Eduardo Julio de la Moya

Model-Based Diagnosis of AC Motors

Abstract This paper presents a model-based fault diagnosis of an AC motor. A linear residual generator is going to be designed from a non-linear model of the AC motor. First, we make up a mathematical model of the motor with identification techniques, in order to use it in the residual generator. As the identified model is defined by only two equations, some operations are required for designing a good residual generator. Finally, this residual generator is validated with an experimental work on a real AC motor. We simulate electrical, mechanical and sensor failures in the motor and obtain hopeful results for the diagnosis of AC motors.

Control Performance Assessment: A General Survey

This paper reviews the different indexes and benchmarks used in the control performance assessment field of industrial processes. They are usually implemented to detect and diagnose malfunctions and disturbances in industrial controllers. This survey is just an overview of the methods and tools used in the control performance assessment/monitoring (CPA/CPM) technology which has been deeply studied over the last two decades.

Assessment of Ziegler-Nichols tuned loops using control performance monitoring indices

Most control and process engineers face a large amount of complex processes found in industry and they usually need tools for assessing their performance. This paper deals with the assessment of real industrial control processes of a pilot process plant using standard control performance monitoring indices. The control loops have been tuned using the well-known Ziegler-Nichols method. A comparison between the implemented indices with those based on the APC (Advanced Process Control) PCS7 library is performed.1

Analysis of Production Systems Using the VS-Diagram

This paper presents a graphical tool, the VS (Virtual Supervisor) states space diagram based on the FPM (Finite Positions Machines) framework in order to analyze a manufacturing system and make it recover from a faulty situation (for example, a deadlock situation) to a safer operations sequence. The VS diagrams are intuitive and easy to obtain and they are shown here as another alternative that can be used to analyze a production system. They can be used as a complementary tool in conjunction with other alternatives based on well-known formalisms such as the Petri nets.

ESQUEMA DE DETECCIÓN DE FALLOS DIFUSO BASADO EN MODELADO LINGÜÍSTICO-PRECISO DE UN MOTOR DE INDUCCIÓN

This paper presents a fuzzy models bank to detect and to identify faults using the multimodel technique, calculating a non-linear fuzzy model for each operation mode of the system. A comparison amongst the output of each model with the actual plant data isolates the faults, i.e., the operation mode of the system (normal or faulty one). Each of the considered fuzzy models is defined by a set of fuzzy rules that explain the system behaviour. These fuzzy models obtained from experimental data can be improved, through the fuzzy rules, in order to use all the characteristics of the fuzzy logic in terms of linguistic capacity (linguistic modelling). The fuzzy models are improved using similarity measurements, reducing the number of rules, eliminating incoherencies, redundancies and increasing their interpretability capacity. This method has been applied to an induction motor, in order to illustrate its behaviour and its applicability. The results shown that this method is able to detect and to identify faults even after the simplification of the models.

Full building of a sun tracker and control

As a complement to the studies of industrial engineering, students have the opportunity of making projects to improve their practical knowledge. In this case a sun tracker was built and controlled using and Arduino microprocessor. The main objective is to build an experimental plant that could be used by the students to practice in control courses. This project has been successfully used in various courses in the educational programs of teaching automatic control.

Performance monitoring of PID controllers through unfalsified control theory

Control performance monitoring indices based in unfalsified control theory are presented in this paper. A control performance condition is defined either in frequency domain or in time domain and an algorithm is developed to continuosly monitor that the performance condition is met and to detect when it is violated. The new indices have been successfully implemented on a pilot-scale industrial plant and compared to standard minimum variance control performance indices.

Modeling and process simulation of a single screw food extruder

This paper presents the design of a model for simulation of a food extruder. The model has been obtained from the energy mathematical model and experimental data of a real extruder. Some process variables having no relation with the energy model have been added to the model. Experimental results show information about extruder performance when some controlled, manipulates and additional variables are modified. The simulations carried out can be used to improve the extrusion process and to obtain initial relations between process and product variables in order to produce new products in an easier and economical manner.

VS-Diagrams Identification and Classification Using Neural Networks

VS (Virtual Supervisor) Diagrams, defined from the FPM (Finite Positions Machines) framework, are used to model, analyze and validate automated manufacturing systems and they are obtained, in a practical way, from the PLC (Programmable Logic Controller) signals. This current paper presents a neural network architecture in order to identify that type of diagrams. It is made up of a supervised Hebb neural network cascade linked to a recurrent Hopfield network.

FAULT DIAGNOSIS AND AUTOMATIC EXTRACTION OF FUZZY RULES IN AC MOTORS

Abstract In this paper a new approach to fault diagnosis in an AC motor is introduced. This system combines a neuro-fuzzy system called FasArt (Fuzzy Adaptive System ART based) and the well-known fuzzy k nearest neighbor algorithm. A set of 15 types of non destructive faults has been tested, reaching a high degree of early fault detection and fault type recognition. Moreover, taking into account the neuro-fuzzy nature of the FasArt model, a set of fuzzy rules, containing the knowledge learnt by the system, has been extracted. These rules permit a transfer of the knowledge from a numerical to a symbolic level where the fuzzy rules describe the fault in linguistic terms that can be interpreted by humans in an easier way.