Frequency DomainFIRFilter Design Using FuzzyAdaptive Simulated Annealing
Antonio Petraglia, Mariane R. Petraglia · 2007
An alternative approach todigital filter design is intoanobjective orcostfunction oftheparameters (inthe presented. Although itisusedinthis paper tofitFIRfilters to present case, nonzero values ofthefilter impulse response), frequency domainspecifications, themethodissuitable to sothatglobally minimizing suchanumerical function means application inother problems ofdigital filter design, wherethe matter under study canbestated asfinding theglobal minimumof tofindtheoptimal filter parameters, compatible withthe anumerical function offilter parameters. Theadopted numericaldesign conditions. Tovalidate theproposed method, wehave optimization algorithm isbaseduponthewell-known simulatedapplied ittothedesign problems presented in(1), and annealing paradigm anditsimplementation knownasfuzzycompared therespective results. Thatpaper presents avery adaptive simulated annealing. Theoverall design methodisas goodandcomprehensive setofresults anditisbased upona follows: starting fromfrequency domainconstraints anda neural network framework. As willbeseen, theASA- parameterized expression ofthefilter family underadaptation, a . f corresponding training setiscreated, an errorfunctiondesigned ilpeft synthesized andaglobal minimization process executed. Atthe reported i (1) end,thepoint that minimizes globally theparticular costfunction athanddetermines theoptimalfilter. II.PROBLEMSTATEMENT