Digital Filter Satisfaction Optimization Design

Digital filter is one of the particular and important classes in DSP (digital signal processing) and discrete time system which converts input digital sequence into different output ones using computing algorithm. It has being the most important means in DSP because of its flexible, convenient, higher precision and reliability. As all know, there are many kinds of modern algorithms in digital filter designing, including genetic algorithms (GA), simulated annealing (SA), tabu search (TS), the ant colony optimization (ACO), neural networks (NNs) and artificial immune algorithm (IA) etc. Nevertheless, each method has its advantage and disadvantage. A right algorithm can not only reduce more runtime but save EMS memory as well.Frequency sampling is one of ways that are used in designing FIR digital filter, but the traditional ways can’t satisfy the designer’s requires. This paper established a optimum design of FIR digital filter by sampling in the intermediate zone based on the multi-criterion satisfactory optimization (MCSO), and discussed the satisfactory optimization.Chapter one gives the simple introduction to the traditional optimization and what would be considered the satisfactory optimization. Chapter two is on the basis of analyzing mankind’s intelligent mode of thinking, proposes the general definition of the satisfactory optimization and the procedure diagram of it. A definition of Satisfactory solution and satisfactory rate function that generally suitable are given. The dissertation introduces a model of MCSO that combines satisfactory optimization with genetic algorithm. Chapter three introduces the linearity constraint condition and frequency respond characteristic of FIR digital filter, and establishes a design way in common use-----Frequency sampling. Chapter four applies MCSO to the optimized design of FIR digital filters, and realizes the optimization of the intermediate zone sample, develops the satisfactory optimum designs of the low FIR digital filter by two points samples in the intermediate zone and the band FIR filter by two points. It indicated that the satisfactory optimization is more perfect than traditional one from the good simulation results. The validity and feasibility of this method are stated.