Cairo Lúcio Nascimento Júnior

GFRBS-PHM: A Genetic Fuzzy Rule-Based System for PHM with improved interpretability

This paper presents an approach to predict the Remaining Useful Life (RUL) of a generic system when a higher level of interpretability of the prediction model is desired. A set of well known computational intelligence techniques such as Decision Trees, Fuzzy Logic, and Genetic Algorithms is used to generate a hybrid model which is called Genetic Fuzzy Rule-Based System (GFRBS) supported by a Decision Tree. The proposed method automatically generates fuzzy rules and tunes the associated membership functions. Accuracy and improved interpretability are achieved during training since they are coded in the fitness function used by the genetic algorithm. The proposed approach is applied to a case study of degradation of aeronautical engines. The task is to estimate the remaining useful life of a commercial aircraft engine using only historical data gathered by the sensors embedded in the engine.

Use of PHM Information and System Architecture for Optimized Aircraft Maintenance Planning

Remaining useful life (RUL) estimations obtained from a prognostics and health monitoring (PHM) system can be used to plan in advance for the repair of components before a failure occurs. However, when system architecture is not taken into account, the use of PHM information may lead the operator to rush to replace a component that would not affect immediately the operation of the system under consideration. This paper presents a methodology for decision support in maintenance planning with application in aeronautical systems. The proposed methodology combines system architecture information and RUL estimations for all components in the system under study, allowing the estimation of an overall system-level RUL (S-RUL). The S-RUL information can be used to support maintenance decisions regarding the replacement of multiple components. For this purpose, the decision problem can be cast into an optimization framework involving the minimization of the component replacement cost under a safety constraint. Two case studies are used to illustrate the S-RUL concept, as well as the proposed optimization methodology.

An experimental validation of reinforcement learning applied to the position control of UAVs

In this paper, we explore the application of Reinforcement Learning (RL) to the derivation of control laws for the flight control of an unmanned aerial vehicle (UAV). The controllers are derived off-line with a simulation and the solutions are ported to an actual aircraft. Experimental results showed that the controllers stabilize the quad-rotor during the path tracking as has been learned in the simulation.

Daily and monthly sugar price forecasting using the mixture of local expert models

This article concerns the application of the Mixture of Local Expert Models (MLEM) to predict the daily and monthly price of the Sugar No. 14 contract in the New York Board of Trade. This technique can be seen as a forecasting method that performs data exploratory analysis and mathematical modeling simultaneously. Given a set of data points, the basic idea is as follows: 1) a Kohonen Neural Network is used to divide the data into clusters of points, 2) several modeling techniques are then used to construct competing models for each cluster, 3) the best model for each cluster is then selected and called the Local Expert Model. Finally, a so-called Gating Network combines the outputs of all Local Expert Models. For comparison purposes, the same modeling techniques are also evaluated when acting as Global Experts, i. e., when the technique uses the entire data set without any clustering.

PHM-based Multi-UAV task assignment

This paper is relating to the application of Integrated Vehicle Health Management (IVHM) concepts based on Prognostics and Health Monitoring (PHM) techniques to Multi-UAV systems. Considering UAV as a mission critical system, it is expected and required to accomplish its operational objectives with minimal unscheduled interruptions. So that, it does make sense for UAV to take advantage of those techniques as enablers for the readiness of multi-UAV. The main goal of this paper is to apply information from a PHM system to support decision making through an IVHM framework. PHM system information, in this case, comprises UAV remaining useful life (RUL) estimations. UAV RUL is computed by means of a fault tree analysis that it is fed by a distribution function from a probability density function relating time and failure probability for each UAV critical components. The IVHM framework, in this case, it is the task assignment based on UAV health condition (RUL information) using the Receding Horizon Task Assignment (RHTA) algorithm. The study case was developed considering a team of electrical small UAVs and pitch control system was chosen as the critical system.

Planning and learning for cooperative construction task with quadrotors

In this paper, we describe a stochastic learning approach for planning of assembly and construction tasks of 3-D structures using multiple quadrotors. A planning framework is proposed to generate different sets of high-level plans for the aerial robots. This architecture demonstrates significant advances in ability to quickly find good solutions for complex construction tasks, considering the real world criteria. The high-level plans are derived off-line using learning and heuristic search algorithms in a simulation environment. This process involves the planning of the sequence of maneuvers for each aerial robot, the sequence of assembly of the desired structure, and the set of trajectories for the quadrotors navigate through the moderately constrained and dynamic environment. Moreover, an efficient conflict resolution for multiple vehicles based on speed planning is proposed. The simulation results of the autonomous aerial robot construction system are presented and the obtained high-level plans are evaluated.

Autonomous feature-based exploration using a low-cost mobile robot

This article is concerned with the solution of the SLAM (Simultaneous Localization And Mapping) problem in a medium scale indoor environment using a low-cost mobile robot that autonomously explores the environment. The low-cost robot was built with a distance measurement subsystem composed of three types of sensors: a wireless webcam with a laser pointer (a visual sensor), two infrared sensors and an ultrasonic sensor. SLAM experiments were performed in small and medium scale environments where the robot operated autonomously. This article shows the results of a SLAM experiment in 55 m long by 2.8 m wide corridor where several artificial walls were used to simulate a more complex environment. The acquired map closely matches the real environment and is also used to navigate the robot.

Treinamento de Redes Neurais para Tarefas de Inspeção Mediante Reforços de Múltiplos Críticos

The main objective of thi s work is to investigate the concept s of game theory in the context of reinforcement learning with multiple critics. The neural net is assumed to perform an inspection task, where desirable input patte rns should produce high ‘accept’ output value s while undesirable input patters are expected to yield high ‘ reject’ output value s. The multiple critics may or may not coope rate, so that game theoretic situations arise. In this context , the stochastic learning automata is shown to converge to Nash equilibrium points or Pareto solutions, depending on the nature of the information state.

MoGFT-I: A Multi-objective Optimization approach for the Cart and Pole control problem

In this article a set of well-known computational intelligence techniques such as Decision Trees, Fuzzy Logic, and Multi-objective Optimization Genetic Algorithm are combined to generate a novel hybrid method which is called MoGFT-I: Multi-objective Genetic Fuzzy Rule Based System supported by a Decision Tree with Improved Interpretability. The output of the proposed supervised learning method is a set of Mamdani-type fuzzy systems which are optimized and distributed along a Pareto curve by considering two conflicting attributes: accuracy and interpretability. The MoGFT-I method is then applied to the Cart and Pole control problem such that a set of feedback controller are designed to control this unstable nonlinear dynamical system.

Steering a quadruped robot: Simulation and experimental results

This article presents a solution for the steering problem of a quadruped robot such that it follows a desired path specified by a set of waypoints. The robot was assembled using the popular Robotis Bioloid Comprehensive Kit and it uses 3 servomotors in each leg. Steering is achieved by using an extra servomotor to turn the robot frontal axis which supports its frontal legs. The proposed path following algorithm firstly selects a target point according to the current robot position and the two waypoints of interest at the time. Then the angle of the robot frontal axis is adjusted to point to this target point. As the robot moves, a new target point is selected. Simulation and experimental results are presented and show that the proposed path following algorithm can successfully steer the robot.