Summary of the invention
Purpose of the present invention is exactly at the deficiencies in the prior art, provide a kind of utilize relative amplitude-phase analysis, can high-quality discrimination circuit fault the analog circuit fault detection method, the accuracy of detection height, the fault coverage height, all effective to catastrophic type and parameter type fault.
For achieving the above object, technical scheme of the present invention is as follows:
Ultimate principle of the present invention is: the fault effect in the mimic channel can characterize with the relative energy of faulty circuit signal and normal circuit signal and the variation of time delay.Catastrophic type fault in the mimic channel is usually expressed as the marked change of relative energy between faulty circuit signal and the normal circuit signal, and parameter type fault is often insensitive to the relative energy variation of faulty circuit signal and normal circuit signal, but postpones to change responsive to the relative time between faulty circuit signal and normal circuit signal.The present invention is based on this fact, in conjunction with this theoretical tool efficiently of multiple mutual wavelet transformation, can bring high-quality fault detect again.
A mimic channel, sample sequence to its faulty circuit and normal circuit carries out again wavelet transformation mutually, its relative amplitude value representation faulty circuit output changes the relative energy of normal circuit output so, and the output of relative phase value representation faulty circuit is to the relative phase skew of normal circuit output.Relative amplitude analytical approach utilization wavelet transformation again-mutually can be in different time frequency domain yardstick extraction faulty circuit signals and the relative signal amplitude between the normal circuit signal, and the relative energy that the relative signal amplitude shows as between faulty circuit signal and the normal circuit signal changes.Utilize the relative amplitude analytical approach can detect effectively in the mimic channel relative energy is changed responsive fault.On the other hand, relative phase analytical approach utilization wavelet transformation again-mutually can be in different time frequency domain yardstick extraction faulty circuit signals and the relative phase information between the normal circuit signal, and relative phase information shows as the time delay between faulty circuit signal and the normal circuit signal.Utilize the relative phase analytical approach can detect effectively in the mimic channel time delay is changed responsive fault.The fault detection method that the present invention proposes requires unknown failure sequence and non-fault sequence to sample in the same triggering sampling time, could be used for extracting relative phase skew and energy variation between sample sequence then.Data acquisition unit all has triggering and timing function now, can finish this requirement well, and this obtains the support that practicality provides objective material conditions for the inventive method in engineering practice.
Basic theories of the present invention is as follows:
The sample sequence of a circuit is X (n), and its continuous wavelet transform may be defined as the convolution of X (n) and female small echo ψ (t), namely
In the following formula, * represents complex conjugate, and Δ t represents sampling interval, and s represents dimensions in frequency, and n is time shifting.Multiply-connected continuous wavelet transformation requires ψ (t) to be multiple female small echo.
The multiple mutual wavelet transformation of sample sequence X and Y is defined as:
Relative amplitude
Can think sequence X and the Y relative energy at a certain dimensions in frequency s and time shifting n;
Phase place
Can think that sequence X and Y are at the relative phase of a certain dimensions in frequency s and time shifting n.Here, the relative phase of burst means the time delay between signal.The relative energy that analog circuit fault shows as between fault sample sequence and the normal sample sequence of circuit changes and the relative phase skew.Multiple mutual wavelet transformation of the present invention adopts the female small echo of Morlet, or the female small echo of Paul.
In analog circuit fault diagnosing, the sine wave that it is F1 that the circuit-under-test input is generally a frequency, circuit output response is confined within the frequency response range F2 usually.Relation (the face formula 1. 2. as follows) according to wavelet scale in the wavelet analysis and equivalent Fourier frequency can obtain the wavelet analysis yardstick SF corresponding with frequency response F2.
The formula 1. s in 2. is the frequency domain yardstick of wavelet transformation, and f is equivalent Fourier frequency, ω
0Be the angular frequency of female small echo, general ω
0=6, m is the exponent number of Paul small echo, general m=3.
Be expressed as circuit normal sequence and failure sequence relative amplitude and the phase value on different frequency yardstick s and time shifting n.Because circuit output response concentrates on small echo dimensions in frequency SF usually, so can from
The middle sensitive information that extracts
In order to overcome the edge effect that wavelet transformation brings, only calculate the extraneous relative amplitude of cone of influence (COI) and phase value.Then sensitive information extraction formula can be expressed as:
Below provide the extracting method of sensitive information with an example forms.
Suppose a linear analogue circuit-under-test, its input stimulus is the sine wave of a 1KHz, and circuit output response is stored under the sampling rate of 100KHz, 2048 points of each waveform storage.We adopt ω
0The female small echo of=6Morlet carries out again wavelet transformation mutually to the output sample sequence, mother wavelet function ψ (t) as shown in the formula:
Because circuit is linear circuit, its circuit output response frequency scope also is 1KHz, so according to Morlet wavelet scale and equivalent Fourier frequency relation, and the dimensions in frequency SF of wavelet transformation in the time of can extrapolating equivalent Fourier frequency and be 1KHz:
Suppose V-Fault and V-Normal be circuit under certain failure condition and all elements for the output of the circuit under its nominal value response, get 2048 sampled points.Multiple mutual small echo dimensions in frequency S, following two formula are adopted in suggestion according to document, and the selection of S must cover SF.The value of S is [s
0, s
1..s
J].
So multiple mutual wavelet transformation
Be the complex element matrix of J*2048, wherein dimensions in frequency is S, and length is J, and time shifting is [1,2 ..2048], and length is 2048, then
It is the complex element matrix of a 1*2048.
Because sample sequence is finite length, directly sample sequence is carried out again wavelet transformation mutually, can cause in beginning and the ending phase of dimensions in frequency S bigger error taking place.A common optimization solution removes it behind wavelet transformation then for before carrying out wavelet transformation sample sequence being carried out the zero padding operation.Sample sequence is carried out zero padding operation here, make that sequence length reaches that original series length approaches the most 2
NIndividual, therefore reduced edge effect and accelerated the Fourier transform process.Cone of influence is defined as again mutually the e times die-away time of wavelet transformation auto-correlation energy spectrum under each dimensions in frequency S.Guarantee to make edge effect decline e e times of die-away time
-2Doubly, guaranteed that edge effect can ignore.
Be the 1*2048 sequence under dimensions in frequency SF, in order to overcome the influence of edge effect, according to the data of saying previously of only getting cone of influence inside.Supposing the beginning under the SF and finishing displacement die-away time is N1, N2, then the sample sequence of extraction time section N1-N2:
So
Be the sample sequence of a 2048-N1-(2048-N2) length, this sequence is the useful fault characteristic information abstraction sequence of relative amplitude-phase analysis method that the present invention is proposed.
Particularly, the fault detection method of a kind of mimic channel that the present invention proposes, concrete steps are as follows:
(1) each component parameter with tested mimic channel is made as nominal parameters, and this tested mimic channel is carried out emulation, obtains circuit output response sequence V-Normal under the nominal parameters.Emulation to this tested mimic channel is carried out in HSPICE, and circuit is input as sine wave.The V-Normal sequence is used as again the benchmark of wavelet transformation mutually.
(2) (the circuit elements device parameters meets Gaussian distribution in the range of tolerable variance of each components and parts nominal parameters of tested mimic channel, in 1 times of scope of its standard deviation, i.e. " 1Sigma "), this tested mimic channel is carried out Monte Carlo simulation, obtain the circuit output response sequence with Monte Carlo simulation number of times same number.The Monte Carlo simulation number of times is M time, and the circuit output response sequence that obtains is M, and wherein V-Monte (i) is expressed as the circuit output response sequence of the i time emulation, 1≤i≤M.
(3) circuit output response sequence under the nominal parameters that obtains in the step (1) is carried out again wavelet transformation mutually, namely
And therefrom extract sensitive information, obtain relative amplitude reference value sequence
With relative phase reference value sequence
With circuit that obtain in the step (2) and Monte Carlo simulation number of times same number output response sequence, respectively with step (1) under the nominal parameters that obtains circuit output response sequence carry out again wavelet transformation mutually, namely
And therefrom extract sensitive information, obtain the relative amplitude simulation value sequence with aforementioned Monte Carlo simulation number of times same number
With with the relative phase simulation value sequence of aforementioned Monte Carlo simulation number of times same number
(4) with relative amplitude simulation value sequence A MP that obtain in the step (3) and aforementioned Monte Carlo simulation number of times same number
Monte(i), respectively with step (3) in the relative amplitude reference value sequence A MP that obtains
ReferenceCarry out normalized, obtain the normal circuit output response relative amplitude value AMP with aforementioned Monte Carlo simulation number of times same number
REF(i).
With relative phase simulation value sequence A RG that obtain in the step (3) and aforementioned Monte Carlo simulation number of times same number
Monte(i), respectively with step (3) in the relative phase reference value sequence A RG that obtains
ReferenceCarry out normalized, obtain the normal circuit output response relative phase value ARG with aforementioned Monte Carlo simulation number of times same number
REF(i).
The normalized formula is as follows:
Wherein
Normal circuit output response relative amplitude value AMP that (5) in step (4), obtain and aforementioned Monte Carlo simulation number of times same number
REF(i) in, select numerical value the maximum as maximal phase to range value max (AMP
REF), select the numerical value reckling as minimum relative amplitude value min (AMP
REF).
Normal circuit output response relative phase value ARG that in step (4), obtain and aforementioned Monte Carlo simulation number of times same number
REF(i) in, select numerical value the maximum as maximal phase to phase value max (ARG
REF), select the numerical value reckling as minimum relative phase value min (ARG
REF).
Above-mentioned steps (1)-(5) are the circuit simulation stage, utilize Monte-Carlo emulation to obtain normal circuit (in the parameter tolerances scope) relative amplitude of response and the scope of phase place, determine minimax relative amplitude value and minimax relative phase value.
(6) the tested mimic channel of unknown failure is surveyed, obtain unknown circuit actual measurement output response sequence V-Measured.
(7) the unknown circuit actual measurement output response sequence V-Measured that obtains in circuit output response sequence V-Normal and the step (6) under the nominal parameters that obtains in the step (1) is carried out again wavelet transformation mutually, namely
And therefrom extract sensitive information, obtain the actual measurement relative amplitude value sequence of unknown failure circuit-under-test
Actual measurement relative phase value sequence with the unknown failure circuit-under-test
(8) with the relative amplitude reference value sequence A MP that obtains in the step (3)
ReferenceActual measurement relative amplitude value sequence AMP with the unknown failure circuit-under-test that obtains in the step (7)
MeasuredCarry out normalized, obtain unknown circuit output response relative amplitude value AMP
Unknown
With the relative phase reference value sequence A RG that obtains in the step (3)
ReferenceActual measurement relative phase value sequence ARG with the unknown failure circuit-under-test that obtains in the step (7)
MeasuredCarry out normalized, obtain unknown circuit output response relative phase value ARG
Unknown
The normalized formula is as follows:
(9) with the unknown circuit output response relative amplitude value AMP that obtains in the step (8)
Unknown, with the maximal phase that obtains in the step (5) to range value max (AMP
REF) and minimum relative amplitude value min (AMP
REF) compare; Simultaneously with the unknown circuit output response relative phase value ARG that obtains in the step (8)
Unknown, with the maximal phase that obtains in the step (5) to phase value max (ARG
REF) and minimum relative phase value min (ARG
REF) compare.
If unknown circuit output response relative amplitude value more than or equal to minimum relative amplitude value, simultaneously smaller or equal to maximal phase to range value, and unknown circuit output response relative phase value more than or equal to minimum relative phase value, (be min (AMP smaller or equal to maximal phase to phase value simultaneously
REF)≤AMP
Unknown≤ max (AMP
REF), while min (ARG
REF)≤ARG
Unknown≤ max (ARG
REF)), the tested mimic channel of this unknown failure non-fault then so.
If unknown circuit output response relative amplitude value (is AMP greater than maximal phase to range value perhaps less than minimum relative amplitude value
Unknown<min (AMP
REF) or AMP
Unknown>max (AMP
REF)), then there is fault (relative amplitude fault) in the tested mimic channel of this unknown failure so; If unknown circuit output response relative phase value (is ARG greater than maximal phase to phase value perhaps less than minimum relative phase value
Unknown<min (ARG
REF) or ARG
Unknown>max (ARG
REF)), then there is fault (relative phase fault) in the tested mimic channel of this unknown failure so.
Compared with prior art, the invention has the beneficial effects as follows: utilize relative amplitude-phase analysis to detect analog circuit fault, singly not to distinguish fault from the amplitude-frequency relation of circuit-under-test test response, also in conjunction with the phase place-frequency characteristic of circuit-under-test test response, the accuracy of detection height, the fault coverage height, all effective to catastrophic type and parameter type fault.
Embodiment 1
As Fig. 1, Fig. 2, shown in Figure 3.Choose international benchmark leapfrog Circuit verification analog circuit fault detection method of the present invention.The fault detection method step of described mimic channel is as follows:
(1) each component parameter with tested mimic channel is made as nominal parameters, i.e. R among Fig. 2
1~R
13=10K Ω, C
1=C
4=10nF, C
2=C
3=20nF, the C among Fig. 3
1=6pF.This tested mimic channel is carried out emulation in HSPICE, input VIN is that amplitude is the 1kHz sine wave of 3V, and output VOUT samples under 100K sps, stores 2048 points at every turn, obtains circuit output response sequence under the nominal parameters.
(2) in the range of tolerable variance of each components and parts nominal parameters of tested mimic channel, this tested mimic channel is carried out Monte Carlo simulation 2000 times, obtain 2000 circuit output response sequences.
(3) circuit output response sequence under the nominal parameters that obtains in the step (1) is carried out again wavelet transformation mutually, and therefrom extract sensitive information, obtain relative amplitude reference value sequence and relative phase reference value sequence.
With 2000 circuit output response sequences that obtain in the step (2), respectively with step (1) under the nominal parameters that obtains circuit output response sequence carry out again wavelet transformation mutually, and therefrom extract sensitive information, obtain 2000 relative amplitude simulation value sequences and 2000 relative phase simulation value sequences.
(4) with 2000 relative amplitude simulation value sequences that obtain in the step (3), respectively with step (3) in the relative amplitude reference value sequence that obtains carry out normalized, obtain 2000 normal circuit output response relative amplitude values.
With 2000 relative phase simulation value sequences that obtain in the step (3), respectively with step (3) in the relative phase reference value sequence that obtains carry out normalized, obtain 2000 normal circuit output response relative phase values.
(5) in 2000 normal circuit output response relative amplitude values that in step (4), obtain, select numerical value the maximum as maximal phase to range value, select the numerical value reckling as minimum relative amplitude value.
In 2000 normal circuit output response relative phase values that in step (4), obtain, select numerical value the maximum as maximal phase to phase value, select the numerical value reckling as minimum relative phase value.
Carry out above-mentioned steps (1)-(5) with the female small echo of Morlet and the female small echo of Paul respectively, obtain the relative amplitude of normal circuit response and the scope of relative phase then, as shown in table 1.As can be seen, when adopting Morlet mother small echo, the relative amplitude scope absolute value of normal circuit is about 25.8, and adopt the female small echo relative amplitude of Paul scope to be about 9.0, thereby can draw and adopt the female small echo of Morlet than Paul small echo better fault resolution to be arranged in the relative amplitude analysis.And for the relative phase analysis, its normal response scope absolute value all is about 18.2, and the effect that these two kinds of female small echos are analyzed relative phase is suitable.
|
Minimum value |
Maximal value |
Relative amplitude value (the female small echo of Morlet) |
-13.498259 |
12.343993 |
Relative phase value (the female small echo of Morlet) |
-8.895469 |
9.295972 |
Relative amplitude value (the female small echo of PAUL) |
-4.740685 |
4.336003 |
Relative phase value (the female small echo of PAUL) |
-8.885899 |
9.293368 |
Table 1 leapfrog circuit normal response relative amplitude and relative phase bounds
(6) 10 known busts and 14 known parameter type faults are injected the leapfrog circuit respectively, obtain 24 corresponding actual measurement output response sequences.
(7) with the actual measurement output response sequence corresponding with 10 busts that obtain in the step (6) respectively with step (1) under the nominal parameters that obtains circuit output response sequence carry out again mutual wavelet transformation with the female small echo of Morlet, and therefrom extract sensitive information, obtain 10 actual measurement relative amplitude value sequences and 10 actual measurement relative phase value sequences.
With the actual measurement output response sequence corresponding with 14 parameter type faults that obtain in the step (6) respectively with step (1) under the nominal parameters that obtains circuit output response sequence carry out again wavelet transformation mutually, multiple mutual wavelet transformation carries out with the female small echo of Morlet and the female small echo of Paul respectively, and therefrom extract sensitive information, obtain 28 actual measurement relative amplitude value sequences and 28 actual measurement relative phase value sequences.
(8) with obtain in the step (7) 10 actual measurement relative amplitude value sequences respectively with step (3) in the relative amplitude reference value sequence that obtains carry out normalized, obtain the output response relative amplitude value of 10 known busts, as shown in table 2.With obtain in the step (7) 10 actual measurement relative phase value sequences respectively with step (3) in the relative phase reference value sequence that obtains carry out normalized, obtain the output response relative phase value of 10 known busts, as shown in table 2.
Table 2 leapfrog circuit typical case bust
With obtain in the step (7) 28 actual measurement relative amplitude value sequences respectively with step (3) in the relative amplitude reference value sequence that obtains carry out normalized, obtain the output response relative amplitude value of 28 known parameters type faults, as shown in table 3.With obtain in the step (7) 28 actual measurement relative phase value sequences respectively with step (3) in the relative phase reference value sequence that obtains carry out normalized, obtain the output response relative phase value of 28 known parameters type faults, as shown in table 3.
Table 3 leapfrog circuit canonical parameter type fault
(9) with the output response relative amplitude value that obtains in the step (8), with the maximal phase that obtains in the step (5) range value and minimum relative amplitude value are compared; With the output response relative phase value that obtains in the step (8), with the maximal phase that obtains in the step (5) phase value and minimum relative phase value are compared simultaneously.
In the table 2, C4-Open breaking phenomena occurs for element C4, and C4-Short short circuit phenomenon occurs for element C4.AMP (Morlet) is illustrated in the relative amplitude value of corresponding fault under the female small echo of Morlet, ARG (Morlet) is illustrated in the relative phase value of corresponding fault under the female small echo of Morlet, and OBS (Morlet) expression is adopted relative amplitude or this fault detect can be come out under the analysis relatively.
For the C4-Open fault, adopt the relative phase analysis this fault detect can well be come out, it shows as the faulty circuit sample sequence and the normal circuit sample sequence has a tangible time delay.For the R10-Open fault, it analyzes responsive to relative amplitude, and insensitive to the relative phase analysis, it shows as the faulty circuit sample sequence has significant change with respect to normal circuit sample sequence energy.For faults such as C4-Short, R10-Short and R12-Short, can detect with relative amplitude and relative phase analysis, its fault not only shows the variation of energy between the sample sequence, has a tangible relative time simultaneously and postpones.Bust often shows that circuit output response energy variation is relatively large, for example R12-Short, R11-Open fault, and it is almost nil that its sample sequence shows as the sampled output signal amplitude in time domain.And for C4-Open, C1-Open fault, its sample sequence amplitude on time domain is almost the same with normal circuit output, but it significantly shows tangible time delay of appearance between faulty circuit and normal circuit output.
In the table 3, C1-6Sigma reduces 6sigma for element C1, and C2+6Sigma increases 6sigma for element C2.AMP (Morlet) is illustrated in the relative amplitude value of corresponding fault under the female small echo of Morlet, ARG (Morlet) is illustrated in the relative phase value of corresponding fault under the female small echo of Morlet, AMP (Paul) is illustrated in the relative amplitude value of corresponding fault under the female small echo of Paul, and ARG (Paul) is illustrated in the relative phase value of corresponding fault under the female small echo of Paul.
When C1 reduces 6sigma, relative amplitude value overrun upper boundary values 12.343993, so the C1-6Sigma fault can be come out with the relative amplitude analyzing and testing.When C3 increases 6sigma, relative phase value overrun lower border value-8.895469, so the C3+6Sigma fault can be come out with the relative phase analyzing and testing.C2+6Sigma and C3-6Sigma fault can be come out by the female small echo of Morlet and the female Wavelet Detection of Paul simultaneously, and the selection of female small echo mainly influences and shows fault resolution aspect.For C1-6Sigma, R1-6Sigma and R2-6Sigma fault, it is more more responsive than relative phase analysis to the relative amplitude analysis.Simultaneously, for C3+6Sigma, C4+6sigma and R6-6sigma fault, it analyzes insensitive to relative amplitude, and can detect this type of fault with the relative phase analysis.For C2+6Sigma and R7-6sigma fault, it can be detected by relative amplitude and relative phase analysis, and wherein the C2+6Sigma fault is more responsive to the relative phase analysis, and the R7-6sigma fault is more responsive to the relative amplitude analysis.