Nicola Matarese

A Method to Point Out Anomalous Input-Output Patterns in a Database for Training Neuro-Fuzzy System with a Supervised Learning Rule

When designing a neural or fuzzy system, a careful preprocessing of the database is of utmost importance in order to produce a trustable system. In function approximation applications, when a functional relationship between input and output variables is supposed to exist, the presence of data where the similar set of input variables is associated to very different values of the output is not always beneficial for the final system to design. A method is presented which can be used to detect anomalous data, namely non-coherent associations between input and output patterns. This technique, by mean of a comparison between two distance matrix associated to the input and output patterns, is able to detect elements in a dataset, where similar values of input variables are associated to quite different output values. A numerical example and a more complex application in the pre-processing of data coming from an industrial database were presented.

Simulation of an Automated Warehouse for Steel Tubes

The paper presents a software which simulates an existing automated warehouse of steel tubes, including all the movements of the tubes packs conveyors as well as the management of all the operations for their storage and recovery for delivery to customers. The reason to develop such simulator lies in the needing to test new stocking strategies, that could overcome the drawback of the ones which are already implemented in the current system. Such strategies and their possible improvements are presented and discussed. As the simulator elaborates as inputs the real data of the tubes production and customer orders exactly in the format in which they are processed by the real automated warehouse management, realistic simulations are possible. Numerical results are shown, which demonstrates that the proposed improvements in the tube packs allocation and reordering successfully increase the warehouse capacity and efficiency through an improved exploitation of the available space.

Process Integration in Energy and Carbon Intensive Industries Through Exploitation of Optimization Techniques and Decision Support

Process industries show an ever-increasing interest in reducing their environmental impact and energy consumption as well as maintaining an acceptable profit. This is particularly true for industries such as the steel one, which is among the highest energy consumers worldwide. Process modelling and optimization are techniques by which this problem can be effectively addressed, particularly if the overall system is optimised as a whole. In this article we describe a model for a discrete dynamic optimization of the process gas network in an integrated steel plant. The main sub-plants are modelled in order to calculate mass and energy balances in different scenarios of operation. The scenarios are then exploited within a multi-objective optimization problem, where cost and CO2 emissions are simultaneously minimised. The optimization is carried out by exploitation of evolutionary algorithms that enable a flexible problem formulation and to effectively generate a set of different trade-off solutions. Application of the model to an industrial case study results in an interesting potential for reduction of CO2 emissions and costs. The described optimisation model is embedded in a more general software tool to help the plant managers in their daily decision-making process.

GADF — Genetic Algorithms for distribution fitting

Distribution fitting is a widely recurring problem in different fields such as telecommunication, finance and economics, sociology, physics, etc. Standard methods often require solving difficult equations systems or investments in specialized software. The paper presents a new approach to distribution fitting that exploits Genetic Algorithms in order to simultaneously identify the distribution type and tune its parameters by exploiting a dataset sampled from the observed random variable and a set of distribution families. The strength of this approach lies in the easiness of the expansion of this set: in fact distributions are simply described by means of their probability density functions and cumulative distribution functions, which are well-known. This approach employs two different score metrics, the Mean Absolute Error and the Kolmogorov-Smirnov test, that are linearly combined in order to find the best fitting distribution. The results obtained in an industrial application are presented and discussed.

Prediction of under pickling defects on steel strip surface

An extremely important part of the finishing line is the pickling process, in which oxides formed during the hot rolling stage are removed from the surface of the steel sheets. The efficiency of the pickling process is mainly dependent on the nature of the oxide present at the surface of the steel, but, also, on process parameters such as bath composition and time duration are relevant. When acid concentration, solution temperatures and line speed are not properly balanced, in fact, sheet defects like under pickling or over pickling may happen and their occurrence does have a very serious effect on cold-reduction performance and surface appearance of the finished product. Furthermore, product damage from handling or improper equipment adjustment can render the steel unsuitable for further processing. This is the reason why it is important that process significant parameters are controlled and maintained as accurately as possible in order to avoid these undesired phenomena. In the present work, a control algorithm, composed by two different modules, i.e. decision tree and rectangular Basis Function Network, has been implemented to aim of predicting pickling defects and suggesting the optimal speed or the admissible speed range of the steel strip in the process line. In this way the most suitable line speed value can be set in an automatic way or by the technical personnel.

Detection of transients in steel casting through standard and ai-based techniques

The detection of transients in the practice of continuous casting within a steel-making industry is a key task for the prediction of final product properties but currently a direct observation of this phenomenon is not available. For this reason in this paper several standard and soft-computing based methods for the detection of transients from plant data will be tested and compared. From the obtained results it emerges that the use of a fuzzy inference system based on experts knowledge achieves very satisfactory results correctly identifying most of the transient events present in the databases provided by different companies.

GA-based solutions comparison for warehouse storage optimization

The paper analyses the issues behind allocation and reordering strategies optimization for an existing automated warehouse for the steelmaking industry. Genetic Algorithms are employed to this purpose by deriving custom chromosome structures as well as ad-hoc crossover and mutation operators. A comparison between three different solutions capable to deal with multi-objective optimization are presented: the first approach is based on a common linear weighting function that combines different objectives; in the second one, a fuzzy system is used to aggregate objective functions, while in the last one the Strength Pareto Evolutionary Algorithm is applied in order to exploit a real multi-objective optimization. These three approaches are described and results are presented in order to highlight benefits and pitfalls of each technique.

A CO2-Management Tool for Integrated Steelworks

The present paper describes an holistic CO2-monitoring system for an integrated steel-making plant. Firstly, the implementation of a centralised Database in the steelwork, supported by a dedicated server, was required as a preparatory step to collect data with regard to the main energy and Carbon containing flows. The implementation of a CO2-monitoring system was necessary in order to measure and account the emissions. The holistic model, validated with real operating data, should be able to monitor and to control the CO2 outlet and is a valid tool to reduce such emissions through the management and optimization of the relevant flows.

Quality Improvement in Hot Dip Galvanizing Line through Hybrid Case-Based Reasoning System

The present paper deals with quality improvement of flat steel sheet surface coming from the continuous Hot Dip Galvanizing (HDG) process. The main idea has been to combine a Case-Based Reasoning (CBR) system, which allows to learn from previous experience, and a module exploiting a Cause Induction in Discrimination tree (CID tree), which allows to identify the process variables of the HDG process which mostly affect the formation of surface defects on the steel sheet. This hybrid system is capable to suggest optimal variability ranges for these variables in order to reduce or avoid defects formation, by using a data mining approach. The joint use of the CBR system and the CID tree methodology allows the identification of defects and the detection of possible causes (i.e. values of some HDG process parameters) on their formation, by tracking them in a knowledge base representing a baseline for reduction of defects formation in future manufacturing.

A procedure for the detection of anomalous input-output patterns

Data preprocessing is a main step in data mining because real data can be corrupted for different causes and high performance data mining systems require high quality data. When a database is used for training a neural network, a fuzzy system or a neuro-fuzzy system, a suitable data selection and pre-processing stage can be very useful in order to obtain a reliable result. For instance, when the final aim of a system trained through a supervised learning procedure is to approximate an existing functional relationship between input and output variables, the database that is exploited in the system training phase should not contain input-output patterns for which the same input or similar input sets are associated to very different values of the output variable. In this paper a procedure is proposed for detecting non-coherent associations between input and output patterns: by comparing two distance matrices associated to the input and output patterns, the elements of the available dataset, where similar values of input variables are associated to quite different output values can be pointed out. The efficiency of the proposed algorithm when pre-processing data coming from an industrial database is presented and discussed together with a statistical assessment of the obtained results.