Machine Learning Models for Suspension System Performance Prediction in Large Dump Trucks

For achieving bulk economic excavation, large dump trucks are being used in majority of the earth moving operations resulting in high impact shovel loading operations (HISLO) which exposes the operators to high levels of whole body vibrations (WBV). As the truck ages, hydro-pneumatic suspension struts loses their capability to effectively attenuate the vibration. A system is required to monitor and predict the performance of the suspension struts in real time. Machine learning and artificial intelligence (AI) has been applied for modeling and predicting the suspension system performance for lighter/smaller vehicles. However, no work has been done to implement machine learning or AI for modelling and predicting the performance of hydro-pneumatic struts in large dump trucks. Therefore, current work is going to be a pioneering effort towards developing machine learning and AI models for solving this problem. Support vector machine (SVM) and regularized dual non-linear regression has been implemented in order to achieve the goal. 3D virtual simulator for CAT793D was used to conduct experiments in MSC.ADMAS environment. The two main recorded outputs were the RMS accelerations in the vertical and horizontal directions at the operator seat for characterizing the performance of the suspension system. During the development and the training of the models, eighty percent (80%) of the total experimental data was used and the remaining 20% was used during the testing and validation of the developed models. SVM model showed the desired accuracy in terms of hydro-pneumatic suspension system performance prediction for large dump trucks. These models can be implemented in the dump truck controller which can then be used to monitor the performance of the suspension system in real-time, and with that proper maintenance and/or replacement can be scheduled by the maintenance personnel. Workplace safety, operator\u27s health and the overall system efficiency can be greatly improved with an implementation of such an intelligent system