CN106841949A - Three-phase asynchronous Ac motor stator insulation on-line monitoring method and device - Google Patents

Abstract

The invention discloses a kind of three-phase asynchronous Ac motor stator insulation on-line monitoring method, step：The voltage of synchronous acquisition motor electrical insulation system, current signal；The voltage of the motor to collecting, current signal are processed using PWM demodulation techniques；Characteristic index extraction is carried out to the signal after treatment, extraction includes MCSA characteristic indexs, current signal statistical-simulation spectrometry index, sequence voltage distribution statisticses index, Park transform characteristics index, three-phase alternating current pressure difference statistical indicator and three-phase alternating current difference statistical indicator；Each index that will be extracted carries out state and examines in advance as the training set of SVMs, and the comprehensive state of output motor insulation system examines result in advance, and the running status to motor electrical insulation system carries out online evaluation.The motor electrical insulation system on-line monitoring method that the present invention is provided, is capable of achieving the permanently effective monitoring to motor state of insulation；Multiclass dielectric features index can be merged, increased the reliability and monitoring accuracy of method, data processing precision is high.

Description

Three-phase asynchronous Ac motor stator insulation on-line monitoring method and device

Technical field

The present invention relates to electrical equipment monitoring technical field, more particularly to a kind of three-phase asynchronous Ac motor stator insulation exists Line monitoring method and device.

Background technology

Generally, three-phase asynchronous Ac motor is the key device in many engineering fields, is widely used at a high speed The fields such as train, Aero-Space, electric vehicle, in the last few years, the security incident of three-phase asynchronous Ac motor takes place frequently, wherein motor The turn-to-turn insulation failure of insulation system is to cause one of the principal element of motor accident.Therefore, it is fixed to three-phase asynchronous Ac motor Insulating sublayer carries out on-line monitoring assessment, significant for improving motor device reliability.At present, for motor insulation event A series of existing detection methods of barrier, existing off-line checking method such as AC impedence method, power attenuation method, differential coil probe method Deng, and the online test method such as test of insulation current method, stator current, magnetic flux leakage, order impedance matrix, high-frequency resistance and circle Between capacity measurement etc..

Above-mentioned motor insulation failure detection method can only assess a certain performance of insulation system, and some detection methods Testing equipment is more complicated, heaviness, and field test is very inconvenient, convenience, measurement sensitivity and applicable situation in test etc. There is certain deficiency in aspect, and analysis to test result is overly dependent upon expert system and professional's experience, to visitor See, show that test result has certain limitation exactly.

The content of the invention

The present invention in the prior art can not be while assess multiple performance and the on-line monitoring technique in detection in order to overcome The problems such as low low precision of reliability, high cost, there is provided a kind of three-phase asynchronous Ac motor stator insulation on-line monitoring method and dress Put.

To achieve these goals, the present invention is realized by following scheme：

A kind of three-phase asynchronous Ac motor stator insulation on-line monitoring method, comprises the following steps：

The voltage of synchronous acquisition motor electrical insulation system, current signal；

The voltage of the motor to collecting, current signal are processed using PWM demodulation techniques；

Characteristic index extraction is carried out to the signal after treatment, extraction includes MCSA characteristic indexs, current signal statistical model Distinguishing indexes, sequence voltage distribution statisticses index, Park transform characteristics index, three-phase alternating current pressure difference statistical indicator and three-phase alternating current Difference between current statistical indicator；

Each index that will be extracted carries out state and examines in advance as the training set of SVMs, output motor insulation system Comprehensive state examine result in advance, the running status to motor electrical insulation system carries out online evaluation.

Used as a kind of embodiment, in step 3, the extraction MCSA characteristic indexs are concretely comprised the following steps：

The current signal of motor will be collected carries out Fourier transformation and the current spectrum after being converted；

When the insulation system state of motor changes, stator winding current can produce the fundamental wave of special characteristic frequency；

By analyzing motor stator current spectrum, detection produces whether the fundamental wave component of special characteristic frequency is high-order component The harmonic wave of frequency, determines whether motor residing running status at present.

As a kind of embodiment, in step 3, the extraction current signal statistical-simulation spectrometry index, including with Lower step：

Current information to collecting is clustered；

The parameter of the mixed Gauss model of estimated current signal, obtains the Gaussian mixtures of current signal, by analysis The Gaussian mixtures under Gaussian mixtures and motor free position under motor health status, show that current signal counts mould Formula distinguishing indexes, the state of decision circuitry insulation system.

Used as a kind of embodiment, the parameter of the mixed Gauss model of the estimated current signal obtains current signal Gaussian mixtures, by analyzing the mixed Gaussian under the Gaussian mixtures under motor health status and motor free position Distribution, draw current signal statistical-simulation spectrometry index, and the state of decision circuitry insulation system is comprised the following steps that：

Initialize the mean μ of each Gaussian Profile_k, variances sigma_kAnd corresponding weight π_k,

Wherein N (x/ μ_j,σ_j) it is j-th Gauss model probability density, wherein μ_jJ-th average of single Gaussian component is represented, σ_jIt is the j covariance matrix of single Gaussian component, π_jThe shared proportion in global mixed Gaussian of j-th single Gaussian component of expression, 0≤π_j≤1；

Probability γ (k) of each Gauss model generation is calculated,

Average, variance and factor of influence according to each Gaussian Profile of Probability estimate, formula are as follows,

π_k=N_k/N (5)

Recurrent formula (3), (4), (5), (6), average, variance and weight factor convergence until trying to achieve Gaussian Profile,

Assuming that H (x) represents the Gaussian mixtures under motor health status, G (x) represents the mixing under motor free position Gaussian Profile, then have：

During p=1, Z represents the manhatton distance between two distributions of H (x) and G (x), and during p=2, Z is represented between two distributions Euclidean distance,

During p=∞, Z represents the Chebyshev's distance between two distributions of H (x) and G (x),

Therefore, Z as the state characteristic quantity of circuit isolation system, can characterize the state of insulation system,

Quantify similarity degree between the two by calculating H (x) and G (x) distribution overlap index between the two.

Used as a kind of embodiment, in step 3, the extraction sequence voltage distribution statisticses index is comprised the following steps that：

Three-phase imbalance voltage or electric current can be resolved into positive-sequence component, negative sequence component and zero-sequence component in the same direction, it is public Formula is as follows：

Wherein, V₀Represent residual voltage, V_PRepresent positive sequence voltage, V_nNegative sequence voltage is represented, in this α=e^j2π/3, j is expressed as Imaginary unit；

Positive sequence voltage, negative sequence voltage, residual voltage are converted into forward-order current, negative-sequence current, zero-sequence current, formula is such as Under：

Wherein, I₀Represent zero-sequence current, I_PRepresent forward-order current, I_nRepresent negative sequence voltage, z_xyCorrespondence motor impedance, V₀Table Show residual voltage, V_PRepresent positive sequence voltage, V_nRepresent negative sequence voltage；

When motor is in health status, according to Kirchhoff's law, the sum of zero of three-phase current, zero-sequence current I₀For 0, corresponding residual voltage V₀It is zero, formula (9) is converted into below equation：

V_p=z₁₁I_p+z₁₂I_n (10)

V_n=z₂₁I_p+z₂₂I_n (11)

In formula (10) (11), electric current and voltage data when being run well according to motor calculate the impedance factor of motor Matrix；

When motor state of insulation changes, then positive sequence voltage or negative sequence voltage and corresponding electric current number now is measured According to according in formula (8), formula (9), formula (10), formula (11), drawing positive sequence voltage V_PWith negative sequence voltage V_n；

When motor breaks down, the positive sequence voltage V of calculating_PWith negative sequence voltage V_nWith positive sequence voltage and the negative phase-sequence electricity of measurement Pressure can there is some difference, the positive sequence voltage V that will be calculated_PWith negative sequence voltage V_nThe positive sequence voltage and negative phase-sequence obtained with measurement The difference of the presence between voltage is used as the characteristic index for judging motor operating state.

Used as a kind of embodiment, in step 3, the extraction Park transform characteristics indexs are comprised the following steps that：

When motor state of insulation changes, stator winding current produces the fundamental wave component and electric current of fault characteristic frequency Park's vector locus can change, according to Park convert, the positive-sequence component of fundamental wave in electric current is converted to dqo coordinates DC component, the negative sequence component of fundamental wave is converted to the AC-frequency component 2f of dqo coordinates₁, by stator current Park Vector Mode square carries out spectrum analysis, and observation whether there is 2f₁Frequency component judges whether stator winding breaks down, using friendship Stream frequency component 2f₁Amplitude size as circuit isolation system fault signature index.

As a kind of embodiment, in step 3, the extraction three-phase alternating current pressure difference statistical indicator, concrete operations Step is as follows：

When motor state of insulation is under health status, three-phase voltage value V is measured₁, appoint when motor state of insulation is in During meaning state, the three-phase voltage value V that measurement is obtained₂, by V₁And V₂Difference V between the two as motor electrical insulation system state Characteristic quantity.

As a kind of embodiment, in step 3, the extraction three-phase alternating current difference statistical indicator, specific steps It is as follows：

When motor state of insulation is under health status, three-phase electricity flow valuve I is measured₁, appoint when motor state of insulation is in During meaning state, the three-phase electricity flow valuve I that measurement is obtained₂, by I₁And I₂Difference I between the two as motor electrical insulation system state Characteristic quantity.

As a kind of embodiment, in step 4, training set of the state index amount that will be obtained as SVMs The state of carrying out is examined in advance, and the running status of output motor insulation system examines result in advance, and concrete operation step is as follows：

According to the six class state characteristic indexs for obtaining, six category feature indexs are：MCSA characteristic indexs, current signal statistics mould Formula distinguishing indexes, sequence voltage distribution statisticses index, Park transform characteristics index, three-phase alternating current pressure difference statistical indicator and three are intersected Stream difference between current statistical indicator, sets up training dataset (ξ_s,ζ_s), ξ_sIt is the value of input variable, ζ_sIt is corresponding input variable value, S It is training set number；

Using gaussian radial basis function：K(ξ_s,ζ_s)=exp (- | ξ_s-ζ|²/σ² ₀), in formula, K (ξ_s,ζ_s) it is core letter Number, σ₀It is the variance of Gaussian function；

From training dataset (ξ_s,ζ_s) SVMs is trained, the state for setting up SVMs examines mould in advance Type；

Diagnostic sample is treated with the SVMs for training carries out diagnosis output, and the state of output motor insulation system is pre- Examine model.

A kind of three-phase asynchronous Ac motor stator insulation on-Line Monitor Device, including frequency converter, motor, data acquisition with deposit Storage module, Data Management Analysis module, pattern-recognition judge module, control computer, controller and load；

According to the ruuning situation of load, the voltage and frequency of control computer regulation frequency converter control the defeated of the motor Go out rotating speed and torque, the Data acquisition and storage module gathers the voltage of the controller, current signal and will collect Voltage, current signal transfer shown and processed to the Data Management Analysis module, and feature is carried out to the signal after treatment Index extraction, the characteristic index of extraction is transferred into the pattern-recognition judge module carries out pattern-recognition, assesses the motor Stator insulation real-time running state；

Described Data acquisition and storage module is arranged on inside the controller, Data acquisition and storage module be used for from The controller of motor obtains the real-time voltage signal and current signal of motor, and in sequence or numbering is stored.

The present invention has advantages below：

A kind of three-phase asynchronous Ac motor stator insulation on-Line Monitor Device, including, frequency converter, motor, data acquisition with Memory module, Data Management Analysis module, pattern-recognition judge module, control computer, controller and load；Can subtest The state that personnel complete motor stator insulated system is examined and fault location in advance, and display state assessment result directly perceived, operating process It is easy to be reliable, be conducive to improving the manufacture and operation of three-phase asynchronous Ac motor stator；

The motor electrical insulation system on-line monitoring method that the present invention is provided, is capable of achieving to the permanently effective of motor state of insulation Monitoring；Multiclass dielectric features index can be merged, so as to increased the reliability and monitoring accuracy of method, with real-time By force, data processing precision is high, core algorithm robustness is good, state estimation accuracy high and the advantages of controllable cost, it is to avoid pass On-line monitoring method sensor installation difficulty is high and poor real for system, and data are not preprocessed, and to be analyzed the precision for causing low Under, and install additional great number cost that sensor causes and it is uncertain the problems such as.

It is that above and other objects of the present invention, feature and advantage can be become apparent, preferred embodiment cited below particularly, And coordinate accompanying drawing, it is described in detail below.

Brief description of the drawings

Fig. 1 is method of the present invention schematic flow sheet；

Fig. 2 is apparatus structure schematic diagram of the invention；

Fig. 3 is the threephase asynchronous machine current signal time domain beamformer collected in the present invention；

Fig. 4 is current signal MCSA spectrograms under the threephase asynchronous machine health status in the present invention；

Fig. 5 is current signal MCSA spectrograms under the threephase asynchronous machine malfunction in the present invention；

Fig. 6 is motor in present invention positive sequence voltage distribution statisticses indicatrix under normal circumstances；

Fig. 7 is the positive sequence voltage distribution statisticses indicatrix under motor in the present invention breaks down；

Fig. 8 is the current signal Park transform characteristics indicatrixs in the present invention；

Fig. 9 is the voltage signal Park transform characteristics indicatrixs in the present invention.

Specific embodiment

Below in conjunction with accompanying drawing, the technical characteristic above-mentioned and other to the present invention and advantage are clearly and completely described, Obviously, described embodiment is only section Example of the invention, rather than whole embodiments.

The invention provides the three-phase asynchronous Ac motor insulation system on-line monitoring method based on PHM, its flow chart is such as Shown in Fig. 1.

S1, the voltage of synchronous acquisition motor electrical insulation system, current signal；

S2, the voltage of motor to collecting, current signal are processed using PWM demodulation techniques；

S3, characteristic index extraction is carried out to the signal after treatment, extraction includes that electric current MCSA characteristic indexs, current signal are united Meter pattern-recognition index, sequence voltage distribution statisticses index, Park transform characteristics index, three-phase alternating current pressure difference statistical indicator and three Cross streams difference between current statistical indicator；

S4, each index that will be extracted carry out state and examine in advance as the training set of SVMs, output motor insulation The comprehensive state of system examines result in advance, and the running status to motor electrical insulation system carries out online evaluation.

A kind of three-phase asynchronous Ac motor stator insulation on-Line Monitor Device, such as Fig. 2 insulate for three-phase asynchronous Ac motor The structural representation of system on-Line Monitor Device, the embodiment include frequency converter 1, motor 2, Data acquisition and storage module 3, Data Management Analysis module 4, pattern-recognition judge module 5, control computer 6, controller 7, load 8.In device actual motion, According to the actual conditions of load 8, the voltage and frequency of the regulation frequency converter 1 of control computer 6, and then phase asynchronous exchange can be controlled The output speed of motor 2 and torque, meet the work requirements of actual loading.The direct acquisition control of Data acquisition and storage module 3 The voltage of device 7, current signal, and send data to Data Management Analysis module 4 and processed；Data Management Analysis module 4 Built-in PWM demodulating algorithms and feature extraction algorithm, can be demodulated and feature extraction to signal, and the characteristic index that will be handled well is passed It is defeated to be estimated to pattern-recognition judge module 5, to judge that motor electrical insulation system whether there is turn-to-turn insulation failure；To finally comment Estimate result to transmit to control computer 6, so as to reach the purpose of the motor stator insulated state of real-time assessment.Additionally, data acquisition With memory module 3, the data that Data Management Analysis module 4 and pattern-recognition judge module 5 will be obtained are transmitted to control with result Computer processed 6 is stored accordingly.

The modules in device are built, logical and test device is adjusted, by Data acquisition and storage module 3 directly from motor Controller in obtain motor voltage signal and current signal, obtain a series of signal data.

Fig. 3 is refer to, expression is the threephase asynchronous machine current signal time domain beamformer for collecting；

How introduction in detail below extracts each characteristic index：

In S3, the extraction electric current MCSA characteristic indexs are concretely comprised the following steps：

The current signal of motor will be collected carries out Fourier transformation and the spectrogram after being converted；

When the insulation system state of motor changes, stator winding current can produce the fundamental wave of special characteristic frequency, By analyzing motor stator current spectrum, detection produces whether the fundamental wave component of special characteristic frequency is the humorous of high-order component frequency Ripple, determines whether motor residing running status at present.If there is fundamental wave and with the frequency phase of constant component in current signal Mutually correspondence, then illustrate that motor have failed.

Fourier transformation treatment is done respectively to current signal under current signal under the health status that collects and malfunction, Current signal MCSA statistical indicators are tried to achieve, as a result as shown in Figure 4,5, is understood in comparison diagram 4 and Fig. 5, the frequency of normal current signal Frequency content is simple in spectrogram, has no high fdrequency component, and there is many places high fdrequency component in the spectrogram of fault-current signal.Therefore Can search whether there is corresponding frequency content on spectrogram, the running status residing at present to determine motor.

In S3, the extraction current signal statistical-simulation spectrometry index is comprised the following steps：

Current information to collecting is clustered；

The parameter of the mixed Gauss model of estimated current signal, obtains the Gaussian mixtures of current signal, by analysis The Gaussian mixtures under Gaussian mixtures and motor free position under motor health status, obtain circuit isolation system State characteristic quantity.

In S3, the parameter of the mixed Gauss model of the estimated current signal obtains the mixed Gaussian point of current signal Cloth, by analyzing the Gaussian mixtures under the Gaussian mixtures under motor health status and motor free position, obtains electricity The state characteristic quantity of road insulation system is comprised the following steps that：

Initialize the mean μ of each Gaussian Profile_k, variances sigma_kAnd corresponding weight π_k,

Wherein N (x/ μ_j,σ_j) it is j-th Gauss model probability density, wherein μ_jJ-th average of single Gaussian component is represented, σ_jIt is the j covariance matrix of single Gaussian component, π_jThe shared proportion in global mixed Gaussian of j-th single Gaussian component of expression, 0≤π_j≤1；

Probability γ (k) of each Gauss model generation is calculated,

Average, variance and factor of influence according to each Gaussian Profile of Probability estimate, formula are as follows,

π_k=N_k/N (5)

Recurrent formula (3), formula (4), formula (5), formula (6), calculating above-mentioned steps, until the average of Gaussian Profile, Variance and weight factor restrain,

Assuming that H (x) represents the Gaussian mixtures under motor health status, G (x) represents the mixing under motor free position Gaussian Profile, then have：

During p=1, Z represents the manhatton distance between two distributions of H (x) and G (x), and during p=2, Z is represented between two distributions Euclidean distance,

During p=∞, Z represents the Chebyshev's distance between two distributions of H (x) and G (x),

Therefore, Z as the state characteristic quantity of circuit isolation system, can characterize the state of insulation system,

Quantify similarity degree between the two by calculating H (x) and G (x) distribution overlap index between the two.

In S3, the extraction sequence voltage distribution statisticses index is comprised the following steps that：

Three-phase imbalance voltage or electric current can be resolved into positive-sequence component, negative sequence component and zero-sequence component in the same direction, it is public Formula is as follows：

Wherein, V₀Represent residual voltage, V_PRepresent positive sequence voltage, V_nNegative sequence voltage is represented, in this α=e^j2π/3, j is expressed as Imaginary unit；

Positive sequence voltage, negative sequence voltage, residual voltage are converted into forward-order current, negative-sequence current, zero-sequence current, formula is such as Under：

Wherein, I₀Represent zero-sequence current, I_PRepresent forward-order current, I_nRepresent negative sequence voltage, z_xyCorrespondence motor impedance, V₀Table Show residual voltage, V_PRepresent positive sequence voltage, V_nRepresent negative sequence voltage；

When motor is in health status, according to Kirchhoff's law, the summation of three-phase current will be zero, zero-sequence current I₀ It is 0, corresponding residual voltage V₀Zero is all, formula (9) is converted into below equation：

V_p=z₁₁I_p+z₁₂I_n (10)

V_n=z₂₁I_p+z₂₂I_n (11)

In formula (10) (11), electric current and voltage data when being run well according to motor calculate the impedance factor of motor Matrix；

When motor state of insulation changes, then positive sequence voltage or negative sequence voltage and corresponding electric current number now is measured According to according in formula (8), formula (9), formula (10), formula (11), drawing positive sequence voltage V_PWith negative sequence voltage V_n；

When motor breaks down, the positive sequence voltage V of calculating_PWith negative sequence voltage V_nWith positive sequence voltage and the negative phase-sequence electricity of measurement Pressure can there is some difference, the positive sequence voltage V that will be calculated_PWith negative sequence voltage V_nThe positive sequence voltage and negative phase-sequence obtained with measurement The difference of the presence between voltage is used as the characteristic index for judging motor operating state.

Electric current according to the threephase asynchronous machine for collecting calculates positive sequence voltage distribution statisticses index with voltage data, calculates Result is as shown in Figure 6,7.Comparison diagram 6 and Fig. 7 understand, the positive sequence voltage of calculating and the positive sequence of measurement in motor under normal circumstances Voltage value coincide substantially, only exists less gap.In the case of electrical fault, the positive sequence voltage of calculating is with measurement just There is larger gap in sequence voltage, and failure is more serious, and gap is bigger.Therefore, can be by the positive and negative sequence voltage being calculated and survey Distribution distance between the positive and negative sequence voltage for measuring is used as the characteristic index for judging motor operating state.

In S3, the extraction Park transform characteristics indexs are comprised the following steps that：

When motor state of insulation changes, stator winding current produces the fundamental wave component and electric current of fault characteristic frequency Park's vector locus can change, according to Park convert, the positive-sequence component of fundamental wave in electric current is converted to dqo coordinates DC component, the negative sequence component of fundamental wave is converted to the 2f of dqo coordinates₁Component, by stator current Park Vector Modes Spectrum analysis square is carried out, observation whether there is AC-frequency component 2f₁To judge whether stator winding breaks down, using exchange frequency Rate component 2f₁Amplitude size as circuit isolation system fault signature index.Referring to Fig. 8,9, to the electricity under health status Current signal under stream signal and malfunction carries out Park conversion, line frequency analysis of spectrum of going forward side by side, and the Park for obtaining two class signals becomes Change characteristic index figure.As it can be observed in the picture that the current signal under health status is a circle under dqo coordinate systems, and malfunction Track of the current signal in dqo coordinate systems can be distorted, and ellipse is turned into by circle；

As can be seen from Figure 9, when stator insulation is under health status, only direct current divides in the spectrogram after signal Park conversion Amount, without frequency content.When stator insulation system jam, there is one in the spectrogram after signal Park conversion by negative phase-sequence 2f₁The frequency content that component causes, and failure is more serious, the frequency content amplitude is bigger.Therefore can be by detecting electric current Park Frequency component in Vector Mode judges whether motor occurs stator insulation failure.

In S3, the extraction three-phase alternating current pressure difference statistical indicator, concrete operation step is as follows：

When motor state of insulation is under health status, three-phase voltage value V is measured₁, appoint when motor state of insulation is in During meaning state, the three-phase voltage value V that measurement is obtained₂, by V₁And V₂Difference V between the two as motor electrical insulation system state Characteristic quantity.

In S3, the extraction three-phase alternating current difference statistical indicator is comprised the following steps that：

When motor state of insulation is under health status, three-phase electricity flow valuve I is measured₁, appoint when motor state of insulation is in During meaning state, the three-phase electricity flow valuve I that measurement is obtained₂, by I₁And I₂Difference I between the two as motor electrical insulation system state Characteristic quantity.

Further, the state index amount that will be obtained carries out state and examines in advance as the training set of SVMs, output electricity The running status of machine insulation system examines result in advance, and concrete operation step is as follows：

According to the six class state characteristic index T for obtaining, training dataset (ξ is set up_s,ζ_s), ξ_sIt is input variable Value, ζ_sIt is corresponding input variable value, S is training set number；

Using gaussian radial basis function：K(ξ_s,ζ_s)=exp (- | ξ_s-ζ|²/σ² ₀), in formula, K (ξ_s,ζ_s) it is kernel function, σ₀It is the variance of Gaussian function；

From a certain amount of training dataset (ξ_s,ζ_s) SVMs is trained, the state for setting up SVMs is pre- Examine model；

Diagnostic sample is treated with the SVMs for training carries out diagnosis output, and the state of output motor insulation system is pre- Examine model.

Although the present invention is disclosed above by preferred embodiment, but the present invention is not limited to, it is any to know this skill Skill person, without departing from the spirit and scope of the present invention, can make a little change and retouching, therefore protection scope of the present invention is worked as It is defined depending on claims scope required for protection.

Claims (10)

1. a kind of three-phase asynchronous Ac motor stator insulation on-line monitoring method, it is characterised in that comprise the following steps：

The voltage of synchronous acquisition motor electrical insulation system, current signal；

The voltage of the motor to collecting, current signal are processed using PWM demodulation techniques；

Characteristic index extraction is carried out to the signal after treatment, extraction includes MCSA characteristic indexs, current signal statistical-simulation spectrometry Index, sequence voltage distribution statisticses index, Park transform characteristics index, three-phase alternating current pressure difference statistical indicator and three-phase alternating current Difference statistical indicator；

Each index that will be extracted carries out state and examines in advance as the training set of SVMs, output motor insulation system it is comprehensive Conjunction state examines result in advance, and the running status to motor electrical insulation system carries out online evaluation.

2. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 3, the extraction MCSA characteristic indexs are concretely comprised the following steps：

The current signal of motor will be collected carries out Fourier transformation and the current spectrum after being converted；

When the insulation system state of motor changes, stator winding current can produce the fundamental wave of special characteristic frequency；

By analyzing motor stator current spectrum, detection produces whether the fundamental wave component of special characteristic frequency is high-order component frequency Harmonic wave, determine whether motor residing running status at present.

3. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 3, the extraction current signal statistical-simulation spectrometry index is comprised the following steps：

Current information to collecting is clustered；

The parameter of the mixed Gauss model of estimated current signal, obtains the Gaussian mixtures of current signal, by analyzing motor The Gaussian mixtures under Gaussian mixtures and motor free position under health status, show that current signal statistical model is known Other index, the state of decision circuitry insulation system.

4. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 3, it is characterised in that described The parameter of the mixed Gauss model of estimated current signal, obtains the Gaussian mixtures of current signal, by analyzing motor health The Gaussian mixtures under Gaussian mixtures and motor free position under state, show that current signal statistical-simulation spectrometry refers to Mark, the state of decision circuitry insulation system is comprised the following steps that：

Initialize the mean μ of each Gaussian Profile_k, variances sigma_kAnd corresponding weight π_k,

$N\left(x\right)=\underset{j=1}{\overset{K}{\Σ}}{\mathrm{\π}}_{j}N(x/{\mathrm{\μ}}_j,{\mathrm{\σ}}_j)---\left(1\right)$

Wherein N (x/ μ_j,σ_j) it is j-th Gauss model probability density, wherein μ_jRepresent j-th average of single Gaussian component, σ_jIt is j The covariance matrix of individual single Gaussian component, π_jThe shared proportion in global mixed Gaussian of j-th single Gaussian component of expression, 0≤π_j ≤1；

Probability γ (k) of each Gauss model generation is calculated,

$\mathrm{\γ}\left(k\right)=\frac{{\mathrm{\π}}_{k}N(x/{\mathrm{\μ}}_k,{\mathrm{\σ}}_k)}{\underset{j=1}{\overset{K}{\Σ}}{\mathrm{\π}}_{j}N(x/{\mathrm{\μ}}_j,{\mathrm{\σ}}_j)}---\left(2\right)$

Average, variance and factor of influence according to each Gaussian Profile of Probability estimate, formula are as follows,

${\mathrm{\μ}}_k=\underset{i=1}{\overset{N}{\Σ}}\mathrm{\γ}\left(k\right)x_i/N_k---\left(3\right)$

${\mathrm{\σ}}_k=\underset{i=1}{\overset{N}{\Σ}}\mathrm{\γ}\left(k\right)(x_i-{\mathrm{\μ}}_k){(x_i-{\mathrm{\μ}}_k)}^T/N_k---\left(4\right)$

π_k=N_k/N (5)

$N_k=\underset{i=1}{\overset{N}{\Σ}}\mathrm{\γ}\left(k\right)---\left(6\right)$

Recurrent formula (3), (4), (5), (6), average, variance and weight factor convergence until trying to achieve Gaussian Profile,

Assuming that H (x) represents the Gaussian mixtures under motor health status, G (x) represents the mixed Gaussian under motor free position Distribution, then have：

$Z=\left|\right|H\left(x\right)-G\left(x\right)|{|}_{L_p},(n=1,2,\mathrm{\∞})---\left(7\right)$

During p=1, Z represents the manhatton distance between two distributions of H (x) and G (x), and during p=2, Z represents the Europe between two distributions Formula distance,

During p=∞, Z represents the Chebyshev's distance between two distributions of H (x) and G (x),

Therefore, Z as the state characteristic quantity of circuit isolation system, can characterize the state of insulation system,

Quantify similarity degree between the two by calculating H (x) and G (x) distribution overlap index between the two.

5. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 3, the extraction sequence voltage distribution statisticses index is comprised the following steps that：

Three-phase imbalance voltage or electric current can be resolved into positive-sequence component, negative sequence component and zero-sequence component in the same direction, formula is such as Under：

$\left[\begin{array}{c}{V}_0\\ V_p\\ V_n\end{array}\right]=\frac{1}{3}\left[\begin{array}{ccc}1& 1& 1\\ 1& \mathrm{\α}& {\mathrm{\α}}^2\\ 1& {\mathrm{\α}}^2& \mathrm{\α}\end{array}\right]\left[\begin{array}{c}{V}_u\\ V_v\\ V_w\end{array}\right]---\left(8\right)$

Wherein, V₀Represent residual voltage, V_PRepresent positive sequence voltage, V_nNegative sequence voltage is represented, in this α=e^j2π/3, j is expressed as imaginary number Unit；

Positive sequence voltage, negative sequence voltage, residual voltage are converted into forward-order current, negative-sequence current, zero-sequence current, formula is as follows：

$\left[\begin{array}{c}{V}_0\\ V_P\\ V_n\end{array}\right]=\left[\begin{array}{ccc}{z}_{00}& z_{01}& z_{02}\\ z_{10}& z_{11}& z_{12}\\ z_{20}& z_{21}& z_{22}\end{array}\right]\left[\begin{array}{c}{I}_0\\ I_p\\ I_n\end{array}\right]---\left(9\right)$

Wherein, I₀Represent zero-sequence current, I_PRepresent forward-order current, I_nRepresent negative sequence voltage, z_xyCorrespondence motor impedance, V₀Represent zero Sequence voltage, V_PRepresent positive sequence voltage, V_nRepresent negative sequence voltage；

When motor is in health status, according to Kirchhoff's law, the sum of zero of three-phase current, zero-sequence current I₀It is 0, institute Corresponding residual voltage V₀It is zero, formula (9) is converted into below equation：

V_p=z₁₁I_p+z₁₂I_n (10)

V_n=z₂₁I_p+z₂₂I_n (11)

In formula (10) (11), electric current and voltage data when being run well according to motor calculate the impedance factor square of motor Battle array；

When motor state of insulation changes, then positive sequence voltage now or negative sequence voltage and corresponding current data are measured, According in formula (8), formula (9), formula (10), formula (11), positive sequence voltage V is drawn_PWith negative sequence voltage V_n；

When motor breaks down, the positive sequence voltage V of calculating_PWith negative sequence voltage V_nPositive sequence voltage and negative sequence voltage meeting with measurement There is some difference, the positive sequence voltage V that will be calculated_PWith negative sequence voltage V_nThe positive sequence voltage and negative sequence voltage obtained with measurement Between presence difference as the characteristic index for judging motor operating state.

6. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 3, the extraction Park transform characteristics indexs are comprised the following steps that：

When motor state of insulation changes, stator winding current produces the fundamental wave component and electric current of fault characteristic frequency Park's vector locus can be changed, and be converted according to Park, and the positive-sequence component of fundamental wave in electric current is converted into dqo coordinates DC component, the negative sequence component of fundamental wave is converted to the AC-frequency component 2f of dqo coordinates₁, sweared by stator current Park Amount mould square carries out spectrum analysis, and observation whether there is 2f₁Frequency component judges whether stator winding breaks down, using exchange Frequency component 2f₁Amplitude size as circuit isolation system fault signature index.

7. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 3, the extraction three-phase alternating current pressure difference statistical indicator, concrete operation step is as follows：

When motor state of insulation is under health status, three-phase voltage value V is measured₁, when motor state of insulation is in free position When, the three-phase voltage value V that measurement is obtained₂, by V₁And V₂Difference V between the two as motor electrical insulation system state characteristic quantity.

8. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 3, the extraction three-phase alternating current difference statistical indicator is comprised the following steps that：

When motor state of insulation is under health status, three-phase electricity flow valuve I is measured₁, when motor state of insulation is in free position When, the three-phase electricity flow valuve I that measurement is obtained₂, by I₁And I₂Difference I between the two as motor electrical insulation system state characteristic quantity.

9. three-phase asynchronous Ac motor stator insulation on-line monitoring method according to claim 1, it is characterised in that in step In rapid 4, the state index amount that will be obtained carries out state and examines in advance as the training set of SVMs, output motor insulation system Running status examines result in advance, and concrete operation step is as follows：

According to the six class state characteristic indexs for obtaining, six category feature indexs are：MCSA characteristic indexs, current signal statistical model are known Other index, sequence voltage distribution statisticses index, Park transform characteristics index, three-phase alternating current pressure difference statistical indicator and three-phase alternating current Stream difference statistical indicator, sets up training dataset (ξ_s,ζ_s), ξ_sIt is the value of input variable, ζ_sIt is corresponding input variable value, S is instruction Practice collection number；

Using gaussian radial basis function：K(ξ_s,ζ_s)=exp (- | ξ_s-ζ|²/σ² ₀), in formula, K (ξ_s,ζ_s) it is kernel function, σ₀ It is the variance of Gaussian function；

From training dataset (ξ_s,ζ_s) SVMs is trained, the state for setting up SVMs examines model in advance；

Diagnostic sample is treated with the SVMs for training carries out diagnosis output, and the state of output motor insulation system examines mould in advance Type.

10. a kind of three-phase asynchronous Ac motor stator insulation on-Line Monitor Device, it is characterised in that including frequency converter, motor, number According to collection and memory module, Data Management Analysis module, pattern-recognition judge module, control computer, controller and load；

According to the ruuning situation of load, the voltage and frequency of control computer regulation frequency converter control the output of the motor to turn Speed and torque, the Data acquisition and storage module gather the voltage of the controller, current signal and the voltage that will be collected, Current signal transfer shown and processed to the Data Management Analysis module, and carrying out characteristic index to the signal after treatment carries Take, the characteristic index of extraction is transferred into the pattern-recognition judge module carries out pattern-recognition, assesses the motor stator exhausted Edge real-time running state；

Described Data acquisition and storage module is arranged on inside the controller, and Data acquisition and storage module is used for from motor Controller obtain motor real-time voltage signal and current signal, and in sequence or numbering stored.