Gerardo Beruvides

Multi-objective optimization based on an improved cross-entropy method. A case study of a micro-scale manufacturing process

The strong points of Estimation-of-Distribution algorithms (EDAs) and specifically cross-entropy methods are widely acknowledged. One of the main advantages of EDAs is that the fusion of prior information into the optimization procedure is straightforward, thereby reducing convergence time when such information is available. This study presents the modified Multi-Objective Cross-Entropy (MOCE+) method, based on a new procedure for addressing constraints: (i) the use of variable cutoff values for selecting the elitist population; and, (ii) filtering of the elitist population after each epoch. We study the proposed method in different test suites and compare its performance with some other well-known optimization methods. The comparative study demonstrates the good figures of merit of the MOCE+ method in complex test suites. Finally, the proposed method is applied to the multi-objective optimization of a micro-drilling process. Two conflicting targets are considered: total drilling time and vibrations on the plane that is perpendicular to the drilling axis. The Pareto front, obtained through the optimization process, is analyzed through quality metrics and the available options in the decision-making process. Overall, the quality metrics of the MOCE+ method were better than the metrics of the other optimization methods considered in this work. The reported optimization of the micro-drilling process with the proposed method could potentially have a direct impact on improvements in industrial efficiency.

Obstacle Recognition Based on Machine Learning for On-Chip LiDAR Sensors in a Cyber-Physical System

Collision avoidance is an important feature in advanced driver-assistance systems, aimed at providing correct, timely and reliable warnings before an imminent collision (with objects, vehicles, pedestrians, etc.). The obstacle recognition library is designed and implemented to address the design and evaluation of obstacle detection in a transportation cyber-physical system. The library is integrated into a co-simulation framework that is supported on the interaction between SCANeR software and Matlab/Simulink. From the best of the authors’ knowledge, two main contributions are reported in this paper. Firstly, the modelling and simulation of virtual on-chip light detection and ranging sensors in a cyber-physical system, for traffic scenarios, is presented. The cyber-physical system is designed and implemented in SCANeR. Secondly, three specific artificial intelligence-based methods for obstacle recognition libraries are also designed and applied using a sensory information database provided by SCANeR. The computational library has three methods for obstacle detection: a multi-layer perceptron neural network, a self-organization map and a support vector machine. Finally, a comparison among these methods under different weather conditions is presented, with very promising results in terms of accuracy. The best results are achieved using the multi-layer perceptron in sunny and foggy conditions, the support vector machine in rainy conditions and the self-organized map in snowy conditions.

Coping with Complexity When Predicting Surface Roughness in Milling Processes: Hybrid Incremental Model with Optimal Parametrization

The complexity of machining processes relies on the inherent physical mechanisms governing these processes including nonlinear, emergent, and time-variant behavior. The measurement of surface roughness is a critical step done offline by expensive quality control procedures. The surface roughness prediction using an online efficient computational method is a difficult task due to the complexity of machining processes. The paradigm of hybrid incremental modeling makes it possible to address the complexity and nonlinear behavior of machining processes. Parametrization of models is, however, one bottleneck for full deployment of solutions, and the optimal setting of model parameters becomes an essential task. This paper presents a method based on simulated annealing for optimal parameters tuning of the hybrid incremental model. The hybrid incremental modeling plus simulated annealing is applied for predicting the surface roughness in milling processes. Two comparative studies to assess the accuracy and overall quality of the proposed strategy are carried out. The first comparative demonstrates that the proposed strategy is more accurate than theoretical, energy-based, and Taguchi models for predicting surface roughness. The second study also corroborates that hybrid incremental model plus simulated annealing is better than a Bayesian network and a multilayer perceptron for correctly predicting the surface roughness.

Artificial cognitive control with self-x capabilities

This computational architecture is inspired and fed by recent progress in neuroscience.The design and implementation of self-learning and self-optimization capabilities.The implementation in a low-cost computational platform to facilitate technology transfer in industry. Nowadays, even though cognitive control architectures form an important area of research, there are many constraints on the broad application of cognitive control at an industrial level and very few systematic approaches truly inspired by biological processes, from the perspective of control engineering. Thus, our main purpose here is the emulation of human socio-cognitive skills, so as to approach control engineering problems in an effective way at an industrial level. The artificial cognitive control architecture that we propose, based on the shared circuits model of socio-cognitive skills, seeks to overcome limitations from the perspectives of computer science, neuroscience and systems engineering. The design and implementation of artificial cognitive control architecture is focused on four key areas: (i) self-optimization and self-leaning capabilities by estimation of distribution and reinforcement-learning mechanisms; (ii) portability and scalability based on low-cost computing platforms; (iii) connectivity based on middleware; and (iv) model-driven approaches. The results of simulation and real-time application to force control of micro-manufacturing processes are presented as a proof of concept. The proof of concept of force control yields good transient responses, short settling times and acceptable steady-state error. The artificial cognitive control architecture built into a low-cost computing platform demonstrates the suitability of its implementation in an industrial setup.

Self-Tuning Method for Increased Obstacle Detection Reliability Based on Internet of Things LiDAR Sensor Models

On-chip LiDAR sensors for vehicle collision avoidance are a rapidly expanding area of research and development. The assessment of reliable obstacle detection using data collected by LiDAR sensors has become a key issue that the scientific community is actively exploring. The design of a self-tuning methodology and its implementation are presented in this paper, to maximize the reliability of LiDAR sensors network for obstacle detection in the ‘Internet of Things’ (IoT) mobility scenarios. The Webots Automobile 3D simulation tool for emulating sensor interaction in complex driving environments is selected in order to achieve that objective. Furthermore, a model-based framework is defined that employs a point-cloud clustering technique, and an error-based prediction model library that is composed of a multilayer perceptron neural network, and k-nearest neighbors and linear regression models. Finally, a reinforcement learning technique, specifically a Q-learning method, is implemented to determine the number of LiDAR sensors that are required to increase sensor reliability for obstacle localization tasks. In addition, a IoT driving assistance user scenario, connecting a five LiDAR sensor network is designed and implemented to validate the accuracy of the computational intelligence-based framework. The results demonstrated that the self-tuning method is an appropriate strategy to increase the reliability of the sensor network while minimizing detection thresholds.

Modeling and Optimization of Mechanical Systems and Processes

This chapter reviews the most commonly used techniques used for modeling and optimizing mechanical systems and processes. Statistical and artificial intelligence based tools for modeling are summarized, pointing their advantages and shortcomings. Also, analytic, numeric and stochastic optimization techniques are briefly explained. Finally, two cases of study are developed in order to illustrate the use of these tools, the first one dealing with the modeling of the surface roughness in a drilling process and the other one, on the multi-objective optimization of a hot forging process.

A Simple Multi-Objective Optimization Based on the Cross-Entropy Method

A simple multi-objective cross-entropy method is presented in this paper, with only four parameters that facilitate the initial setting and tuning of the proposed strategy. The effects of these parameters on improved performance are analyzed on the basis of well-known test suites. The histogram interval number and the elite fraction had no significant influence on the execution time, so their respective values could be selected to maximize the quality of the Pareto front. On the contrary, the epoch number and the working population size had an impact on both the execution time and the quality of the Pareto front. Studying the rationale behind this behavior, we obtained clear guidelines for setting the most appropriate values, according to the characteristics of the problem under consideration. Moreover, the suitability of this method is analyzed based on a comparative study with other multi-objective optimization strategies. While the behavior of simple test suites was similar to all methods under consideration, the proposed algorithm outperformed the other methods considered in this paper in complex problems, with many decision variables. Finally, the efficiency of the proposed method is corroborated in a real case study represented by a two-objective optimization of the microdrilling process. The proposed strategy performed better than the other methods with a higher hyperarea and a shorter execution time.

Monitoring tool usage on the basis of sensory information in micro-drilling operations

Real-time monitoring of tool wear in microdrilling is a key factor in improving the performance of micro-machining operations. This study presents the basis to design a real-time monitoring system for microdrilling operations. The experimental research has the objective of determining both the minimum type of necessary information and the axial positioning of the sensors that provide information on the conditions of the cutting tool. The behavior of three signals in the time domain measured by three piezoelectric sensors is examined: z-component of the force, recorded on a multi-component dynamometer; and, vibration on the y and z-axes, captured by two piezoelectric accelerometers. The sensory information is analyzed and processed with the aim of identifying possible temporal behavioral patterns in the signal features and cutting tool usage. The study of microdrilling operations in the time domain provides clear evidences that certain trends in the temporal behavior of a series of signal features generated by cutting force and vibrations can be related to increased usage or wear of the cutting tool.

A fuzzy-genetic system to predict the cutting force in microdrilling processes

This paper presents the modeling of thrust force in microdrilling processes of five commonly used alloys (titanium-based, tungsten-based, aluminum-based and invar). The process was carried out by peck drilling and the influence of five parameters (drill diameter, cutting speed, feed rate, one-step feed length and total drilling length) on the behavior of the thrust force was considered. A fuzzy system was used for describing these relationships and genetic algorithms were used for fitting the parameters of the model from the experimental data. Finally a comparison with a traditional cutting model obtained with a regression model was made showing both models a similar correlation values (R2), 0.84 for the regression model and 0.86 for the fuzzy-genetic system. However, the fuzzy model showed a better generalization capability (> 0.9) than the regression model, (very poor, near to 0).

Self-Tuning Method for Increasing Reliability in Obstacle Detection based on Internet-of-Things LiDAR Sensor Models

Nowadays, the research and development of on-chip LiDAR sensors for vehicle collision avoidance is growing very fast. Therefore, the assessment of the reliability in obstacle detection using the information provided by LiDAR sensors has become a key issue to be explored by the scientific community. This paper presents the design and implementation of a self-tuning method in order to maximize the reliability of an Internet-of-Things sensors network and to minimize the number of sensors to localize with the required accuracy obstacles by a detection threshold. In order to achieve this goal, models that predict accuracy (i.e., prediction error) for object localization using data collected by LIDAR sensors are designed and implemented in Webots Automobile 3D simulation tool. The approach is based on combining different techniques. Firstly, point-cloud clustering technique and an error prediction model library composed by a multilayer perceptron neural network with backpropagation, k-nearest neighbors and linear regression are explored. Secondly the above-mentioned techniques for modeling are also combined with a supervised and reinforcement machine learning technique, Q-learning in order to minimize the detection threshold. In addition, a IoT driving assistance simulated scenario with a LiDAR sensor network is designed in order to validate the prediction model and the optimal configuration of the sensor network to guarantee reliability in obstacle localization. The results demonstrate that the self-tuning method is appropriate to increase the reliability of the sensor network whereas minimizing the detection threshold.