Journal of Science & Technology 101 (2014) 007-011 The Study of Improving the Hedge Algebra Based Controller Nguyen Huu Cong^', Ngo Kien Trun^, Nguyen Tien Du^ ' Thai Nguyen University, Tan TInnh Ward, Thai Nguyen City ^Thai Nguyen University of Technology Received February 18, 2014 accepted- April 22, 2014 Abstract In recent years, studies of hedge algebra applications in the field of control have been remarkably successful Using soft calculating tools such as fuzzy logic and hedge algebra in control has some advantages such as (1) they can control the objects with incomplete information; (2) if the controllers are well-designed, they will be enough intelligent to control nonlinear objects However, the above controllers also contain some disadvantages including huge complexity of the algonthm, longer calculating time than PID controller's: so they can not control the objects requinng fast response This paper presents the idea and algonthm to design and implement the hedge algebra based controller which has been improved to reduce the complexity and calculating time of the controller as well as open the way for real application Keywords' Hedge algebra based controller, reduce complexity, reduce calculating time Introduction The intelligent controllers are increasingly applied in industrial systems Reducing the calculatmg time and simplifying programming for the chosen microprocessor (controller) in fact are always concerned and studied by designers with different methods Hedge Algebra based Controller (HAG) is a new approach in calculating of the fuzzy conh-oller and is usually designed with inputs including error and derivative of error [2,3] Designing HAG is based on fuzzy model with many control rules leading to the requirement; the chosen microprocessor must have a huge memory capacity for direct access and long calculating time, A developed idea is if all states in the phase space of error are put into the input of the controller (such as PID), the rules and the computational complexity of the controller will he able to he reduced From that idea, the previous researchers designed a fiizzy controller with inputs including error, derivative of error and integral of error and achieved certain success As a result, a lot of rules to design controller were reduced, for example, from 75 to 27 control rules [1] However, this work sfill has existed limitations' (I) The interpolation to calculate the output value is very difficult because it is implemented in 4dimensional space with lots of calculations and complexibly mathematical formulas • Con-esponding Author: Tel; (+84) 913589758 Email: conghn@tnu edu.vn (2) The number of rules is still high, so programming in designing controller is still quite complicated With the approach of Hedge Algebra (HA), we can combine (with weights or without weights) the semantic quantify values of the input variables (mapping: R^ -* R) and define the input/output relationship in 2-dimensional space Therefore, the limitation (1) of Che fiizzy controller can be overcome and the number of control rules (2) can be reduced compared with fuzzy control, Improving the Hedge algebra based controller This paper studies the method to design HAG and proposes improved algorithm for controller The results and the efficiency can be expressed through a control problem for a nonlinear object in practice 2.L Method controller design Hedge algebra based 2.1 Brief introduction of Hedge algebra HA IS the development of logical thought of language [2,3] With the input - output relationship of the fuzzy logic the membership function must be determined discretely, the HA with an algebraic structure as a fiinction relationship allows the formation of a set of infinitely large linguistic values so that the obtained structure can simulate semantics of the language well to help the inference process of human Journal of Science & Technology 101 (2014) 007-011 ExampleConsider a set of linguistic intervals which is a linguistic domain of "TEMPERATURE" truth variable including: T = dom(TEMPERATURE) = {Large, Small, very Large, very Small, more Large, more Small, approximately Large, approximately Small, little Large, little Small, less Large, less Small, very more Large, very more Small, very possible Large, very possible Small, } Then the Imguistic domain T=dom(TEMPERATURE) can be considered as an abstract algebra: AT - (T, G, H,