The Abrasive Water Jet milling process is demonstrated to be an efficient technology for milling low machinability materials. Although its capability is demonstrated, the industrial application of this technology requires a depth control, for which a work focused in process modelling is needed. This research work introduces a model to predict the kerf shape in AWJ slot milling in Aluminium 7075-T651 in terms of four important process parameters: pressure, abrasive mass flow rate, stand-off distance and traverse feed rate. A hybrid evolutionary approach was employed for modelling the profile through two parameters: the maximum cutting depth and the full width at half maximum. Both the maximum depth and the width were also modelled as a function of aforementioned process parameters based on Analysis of Variance and regression techniques. Combination of two models resulted in an adequate strategy to predict the kerf shape for different machining conditions.

Many researchers demonstrated the capability of Abrasive Waterjet (AWJ) technology for precision milling operations. However, the concurrence of several input parameters along with the stochastic nature of this technology leads to a complex process control, which requires a work focused in process modelling. This research work introduces a model to predict the kerf shape in AWJ slot milling in Aluminium 7075-T651 in terms of four important process parameters: the pressure, the abrasive flow rate, the stand-off distance and the traverse feed rate. A hybrid evolutionary approach was employed for kerf shape modelling. This technique allowed characterizing the profile through two parameters: the maximum cutting depth and the full width at half maximum. On the other hand, based on ANOVA and regression techniques, these two parameters were also modelled as a function of process parameters. Combination of both models resulted in an adequate strategy to predict the kerf shape for different machining conditions.

Classification methods have been widely used during last years in order to predict patterns and trends of interest in data. In present paper, a multiclassifier approach that combines the output of some of the most popular data mining algorithms is shown. The approach is based on voting criteria, by estimating the confidence distributions of each algorithm individually and combining them according to three different methods: confidence voting, weighted voting and majority voting. To illustrate its applicability in a real problem, the drill wear detection in machine-tool sector is addressed. In this study, the accuracy obtained by each isolated classifier is compared with the performance of the multiclassifier when characterizing the patterns of interest involved in the drilling process and predicting the drill wear. Experimental results show that, in general, false positives obtained by the classifiers can be slightly reduced by using the multiclassifier approach.

This paper presents the objectives and current activities in the course of an FP7 project, entitled EnergyWarden (www.energywarden.net). The project broadly aims at an integrated view of energy management and renewable technology, as deployed in the building domain. EnergyWarden will deliver an integrated toolset including three distinct modules: A Simulator (EW-S), to run long and short time simulations of the renewable energy production at the hourly, daily and yearly level, a controller (EW-C) to control energy flows locally at the building or at a district level and a user information module (EW-U) to deliver a detailed profile of the building energy production and use, to provide a real time calculation of CO2 savings and to assess energy module performance, based on real time data and according to existing standards.

Both complexity and lack of knowledge associated to physical processes makes physical models design an arduous task Frequently, the only available information about the physical processes are the heuristic data obtained from experiments or at best a rough idea on what are the physical principles and laws that underlie considered physical processes Then the problem is converted to find a mathematical expression which fits data There exist traditional approaches to tackle the inductive model search process from data, such as regression, interpolation, finite element method, etc Nevertheless, these methods either are only able to solve a reduced number of simple model typologies, or the given black-box solution does not contribute to clarify the analyzed physical process In this paper a hybrid evolutionary approach to search complex physical models is proposed Tests carried out on a real-world industrial physical process (abrasive water jet machining) demonstrate the validity of this approach.

The success of intelligent diagnosis systems normally depends on the knowledge about the failures present on monitored systems. This knowledge can be modelled in several ways, such as by means of rules or probabilistic models. These models are validated by checking the system output fit to the input in a supervised way. However, when there is no such knowledge or when it is hard to obtain a model of it, it is alternatively possible to use an unsupervised method to detect anomalies and failures. Different unsupervised methods (HCL, K-Means, SOM) have been used in present work to identify abnormal behaviours on the system being monitored. This approach has been tested into a real-world monitored system related to the railway domain, and the results show how it is possible to successfully identify new abnormal system behaviours beyond those previously modelled well-known problems.

Both complexity and lack of knowledge associated to physical processes makes physical models design an arduous task. Frequently, the only available information about the physical processes are the heuristic data obtained from experiments or at best a rough idea on what are the physical principles and laws that underlie considered physical processes. Then the problem is converted to find a mathematical expression which fits data. There exist traditional approaches to tackle the inductive model search process from data, such as regression, interpolation, finite element method, etc. Nevertheless, these methods either are only able to solve a reduced number of simple model typologies, or the given black-box solution does not contribute to clarify the analyzed physical process. In this paper a hybrid evolutionary approach to search complex physical models is proposed. Tests carried out on both theoretical and real-world physical processes demonstrate the validity of this approach.

This article introduces an approach to anomaly intrusion detection based on a combination of supervised and unsupervised machine learning algorithms. The main objective of this work is an effective modeling of the TCP/IP network traffic of an organization that allows the detection of anomalies with an efficient percentage of false positives for a production environment. The architecture proposed uses a hierarchy of Self-Organizing Maps for traffic modeling combined with Learning Vector Quantization techniques to ultimately classify network packets. The architecture is developed using the known SNORT intrusion detection system to preprocess network traffic. In comparison to other techniques, results obtained in this work show that acceptable levels of compromise between attack detection and false positive rates can be achieved.

This paper describes a genetic system for designing and training feed-forward artificial neural networks to solve any problem presented as a set of training patterns. This system, called GANN, employs two interconnected genetic algorithms that work parallelly to design and train the better neural network that solves the problem. Designing neural architectures is performed by a genetic algorithm that uses a new indirect binary codification of the neural connections based on an algebraic structure defined in the set of all possible architectures that could solve the problem. A crossover operation, known as Hamming crossover, has been designed to obtain better performance when working with this type of codification. Training neural networks is also accomplished by genetic algorithms but, this time, real number codification is employed. To do so, morphological crossover operation has been developed inspired on the mathematical morphology theory. Experimental results are reported from the application of GANN to the breast cancer diagnosis within a complete computer-aided diagnosis system.

This paper proposes a new learning method based on evolutionary techniques to train artificial neural networks for playing chess. Instead of generating the following movement, artificial neural networks are proposed to accomplish the evaluation of each position generated by a search algorithm. A real-coded genetic algorithm combined with an improved version of the morphological crossover operator. has been employed to train the neural networks. A self-play algorithm is applied to calculate the fitness of the individuals, which represent a set of weights and biases of a neural architecture. This is a non-supervised approach that permits to increase the play level of the chess engine without expert, knowledge.

The goal of this work is to propose a novel approach to function optimisation by evolutionary techniques, in particular, real-coded genetic algorithms. A new genetic crossover operator, suitable for real codification, has been designed. This operator is called morphological crossover as it is based on mathematical morphology theory. The morphological crossover includes a new genetic diversity measure that has low computational cost. This operator is presented along with the resolution of a set of optimisation problems, including neural network training. The results are compared to other optimisation approaches as gradient descent methods or binary and real-coded genetic algorithms using different crossover operators. These tests show that the properties exhibited by the proposed operator when using real-coded genetic algorithms give higher convergence speed and less probability of being trapped in a local optimum.

This paper proposes a new approach for constructing fuzzy knowledge bases using evolutionary methods. We have designed a genetic algorithm that automatically builds neuro-fuzzy architectures based on a new indirect encoding method. The neuro-fuzzy architecture represents the fuzzy knowledge base that solves a given problem; the search for this architecture takes advantage of a local search procedure that improves the chromosomes at each generation. Experiments conducted both on artificially generated and real world problems confirm the effectiveness of the proposed approach.

This paper describes a new evolutionary system known as ADANNET for the generation and adaptation of feed-forward artificial neural networks to solve any problem presented as a set of training patterns. ADANNET synthesizes the structure of the network that better solves the given problem and, parallelly, accomplishes the training process, Both processes use new techniques based in genetic algorithms. Basic-architectures codification method and a specialized crossover operator (the Hamming crossover) for this type of codification have been developed to solve the neural architecture design process, while a new a new crossover operator for real-coded genetic algorithms (mathematical morphology crossover) has been designed for adapting the network. Several experiments have been made to show that ADANNET obtains the smallest neural architecture that solves the given problem.

This work presents a new system for real time breast abnormalities detection that could be related to a carcinoma, taking as input a digitized mammography, in order to assist radiologists in their mammography interpretation task. The system built has been designed to the parallel detection of microcalcifications and breast masses. Algorithms based on mathematical morphology combined with dynamic statistical methods are employed in microcalcifications detection. Histogram analysis of the digitized mammogram and a modified version of the watershed algorithm have also been used for breast masses localization. The output given by the system consists on a set of suspicious regions of being a carcinoma located in the original digitized image. .A clinical database has been built for testing purposes comprising 690 mammographic studies for which surgical verification is available, 392 of them obtained in 1997, and the rest in 1998.

Inverse protein folding or protein design stands for searching a particular amino acids sequence whose native structure or folding matches a pre specified target. The problem of finding the corresponding folded structure of a particular sequence is, per se, a hard computational problem. We use a genetic algorithm for searching the space of potential sequences, and the fitness of each individual is measured with the output of a second GA performing a minimization process in the space of structures. Using an off-lattice protein-like 2D model, we show how the implemented techniques are able to obtain a variety of sequences attaining the target structures proposed.