Application Example of Deep Echo State Neural Networks Case Study: Prediction of Mobile Hydraulic Crane’s Pressure and Ecu Temperatures

Abstract

Real data taken from the field can be used as design parameters in engineering studies. Alternatively, the calculated and analyzed values should be verified by field tests. However, waiting for data from the field for design parameters can sometimes take a very long time. This makes engineering solutions too long or impossible. In the same way, there may be tests that are difficult to test in design verifications, require cost, and create security problems. This study sought solutions to the problems described using the DESN model in two different data sets. In the study, deep Echo State neural network analysis was performed on two different data sets. As data, the pressures formed in the cylinder during the lifting and lowering of 6 different loads by a truck-mounted mobile crane and the 4- month device temperature of the electronic control unit in an overhead crane were recorded. Echo State Network application was made on these records with deep learning. After training with 80% of the data, the DeepESN model was tested with 20%, and these results were evaluated.