A neural networks application for the study of the influence of transport conditions on the working performance

[摘要] This paper presents a study about the factors that influence the working performances of workers in the automotive industry. These factors regard mainly the transportations conditions, taking into account the fact that a large number of workers live in places that are far away of the enterprise. The quantitative data obtained from this study will be generalized by using a neural network, software simulated. The neural network is able to estimate the performance of workers even for the combinations of input factors that had been not recorded by the study. The experimental data obtained from the study will be divided in two classes. The first class that contains approximately 80% of data will be used by the Java software for the training of the neural network. The weights resulted from the training process will be saved in a text file. The other class that contains the rest of the 20% of experimental data will be used to validate the neural network. The training and the validation of the networks are performed in a Java software (TrainAndValidate Java class). We designed another Java class, Test.Java that will be used with new input data, for new situations. The experimental data collected from the study. The software that simulated the neural network. The software that estimates the working performance, when new situations are met. This application is useful for human resources department of an enterprise. The output results are not quantitative. They are qualitative (from low performance to high performance, divided in five classes).

[发布机构] University of Piteti, Manufacturing and Industrial Management Department, Târgul din Vale Street No. 1, Piteti, Romania^1;University of Piteti, Department of Electronics, Computers, Communications and Electrical Engineering, Târgul din Vale Street No. 1, Piteti, Romania^2

[效力级别] 运输工程 [学科分类] 工业工程学

[关键词] Input datas;Input factors;Java class;Java software;Neural networks applications;Quantitative data;Training process;Working performance