltage abnormal duration, and acoustic wave signals of each channel during the laser multi-channel sequential power-on process; The calculation module is configured to calculate the voltage variation amplitude and voltage abnormal duration of each channel according to the driving voltage data, and mark the channels with voltage variation amplitude exceeding a preset voltage variation threshold or voltage abnormal duration exceeding a preset time threshold as voltage abnormal channels; The transformation module is configured to perform short-time Fourier transform on each acoustic wave signal to obtain a time-frequency spectrum, and extract acoustic feature waveforms from the time-frequency spectrum, and mark the channels with acoustic feature waveforms exceeding a preset short-circuit acoustic feature template or a preset arc acoustic feature template as acoustic abnormal channels; The construction module is configured to construct an abnormal correlation matrix according to the channels marked as voltage abnormal channels and acoustic abnormal channels, and select correlation groups with abnormal correlation strength exceeding a preset strong correlation threshold from the abnormal correlation matrix to determine a failure channel set; The matching module is configured to match the spatial distribution characteristics of the failure channel set with historical failure cases according to the historical failure feature database to obtain an identification report containing a failure positioning location, an abnormal type classification result, and a short-circuit risk warning level.

9. A computing device, comprising: The processing assembly and the storage assembly are included; the storage assembly stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing assembly to realize the laser multi-channel sequential power-on failure detection method of any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer program is stored in the computer, and when the computer program is executed by the computer, the laser multi-channel sequential power-on failure detection method of any one of claims 1-7 is realized.

Images