CN106127184A - A kind of gear case of blower method for diagnosing faults - Google Patents

Abstract

The invention discloses a kind of gear case of blower method for diagnosing faults, multiple test points of gear case of blower are installed vibrating sensor composition multisensor syste；By multiple sensor acquisition to each road vibration data carry out anti-aliasing filter process, anti-aliasing filter uses wavelet threshold denoising method；Use Information Entropy Feature Extraction method based on intrinsic mode function to extract the feature of each road vibration data, each road vibration data characteristic vector T tried to achieve input BP neutral net is completed fault attribute judgement；After the attribute court verdict of each sub-BP network processes, calculate the broad sense of various fault attribute basic confidence distribution under each evidence, it is assumed that BP neutral net is output as A=[a₁,a₂,...a_k], then the broad sense confidence obtained by this evidence is assigned as m=[m₁,m₂,...m_j]；Use DSmT decision-making theory, fault attribute Decision fusion will be carried out by each laissez-passers according to the broad sense confidence distribution obtained thus obtain gear case of blower fault type.

Description

A kind of gear case of blower method for diagnosing faults

Technical field

The invention belongs to technical field of wind power, particularly to a kind of gear case of blower method for diagnosing faults.

Background technology

The many designs of wind energy turbine set, at the mountain area farther out from residential block or sea, are safeguarded relatively difficult, and blower fan stops because of fault simultaneously Wind energy turbine set operator is caused the biggest loss by National Games.And the appearance of fan trouble all can be with a mistake from faint to serious Journey, respectively measure fault occurs in early days faint when sensor acquisition to data may produce conflict, affect to therefore The judgement of barrier type.

Dezert-Smarandache Theory (DSmT) theory is that the scholars such as Dezert and Smarandsche are 2002 A kind of effective evidences conflict combinatorial theory proposed in year, it mainly processes high uncertainty, high conflict and coarse information Source evidence.The multiple combination rule such as PCR1, PCR2, PCR3, PCR4, PCR5 and PCR6 are gradually proposed in development subsequently Then.Wherein, PCR5 is more accurate a kind of conflict distribution method.PCR6 is by Arnaud Martin and Christophe The scholar such as Osswald propose as a kind of fusion rule substituting PCR5 as information source s > 2, it can obtain more preferable than PCR5 Fusion results, so when information source many with two time, general replace PCR5 rule of combination by PCR6 rule.

Summary of the invention

It is an object of the invention to design one based on gear case of blower Incipient Fault Diagnosis system theoretical for DSmT.

The technical scheme is that a kind of gear case of blower method for diagnosing faults comprises the following steps:

Step one, gear case of blower multiple test points on install vibrating sensor composition multisensor syste；

Step 2, by multiple sensor acquisition to each road vibration data carry out anti-aliasing filter process, anti-aliasing filter Use wavelet threshold denoising method；

Step 3, uses Information Entropy Feature Extraction method based on intrinsic mode function to extract the spy of each road vibration data Levy, first filtered data are carried out EMD decomposition, then choose several IMF components front of decomposition, obtain the total of each IMF ENERGY E_i, construct characteristic vector T

$E={\left(\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\left|E_i\right|}^2\right)}^{1/2}$

T=[E₁/E,E₂/E,...,E_n/E]；

Step 4, completes fault attribute judgement by characteristic vector T tried to achieve by each road vibration data input BP neutral net；

Step 5, after the attribute court verdict of each sub-BP network processes, calculates various fault attribute under each evidence The basic confidence of broad sense is distributed, it is assumed that BP neutral net is output as:

A=[a₁,a₂,...a_k]

The broad sense confidence then obtained by this evidence is assigned as:

$m_j=a_j/\left(\underset{j=1}{\overset{k}{\mathrm{\Σ}}}{a}_j\right)$

M=[m₁,m₂,...m_j]；

Step 6, uses DSmT decision-making theory, will be carried out fault attribute certainly by each laissez-passers according to the broad sense confidence distribution obtained Plan merges thus obtains gear case of blower fault type.

When DSmT merges, if evidence source is two, use PCR5 fusion rule；If evidence source more than two, use PCR6 fusion rule.

The present invention extracts from multi-sensor information, EMD Energy Decomposition, and IMF comentropy is extracted, and characteristic vector builds, neural Network training and DSmT Decision fusion thus obtain gear case of blower initial failure type.The present invention draws DSmT decision-making theory Enter in the middle of gear case of blower Incipient Fault Diagnosis system, in conjunction with BP neutral net thus construct gear case of blower initial failure and examine Disconnected model.

Accompanying drawing explanation

Fig. 1 is the system flow chart of the present invention.

Fig. 2 is fault diagnosis model algorithm flow chart in the embodiment of the present invention.

Detailed description of the invention

As in figure 2 it is shown, the gear case of blower method for diagnosing faults step of the present invention is:

1. at the multiple key position of gear case of blower, multiple vibrating sensors composition multisensor syste is installed.

2. by multiple sensor acquisition to data carry out anti-aliasing filter process, anti-aliasing filter uses wavelet threshold to go Make an uproar method.

3. use Information Entropy Feature Extraction method based on intrinsic mode function to extract the feature of each road vibration data.First Filtered data are carried out EMD decomposition, then chooses several IMF components front of decomposition, obtain the gross energy E of each IMF_i, Construct characteristic vector T

$E={\left(\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\left|E_i\right|}^2\right)}^{1/2}$

T=[E₁/E,E₂/E,...,E_n/E]

4. characteristic vector T tried to achieve by each road vibration data input BP neutral net is completed fault attribute judgement；

After the attribute court verdict of the most each sub-BP network processes, calculate the generalized base of various fault attribute under each evidence This confidence is distributed.Assume that BP neutral net is output as:

A=[a₁,a₂,...a_k]

The broad sense confidence then obtained by this evidence is assigned as:

$m_j=a_j/\left(\underset{j=1}{\overset{k}{\mathrm{\Σ}}}{a}_j\right)$

M=[m₁,m₂,...m_j]

6. use DSmT decision-making theory, fault attribute Decision fusion will be carried out by each laissez-passers according to the broad sense confidence distribution obtained Thus obtain gear case of blower fault type.When DSmT merges, if evidence source is two, use PCR5 fusion rule；If card According to source more than two, use PCR6 fusion rule.

Claims (2)

1. a gear case of blower method for diagnosing faults, it is characterised in that comprise the following steps:

Step one, gear case of blower multiple test points on install vibrating sensor composition multisensor syste；

Step 2, by multiple sensor acquisition to each road vibration data carry out anti-aliasing filter process, anti-aliasing filter uses Wavelet threshold denoising method；

Step 3, uses Information Entropy Feature Extraction method based on intrinsic mode function to extract the feature of each road vibration data, first First filtered data are carried out EMD decomposition, then choose several IMF components front of decomposition, obtain the gross energy of each IMF E_i, construct characteristic vector T

$E={\left(\underset{i=1}{\overset{n}{\Σ}}\right|E_i{|}^2)}^{1/2}$

T=[E₁/E,E₂/E,...,E_n/E]；

Step 4, completes fault attribute judgement by characteristic vector T tried to achieve by each road vibration data input BP neutral net；

Step 5, after the attribute court verdict of each sub-BP network processes, calculates the broad sense of various fault attribute under each evidence Basic confidence distribution, it is assumed that BP neutral net is output as:

A=[a₁,a₂,...a_k]

The broad sense confidence then obtained by this evidence is assigned as:

$m_j=a_j/\left(\underset{j=1}{\overset{k}{\Σ}}{a}_j\right)$

M=[m₁,m₂,...m_j]；

Step 6, uses DSmT decision-making theory, melts being carried out fault attribute decision-making by each laissez-passers according to the broad sense confidence distribution obtained Close thus obtain gear case of blower fault type.

2. gear case of blower method for diagnosing faults as claimed in claim 1, it is characterised in that when DSmT merges, if evidence Source is two, uses PCR5 fusion rule；If evidence source more than two, use PCR6 fusion rule.