Printer Friendly
The Free Library
23,375,127 articles and books


Probabilistic intelligent fault diagnosis in television receiver circuit using visual symptoms.

Introduction

Analyzing system status, identifying possible faults, and arriving at diagnosis decision rapidly in complex system environment are challenging tasks even for skilled and experienced engineers. It is difficult to arrive timely at effective diagnosis conclusion from the great deal of information generally produced by the complex systems in the form of monitored parameters. Thus, fault diagnosis is a challenging area of research. Presently, fault diagnosis lays emphasis on the study of the individual faulty component and the resultant symptoms manifested. There is still no complete and fool-proof generalized methodology of fault diagnosis, i.e. there are obvious shortfalls in knowledge for expressing, collecting and computing the symptomatic data. Consequently, it is difficult to arrive at a rational decision in a timely manner for the fault diagnosis of sophisticated systems. Such a decision requires the knowledge structure for expressing the relations between the causes and effects and also for dealing with complex continuous and discrete variables. The monitoring and fault diagnosis strategies of analog circuits have therefore, moved in recent years from traditional fault finding methodologies to methodologies that utilize Artificial Intelligence (AI) techniques.

Faults means, in general any change in the value of an element with respect to its nominal value which can cause the failure of the whole circuit and system. The faults can be categorized as catastrophic faults (hard faults) or deviation faults (soft faults). Catastrophic faults occur when the faulty element produces either a short circuit or an open circuit. The deviation faults occur when the faulty element deviates from its nominal value without reaching its extreme bounds. Thus, deviation faults are incipient in nature and can be prevented from going into catastrophes, if detected timely.

An intelligent fault diagnosis methodology should effectively assist the system operator in quickly and accurately locating fault. It is based on selective analysis of fault data and use of a priori information about faults. Probabilistic techniques are often used for representation and aggregation of information in Intelligent Fault Diagnosis. Uncertainty is inherent in many fault diagnosis problems because the real world does not operate as a Boolean system. Probabilistic reasoning is one of the techniques employed in artificial intelligence to handle uncertainty. Probabilistic reasoning is used in Intelligent Fault Diagnosis for the incorporation of a priori information, such as past history of system performance, occurrence of chance events and working age of system components along with their shelf life [01].

Intelligent Fault Diagnosis

As already pointed out, fault diagnosis is a challenging area of research in complex system analysis. In the rapidly developing technological world, efficient fault diagnosis is important, because the complexity of systems is continuously growing and product life cycle is reducing. Many heuristic techniques have been developed recently for Intelligent Fault Diagnosis, but the industrial acceptance of these techniques, particularly in cost sensitive areas, has not been high. This is due to high installation cost of Intelligent Fault Diagnosis systems. It is hence, necessary to provide cost effective solutions. Different types of Intelligent Fault Diagnosis systems are hence being developed for different application domains [02].

Although the fundamental concepts of the various Intelligent Fault Diagnosis Systems are generally similar, but several variations of these approaches are used for diagnosing the faults in the various complex industrial systems. They are based on the following emerging technologies: --

1-Rule Based Systems--These are expert systems that implement human like reasoning.

2-Model Based Systems--These systems have a mathematical model of plant, which is used to predict and compare the performance of the actual plant [03].

3-Machine Learning Based Systems--These are heuristic methods based on the following approaches for Fault Diagnosis

A] Case-based Learning Systems

B] Explanation based Learning Systems

C] Learning Inductive Systems [04].

4-Time Series Analysis Based Systems--These techniques use historical data to predict future events.

5-Fuzzy Logic Based Systems--They include fuzzy diagnosis matrix, clustering, evaluation, segmentation and learning.

6-Neural Networks Based Systems--These can effectively utilize on-line knowledge of system operation by self-learning.

7-Multi-source and Multi-sensor Data Fusion Based Systems- They use a combination of numeric, logic and linguistic information.

8-Probabilistic Reasoning Based Systems--They employ Bayesian networks, Fisher's discrimination functions and Dempster--Shafer's evidence theory [05-06].

Intelligent Fault Diagnosis Techniques often employ Model Based Approach. Model Based Fault Diagnosis comprises of three stages:

A] Model Development from historical and operational knowledge.

B] Model Application for fault identification.

C] Model Refinement for adapting to changing situations and providing better performance in future.

Probabilistic reasoning system is often employed in stages A and B to incorporate knowledge acquired from experts into the model.

Proposed Model of Probabilistic Intelligent Fault Diagnosis

Various types of probabilities such as a priori, posteriori, conditional and epistemic probabilities are used for different fault diagnosis applications. The modern theory of probability includes several emerging techniques, such as Bayesian approach, Evidential Reasoning and Dempster-Shafer theory. Probabilistic models have been found to be useful for Intelligent Fault Diagnosis in real world applications, where the observed system data is plagued by various forms of uncertainty and measurement inaccuracy. The generalized block diagram of Probabilistic Intelligent Fault Diagnosis system is shown in figure 1:

[FIGURE 1 OMITTED]

As depicted in figure 1, Probabilistic reasoning is used in the decision-making module for arriving at a rational decision. Probabilistic Intelligent Fault Diagnosis approach makes use of the historical record of system performance in the form of a sequence of sets of system data so as to progressively identify and locate the likely system faults [07, 08]. The system data obtained from complex systems are often random. Thus, the number of sets of such data required to identify a fault location is subjective to the particular application. The data sets used in the diagnosis may be obtained from a system in off-line mode by employing special diagnostic tests. Alternatively, the data sets used in the diagnosis may be obtained from a system in on-line mode during normal operation. Probabilistic Intelligent Fault Diagnosis technique can be used for both types of faults, i.e., the faults, which are continuous in nature and always present or the faults that may be intermittent.

Moreover, probabilistic techniques are particularly useful for fault diagnosis in the cases of multiple faults. The authors of this paper propose to use probabilistic technique for probabilistic fault diagnosis in a problem with multiple faults depicted by multiple observed symptoms [9].

Methodology For Multiple Faults And Multiple Symptoms Probabilistic Intelligent Fault Diagnosis

The application of probabilistic reasoning is expected to be useful for Intelligent Fault Diagnosis in a complex system with multiple faults that are diagnosed using multiple symptoms. The major steps in this approach are as follows:

1] The basic information about the identifiable sub-systems of the complex system is required to be obtained from the data base of system operation and by discussion with field experts. The knowledge and experience of the expert is used to assign the basic probabilities of occurrence of faults in various sections, as depicted by different observed symptoms. This information may be organized in the form of Probabilistic Fault Diagnosis Matrix, which serves as the initial model.

2] The group of faults which may occur in combination and the group of symptoms which may be found simultaneously in the system are identified. This would be employed to formulate the Multiple Faults and Multiple Symptoms problem for the Intelligent Fault Diagnosis System [10].

3] The joint probabilities of Multiple Faults for each symptom can be evaluated as follows:

p(FGi, FGj)=p(FGi) *p(FGj)/Rationalization Factor (1)

Where, Rationalization Factor = Sum of products of probabilities for all possible fault groups

Here, p(FGi) represents the probability of the Ith faults and p(FGi,FGj) represents the joint probability of both faults FGi and FGj and "," represents "AND" [11].

Illustrative Example

The authors of this paper have carried out an application of Probabilistic Intelligent Fault Diagnosis for television receiver system. The motivation for selection of this problem is that the television is commonly used sufficiently complex analog equipment. Identifying faults in a television receiver is quite challenging and needs Intelligent Fault Diagnosis System. It is extremely difficult to find experts for detecting the faults in television sets, particularly in remote areas. It would hence be useful, if Automatic Fault Diagnosis System could be developed for Television Receivers. As a preliminary step towards this, the fault diagnosis for a few symptoms has been presented as illustrative example of Probabilistic Intelligent Fault Diagnosis Methodology. Three observed fault symptoms have been used to identify the faults in three different sections of television receiver i.e. in the Radio Frequency section (R.F.section), Intermediate Frequency section (I.F.section) and Automatic Gain Controller section (AGC section).Thus, a useful approach for solving multiple faults and multiple symptom fault diagnosis problems has been illustrated. The probabilities of fault occurrence for observed symptoms have been considered on the basis of field experience.

The data considered in this example of Probabilistic Intelligent

Fault Diagnosis System for TV Receiver is given in Table 1:

In this system:

S1, S2 and S3: represent Symptoms observed on the faceplate of picture tube.

F1, F2 and F3: represent probable faulty stages in the television circuits.

The Probabilistic fault diagnosis approach can be explained as follows:

Step One:

Consider that the faulty symptom observed on the faceplate of picture tube as S1, i.e., Negative Picture:

Table 1 indicates that maximum probability of fault occurrence is in F1 i.e., R.F.

Stage.

Step Two:

Consider that symptom S2, i.e., Weak Picture is also observed. It is observed from Table 1 that this system suggests that maximum possibility of fault occurrence is in AGC section. Hence, it has become difficult to arrive at rational decision regarding the occurrence of fault in presence of both symptoms S1 and S2. The overall Probabilities of faults have to be evaluated for the combined symptoms {S1, S2}.

The Probability of faulty stage can be found out from first two rows of Table1 as follows:

Probability of occurrence of fault in different stages is calculated by taking the product of individual probabilities for the two symptoms as follows:

p(F1)= (0.5*0.3) = {0.15}

p(F2)= (0.3*0.3) = {0.09}

p(F3)= (0.2*0.4) = {0.08}

Thus, [SIGMA] p(F1+F2+F3) = 0.32

It is now required to rationalize the probability values as follows:

p(F1) = F1/[SIGMA]F =0.15/0.32 = 0.468

p(F2) = F2/[SIGMA]F =0.09/0.32 = 0.281

p(F3) = F3/[SIGMA]F =0.08/0.32 = 0.25

Thus, fault F1 has maximum probability (0.468) and appears to be the most likely fault.

When all three symptoms are considered simultaneously, the following values of probability are obtained for the various faults:

p(F1) = F1/[SIGMA]F = 0.306

p(F2) = F2/[SIGMA]F = 0.367

p(F3) = F3/[SIGMA]F = 0.327

Thus, the probabilistic diagnostic system yields that fault F2, i.e., I.F. Stage, has maximum probability (0.367) and appears to be the most likely fault.

Results

From above example it can be inferred that:

1] If faulty symptom observed on the faceplate of picture tube is Negative Picture i.e.S1, then maximum possibility of occurrence of fault is in F1 i.e. R.F.Stage.

2] If faulty symptom observed on the faceplate of picture tube is Weak Picture i.e. S2, then maximum possibility of occurrence of fault is in F3 i.e. AGC Stage.

3] Probabilistic Diagnosis approach removed the difficulty to arrive at rational decision regarding the occurrence of fault in presence of both S1 and S2. It shows that for faulty symptom S1 and S2, it is more likely that fault is in R.F.Stage than AGC Stage. Also it is important to note that there is more possibility of occurrence of fault in I.F. Stage than in AGC Stage.

4] Finally, when all the three symptoms are considered simultaneously, the probabilistic diagnostic system reveals that the fault F2 i.e. I.F. Stage has maximum probability (0.367) and appears to be the most likely fault.

It is clearly depicted by the illustrative example that the proposed innovative probabilistic fault diagnosis methodology is useful for Intelligent Fault Diagnosis in analog circuit.

Conclusion

Intelligent Fault diagnosis has become an integral part of effective automation systems. This paper hence describes a method for detecting and locating faults in Television Receiver Circuits. Probabilistic Technique for Intelligent Fault Diagnosis has been applied to locate the faulty stage for the number of symptoms in the television receiver. The paper shows how probabilistic theory with different techniques can be used to identify the most probable faulty stage from the available data.

This study suggests that Intelligent Fault Diagnosis should focus on the following aspects:--

1] Time required for fault detection should be as low as possible.

2] Uncertainties in data should be suitably accounted for, in the process of decision making for Fault Diagnosis.

3] The process of Fault Diagnosis should be iterative, so as to allow the initiation of the diagnosis with available data and to modify the results whenever more data becomes available.

Direction for Future Research

Research on Fault Diagnosis in industrial system has yielded a large number of theories and methods. However, the Fault Diagnosis is always a tedious work. Analysis of failure reasons is an essential part of Fault Diagnosis system to promote future improvement. After systematic failures are detected during the testing phase, it is desirable to pinpoint the location of failure as well as the parameter that is causing the failure. The authors are currently engaged in developing Intelligent Fault Diagnosis techniques for this purpose.

References

[1] Z. Wang, M. Sodowska, K.Tasai, J.Rajsia, Analysis and Methodology for Multiple Fault Diagnosis, IEEE Transaction on Computer Aided Design of Integrated Circuits and System, Volume 25, No. 3,pp. 558-575, March 2006.

[2] W. G. Fenton, T.M. McGinnity, L. P. Maguirer, Fault Diagnosis of Electronic Systems using Intelligent Techniques: A Review, IEEE Transaction on Systems, Man and Cybernetics part C, Vol 31, No 03, pp 269-281, Aug 2001

[3] G.F. Luger, W.A Stubblefield, Artificial Intelligence Structures and Strategies for Complex Problem Solving, Addison -Wesley 1998.

[4] L.A.Zadeh, A Simple View of Dempster-Shafer Theory of Evidence and it's Implication for the Rule of Combination, AI Magazine, 7(2), pp 85-90, 1986.

[5] C.Dannt, S.S.Jose, On the Applicability of State of the Art Fault Diagnosis Methodologies to Simple and Complex Systems, The Annals of Dun area De Joss University of Galati, FASCICLE -III, 2005.

[6] Malgorzata Steinder,, Probabilistic Fault Localization in Communication Systems Using Belief Networks, IEEE/ACM Transactions on Networking, Vol 12, No 5, pp 1-14, Oct 2004.

[7] J.W.Bandler, Fault Diagnosis of Analog Circuits, Proceedings of the IEEE, Vol 73, No 08, August 1985.

[8] L. Fang, P.K. Nikolaou, S.Ozev, Parametric Fault Diagnosis for Analog Circuits using a Bayesian Framework, VLSI Test Symposium 2006, Proceeding of 24th IEEE, 30 April -4 May 2006.

[9] Satoh, M.S. Shaikh, Y. Dote, Fault Diagnosis for Dynamical Systems using Soft Computing IEEE International Conference on System, Man and Cybernetic 2001, Volume 3, pp 1448-1452, 7-10 Oct 2001.

[10] J.Vajpai, S.B.Dhoot, Intelligent Fault Diagnosis: State of Art and Application, IEEE sponsored International Conference on Recent Application of Soft Computing in Engineering and Technology, IET Alwar(Raj.), Dec. 22-23 2007.

[11] J.Vajpai, S.B.Dhoot, Intelligent Fault Diagnosis: Future Trend in Automation, National Conference on Futuristic Trends in Engineering and Technology, JCDV Sirsa (Har), Jan 28-29 2008.

Dr. (Mrs.) Jayashri Vajpai (1) and Sachindra Dhoot (2)

(1) Associate Professor, JNVU, Jodhpur, (Rajasthan) India. E-mail: jvajpai@gmail.com

(2) Senior Lecturer, Aurangabad (Maharashtra) India. E-mail: sbd_iste@rediffmail.com

Manuscript submitted to International Journal of Applied Engineering Research on 04 May 2009. Dr. (Mrs.) Jayashri Vajpai is Associate Professor in Electrical Engineering Department, Faculty of Engineering College, Jai Narain Vyas University, Jodhpur, (Rajasthan) India. jvajpai@gmail.com. Sachindra Dhoot is Research Scholar under (QIP Scheme) at JNVU, Jodhpur and Senior Lecturer in Electronics and Communication Engineering of Government Polytechnic, Aurangabad (Maharashtra) India. sbd_iste@rediffmail.com
Table 1: Probabilistic Fault Diagnosis Matrix

Symptoms        Faults    R.F.           I.F.           AGC
[down arrow]    [right    Section (F1)   Section (F2)   Section (F3)
                arrow]

Negative Picture (S1)     0.5            0.3            0.2
Weak Picture (S2)         0.3            0.3            0.4
Snowy Picture (S3)        0.2            0.4            0.4
COPYRIGHT 2009 Research India Publications
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

 Reader Opinion

Title:

Comment:



 

Article Details
Printer friendly Cite/link Email Feedback
Author:Vajpai, Jayashri; Dhoot, Sachindra
Publication:International Journal of Applied Engineering Research
Article Type:Report
Geographic Code:1USA
Date:Aug 1, 2009
Words:2739
Previous Article:Case frame generation for Sanskrit language--a knowledge representation tool.
Next Article:Critical factors to be monitored for successful completion of construction projects.
Topics:



Related Articles
Retinal hemorrhages in 4 patients with dengue fever.
European test symposium (ETS 2005); proceedings.
Study: 44% of patients had visual auras with migraine attacks.
Visual symptoms of migraine underestimated.
How many people would benefit from a smart wheelchair?
Clinical Decision Making VI: diagnosis of visual loss; posterior segment.
Specific learning difficulties and vision.

Terms of use | Copyright © 2014 Farlex, Inc. | Feedback | For webmasters