13B 



Having the cluster levelled and aligned to North, the N/S accelerometer 

 detects only accelerations in a N/S direction. Likewise the E/W 

 and vertical accelerometers are pointing E/W and vertical respectively. 

 An acceleration in the N/S direction would be detected by the N/S 

 accelerometer and if this acceleration was integrated, it would 

 produce a velocity in the N/S direction. An integration of this 

 velocity would result in a displacement N/S. Similarly, if acceler- 

 ations occured in the E/W and vertical directions, double integration 

 in both channels would produce displacements E/W and vertical. 



Since it is necessary that the cluster always points North and is 

 aligned to the local vertical, there is a coupling between the N/S 

 accelerometer and the E/W gyroscope. This coupling between the 

 instruments becomes the basis of a feed-back loop and is called a 

 Schuler loop. One property of this loop is that it has a natural 

 period of 84 minutes. Thus any instrument errors, scaling errors, 

 computing errors, misalignment errors, etc., manifest themselves 

 usually as velocity errors whose period is 84 minutes. 



A typical inertial system has errors in position of 1 NM (1.8km) 

 in one hour. The velocity error associated with the position error 

 has a long period. It is this fact which offers a solution to the 

 problem of attaining centimeters accuracy. The solution takes the 

 form of sampling the Schuler loop velocity error frequently. The 

 oscillation can, in short time intervals, be adequately represented 

 by a low order function: a + bt + ct^ + dt^ where, a, b, c and d 

 are constants and t is time. This simple model can then be used to 

 represent the Schuler velocity errors. The accuracy of the model 

 depends on two factors: (a) the frequency of sampling and (b) the 

 accuracy and resolution of velocity error measurement. The first 

 factor is influenced by the mode of operation, and the second factor 

 depends critically on some standard of velocity with which to compare 

 the error velocity. The standard velocity chosen here is zero 

 velocity - stopped. This has a unique quality of being absolute and 

 capable of infinite resolution. In practise, therefore the inertial 

 system has to be stopped every minute say, and the error measured. 

 A model is constructed to represent the error velocity and the net 

 system velocity is inferred from the model velocity and measured 

 velocity. Laboratory experiments were completed with the system 

 stationary on a bench, the error velocity was sampled at various 

 times and the resultant positional error logged. The error is about 

 3cm (1.1 in.) for one minute time intervals, 15cm (5.9 in.) for 

 two minutes and 80cm (32 in.) for three minutes. 



The INS is fully contained in one unit measuring 38cm x 33cm x 30cm 

 (15 in. x 13 in. x 11 in.). The system also has a unit called GENIE 

 (General Electronic Navigation Interface Equipment) . Its purpose is 

 to convert the basic INS into an accurate navigation system suitable 

 for submarines. Fundamentally, it receives velocity information from 

 the INS, and with its own computer performs the processes necessary 

 to model the error velocity, and integrates to produce displacements. 

 Displacements N/S, E/W and vertical are calculated to around 1mm and 



