CN103278772A - Method and device for fault diagnosis of wind generating set based on evidence entropy - Google Patents

Abstract

The invention discloses a method and device for fault diagnosis of a wind generating set based on evidence entropy. The method comprises the following steps: obtaining a basic probability function value of each evidence for each fault mode, calculating and obtaining a weight of each evidence through the evidence entropy and a weight calculation formula according to the basic probability function value of each evidence for each fault mode, conducting weighted mean on the basic probability function value of each evidence through each obtained weight, obtaining a basic probability function value of an average evidence, calculating the overall degree of deviation of an ith evidence and the average evidence, regulating original evidences by the adoption of the obtained overall degree of deviation, fusing the regulated evidences through a fusion formula, and obtaining a final diagnosis result. The method and device reduces conflicting factors of the evidences to a certain extent, and makes the diagnosis result more accurate.

Description

Wind power generating set method for diagnosing faults and device based on the evidence entropy

Technical field

The present invention particularly relates to a kind of wind power generating set method for diagnosing faults and device based on the evidence entropy about a kind of wind power generating set method for diagnosing faults and device.

Background technology

Along with the wind power generating set installed capacity does not increase very much, its ratio in electrical network is also improving year by year, and therefore before fault causes the accident it being carried out fault diagnosis also becomes extremely important.Based on the information fusion method of the evidence theory advantage with aspects such as its expression at uncertain information, tolerance and combinations, in fault diagnosis field, be used widely, be proved to be a kind of effective method.Yet it is independence and mutual exclusion that evidence theory requires the element of its framework of identification, but often it contains bigger conflict, how to realize under the high conflict situations of evidence that effective fusion of multi-source information is a problem that presses for solution.

Summary of the invention

For overcoming the deficiency that above-mentioned prior art exists, the present invention's purpose is to provide a kind of wind power generating set method for diagnosing faults and device based on the evidence entropy, the evidence importance difference that obtains according to a plurality of sensors, the importance parameter that adopts evidence entropy principle to obtain each evidence is weight, merge with the evidence of Dempster rule of combination after to weighting then and obtain final diagnostic result, reduced the conflict factor of evidence to a certain extent, made that the result of diagnosis is more accurate.

For reaching above-mentioned and other purpose, the present invention proposes a kind of wind power generating set method for diagnosing faults based on the evidence entropy, comprises the steps:

Step 1 obtains each bar evidence to the basic probability function value of each fault mode;

Step 2 utilizes evidence entropy and weights computing formula to calculate the weights that obtain each bar evidence according to each bar evidence to the basic probability function value of each fault mode;

Step 3, the weights that utilization obtains are weighted on average the basic probability function value of each bar evidence, obtain the basic probability function value of average evidence;

Step 4 is calculated i bar evidence and the average evidence population deviation degree of obtaining;

Step 5 utilizes the population deviation degree that obtains to adjust original evidence;

Step 6 utilizes fusion formula to merge to the evidence after adjusting, and obtains final diagnostic result.

Further, in step 1, suppose that a total P sign territory produces evidence, for failure domain Q=(Q ₁, Q ₂... Q _NOne total P bar evidence, obtain P * N basic probability function value:

m _i＝(m _1，i，m _2，i...m _N，i) (i＝1，2，…，P)

Wherein, m _iBe that i bar evidence is to the basic probability function of whole fault mode, m _{1, i}, m _{2, i}... m _{N, i}Be i bar evidence to the basic probability function value of each fault mode,

Further, this evidence entropy is:

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i},i=\mathrm{1,2},...P$

Further, this weights computing formula is:

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}$

Further, in step 4, the population deviation degree is calculated as follows:

I bar evidence m _iWith average evidence the proposition { Q _iOn deviation be designated as

M so _iWith

Deviation on whole proposition collection can be expressed as ε _i=(ε _{1, i}, ε _{2, i}. ... ε _{N, i})

${\mathrm{\δ}}_i=\frac{\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left({\mathrm{\ϵ}}_{k,i}\right)}^2}{N}$

δ wherein _iBe evidence m _iWith average evidence

Between the population deviation degree.

Further, this fusion formula is:

$m\left(A\right)=(m_1\⊕m_2)\left(A\right)=\frac{\underset{B\∩C=A}{\mathrm{\Σ}}{m}_1\left(B\right)m_2\left(C\right)}{1-k}$

Wherein

The conflict factor or the contradiction factor for evidence.

For reaching above-mentioned and other purpose, the present invention also provides a kind of wind power generating set trouble-shooter based on the evidence entropy, comprises at least:

The weights computing module obtains each bar evidence to the basic probability function value of each fault mode, and according to the basic probability function value of each bar evidence to each fault mode, utilizes evidence entropy and weights computing formula to calculate the weights that obtain each bar evidence;

The weighted mean module, the weights that utilization obtains are weighted on average the basic probability function value of each bar evidence, obtain the basic probability function value of average evidence;

The deviation acquisition module calculates i bar evidence and the average evidence population deviation degree of obtaining;

Adjusting module utilizes the population deviation degree that obtains to adjust original evidence; And

Fusion Module utilizes fusion formula to merge to the evidence after adjusting, and improves the degree of confidence of fault mode, obtains final diagnostic result.

Further, weights computing module hypothesis one total P sign territory produces evidence, for failure domain Q={Q ₁, Q ₂... Q _NOne total P bar evidence, obtain P * N basic probability function value:

m _i＝(m _1，i，m _2，i...m _N，i) (i＝1，2，…，P)

Wherein, m _iBe that i bar evidence is to the basic probability function of whole fault mode, m _{1, i}, m _{2, i}... m _{N, i}Be i bar evidence to the basic probability function value of each fault mode,

Further, this evidence entropy and weights computing formula are respectively:

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i}$ i＝1，2，…P

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}$

Further, this fusion formula is:

$m\left(A\right)=(m_1\⊕m_2)\left(A\right)=\frac{\underset{B\∩C=A}{\mathrm{\Σ}}{m}_1\left(B\right)m_2\left(C\right)}{1-k}$

Wherein

The conflict factor or the contradiction factor for evidence.

Compared with prior art, a kind of wind power generating set method for diagnosing faults based on the evidence entropy of the present invention, the evidence importance difference that obtains according to a plurality of sensors, the importance parameter that adopts evidence entropy principle to obtain each evidence is weight, merge with the evidence of Dempster rule of combination after to weighting and to obtain final diagnostic result, the present invention can reduce the conflict factor of evidence to a certain extent, makes that the result of diagnosis is more accurate.

Description of drawings

Fig. 1 is the flow chart of steps of a kind of wind power generating set method for diagnosing faults based on the evidence entropy of the present invention;

Fig. 2 is the system architecture diagram of a kind of wind power generating set trouble-shooter based on the evidence entropy of the present invention.

Embodiment

Below by specific instantiation and accompanying drawings embodiments of the present invention, those skilled in the art can understand other advantage of the present invention and effect easily by the content that this instructions discloses.The present invention also can be implemented or be used by other different instantiation, and the every details in this instructions also can be based on different viewpoints and application, carries out various modifications and change under the spirit of the present invention not deviating from.

Before introducing the present invention, earlier concept involved in the present invention, theory and Dempster rule of combination are done one and simply introduce:

Make Θ={ θ ₁, θ ₂... θ _nBe a framework of identification (domain), in fault diagnosis, Θ is limited single fault set of patterns, θ wherein _iRepresent i fault mode.

If (Ω, F P) are a probability space, (Θ, B _Θ) be a measurable space, then defining random set is a collection value mapping: X: Ω → 2 ^Θ

In the formula: 2 ^ΘBe the power set of Θ, φ is the σ territory, B _ΘBe 2 ^ΘOn the σ territory.Elementary probability assignment BPA is a function m: 2 ^Θ→ [0,1], it satisfies: (1) has m (φ)=0 to empty set φ; (2) right The power set 2 here ^ΘThe set of being formed by Θ and all subclass.The core of evidence theory is its composite formula, supposes m ₁And m ₂Be to be defined in 2 ^ΘOn 2 mass functions, its composite formula is (also being defined as the Dempster rule of combination):

$m\left(A\right)=(m_1\⊕m_2)\left(A\right)=\frac{\underset{B\∩C=A}{\mathrm{\Σ}}{m}_1\left(B\right)m_2\left(C\right)}{1-k}---\left(1\right)$

$k=\underset{B\∩C=\mathrm{\φ}}{\mathrm{\Σ}}{m}_1\left(B\right)m_2\left(C\right)---\left(2\right)$

Wherein k is called the conflict factor or the contradiction factor of evidence, has reflected the conflict relationship between the evidence.

Suppose that a total P sign territory produces evidence, for framework of identification (failure domain) Q={Q ₁, Q ₂... Q _NOne total P bar evidence, P * N basic probability function value.

m _i＝(m _1，i，m _2，i...m _N，i) (i＝1，2，…，P)

Satisfy according to the theoretical elementary probability assignment condition of D-S

Evidence for different feasible degree distributes different weight w _i, and weights are carried out normalized, satisfy condition

Here weights can be thought probability, introduce concept of information entropy, are used for determining weights.

Definition (evidence entropy)

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i}$ i＝1，2，...P (3)

Work as H _iWhen smaller, can think that according to information entropy theory this evidence is to { q in the failure domain _iDegree of support bigger, namely favourable to this decision-making, therefore can distribute higher value of this evidence, represent that this evidence is important to judged result, the effect in fusion process is bigger.Here the weighting value is:

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}---\left(4\right)$

After the normalized, can satisfy

The entropy H that from (3) formula, works as certain sign territory as can be seen _iHour, can obtain bigger weights, on the contrary opposite.The characteristics of this formula are make authoritative big evidence obtain bigger weights, and the weak relatively part of evidence to obtain less weights when evidence has bigger conflict.When evidence importance is consistent, the mean allocation weights.

Fig. 1 is the flow chart of steps of a kind of wind power generating set method for diagnosing faults based on the evidence entropy of the present invention.As shown in Figure 1, a kind of wind power generating set method for diagnosing faults based on the evidence entropy of the present invention comprises the steps:

Step 101 supposes that a total P sign territory produces evidence, for framework of identification (failure domain) Q={Q ₁, Q ₂... Q _NOne total P bar evidence, obtain P * N basic probability function value (BPA):.

m _i＝(m _1，i，m _2，i...m _N，i) (i＝1，2，…，P)

Wherein, elementary probability assignment BPA satisfies

Step 102 according to the basic probability function value of each bar evidence to each fault mode, is calculated the weight w that obtains each bar evidence according to evidence entropy and weights computing formula.

The evidence entropy:

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i}$ i＝1，2，…P

Work as H _iWhen smaller, can think that according to information entropy theory this evidence is to { q in the failure domain _iDegree of support bigger, namely favourable to this decision-making, therefore can distribute higher value of this evidence, represent that this evidence is important to judged result, the effect in fusion process is bigger.The weights computing formula is:

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}$

After the normalized, can satisfy

The entropy H that from the formula of evidence entropy, works as certain sign territory as can be seen _iHour, can obtain bigger weights, on the contrary opposite.The characteristics of this formula are make authoritative big evidence obtain bigger weights, and the weak relatively part of evidence to obtain less weights when evidence has bigger conflict.When evidence importance is consistent, the mean allocation weights.

Step 103, the weights that utilization obtains are weighted on average the basic probability function value (BPA) of each bar evidence, obtain the basic probability function value of average evidence:

$\stackrel{\‾}{m}=\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{w}_i{m}_i=(\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{w}_i{m}_{1.i},...\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{w}_i{m}_{N.i}).$

Step 104 is calculated i bar evidence and the average evidence population deviation degree of obtaining.

I bar evidence m _iWith average evidence the proposition { Q _iOn deviation be designated as:

${\mathrm{\ϵ}}_{q,i}=m_{q,i}-\stackrel{\‾}{m_q}$

M so _iWith

Deviation on whole proposition collection can be expressed as:

ε _i＝(ε _1，i，ε _2，i…ε _N，i)

${\mathrm{\δ}}_i=\frac{\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left({\mathrm{\ϵ}}_{k,i}\right)}^2}{N}$

Following formula δ _iIn, δ _iBe evidence m _iWith average evidence

Between the population deviation degree.

Step 105 utilizes the population deviation degree that obtains to adjust original evidence, and the evidence after the adjustment is designated as m ' _i, then

m′ _i＝(m _k，i-ε _k，iδ _i k＝1，2，...N)

Step 106 utilizes the fusion formula of formula (1) to merge to the evidence after adjusting, and improves the degree of confidence of fault mode, obtains final diagnostic result.

Fig. 2 is the system architecture diagram of a kind of wind power generating set trouble-shooter based on the evidence entropy of the present invention.As shown in Figure 2, a kind of wind power generating set trouble-shooter based on the evidence entropy of the present invention comprises: weights computing module 201, weighted mean module 202, deviation acquisition module 203, adjusting module 204 and Fusion Module 205 at least.

Wherein, weights computing module 201 obtains each bar evidence to the basic probability function value (BPA) of each fault mode, and according to the basic probability function value of each bar evidence to each fault mode, utilize evidence entropy and weights computing formula to calculate the weight w that obtains each bar evidence.In preferred embodiment of the present invention, suppose that a total P sign territory produces evidence, for framework of identification (failure domain) Q={Q ₁, Q ₂... Q _NOne total P bar evidence, obtain P * N basic probability function value (BPA):.

m _i＝(m _1，i，m2 _，i...m _N，i) (i＝1，2，…，P)

Wherein, elementary probability assignment BPA satisfies

The evidence entropy:

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i}$ i＝1，2，...P

Work as H _iWhen smaller, can think that according to information entropy theory this evidence is to { q in the failure domain _iDegree of support bigger, namely favourable to this decision-making, therefore can distribute higher value of this evidence, represent that this evidence is important to judged result, the effect in fusion process is bigger.The weights computing formula is:

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}$

After the normalized, can satisfy

The entropy H that from the formula of evidence entropy, works as certain sign territory as can be seen _iHour, can obtain bigger weights, on the contrary opposite.The characteristics of this formula are make authoritative big evidence obtain bigger weights, and the weak relatively part of evidence to obtain less weights when evidence has bigger conflict.When evidence importance is consistent, the mean allocation weights.

Weighted mean module 202, the weights that utilization obtains are weighted on average the basic probability function value (BPA) of each bar evidence, obtain the basic probability function value of average evidence:

$\stackrel{\‾}{m}=\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{w}_i{m}_i=(\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{w}_i{m}_{1.i},...\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{w}_i{m}_{N.i}).$

Deviation acquisition module 203 calculates i bar evidence and the average evidence population deviation degree of obtaining.

I bar evidence m _iWith average evidence the proposition { Q _iOn deviation be designated as:

${\mathrm{\ϵ}}_{q,i}=m_{q,i}-\stackrel{\‾}{m_q}$

M so _iWith

Deviation on whole proposition collection can be expressed as:

ε _i＝(ε _1，i，ε _2，i…ε _N，i)

${\mathrm{\δ}}_i=\frac{\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left({\mathrm{\ϵ}}_{k,i}\right)}^2}{N}$

Following formula δ _iIn, δ _iBe evidence m _iWith average evidence

Between the population deviation degree.

Adjusting module 204 utilizes the population deviation degree that obtains to adjust original evidence, and the evidence after the adjustment is designated as m ' _i, then

m′ _i＝(m _k，i-ε _k，iδ _i k＝1，2，...N)

Fusion Module 205 utilizes the fusion formula of formula (1) to merge to the evidence after adjusting, and improves the degree of confidence of fault mode, obtains final diagnostic result.

In sum, a kind of wind power generating set method for diagnosing faults based on the evidence entropy of the present invention, the evidence importance difference that obtains according to a plurality of sensors, the importance parameter that adopts evidence entropy principle to obtain each evidence is weight, merge with the evidence of Dempster rule of combination after to weighting and to obtain final diagnostic result, the present invention can reduce the conflict factor of evidence to a certain extent, makes that the result of diagnosis is more accurate.

Above-described embodiment is illustrative principle of the present invention and effect thereof only, but not is used for restriction the present invention.Any those skilled in the art all can be under spirit of the present invention and category, and above-described embodiment is modified and changed.Therefore, the scope of the present invention should be listed as claims.

Claims (10)

1. the wind power generating set method for diagnosing faults based on the evidence entropy comprises the steps:

Step 1 obtains each bar evidence to the basic probability function value of each fault mode;

Step 2 utilizes evidence entropy and weights computing formula to calculate the weights that obtain each bar evidence according to each bar evidence to the basic probability function value of each fault mode;

Step 3, the weights that utilization obtains are weighted on average the basic probability function value of each bar evidence, obtain the basic probability function value of average evidence;

Step 4 is calculated i bar evidence and the average evidence population deviation degree of obtaining;

Step 5 utilizes the population deviation degree that obtains to adjust original evidence;

Step 6 utilizes fusion formula to merge to the evidence after adjusting, and obtains final diagnostic result.

2. a kind of wind power generating set method for diagnosing faults based on the evidence entropy as claimed in claim 1 is characterized in that, in step 1, supposes that a total P sign territory produces evidence, for failure domain Q={Q ₁, Q ₂... Q _NOne total P bar evidence, obtain P * N basic probability function value:

m _i＝(m _1，i，m _2，i...m _N，i) (i＝1，2，...，P)

Wherein, m _iBe that i bar evidence is to the basic probability function of whole fault mode, m _{1, i}, m _{2, i}... m _{N, i}Be i bar evidence to the basic probability function value of each fault mode,

3. a kind of wind power generating set method for diagnosing faults based on the evidence entropy as claimed in claim 2 is characterized in that this evidence entropy is:

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i}$ i＝1，2，...P

4. a kind of wind power generating set method for diagnosing faults based on the evidence entropy as claimed in claim 3 is characterized in that this weights computing formula is:

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}$

5. a kind of wind power generating set method for diagnosing faults based on the evidence entropy as claimed in claim 4 is characterized in that in step 4, the population deviation degree is calculated as follows:

I bar evidence m _iWith average evidence the proposition { Q ₁On deviation be designated as

M so _iWith Deviation on whole proposition collection can be expressed as ε _i=(ε _{1, i}, ε _{2, i}ε _{N, i})

${\mathrm{\δ}}_i=\frac{\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left({\mathrm{\ϵ}}_{k,i}\right)}^2}{N}$

δ wherein _iBe evidence m _iWith average evidence

Between the population deviation degree.

6. a kind of wind power generating set method for diagnosing faults based on the evidence entropy as claimed in claim 5 is characterized in that this fusion formula is:

$m\left(A\right)=(m_1\⊕m_2)\left(A\right)=\frac{\underset{B\∩C=A}{\mathrm{\Σ}}{m}_1\left(B\right)m_2\left(C\right)}{1-k}$

Wherein

The conflict factor or the contradiction factor for evidence.

7. wind power generating set trouble-shooter based on the evidence entropy comprises at least:

The weights computing module obtains each bar evidence to the basic probability function value of each fault mode, and according to the basic probability function value of each bar evidence to each fault mode, utilizes evidence entropy and weights computing formula to calculate the weights that obtain each bar evidence;

The weighted mean module, the weights that utilization obtains are weighted on average the basic probability function value of each bar evidence, obtain the basic probability function value of average evidence;

The deviation acquisition module calculates i bar evidence and the average evidence population deviation degree of obtaining;

Adjusting module utilizes the population deviation degree that obtains to adjust original evidence; And

Fusion Module utilizes fusion formula to merge to the evidence after adjusting, and improves the degree of confidence of fault mode, obtains final diagnostic result.

8. a kind of wind power generating set trouble-shooter based on the evidence entropy as claimed in claim 7 is characterized in that, weights computing module hypothesis one total P sign territory produces evidence, for failure domain Q={Q ₁, Q ₂... Q _NOne total P bar evidence, obtain P * N basic probability function value:

m _i＝(m _1，i，m _2，i...m _N，i) (i＝1，2，...，P)

Wherein, m _iBe that i bar evidence is to the basic probability function of whole fault mode, m _{1, i}, m _{2, i}... m _{N, i}Be i bar evidence to the basic probability function value of each fault mode,

9. a kind of wind power generating set trouble-shooter based on the evidence entropy as claimed in claim 8, it is characterized in that: this evidence entropy and weights computing formula are respectively:

$H_i=-\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{m}_{j,i}{\log }_2{m}_{j,i}$ i＝1，2，...P

$w_i=\frac{1/H_i}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}\frac{1}{H_i}}$

10. a kind of wind power generating set trouble-shooter based on the evidence entropy as claimed in claim 9 is characterized in that this fusion formula is:

$m\left(A\right)=(m_1\⊕m_2)\left(A\right)=\frac{\underset{B\∩C=A}{\mathrm{\Σ}}{m}_1\left(B\right)m_2\left(C\right)}{1-k}$

Wherein

The conflict factor or the contradiction factor for evidence.

Images