Time Series/Kalman Filter
- To: mathgroup at smc.vnet.net
- Subject: [mg16910] Time Series/Kalman Filter
- From: asdsa <asdsa at hotmail.com>
- Date: Tue, 6 Apr 1999 01:27:35 -0400
- Organization: None
- Sender: owner-wri-mathgroup at wolfram.com
I don't know if this is a good place to post my request. If not, perhaps somebody can be kind enough to direct me to a more suitable newsgroup. I have two series of measurements (measured over time) which are used to assess the efficiency of a manufacturing process. I measure A at the head of the process (raw input material) and B at the tail of the process (finished output product). It usually takes about 20 minutes for A to become B. I need to know: how much A is needed to make a unit of B at any instant during the process. The ratio B/A works well providing the process is in a steady state, but this is rarely the case. Sometimes the head cannot keep up with the tail, or vice versa. Sometimes the process breaks somewhere in the middle and the ratio B/A gets distorted, especially early in the process (shift). What is the best way to model B/A when A and B are taken at different locations (times) in a complex process. The term "Kalman Filter" has been suggested to me. Any guidance would be greatly appreciated. Please feel free to email me at shawburt at nbnet.nb.ca