Download - Academy Publisher

More documents

Recommendations

Info

ISBN 978-952-5726-09-1 (Print) Proceedings of the Second International Symposium on Networking and Network Security (ISNNS ’10) Jinggangshan, P. R. China, 2-4, April. 2010, pp. 116-119 Study on the Failure Identifiability of Disaster Detection System Xiaoyu Wang 1 , and Kaiquan Wang 2 1 Changzhou Iinstitute of Litht Industry Technology,Changzhou 213164, China Email: wxy730@163.com 2 Jiangsu Polytechnic University,Changzhou 213164, China Email: wkq@jpu.edu.cn Abstract—There is a serious potential risk in the petrochemical production process, the Disaster Detection System(DDS) is of key equipment to monitor the production status, prevent and control the potential risk. However, the Initiating failure and Inhibiting failure would occur in the period of the Disaster Detection System(DDS) working. In order to identify the two failures by DDS itself, the paper proposed a mathematic method for finding the fault test codes based on the theory of Bayle derivation, and verified it on the FCC (Fluidized Catalytic Cracking Unit) of a petrochemical company. The results evidence that according to the property of the Boolean difference, the codes are available to fined DDS failures; in order to identification of DDS failure it is necessary to properly design the combinational logic unit for ensure it has the fault test codes; if the DDS failure that had been judged to be identification, the fault test codes should always be input to DDS timely to implement self fault test so that its failure may be able to find. Through such means the DDS functions should be achieved correctly. Index Terms—Disaster Detection System; Initiating failure; Inhibiting failure; Identification; Fluidized Catalytic Cracking Unit I. INTRODUCTION There is a serious potential risk in the petrochemical production process, the Disaster Detection System(DDS) is of key equipment to monitor the production status, prevent and control the potential risk. However, the initiating failure and inhibiting failure would occur in the period of DDS’ working. The former because of "undetected", mistakenly believe that a critical condition is safety so that neither recognize the dangers nor prevention the accidents in time; the latter because of "false detection", in contrast consider the safety state as dangerous, forced interruption of production, caused the cut-off losses. [1] In order to achieve the overall security of the production process, DDS must have self-testing capabilities to identify the two types of fault. The core component of DDS in petrochemical production (such as Disrubution Control System-DCS, Emergency Shutdown System-ESD) is a combination logic unit, so the problem must be selution in study of DDS failure identifiability is that whether the information carrying two types of fault can be correctly detected in the output of the unit. Specifically, for a tertain combination © 2010 ACADEMY PUBLISHER AP-PROC-CS-10CN006 116 logic unit, can we found out some representatives test code while these code be input to the unit then the failure signal may be detected in the outpu. If such a test code [2] [3] exists, the DDS failures can be identification. The approach of seeking the relationship of code between input and output in combinational logic unit was proposed firstly by prf. Akers. [4] Based on approach and Boolean difference operation, this paper will study the identificability of DDS’ failure. II. PRINCIPLE OF DDS FAULT IDENTIFICATION [5] A. The properties of Boolean difference Assume that there is a combinational logic unit, and its initial input Boolean variable is recorded as x 1 , x2 ,...... xn , output Boolean function denoted by f ( x1, x2 ,...... xn ) , then df( X) = f( x1, x2,... xi −1, xi , xi + 1,... xn ) ⊕ f ( x1, x2,... xi −1, xi , xi + 1,... xn ) dxi (1) known as Boolean difference f (X ) to x i . Boolean difference has two important properties. Property 1: If f ( x1, x2,... xi− 1,0, xi+ 1,..., xn ) = f ( x1, x2,... xi− 1,1, xi+ 1,..., xn) then df ( X ) = 0 dx i There is no effect on the output while x i change, so that the failure at x i is unpredictable. Property 2: If f ( x1, x2,... xi− 1,0, xi+ 1,..., xn) ≠ f ( x1, x2,... xi −1,1, xi+ 1,..., xn) t hen df ( X ) = 1 dx i There is effect on the output while x i change, so that the failure at x i is predictable. B. Calculation of failure test code 1) Logical unit failure mode
Combinational logic unit failures in two ways: Failure Mode 1 - Fixed 1 fault ( s − a −1) . While the logic unit has AND gate and Boolean algebraic expression is Z = x x (2) 1 2x3 Then any one of element (such as x 1 ) take 1, Z will not change, that is, expression (2) can not find the failure mode 1 of x 1 , so that x 1 is Fixed 1 fault ( s − a −1) of Boolean algebraic expression Z . Similarly, x 2 , x 3 is also Fixed 1 fault ( s − a −1) . Failure Mode 2 - Fixed 0 fault ( s − a − 0) . While the logic unit has OR gate and Boolean algebraic expression is Z x + x + = (3) 1 2 x3 x ) take 0, Z will Then any one of element (such as 1 not change, that is, expression (3) can not find the failure mode 2 of x 1 , so that x 1 is Fixed 0 fault ( s − a − 0) of Boolean algebraic expression Z . Similarly, x 2 , x 3 is also Fixed 0 fault ( s − a − 0) of Boolean algebraic expression Z . 2) Calculation of single failure test code If a , a ,..., a ) ( 1 2 n T is the test code of Fixed 0 fault in measured channel X ( x1, x2,..., xn) , then the formula established: df ( X ) X ( x1, x2,..., xn) × T ( a , ,..., ) = 1 1 a2 a (4) n dx Where a = 0 or 1. i ( a1, a2,..., an i If T ) is the test code of Fixed 1 X ( x , ,..., ) fault in measured channel 1 x2 xn , then the formula established: df ( X ) X ( x1, x2,..., xn) × T ( a , ,..., ) = 1 1 a2 a (5) n dx Where a = 0 or 1. i i Application the Property 1 and Property 2 of Boolean difference, it would be appraised whether the Fixed 1 fault ( s − a −1) and Fixed 0 fault ( s − a − 0) may be measured on the channel X x , x ,..., x ) . ( 1 2 n Applications formula (4) and (5) , the single failure test code of the Fixed 1 fault ( s − a −1) and Fixed 0 fault ( s − a − 0) on the unit X ( x1, x2,..., xn) would be calculated. For example, added the single test code T ( a i ) to the input of the measured unit, if the unit output Z = 1, then there is no failure of ( s − a −1) in X x , x ,..., x ) , otherwise there is. ( 1 2 n Similarly, the multi-fault test code T ( a1, a2 ,..., an ) can be calculated, not repeat them due to space limitations. III. EXAMPLE There is a petrochemical company using ESD as DDS in its FCCU-II sets. [6][7] The main logic diagram of the combination logic unit shown in Fig.1 .[8] Let's examine the identification of the DDS’ failure. A.Failure identifiability of combination logic unit There are three kinds of basic logic unit in fig.1 (Fig. 2). [9][10] For fig. 2 (a), because of f ( X ) = x1x2 + x1x3 + x2x3 , then df ( X) = ( x1x 2 + x1x 3 + x2x3 ) ⊕( x1x 2 + x1x 3 + x2x3 ) dx 1 = x + 2 + x3 x2 x3 So that the fault test expressions: df ( X ) x 1 = x1 ( x2 dx1 + x3 + x2 x3 ) (6-1) df ( X ) x 1 = x1 ( x2 dx1 + x3 + x2 x3 ) (6-2) By symmetry, df ( X ) x 2 = x2 ( x1 + x3 + x1x3 ) dx (7-1) 1 df ( X ) x 2 dx1 x2 ( x1 + x3 + x2 x3 ) df ( X ) x 3 dx1 x3 ( x1 + x2 + x1x2 ) df ( X ) x 3 x3 ( x1 + x2 + x1x2 ) dx 1 = (7-1) = (8-1) = (8-1) Similarly, for fig. 2 (b) and fig. 2 (c), may also find thier fault test expressions. Choosed different value of x i ( x i = 1 or 0) and fixed others in these pairs of expressions (for example in expression (6-1)and (6-2)), if the value of these pairs expression do chuange, then x i is the single fault test code, and this set of variables is called combinational fault test code. Tab.1 lised out all of the combinational fault test code of the logic unit of fig.2 (a), fig.2 (b) and fig.2 (c). B.Faults identification of combination of logic unit Based on the analysis above, the fault of the DDS can be identification as follow. In Figure 1, x + x A = ( x1 x2 + x1x3 + x2x3) x4 + xB = x26 x4 x + x + x B = ( x5 + x6 + x7x8 ) x9 + xC + xD = x27x9 x C = ( x x + x = x x + x 10 x11 + x10x12 + x11x12) 13 24 28 13 C B 24 D 117
Page 1 and 2:
Proceedings The Second Internationa
Page 3 and 4:
Table of Contents Message from the
Page 5 and 6:
Zuming Xiao, Zhan Guo, Bin Tan, and
Page 7 and 8:
Message from the Symposium Chairs T
Page 9 and 10:
Second International Symposium on N
Page 11 and 12:
ISBN 978-952-5726-09-1 (Print) Proc
Page 13 and 14:
model that can deal with time serie
Page 15 and 16:
ISBN 978-952-5726-09-1 (Print) Proc
Page 17 and 18:
training for the Wushu competition
Page 19 and 20:
ISBN 978-952-5726-09-1 (Print) Proc
Page 21 and 22:
information, called weak uncertain
Page 23 and 24:
Student side Student side Student s
Page 25 and 26:
ISBN 978-952-5726-09-1 (Print) Proc
Page 27 and 28:
If we define element of student as
Page 29 and 30:
ISBN 978-952-5726-09-1 (Print) Proc
Page 31 and 32:
QoS, each frame data is divided int
Page 33 and 34:
ISBN 978-952-5726-09-1 (Print) Proc
Page 35 and 36:
for Imaging Two and Three Phase Flo
Page 37 and 38:
ISBN 978-952-5726-09-1 (Print) Proc
Page 39 and 40:
A. Profiling&following control algo
Page 41 and 42:
Research and Realization about Conv
Page 43 and 44:
PDF document structure is a tree st
Page 45 and 46:
ISBN 978-952-5726-09-1 (Print) Proc
Page 47 and 48:
support 10 Mb / s. But ENC28J60 onl
Page 49 and 50:
ISBN 978-952-5726-09-1 (Print) Proc
Page 51 and 52:
condition the first byte output of
Page 53 and 54:
ISBN 978-952-5726-09-1 (Print) Proc
Page 55 and 56:
According to maximum membership deg
Page 57 and 58:
ISBN 978-952-5726-09-1 (Print) Proc
Page 59 and 60:
ubber according to the mass ratio o
Page 61 and 62:
Ⅲ. AN IMPROVED DNA ALGORITHM FOR
Page 63 and 64:
Ⅴ.CONCLUSION REMARKS DNA computer
Page 65 and 66:
its first child q 1 on the left, th
Page 67 and 68:
chains can greatly improve efficien
Page 69 and 70:
II. RELATED WORK A. Mobile Service
Page 71 and 72:
special services. SOAP is used to b
Page 73 and 74:
detection methods of DDoS attacks m
Page 75 and 76: Step4 calculate the new subordinate
Page 77 and 78: Figure 1. An analysis of the partit
Page 79 and 80: ISBN 978-952-5726-09-1 (Print) Proc
Page 81 and 82: The preceding three formulas can be
Page 85 and 86: And the decay speed of buffer seque
Page 89 and 90: distance of view point. Given that
Page 93 and 94: Ⅳ. EVALUATION OF BLENDED LEARNING
Page 97 and 98: symmetric with respect to the origi
Page 101 and 102: each sample belongs to each categor
Page 105 and 106: A. Data Preparing To generate our t
Page 109 and 110: oth α and β . determination of Qu
Page 113 and 114: Strong earthquake 0.1< M L
Page 117 and 118: indirect causes, and the logical re
Page 119 and 120: egression. Granger causality test i
Page 121 and 122: B. Evaluation We have evaluated thi
Page 123 and 124: used as a source and neighboring ce
Page 125: Figure 3. Drainage networks generat
Page 129 and 130: Tabal.1 Combinational fault logic t
Page 131 and 132: under the endorsement of both the m
Page 133 and 134: TABLE I. DESCRIPTION OF PROPOSITION
Page 135 and 136: Figure 2. The process of the invers
Page 137 and 138: Figure 9. (a)the original image.(b)
Page 141 and 142: In order to reduce to the number of
Page 145 and 146: that studying being going to be to
Page 147 and 148: Reference[10] analyzed the evolutio
Page 149 and 150: module, communication module, apper
Page 151 and 152: After the comprehensive performance
Page 153 and 154: system testing can be seen that the
Page 155 and 156: III. AUTONOMIC RESOURCE ALLOCATION
Page 157 and 158: The QoE i (T i ) is the ith user’
Page 161 and 162: nodes will be formed one cluster, t
Page 165 and 166: Where n is the number of data point
Page 169 and 170: Keyboard event Application User mod
Page 173 and 174: SQL Azure will eventually include a
Page 177 and 178:
The nodes in the suffix tree are dr
Page 179 and 180:
[3] Y. Li, S. M. Chung, and J. D. H
Page 181 and 182:
some drilling fluid produces hydrog
Page 183 and 184:
"normalization", whose membership b
Page 185 and 186:
and minimum structural elements in
Page 187 and 188:
esults of the spatial transform par
Page 189 and 190:
B. Research on protocol actions Com
Page 191 and 192:
ACKNOWLEDGMENT This work is funded
Page 193 and 194:
A. Problem Description In a small b
Page 195 and 196:
[4] Huang, Hung, and J. Y. jen Hsu.
Page 197 and 198:
Further by calculating, the followi
Page 199 and 200:
TABLE II. THE CONCENTRATION BETWEEN
Page 201 and 202:
private key that obtained by using
Page 203 and 204:
ISBN 978-952-5726-09-1 (Print) Proc
Page 205 and 206:
ontology, and the domain dictionary
Page 207 and 208:
ISBN 978-952-5726-09-1 (Print) Proc
Page 209 and 210:
Eq.4 is NP-hard and can be solved b
Page 211 and 212:
ISBN 978-952-5726-09-1 (Print) Proc
Page 213 and 214:
Where: G i is the selection field o
Page 215 and 216:
If there is X j in a generation, wh
Page 217 and 218:
Theorem 2.4([10]). Let L 1 and L 2
Page 219 and 220:
The relation between the lattice im
Page 221 and 222:
Figure 2. Example of single-step de
Page 223 and 224:
evocation and the fourth group stor
Page 225 and 226:
focused mainly on rule-based forms
Page 227 and 228:
Ⅵ. CONCLUSION Figure 6. The Flask
Page 229 and 230:
EDCF in comparison with DCF, has so
Page 231 and 232:
B. Simulation results and analysis
Page 233 and 234:
ISBN 978-952-5726-09-1 (Print) Proc
Page 235 and 236:
in routing table. When search resou
Page 237 and 238:
ISBN 978-952-5726-09-1 (Print) Proc
Page 239 and 240:
others through the selective emotio
Page 241 and 242:
such as weather information, techno
Page 243 and 244:
the opening window, a related page
Page 245 and 246:
1) Process context storage areas. I
Page 247 and 248:
V. CONCLUSION In this thesis, sCPU-
Page 249 and 250:
main types of horizontal search env
Page 251 and 252:
Enterprise Portal security provides
Page 253 and 254:
() t = [ S () t S () t ] T N S ,...
Page 255 and 256:
[2] Kumaravel, N., and Kavitha, V.,
Page 257 and 258:
output, so BP network has been wide
Page 259 and 260:
input to the artificial neural netw
Page 261 and 262:
(2) In a process (tokens from outsi
Page 263 and 264:
The reduction process consists of t
Page 265 and 266:
all eight normal vectors are identi
Page 267 and 268:
is B object , and the number of emp
Page 269 and 270:
xi, j y' = εα ( (1 + tanh( )) −
Page 271 and 272:
Error 8000 7000 6000 5000 4000 3000
Page 273 and 274:
Architecture (AMBA) a new bus archi
Page 275 and 276:
In order to ensure the smooth proce
Page 277 and 278:
ISBN 978-952-5726-09-1 (Print) Proc
Page 279 and 280:
In this paper, we adopt two Sobel o
Page 281 and 282:
ISBN 978-952-5726-09-1 (Print) Proc
Page 283 and 284:
four layers.The far right of the gr
Page 285 and 286:
ISBN 978-952-5726-09-1 (Print) Proc
Page 287 and 288:
TABLE II. OPERATIONAL EMPLOYEE TABL
Page 289 and 290:
A. Object-relational Type To audit
Page 291 and 292:
to execution time. Also, DBA would
Page 293 and 294:
Liping Chen .......................
show all

Download - Academy Publisher

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?