|
SUBOPTIMAL ALGORITHM FOR BOX-JENKINS MODEL PARAMETER ESTIMATION
Nenad Dodić
The Box-Jenkins model is suitable to describe a large class of linear systems of which measuring channels are corrupted by noise of particular spectral properties. The paper discusses the Box-Jenkins time-discrete model parameter estimation and suggests a suboptimal real-time algorithm that gives approximately minimum varinace unbiased estimates of model parameters. The form of the algorithm is suitable for direct computer implementation. Parameters of analogue or hybrid system may be estimated indirectly, using a particular bilaterally unique transformation of the analogue mathematical model into a discrete one and applying the presented algorithm to the transformed model. The significance of the presented algorithm is thus increased in practice...
Ključne reči: Parameter estimation, Box-Jenkins model, discrete system, analogue system, algorithm, flow chart, computer, response measuring. |