Search

CN-121995198-A - Waveform characteristic visualization method and system based on high-low temperature test

CN121995198ACN 121995198 ACN121995198 ACN 121995198ACN-121995198-A

Abstract

The invention provides a waveform characteristic visualization method and a system based on high-low temperature test, which are characterized in that a Hall effect sensor is used for collecting current time sequence data under the working condition of the high-low temperature test of a chip, a sliding window is used for filtering and resampling pretreatment, a local polynomial fitting is used for generating a continuous smooth current data fitting waveform, the fitting waveform is divided according to a time window, a wave crest and wave trough seed point is extracted, a pulse width and period characteristic data set is obtained, an inherent time scale decomposition algorithm is called for stripping waveform PR components, the PR component density degree data set is obtained through discretization, a time domain and frequency domain characteristic calculation disorder coefficient, a trend coefficient and a warp index are fused, a multidimensional characteristic data set is constructed, a time domain distribution curve and a thermodynamic diagram visualization chart are drawn through characteristic mapping and normalization treatment, and the waveform characteristic visualization data set is generated, so that the extraction and coupling of current waveform characteristics are realized, and visual and reliable data are provided for the analysis of the working condition of the high-low temperature test of the chip.

Inventors

  • WANG RUOYU

Assignees

  • 上海电洋材料科技有限公司

Dates

Publication Date
20260508
Application Date
20260305

Claims (9)

  1. 1. The waveform characteristic visualization method based on the high-low temperature test is characterized by comprising the following steps of: s1, acquiring a chip current time sequence data sequence based on a Hall effect sensor, preprocessing based on the chip current time sequence data sequence, and fitting to acquire a current data fitting waveform; S2, dividing the fitting waveform of the current data according to time windows, and recording the wave crest and wave trough seed points of the waveform in each window to obtain a sub-window pulse width basic data set and a sub-window waveform period characteristic data set; s3, carrying out waveform decomposition and discretization based on the sub-window pulse width basic data set to obtain a sub-window PR component density degree data set; S4, analyzing disorder based on the sub-window pulse width basic data set and the sub-window waveform periodic characteristic data set, enhancing based on the sub-window PR component density degree data set, and acquiring a sub-window waveform pulse width sequence warp index data set based on an analysis result and an enhancement result; And S5, carrying out feature fusion on the basis of the window-divided waveform pulse width sequence warp index data set, the window-divided PR component density data set and the window-divided waveform pulse width sequence warp index data set to obtain a waveform feature visualization data set.
  2. 2. The method for visualizing waveform characteristics based on high and low temperature testing as in claim 1, wherein said acquiring a chip current time series data sequence based on a hall effect sensor comprises: placing the chip into high-low temperature test equipment, deploying a Hall effect sensor, setting a preset sampling interval, continuously collecting working current signals of the chip under high-low temperature test working conditions, and obtaining an initial chip current time sequence data sequence; based on the initial chip current time sequence data sequence, correcting peak abnormal values in the current signals by adopting a sliding window filtering method, and acquiring the current time sequence data sequence by resampling the time resolution of the unified signals.
  3. 3. The method for visualizing waveform features based on high and low temperature test as in claim 2, wherein said fitting based on chip current time series data sequence to obtain a current data fitting waveform comprises: based on the current time sequence data sequence, carrying out local neighborhood window division and kernel function weight configuration to generate a local fitting data set with weight constraint; Calculating the weight coefficient and the basis function mapping value of each sampling point in the local fitting data set to generate a polynomial coefficient matrix, and carrying out point-by-point fitting on discrete sampling points according to the polynomial coefficient matrix to obtain a local continuous fitting segment; And inputting the global continuous waveform curve into a noise filter layer of the model, eliminating the sawtooth effect caused by discrete sampling, and obtaining the fitting waveform of the current data.
  4. 4. The method for visualizing waveform features based on high and low temperature test as in claim 1, wherein the step of dividing the waveform based on current data fitting by time window and recording the peak and trough seed points of the waveform in each window to obtain the sub-window pulse width basic data set and the sub-window waveform period feature data set comprises the steps of: setting window length and sliding step length based on current data fitting waveforms, uniformly dividing the current data fitting waveforms into a plurality of continuous and non-overlapping time windows, establishing a corresponding relation between each window and a waveform segment, and obtaining a window-dividing current waveform data set; Identifying wave crest and wave trough seed points of the wave form in each window by adopting an extremum detection algorithm based on the current waveform data set of each window, generating a wave crest sequence and a wave trough sequence according to time sequence, calculating the time span between adjacent wave crests, determining pulse width parameters of each period, and obtaining a basic data set of the pulse width of each window; and calculating the rising edge slope and the falling edge slope of the waveform in each window based on the current waveform data set of the sub-window, and counting the waveform amplitude change rate in the period to obtain the waveform period characteristic data set of the sub-window.
  5. 5. The method for visualizing waveform features based on high and low temperature testing as in claim 1, wherein said performing waveform decomposition and discretization based on windowed pulse width basic dataset, obtaining windowed PR component density dataset comprises: calling an inherent time scale decomposition algorithm based on the current data fitting waveform, decomposing the current fitting waveform in a sub-window into PR components with preset layers, and realizing layered stripping of waveform frequency domain characteristics to obtain a multi-scale PR component data set; And discretizing continuous PR components by adopting an equidistant value method based on the multi-scale PR component data set to ensure that the discretized data quantity is consistent with the current time sequence data quantity of the corresponding window, calculating the absolute deviation of each PR component data and the component mean value, and obtaining the window PR component density degree data set.
  6. 6. The method for visualizing waveform features based on high and low temperature testing as in claim 1, wherein said analyzing the clutter based on the windowed pulse width basic dataset and the windowed waveform period feature dataset, enhancing based on the windowed PR component intensity dataset, and obtaining the windowed pulse width sequence warp index dataset based on the analysis and result and the enhancement result comprises: Based on a sub-window pulse width basic data set and a sub-window waveform periodic characteristic data set, counting the mean value and variance of all periodic pulse widths in each window, introducing preset adjustment parameters to avoid denominator to be zero, calculating the relative deviation absolute value of each periodic pulse width and the mean value, and calculating the mean value of all deviation absolute values to obtain a sub-window waveform pulse width disorder coefficient data set; calculating the difference of the density degree of PR components of different levels in each window based on the window PR component density degree data set, constructing a divergence matrix among the components, and obtaining a window PR component divergence characteristic data set; and carrying out sequence warp index fusion based on the sub-window waveform pulse width disorder coefficient data set and the sub-window PR component divergence characteristic data set to obtain a sub-window waveform pulse width sequence warp index data set.
  7. 7. The method for visualizing waveform features based on high and low temperature testing as in claim 6, wherein said performing sequence warp index fusion based on a windowed waveform pulse width disorder coefficient dataset and a windowed PR component divergence feature dataset, obtaining a windowed waveform pulse width sequence warp index dataset comprises: Based on a windowed waveform pulse width disorder coefficient data set and a windowed PR component divergence characteristic data set, constructing a time sequence of disorder coefficients of each window according to time sequence, uniformly dividing the sequence into a plurality of sub-windows, calculating the ratio of the range of the disorder coefficients in each sub-window to the corresponding time interval, solving the average value of the ratio of all the sub-windows, and obtaining a windowed waveform pulse width sequence trend coefficient data set; And carrying out window-by-window multiplication operation on the trend coefficient and the PR component density degree based on the window-by-window waveform pulse width sequence trend coefficient data set and the window-by-window PR component density degree data set, so as to realize the coupling of time domain and frequency domain characteristics and obtain a window-by-window waveform pulse width sequence warp index data set.
  8. 8. The method for visualizing waveform features based on high and low temperature test as in claim 1, wherein said feature fusion is performed based on a windowed waveform pulse width sequence warp index dataset, a windowed PR component density dataset, and a windowed waveform pulse width sequence warp index dataset, and obtaining the waveform feature visualization dataset comprises: Based on a window-dividing waveform pulse width sequence warp index data set, a window-dividing PR component density data set and a window-dividing waveform pulse width sequence warp index data set, establishing a mapping relation between a time window and various characteristic parameters, constructing a characteristic parameter topological association matrix, and acquiring a waveform characteristic original data set; Mapping the numerical range of various characteristic parameters to a [0-1] interval by adopting a minimum and maximum normalization method on the waveform characteristic original data set, eliminating the dimensional difference of different characteristic dimensions, acquiring the waveform characteristic data set, and drawing a visual chart based on the waveform characteristic data set to serve as the waveform characteristic visual data set.
  9. 9. A waveform characteristic visualization system based on high-low temperature test is characterized by comprising, The current data acquisition module comprises a Hall effect sensor for acquiring a chip current time sequence data sequence; The waveform time domain feature extraction module is connected with the current data acquisition module and is used for dividing the fitting waveform of the current data according to time windows, recording wave crest and wave trough seed points of the waveform in each window and acquiring a sub-window pulse width basic data set and a sub-window waveform period feature data set; the waveform frequency domain feature extraction module is connected with the current data acquisition module and is used for carrying out waveform decomposition and discretization based on the windowing pulse width basic data set to obtain a windowing PR component density degree data set; The characteristic quantization fusion module is respectively connected with the waveform time domain characteristic extraction module and the waveform frequency domain characteristic extraction module and is used for analyzing messiness based on the sub-window pulse width basic data set and the sub-window waveform period characteristic data set, enhancing based on the sub-window PR component density degree data set and acquiring a sub-window waveform pulse width sequence warp index data set based on an analysis result and an enhancement result.

Description

Waveform characteristic visualization method and system based on high-low temperature test Technical Field The invention belongs to the technical field of chip measurement, and particularly relates to a waveform characteristic visualization method and system based on high-low temperature testing. Background Along with the rapid development of the semiconductor industry, the application scene of the chip is diversified, and the severe requirements are put on the stability and the reliability of the chip under the high-low temperature extreme environment from consumer electronics to the fields of industrial control, aerospace and the like. The high-low temperature test is used as a verification link before the chip leaves the factory, and can simulate the working state of the chip under different temperature working conditions and judge whether the performance of the chip meets the standard. The current mainstream chip high-low temperature test method is mostly dependent on the combination mode of a three-temperature separator and a heat flow cover, and the method has certain advantages in the chip test of a large quantity and a small variety, but has limitations when facing the test requirements of a small quantity and a plurality of varieties. The traditional method has a rough processing mode on the current time sequence data, and the sawtooth effect brought by discrete sampling can mask the real characteristics of the waveform, influence the accuracy of subsequent analysis and hardly reflect the chip state. Disclosure of Invention In order to solve the technical problems, the invention provides a waveform characteristic visualization method and a waveform characteristic visualization system based on high-low temperature testing, which are used for solving the technical problems that in the prior art, the traditional measuring method is rough in processing mode of current time sequence data, the sawtooth effect caused by discrete sampling can mask the real characteristics of waveforms, the accuracy of subsequent analysis is affected, and the chip state is difficult to reflect. The invention discloses a waveform characteristic visualization method and system based on high-low temperature test, which are aimed at and have the efficacy, and are achieved by the following specific technical means: a waveform characteristic visualization method based on high-low temperature test comprises the following steps: s1, acquiring a chip current time sequence data sequence based on a Hall effect sensor, preprocessing based on the chip current time sequence data sequence, and fitting to acquire a current data fitting waveform; S2, dividing the fitting waveform of the current data according to time windows, and recording the wave crest and wave trough seed points of the waveform in each window to obtain a sub-window pulse width basic data set and a sub-window waveform period characteristic data set; s3, carrying out waveform decomposition and discretization based on the sub-window pulse width basic data set to obtain a sub-window PR component density degree data set; S4, analyzing disorder based on the sub-window pulse width basic data set and the sub-window waveform periodic characteristic data set, enhancing based on the sub-window PR component density degree data set, and acquiring a sub-window waveform pulse width sequence warp index data set based on an analysis result and an enhancement result; And S5, carrying out feature fusion on the basis of the window-divided waveform pulse width sequence warp index data set, the window-divided PR component density data set and the window-divided waveform pulse width sequence warp index data set to obtain a waveform feature visualization data set. According to a preferred embodiment, the acquiring the chip current time series data sequence based on the hall effect sensor includes: placing the chip into high-low temperature test equipment, deploying a Hall effect sensor, setting a preset sampling interval, continuously collecting working current signals of the chip under high-low temperature test working conditions, and obtaining an initial chip current time sequence data sequence; based on the initial chip current time sequence data sequence, correcting peak abnormal values in the current signals by adopting a sliding window filtering method, and acquiring the current time sequence data sequence by resampling the time resolution of the unified signals. According to a preferred embodiment, the fitting based on the chip current time series data sequence to obtain a current data fitting waveform includes: based on the current time sequence data sequence, carrying out local neighborhood window division and kernel function weight configuration to generate a local fitting data set with weight constraint; Calculating the weight coefficient and the basis function mapping value of each sampling point in the local fitting data set to generate a polynomial coefficient matrix, and carrying