CN105160170A - Solid state power amplification fault diagnosis method - Google Patents

Abstract

The present invention belongs to the field of microwave solid state power amplifier, relates to a solid state power amplification and fault diagnosis, and provides a solid state power amplification fault diagnosis method based on a fuzzy fault tree and a Bayesian network. The method combines the advantages of both fuzzy fault tree and Bayesian network, converting the T S fuzzy fault tree into the Bayesian network, then performing derivation according to a rule of Bayesian network to obtain an important degree of each basis event, namely a component fault, and finally performing fault dropping according to the magnitude of important degree, so as to achieve fault diagnosis. The solid state power amplification fault diagnosis method provided in the present invention not only fills the blank of solid state power amplication fault diagnosis field, but also is simple in calculation and more practical due to combination of the advantages of both fuzzy fault tree and Bayesian network compared with the traditional fault diagnosis method.

Description

A kind of solid state power amplifier method for diagnosing faults

Technical field

The present invention relates to solid state power amplifier and fault diagnosis, provide a kind of solid state power amplifier method for diagnosing faults based on fuzzy fault tree and Bayesian network.Belong to microwave solid-state power amplifier field.

Background technology

Along with the development of electronic technology, the frequency of operation of observing and controlling, communication, radar develops into millimeter wave frequency band, the power amplifier unit being in transmitting chain end is the nucleus equipment in transmitting chain, has decisive influence to the aspect such as Measure Precision, communication quality, radius of action of system.

Solid state power amplifier is as critical component wherein, and its importance is self-evident, only the normal work of guaranteed solid state power amplifier, could realize the normal operation of whole system.In order to ensure the normal work of solid state power amplifier, just need when solid state power amplifier breaks down, make rapid reply, this just needs to carry out fault diagnosis to solid state power amplifier, and does not almost have about the method for diagnosing faults of solid state power amplifier.Therefore a kind of method carrying out fault diagnosis to solid state power amplifier is needed in the industry.

Summary of the invention

For above-mentioned existing problems or deficiency, the invention provides a kind of solid state power amplifier method for diagnosing faults.The method, based on fuzzy fault tree and Bayesian network, combines both advantages, specifically comprises the steps:

Step 1, build a T-S fuzzy fault tree according to solid state power amplifier theory of constitution;

Step 2, T-S fuzzy fault tree is converted to Bayesian network:

By each event in T-S fuzzy fault tree and the node one_to_one corresponding in Bayesian network, the corresponding root node of bottom event, the corresponding intermediate node of intermediate event, the corresponding leaf node of top event;

Step 3, T-S door rule is converted into the conditional probability that solid state power amplifier fault in Bayesian network occurs:

According to the rule of T-S door

$P(\mathrm{\β}={\mathrm{\β}}^y|{\mathrm{\α}}_1={\mathrm{\α}}_1^{x_1},{\mathrm{\α}}_2={\mathrm{\α}}_2^{x_2},......,{\mathrm{\α}}_n={\mathrm{\α}}_n^{x_n})=P^L\left({\mathrm{\β}}^y\right)$ Formula 1,

Obtain the conditional probability of solid state power amplifier fault under different situations, wherein α _i(0 < i≤n) represents bottom event, and n is the number of bottom event, and β represents intermediate event or top event, x _i(0 < i≤k _i) represent α _ifault degree, k _irepresent α _ifault degree divide number, β ^yrepresent β fault degree, fault degree is divided into 3 classes, represents non-fault, semifault, complete failure respectively with 0,0.5,1.

Step 4, to derive according to the rule of Bayesian network:

By event posterior probability formula

$P\left(\mathrm{\&beta;}={\mathrm{\&beta;}}^y\right|{\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})=\frac{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j},\mathrm{\&beta;}={\mathrm{\&beta;}}^y)}{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})}=\frac{\underset{{\mathrm{\&alpha;}}_1,{\mathrm{\&alpha;}}_2,{\mathrm{\&alpha;}}_3,...,{\mathrm{\&alpha;}}_n}{\mathrm{\&Sigma;}}P({\mathrm{\&alpha;}}_1,{\mathrm{\&alpha;}}_2,...,{\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j},...,{\mathrm{\&alpha;}}_n,\mathrm{\&beta;}={\mathrm{\&beta;}}^y)}{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})}$ Formula 2

Obtain the probability that when specific top event occurs, bottom event occurs.By probabilistic compct formula:

$C_{{\mathrm{\&beta;}}^y}^{PR}\left({\mathrm{\&alpha;}}_j\right)=\frac{1}{k_j-1}\underset{x_j=1}{\overset{k_j}{\&Sigma;}}{C}_{{\mathrm{\&beta;}}^y}^{PR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})$ Formula 3

Obtain bottom event α _ibe β about top event state ^yprobabilistic compct, wherein

$C_{{\mathrm{\&beta;}}^y}^{PR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})=P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y|{\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})-P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y|{\mathrm{\&alpha;}}_j=0)$ Formula 4;

By criticality importance formula:

$C_{{\mathrm{\&beta;}}^y}^{CR}\left({\mathrm{\&alpha;}}_j\right)=\frac{1}{k_j-1}\underset{x_j=1}{\overset{k_j}{\&Sigma;}}{C}_{{\mathrm{\&beta;}}^y}^{CR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})$ Formula 5

Obtain bottom event α _ibe β about top event state ^ycriticality importance, wherein

$C_{{\mathrm{\&beta;}}^y}^{CR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})=\frac{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})C_{{\mathrm{\&beta;}}^y}^{PR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})}{P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y)}$ Formula 6.

Step 5, the criticality importance of trying to achieve according to step 4, sort from big to small by it, when breaking down, carries out malfunction elimination according to this order.

In sum, the method that the present invention proposes calculates simple, is easy to realize, and is a kind of solid state power amplifier method for diagnosing faults efficiently, has filled up the blank of microwave solid-state power amplifier field fault diagnosis.

Accompanying drawing explanation

Fig. 1 is a kind of solid state power amplifier theory of constitution;

Fig. 2 is solid state power amplifier fault tree models;

Fig. 3 is Bayesian network model;

Reference numeral: α ₁-bottom event cable fault, α ₂-fan failure, α ₃-power fail, α ₄-temperature over-range fault, α ₅-electric current transfinites fault, α ₆-forward power transfinites fault, α ₇-reflective power transfinites fault, β ₁-intermediate event final block fault, β-top event solid state power amplifier fault.

Embodiment

Embodiment 1

According to a kind of theory of constitution of solid state power amplifier, as shown in Figure 1, the T-S fuzzy fault tree of solid state power amplifier is built, as shown in Figure 2.The foundation of T-S door rule needs first to divide fault degree, here three classes are divided into, non-fault, semifault, complete failure is represented respectively with 0,0.5,1, as shown in Table 3, and the probability broken down just needs in conjunction with existing data, or carry out a large amount of experiments, and then build T-S door rule, as shown in table one, table two.Wherein α ₁, α ₂, α ₃, α ₄, α ₅, α ₆, α ₇represent that bottom event cable fault, fan failure, power fail, temperature over-range fault, electric current transfinite fault, the reflective power of fault, forward power that transfinite transfinites fault respectively, β ₁represent intermediate event final block fault, β represents top event solid state power amplifier fault.

Then T-S fuzzy fault tree is converted to Bayesian network, as shown in Figure 3.

Derive according to Bayesian network rule again.Tried to achieve the posterior probability of event by formula 2, try to achieve each bottom event criticality importance by formula 5, as shown in Table 4.

Finally by the criticality importance of each bottom event by sorting from big to small, when carrying out fault diagnosis, according to this order carry out malfunction elimination, finally realize fault diagnosis.

The criticality importance of each bottom event of gained is calculated, α according to formula 5 ₅criticality importance maximum, namely electric current transfinites fault having the greatest impact for solid state power amplifier fault, and the reason that is most possibly causing solid state power amplifier to break down is that electric current transfinites fault, so when solid state power amplifier breaks down, preferentially get rid of α ₁electric current transfinites fault, then gets rid of α successively ₂fan failure, α ₆forward power transfinites fault, α ₇reflective power transfinites fault, α ₃power fail, α ₄temperature over-range fault, α ₁cable fault, realizes fault diagnosis.

Table one power amplifier fault T-S door rule (part)

Table two final block fault T-S door rule (part)

The each event of table three probability of malfunction example (k=0.000001) in various degree

The criticality importance of each bottom event of table four

Claims (1)

1. a solid state power amplifier method for diagnosing faults, specifically comprises the steps:

Step 1, according to solid state power amplifier theory of constitution, divide its fault degree, then build its T-S fuzzy fault tree;

Step 2, the T-S fuzzy fault tree built are converted to Bayesian network:

By each event in T-S fuzzy fault tree and the node one_to_one corresponding in Bayesian network, the corresponding root node of bottom event, the corresponding intermediate node of intermediate event, the corresponding leaf node of top event;

Step 3, T-S door rule is converted into the conditional probability that in Bayesian network, solid state power amplifier fault occurs:

According to the rule of T-S door

$P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y|{\mathrm{\&alpha;}}_1={\mathrm{\&alpha;}}_1^{x_1},{\mathrm{\&alpha;}}_2={\mathrm{\&alpha;}}_2^{x_2},......,{\mathrm{\&alpha;}}_n={\mathrm{\&alpha;}}_n^{x_n})=P^L\left({\mathrm{\&beta;}}^y\right)$ Formula 1,

Obtain the conditional probability of solid state power amplifier fault under different situations, wherein α _i(0<i≤n) represents bottom event, and n is the number of bottom event, and β represents intermediate event or top event, x _i(0<i≤k _i) represent α _ifault degree, k _irepresent α _ifault degree divide number, β ^yrepresent β fault degree, fault degree is divided into 3 classes, represents non-fault, semifault, complete failure respectively with 0,0.5,1;

Step 4, to derive according to the rule of Bayesian network:

By event posterior probability formula

$P\left(\mathrm{\&beta;}={\mathrm{\&beta;}}^y\right|{\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})=\frac{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j},\mathrm{\&beta;}={\mathrm{\&beta;}}^y)}{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})}=\frac{\underset{{\mathrm{\&alpha;}}_1,{\mathrm{\&alpha;}}_2,{\mathrm{\&alpha;}}_3,...,{\mathrm{\&alpha;}}_n}{\mathrm{\&Sigma;}}P({\mathrm{\&alpha;}}_1,{\mathrm{\&alpha;}}_2,...,{\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j},...,{\mathrm{\&alpha;}}_n,\mathrm{\&beta;}={\mathrm{\&beta;}}^y)}{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})}$ Formula 2

Obtain the probability that when specific top event occurs, bottom event occurs, by probabilistic compct formula:

$C_{{\mathrm{\&beta;}}^y}^{PR}\left({\mathrm{\&alpha;}}_j\right)=\frac{1}{k_j-1}\underset{x_j=1}{\overset{k_j}{\&Sigma;}}{C}_{{\mathrm{\&beta;}}^y}^{PR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})$ Formula 3

Obtain bottom event α _ibe β about top event state ^yprobabilistic compct, wherein

$C_{{\mathrm{\&beta;}}^y}^{PR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})=P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y|{\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})-P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y|{\mathrm{\&alpha;}}_j=0)$ Formula 4;

By criticality importance formula:

$C_{{\mathrm{\&beta;}}^y}^{CR}\left({\mathrm{\&alpha;}}_j\right)=\frac{1}{k_j-1}\underset{x_j=1}{\overset{k_j}{\&Sigma;}}{C}_{{\mathrm{\&beta;}}^y}^{CR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})$ Formula 5

Obtain bottom event α _ibe β about top event state ^ycriticality importance, wherein

$C_{{\mathrm{\&beta;}}^y}^{CR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})=\frac{P({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})C_{{\mathrm{\&beta;}}^y}^{PR}({\mathrm{\&alpha;}}_j={\mathrm{\&alpha;}}_j^{x_j})}{P(\mathrm{\&beta;}={\mathrm{\&beta;}}^y)}$ Formula 6;

Step 5, the criticality importance of trying to achieve according to step 4, sort from big to small by it, when breaking down, carries out malfunction elimination according to this order.