This paper presents a novel feature selection method based on the conditional mutual information (CMI). The proposed High Order Conditional Mutual Information Maximization (HOCMIM) incorporates high order dependencies into procedure and has straightforward interpretation due to its bottom-up derivation. HOCMIM is derived from CMI's chain expansion expressed as maximization optimization problem. problem solved using greedy search procedure, which speeds up entire process. experiments are run...

The objective of Industry 5.0 is to (re)centre the human operator amidst digital process automation. This requires new data processing technologies that extract expertise and integrate it with advanced modelling techniques enhance human-computer collaboration. In this work, we present an integrated systematic approach combines contemporary technology knowledge gained from engineers. Specifically, develop, investigate, compare data- expert-driven approaches for selecting variables a real-time...

New residential neighborhoods are often supplied with heat via district heating systems (DHS). Improving the energy efficiency of a DHS is critical for increasing sustainability and satisfying user requirements. In this paper, we present HELIOS, dedicated artificial intelligence (AI) model designed specifically modeling load in DHS. HELIOS leverages combination established physical principles expert knowledge, resulting superior performance compared to existing state-of-the-art models....

This paper proposes a mixture of univariate linear regression models (MULRM) to be applied in time-varying scenarios, and its application soft sensor problems. Offline online solutions MULRM will obtained using the Expectation-Maximization Algorithm. A forgetting factor introduced solution discount information already learned data, so that it can time varying settings. The proposed method allows recursive any problem, without necessity storing past value data. then two real-world prediction...

The paper proposes a methodology to online self-evolve direct fuzzy logic controllers (FLCs), deal with unknown and time-varying dynamics. proposed self-designs the controller, where control rules can be added or removed considering predefined criterion. aims reach structure easily interpretable by human operators. FLC is defined univariate rules, each input variable represented set of improving interpretability ability learned controller. self-evolving methodology, when process under...

Abstract Improving river water quality requires a thorough understanding of the relationship between constituent concentration and discharge during runoff events (i.e., C‐Q hysteresis), which may be strongly non‐linear. Analysis hysteresis on large temporal scales provides unprecedented insights into event dynamics long‐term trends in surface groundwater. Despite increasing availability time series data quality, there are still limited quantitative modeling frameworks that enable this...

This paper proposes a new method for input variable and delay selection (IVDS) Soft Sensors (SS) design. The IVDS algorithm is composed by the following steps: (1) Time selection; (2) Identification exclusion of redundant variables; (3) Best variables subset selection. proposed in this work performs through two distinct methods, mutual information (MI) applied to multilayer perceptron (MLP) based approach performed. It shown case studies that application before applying increases...

Abstract Deep learning (DL) has captured the attention of community with an increasing number recent papers in regression applications, including surveys and reviews. Despite efficiency good accuracy systems high-dimensional data, many DL methodologies have complex structures that are not readily transparent to human users. Accessing interpretability these models is essential factor for addressing problems sensitive areas such as cyber-security systems, medical, financial surveillance,...

In this paper a study is made of the problem classifying scenarios, in terms semantic categories, based on data gathered from sensors mounted on-board mobile robots operating indoors. Once are transformed to feature space, supervised classification performed by probabilistic approach called Dynamic Bayesian Mixture Models (DBMM). This combines class-conditional probabilities learning models and incorporates past inferences. work, several experiments multi-class place reported publicly...

This paper proposes a method for fault detection and replacement of the sensor responsible by measurement burning zone temperature in rotary cement kiln. The control is crucial kiln therefore quality, pollutant emissions, consumed energy. However flying dust within can block pyrometer sensor, causing faults sensor. Exploring analytical redundancy that usually exist industrial processes, proposed methodology uses neural network trained using an online sequential extreme learning machine to...

The performance and efficiency of photovoltaic (PV) modules are significantly impacted by their operating temperature. Therefore, accurately estimating the PV module temperature ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$T_{m}$</tex-math></inline-formula> ) is a crucial factor in assessment systems. This article introduces hybrid model for estimation that combines both physical data-driven modeling....

In this paper, a new isolated DC-DC boost converter based on three-state switching cell (3SSC) is proposed. The mentioned allows the use of only two windings in isolation transformer, as well as, series connection one DC current blocking capacitor to avoid its saturation. Other relevant characteristics are, voltage across controlled switches low, which utilization lower drain-to-source conduction resistances (RDSon) MOSFETs, and through autotransformer winding almost continuous minimizing...

This paper proposes a method for Soft Sensors design using Multilayer Perceptron model based on co-evolutionary genetic algorithms, called CEV-MLP. jointly and automatically selects the best input variables configuration of network prediction setting. The CEV-MLP is constituted by three levels, first level respective delays set, second composed parameters hidden layers to be optimized (number neurons in transfer function), third combination level. was successfully applied, compared with two...

The paper proposes a method to select the best variables and respective time-lags for industrial applications when objective is estimation of target variable using information content empirical data. No further assumed about process. problem jointly selecting treated as selection problem. This assumption implies an increase input dimensionality multicollinearity into space. Then, multidimensional mutual estimator based on l-nearest neighbor algorithm used in forward search procedure...

Industries are faced with the choice of suitable process control policies to improve costs, quality and raw material consumption. In paper pulp industry, it is important estimate quickly Chemical Oxygen Demand (COD), a parameter that highly correlated product quality. Soft Sensors (SSs) have been established as alternative hardware sensors laboratory measurements for monitoring purposes. However, in real setups often difficult get sufficient data SS development. This work proposes Ensemble...

A common step in most of water treatment plants is the chemical coagulation. The coagulation process destabilizing colloidal particles suspended raw by addition coagulants. Generally, determination quantity coagulant to be added made manually jar tests. However, manual control has slow response changes and it requires intensive laboratory analysis. To reduce effort improve change quality, this work proposes dosage using dynamic neural network modeling available sensors as input model. case...