An artificial neural networks approach to product cost estimation. The case study for electric motor

Abstract

The aim of this paper is to present, in theoretical and application terms, artificial neural networks (ANNs) as a method of estimating the product cost. The first part of the article reviews the methods used to estimate the product cost. The basic approaches to the problem of product cost estimation, presented by various authors, were described. In the second part an empirical study using artificial neural networks was conducted. Two research methods were used in this paper: literature analysis and empirical research carried out in the form of an extensive case study. The test object is a new generation induction motor. The main research problem of the article is the modelling of artificial neural networks for the estimation process of product costs with advanced production technology. The test procedures focus on the application aspects. The conclusions discuss the usefulness and advantages of using ANN models in estimating the costs of products.

Comparison of Product Life Cycle Cost Estimating Models Based on Neural Networks and Parametric Techniques—A Case Study for Induction Motors

“…Also, the enrichment of MLP with subsequent hidden layers resulted in an increase in the estimation error. This is confirmed by the necessity of keeping the model as simple as possible (avoiding its excessive complexity), a need often postulated in the literature [41].…”

“…The training parameters and network structure ( Table 5) were selected automatically by the simulator to minimize the error function-the sum of squares (SOS). Figure 6 presents estimation errors (for the data from the test set) generated by the MLP and the RBF network (the development of [41]). For both types of ANN, there is a noticeable increase in the estimation error for the lower-cost engines.…”

“…The reason for such a low precision can be an insufficient number of cases of the low cost of the product life cycle in the training set. This may indicate an increased risk of unacceptable low precision of estimation if the product's cost strongly differs from the average value [41].…”

“…Such a division was related to the amount of available data (different types of induction motors) and constituted a compromise between the desired high number of teaching resources (required for more effective training and lower estimation errors) and the number of cases in test resources (necessary for reliable evaluation of the model's operation). It was also confirmed in the empirical research described in [41]. Two types of ANNs, with unidirectional data flow, were tested: MLP and radial basis function (RBF) networks ( [42,43]).…”