An Intelligent Controller based on LAMDA

Abstract

In this paper, a new intelligent controller, which is based on the machine learning algorithm called LAMDA (Learning Algorithm for Multivariable Data Analysis) is presented. This algorithm can be used for supervised and unsupervised learning, and it has been applied in the detection of functional states of systems. LAMDA is based on the calculation of the Global Adequacy Degree (GAD) of an individual to a class in order to determine its degree of similarity. In the training phase is established the centers of the classes, which correspond to the functional states of the system. However, being a fuzzy classification algorithm, it does not have an inference method; which is proposed in this work, assigning a numerical value at the output of the controller when the system is in one of the classes. The idea is that the controller can take the system from a current functional state to a desired one. Our proposal has been tested in two case studies, where it is demonstrated its versatility in the implementation of PD and PI controllers. Finally, its performance is compared to other controllers.