# Can you guide me about the purpose of fine alignment in inertial navigation systems?

I am new in inertial navigation field. I have some confusion about fine alignment process of INS. I want to ask that what is purpose of fine alignment?

I know by process of alignment we find the initial attitude matrix from body frame to navigation frame (reference frame) which normally take place by two step

1. Coarse alignment
2. Fine alignment

a. what is major difference between both steps?

b. how fine alignment refine the output of coarse alignment? Does Fine alignment minimize the errors to get more accurate output? or Does Fine alignment re correct the initial attitude estimate by coarse alignment?

c. why sensor bias like accelerator and gyro bias and lever arm error are estimated in Fine alignment process.

• Welcome to Av.SE
– Ralph J
Jan 23, 2019 at 14:48
• INS alignment is a convergent/asymptotical process with no completion. The more you wait, the higher the precision, see Why does IRS alignment take so much time?
– mins
Jan 25, 2019 at 21:57
• @mins Actually it is not asymptotic. As you integrate for longer periods, errors in the integration process add up and cause drift, that is, random walks rarely end up at the starting point. Same effect as clocks: the longer you wait after setting a clock, the more wrong it gets. The measure is known as Allan Variance, see section 5, for example. Feb 23, 2019 at 3:46
• @user71659: Seems convergent.
– mins
Feb 23, 2019 at 13:00
• @mins Look at fig 11 in my linked document. At 1000+ seconds for that model, error starts to trend upwards. It's intuitive when you think about it, the longer you run an IRS, the more it drifts. Feb 23, 2019 at 23:53

I was a Doppler Radar & INS specialist for the Air Force in the 80s, and learned systems for the F4 (flying brick) and the triple carousel system for the C-5A. These systems used mechanical gyros in combination with accelerometers to establish a "Stable Platform". The carousel idea was state of the art then and rotated 1 revolution per hour to greatly reduce drift.

BATH was used for course alignment, which is an acronym for Best Available True Heading and local gravity to start level. It would then spin up and do fine alignment by sensing Earth rotation to determine True North. Typically this took about 30 minutes. Modern systems us laser gyros which aren't affected by mechanical drift.

• This answer does not seem to address the questions. It is tangential or informational at best. Oct 21, 2019 at 16:18

Coarse alignment is using accelerometers to work out an approximate level by sensing earths gravity by the "non vertical accleromnmeter. BUt that "vertical" is influenced by accelerometer error, earth rotation and local effects, the masses of Hills tides etc do change the local vector, so having gained an approximate vertical from the accelerometers, the true vertical and true north can then be measured by earth rotation and the longe rtyhe fine alignment the better,

In some systems, which know they are stationary, the coarse alignment can be used to calculate gyrodrift, since there should be no rate on the Pitch and roll axes, and heading only earth rate, (for which you need to know latitude..) (Function is sin of latitude)

BUT GPS makes this so much easier (esp when moving) since both position is known and velocities. Stored heading if still valid (ie Ac not moved) helps, but GPS track on the ground = heading, and two GPS antennae spaced apart along the fuselage can also give heading when not moving.

With GPS since its acts as a low drift position system, you can use the data to reduce errors, correct offsets and drifts, but with long time constants ie effectively is "fine alignment" in flight..,.as long as the AC is not manoeuvring.

For strap down systems, all the above applies to the resolved axis in "earth units" and not the physical axis of the sensors,