2004 2 59 :1000-4653 (2004) 02-0001 - 06 NAV IGA TION OF CHINA No. 2 J un. 2004 Serial No. 59, (, 116026) :,,, 100,, : ; ; ; ; :U676. 1 :A Data Mining to Human Factors Based on Ship Coll ision Accident Survey Reports :2004202225 L IU Zheng2jiang, W U Zhao2lin (Dalian Maritime U niversity, Dalian 116026, China) Abstract : The human action of fault in ship collision could be identified roughly by accident statistics. The difficulty re2 search workers face presently is to set up the relationship between the human fault and its influencing factors due to the lack of sufficient data and approaches. In this paper, based on the data collected from 100 accident survey reports, the as2 sociation rules of data mining technique is applied to establish the relationship between the human fault and its influencing factors. The results indicate that this approach is feasible. Key words : Waterway transportation ; Ship collision ; Study ; Data mining ; Human factor ; Association rules, [5 ], [1, 90 % ],, 60 %, 30 %, [2 ],, 1 (Data Mining) 1. 1 [3,4 ],,, : (19592),,,,
2 2004 2 s - count ( A B ) c ( A B ) = P( B ga ) = count ( A ), (2),, 1, 1 [6 ],,,, 2, ( c min ), ( s min ), 1 [7 ], A, ( c) P( B A ) A B, ( s) P( B A ),, 1. 3 1. 2,, 1,, 90 %,, [13 ] : I = { i 1, i 2,, i m }, (item) 1 D, 1 T I 1, T I 1,, TID X 1 I, X T, T X A B, A I, B I, A B = D c % A B, A B D :, c D s % A B, A B D s c A B P ( B A ), s D A B, P ( A B ),, : s ( A B ) = P( A B ) = s - count ( A B ) count ( D) (1), B 1 :, :, :
,;,, :,, : ; 2,, 30,, 2, ARPA [8,9,10,11 ], 12 4,, 2 ( ) :, : 3 3 3. 1, 3. 2, 100, 1993 2002 10,, - [12218 ] 3 : 3. 2. 1 :, 4,,, ; :,, ;,,,,, : 2,
4 2004 2 3 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 - - 2 1 2 3 13 9 7-37 MAIB - - - - - - - 7 5 2 14 TSB 3 4 4 2 1 3 1 - - - 18 A TSB - 1 2 7 3 2 1 2 1-19 USCG 2 1 - - - - - - - - 3 N TSB - - - - 1 2 1 - - 1 5 TAIC - - - 1 - - - - 1 1 3 SHK - - - - - 1 - - - - 1 5 6 8 11 7 11 16 18 14 4 100 :MAIB2 ; TSB2 ;A TSB2 ;USCG2 ; N TSB2 ; TAIC2 ;SHK2 : ; ; ; 3. 2. 2, 4 MAIB TSB A TSB USCG N TSB TAIC SHK 3. 2. 3,, D 100, count ( D) = 100,,,, TID, TID 1, Schroder [19 ] 2,,100 I = { i 1, i 2,, i m },,,, ( ) 12, 30, m = 42 4 4. 1 100 D 1 T I 1, T I A B I, A, B, A T, B T T A B
5 ( s min 20 %, c min 30 %) s min ( %) c min ( %) s min ( %) c min ( %) 56 36 38 75 55 67 36 71 54 66 25 49 39 64 75 49 30 46 25 38 30 36 ( ) (18) 35 29 35 ( ) 37 (19) 27 32 26 31 30 79 21 84 27 71 ( ) (18) 72 24 63 ( ) (19) (26) 22 71 33 75 20 65 34 77 ( ) (19) 60 30 68 ( ) (17) 55 22 50 ( ) (16) 52 ( ) (16) 36 ( ) (15) 94 34 79 ( ) (12) 75 32 74 ( ) (12) 75 31 72 ( ) (9) 56 23 53 ( ) (9) 82 36 80 ( ) (7) 64 33 73 ( ) (6) 55 28 62 ( ) (6) 55 21 47 ( ) (6) 55 ( ) (19) 42 64 75 51 76 60 68 48 72 57 65 44 66 34 39 24 36 31 60 21 31 30 34 21 31 30 34 ( ) (19) (28) 29 32 ( ) (18) (27) 27 31 ( ) (18) (27) : B 2,, B 30, 12 30, A B, (3) (4) : (1), : 5 (2), A A 1, A 2,, A 12, B B 1, s ( A i B j ) = P( A i B j ) = s - count ( A i B j ) count ( D) (3)
6 2004 2 i = 1,2,,12 ; j = 1,2,,30 c ( A i B j ) = P( B j ga i ) = s - count ( A i B j ) count ( A i ) i = 1,2,,12 ; j = 1,2,,30 (3) (4) (4) 9 (3) 4 (1), 10 (2) 3 (1) 8 (4) 2 (3), ( Perfor2 mance Shaping Factors,PSFs) [20 ] 6 (1) 2 (2) 3 (3) (1) 4. 2, 100, 20 % [ 1 ] Department of Transport, U K. The Human element in shipping casualties[ R ]. London : HMSO, 1991. 30 %, 5 [ 2 ] Kuo C. Managing ship safety[ M ]. London : LL P Ref2 5, erence Publishing, 1998. [ 3 ]. [J ].,, 1997, 23 (1) : 41-45., [ 4 ] Curtis R G. Determination of Mariners Reaction Times 50 %, [J ]. Journal of Navigation, 1980, 31 : 409-417. :, [ 5 ] Fagan L. Improving the quality of information in the 20 %, Marine transportation system : An exercise in risk reduc2 30 %, tion [ D ]. Http :/ / students. washington. edu/ lfagan/ complete. html/. 1999. [ 6 ],,. [ M ]. :,2002. 5, [ 7 ]. [J ]., ( ), 2002, 26 2 (6) :758-760. 30, 10 [ 8 ] Moore W H,Bea R G. Management of human errors in, 10, operation of marine systems [ R ]. Berkeley, CA : Uni2 : versity of California at Berkeley, 1993. ; : ; [ 9 ] Landsburg A C, Ferguson S J, Pillsbury C, Ellingstad : ; V. Accident causation2based human factors taxonomy : [J ]. BIMCO BULL ETIN, 1999, 94 (1) : 20-25. [10 ] International Maritime Organization. Amendments to 6 the code for the investigation of marine casualties and in2 5,,, 6 cidents ( Resolution A. 849 ( 20) ) ( Resolution A. 884 (21) ) [ Z]. London : Author, 2000., [11 ] Rothblum A M, et al. Human factors in incident inves2, tigation and analysis[ C]. 2nd international workshop on, human factors in offshore operations ( HFW2002 )., Houston, Texas, 2002., ( 14 )
14 2004 2 1 titude Estimation without Rata Sensors [J ]. Journal of,, Guidance, Control, and Dynamics, 1997, 20 (3) : 522-527. 1, [ 4 ] Y Oshman, F L Markley. Sequential Gyroless Attitude and Attitude2Rate Estimation from Vector Observations 1 ( ) [J ]. Acta Astronautical, 2000, 46(7) : 449-463. [ 5 ] M J Caruso. Application of Magnetoresistive Sensors in ( ) 0. 760 7 0. 478 8 ( ) 0. 404 5 0. 159 9 ( ) 0. 409 4 0. 159 4 ( / s) - 2. 661 8 ( / s) - 1. 517 9 ( / s) - 0. 879 6 5 Navigation Systems[J ]. Sensors and Actuators, 1997, SAE SP21220 : 15-21. [ 6 ],. [ M ]. :, 1995. [ 7 ] M J Caruso. Applications of Magnetic Sensors for Low Cost Compass Systems [ EB/ OL ]. http :/ / www. ssec. honeywell. com/ magnetic/ datasheets/ lowcost. pdf. [ 8 ] S J J ulier, J K Uhlmann, H F Durrant2Whyte. A New Approach for Filtering Nonlinear Systems[ A ]. Proceed2 ings of the 1995 American Control Conference [ C ],, Seattle,WA, 1995, 3 : 1628-1632. [ 9 ] S J J ulier, J K Uhlmann. A New Extension of the U KF Kalman Filter to Nonlinear Systems[ A ]. The Proceed2, ings of AeroSense : The 11th International Symposium on Aerospace/ Defense Sensing, Simulation and Controls 1 [ C], SPIE, Orlando FL, USA, 1997, 182-193.,, [10 ] S J J ulier, J K Uhlmann, H F Durrant2Whyte. A New Method for the Nonlinear Transformation of Means and Covariances in Filters and Estimators [J ]. IEEE Trans2 actions on Automatic Control, 2000, 45 ( 3) : 477-482. [ 1 ] M Challa, G Natanson, C Wheeler. Simultaneous De2 [11 ] Simon Haykin. Kalman Filtering and Neural Networks termination of Spacecraft Attitude and Rates Using only [ M ]. Wiley, 2001. a Magnetometer [ A ]. Proceedings of the AIAA/ AAS [12 ] A Huster, S M Rock. Relative Position Sensing by Fus2 Astrodynamics Specialist Conference [ C ]. San Diego, ing Monocular Vision and Inertial Rate Sensors [ A ]. CA, 1996,7 :AIAA96-3630. Proceedings of ICAR 2003 The 11th International Con2 [ 2 ],,. ference on Advanced Robotics[ C ]. Coimbra, Portugal, [J ]., 2001, 19 (2) : 34-37. 2003, (6,7) :1562-1567. [ 3 ] J L Crassidis, F L Markley. Predictive Filtering for At2 ( 6 ) [12 ]. [17 ] United States Coast Guard. http :/ / www. uscg. mil/ hg/ [ Z]. 1995,1996,1997,1998, 1999, 2000, 2001. [13 ] Marine Accident Investigation Branch, U K. http :/ / www. dft. gov. uk/ reports/ pdf [14 ] The Transportation Safety Board of Canada. http :/ / www. tsb. gc. ca/ en/ reports/ marine/ [15 ] Australian Transport Safety Bureau. http :/ / www. atsb. gov. au/ marine/ incident/ index. [16 ] National Transportation Safety Board of United States. http :/ / www. ntsb. gov/ publication/ g - m/ moa/ collisions. htm [18 ] Transportation Accident Investigation Commission, http :/ / www. taic. org. nz/ marine/ index. html [19 ] Schroder J U, Zade G. The impact of marine casualty investigation on maritime administration and maritime e2 ducation and training[ C]. World Maritime University, Malmo, Sweden, 2002. [20 ] Whalley S. Modeling human errors[ Z]. In : Wegemt, 1993,1 (9) :1223. Lisbon : a European association of u2 niversities in marine technology.