Electronic component monitoring system based on big data

Abstract

The invention relates to the technical field of equipment state monitoring, in particular to an electronic component monitoring system based on big data, which comprises a multi-frequency response analysis module, a frequency domain feature reconstruction module, a stress deviation trend module, a coupling correlation structure module and a grade mapping generation module. According to the invention, through continuous data acquisition and analysis based on multi-frequency response information, electrical parameter changes of components are tracked in real time, change trends under various frequencies are continuously monitored and accurately identified, and through analysis and extraction of frequency domain data features and combination of stress deviation trend analysis, tiny changes caused by external environment changes are identified. According to the electrical parameter change of different time slices, the long-term health state of the component is comprehensively reflected, a dynamic evolution model of the component is constructed, a multi-dimensional data association analysis method is adopted, and the component degradation process is accurately evaluated by judging the relevance among different electrical parameters. And scientific and accurate decision support is provided for component maintenance and replacement.

Description

Technical Field

[0001] This invention relates to the field of equipment condition monitoring technology, and in particular to an electronic component monitoring system based on big data. Background Technology

[0002] Equipment condition monitoring technology is a technical system that revolves around the acquisition of status, analysis of operating parameters, health assessment, and anomaly identification of industrial equipment, electronic components, and related operating entities. Specifically, it includes the continuous acquisition of operating parameters of the monitored object, the analysis and structured organization of the acquired data, the identification of the operating status characteristics of the equipment based on predetermined criteria, and the identification and judgment of aging, performance changes, or potential anomalies during equipment operation. It typically involves multi-source data acquisition, quantitative description of operating status, status modeling based on historical data, and tracking and analysis of the status evolution of the monitored object, forming a technical framework for real-time monitoring and systematic management of equipment status.

[0003] Traditional electronic component monitoring systems refer to systems that acquire, organize, and determine the physical and electrical parameters of components such as resistors, capacitors, and transistors during use. Typically, these systems involve manual testing or basic measuring instruments to measure parameters such as resistance, capacitance, withstand voltage, leakage current, or pin continuity. Specifically, this includes measuring the static electrical parameters of components with a multimeter, scanning key indicators of components one by one with a dedicated tester, observing changes in component parameters by applying working conditions with a constant voltage or current source, or determining the status of components by recording impedance changes at specific frequency bands. The monitoring method relies on component-by-component measurement and single-point parameter recording, and determines whether the component is in a normal state by comparing the difference between the measured parameters and the standard parameters.

[0004] Traditional electronic component monitoring systems rely on single-point parameter detection, including using multimeters to measure electrical parameters such as resistance, capacitance, and leakage current. They assess component status based on single measurements. Due to low measurement frequency, discrete and discontinuous data, they cannot capture dynamic changes in components in real time. When component operating conditions change, they struggle to identify subtle anomalies or trends, cannot effectively track changes in components under different operating environments, and lack quantitative analysis of the long-term evolution of component status. This results in the inability to detect potential performance degradation or failure risks in a timely manner. Furthermore, the lack of multi-dimensional performance evaluation affects the reliability and accuracy of the system. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a big data-based electronic component monitoring system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an electronic component monitoring system based on big data includes: The multi-frequency response analysis module uses the multi-frequency response information of components to analyze the voltage and current of monitoring nodes at multiple frequencies, compare the changing trends of voltage at adjacent frequencies, calculate the current amplitude at adjacent frequencies, identify abnormal fluctuations and filter continuously changing regions, statistically analyze the frequency domain structure characteristics formed by the frequency link, and obtain multi-frequency difference gradient sequence data. The frequency domain feature reconstruction module compares the same frequency direction and organizes the consistent link according to the multi-frequency difference gradient sequence data, calculates the amplitude fluctuation to extract the change segment, and integrates it into the change path in time to obtain the frequency domain dynamic trajectory structure data. The stress offset trend module uses the frequency domain dynamic trajectory structure data to compare the differences between adjacent events and the offset direction based on the change in electrical parameters after the applied electrical stress, calculates the offset amplitude shape, and determines the stage trend according to the continuity of direction and amplitude to obtain stress-driven trend data. The coupled correlation structure module uses the stress-driven trend data to compare the directions of adjacent time points and combine consistent terms to calculate the change ratio based on the changes in conduction, leakage, temperature rise, and power consumption. It then constructs a multi-dimensional point set and judges the coupling relationship based on the clustering characteristics to obtain the linkage degradation structure information. The grading mapping generation module utilizes the linked degradation structure information, compares the clustered segments with the grading condition interval records based on the electrical parameter combination characteristics of each time slice, calculates the distribution of the grading on the time axis, judges the continuity of segments of the same grading to filter clustering segments, classifies them according to grading order, and constitutes degradation grading mapping data.

[0007] As a further aspect of the present invention, the multi-frequency difference gradient sequence data includes frequency point difference amplitude sequence, frequency direction change sequence, and frequency domain continuity marker information; the frequency domain dynamic trajectory structure data includes time series frequency node link information, frequency band change morphology information, and trajectory structure index information; the stress-driven trend data includes external stress event sequence identifiers, key electrical parameter offset direction information, and stage offset trend description data; the linkage degradation structure information includes a multi-parameter linkage node set, parameter coupling strength description information, and degradation linkage topology structure data; and the degradation level mapping data includes a time slice level identifier set, level segment division information, and degradation level distribution description data.

[0008] As a further aspect of the present invention, the multi-frequency response analysis module includes: The frequency response acquisition submodule uses the multi-frequency response information of components to analyze the multi-frequency voltage and current responses generated by the monitoring nodes, compares the voltage changes of adjacent frequency points and records the direction, extracts the current change amplitude of adjacent frequency points and organizes it into a data sequence sorted by frequency, organizes the frequency point data within the same time slice, and generates frequency point difference characterization quantity. The frequency link filtering submodule determines the correspondence between the continuous frequency point direction record and the current change amplitude based on the frequency point difference characterization quantity, performs quantification processing on the frequency link group, calculates the frequency link group linkage strength quantity, filters continuous frequency link groups, and generates frequency link linkage measurement information. The frequency domain gradient generation submodule calls the frequency chain linkage metric information, analyzes the distribution of each frequency chain group in the same time slice, judges abnormal jumps and filters continuously changing regions, performs statistics on all frequency links in the same time slice, forms frequency domain structural features, and generates multi-frequency difference gradient sequence data.

[0009] As a further aspect of the present invention, the frequency domain feature reconstruction module includes: The node orientation chain construction submodule acquires the multi-frequency differential gradient sequence data, analyzes the orientation changes of consecutive time slices at each frequency, compares the orientation records of the same frequency in adjacent time slices and identifies nodes with the same orientation, organizes nodes with the same orientation to form node links, and generates node orientation chain data. The amplitude fluctuation extraction submodule extracts the amplitude change corresponding to each link based on the node direction link data, and extracts significant segments by calculating the fluctuation of the amplitude change in the link to obtain amplitude fluctuation segment data. The continuous path integration submodule determines the temporal continuity of adjacent segments based on the amplitude fluctuation segment data and integrates them into a continuous changing path. It then organizes the changing path into a frequency domain changing structure and generates frequency domain dynamic trajectory structure data.

[0010] As a further aspect of the present invention, the process of extracting significantly changed segments specifically includes: Based on the node direction link data, calculate the amplitude change of each node in the link within a predetermined frequency range, and detect the volatility based on the standard deviation of the node amplitude change. If the standard deviation exceeds twice the preset benchmark threshold, the target segment is determined to be a segment with significant changes. Based on historical monitoring data, the amplitude change of the target electronic component under normal working conditions is calculated, and the standard deviation of the target amplitude is set as the preset benchmark threshold.

[0011] As a further aspect of the present invention, the stress offset trend module includes: The electrical parameter difference analysis submodule acquires the frequency domain dynamic trajectory structure data, analyzes the changes in key electrical parameters before and after each applied electrical stress event, calculates the differences in electrical parameters between adjacent events, compares the magnitude of continuous electrical parameter changes, and obtains electrical parameter change difference data. The offset direction recording submodule compares the offset direction in continuous events and records the direction sequence based on the difference data of electrical parameter changes, calculates the change pattern of offset amplitude in the event sequence, judges the stage offset trend based on the continuity of direction and the coherence of amplitude, and obtains offset direction sequence data. The trend expression construction submodule identifies the connection relationship of the offset trend on the time axis and forms continuous change segments based on the offset direction sequence data. It then organizes the change segments to construct the trend expression under stress and generates stress-driven trend data.

[0012] As a further aspect of the present invention, the coupling and association structure module includes: The electrical parameter direction identification submodule uses the stress-driven trend data to analyze the change direction of each electrical parameter at adjacent moments based on the changes in conduction, leakage, temperature rise, and power consumption within the monitoring period. It compares the difference signs at adjacent moments and records the signs that match, filters the electrical parameter combinations with the same direction in the monitoring sequence, and generates electrical parameter direction-matching combination data. The proportional strength determination submodule, based on the consistent combination data of the electrical parameters, calls the change amplitude of each electrical parameter recorded in continuous time slices, calculates the difference relationship in multiple time slices, judges the consistency and direction continuity of the amplitude change, calculates the cooperative strength coefficient of the electrical parameter combination, and obtains the coupling proportional strength evaluation data. The aggregation structure construction submodule constructs a multi-dimensional point set based on the coupling ratio strength evaluation data and the correlation strength between multiple pairs of combinations. Based on the aggregation characteristics of the point set in the multi-dimensional coordinates, it determines the coupling relationship between electrical parameters, organizes the coupling relationship, and generates linkage degradation structure information.

[0013] As a further aspect of the present invention, the level mapping generation module includes: The combined feature analysis submodule acquires the linked degradation structure information, analyzes the electrical parameter combination characteristics of each time slice within the monitoring period, compares the change amplitude of electrical parameters in adjacent time slices, and extracts the fluctuation pattern of the clustered segments in each time slice to generate electrical parameter combination fluctuation feature data. The grade classification submodule compares the clustered segments with the electrical parameter combination intervals in the grade conditions based on the electrical parameter combination fluctuation characteristic data, determines and records the grade classification of each segment, and generates segment grade data. The clustering segment organization submodule calculates the distribution of each level of segment on the time axis based on the segment level data, determines the continuity of segments of the same level on the time axis and filters them to form cluster segments, organizes the cluster segments in order of level, and generates degradation level mapping data.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, continuous data acquisition and analysis based on multi-frequency response information are used to track changes in the electrical parameters of components in real time, continuously monitor and accurately identify the changing trends at each frequency, analyze and extract frequency domain data features, and combine stress offset trend analysis to identify minute changes caused by changes in the external environment. Based on the changes in electrical parameters at different time slices, the long-term health status of components is comprehensively reflected, a dynamic evolution model of components is constructed, and a multi-dimensional data correlation analysis method is used to accurately assess the degradation process of components by judging the correlation between different electrical parameters, providing scientific and accurate decision support for component maintenance and replacement. Attached Figure Description

[0015] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the multi-frequency response analysis module of the present invention; Figure 3 This is a flowchart of the frequency domain feature reconstruction module of the present invention; Figure 4 This is a flowchart of the stress offset trend module of the present invention; Figure 5 This is a flowchart of the coupled and related structural modules of the present invention; Figure 6 This is a flowchart of the level mapping generation module of the present invention. Detailed Implementation

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

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

[0018] Example 1 Please see Figure 1 A big data-based electronic component monitoring system includes: The multi-frequency response analysis module uses the multi-frequency response information of components to analyze the voltage and current of monitoring nodes at multiple frequencies, compare the changing trends of voltage at adjacent frequencies, calculate the current amplitude at adjacent frequencies, identify abnormal fluctuations and filter continuously changing regions, statistically analyze the frequency domain structure characteristics formed by the frequency link, and obtain multi-frequency difference gradient sequence data. The frequency domain feature reconstruction module compares the same frequency direction and organizes the consistent links based on the multi-frequency difference gradient sequence data, calculates the amplitude fluctuations to extract the change segments, and integrates them into the change path in time to obtain the frequency domain dynamic trajectory structure data. The stress offset trend module uses frequency domain dynamic trajectory structure data, compares the differences between adjacent events and the offset direction based on the change of electrical parameters after the application of external electrical stress, calculates the offset amplitude shape, and determines the stage trend according to the continuity of direction and amplitude to obtain stress-driven trend data. The coupled and associated structure module utilizes stress-driven trend data, compares the directions of adjacent time points and combines consistent terms to calculate the change ratio based on the changes in conduction, leakage, temperature rise and power consumption, constructs a multi-dimensional point set, and judges the coupling relationship based on the clustering characteristics to obtain the linkage degradation structure information. The grading mapping generation module utilizes the linked degradation structure information, compares the clustered segments with the grading condition interval records based on the electrical parameter combination characteristics of each time slice, calculates the distribution of the grading on the time axis, judges the continuity of segments of the same grading to filter clustering segments, classifies them according to grading order, and constitutes degradation grading mapping data.

[0019] Multi-frequency differential gradient sequence data includes frequency point differential amplitude sequences, frequency direction change sequences, and frequency domain continuity marker information; frequency domain dynamic trajectory structure data includes time series frequency node link information, frequency band change morphology information, and trajectory structure index information; stress-driven trend data includes external stress event sequence identifiers, key electrical parameter offset direction information, and stage offset trend description data; linkage degradation structure information includes multi-parameter linkage node sets, parameter coupling strength description information, and degradation linkage topology data; degradation level mapping data includes time slice level identifier sets, level segment division information, and degradation level distribution description data.

[0020] Please see Figure 2 The multi-frequency response analysis module includes: The frequency response acquisition submodule uses the multi-frequency response information of components to analyze the multi-frequency voltage and current responses generated by the monitoring nodes, compares the voltage changes of adjacent frequency points and records the direction, extracts the current change amplitude of adjacent frequency points and organizes it into a data sequence sorted by frequency, organizes the frequency point data within the same time slice, and generates frequency point difference characterization quantity. Utilizing the multi-frequency response information of components, specifically within a time slice Inside, for one Device application , , and Excitation at four frequency points was used to acquire the multi-frequency voltage and current responses generated by the monitoring nodes. The collected data is shown in Table 1. The multi-frequency voltage and current responses generated by the monitoring nodes were analyzed, and operations were performed to compare the voltage changes of adjacent frequency points, for example... and The voltage change is The direction of change is recorded as "negative". and The voltage change is The recording direction is "negative". and The voltage change is Recording the direction as "negative" yields the voltage direction sequence. Next, the amplitude of current change at adjacent frequency points is extracted. and The amplitude of the current change is , and The amplitude of the current change is , and The amplitude of the current change is The extracted amplitude values ​​are then organized into a data sequence sorted by frequency. Finally, for the same time slice Frequency data within and The data is processed to generate frequency point difference characterization quantities.

[0021] Table 1 Time-slice multi-frequency response data table

[0022] As shown in Table 1, this data is Time-slice acquisition Respond to the raw data.

[0023] The frequency link filtering submodule determines the correspondence between continuous frequency point direction records and current change amplitudes based on frequency point difference characteristics, and quantifies the frequency link group using the following formula: ; Calculate the frequency chain linkage strength, filter consecutive frequency chain groups, and generate frequency chain linkage measurement information. in, For the first The first frequency chain The normalized voltage difference components at each frequency point are obtained by acquiring the voltage response differences between adjacent frequency points and performing normalization processing on the maximum absolute amplitude among all difference terms in the chain. For the first The first frequency chain The normalized current difference components at each frequency point are obtained by acquiring the current response differences between adjacent frequency points and performing normalization processing on the maximum absolute amplitude among all difference terms in the chain. The frequency index is the number of a frequency point within the frequency chain, obtained by collecting multi-frequency data from monitoring nodes. The number of frequency points in a single frequency chain is obtained by recording the total number of frequency points in the chain that participate in the differential operation. The frequency chain numbering corresponding to the voltage differential sequence is obtained by grouping the voltage response by frequency. The frequency chain number corresponding to the current difference sequence is obtained by grouping the current response by frequency. This refers to the linkage strength of the frequency chain group; Based on the frequency point difference characterization quantity, i.e. as well as The correspondence between the direction records of continuous frequency points and the amplitude of current changes is determined, and the frequency chain is quantized using the formula. The calculation of the frequency chain linkage strength is as follows, including the parameters in the formula and the calculation process: This refers to the frequency chain linkage strength. For the first The first frequency chain Normalized voltage difference components at each frequency point For the first The first frequency chain Normalized current difference components at each frequency point The frequency point number within the frequency chain. , This refers to the number of frequency points in a single frequency chain. , This refers to the frequency chain number corresponding to the voltage differential sequence. , This represents the frequency chain number corresponding to the current difference sequence. First, obtain the original voltage differential sequence. The maximum absolute magnitude in this chain is Normalization was performed to obtain Next, the original current difference sequence is obtained. , , ,get The maximum absolute magnitude in this chain is Normalization was performed to obtain The calculation logic of the formula lies in: the numerator term The amplitude "covariance" of the two chains at each frequency point is characterized by calculating the sum of the normalized VI differential amplitude products. The first term in the denominator... With the second item Quantify the overall change in "energy" of the voltage chain and current chain separately, the third term in the denominator Quantify the "differences" between the two chains in their normalized form, overall The value quantifies the coupling level of the VI chain group by the ratio of "covariance" to "total energy and difference". The advantage of the formula is that it constructs a normalized linkage strength by combining three parts: the correlation of differential amplitude at the same frequency point (numerator), the overall difference within the chain (the first and second terms in the denominator), and the differential offset difference between chains (the third term in the denominator). It can stably quantize the synchronization of voltage and current responses in the frequency domain, unaffected by the absolute amplitude of the original signal, thereby accurately identifying frequency chains with true coupling relationships and substituting the values ​​for calculation: Numerator ; The first term in the denominator ; The second term in the denominator ; The third term in the denominator ; ; After the calculation is completed, Calculated value Linkage threshold with preset frequency chain group Comparison, The settings are based on: data collection Taiwan Health exist Under different working conditions Value, total indivual Value samples, statistical distribution, and found of The values ​​are all lower than ,Right now To ensure that only significant linkages are selected, Set as ,because Therefore, the frequency chain group is determined to be a continuous frequency chain group. Continuous frequency chain groups are then filtered to generate frequency chain linkage metric information. Among them, the frequency chain group linkage strength is a non-negative real number quantified based on the response synchronization degree, amplitude correspondence, and offset difference correlation of the voltage differential sequence and the current differential sequence within the same time slice. It is used to characterize the coupling level of differential behavior between the two frequency chains within the same time slice. This value is composed of three parts: differential amplitude correlation at the same frequency point, the overall differential component within the chain, and the differential offset difference between the chains. It can describe the synchronous change trend characteristics of the frequency chain group under a multi-frequency response environment and can be used to sort, filter, merge, or separate different frequency chain groups in subsequent processing to form a structurally complete and directional frequency domain change path.

[0024] The frequency domain gradient generation submodule calls the frequency chain linkage measurement information, analyzes the distribution of each frequency chain group in the same time slice, judges abnormal jumps and filters continuously changing areas, performs statistics on all frequency links in the same time slice, forms frequency domain structural features, and generates multi-frequency differential gradient sequence data. Invoke frequency chain linkage metric information, i.e., time slice. of And the state is "continuous", and there is another frequency chain ( to )of Furthermore, the state is "discontinuous," analyzing each frequency chain group in the same time slice. The distribution within, i.e. To identify abnormal fluctuations, retrieve the previous time slice. Data ,calculate Change in value and Set abnormal fluctuation threshold This threshold is analyzed steady-state operation Hours (total) Within a time slice The statistical distribution of the value changes is obtained, and the calculation is performed. Standard deviation of the sequence ,Will Set as ,Compare Change and threshold and ,determination If there are no abnormal fluctuations at any time, the next step is to filter for areas of continuous change. This operation is based on... , exist At that time (greater than) ),exist At that time (greater than) ),therefore The chain group it belongs to is determined to be a continuously changing region. exist and All less than Not selected, for the same time slice Statistics were compiled on all frequency links within the region, and the results show: continuously changing regions. One, non-contiguous region One, forming The frequency domain structure characteristics at time t, which are expressed as This generates multi-frequency differential gradient sequence data.

[0025] Please see Figure 3 The frequency domain feature reconstruction module includes: The node orientation chain construction submodule acquires multi-frequency differential gradient sequence data, analyzes the orientation changes of consecutive time slices at each frequency, compares the orientation records of the same frequency in adjacent time slices and identifies nodes with the same orientation, organizes nodes with the same orientation to form node links, and generates node orientation chain data. Obtain multi-frequency difference gradient sequence data, specifically extracting the original frequency point difference characteristics contained therein, in order to... - Node (frequency point) and Taking the difference between them as an example, we analyze the node in continuous time slices. Changes in direction, retrieve Voltage direction record at time middle - The node's direction is "negative" (see paragraph 1). Real-time collection Calculate the difference The direction is "negative". Real-time collection Calculate the difference The direction is "positive". Real-time collection Calculate the difference When the direction is "negative", we get - The direction sequence of the nodes is Compare the same frequencies in adjacent time slices ( - Record and identify nodes with the same orientation. and Both are negative, and they are in the same direction. and The change from "negative" to "positive" has an inconsistent direction. and The change from "positive" to "negative" indicates a misalignment in direction. Nodes with consistent directions are then organized to form a node link, thus obtaining the link. At the same time, for - The nodes perform the same operation to obtain their direction sequence. Comparison revealed to Always consistent, organized into a chain Generate node direction link data.

[0026] The amplitude fluctuation extraction submodule extracts the amplitude change corresponding to each link based on the node direction link data. By calculating the fluctuation of the amplitude change in the link, it extracts the segments with significant changes and obtains amplitude fluctuation segment data. Extract based on node direction link data Extract the amplitude change corresponding to each link, i.e. - Node at to The amplitude of current change during the time slice, among which, The time range is , Real-time collection , amplitude . Real-time collection , amplitude . Time (Peak) Collection , amplitude This leads to the link. Amplitude sequence: To detect the volatility of link amplitude changes, the standard deviation (population standard deviation) is used for calculation, i.e., the denominator is the sample size. First, calculate the mean of the amplitude sequence: ; Calculate the population standard deviation accordingly: Based on historical monitoring data, the amplitude fluctuation of the target electronic component (MOSFET) under normal operating conditions is obtained, and the standard deviation of its amplitude change is used as a preset benchmark threshold. Specifically, data is collected... A health status time slice Changes in node amplitude form a healthy sequence: Calculate its mean: ; Calculate the population standard deviation: Set a preset baseline threshold: If the standard deviation of the target link Exceeding twice the benchmark threshold (i.e. If the value is 0, then the link is determined to be a significantly changed segment. The current calculation result is: Therefore, the link (correspond The time period is identified as a segment with significant changes, and then data on the amplitude fluctuation segment is obtained.

[0027] The continuous path integration submodule determines the temporal continuity of adjacent segments based on amplitude fluctuation segment data and integrates them into a continuous changing path. It then organizes the changing path into a frequency domain change structure and generates frequency domain dynamic trajectory structure data. Based on the amplitude fluctuation segment data, i.e. And assuming that two other fragments were obtained through the same processing flow. and Determine the temporal continuity of adjacent segments. The end time is , The start time is , Adjacent ,determination and It has temporal continuity. The end time is , The start time is , The two events overlap in time and are not continuous. It is "Not Significant" and does not participate in the integration. and For a continuously changing path, this path represents Frequency domain response anomaly from - node( - ) "Transferred" to - node( - The change path is structurally organized and a frequency domain change structure is formed. Record as Generate frequency domain dynamic trajectory structure data.

[0028] Please see Figure 4 The stress offset trend module includes: The electrical parameter difference analysis submodule acquires frequency domain dynamic trajectory structure data, analyzes the changes in key electrical parameters before and after each applied electrical stress event, calculates the differences in electrical parameters between adjacent events, compares the magnitude of continuous electrical parameter changes, and obtains electrical parameter change difference data. Acquire frequency domain dynamic trajectory structure data, i.e. The marked - Frequency domain anomalies within a time period, analysis The applied electrical stress events experienced during this period, such as in Stress events at all times ( Overvoltage pulse), in Apply at any time ,exist Apply at any time The changes in key electrical parameters before and after each applied electrical stress event were analyzed, and the on-resistance was selected as the key electrical parameter. ,Measurement After it happened Stable value ,Measurement After it happened Stable value ,Measurement After it happened Stable value Calculate the difference in electrical parameters between adjacent events. , Compare the magnitude of the continuous electrical parameter changes, i.e. and We obtained data on the differences in electrical parameter changes.

[0029] The offset direction recording submodule compares the offset direction in continuous events and records the direction sequence based on the difference data of electrical parameter changes. It calculates the change pattern of offset amplitude in the event sequence, judges the stage offset trend based on the continuity of direction and the coherence of amplitude, and obtains the offset direction sequence data. Based on the differences in electrical parameter changes, i.e., the sequence Comparing consecutive events ( and The offset direction in ) The value is positive, and the offset direction is "positive". The value is positive, the offset direction is "positive", and the direction sequence is recorded. Calculate the change pattern of the offset magnitude in the event sequence, i.e. Become The form is "increase", based on directional continuity ( Sequence continuity is "positive" and amplitude coherence ( arrive (For continuous increase), judge the phased deviation trend, determine. to period When the trend is "positive acceleration offset", the offset direction sequence data is obtained.

[0030] The trend expression construction submodule identifies the connection relationship of the offset trend on the time axis based on the offset direction sequence data and forms continuous change segments. It organizes the change segments, constructs the trend expression under stress, and generates stress-driven trend direction data. Based on the offset direction sequence data, i.e. - It exhibits a "positive acceleration shift" trend, and supplements... event( )and event( (data) back , back ,calculate , , - The differential sequences are Trend It was determined to be "positive acceleration deviation". - The difference is Trend The trend is identified as "positive deceleration offset," and the connection between the offset trend and the time axis is determined. Ending Trend Beginning Both in Connect and form a continuous variation segment. The change segments are organized to construct a trend expression under stress, i.e. Under stress to Degradation is accelerated under the influence of the action, Degradation slows down, generating stress-driven trend data.

[0031] Please see Figure 5 The coupled and related structural modules include: The electrical parameter direction identification submodule utilizes stress-driven trend data to analyze the direction of change of each electrical parameter at adjacent moments based on the changes in conduction, leakage, temperature rise, and power consumption within the monitoring period. It compares the difference signs of adjacent moments and records the signs that match, filters the electrical parameter combinations with the same direction in the monitoring sequence, and generates electrical parameter direction-consistent combination data. Using stress-driven trend data Based on the monitoring period (e.g.) to Between The change in conduction at time ( ) ), and the amount of deviation () ), temperature rise change ( ) and power consumption change ( ), analyze adjacent time points ( and The direction of change of each electrical parameter, The time data is as follows: (direction "positive"), (direction "positive"), (Negative direction) (Direction "positive"), compare the signs of the differences between adjacent time points and record the signs that match. and Consistent signs (positive, positive). and Inconsistent signs (positive, negative) and Consistent signs (positive, positive). and Matching signs (positive, positive) are used to filter electrical parameter combinations with consistent directions in the monitoring sequence. The combination of moments is , , Generate combined data with consistent electrical parameter directions.

[0032] The proportional intensity determination submodule combines data based on the consistent direction of electrical parameters, retrieves the change amplitude recorded for each electrical parameter in continuous time slices, calculates the difference relationship across multiple time slices, and determines the consistency and continuity of amplitude changes using the following formula: ; Calculate the synergistic strength coefficient of the electrical parameter combination to obtain coupling ratio strength evaluation data; in, For the first Electrical parameters at time The normalized variation amplitude is obtained by acquiring the electrical parameter at time [time]. The original change is then divided by the reference change of the same type of electrical parameter within the monitoring interval to obtain the result. For the first Electrical parameters at time The normalized variation amplitude is obtained by acquiring the electrical parameter at time [time]. The original change is then divided by the reference change of the same type of electrical parameter within the monitoring interval to obtain the result. To avoid extremely small dimensionless quantities with a denominator of zero, For the first Electrical parameters at time The trend continuation marker value is obtained by identifying the directional consistency of the normalized change amplitude of the electrical parameter in adjacent time steps. For the first Electrical parameters at time The trend continuation marker value is obtained by identifying the directional consistency of the normalized change amplitude of the electrical parameter in adjacent time steps. For the first With the Synergy strength coefficient of electrical parameter combination This is the first index in the electrical parameter combination, obtained by numbering the types of electrical parameters in the monitoring sequence. This is the second index in the electrical parameter combination, obtained by numbering the types of electrical parameters in the monitoring sequence. The time slice number is obtained by sequentially numbering the sampling points of the monitoring period. The number of time slices used in the average calculation is obtained by statistically normalizing the total number of samples in the change amplitude sequence. Based on the consistent combination of electrical parameters, i.e. Moment Combinations of these parameters, calling each electrical parameter in a continuous time slice ( ,Right now The recorded changes in amplitude are analyzed, and the difference relationship across multiple time slices is calculated to determine the consistency and directional continuity of the amplitude changes. A formula is then used. Calculate the synergistic strength coefficient of the electrical parameter combination, in order to (R), (I) as an example , Set as , For the first Electrical parameters at time The normalized variation range, For the first Electrical parameters at time The normalized variation range, and The value is used to mark the continuation of the trend. The synergy strength coefficient, Time slice number First obtain The original change: , , , , , ,set up Reference change (the monitoring interval) (mean of change) is , Reference change (the monitoring interval) (mean of change) is Calculate the normalized amplitude: , , , , , Identify trend continuity. time and All are positive, therefore hour and At this time, the weight term The calculation logic of the formula is as follows: The term is used to quantify the two normalized magnitudes. and The similarity, when and When they are very close, the denominator approaches 0, the term value is extremely large, indicating that the amplitude changes are highly consistent. The item is used as a trend weight when and Both are in the process of continuing the trend ( This item is To ensure the calculation is valid, By taking the root mean square of the time series, a representation is obtained. and The stability coefficient value of the degree of coordination in magnitude and trend is beneficial because the formula introduces a normalized change magnitude. This eliminates the influence of different electrical parameter dimensions and utilizes The structure greatly amplifies the weight of combinations with similar magnitude changes (i.e., denominators approaching 0), and this is combined with trend continuation markers. , making The coefficients are highly sensitive to combinations of electrical parameters with strong synchronization and consistent trends (i.e., highly coupled degradation mechanisms), which can be applied to numerical calculations: hour, ; hour, ; hour, ; ; The result Much greater than the synergy strength threshold , The setting is based on: by considering known degradation mechanisms that are irrelevant (such as...) (related to packaging hermeticity) and strongly correlated with (e.g.) and The electrical parameters of ) are used for Calculations and statistics revealed unrelated pairs Strong correlation Therefore, it is set As a criterion for judging strong synergy, ,show and The system exhibits a strong synergistic relationship, yielding coupling ratio strength assessment data. The synergistic strength coefficient of the electrical parameter combination describes whether two electrical parameters exhibit synchronous changes, mutual influence, or co-driving relationships during component degradation. The practical effect of this coefficient is to help the monitoring system identify "which electrical parameters exhibit mutual attraction or co-evolution characteristics during degradation," providing a foundation for subsequently constructing a multi-dimensional linkage degradation structure, enabling the system to characterize the coupling mechanism of component degradation at the parameter level.

[0033] The clustering structure construction submodule constructs a multi-dimensional point set based on the coupling ratio strength evaluation data and the correlation strength between multiple pairs of combinations. Based on the clustering characteristics of the point set in the multi-dimensional coordinates, it judges the coupling relationship between electrical parameters, organizes the coupling relationship, and generates linkage degradation structure information. Based on the coupling ratio strength assessment data, i.e. (Strong), and assume (powerful), (powerful), (weak), (weak), (weak), through the strength of the association between multiple pairs of combinations ( (value), construct a with The four electrical parameters are nodes, The graph structure, with values ​​representing edge weights, determines the coupling relationship between electrical parameters based on the clustering characteristics of point sets in multidimensional coordinates. Specifically, it sets... Filter all Strongly correlated edges, (R,I), (R,P), (I,P), are in the graph. The three nodes are tightly connected by strong edges, forming a A fully connected subgraph of nodes (clustering feature), and Nodes and The edge weights between them are all less than In a free state, the coupling relationships are restructured and determined. Forming a linked degradation cluster, As independent parameters, generate linked degradation structure information.

[0034] Please see Figure 6 The level mapping generation module includes: The combined feature analysis submodule acquires information on the linked degradation structure, analyzes the combined electrical parameters of each time slice within the monitoring period, compares the change amplitude of electrical parameters in adjacent time slices, extracts the fluctuation pattern of the clustered segments in each time slice, and generates combined electrical parameter fluctuation feature data. Obtain information on the linked degradation structure, i.e. Linked clusters, analysis within the monitoring period ( The electrical parameter combination characteristics of each time slice are analyzed, the variation amplitude of electrical parameters in adjacent time slices is compared, and the electrical parameters in each time slice are extracted. () the clustering fragments () The wave pattern of the cluster is calculated by normalizing the amplitude of all electrical parameters within the cluster. The mean (cluster centroid) is used to characterize it. When (t=1), (Assuming) centroid , At (t=2), (Assuming) centroid , At (t=3), (Assuming), center of mass The obtained centroid sequence This refers to the wave pattern of the cluster, generating electrical parameter combination wave characteristic data.

[0035] The grade classification submodule compares the clustered segments with the electrical parameter combination intervals in the grade conditions based on the electrical parameter combination fluctuation characteristic data, determines and records the grade classification of each segment, and generates segment grade data. Based on the combined wave characteristics data of electrical parameters, i.e., the cluster centroid sequence Compare the clustered segments (centroid values ​​of each time slice) with the electrical parameter combination intervals in the grading conditions, which are obtained through accelerated degradation experiments ( Calibration, Experimental data show that: center of mass correspond In good health (Level 1). Corresponds to early degradation (level 2). Corresponding to mid-game degeneration (level 3). For severe degradation (level 4), determine and record the level classification of each segment. , It belongs to level 2. , It belongs to level 3. , It is classified as Level 2, and fragment level data is generated.

[0036] The clustering segment organization submodule calculates the distribution of each level of segment on the time axis based on the segment level data, determines the continuity of segments of the same level on the time axis and filters them to form cluster segments, organizes the cluster segments in order of level, and generates degradation level mapping data. Based on fragment level data, i.e. The corresponding level sequence is and expanded to to common A time slice is used to obtain the hierarchical sequence. Calculate the distribution of each level's segment on the timeline, and determine the continuity of segments of the same level on the timeline. Continuous , Continuous , for (Length is 1), Continuous , Continuous Set a minimum continuous length threshold Select and form cluster segments. (L1, length 3) is selected. (L2, length 3) is selected. (L3, length 1) is discarded. (L2, length 2) is selected. (L1, length 2) is selected. The cluster segments are then sorted in hierarchical order. L1 segment: L2 section: L3 level segment due to fragment Data whose length does not meet the minimum continuous length threshold is discarded and marked as an empty set, forming the degradation level mapping data.

[0037] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A big data-based electronic component monitoring system, characterized in that, The system includes: The multi-frequency response analysis module uses the multi-frequency response information of components to analyze the voltage and current of monitoring nodes at multiple frequencies, compare the changing trends of voltage at adjacent frequencies, calculate the current amplitude at adjacent frequencies, identify abnormal fluctuations and filter continuously changing regions, statistically analyze the frequency domain structure characteristics formed by the frequency link, and obtain multi-frequency difference gradient sequence data. The frequency domain feature reconstruction module compares the same frequency direction and organizes the consistent link according to the multi-frequency difference gradient sequence data, calculates the amplitude fluctuation to extract the change segment, and integrates it into the change path in time to obtain the frequency domain dynamic trajectory structure data. The stress offset trend module uses the frequency domain dynamic trajectory structure data to compare the differences between adjacent events and the offset direction based on the change in electrical parameters after the applied electrical stress, calculates the offset amplitude shape, and determines the stage trend according to the continuity of direction and amplitude to obtain stress-driven trend data. The coupled correlation structure module utilizes the stress-driven trend data to compare the directions of adjacent time points and combine consistent terms to calculate the change ratio based on the changes in conduction, leakage, temperature rise, and power consumption. It then constructs a multi-dimensional point set, determines the coupling relationship based on the clustering characteristics, and obtains the linkage degradation structure information.

2. The electronic component monitoring system based on big data according to claim 1, characterized in that, The multi-frequency difference gradient sequence data includes frequency point difference amplitude sequence, frequency direction change sequence, and frequency domain continuity marker information. The frequency domain dynamic trajectory structure data includes time series frequency node link information, frequency band change morphology information, and trajectory structure index information. The stress-driven trend data includes external stress event sequence identifier, key electrical parameter offset direction information, and stage offset trend description data. The linkage degradation structure information includes multi-parameter linkage node set, parameter coupling strength description information, and degradation linkage topology structure data.

3. The electronic component monitoring system based on big data according to claim 1, characterized in that, The multi-frequency response analysis module includes: The frequency response acquisition submodule uses the multi-frequency response information of components to analyze the multi-frequency voltage and current responses generated by the monitoring nodes, compares the voltage changes of adjacent frequency points and records the direction, extracts the current change amplitude of adjacent frequency points and organizes it into a data sequence sorted by frequency, organizes the frequency point data within the same time slice, and generates frequency point difference characterization quantity. The frequency link filtering submodule determines the correspondence between the direction records of consecutive frequency points and the amplitude of current changes based on the frequency point difference characterization quantity, and performs quantization processing on the frequency link group using the formula: ； Calculate the frequency chain linkage strength, filter consecutive frequency chain groups, and generate frequency chain linkage measurement information. in, For the first The first frequency chain Normalized voltage difference components at each frequency point For the first The first frequency chain Normalized current difference components at each frequency point The frequency point number within the frequency chain. This refers to the number of frequency points in a single frequency chain. This refers to the frequency chain number corresponding to the voltage differential sequence. This represents the frequency chain number corresponding to the current difference sequence. This refers to the linkage strength of the frequency chain group; The frequency domain gradient generation submodule calls the frequency chain linkage metric information, analyzes the distribution of each frequency chain group in the same time slice, judges abnormal jumps and filters continuously changing regions, performs statistics on all frequency links in the same time slice, forms frequency domain structural features, and generates multi-frequency difference gradient sequence data.

4. The electronic component monitoring system based on big data according to claim 3, characterized in that, The frequency domain feature reconstruction module includes: The node orientation chain construction submodule acquires the multi-frequency differential gradient sequence data, analyzes the orientation changes of consecutive time slices at each frequency, compares the orientation records of the same frequency in adjacent time slices and identifies nodes with the same orientation, organizes nodes with the same orientation to form node links, and generates node orientation chain data. The amplitude fluctuation extraction submodule extracts the amplitude change corresponding to each link based on the node direction link data, and extracts significant segments by calculating the fluctuation of the amplitude change in the link to obtain amplitude fluctuation segment data. The continuous path integration submodule determines the temporal continuity of adjacent segments based on the amplitude fluctuation segment data and integrates them into a continuous changing path. It then organizes the changing path into a frequency domain changing structure and generates frequency domain dynamic trajectory structure data.

5. The electronic component monitoring system based on big data according to claim 4, characterized in that, The process of extracting significantly changed segments is as follows: Based on the node direction link data, calculate the amplitude change of each node in the link within a predetermined frequency range, and detect the volatility based on the standard deviation of the node amplitude change. If the standard deviation exceeds twice the preset benchmark threshold, the target segment is determined to be a segment with significant changes. Based on historical monitoring data, the amplitude change of the target electronic component under normal working conditions is calculated, and the standard deviation of the target amplitude is set as the preset benchmark threshold.

6. The electronic component monitoring system based on big data according to claim 4, characterized in that, The stress offset trend module includes: The electrical parameter difference analysis submodule acquires the frequency domain dynamic trajectory structure data, analyzes the changes in key electrical parameters before and after each applied electrical stress event, calculates the differences in electrical parameters between adjacent events, compares the magnitude of continuous electrical parameter changes, and obtains electrical parameter change difference data. The offset direction recording submodule compares the offset direction in continuous events and records the direction sequence based on the difference data of electrical parameter changes, calculates the change pattern of offset amplitude in the event sequence, judges the stage offset trend based on the continuity of direction and the coherence of amplitude, and obtains offset direction sequence data. The trend expression construction submodule identifies the connection relationship of the offset trend on the time axis and forms continuous change segments based on the offset direction sequence data. It then organizes the change segments to construct the trend expression under stress and generates stress-driven trend data.

7. The electronic component monitoring system based on big data according to claim 6, characterized in that, The coupled association structure module includes: The electrical parameter direction identification submodule uses the stress-driven trend data to analyze the change direction of each electrical parameter at adjacent moments based on the changes in conduction, leakage, temperature rise, and power consumption within the monitoring period. It compares the difference signs at adjacent moments and records the signs that match, filters the electrical parameter combinations with the same direction in the monitoring sequence, and generates electrical parameter direction-matching combination data. The proportional strength determination submodule, based on the consistent combination data of the electrical parameters, calls the change amplitude of each electrical parameter recorded in continuous time slices, calculates the difference relationship in multiple time slices, judges the consistency and direction continuity of the amplitude change, calculates the cooperative strength coefficient of the electrical parameter combination, and obtains the coupling proportional strength evaluation data. The aggregation structure construction submodule constructs a multi-dimensional point set based on the coupling ratio strength evaluation data and the correlation strength between multiple pairs of combinations. Based on the aggregation characteristics of the point set in the multi-dimensional coordinates, it determines the coupling relationship between electrical parameters, organizes the coupling relationship, and generates linkage degradation structure information.

8. The electronic component monitoring system based on big data according to claim 1, characterized in that, The system also includes: The grade mapping generation module utilizes the linked degradation structure information, compares the clustered segments with the grade condition interval records according to the electrical parameter combination characteristics of each time slice, calculates the distribution of the classification on the time axis, judges the continuity of segments of the same grade to filter clustering segments, classifies them according to grade order, and constitutes degradation grade mapping data. The degradation level mapping data includes a set of time slice level identifiers, level segment division information, and degradation level distribution description data.

9. The electronic component monitoring system based on big data according to claim 8, characterized in that, The level mapping generation module includes: The combined feature analysis submodule acquires the linked degradation structure information, analyzes the electrical parameter combination characteristics of each time slice within the monitoring period, compares the change amplitude of electrical parameters in adjacent time slices, and extracts the fluctuation pattern of the clustered segments in each time slice to generate electrical parameter combination fluctuation feature data. The grade classification submodule compares the clustered segments with the electrical parameter combination intervals in the grade conditions based on the electrical parameter combination fluctuation characteristic data, determines and records the grade classification of each segment, and generates segment grade data. The clustering segment organization submodule calculates the distribution of each level of segment on the time axis based on the segment level data, determines the continuity of segments of the same level on the time axis and filters them to form cluster segments, organizes the cluster segments in order of level, and generates degradation level mapping data.

Images