-6000 -10000 2000 4000 6000 x/mm (c) Fault model:three, score=0.3957 5000 -8000 -5000 0 5000 x

Case 3b: Suitable encoder stuck, fault model = three (time Step 13).(a) Fault model:1, score=0.0955 5000 (b) Fault model:two, score=0.0460y/mm-y/mm--1000 2000 3000 x/mm (c) Fault model:three, score=0.5226-10000 -2000 4000 x/mm (d) Fault model:four, score=0.0148y/mm-y/mm--1000 2000 x/mm-10000 -0 x/E other members of its variety II turn class (e.g. mmFigure 19. Case 4a:Gyroscope stuck, fault model = four (time Step 16).(a) Fault model:1, score=0.0610 6000 4000 y/mm y/mm 2000 0 -2000 -4000 -5000 5000 10000 x/mm (c) Fault model:3, score=0.0285 6000 4000 y/mm y/mm 2000 0 -2000 -4000 -5000 0 5000 x/mm 10000 0 (b) Fault model:2, score=0.0265 6000 4000 2000 0 -2000 -4000 2000 4000 6000 x/mm (d) Fault model:4, score=0.0940 6000 4000 2000 0 -2000 -4000 -5000 0 x/mm 5000Sensors 2014, 14 Figure 22. Case 4b: Gyroscope stuck, fault model = 4 (time Step 17).(a) Fault model:1, score=0.0394 4000 (b) Fault model:2, score=0.0452y/mmy/mm-2000 4000 6000 x/mm (c) Fault model:three, score=0.0450-2000 4000 6000 x/mm (d) Fault model:4, title= j.1399-3046.2011.01563.x score=0.1238y/mmy/mm-2000 -5000 x/mm-2000 -5000 x/mmFigure 23. Robust dead reckoning result for Case 4a (time Step 16).(a)RDR[-0.21,-0.05,0.24], score=0.0950 (b)DR[-0.21,-0.00,0.00], score=0.0610 6000 6000 4000 y/mm y/mm 2000 0 -2000 -4000 -5000 4000 2000 0 --4000 0 5000 -5000 0 5000 10000 x/mm x/mm (c)PSM[1.36,-3.39,0.06], score=0.0922 (D)ICP[-8.26,48.21,0.27], score=0.6516 6000 6000 4000 4000 y/mm 2000 0 -2000 0 5000 x/mm 10000 -4000 -5000 0 x/mmy/mm2000 0 -2000 -4000 -Sensors 2014, 14 Figure 24. Robust dead reckoning result for Case 4b (time Step 17).(a)RDR[0.45,0.11,0.23], score=0.1259 4000 (b)DR[0.46,0.00,0.00], score=0.0394y/mmy/mm-2000 5000 10000 0 2000 4000 6000 x/mm x/mm (c)PSM[-10.59,-10.71,0.29], score=0.1967(D)ICP[18.85,48.32,0.28], score=0.6150 4000 4000-2000 -y/mmy/mm-2000 -5000 x/mm-2000 -5000 x/mm6.five. Diagnosability You can find 3 common situations which might be challenging to diagnose. The initial is that the kinematics of distinct models includes a equivalent functionality. The second will be the that real state has not been modeled inside the technique. The third is that the readings with the external sensors are seriously broken. Figure 25 shows the case of `no faults'; in this case, every single model is right, due to the fact of redundancy title= 2229-5151.84803 amongst the gyroscope and encoders. For that reason, every single model has similar weights. Figure 26 shows the case of `two faults', that's to say, two sensors are faulty simultaneously. Given that none of the models given by the method can govern the circumstance, the weights of all 4 models are insignificant. Figure 25. Diagnosability: No faults.(a) Fault model:1, score=0.5538 4000 (b) Fault mode.

Zuletzt geändert am 28. März 2018 um 21:51