Reading integrity monitoring techniques for vision navigation systems - 5 Results

5 result

The algorithm passed 50 Monte Carlo experiments were carried out to verify . Each experiment has a total of 100 Measured values and different number of faults . From the figure 4.2 You know , after 100 Test... Iterations ,$\vec{P}$ The sum of converges to 4.5. After convergence , All five fault measurements can be isolated , There are no false positives and other measurements . It should be noted that , The algorithm does not need to run to complete convergence to isolate faults , But you can assume that every measurement is included in the test , Run with fewer iterations . However , If it doesn't converge , The possibility of well isolated measurements will be reduced , But as a trade-off of the computing time required for additional testing , This is acceptable .

As the fault measurement increases ,$\vec{P}$ The sum of , Pictured 4.3 and 4.4 Shown ,$\sum \vec{P}$ Converge to 8 and 10.8, Respectively 10 and 15 A fault measurement . If more experiments are carried out ,$\sum \vec{P}$ The steady-state value of is relative to the figure 4.5 The number of fault measurements in , You can see a nonlinear relationship . This relationship is similar to the relationship between the probability of passing the test and the number of fault measurements , Pictured 4.1 Shown .

chart 4.1 Suppose there are 100 Measurements , In a given stator set 5 Measurements , The probability of getting a good subset of measurements

 

Pictured 4.6 Shown , When $\sum \vec{P}$ When plotting the steady-state value relative to the probability of selecting a good subset , We found a linear relationship . This relationship depends on the total number of measurements $N$ And the number of measurements in the test subset $n$, But it's still linear . This provides a simple tool to find the number of fault measurements .

chart 4.2 Error probability vector P Average sum and test iteration , Given 50 Run Monte Carlo experiment ,100 Of the measurements 5 One is faulty .

 

chart 4.3 Probability error vector P Average sum and test iteration , Given 50 Run Monte Carlo experiment ,100 Of the measurements 10 One is faulty .

 

chart 4.4 Error probability vector P Average sum and test iteration , Given 50 Run Monte Carlo experiment ,100 Of the measurements 15 One is faulty .

 

chart 4.5 Error probability vector P Steady state average sum and 100 Number of fault measurements out of total measurements .

 

chart 4.6 Error probability vector P The steady-state average sum of And Yes 100 The number of different measurements and faults , The probability of getting a good data subset .

 

6 Conclusion

The work of this paper reveals the error probability vector $\sum \vec{P}$ And the probability of selecting a good subset of measurements . Given the number of measurements and the size of the subset and other variables , Finding a convenient linear relation equation will be studied .

All in all , The visual system has proven to be GPS A good alternative to navigation . However , Due to the change of navigation environment , The quantity and quality of available measurements vary . This leads to a high probability of single failure and multiple failures in any case . If these systems are to be certified for aviation applications , The ability to detect and eliminate these faults is essential . This paper presents a method , This method provides more information in case of multiple faults , It allows to determine the number of faults and identify faults .

7 reference

A little