Alexandre Manuel Mota

Artificial neural networks and neuro-fuzzy systems for modelling and controlling real systems: a comparative study

Abstract This article presents a comparison of artificial neural networks and neuro-fuzzy systems applied for modelling and controlling a real system. The main objective is to model and control the temperature inside of a kiln for the ceramic industry. The details of all system components are described. The steps taken to arrive at the direct and inverse models using the two architectures: adaptive neuro fuzzy inference system and feedforward neural networks are described and compared. Finally, real-time control results using internal model control strategy are presented. Using available Matlab software for both algorithms, the objective is to show the implementation steps for modelling and controlling a real system. Finally, the performances of the two solutions were compared through different parameters for a specific real didactic case.

Maximum power point tracker applied in batteries charging with PV panels

This work deals with the design and implementation prototype of a real time maximum power point tracker (MPPT) for photovoltaic panel (PV), aiming to improve energy transfer efficiency. This paper describes also the charging process of lead- acid batteries integrating the MPPT algorithm making an charging autonomous system that can be used to feed any autonomous application. The photovoltaic system exhibits a non-linear i-v characteristic and its maximum power point varies with solar insolation and temperature. To control the maximum transfer power from a PV panel the Perturbation and Observation (P&O) MPPT algorithm is executed by a simple microcontroller ATMEL ATTINY861V using the PV voltage and current information and controlling the duty cycle of a pulse width modulation (PWM) signal applied in to a DC/DC converter. The schematic and design of the single-ended primary inductance converter (SEPIC) is presented. This DC/DC converter is chosen because the input voltage can be higher or lower than the output voltage witch presents obvious design advantages. With the P&O MPPT algorithm implemented and executed by the microcontroller, the different charging stages of a lead-acid battery are described, showed and executed Finally, experimental results of the performance of the designed P&O MPPT algorithm are presented and compared with the results achieved with the direct connection of the PV panel to the battery.

B-Live - A Home Automation System for Disabled and Elderly People

This paper describes the architecture, operation and implementation of the B-Live home automation system. This system has been developed at Micro I/O for assisting elderly and disabled people in their homes. The paper also discusses the demonstrator deployed at the CMRRC Rovisco Pais and proposes an improved information exchange mechanism for the B-Live system.

A sliding window solution for the on-line implementation of the Levenberg-Marquardt algorithm

The Levenberg-Marquardt algorithm is considered as the most effective one for training artificial neural networks but its computational complexity and the difficulty to compute the trust region have made it very difficult to develop a true iterative version to use in on-line training. The algorithm is frequently used for off-line training in batch versions although some attempts have been made to implement iterative versions. To overcome the difficulties in implementing the iterative version, a batch-sliding window with Early Stopping, which uses a hybrid Direct/Specialized evaluation procedure, is proposed and tested with a real system. m.

Implementation of a stand-alone photovoltaic lighting system with MPPT battery charging and LED current control

This paper presents a efficient stand-alone battery photovoltaic (PV) lighting system which can provide functional illumination based on power light-emitting diodes (PLEDs). PLED with specific features of small size, long life and high-brightness light will be choices of future light sources. The PLEDs are feed using a lead acid battery that is charged with a PV panel. This paper presents an interface board with a ATMEL ATTINY861V microcontroller, a single-ended primary inductance converter (SEPIC) and input and output voltage and current measurements. The microcontroller runs the perturbation and observation (P&O) maximum power point tracker (MPPT) algorithm used in battery charging process. This algorithm makes the system more efficient. In order to control the PLEDs current, an equal interface board is used running now the proportional integral (PI) current PLED control algorithm adjusting the wanted level of light in the PLEDs array. The two identical interfaces boards based in the economic microprocessor achieved very good results in battery charging and discharge supervision improving the efficiency a life-time of the lead acid batteries (first board) and presents good current control results in the PLEDs array (second equal board) providing a constant PLEDs light even in voltage battery variations or discharges. The stand alone lighting system monitors the surround area and if it feels any movement it changes the level of light of PLEDs from signal light to illumination light. After a pre-defined time it comes back to signal light, saving energy.

Implementing the Levenberg-Marquardt Algorithm on-line: a Sliding Window Approach with Early Stopping

Abstract The Levenberg-Marquardt algorithm is considered as the most effective one for training Artificial Neural Networks but its computational complexity and the difficulty to compute the trust region have made it very difficult to develop a true iterative version to use in on-line training. The algorithm is frequently used for off-line training in batch versions although some attempts have been made to implement iterative versions. To overcome the difficulties in implementing the iterative version, a batch sliding window with Early Stopping version, which uses a hybrid Direct/Specialized evaluation procedure is proposed and tested with a real system.

Thermoelectric generator using water gas heater energy for battery charging

This work deals with the design and a prototype implementation of a maximum power point tracker (MPPT) for a thermoelectric (TE) module aiming to improve energy conversion efficiency in battery charging. This system uses TE devices that directly convert heat energy from a water gas heater to electricity to charge a battery. The TE module exhibits a non-linear i-v characteristic and its maximum power point varies with the change of its surfaces temperatures. A SEPIC (single-ended primary inductance converter) DC-DC converter is applied and controlled by a microcontroller and to achieve the maximum power point tracking it is used the perturbation and observation (P&O) algorithm. The microcontroller will also control the charging process of lead-acid battery making an autonomous system that is used to feed the necessary electronics that controls an autonomous gas water heater system. The objectives of this work are to study the principle of TE power generation and to design and develop a TE battery charger that uses heat of a water gas heater system making an autonomous electrical system. The two different charging stages of a lead-acid battery lead us to a good interface board. Finally, experimental results performance of the first stage of charge using P&O MPPT algorithm are presented and compared with the results achieved with the direct connection of the TE module to the battery.

Adapting the sampling period of a real-time adaptive distributed controller to the bus load

This paper presents a new method to allow more flexibility on the scheduling of messages across a fieldbus in overload situations. The method proposes the on-line adaptation of the sampling period to the bus load, trying to keep it as low as possible during overloads. The sampling period is allowed to change inside an interval corresponding to 4 to 10 samples per rise time. To assess the solution herein presented, a distributed adaptive control system was implemented in TrueTime, using the FTT-CAN protocol, for the communication infrastructure, a pole-placement controller and a model for the identification of plant parameters that takes into account the network-induced jitter

Comparing control strategies for autonomous line-tracking robots

Autonomous mobile robotics is a very exciting area for students particularly for those who attend courses on electronics. The authors have been involved in several activities in this area together with students of the University of Aveiro. In particular, one of such activities is to build robots to follow a line drawn on the floor. In order to do this efficiently a simulator has been implemented and used to test the influence of different control approaches. This article presents a brief description of the MATLAB-based robot model and line-tracking simulator. It then compares several different control approaches in terms of resulting integral absolute error (IAE) and integral squared error (ISE), easiness of tuning and complexity of the respective code. The compared approaches are: proportional, proportional-derivative, proportional-integral-derivative, fuzzy, table-based fuzzy, self-organising fuzzy and neural-net inverse model based.

NAT2TEST SCR : Test case generation from natural language requirements based on SCR specifications

Formal models are increasingly being used as input for automated test generation strategies. Software Cost Reduction (SCR), for example, was designed to detect and correct errors during the requirements phase, also allowing test generation. However, the syntax of SCR and other formalisms are not trivial for non-experts. In this work, we present a strategy for test case generation from natural language requirements that uses SCR as an intermediate and hidden formalism. To minimize textual ambiguity, requirements are written according to a controlled natural language. Syntactically valid requirements are mapped into their semantic representation using case frames, from which SCR specifications are derived. These specifications are then used by the T-VEC tool to generate tests cases. Our strategy was evaluated in four different domains: (i) a vending machine (toy example); (ii) a control system for safety injection in a nuclear power plant (publicly available), (iii) one example provided by our industrial partner Embraer; and (iv) the turn indicator system of Mercedes vehicles (publicly available). As a baseline we considered random testing, and, in general, our strategy outperformed it in terms of performance and mutant-based strength analysis.