Monitorización de estructuras aeronáuticas mediante técnicas de inteligencia artificial

Una de las barreras para la aplicación de las técnicas de monitorización de la integridad estructural (SHM) basadas en ondas elásticas guiadas (GLW) en aeronaves es la influencia perniciosa de las condiciones ambientales y de operación (EOC). En esta tesis se ha estudiado dicha influencia y la compensación de la misma, particularizando en variaciones del estado de carga y temperatura. La compensación de dichos efectos se fundamenta en Redes Neuronales Artificiales (ANN) empleando datos experimentales procesados con la Transformada Chirplet. Los cambios en la geometría y en las propiedades del material respecto al estado inicial de la estructura (lo daños) provocan cambios en la forma de onda de las GLW (lo que denominamos característica sensible al daño o DSF). Mediante técnicas de tratamiento de señal se puede buscar una relación entre dichas variaciones y los daños, esto se conoce como SHM. Sin embargo, las variaciones en las EOC producen también cambios en los datos adquiridos relativos a las GLW (DSF) que provocan errores en los algoritmos de diagnóstico de daño (SHM). Esto sucede porque las firmas de daño y de las EOC en la DSF son del mismo orden. Por lo tanto, es necesario cuantificar y compensar el efecto de las EOC sobre la GLW. Si bien existen diversas metodologías para compensar los efectos de las EOC como por ejemplo “Optimal Baseline Selection” (OBS) o “Baseline Signal Stretching” (BSS), estas, se emplean exclusivamente en la compensación de los efectos térmicos. El método propuesto en esta tesis mezcla análisis de datos experimentales, como en el método OBS, y modelos basados en Redes Neuronales Artificiales (ANN) que reemplazan el modelado físico requerido por el método BSS. El análisis de datos experimentales consiste en aplicar la Transformada Chirplet (CT) para extraer la firma de las EOC sobre la DSF. Con esta información, obtenida bajo diversas EOC, se entrena una ANN. A continuación, la ANN actuará como un interpolador de referencias de la estructura sin daño, generando información de referencia para cualquier EOC. La comparación de las mediciones reales de la DSF con los valores simulados por la ANN, dará como resultado la firma daño en la DSF, lo que permite el diagnóstico de daño. Este esquema se ha aplicado y verificado, en diversas EOC, para una estructura unidimensional con un único camino de daño, y para una estructura representativa de un fuselaje de una aeronave, con curvatura y múltiples elementos rigidizadores, sometida a un estado de cargas complejo, con múltiples caminos de daños. Los efectos de las EOC se han estudiado en detalle en la estructura unidimensional y se han generalizado para el fuselaje, demostrando la independencia del método respecto a la configuración de la estructura y el tipo de sensores utilizados para la adquisición de datos GLW. Por otra parte, esta metodología se puede utilizar para la compensación simultánea de una variedad medible de EOC, que afecten a la adquisición de datos de la onda elástica guiada. El principal resultado entre otros, de esta tesis, es la metodología CT-ANN para la compensación de EOC en técnicas SHM basadas en ondas elásticas guiadas para el diagnóstico de daño. ABSTRACT One of the open problems to implement Structural Health Monitoring techniques based on elastic guided waves in real aircraft structures at operation is the influence of the environmental and operational conditions (EOC) on the damage diagnosis problem. This thesis deals with the compensation of these environmental and operational effects, specifically, the temperature and the external loading, by the use of the Chirplet Transform working with Artificial Neural Networks. It is well known that the guided elastic wave form is affected by the damage appearance (what is known as the damage sensitive feature or DSF). The DSF is modified by the temperature and by the load applied to the structure. The EOC promotes variations in the acquired data (DSF) and cause mistakes in damage diagnosis algorithms. This effect promotes changes on the waveform due to the EOC variations of the same order than the damage occurrence. It is difficult to separate both effects in order to avoid damage diagnosis mistakes. Therefore it is necessary to quantify and compensate the effect of EOC over the GLW forms. There are several approaches to compensate the EOC effects such as Optimal Baseline Selection (OBS) or Baseline Signal Stretching (BSS). Usually, they are used for temperature compensation. The new method proposed here mixes experimental data analysis, as in the OBS method, and Artificial Neural Network (ANN) models to replace the physical modelling which involves the BSS method. The experimental data analysis studied is based on apply the Chirplet Transform (CT) to extract the EOC signature on the DSF. The information obtained varying EOC is employed to train an ANN. Then, the ANN will act as a baselines interpolator of the undamaged structure. The ANN generates reference information at any EOC. By comparing real measurements of the DSF against the ANN simulated values, the damage signature appears clearly in the DSF, enabling an accurate damage diagnosis. This schema has been applied in a range of EOC for a one-dimensional structure containing single damage path and two dimensional real fuselage structure with stiffener elements and multiple damage paths. The EOC effects tested in the one-dimensional structure have been generalized to the fuselage showing its independence from structural arrangement and the type of sensors used for GLW data acquisition. Moreover, it can be used for the simultaneous compensation of a variety of measurable EOC, which affects the guided wave data acquisition. The main result, among others, of this thesis is the CT-ANN methodology for the compensation of EOC in GLW based SHM technique for damage diagnosis.

Peptide design by artificial neural networks and computer-based evolutionary search

A technique for systematic peptide variation by a combination of rational and evolutionary approaches is presented. The design scheme consists of five consecutive steps: (i) identification of a “seed peptide” with a desired activity, (ii) generation of variants selected from a physicochemical space around the seed peptide, (iii) synthesis and testing of this biased library, (iv) modeling of a quantitative sequence-activity relationship by an artificial neural network, and (v) de novo design by a computer-based evolutionary search in sequence space using the trained neural network as the fitness function. This strategy was successfully applied to the identification of novel peptides that fully prevent the positive chronotropic effect of anti-β1-adrenoreceptor autoantibodies from the serum of patients with dilated cardiomyopathy. The seed peptide, comprising 10 residues, was derived by epitope mapping from an extracellular loop of human β1-adrenoreceptor. A set of 90 peptides was synthesized and tested to provide training data for neural network development. De novo design revealed peptides with desired activities that do not match the seed peptide sequence. These results demonstrate that computer-based evolutionary searches can generate novel peptides with substantial biological activity.

Self-organized phase transitions in neural networks as a neural mechanism of information processing.

Transitions between dynamically stable activity patterns imposed on an associative neural network are shown to be induced by self-organized infinitesimal changes in synaptic connection strength and to be a kind of phase transition. A key event for the neural process of information processing in a population coding scheme is transition between the activity patterns encoding usual entities. We propose that the infinitesimal and short-term synaptic changes based on the Hebbian learning rule are the driving force for the transition. The phase transition between the following two dynamical stable states is studied in detail, the state where the firing pattern is changed temporally so as to itinerate among several patterns and the state where the firing pattern is fixed to one of several patterns. The phase transition from the pattern itinerant state to a pattern fixed state may be induced by the Hebbian learning process under a weak input relevant to the fixed pattern. The reverse transition may be induced by the Hebbian unlearning process without input. The former transition is considered as recognition of the input stimulus, while the latter is considered as clearing of the used input data to get ready for new input. To ensure that information processing based on the phase transition can be made by the infinitesimal and short-term synaptic changes, it is absolutely necessary that the network always stays near the critical state corresponding to the phase transition point.

Aplicação de técnicas inteligentes com análise no domínio do tempo para reconhecimento de defeitos em motores de indução trifásicos

Os motores de indução trifásicos são os principais elementos de conversão de energia elétrica em mecânica motriz aplicados em vários setores produtivos. Identificar um defeito no motor em operação pode fornecer, antes que ele falhe, maior segurança no processo de tomada de decisão sobre a manutenção da máquina, redução de custos e aumento de disponibilidade. Nesta tese são apresentas inicialmente uma revisão bibliográfica e a metodologia geral para a reprodução dos defeitos nos motores e a aplicação da técnica de discretização dos sinais de correntes e tensões no domínio do tempo. É também desenvolvido um estudo comparativo entre métodos de classificação de padrões para a identificação de defeitos nestas máquinas, tais como: Naive Bayes, k-Nearest Neighbor, Support Vector Machine (Sequential Minimal Optimization), Rede Neural Artificial (Perceptron Multicamadas), Repeated Incremental Pruning to Produce Error Reduction e C4.5 Decision Tree. Também aplicou-se o conceito de Sistemas Multiagentes (SMA) para suportar a utilização de múltiplos métodos concorrentes de forma distribuída para reconhecimento de padrões de defeitos em rolamentos defeituosos, quebras nas barras da gaiola de esquilo do rotor e curto-circuito entre as bobinas do enrolamento do estator de motores de indução trifásicos. Complementarmente, algumas estratégias para a definição da severidade dos defeitos supracitados em motores foram exploradas, fazendo inclusive uma averiguação da influência do desequilíbrio de tensão na alimentação da máquina para a determinação destas anomalias. Os dados experimentais foram adquiridos por meio de uma bancada experimental em laboratório com motores de potência de 1 e 2 cv acionados diretamente na rede elétrica, operando em várias condições de desequilíbrio das tensões e variações da carga mecânica aplicada ao eixo do motor.

Combining visual features and Growing Neural Gas networks for robotic 3D SLAM

The use of 3D data in mobile robotics provides valuable information about the robot’s environment. Traditionally, stereo cameras have been used as a low-cost 3D sensor. However, the lack of precision and texture for some surfaces suggests that the use of other 3D sensors could be more suitable. In this work, we examine the use of two sensors: an infrared SR4000 and a Kinect camera. We use a combination of 3D data obtained by these cameras, along with features obtained from 2D images acquired from these cameras, using a Growing Neural Gas (GNG) network applied to the 3D data. The goal is to obtain a robust egomotion technique. The GNG network is used to reduce the camera error. To calculate the egomotion, we test two methods for 3D registration. One is based on an iterative closest points algorithm, and the other employs random sample consensus. Finally, a simultaneous localization and mapping method is applied to the complete sequence to reduce the global error. The error from each sensor and the mapping results from the proposed method are examined.

3D model reconstruction using neural gas accelerated on GPU

In this work, we propose the use of the neural gas (NG), a neural network that uses an unsupervised Competitive Hebbian Learning (CHL) rule, to develop a reverse engineering process. This is a simple and accurate method to reconstruct objects from point clouds obtained from multiple overlapping views using low-cost sensors. In contrast to other methods that may need several stages that include downsampling, noise filtering and many other tasks, the NG automatically obtains the 3D model of the scanned objects. To demonstrate the validity of our proposal we tested our method with several models and performed a study of the neural network parameterization computing the quality of representation and also comparing results with other neural methods like growing neural gas and Kohonen maps or classical methods like Voxel Grid. We also reconstructed models acquired by low cost sensors that can be used in virtual and augmented reality environments for redesign or manipulation purposes. Since the NG algorithm has a strong computational cost we propose its acceleration. We have redesigned and implemented the NG learning algorithm to fit it onto Graphics Processing Units using CUDA. A speed-up of 180× faster is obtained compared to the sequential CPU version.

Designing Regularizers and Architectures for Recurrent Neural Networks

Cette thèse contribue a la recherche vers l'intelligence artificielle en utilisant des méthodes connexionnistes. Les réseaux de neurones récurrents sont un ensemble de modèles séquentiels de plus en plus populaires capable en principe d'apprendre des algorithmes arbitraires. Ces modèles effectuent un apprentissage en profondeur, un type d'apprentissage machine. Sa généralité et son succès empirique en font un sujet intéressant pour la recherche et un outil prometteur pour la création de l'intelligence artificielle plus générale. Le premier chapitre de cette thèse donne un bref aperçu des sujets de fonds: l'intelligence artificielle, l'apprentissage machine, l'apprentissage en profondeur et les réseaux de neurones récurrents. Les trois chapitres suivants couvrent ces sujets de manière de plus en plus spécifiques. Enfin, nous présentons quelques contributions apportées aux réseaux de neurones récurrents. Le chapitre \ref{arxiv1} présente nos travaux de régularisation des réseaux de neurones récurrents. La régularisation vise à améliorer la capacité de généralisation du modèle, et joue un role clé dans la performance de plusieurs applications des réseaux de neurones récurrents, en particulier en reconnaissance vocale. Notre approche donne l'état de l'art sur TIMIT, un benchmark standard pour cette tâche. Le chapitre \ref{cpgp} présente une seconde ligne de travail, toujours en cours, qui explore une nouvelle architecture pour les réseaux de neurones récurrents. Les réseaux de neurones récurrents maintiennent un état caché qui représente leurs observations antérieures. L'idée de ce travail est de coder certaines dynamiques abstraites dans l'état caché, donnant au réseau une manière naturelle d'encoder des tendances cohérentes de l'état de son environnement. Notre travail est fondé sur un modèle existant; nous décrivons ce travail et nos contributions avec notamment une expérience préliminaire.

Designing Regularizers and Architectures for Recurrent Neural Networks

Cette thèse contribue a la recherche vers l'intelligence artificielle en utilisant des méthodes connexionnistes. Les réseaux de neurones récurrents sont un ensemble de modèles séquentiels de plus en plus populaires capable en principe d'apprendre des algorithmes arbitraires. Ces modèles effectuent un apprentissage en profondeur, un type d'apprentissage machine. Sa généralité et son succès empirique en font un sujet intéressant pour la recherche et un outil prometteur pour la création de l'intelligence artificielle plus générale. Le premier chapitre de cette thèse donne un bref aperçu des sujets de fonds: l'intelligence artificielle, l'apprentissage machine, l'apprentissage en profondeur et les réseaux de neurones récurrents. Les trois chapitres suivants couvrent ces sujets de manière de plus en plus spécifiques. Enfin, nous présentons quelques contributions apportées aux réseaux de neurones récurrents. Le chapitre \ref{arxiv1} présente nos travaux de régularisation des réseaux de neurones récurrents. La régularisation vise à améliorer la capacité de généralisation du modèle, et joue un role clé dans la performance de plusieurs applications des réseaux de neurones récurrents, en particulier en reconnaissance vocale. Notre approche donne l'état de l'art sur TIMIT, un benchmark standard pour cette tâche. Le chapitre \ref{cpgp} présente une seconde ligne de travail, toujours en cours, qui explore une nouvelle architecture pour les réseaux de neurones récurrents. Les réseaux de neurones récurrents maintiennent un état caché qui représente leurs observations antérieures. L'idée de ce travail est de coder certaines dynamiques abstraites dans l'état caché, donnant au réseau une manière naturelle d'encoder des tendances cohérentes de l'état de son environnement. Notre travail est fondé sur un modèle existant; nous décrivons ce travail et nos contributions avec notamment une expérience préliminaire.

MULTIPRED: A computational system for prediction of promiscuous HLA binding peptides

MULTIPRED is a web-based computational system for the prediction of peptide binding to multiple molecules ( proteins) belonging to human leukocyte antigens (HLA) class I A2, A3 and class II DR supertypes. It uses hidden Markov models and artificial neural network methods as predictive engines. A novel data representation method enables MULTIPRED to predict peptides that promiscuously bind multiple HLA alleles within one HLA supertype. Extensive testing was performed for validation of the prediction models. Testing results show that MULTIPRED is both sensitive and specific and it has good predictive ability ( area under the receiver operating characteristic curve A(ROC) > 0.80). MULTIPRED can be used for the mapping of promiscuous T-cell epitopes as well as the regions of high concentration of these targets termed T-cell epitope hotspots. MULTIPRED is available at http:// antigen.i2r.a-star.edu.sg/ multipred/.

The applicability of recurrent neural networks for biological sequence analysis

Selection of machine learning techniques requires a certain sensitivity to the requirements of the problem. In particular, the problem can be made more tractable by deliberately using algorithms that are biased toward solutions of the requisite kind. In this paper, we argue that recurrent neural networks have a natural bias toward a problem domain of which biological sequence analysis tasks are a subset. We use experiments with synthetic data to illustrate this bias. We then demonstrate that this bias can be exploitable using a data set of protein sequences containing several classes of subcellular localization targeting peptides. The results show that, compared with feed forward, recurrent neural networks will generally perform better on sequence analysis tasks. Furthermore, as the patterns within the sequence become more ambiguous, the choice of specific recurrent architecture becomes more critical.

Tracking the states of a nonlinear and nonstationary system in the weight-space of artificial neural networks

We propose a novel interpretation and usage of Neural Network (NN) in modeling physiological signals, which are allowed to be nonlinear and/or nonstationary. The method consists of training a NN for the k-step prediction of a physiological signal, and then examining the connection-weight-space (CWS) of the NN to extract information about the signal generator mechanism. We de. ne a novel feature, Normalized Vector Separation (gamma(ij)), to measure the separation of two arbitrary states i and j in the CWS and use it to track the state changes of the generating system. The performance of the method is examined via synthetic signals and clinical EEG. Synthetic data indicates that gamma(ij) can track the system down to a SNR of 3.5 dB. Clinical data obtained from three patients undergoing carotid endarterectomy of the brain showed that EEG could be modeled (within a root-means-squared-error of 0.01) by the proposed method, and the blood perfusion state of the brain could be monitored via gamma(ij), with small NNs having no more than 21 connection weight altogether.

Detecting and sorting targeting peptides wtih neural networks and support vector machines

This paper presents a composite multi-layer classifier system for predicting the subcellular localization of proteins based on their amino acid sequence. The work is an extension of our previous predictor PProwler v1.1 which is itself built upon the series of predictors SignalP and TargetP. In this study we outline experiments conducted to improve the classifier design. The major improvement came from using Support Vector machines as a "smart gate" sorting the outputs of several different targeting peptide detection networks. Our final model (PProwler v1.2) gives MCC values of 0.873 for non-plant and 0.849 for plant proteins. The model improves upon the accuracy of our previous subcellular localization predictor (PProwler v1.1) by 2% for plant data (which represents 7.5% improvement upon TargetP).

Robustness of mathematical models for biological systems

The robustness of mathematical models for biological systems is studied by sensitivity analysis and stochastic simulations. Using a neural network model with three genes as the test problem, we study robustness properties of synthesis and degradation processes. For single parameter robustness, sensitivity analysis techniques are applied for studying parameter variations and stochastic simulations are used for investigating the impact of external noise. Results of sensitivity analysis are consistent with those obtained by stochastic simulations. Stochastic models with external noise can be used for studying the robustness not only to external noise but also to parameter variations. For external noise we also use stochastic models to study the robustness of the function of each gene and that of the system.

Representation and learning of visual information for pose recognition

Recovering position from sensor information is an important problem in mobile robotics, known as localisation. Localisation requires a map or some other description of the environment to provide the robot with a context to interpret sensor data. The mobile robot system under discussion is using an artificial neural representation of position. Building a geometrical map of the environment with a single camera and artificial neural networks is difficult. Instead it would be simpler to learn position as a function of the visual input. Usually when learning images, an intermediate representation is employed. An appropriate starting point for biologically plausible image representation is the complex cells of the visual cortex, which have invariance properties that appear useful for localisation. The effectiveness for localisation of two different complex cell models are evaluated. Finally the ability of a simple neural network with single shot learning to recognise these representations and localise a robot is examined.

Bayesian retrieval of scatterometer wind fields

The retrieval of wind fields from scatterometer observations has traditionally been separated into two phases; local wind vector retrieval and ambiguity removal. Operationally, a forward model relating wind vector to backscatter is inverted, typically using look up tables, to retrieve up to four local wind vector solutions. A heuristic procedure, using numerical weather prediction forecast wind vectors and, often, some neighbourhood comparison is then used to select the correct solution. In this paper we develop a Bayesian method for wind field retrieval, and show how a direct local inverse model, relating backscatter to wind vector, improves the wind vector retrieval accuracy. We compare these results with the operational U.K. Meteorological Office retrievals, our own CMOD4 retrievals and a neural network based local forward model retrieval. We suggest that the neural network based inverse model, which is extremely fast to use, improves upon current forward models when used in a variational data assimilation scheme.