Fast prediction method for fatigue life of pump truck boom structure based on ensemble learning model

DONG Qing; SU Youcheng; XU Gening; SHE Lingjuan; CHANG Yibin

doi:10.16579/j.issn.1001.9669.2026.06.014

Abstract：[Objective]To rapidly and accurately assess the fatigue life of in-service concrete pump truck boom structures,a fatigue life prediction method based on an ensemble learning model is proposed,utilizing monitoring data and machine learning techniques.[Methods]Firstly,a concrete pump truck information acquisition system was employed to obtain functional and performance characteristics during the operational phase of the pump truck.Through data preprocessing and transformation,a sample dataset of stress range under typical working conditions,denoted as O,was generated.From the perspective of complementary advantages,a Stacking model for stress range prediction was constructed using gradient boosting decision tree(GBDT),random forest(RF),extra trees(ET),adaptive boosting(Adaboost),and sequential learners.Subsequently,kernel density estimation sampling(KDES)was utilized to extract functional characteristics of the pump truck's operation within specific service cycles,which were then input into the established Stacking model to predict the stress range dataset for the boom structure.Furthermore,using Matlab as the computational platform and integrating fracture mechanics theory,rapid predictions of fatigue life for the boom structure were achieved.Reliability analysis was conducted to ascertain the reliability of the corresponding fatigue life predictions,thereby enhancing the credibility of the results.Finally,taking a 56X-6RZ model concrete pump truck from a certain company as an example,the feasibility of the proposed method was validated through comparisons with single machine learning models.[Results]The proposed method provides a basis for determining maintenance cycles and retirement decisions for pump trucks based on fatigue life assessments.