Selected Methods of Predicting Financial Health of Companies: Neural Networks Versus Discriminant Analysis

This paper focuses on the financial health prediction of businesses. The issue of predicting the financial health of companies is very important in terms of their sustainability. The aim of this paper is to determine the financial health of the analyzed sample of companies and to distinguish financially healthy companies from companies which are not financially healthy. The analyzed sample, in the field of heat supply in Slovakia, consisted of 444 companies. To fulfil the aim, appropriate financial indicators were used. These indicators were selected using related empirical studies, a univariate logit model and a correlation matrix. In the paper, two main models were applied—multivariate discriminant analysis (MDA) and feed-forward neural network (NN). The classification accuracy of the constructed models was compared using the confusion matrix, error type 1 and error type 2. The performance of the models was compared applying Brier score and Somers’ D. The main conclusion of the paper is that the NN is a suitable alternative in assessing financial health. We confirmed that high indebtedness is a predictor of financial distress. The benefit and originality of the paper is the construction of an early warning model for the Slovak heating industry. From our point of view, the heating industry works in the similar way in other countries, especially in transition economies; therefore, the model is applicable in these countries as well.

Experimental Investigation of Laser Surface Transformation Hardening of 4340 Steel Spur Gears

This paper presents an experimental investigation of laser surface transformation hardening (LSTH) of 4340 steel spur gears using regression analysis. The experimental work is focused on the effects of various LSTH parameters on the hardness profile shape and the hardened depth variation. The investigations are based on a structured design of experiments and improved statistical analysis tools. The experimentations are carried out on AISI 4340 steel spur gears using a commercial 3 kW Nd:YAG laser system. Laser power, scanning speed, and rotation speed are used as process parameters to evaluate the variation of the hardened depth and to identify the possible relationship between the process parameters and the hardened zone physical and geometrical characteristics. Based on the experimental data and analysis of variance, the direct and interactive contributions of the process parameters on the variation of the hardness profile shape and the hardened depth are analyzed. The main effects and the interaction effects are also evaluated. The results reveal that all the process parameters are relevant. The cumulative contribution of the three parameters in the hardened depth variation represents more than 80% with a clear predominance of laser power. The contribution of the interactions between the parameters represents 12% to 16%. The resulting hardness values are relatively similar for all the experimental tests with about 60 HRC. The evaluation of the produced regression models for hardened depth prediction shows limited performance suggesting that the predictive modeling process can be improved.

Vacuum Thermoforming Process: An Approach to Modeling and Optimization Using Artificial Neural Networks

In the vacuum thermoforming process, the group effects of the processing parameters, when related to the minimizing of the product deviations set, have conflicting and non-linear values which make their mathematical modelling complex and multi-objective. Therefore, this work developed models of prediction and optimization using artificial neural networks (ANN), having the processing parameters set as the networks’ inputs and the deviations group as the outputs and, furthermore, an objective function of deviation minimization. For the ANN data, samples were produced in experimental tests of a product standard in polystyrene, through a fractional factorial design (2^k-p). Preliminary computational studies were carried out with various ANN structures and configurations with the test data until reaching satisfactory models and, afterwards, multi-criteria optimization models were developed. The validation tests were developed with the models’ predictions and solutions showed that the estimates for them have prediction errors within the limit of values found in the samples produced. Thus, it was demonstrated that, within certain limits, the ANN models are valid to model the vacuum thermoforming process using multiple parameters for the input and objective, by means of reduced data quantity.

Implementation of Neural Network Models for Parameter Estimation of a PEM-Electrolyzer

Predictive models were built using neural networks for hydrogen flow rate, electrolyzer system-efficiency and stack-efficiency respectively. A comprehensive experimental database forms the foundation for the predictive models. It is argued that, due to the high costs associated with the hydrogen measuring equipment; these reliable predictive models can be implemented as virtual sensors. These models can also be used online for monitoring and safety of hydrogen equipment. The quantitative accuracy of the predictive models is appraised using statistical techniques. These mathematical models are found to be reliable predictive tools with an excellent accuracy of ±3% compared with experimental values. The predictive nature of these models did not show any significant bias to either over prediction or under prediction. These predictive models, built on a sound mathematical and quantitative basis, can be seen as a step towards establishing hydrogen performance prediction models as generic virtual sensors for wider safety and monitoring applications.

Multicriteria meta-heuristics for AGV dispatching control based on computational intelligence

In many manufacturing environments, automated guided vehicles are used to move the processed materials between various pickup and delivery points. The assignment of vehicles to unit loads is a complex problem that is often solved in real-time with simple dispatching rules. This paper proposes an automated guided vehicles dispatching approach based on computational intelligence. We adopt a fuzzy multicriteria decision strategy to simultaneously take into account multiple aspects in every dispatching decision. Since the typical short-term view of dispatching rules is one of the main limitations of such real-time assignment heuristics, we also incorporate in the multicriteria algorithm a specific heuristic rule that takes into account the empty-vehicle travel on a longer time-horizon. Moreover, we also adopt a genetic algorithm to tune the weights associated to each decision criteria in the global decision algorithm. The proposed approach is validated by means of a comparison with other dispatching rules, and with other recently proposed multicriteria dispatching strategies also based on computational Intelligence. The analysis of the results obtained by the proposed dispatching approach in both nominal and perturbed operating conditions (congestions, faults) confirms its effectiveness.

Layered neural networks based analysis of radon concentration and environmental parameters in earthquake prediction

A layered neural network (LNN) has been employed to estimate the radon concentration in soil related to the environmental parameters. This technique can find any functional relationship between the radon concentration and the environmental parameters. Analysis of the data obtained from a site in Thailand indicates that this approach is able to differentiate time variation of radon concentration caused by environmental parameters from those arising by anomaly phenomena in the earth (e.g. earthquake). This method is compared with a linear computational technique based on impulse responses from multivariable time series. It is indicated that the proposed method can give a better estimation of radon variations related to environmental parameters that may have a non-linear effect on the radon concentration in soil, such as rainfall.