CN102590762B - Information entropy principle-based method for fault diagnosis of switch power supply - Google Patents

Abstract

The invention discloses an information entropy principle-based method for fault diagnosis of a switch power supply; therefore, defects that a current diagnostic method needs more test points, a programming algorithm is complex, there are a few diagnosable fault types, and accurate positioning can not be realized can be overcome. When a fault diagnosis is carried out, a magnetic leakage signal of a switch power supply board magnetic element is obtained; a spectral entropy characteristic Hf, a time domain entropy characteristic Ht, a peak-to-peak value characteristic Vpp, a mean value characteristic a, a root mean square characteristic r, and a variance characteristic sigma of the magnetic leakage signal are extracted; and all the extracted characteristic values are compared with characteristic values in a characteristic value table that is established before the diagnosis, so that it is determined whether there is a fault on the switch power supply and what a type of the fault is. According to the invention, there is a few test points that are needed according to the method; and the fault diagnosis of the power supply can be realized only needing the magnetic leakage signal of the power supply board magnetic element. And the method is especially suitable for an occasion on which a contact type fault diagnosis can not be carried out as well as can be applied to system tests and fault diagnoses of various switch power supplies.

Description

Switching Power Supply method for diagnosing faults based on information entropy principle

Technical field

The invention belongs to technical field of measurement and test, relate to a kind of method for diagnosing faults, by Switching Power Supply key node signal is carried out to information entropy extraction, can realize the fault diagnosis of Switching Power Supply.

Background technology

Switching Power Supply product is widely used in the fields such as industrial automatic control, military industry equipment, research equipment, LED illumination, industrial control equipment, and Switching Power Supply is generally the part occurred frequently of fault, Switching Power Supply is carried out to fault diagnosis quickly and easily significant.

The fault diagnosis of Switching Power Supply has manual method and automatic diagnosis method at present.

Manual method is to utilize the directly voltage signal of measuring switch power supply interdependent node such as multimeter or oscillograph, and by analysis node voltage waveform, judgement Switching Power Supply is in what malfunction.This method complex operation, efficiency is low; And in Switching Power Supply, contain high-pressure section, if misoperation meeting produces injury to equipment under test and tester.

Automatic diagnosis method is mainly to utilize test needle-bar, the equipment such as program control flying needle, by test probe, contact and obtain node voltage signal with respective nodes in circuit, again these information exchanges are crossed to various conversion, use neural network, expert system etc. to carry out intelligent diagnostics, and then judge the malfunction of Switching Power Supply.The method of existing diagnosis, diagnosable fault type relatively limits to, and diagnostic result can not accurately be located abort situation, and required test node is many, causes arithmetic programming complicated.These have all restricted the application of said method.

Summary of the invention

The object of the invention is to the deficiency for above-mentioned prior art, proposed a kind of Switching Power Supply method for diagnosing faults based on information entropy principle, improved the discrimination of fault, diagnosable failure mode and the reliability and the efficiency that have improved fault diagnosis.

For achieving the above object, the Switching Power Supply method for diagnosing faults based on information entropy principle of the present invention, comprises the steps:

(1) set up the list of feature values of various each feature distributed areas of experiment Switching Power Supply plate magnetic leakage signal:

(1a) choose respectively a trouble-free normal experiment Switching Power Supply plate and be labeled as B ₀there is the experiment Switching Power Supply plate of different faults type with m piece, respectively mark B ₁, B ₂..., B _m, m≤100, wherein fault type comprises: the short circuit of output commutation diode, feedback power supply diode open-circuit, current sample filtering circuit resistance disconnect, shake tank capacitance open circuit, feedback divider resistance open circuit and output terminal short circuit;

(1b) adopt cordless to obtain the magnetic leakage signal of various experiment Switching Power Supply plate magnetic elements, be respectively s ₀, s ₁..., s _m, s wherein ₀for trouble-free normal experiment Switching Power Supply plate B ₀magnetic leakage signal, s ₁be the 1st out of order experiment Switching Power Supply plate B ₁magnetic leakage signal, the like, s _mbe m out of order experiment Switching Power Supply plate B _mmagnetic leakage signal;

(1c) use information entropy principle to extract experiment Switching Power Supply plate magnetic element magnetic leakage signal s _itime domain entropy feature in time domain and the spectrum entropy feature in frequency domain, and extract magnetic leakage signal s by the method for statistical study _ipeak-to-peak value feature, characteristics of mean, root mean square feature and Variance feature, obtain testing Switching Power Supply plate B ₀to B _meach feature of magnetic leakage signal, wherein test Switching Power Supply plate B _ithe feature of magnetic leakage signal comprise: spectrum entropy H _fi, time domain entropy H _ti, peak-to-peak value V _ppi, average a _i, root mean square r _iand variances sigma _i, i=0,1,2 ..., m;

(1d) change the measuring position of non-contact magnetically probe, repeating step (1b)-(1c) g time, 10≤g≤20, the experiment Switching Power Supply plate B that step (1c) is extracted _ieach feature of magnetic leakage signal, obtain respectively average and the maximum deviation of g data, then obtain the distributed area of each feature, and the distributed area of each each feature of experimental power supply plate is write in table, i=0,1,2,, m, sets up the list of feature values of respectively testing Switching Power Supply plate magnetic leakage signal;

(2) to tested Switching Power Supply plate whether fault and fault type diagnose:

(2a) adopt cordless to obtain the magnetic leakage signal of tested Switching Power Supply plate magnetic element;

(2b) use information entropy principle to extract the time domain entropy feature H in tested Switching Power Supply plate magnetic element magnetic leakage signal time domain _twith the spectrum entropy feature H in frequency domain _f, and by the method for statistical study, extract the peak-to-peak value feature V of this Switching Power Supply plate magnetic leakage signal _pp, characteristics of mean a, root mean square feature r and Variance feature σ;

(2c) read the feature distributed area of the first experiment Switching Power Supply plate magnetic leakage signal in the list of feature values, judge that each feature of tested power panel is whether all in the distributed area of the individual features of this experiment Switching Power Supply plate magnetic leakage signal, if all in the distributed area of individual features, the fault of tested Switching Power Supply plate is fault corresponding to this experiment Switching Power Supply plate, otherwise read lower a kind of feature distributed area of testing Switching Power Supply plate magnetic leakage signal in the list of feature values, judge successively, until determine the fault type of tested Switching Power Supply plate, if searched the list of feature values, still cannot find the experiment Switching Power Supply plate with tested Switching Power Supply with same fault, by manual method, carrying out fault determines, and its fault signature is filled in the list of feature values.

The present invention compares with existing method for diagnosing faults, and tool has the following advantages:

1) the present invention is owing to having utilized information entropy principle, time domain entropy feature in magnetic leakage signal time domain and the spectrum entropy feature in frequency domain have been extracted, and peak-to-peak value feature, characteristics of mean, root mean square feature and the Variance feature of magnetic leakage signal in conjunction with the method for statistical study, have been extracted, make the signal characteristic difference between dissimilar fault obvious, make this diagnostic method there is arithmetic programming simple, fault recognition rate is high, the many advantages of failure mode that can identify.

2) the present invention is owing to before feature extraction, magnetic leakage signal having been carried out to normalized, measurand and the uncertain impact bringing of test macro distance have been eliminated, simplify the design of test macro, improved the reliability and the range of application that has expanded this method of testing of this method of testing.

Below in conjunction with accompanying drawing, the present invention is further illustrated:

Accompanying drawing explanation

Fig. 1 is the list of feature values Establishing process figure in method for diagnosing faults of the present invention;

Fig. 2 is the Troubleshooting Flowchart in method for diagnosing faults of the present invention.

Embodiment

Switching Power Supply method for diagnosing faults based on information entropy principle of the present invention, implementation step is as follows:

Step 1, sets up the list of feature values of various each feature distributed areas of experimental power supply plate magnetic leakage signal.

With reference to Fig. 1, being implemented as follows of this step:

(1) choose the experiment switching circuit board of various different faults types

Choose respectively a trouble-free normal experiment Switching Power Supply plate and be labeled as B ₀there is the experiment Switching Power Supply plate of different faults type with m piece, respectively mark B ₁, B ₂..., B _m, m≤100, wherein fault type comprises: the short circuit of output commutation diode, feedback power supply diode open-circuit, current sample filtering circuit resistance disconnect, shake tank capacitance open circuit, feedback divider resistance open circuit and output terminal short circuit;

(2) obtain the magnetic leakage signal of various experiment Switching Power Supply plate magnetic elements

(2a) at experiment Switching Power Supply plate B ₀fixed resistance of upper connection, as the load of Switching Power Supply, and can guarantee that power panel works under non-failure conditions, and gives this power panel power supply;

(2b) non-contact magnetically probe is placed in to experiment Switching Power Supply plate B ₀magnetic element near any position of 1～5cm, measure the magnetic leakage signal of magnetic element, and the simulating signal of measuring amplified to conditioning and sampling, obtain and test Switching Power Supply plate B ₀the Serial No. s of magnetic leakage signal ₀;

(2c), by the same method of step (2a) to (2b), obtain all the other experiment Switching Power Supply plate B ₁to B _mmagnetic leakage signal, obtain testing Switching Power Supply plate B _imagnetic leakage signal s _i, i=0,1,2 ..., m;

(3) extract and respectively test Switching Power Supply plate B ₀to B _mmagnetic leakage signal feature

(3a) use information entropy principle to extract experiment Switching Power Supply plate B _imagnetic element magnetic leakage signal s _itime domain entropy feature H in time domain _tiwith the spectrum entropy feature H in frequency domain _fi, i=0,1,2 ..., m:

(3a1) to the magnetic leakage signal Serial No. s after sampling ₀carry out filtering, search maximum value and minimal value in magnetic leakage signal Serial No., be normalized, the Serial No. obtaining after magnetic leakage signal normalization is D={d ₁, d ₂..., d _k..., d _n, N is the length of Serial No. D, N=100000, wherein k data d in D _kbe calculated as follows:

$d_k=\frac{d_{\mathrm{pk}}-d_{\min }}{d_{\max }-d_{\min }}$

D wherein _pkfor k data point in the magnetic leakage signal Serial No. after sampling filter, d _maxfor the maximum value of this Serial No., d _minfor the minimal value of this Serial No., in the magnetic leakage signal sequence D obtaining after normalization, k data dk meets: 0≤d _k≤ 1, k=1,2 ..., N;

(3a2) interval [0,1] is divided into n minizone, n=100, each interval size is 1/n, calculate the numeric distribution scope of each minizone, that is: first interval [0,1/n), second interval [1/n, 2/n) ..., n interval [99/n, 1], according to data d in magnetic leakage signal sequence D _ksize, count the data amount check m in the magnetic leakage signal sequence D comprising in each minizone _j, j=1,2 ..., n, according to m _jwith the length N of sequence D, obtain the data amount check m of each minizone _jthe ratio p of shared total data number _tj:

$p_{\mathrm{tj}}=\frac{m_j}{N},j=\mathrm{1,2},\·\·\·,n$

By p _tjsubstitution information entropy formula, obtains the time domain entropy H of magnetic leakage signal _t:

$H_t=-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{p}_{\mathrm{tj}}\log p_{\mathrm{tj}}$

When calculating, for making this formula meaningful all the time, stipulate p _tj=0 o'clock,

(3a3) the spectrum entropy H of magnetic leakage signal sequence D after calculating normalization _f, sequence D is carried out to Fourier transform, the sequence F={F after being converted ₁, F ₂..., F _k..., F _n, by the element F in sequence F _ksquare obtain the energy of each frequency content | F _k| ², k=1,2 ..., N, obtains summation after all elements square the gross energy of magnetic leakage signal, the ratio p of the shared gross energy of energy of each element _fkfor:

$p_{\mathrm{fk}}=\frac{{\left|F_k\right|}^2}{\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left|F_k\right|}^2},k=\mathrm{1,2},\·\·\·,N$

By p _fksubstitution information entropy formula, obtains the spectrum entropy H of magnetic leakage signal _f:

$H_f=-\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{p}_{\mathrm{fk}}\log p_{\mathrm{fk}}$

When calculating, for making this formula meaningful all the time, stipulate p _fk=0 o'clock,

(3a4), according to (3a1) to (3a3) same method, extract all the other experiment Switching Power Supply plate B ₁to B _mmagnetic leakage signal spectrum entropy feature and time domain entropy feature, obtain testing Switching Power Supply plate B _ithe spectrum entropy feature H of magnetic leakage signal _fiwith time domain entropy feature H _ti, i=0,1,2 ..., m;

(3b) method of use statistical study is extracted the peak-to-peak value feature V of this power panel magnetic leakage signal _ppi, characteristics of mean a _i, root mean square feature r _iwith Variance feature σ _i, i=0,1,2 ..., m:

(3b1) to the magnetic leakage signal Serial No. s after sampling ₀carry out filtering, filtering spiking, and calculate this magnetic leakage signal peak-to-peak value V _pp0;

(3b2) search magnetic leakage signal Serial No. s after filtering ₀in maximum value and minimal value, be normalized, the Serial No. obtaining after magnetic leakage signal normalization is D={d ₁, d ₂..., d _k..., d _n, N is the length of Serial No. D, N=100000, wherein k data d in D _kbe calculated as follows:

$d_k=\frac{d_{\mathrm{pk}}-d_{\min }}{d_{\max }-d_{\min }}$

D wherein _pkfor k data point in the magnetic leakage signal Serial No. after sampling filter, d _maxfor the maximum value of this Serial No., d _minfor the minimal value of this Serial No., k data d in the magnetic leakage signal sequence D obtaining after normalization _kmeet: 0≤d _k≤ 1, k=1,2 ..., N;

(3b3) obtain the average a of the D of Serial No. ₀,

(3b4) obtain the root mean square r of the D of Serial No. ₀,

(3b5) obtain the variances sigma of the D of Serial No. ₀,

(3b6), according to (3b1) to (3b5) same method, extract all the other experiment Switching Power Supply plate B ₁to B _mmagnetic leakage signal spectrum peak-to-peak value feature, characteristics of mean, root mean square feature, Variance feature, obtain testing Switching Power Supply plate B _ithe peak-to-peak value feature V of magnetic leakage signal _ppi, characteristics of mean a _i, root mean square feature r _iwith Variance feature σ _i, i=0,1,2 ..., m.

(4) set up the list of feature values of respectively testing Switching Power Supply plate magnetic leakage signal

(4a) change the measuring position of non-contact magnetically probe, repeating step (2)-(3) g time, 10≤g≤20, make each test every kind of feature of Switching Power Supply plate, all obtain g data;

(4b) to experimental power supply plate B _ithe g of each a feature data are averaged respectively and maximum deviation, then obtain the distributed area of each feature:

(4b1) obtain experimental power supply plate B _ithe average of g spectrum entropy characteristic

the average of time domain entropy feature

the average of peak-to-peak value feature

the average of characteristics of mean

the average of root mean square feature

the average of Variance feature

i=0,1,2 ..., m;

(4b2) obtain experimental power supply plate B _ithe maximum deviation Δ H of g spectrum entropy characteristic _fi, i.e. Δ H _figet g spectrum entropy characteristic and average

in poor absolute value maximum one; In like manner obtain experimental power supply plate B _ithe maximum deviation Δ H of time domain entropy feature _ti, the maximum deviation Δ V of peak-to-peak value feature _ppi, the maximum deviation Δ a of characteristics of mean _i, the maximum deviation Δ r of root mean square feature _ithe maximum deviation Δ σ of Variance feature _i, i=0,1,2 ..., m;

(4b3) obtain experimental power supply plate B _ithe distributed area of spectrum entropy feature

the distributed area of time domain entropy feature the distributed area of peak-to-peak value feature

the distributed area of characteristics of mean

the distributed area of root mean square feature

the distributed area of Variance feature i=0,1,2 ..., m;

(4c) distributed area of each each feature of experimental power supply plate is write in table, set up the list of feature values of respectively testing Switching Power Supply plate magnetic leakage signal.

Step 2, to tested power panel, fault and fault type are diagnosed

With reference to Fig. 2, being implemented as follows of this step:

(1) adopt cordless to obtain the magnetic leakage signal of tested Switching Power Supply plate magnetic element;

(2) extract each feature of tested Switching Power Supply plate magnetic leakage signal, use information entropy principle to extract the time domain entropy feature H in the magnetic leakage signal time domain of tested Switching Power Supply plate magnetic element _twith the spectrum entropy feature H in frequency domain _f, and by the method for statistical study, extract the peak-to-peak value feature V of this power panel magnetic leakage signal _pp, characteristics of mean a, root mean square feature r and Variance feature σ;

(3) search the list of feature values and carry out fault diagnosis

Read the feature distributed area of the first experiment Switching Power Supply plate magnetic leakage signal in the list of feature values, judge that each feature of tested power panel is whether all in the interval of the individual features of this experiment Switching Power Supply plate magnetic leakage signal, if all in the interval of individual features, the fault of tested power panel is fault corresponding to this experiment Switching Power Supply plate, otherwise read lower a kind of feature distributed area of testing Switching Power Supply plate magnetic leakage signal in the list of feature values, judge successively, until determine the fault type of tested Switching Power Supply plate; If searched the list of feature values, still cannot find the experiment Switching Power Supply plate with tested Switching Power Supply with same fault, by manual method, carry out fault and determine, and its fault signature is filled in the list of feature values.

Use diagnostic method of the present invention to carry out diagnostic test to a single-end flyback switching power supply, result shows, diagnostic method of the present invention can be diagnosed the common various faults of Switching Power Supply, and the accuracy of fault diagnosis is up to more than 90%.

This embodiment is with reference to explanation, not form any restriction to content of the present invention to of the present invention, uses other signal being directly proportional to magnetic leakage signal on Switching Power Supply plate, as PWM drives signal, adopts the diagnostic method of this invention also can carry out fault diagnosis.

Claims (4)

1. the Switching Power Supply method for diagnosing faults based on information entropy principle, comprises the following steps:

(1) set up the list of feature values of various each feature distributed areas of experiment Switching Power Supply plate magnetic leakage signal:

(1a) choose respectively a trouble-free normal experiment Switching Power Supply plate and be labeled as B ₀there is the experiment Switching Power Supply plate of different faults type with m piece, respectively mark B ₁, B ₂..., B _m, m≤100, wherein fault type comprises: the short circuit of output commutation diode, feedback power supply diode open-circuit, current sample filtering circuit resistance disconnect, shake tank capacitance open circuit, feedback divider resistance open circuit and output terminal short circuit;

(1b) adopt cordless to obtain the magnetic leakage signal of various experiment Switching Power Supply plate magnetic elements:

(1b1) at experiment Switching Power Supply plate B _ifixed resistance of upper connection, as the load of Switching Power Supply, and can guarantee that Switching Power Supply plate works under non-failure conditions, and gives this Switching Power Supply plate power supply, i=0, and 1,2 ..., m;

(1b2) non-contact magnetically probe is placed in to experiment Switching Power Supply plate B _imagnetic element near any position of 1～5cm, measure the magnetic leakage signal of magnetic element, and the simulating signal of measuring amplified to conditioning and sampling, obtain and test Switching Power Supply plate B _ithe Serial No. s of magnetic leakage signal _i, i=0,1,2 ..., m;

(1c) use information entropy principle to extract experiment Switching Power Supply plate magnetic element magnetic leakage signal s _itime domain entropy feature in time domain and the spectrum entropy feature in frequency domain, and extract magnetic leakage signal s by the method for statistical study _ipeak-to-peak value feature, characteristics of mean, root mean square feature and Variance feature, obtain testing Switching Power Supply plate B ₀to B _meach feature of magnetic leakage signal, wherein test Switching Power Supply plate B _ithe feature of magnetic leakage signal comprise: spectrum entropy H _fi, time domain entropy H _ti, peak-to-peak value V _ppi, average a _i, root mean square r _iand variances sigma _i, i=0,1,2 ..., m;

(1d) change the measuring position of non-contact magnetically probe, repeating step (1b)-(1c) g time, 10≤g≤20, the experiment Switching Power Supply plate B that step (1c) is extracted _ieach feature of magnetic leakage signal, obtain respectively average and the maximum deviation of g data, then obtain the distributed area of each feature, and the distributed area of each each feature of experimental power supply plate is write in table, i=0,1,2,, m, sets up the list of feature values of respectively testing Switching Power Supply plate magnetic leakage signal;

(2) to tested Switching Power Supply plate whether fault and fault type diagnose:

(2a) adopt cordless to obtain the magnetic leakage signal of tested Switching Power Supply plate magnetic element;

(2b) use information entropy principle to extract the time domain entropy feature H in tested Switching Power Supply plate magnetic element magnetic leakage signal time domain _twith the spectrum entropy feature H in frequency domain _f, and by the method for statistical study, extract the peak-to-peak value feature V of this Switching Power Supply plate magnetic leakage signal _pp, characteristics of mean a, root mean square feature r and Variance feature σ;

(2c) read the feature distributed area of the first experiment Switching Power Supply plate magnetic leakage signal in the list of feature values, judge that each feature of tested power panel is whether all in the interval of the individual features of this experiment Switching Power Supply plate magnetic leakage signal, if all in the interval of individual features, the fault of tested power panel is fault corresponding to this experiment Switching Power Supply plate, otherwise read lower a kind of feature distributed area of testing Switching Power Supply plate magnetic leakage signal in the list of feature values, judge successively, until determine the fault type of tested Switching Power Supply plate, if searched the list of feature values, still cannot find the experiment Switching Power Supply plate with tested Switching Power Supply with same fault, by manual method, carrying out fault determines, and its fault signature is filled in the list of feature values.

2. a kind of Switching Power Supply method for diagnosing faults based on information entropy principle according to claim 1, wherein the described use information entropy principle of step (1c) extracts the time domain entropy feature H in experiment Switching Power Supply plate magnetic element magnetic leakage signal time domain _twith the spectrum entropy feature H in frequency domain _f, carry out as follows:

(1c1) the magnetic leakage signal Serial No. after sampling is carried out to filtering, search maximum value and minimal value in magnetic leakage signal Serial No., be normalized, the Serial No. obtaining after magnetic leakage signal normalization is D={d ₁, d ₂..., d _k..., d _n, N is the length of Serial No. D, N=100000, wherein k data d in D _kbe calculated as follows:

$d_k=\frac{d_{\mathrm{pk}}-d_{\min }}{d_{\max }-d_{\min }}$

D wherein _pkfor k data point in the magnetic leakage signal Serial No. after sampling filter, d _maxfor the maximum value of this Serial No., d _minfor the minimal value of this Serial No., k data d in the magnetic leakage signal sequence D obtaining after normalization _kmeet: 0≤d _k≤ 1, k=1,2 ..., N;

(1c2) the time domain entropy H of magnetic leakage signal sequence D after calculating normalization _t, interval [0,1] is divided into n minizone, n=100, each interval size is 1/n, calculates the numeric distribution scope of each minizone, that is: first interval [0,1/n), second interval [1/n, 2/n),, n interval [99/n, 1], according to the size of data in magnetic leakage signal sequence D, count the data amount check m in the magnetic leakage signal sequence D comprising in each minizone _j, j=1,2 ..., n, according to m _jwith the length N of sequence D, obtain the ratio p of the shared total data number of data amount check of each minizone _tj:

$p_{\mathrm{tj}}=\frac{m_j}{N},j=\mathrm{1,2},...,n$

Will _ptjsubstitution information entropy formula, obtains the time domain entropy H of magnetic leakage signal _t:

$H_t=-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{p}_{\mathrm{tj}}{\log p}_{\mathrm{tj}}$

When calculating, for making this formula meaningful all the time, stipulate p _tj=0 o'clock,

(1c3) the spectrum entropy H of magnetic leakage signal sequence D after calculating normalization _f, sequence D is carried out to Fourier transform, the sequence F={F after being converted ₁, F ₂..., F _k..., F _n, by the element F in sequence F _ksquare obtain the energy of each frequency content | F _k| ², k=1,2 ..., N, obtains summation after all elements square the gross energy of magnetic leakage signal, and the energy of each element accounts for the ratio p of gross energy _fkfor:

$p_{\mathrm{fk}}=\frac{{\left|F_k\right|}^2}{\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left|F_k\right|}^2},k=\mathrm{1,2},...,N$

By p _fksubstitution information entropy formula, obtains the spectrum entropy H of magnetic leakage signal _f:

$H_f=-\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{p}_{\mathrm{fk}}{\log p}_{\mathrm{fk}}$

When calculating, for making this formula meaningful all the time, stipulate p _fk=0 o'clock,

3. a kind of Switching Power Supply method for diagnosing faults based on information entropy principle according to claim 1, wherein the method for the described use statistical study of step (1c) is extracted the peak-to-peak value feature V of this power panel magnetic leakage signal _pp, characteristics of mean a, root mean square feature r and Variance feature σ, carry out as follows:

(1c.1) the magnetic leakage signal Serial No. after sampling is carried out to filtering, filtering spiking, and calculate this magnetic leakage signal peak-to-peak value V _pp;

(1c.2) search maximum value and the minimal value in magnetic leakage signal Serial No. after filtering, be normalized, the Serial No. obtaining after magnetic leakage signal normalization is D={d ₁, d ₂..., d _k..., d _n, N is the length of Serial No. D, N=100000, wherein k data d in D _kbe calculated as follows:

$d_k=\frac{d_{\mathrm{pk}}-d_{\min }}{d_{\max }-d_{\min }}$

D wherein _pkfor k data point in the magnetic leakage signal Serial No. after sampling filter, d _maxfor the maximum value of this Serial No., d _minfor the minimal value of this Serial No., k data d in the magnetic leakage signal sequence D obtaining after normalization _kmeet: 0≤d _k≤ 1, k=1,2 ..., N;

(1c.3) obtain the average a of the D of Serial No.,

(1c.4) obtain the root mean square r of the D of Serial No.,

(1c.5) obtain the variances sigma of the D of Serial No.,

4. a kind of Switching Power Supply method for diagnosing faults based on information entropy principle according to claim 1, wherein step (1d) described to experimental power supply plate B _ithe g of each a feature data are averaged respectively and maximum deviation, then obtain the distributed area of each feature, carry out as follows:

(1d1) obtain experimental power supply plate B _ithe average of g spectrum entropy characteristic

the average of time domain entropy feature

the average of peak-to-peak value feature

the average of characteristics of mean

the average of root mean square feature

the average of Variance feature

i=0,1,2 ..., m;

(1d2) obtain experimental power supply plate B _ithe maximum deviation Δ H of g spectrum entropy characteristic _fi, i.e. Δ H _figet g spectrum entropy characteristic and average in poor absolute value maximum one; In like manner obtain experimental power supply plate B _ithe maximum deviation Δ H of time domain entropy feature _ti, the maximum deviation Δ V of peak-to-peak value feature _ppi, the maximum deviation Δ a of characteristics of mean _i, the maximum deviation Δ r of root mean square feature _i, the maximum deviation Δ σ of Variance feature _i, i=0,1,2 ..., m;

(1d3) obtain experimental power supply plate B _ithe distributed area of spectrum entropy feature

the distributed area of time domain entropy feature

the distributed area of peak-to-peak value feature

the distributed area of characteristics of mean

the distributed area of root mean square feature

the distributed area of Variance feature

Images