Designing digital IIR low-pass differentiators with multi-objective optimization
Ståle Andreas van Dorp Skogstad, Sverre Holm, Mats Erling Hovin
Abstract
Ståle Andreas van Dorp Skogstad, Sverre Holm, Mats Erling Hovin
Abstract
In this paper we examine the possibility of designing IIR low-pass differentiators by approaching it as a weighted multi-objective optimization problem and solving it with an unbiased metaheuristic search algorithm. By collecting several solutions with different sets of weights we are able to make a thorough comparison of different design strategies. We present possible designs that are realizable with (1) cascading classical IIR low-pass filters with appropriate operators, and (2) non-cascaded general IIR differentiator designs. Elliptical filters are found to be the most magnitude-optimal among the first type. However, the non-cascaded approach found more optimal IIR differentiators at the expense of a more complicated search. Finally, we present some non-cascaded general designs that compare favorably with the available designs given in literature, and which we reason are nearly optimal.
OpenAlex reports 10 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In this paper we examine the possibility of designing IIR low-pass differentiators by approaching it as a weighted multi-objective optimization problem and solving it with an unbiased metaheuristic search algorithm. By collecting several solutions with different sets of weights we are able to make a thorough comparison of different design strategies. We present possible designs that are realizable with (1) cascading classical IIR low-pass filters with appropriate operators, and (2) non-cascaded general IIR differentiator designs. Elliptical filters are found to be the most magnitude-optimal among the first type. However, the non-cascaded approach found more optimal IIR differentiators at the expense of a more complicated search. Finally, we present some non-cascaded general designs that compare favorably with the available designs given in literature, and which we reason are nearly optimal.
Key concepts: Infinite impulse response, Differentiator, 2D Filters, Computer science, Optimal design, Mathematical optimization, Algorithm, Digital filter