8-bit MicrocontrollersのトピックRe: elevator displacement

elevator displacement

dimor — Mon, 24 Jan 2011 20:14:08 GMT

hey,

i've been given a task to measure elevator displacement with an accelerometer (Olimex mod-mma7260 - mma7260+at91sam7s).

i've been reading this application note http://www.freescale.com/files/sensors/doc/app_note/AN3397.pdf regarding implementing positioning algorithms using low end accelerometers. it's been stated several times that these algorithms are useful only for situations where the displacement precision is not extremely critical.

what's extremely critical? i want to calculate the displacement to figure out the current floor while the elevator is moving, is it possible to do it with the mma7260 and these algorithms?

thanks

Re: elevator displacement

bigmac — Tue, 25 Jan 2011 02:19:10 GMT

Hello, and welcome to the forum.

I don't think that this concept will work. Let's assume that you have a building with say, 50 floors. You would therefore require a distance resolution of better than 2 percent for a journey from bottom to top, or vice versa. My gut feels that this requirement is at least an order of magnitude better than might be achieved with an accelerometer.

The problem with deriving an absolute position using acceleration measurements is that a double integration process is required, firstly to derive the velocity from the many acceleration samples, and then to derive the position from the velocity characteristic. This is subject to many cumulative errors. You would also need to start from a known reference position, and frequently recalibrate whenever located at this position.

A further consideration for this application is that only z-axis readings are relevant, but this axis will be subject to an offset of 1g, which may reduce measurement resolution. The delta-g about this offset will represent the acceleration of the lift/elevator. I presume that mid-journey, there will be constant velocity, resulting in zero delta-g. Therefore position must be estimated from the derived velocity and elapsed time. Maybe the velocity error could be in the region of 10 percent, but I am guessing. However, since there is a tendency for the velocity to become zero after a deceleration, this does allow for the velocity to be recalibrated.

Regards,

Mac

Re: elevator displacement

dimor — Tue, 08 Feb 2011 03:17:08 GMT

hey,

first of all - thanks for replying.

first i'm trying it on a 4 floor building.

i've managed to get everything done. i calculate the displacement correctly and pretty accurately (but then again, it's only 4 floors), but i do have one problem - the distance measured upwards isn't equal to the distance measured downwards.

a distance of one floor journey is 3 meters when going up, but 2.7 meters when going down.

i must say all the measures are fairly constant.

i'm pretty sure it's gravity related, just not sure what. anyone got an idea?