Aki Sorsa

Real-coded genetic algorithms and nonlinear parameter identification

In this study, real-coded genetic algorithms are used in the parameter identification of the macroscopic Chemostat model. The Chemostat model utilized in this work is nonlinear having two distinct operating areas. Thus, the model is identified separately for both operating areas. The process simulator is used to generate data for the parameter identification. The optimizations with genetic algorithms are repeated with 200 different initial populations to guarantee the validity of the results. The parameter identification with genetic algorithms performs well giving accurate results.

Sensor Validation And Outlier Detection Using Fuzzy Limits

In a continuous industrial process, the accuracy and reliability of process and analytical measurements create the basis for control system performance and ultimately for product uniformity. Validation of measured values is the key and a prerequisite to guarantee reliable measurements for process control. This application introduces the use of standard deviation and density function-based absolute limits. Limits are used to cut off outliers and weigh the reliability of the on-line measurement against more reliable, but seldom made, laboratory analysis. Absolute limits are accomplished with constant or adaptively updating fuzzy limits. The adaptive fuzzy limits are recursively updated in real time when a new measured value and reference analysis become available.

Barkhausen noise characterisation during elastic bending and tensile-compression loading of case-hardened and tempered samples

This study examined the Barkhausen noise (BN) response of carburising case-hardened steel with varying tempering stages. The test material was loaded in bending and in alternating loading. The aim of the study was to obtain relevant multiparameter BN data from different loading conditions and to investigate the effect of applied stress on the BN response. The test bar series was made from case-hardened steel. Different tempering parameters were used to vary the surface hardness and the surface residual stresses of the studied series of test bars. In the bending tests, the samples were subjected to incrementally applied loading in the purely elastic deformation region. In addition, uniaxial stepwise loading with tensile and compressive stress was applied to selected samples simultaneously with the BN measurements. The BN measurements were performed under different loading conditions along with X-ray diffraction strain measurements in bending. The results revealed linear behaviour between the reciprocal root mean square value and the stress values obtained with strain gages and X-ray diffraction for the tempered samples.

Comparison of feature selection methods applied to Barkhausen noise data set

Abstract This paper discusses the feature selection problem and provides results from feature selection task in the field of non-destructive testing. The features are selected for quantitative prediction of residual stress from the Barkhausen noise signal. It is stated in the literature that the model behaviour depends on the used features and thus the selection must be carried out carefully. The selection methods studied in this paper are forward-selection, backward-elimination, simulated annealing and genetic algorithms. The used data set is divided into training and external validation data sets. The training data set is used in feature selection. The selection algorithms utilize leave-multiple-out cross-validation procedure in deciding which features are selected. The results show that backward-elimination performs poorly while the other three methods provide reasonable results. The results from the selection indicate that the stochastic methods outperform forward-selection but the external validation shows that in this case forward-selection provides results comparable to the studied more advanced methods. Even though the results indicate that forward-selection gives comparable results, it has been shown in the literature that stochastic methods are more likely to find the global optimum and thus should be used especially when the problem complexity increases.

Feature selection from Barkhausen noise data using genetic algorithms with cross-validation

Barkhausen noise is used in non-destructive testing of ferromagnetic materials. It has been shown to be sensitive to material properties but the reported results are more or less qualitative. The quantitative prediction of the material properties from the Barkhausen noise signal is challenging. In order to develop reliable models, the feature selection is critical. The feature selection method applied in this study utilizes genetic algorithms with cross-validation based objective function. Cross-validation is used because the amount of data is limited. The results show that genetic algorithms can be successfully applied to feature selection. The obtained results are reliable and rather consistent with the results obtained earlier.

State detection of a wastewater treatment plant

Abstract This paper describes a simple rule-based approach for the state detection in a biological waste water treatment plant. The plant shows bi-stable behaviour that makes its control a challenging and difficult task. The good operating point is difficult to reach and easy to lose. The approach combines the mathematical model of the plant and the available measurement information. After the state detection, the control system uses the model developed for the operation point in question and calculates the outlet substrate concentration. The approach is tested by simulations with the Chemostat -model where the kinetics follows Haldane-kinetics.

Case depth verification of hardened samples with Barkhausen noise sweeps

An interesting topic of recent Barkhausen noise (BN) method studies is the application of the method to case depth evaluation of hardened components. The utilization of BN method for this purpose is based on the difference in the magnetic properties between the hardened case and the soft core. Thus, the detection of case depth with BN can be achieved. The measurements typically have been carried out by using low magnetizing frequencies which have deeper penetration to the ferromagnetic samples than the conventional BN measurement. However, the penetration depth is limited due to eddy current damping of the signal. We introduce here a newly found sweep measurement concept for the case depth evaluation. In this study sweep measurements were carried out with various magnetizing frequencies and magnetizing voltages to detect the effect of different frequency and voltage and their correspondence to the actual case depth values verified from destructive characterization. Also a BN measurement device that has an...

Barkhausen noise-magnetizing voltage sweep measurement in evaluation of residual stress in hardened components

In this study, Barkhausen noise (BN) magnetizing voltage sweep (MVS) measurement is used to evaluate non-destructively the surface residual stress state of hardened components. A new computational feature, where the maximum slope of the sweep is divided by the corresponding magnetizing voltage, is introduced. The results show that this feature has a linear relationship with the residual stress state of the samples. The determination of residual stresses during online production of components is a highly recognized task because tensile stresses may be detrimental to the component. In this study, two sets of hardened samples are used in the analysis. A linear relationship is observed in each sample set indicating that the new feature is applicable in assessment of surface residual stresses of the components.

An Attempt to Find an Empirical Model between Barkhausen Noise and Stress

A nonlinear empirical model between stress and Barkhausen noise is identified in this study. The identification procedure uses a genetic algorithm followed by a Nelder-Mead optimization procedure. The model is identified with the data set where an external load is applied to RAEX400 low alloyed hot-rolled steel samples. The results of the study show that the identified model performs well in stress predictions. The identified model includes three terms which are in accordance with the literature.

Case Studies for Genetic Algorithms in System Identification Tasks

In this paper, genetic algorithms are used in system identification with reference to two case studies. The first case considers a structure identification problem of a black-box model. The identified black-box model is dedicated to the prediction of residual stress based on Barkhausen noise measurement. To find the most suitable model structure, a genetic algorithm with a cross-validation based objective function is utilized. The second case studies a parameter identification problem and uses a model of a biological reactor. The Chemostat model utilized in this work is nonlinear with two distinct operating areas and the model is identified separately for both operating areas. Optimization with genetic algorithms is repeated many times in both cases to guarantee the validity of results. The results from both cases are good, indicating that genetic algorithms can be used in system identification tasks.