Luís Rato

Cascade control of a distributed collector solar field

Abstract This paper reports experimental results on the cascade control of a distributed collector solar field. The control problem consists of keeping constant the field outlet oil temperature by acting on the circulating oil flow used for heat transfer. In the inner loop an adaptive model based predictive controller exploiting the information conveyed by accessible disturbances (radiation changes and inlet oil temperature) is used, while in the outer loop a PID is employed. The need for adaptive control arises from the time varying behaviour of the plant. Due to the generality of the methods employed, the experience reported is relevant to a wide class of industrial processes.

ADAPTIVE AND NON-ADAPTIVE MODEL PREDICTIVE CONTROL OF AN IRRIGATION CHANNEL

The performance achieved with both adaptive and non-adaptive Model Predictive Control (MPC) when applied to a pilot irrigation channel is evaluated. Several control structures are considered, corresponding to various degrees of centralization of sensor information, ranging from local upstream control of the different channel pools to multivariable control using only proximal pools, and centralized multivariable control relying on a global channel model. In addition to the non-adaptive version, an adaptive MPC algorithm based on redundantly estimated multiple models is considered and tested with and without feedforward of adjacent pool levels, both for upstream and downstream control. In order to establish a baseline, the results of upstream and local PID controllers are included for comparison. A systematic simulation study of the performances of these controllers, both for disturbance rejection and reference tracking is shown.

Control of a water delivery canal with cooperative distributed MPC

This article addresses the problem of controlling pool levels in a water delivery canal using a novel cooperative distributed MPC control algorithm that incorporates stability constraints. According to a distributed control strategy, a local control agent is associated to all canal gates (actuators). In order to achieve cooperative action, each control agent computes the corresponding gate position (manipulated variable) by performing the minimization of a cost function that considers not only its local control objectives, but also the ones of their immediate neighbors. For this purpose, a MPC algorithm with stability constraints is used (SIORHC). At the beginning of each sampling interval, local control agents exchange information with their neighbors and adjust their decisions in an iterative way. The resulting distributed MPC is denoted D-SIORHC and yields a stable closed-loop. Experimental results are provided to show the influence of the controller configuration parameters on the resulting performance.

Multi-Platform Controller Interface for SCADA Application

Abstract This paper concerns the development of a SCADA-Controller Interface (SCI) application for an open-channel experimental facility. Water delivery canals are complex and spatially distributed systems. The proposed application is to be applied to test control algorithms developed by several research groups with different technical approaches. The proposed interface allows the development of controllers in different environments – C/C++, MATLAB/Simulink, and GNU Prolog – and may be easily extended to other environments. The experimental facilities with the used instrumented canal, the programmable logic controller (PLC) network and the SCADA system are also described in this paper. Finally, some software experimental results are presented.

Application of RotaSVM for HLA Class II Protein-Peptide Interaction Prediction

In this article, the recently developed RotaSVM is used for accurate prediction of binding peptides to Human Leukocyte Antigens class II (HLA class II) proteins. The HLA II - peptide complexes are generated in the antigen presenting cells (APC) and transported to the cell membrane to elicit an immune response via T-cell activation. The understanding of HLA class II protein-peptide binding interaction facilitates the design of peptide-based vaccine, where the high rate of polymorphisms in HLA class II molecules poses a big challenge. To determine the binding activity of 636 non-redundant peptides, a set of 27 HLA class II proteins are considered in the present study. The prediction of HLA class II - peptide binding is carried out by an ensemble classifier called RotaSVM. In RotaSVM, the feature selection scheme generates bootstrap samples that are further used to create a diverse set of features using Principal Component Analysis. Thereafter, Support Vector Machines are trained with these bootstrap samples with the integration of their original feature values. The effectiveness of the RotaSVM for HLA class II protein-peptide binding prediction is demonstrated in comparison with other traditional classifiers by evaluating several validity measures with the visual plot of ROC curves. Finally, Friedman test is conducted to judge the statistical significance of RotaSVM in prediction of peptides binding to HLA class II proteins.

Distributed LQG control of a water delivery canal with feedforward from measured consumptions

This work addresses the design of distributed LQG controllers for water delivery canals that include feedforward from local farmer water consumptions. The proposed architecture consists of a network of local control agents, each connected to one of the canal pools and sharing information with their neighbors in order to act in a coordinated way. In order to improve performance, the measurement of the outflow from each pool is used as a feedforward signal. Although the feedforward action is local, it propagates due to the coordination procedure. The paper presents the distributed LQG algorithm with feedforward and experimental results in a large scale pilot water delivery canal.

Multivariable and Distributed LQG Control of a Water Delivery Canal

AbstractThis paper addresses the problem of the development of a distributed linear quadratic Gaussian (LQG) controller for a water delivery canal. The control structure proposed relies on a set of LQG control agents interconnected through a communication network. Each of these local control agents controls a canal reach made of a pool and the corresponding downstream gate and receives information (output signal and control moves) only from the corresponding canal reach and the ones that are adjacent to it. An algorithm is proposed to achieve coordinated action of the different local control agents. This distributed control structure is compared with centralized multivariable LQG control. Several aspects with incidence on performance are addressed, including the modification of the quadratic cost to ensure a constraint on closed-loop poles, the use of a nonlinear filter to limit noise effects, and the impact of a quantization commonly forced in gate position. Experimental results obtained in a pilot canal...

Analysis of Pancreas Histological Images for Glucose Intolerance Identification Using Wavelet Decomposition

Subtle structural differences can be observed in the islets of Langerhans region of microscopic image of pancreas cell of the rats having normal glucose tolerance and the rats having pre-diabetic (glucose intolerant) situations. This paper proposes a way to automatically segment the islets of Langerhans region from the histological image of rat’s pancreas cell and on the basis of some morphological feature extracted from the segmented region the images are classified as normal and pre-diabetic. The experiment is done on a set of 134 images of which 56 are of normal type and the rests 78 are of pre-diabetic type. The work has two stages: primarily, segmentation of the region of interest (roi), i.e., islets of Langerhans from the pancreatic cell and secondly, the extraction of the morphological features from the region of interest for classification. Wavelet analysis and connected component analysis method have been used for automatic segmentation of the images. A few classifiers like OneRule, Naive Bayes, MLP, J48 Tree, SVM, etc, are used for evaluation among which MLP performed the best.

RotaSVM: A New Ensemble Classifier

In this paper, an ensemble classifier, namely RotaSVM, is proposed that uses recently developed rotational feature selection approach and Support Vector Machine classifier cohesively. The RotaSVM generates the number of predefined outputs of Support Vector Machines. For each Support Vector Machine, the training data is generated by splitting the feature set randomly into \(\mathcal{S}\) subsets. Subsequently, principal component analysis is used for each subset to create new feature sets and all the principal components are retained to preserve the variability information in the training data. Thereafter, such features are used to train a Support Vector Machine. During the testing phase of RotaSVM, first the rotation specific Support Vector Machines are used to test and then average posterior probability is computed to classify sample data. The effectiveness of the RotaSVM is demonstrated quantitatively by comparing it with other widely used ensemble based classifiers such as Bagging, AdaBoost, MultiBoost and Rotation Forest for 10 real-life data sets. Finally, a statistical test has been conducted to establish the superiority of the result produced by proposed RotaSVM.

A Fast Algorithm for Automatic Segmentation of Pancreas Histological Images for Glucose Intolerance Identification

This paper describes a novel fast algorithm for automatic segmentation of islets of Langerhans and β-cell region from pancreas histological images for automatic identification of glucose intolerance. Here, LUV colour space and connected component analysis are used on 134 images among which 56 are of normal and rest 78 are of prediabetic type. The paper also talks about a supervised learning approach for classifying the images based on their morphological features. In the present work, we have introduced a modern classifier weighted ELM (Extreme Learning Machine) for prediabetes identification. Performances of weighted ELM are comparable with all the present-day’s robust classifiers such as Support Vector Machines (SVM), Multilayer Perceptron (MLP), etc. We have also compared the result with traditional ELM and observed better performance in the present skewed dataset with substantial improvement in training time.