Identify position based on voltage data

[artz]

Hi,

I am using an accelerometer to determine a patient's position on a bed. Five positions need to be identified. In the table below, corresponding accelerometer voltages were measured in each of the five positions.

Orientation	Vx	Vy	Vz
Supine flat	1.666	1.68	1.007
Supine 10˚	1.657	1.88	1.036
Prone     	1.656	1.681	2.34
Left side	1.671	1.009	1.615
Right side	1.687	2.325	1.675

Can anyone in the Forum identify the logic needed where the inputs would be Vx, Vy, and Vz and the ouput would be position?

A nested if statement would seem to be the way to do this or with VBA. Any help is greatly appreciated.

Thanks,

Art

 

[Joe4]

Will all three of the Vx, Vy, and Vz values ALWAYS EXACTLY match one of the 5 sets listed above?
Or is there every any variance that needs to be accounted for?
If there may be variance, how much is allowed?

 

[artz]

Hi Joe,

Thanks for your response. Unfortunately, with real world variances and tolerances would make the answer a no, these values would not be exact match matches every time. The device itself has a specification on repeatability of +/- 2% and an absolute specification of +/- 3%. After I posted, I was sort of thinking that this may more of a kind of fuzzy logic problem more than a simple if x, then y kind of problem.

Is that a scenario that Excel can work with?

Thanks,

Art

 

[LlebKcir]

Hi Joe,

Thanks for your response. Unfortunately, with real world variances and tolerances would make the answer a no, these values would not be exact match matches every time. The device itself has a specification on repeatability of +/- 2% and an absolute specification of +/- 3%. After I posted, I was sort of thinking that this may more of a kind of fuzzy logic problem more than a simple if x, then y kind of problem.

Is that a scenario that Excel can work with?

Thanks,

Art

Will there be a range for each of the axis values? Aslo what is the expected range output? Are you expecting the answer in a matrix format V<x, y, z> or some other format?

 

[Joe4]

I am thinking that we can probably create a User Defined Function via VBA to do this (there may be a way to do it without VBA, but it is beyond my formula writing capabilities).

So, what is your accepted variance level, +/- 2% or +/- 3%?

 

[kweaver]

Using +/- 2%, does this do it?

Excel 2010
	A	B	C	D
1	Orientaton	Vx	Vy	Vz
2	Supine flat	1.666	1.68	1.007
3	Supine 10"	1.657	1.88	1.036
4	Prone	1.656	1.681	2.34
5	Left side	1.671	1.009	1.615
6	Right side	1.687	2.325	1.675
7
8	Input:	1.656	1.89	1.034
9	Position:	Supine 10"
Sheet21

Cell Formulas
Range	Formula
B9	{=INDEX(A2:A6,MATCH(1,(B2:B6<=B8*1.02)*(B2:B6>=B8*0.98)*(C2:C6<=C8*1.02)*(C2:C6>=C8*0.98)*(D2:D6<=D8*1.02)*(D2:D6>=D8*0.98),0),1)}
Press CTRL+SHIFT+ENTER to enter array formulas.

 

[Joe4]

Nice formulaic solution, kweaver!

 

[artz]

Hi kweaver,

Thanks, works fine except one issue that Joe asked about. The data will be recorded in a matrix. It will be sampled at 1Hz (one time per second) over an 8 hour period. That's 28,800 rows of Vx, Vy, and Vz! The time will be variable; as short as an hour or less to a maximum of 8 hours.

Can your formulaic approach handle a variable record length?

Thanks,

-Art

 

[kweaver]

Art, I don't understand the matrix issue. Are you saying that there would possibly be 28,800 rows of input measurements by 3 measurements and you want to analyze EACH of them per the 1 I did?
Please explain.

 

[kweaver]

OR...do each of the V three columns have a range?