Document Sample

(IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 Fault Diagnosis Algorithm for Analog Electronic Circuits based on Node-Frequency Approach S.P. Venu Madhava Rao madhavaraosp@gmail.com Dr. N. Sarat Chandra Babu & Dr. K. Lal Kishore Abstract: In this paper we present a novel approach to In [5] a new method in the construction of fault analog electronic circuits fault diagnosis based on dictionary is proposed where a combination of selection of both nodes and frequency for the first time sensitivity based and information channel based as far as we know. Two fault isolation and localization approaches are used. Also the construction of integer algorithms are presented in this paper. The first coded fault dictionary using Quasi-Hamming distance algorithm selects nodes and frequencies which isolate is proposed in this paper. Heuristic methods using all or desired number of faults. The second algorithm evolutionary computation in combination with the presented converts the fault dictionary contents into Fuzzy logic is presented in [6], the main purpose of binary form. Importantly this helps in the automation such a combination is to generate an optimized of the fault diagnosis process. frequency test set and also ambiguity sets are provided to avoid take care of tolerance effects. An SBT based Keywords: Fault Dictionary, Fault Isolation Table, approach is proposed in [7] where the fault dictionary Binary dictionary, singletons. is constructed using test node voltages and the method used to approximate is Section wise piecewise linear I. Introduction (SPLF) method. A procedure for the selection of test Analog Fault Diagnosis has been of immense frequencies is presented in [8]. This is based on the research interest for the past three decades and evaluation of algebraic indices and the inverse norm of continues to sustain the same zeal even today. The a sensitivity matrix of the circuit under test. In [9], [10] main challenges today in analog fault diagnosis are to and [11], fault diagnosis based on different types of design universally accepted fault models, cost neural networks has been proposed. In [12] knowledge effective, faster and accurate diagnosis of faults. base and fuzzy logic have been used in fault diagnosis. Importantly all this is desired even in the presence of The knowledge base is developed in two ways, one by inherent characteristics of analog circuits like simulations and the second is based on heuristic tolerances, non linearity, in accessible test nodes etc. symptoms observed by the operator. In [13] the ambiguity sets are divided based on the lowest error There are two categories of analog circuit fault probability in the construction of fault dictionaries is diagnosis: Simulation before test (SBT) and proposed. This paper used Monte Carlo techniques for Simulation after test (SAT) [1]. The SBT approach sensitivity analysis. In [14] a fault threshold function involves the generation of fault dictionary by and a fault criterion have been proposed for the fault simulating the circuit and then using pattern diagnosis of circuits with tolerance. An algorithm is recognition to identify the faults. This is the most proposed in [15], which aims to reduce the size of the popular method adopted. In SAT approach sufficient fault dictionary. In [16] and [17] different methods and measurements are needed to identify faulty parameters. algorithms are used to reduce the size of the fault In the SBT approach construction of fault dictionary is dictionaries. In [18] time slot specification based an efficient method. Different test measurements like approach is used in analog fault diagnosis. For this node voltages, current sources, branch currents, built in current sensors and test point insertion is used. frequency measurements etc are used in the A sensitivity based approach using randomized construction of fault dictionaries [2]. There are some algorithms is used to diagnose soft faults in [19].In algorithms developed to find out testable [20] the algorithm proposed tries to find the minimum measurements using numerical approach in [3] and [4]. number of test point for maximum fault isolation. This approach is based on information measure of the test 291 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 points. The diagnosis proposed in this paper [21] is ambiguity sets. Then the original readings are replaced based on global sensitivity analysis method. Also by integer numbers indicative of the ambiguity set to fuzzy logic is used to obtain the sensitivity curves. In which these values belong. [22] an efficient method is applied in the selection of test nodes. This is done by searching for the minimum The test frequency set is represented by f1 to fM, where entropy index based on the available test points. An N is the number of frequencies chosen. efficient graph based method is proposed in [23].This method can be used to select optimum test point The nodes are represented by n1 to nP, where P selection and also can be sued to build DFT. Efficient represents the total number of nodes. Inclusion methods and Exclusion methods are The faults are represented by F0(nominal value) to FN, proposed in [24] to select or de select test nodes, in where N represents the total number of faults. other words the faster selection of optimum test points. A novel multi frequency approach is proposed in [25] Algorithm 1: which drastically reduce the number of test frequencies needed to achieve maximum fault diagnosis. The Step 1: Select the test frequency set (f1 to fM). reduction achieved is better than any known methods. The method proposed in [26] consists of two parts. Step 2: Select the test nodes (n1 to nP) which are One is the creation of fault dictionary consisting of accessible for each frequency. nominal and faulty states of the components and Step 3: Note the actual readings of the circuit for the second is a novel fault detection and localization test frequency set and nodes chosen in steps 1 and 2. algorithm. Step 4: Form the integer coded dictionary using the This paper proposes a novel approach where both test ambiguity sets. node and multi frequency techniques are used. This approach is used to diagnose all the faults or the Step 5: Identify unique integer codes called singletons desired number of faults. for each row i.e. for each of the nodes selected. II Node-Frequency Approach Step 6: Identify the node (nK) which has maximum number of singletons for a frequency fJ., where 1<K≤P In the analog fault diagnosis the prominent methods and 1<J≤M. Select this node-frequency (nK, fJ) pair. If used are multi node or multi frequency measurements. more than one node satisfies this condition, then go to The research so far has been on developing methods to step 9. find out optimum number of test nodes or test frequencies that can identify the desired faults. This in Step 7: If the number of singletons is equal to N+1, some cases leads to more number of measurements then go to step 12. If else go to next step 8. being made thus drastically increasing the size of the dictionary. Step 8: Call Algorithm 2, to form binary dictionary which helps in identifying other nodes from the In this paper we have taken basically nodal analysis remaining (P-1) nodes belonging to the frequency fJ, and then a choice of test frequencies is made based on which can identify different faults. If all faults are [27]. The proposed algorithm selects the nodes and isolated then go to step 12. frequencies which isolate all or desired faults. Step 9: Find the total number of singletons for each In this paper two algorithms are presented. The first test frequency. Then choose the node belonging to the algorithm is for fault isolation and localization. The frequency which has the maximum number of second algorithm converts the integer coded fault singletons. If more than one frequency satisfies this dictionary into a binary dictionary which helps in condition choose any one of the nodes randomly. faster fault isolation. Step 11: If all the faults or desired number of faults are The actual measurements of the CUT are noted down not isolated, then repeat steps from 6 with the next and these values are normalized if necessary. From highest number of singletons. these values we form ambiguity sets. Now we construct another table called integer coded table using Step 12: Stop 292 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 Algorithm 2: Table 1: Actual readings of the imaginary CUT Step 1: Replace all the singletons by the value ‘1’ and Node Nominal Fault-1 Fault-2 Fault-3 others by ‘0’ in the integer coded table, resulting in a binary table. Node-1 1.22 0.33 0.78 0.34 Step 2: If nK is the node chosen, then calculate (nM- Node-2 1.33 0 0.09 2.1 nK), where 1<M≤P, thus forming another table called Node-Wise Fault Isolation table. This results in three Node-3 1.45 0 0.99 2.5 values 0,-1 or 1. The value ‘0’ indicates that the fault has been identified by both nM and nK or both the nodes did not isolate the fault, whereas ‘-1’ indicates As seen from the Table 1 above, we see that for that the fault has been isolated by only nK and ‘1’ is an node -1 measurement, fault-1 and fault -3 have almost indication that the fault has been identified by the node the same value and thus belong to the same ambiguity of interest i.e. nM. Therefore choose the node nM which group. Also these two values are the least among all has maximum number of 1’s. and are assigned values ‘1’. The other values do not Step 3: Check the total number of faults isolated by the belong to any ambiguity group and are assigned values 2 for fault-2 and 3 for nominal, based on the ascending nodes nk and nM. If this sum is equal to P, then Stop, otherwise choose the node which has the next highest range of the values. Using the same procedure for all the remaining nodes, integer coded fault dictionary is number of 1’s. formed and is shown in Table 2. Step 4: Repeat step 3 till the desired fault isolation is Table 2: Integer Coded Fault Dictionary achieved or no further isolation is possible. Step 5: Return to Algorithm 1 Node Nominal Fault-1 Fault-2 Fault-3 III. Integer coded dictionary based on ambiguity sets Node-1 3 1 2 1 The formation of the Integer coded dictionary based on Node-2 2 1 1 3 ambiguity sets is illustrated by an example in this Node-3 3 1 2 4 section. Assume that the actual readings of an imaginary circuit under test are given in Table 1 below. In the Table 2, we see that node-1 has 2 singletons, node-2 has 2 singletons and node 3 has 4 singletons. IV Illustration The circuit used here is a 2nd order Butterworth High Pass Filter as shown in Fig. 1. The circuit has been simulated using Tina Spice software. The faults chosen are taken as 50% increase or decrease in the component values. Thus CUT has been simulated for these faults by changing the component values by ±50%. 293 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 Figure 1: Second Order Butterworth-High Pass Filter R2 R3 3 V1 15 4 2 - 1 6 4 C1 C2 2 3 + OP1 + 7 V2 15 VG1 + R5 10k R1 R4 Using the Step1 from the Algorithm 1, we have that all these nodes are accessible. Using Step3 chosen the test frequency fT= {500Hz, 800 Hz, and 4, the CUT has been simulated and the 1000Hz, 1200Hz, and 1500Hz}. From Step 2, integer coded dictionary as shown in Table 3 is we have chosen four nodes with the assumption formed based on the actual readings. Table 3: Integer coded Dictionary for the HP Filter Frequency=500Hz Nodes/Faults F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 N1 5 7 2 4 7 6 4 8 3 8 1 6 4 N2 2 3 1 2 3 2 2 5 6 4 1 4 1 N3 2 3 1 2 3 2 2 5 6 4 1 4 1 N4 5 6 2 4 9 7 3 10 1 7 1 8 1 Frequency =800 Hz N1 7 6 3 5 11 10 4 12 2 9 1 8 4 N2 6 5 4 5 9 7 5 10 1 7 3 8 2 N3 6 5 4 5 9 7 5 10 1 7 3 8 2 294 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 N4 7 6 5 5 12 10 4 11 1 8 3 9 2 Frequency=1000Hz N1 4 3 5 3 8 6 3 7 1 5 2 5 3 N2 6 5 7 5 9 8 4 10 1 7 3 8 2 N3 6 5 7 5 9 8 4 10 1 7 3 8 2 N4 6 5 7 4 12 10 3 11 1 8 3 9 2 Frequency= 1200 Hz N1 5 3 9 4 10 7 3 8 1 6 2 6 4 N2 6 4 9 5 11 8 4 10 1 6 3 7 2 N3 6 4 9 5 11 8 4 10 1 6 3 7 2 N4 6 5 9 4 11 9 3 10 1 7 4 8 2 Frequency=1500Hz N1 5 2 8 3 7 6 2 6 1 5 3 5 4 N2 4 3 9 3 8 6 3 7 1 4 3 5 2 N3 4 3 9 3 8 6 3 7 1 4 3 5 2 N4 7 5 11 4 11 10 3 9 1 7 6 8 2 The number of singletons for each node for the As seen from the Table 4, node1 and node 4 of whole frequency set is calculated (Step 5) and frequency set f2, and node 4 of frequency set f3 tabulated in Table 4. Here the frequencies are have maximum number of singletons equal to f1=500Hz, f2=800Hz, f3=1000Hz, f4=1200Hz 11, i.e. these nodes can isolate 11 of the total and f5=1500Hz. thirteen faults. We have chosen node 1(or even node 4 can be chosen) of the frequency set f2 i.e. Table 4: Total number of singletons 800Hz as it has maximum number of total singletons (step 9). As the condition mentioned Node/Freq f1 f2 f3 f4 f5 in step 7 is not satisfied, binary table is formed as per step 8. N1 4 11 6 7 4 The binary table is formed replacing Table 3 N2 2 8 7 9 7 contents by either ‘0’ or ‘1’. All the singletons N3 2 8 7 9 7 are replaced by ‘1’and ambiguity sets by ‘0’ (step 1 of Algorithm 2). The binary fault N4 8 11 11 9 9 dictionary is shown in Table 5. Total 16 38 31 34 27 After the execution of the step 2(Algorithm 2), the results are shown in the Node-wise Fault isolation Table 6. 295 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 Table 5: Binary Dictionary Frequency: 800Hz Nodes F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 N1 1 1 1 1 1 1 0 1 1 1 1 1 0 N2 1 0 1 0 1 0 0 1 1 0 1 1 1 N3 1 0 1 0 1 0 0 1 1 0 1 1 1 N4 1 1 0 0 1 1 1 1 1 1 1 1 1 Table 6: Node-Wise Fault Isolation Table Frequency: 800Hz Nodes F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 N1 1 1 1 1 1 1 0 1 1 1 1 1 0 N2-N1 0 -1 0 -1 0 -1 0 0 0 -1 0 0 1 N3- N1 0 -1 0 -1 0 -1 0 0 0 -1 0 0 1 N4- N1 0 0 -1 -1 0 0 1 0 0 0 0 0 1 From the Binary dictionary of Table 5, we can is achieved by a single test frequency of 800Hz see that the faults isolated are F0, F1, F2, F3, F4, F5, and nodes 1 and 4. F7, F8, F9, F10, and F11, where as faults F6 and F12 are not isolated. The faults not isolated is V. Conclusions deduced from the ‘0’ entry in the corresponding columns. As seen from the Table 6, the total In this paper we have presented a novel method number of 1’s is two for (N4-N1), one for (N3- using node-frequency approach in analog fault N1) and (N2-N1). So we choose the (N4-N1) diagnosis. We have presented two algorithms, column, i.e. node 4 is chosen. The faults isolated the first one for choosing the frequencies and by this node 4 are F6 and F12. As seen these are nodes for the desired fault isolation and the the faults which are not isolated by node 1. second is for the generation of binary dictionaries. The effectiveness of these two In the example discussed in this paper we have algorithms was demonstrated using a HP filter been able to achieve 100% fault diagnosis. This circuit. 296 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 References [9] Wei- Qiang Zhang and Chen Xu, “Improved algorithms for circuit fault [1] Tsung-Chih Lin, “Analog circuit fault diagnosis based on wavelet packet and diagnosis under parameter variations based neural network”, International Symposium on Type-2 Fuzzy logic systems”, on Non linear dynamics, Journal of International Journal of Innovative Physics: Conference series 96, pp. 1-7, Computing, Information and Control, Vol. 2008. 6, No. 5, pp. 2137, May 2010. [10] Farzan Aminian Mehran Aminian and [2] Jansuz A. Starzyk, Jing Pang, Stefano H.W.Collins, “Analog fault diagnosis of Manetti and Maria Cristina Piccirilli and actual circuits using Neural networks”, Giulio Fedi, “ Finding ambiguity groups in low testability analog circuits”, IEEE IEEE Transactions on Instrumentation and Transactions on circuits and systems-I: Measurements, Vol. 51, No.3, pp. 544-549, Fundamental theory and applications, vol. June 2002. 47, NO.8, August 2000.pp.1125-1137. [3] G.Luculano,A.Liberatore,S.Manetti and [11] K.Mohammadi, A.R. Mohseni M.Marini, “Multi frequency measurement Monfarad and A.Molaei Nejad, “Fault of testability with application to large diagnosis of analog circuits with tolerances linear analog systems”, IEEE Transactions by using RBF and BP Neural Networks”, on Circuits and Systems, Vol. CAS-23, pp. Student Conference on Research 644-648, June 1986. [4] M.Catelani, G.Luculano, A.Liberatore, Development Proceedings, IEEE, pp. 317- S.Manetti and M.Marini, “Improvements to 321, 2002. numerical testability evaluation”, IEEE Transactions on Instrumentation and [12] Lamiaa Mohamed and A.S. Ibrahim, Measurements, Vol. IM-36, pp. 902-907, “Model based Fault diagnosis via December 1987. parameter estimation using knowledge base [5] Jerzy Rutkowski and Jan Machniewski, “ and Fuzzy logic approach”, IEEE Integer code DC fault dictionary”, ISCAS MELECON, pp. 505-509, May 2002. 2000- IEEE International Symposium on Circuits and Systems, May 28-31.pp [13] Jinyan Cai and M.S.Alam, “An 713-716 algorithm dividing ambiguity sets for [6] P.Jantos, D.Grzechca, T.Golenek and J.Rutkowski, “Heuristic methods to test analog fault dictionary”, IEEE, 2002.pp89- frequencies optimization for analogue 92 circuit diagnosis”, Bulletin of the Polish Academy of Sciences, Technical Sciences, [14] Peng Minfing and He Yigang, “Fault Vol. 56, No. 1, pp. 29-38, 2008. dictionary diagnosis based on branch [7] S.Halgas, “Multiple fault diagnosis of non screening for tolerance circuits”, ICSP’04 linear circuits using the fault dictionary proceedings, pp. 1488-1491. approach”, Bulletin of the Polish Academy of Sciences, Technical Sciences, [15] P.Bernardi, M.Grosso, M.Rabaudengo Vol. 56, No. 1, 2008. and M.Sonza Reorda, “A pattern ordering [8] Franseco Grasso, Antonio Luchetto, algorithm fro reducing the size of fault Stefano Manetti and Maria Cristina dictionaries”, Proceedings of the 24th IEEE Piccirilli, “A method for the automatic selection of test frequencies in VLSI Test Symposium (VTS’06), 2006. analog fault diagnosis”, IEEE Transactions on Instrumentation and [16] David B.Lavo and Tracy Larrabee, Measurement, Vol. 56, No. 6, “Making cause-effect cost effective: Low- December 2007 Resolution fault dictionaries”, ITC 297 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 8, No. 4, July 2010 International Test Conference, IEEE, pp. fault diagnosis of unpowered circuit 278-286, 2001. boards”, IEEE transactions on circuits and [17] Baris Arslan and Alex Orailoglu, “Fault systems-II: Analog and Digital signal dictionary size reduction through test processing, vol.47, No. 10, October 2000. response superposition”, IEEE pp 977-987. International Conference on Computer Design: VLSI in Computers and [24] V.C.Prasad and N.Sarat Chandra Babu, Processors, IEEE, 2002. “Selection of Test nodes for analog fault diagnosis in dictionary approach”, IEEE [18] Shambhu Upadhyaya, Jae Min Lee and transactions on Instrumentation and Padmanabhan Nair, “Tie slot specification measurements, vol. 49, NO.6, pp.1289- 1297, December 2000. based approach to analog fault diagnosis [25] N.Sarat Chandra Babu, V.C.Prasad, S.P. using built-in current sensors and test point Venu Madhava Rao and K.L.Kishore, insertion”, Proceedings of the 11th Asian “Multi-Frequency approach to fault Test symposium (ATS’02), 2002. dictionary of linear analog fault diagnosis”, [19] Cesare Alippi, Marcantonio Catelani, Journal of Circuits, Systems and Ada Fort and Marco Mugnaini, “SBT Fault Computers, Vol. 17, No. 5, pp. 905-928, Diagnosis in analog electronic circuits: A October 2008. sensitivity-based approach by randomized algorithms”, IEEE Transactions on [26] Zbigniew Czaja, “A fault diagnosis Instrumentation and measurement, Vol. 51, algorithm of Analog circuits based on No. 5, pp. 1116-1124, October 2002. Node-Voltage relation”, 12th IMEKO TC1 [20] Kranthi K.Pinjala and Bruce C.Kim, & TC7 Joint Symposium on Man Science “An approach for selection of test points & Measurement, pp. 297-304, 2008. for analog fault diagnosis”, Proceedings of [27] S.Seshu and R.Waxman, “Fault isolation in conventional linear systems- A the 18th IEEE International Symposium on feasibility study”, IEEE Transactions on Defect and Fault Tolerance in VLSI Reliability, R-15, pp. 11-16, 1986. systems (DFT’03), 2003. [21] C.Alippi, M.Catelani, A Fort, M.Mugnaini, “Automatic selection of test frequencies for the diagnosis of soft faults in Analog circuits”, IEEE Instrumentation and Measurement Technology conference, May 2002. pp 1503-1508 [22] Jansuz A. Starzyk, Dong Liu, Zhi-Hong Liu, Dale E. Nelson and Jerzy O. Rutkowski, “Entropy based optimum test points selection for analog fault dictionary techniques”, IEEE Transactions on Instrumentation and Measurement, Vol. 53, No. 3, pp. 754-761, June 2004. [23] Jiun-Lang Huang and Kwang-Ting Cheng, “Test point selection for analog 298 http://sites.google.com/site/ijcsis/ ISSN 1947-5500

DOCUMENT INFO

Shared By:

Categories:

Tags:
IJCSIS, call for paper, journal computer science, research, google scholar, IEEE, Scirus, download, ArXiV, library, information security, internet, peer review, scribd, docstoc, cornell university, archive, Journal of Computing, DOAJ, Open Access

Stats:

views: | 109 |

posted: | 8/13/2010 |

language: | English |

pages: | 8 |

Description:
The International Journal of Computer Science and Information Security is a monthly periodical on research articles in general computer science and information security which provides a distinctive technical perspective on novel technical research work, whether theoretical, applicable, or related to implementation.
Target Audience: IT academics, university IT faculties; and business people concerned with computer science and security; industry IT departments; government departments; the financial industry; the mobile industry and the computing industry.
Coverage includes: security infrastructures, network security: Internet security, content protection, cryptography, steganography and formal methods in information security; multimedia systems, software, information systems, intelligent systems, web services, data mining, wireless communication, networking and technologies, innovation technology and management.
Thanks for your contributions in July 2010 issue and we are grateful to the reviewers for providing valuable comments. IJCSIS July 2010 Issue (Vol. 8, No. 4) has an acceptance rate of 36 %.

OTHER DOCS BY ijcsiseditor

Docstoc is the premier online destination to start and grow small businesses. It hosts the best quality and widest selection of professional documents (over 20 million) and resources including expert videos, articles and productivity tools to make every small business better.

Search or Browse for any specific document or resource you need for your business. Or explore our curated resources for Starting a Business, Growing a Business or for Professional Development.

Feel free to Contact Us with any questions you might have.