For　better　solving　some　complicated　problems　in　fuzzy　automata　hierarchy,　simultaneously,　in　order　to　accomplish　better　task　for　fuzzy　signal　processing,　this　paper　presents　a　kind　of　new　automaton-fuzzy　infinite-state　automaton.　The　basic　extracted　frame　of　fuzzy　infinite-state　automaton　is　introduced　by　using　neural　networks.　To　the　extracted　fuzzy　infinite-state　automaton,　this　paper　describes　that　it　is　equivalent　to　fuzzy　finite-state　automaton,　and　its　convergence　and　stability　on　its　hierarchy　system　will　be　discussed.　Finally,　the　simulation　is　carried　on　and　the　simulation　results　show　that　the　states　of　fuzzy　infinite-state　automaton　converge　to　some　stable　states　with　extraction　frame　and　training　for　weights　what　this　paper　provides　at　last.　Finally,　some　problems　of　fuzzy　infinite-state　automaton　and　neural　networks　to　be　solved　and　development　trends　are　discussed.　These　researches　will　not　only　extend　further　automata　hierarchy,　but　also　increase　a　new　tool　for　application　of　fuzzy　signal　processing.　It　is　an　important　base　in　the　application　of　fuzzy　automata　theory.