Search

Here, this article reports about the design of a soft sensor (SS) able to monitor the hazardous gases in industrial plants. The SS is designed to estimate the gas concentrations by means of the measurements coming from an array of sensors, avoiding at the same time the humidity and temperature influence on array outputs. The SS has been designed with a data-driven approach, using a set of experimental data acquired in a laboratory. The design methodologies of two different SSs are compared, with the aim of obtaining both the good performance and a low computational complexity. As a first approach, a principal component analysis (PCA) has been performed to exploit the high correlation among some of the measures coming from the sensor array. A classical multilayer perceptron neural network is then trained to estimate the relationships between the PCA outputs and the gas concentrations. As a second approach, a deep belief network (DBN) has been considered. The data here reported show a good accuracy in the evaluation of several gas concentrations, even in the presence of noised measurements, allowing an efficient risk warning. Even if both the methods gave a similar performance, a lower number of model parameters and a lower noise sensitivity are obtained when using DBNs.

Fractional calculus is a mathematical framework that has attracted considerable interest from mathematicians, physicists, and engineers. Among its applications, the use of fractional calculus in the automatic control field has led to interesting results, such as more robust controllers, compared to their integer-order counterparts. The proposed work utilizes the physical realization of a solid-state fractional-order capacitor for the implementation of a fractional-order lead compensator. The proposed capacitor is realized using a carbon black-based dielectric. Therefore, a fully analog closed-loop system implementation is realized. A suitable case study is conducted to validate the controller performance, both from simulations and experimentally. The obtained results further confirm the possibility of realizing and applying a fully analog fractional-order controller.

Data selection is a critical issue in data-driven soft sensor design. The paper proposes a new method for data selection based on a feature extraction step, followed by data selection algorithms. The method has been applied to an industrial case study, i.e., the estimation of the quality of processed wastewater produced by a Sour Water Stripping plant working in a refinery. The paper reports the results obtained with different data selection algorithms. The comparison has been performed both by using raw data and the feature extraction phase.

The knowledge of the electrochemical processes inside a Fuel Cell (FC) is useful for improving FC diagnostics, and Electrochemical Impedance Spectroscopy (EIS) is one of the most used techniques for electrochemical characterization. This paper aims to propose the identification of a Fractional-Order Transfer Function (FOTF) able to represent the FC behavior in a set of working points. The model was identified by using a data-driven approach. Experimental data were obtained testing a Proton Exchange Membrane Fuel Cell (PEMFC) to measure the cell impedance. A genetic algorithm was firstly used to determine the sets of fractional-order impedance model parameters that best fit the input data in each analyzed working point. Then, a method was proposed to select a single set of parameters, which can represent the system behavior in all the considered working conditions. The comparison with an equivalent circuit model taken from the literature is reported, showing the advantages of the proposed approach.

Soft Sensors (SSs) are inferential dynamical models employed in industries to perform prediction of process hard-to-measure variables based on their relation with easily accessible ones. They allow implementation of real-time control and monitoring of the plants and present other advantages in terms of costs and efforts. Given the complexity of industrial processes, these models are generally designed with data-driven black-box machine learning (ML) techniques. ML methods work well only if the data on which the prediction is performed share the same distribution with the one on which the model was trained. This is not always possible, since plants can often show new working conditions. Even similar plants show different data distributions, making SSs not scalable between them. Models should then be created from scratch with highly time-consuming procedures. Transfer Learning (TL) is a field of ML that re-uses the knowledge from one task to learn a new different, but related, one. TL techniques are mainly used for classification tasks. Only recently TL techniques have been adopted in the SS field. The proposed survey reports the state of the art of TL techniques for nonlinear dynamical SSs design. Methods and applications are discussed and the new directions of this research field are depicted.

Soft Sensors (SSs) are mathematical models that allow real-time estimation of hard-to-measure variables as a function of easy-to-measure ones in an industrial process, emulating the behavior of existing sensors when they are, for instance, taken off for maintenance. The Sulfur Recovery Unit (SRU) from a refinery is taken in exam. Recurrent Neural Networks (RNN) can capture the nonlinearity of such process but present a high complexity training and a very time-consuming structure optimization. For this reason, strategies to use pre-existing models are here examined by testing the transferability of the SSs between two parallel lines of the process.

The design and application of Soft Sensors (SSs) in the process industry is a growing research field, which needs to mediate problems of model accuracy with data availability and computational complexity. Black-box machine learning (ML) methods are often used as an efficient tool to implement SSs. Many efforts are, however, required to properly select input variables, model class, model order and the needed hyperparameters. The aim of this work was to investigate the possibility to transfer the knowledge acquired in the design of a SS for a given process to a similar one. This has been approached as a transfer learning problem from a source to a target domain. The implementation of a transfer learning procedure allows to considerably reduce the computational time dedicated to the SS design procedure, leaving out many of the required phases. Two transfer learning methods have been proposed, evaluating their suitability to design SSs based on nonlinear dynamical models. Recurrent neural structures have been used to implement the SSs. In detail, recurrent neural networks and long short-term memory architectures have been compared in regard to their transferability. An industrial case of study has been considered, to evaluate the performance of the proposed procedures and the best compromise between SS performance and computational effort in transferring the model. The problem of labeled data scarcity in the target domain has been also discussed. The obtained results demonstrate the suitability of the proposed transfer learning methods in the design of nonlinear dynamical models for industrial systems.

Failure prognostics can improve industrial systems’ availability and reliability by determining the occurrence of failure and estimating the system’s remaining useful life (RUL) before deterioration. This paper presents a prognostic method based on data-driven and degradation model approaches to accurately predict the RUL of a scale replica system under different operating conditions. For this purpose, vibration data collected off a faulty system over many days is analyzed in both time and frequency domains to excavate data features related to the degradation and then identify the health indicator (HI) and the failure mechanism. At that point, the constructed HI, thanks to the feature importance ranking and principal component analysis (PCA) techniques, is used as a mathematical input for an exponential degradation model. Finally, the results are validated using real-world vibration-based degradation information leading to an effective predictive strategy for industrial systems.

Soft Sensors (SSs) are inferential models which are widely used in industry. They are generally built through data-driven approaches that exploit industry historical databases. Selection of input variables is one of the most critical issues in SSs design. This paper aims at highlighting difficulties arising from the implementation of data-driven input selection methods when solving real-world case studies. A procedure is, therefore, proposed for input selection, based on both data-driven and expert-driven input selection methods. The procedure allows designing SSs with good prediction accuracy and a low number of inputs. The design of an SS for a real-world industrial process is used. The results reported show that the selection methods proposed in literature do not give consistent results when applied to the considered case study. The key role for plant expert knowledge emerges, outlining the opportunity of judicious use of automatic data-driven procedures.

In this paper, a numerical procedure to obtain the pseudo-polynomial characteristic equation of a commensurate time-delay system is proposed. The method is formulated in terms of an interpolation problem, and it is based on the generation of a suitable set of random numbers. Some results, showing how the random number generation method affects the procedure convergence are also reported. This aspect is investigated in the paper and a suitable strategy to guarantee the problem solution is proposed. A numerical example is reported to outline the method behavior.

The current trend of automation and data exchange in manufacturing technologies is commonly referred to as the fourth industrial revolution, and known as Industry 4.0. Its “pillars” are Cyber-Physical Systems (CPS), Cognitive Computing (CC), Internet of Things (IoT) and Cloud. In such a scenario, to know both technologies and processes supporting digital business is mandatory. In this paper we present our “vision”, i.e., an approach integrating Business Intelligence and Machine Learning and addressed to CIOs and IT leaders in order to build a complete digital business technology platform with generation of linked data knowledge.

In this paper a dielectric structure made with Carbon Black has been designed to realize a Fractional Order Element (FOE). This device is hence used as capacitor to build a fractional-order Wien oscillator. FOE experimental frequency characterization, theoretical and experimental simulations on Wien oscillator are provided.

The prediction of failures in rotating machines is an important issue in industries to improve safety, to reduce the cost of maintenance and to prevent accidents. In this paper a predictive maintenance algorithm, based on the analysis of the orbits shape of the rotor shaft is proposed. It is based on an autonomous image pattern recognition algorithm, implemented by using a Convolutional Neural Network (CNN). The CNN has been designed, by using a suitable database, to recognize the orbits shape, allowing both fault detection and classification.

In this paper, a novel coupling scheme is applied to the synchronization of hyperchaotic systems. The coupling scheme is based on a simple linear dynamical system driven by a suitable signal, accounting for the synchronization error. The proposed approach allows to enhance the synchronization properties of classical hyperchaotic systems, reducing the complexity of the error signal. An analysis of the effect of the choice of the linear coupling system parameters will be presented showing how they are able to control both the synchronization strength and the settling time to the synchronous behavior, leading to wider ranges for the coupling coefficient.

This paper proposes a model of the dynamics of credit contagion through non-performing loans on financial networks. Credit risk contagion is modeled in the context of the classical SIS (Susceptibles-Infected-Susceptibles) epidemic processes on networks but with a fundamental novelty. In fact, we assume the presence of two different classes of infected agents, and then we differentiate the dynamics of assets subject to idiosyncratic risk from those affected by systemic risk by adopting a SIIS (Susceptible-Infected1-Infected2-Susceptible) model. In the recent literature in this field, the effect of systemic credit risk on the performance of the financial network is a hot topic. We perform numerical simulations intended to explore the roles played by two different network structures on the long-term behavior of assets affected by systemic risk in order to analyze the effect of the topology of the underlying network structure on the spreading of systemic risk on the structure. Random graphs, i.e., the Erd&ouml, s&ndash, R&eacute, nyi model, are considered &ldquo, benchmark&rdquo, network structures while core-periphery structures are often indicated in the literature as idealized structures, although they are able to capture interesting, specific features of real-world financial networks. Moreover, as a matter of comparison, we also perform numerical experiments on small-world networks.

In this paper, a black-box model, suitable for being used as a Soft Sensor for an Ionic Polymer-Metal Composite, is introduced. Applications of such materials, in many fields, such as bio-inspired robotics, medicine, and aerospace, have been reported. The proposed Soft Sensor is a data-driven Nonlinear Finite Impulse Response model, implemented by using a Deep Neural Network, where the output is estimated only on the basis of input past samples. The model has been selected with the aim of eliminating any hardware position sensor and, therefore, reducing the complexity of the positioning system.