Dynamic Load Identification for Mechanical Systems: A Review

Abstract

Due to the great challenges of measuring forces directly, identifying dynamic loads based on accessible responses is a crucial problem in engineering, helping ensure integrity and reliability of mechanical structures. Dynamic load identification is a difficult inverse problem due to matrix ill-posedness, noise-sensitivity and computational scale, especially in uncertain structures. Unexpected inaccurate or non-unique solutions may be found if these problems are not well addressed. During the past decades, many methods have been proposed to deal with these problems. This paper tries to provide a comprehensive review of techniques for dynamic load identification, including under ill-posedness and uncertain parameter processing; with an emphasis on the statistical, data science, machine learning, and artificial intelligence aspects. Classical physics-based dynamic load identification theories in frequency and time domain are also introduced. Research challenges and prospects of dynamic load identification in mechanical systems are discussed finally. This review may offer guidelines for dynamic load identification in practical complex structures, as well as possibilities for further researches. Some methods could have broader applicability to other inverse problems.