Research And Engineering Application Of Impact Load Identification Technology Based On Sparse Regularization Method

Mechanical structures are often subjected to impact loads in actual engineering applications,which may cause damage defects and fatigue damage to mechanical components.In order to ensure the accuracy of the design of mechanical structures and the safety and reliability of their use,the impact loads acting on mechanical structures need to be accurately obtained.In practice,however,it is difficult to measure these impact loads by direct methods,so they need to be reconstructed by indirect methods such as load identification.The inverse problem of load recognition is uncomfortable and requires a regularization method to improve its pathological degree.In this thesis,a fundamental study of impact load recognition is carried out based on a sparse regularization algorithm based on the truncated Newton interior point method around cantilever plates,simply supported beams,and wheel-constrained damped noise reduction plate structures.Based on the above ideas,the main research work of this thesis includes:（1）An impulse response function-based model of the impact load identification positive problem is established,and for the sparsity of the impact load signal in the time domain,the l₁-parametric regularization term is introduced for sparse constraint to turn the original problem into an unconstrained convex optimization problem,and the objective function of the sparse regularization algorithm is solved by using the truncated Newton interior point method:first,the preconditioned conjugate gradient algorithm is used to calculate the search direction;then the backward linear search is used to find the minimum number of backward steps to determine the optimal computational step;finally,a specific truncation criterion is used to determine the termination condition of the algorithm so as to obtain an approximate solution that satisfies the condition.（2）The fundamental research work of impact load identification is carried out with cantilevered plate and simply supported beam foundation structure as the test object.The sparse regularization method is used to reconstruct the loads under single-point impact condition and multi-point impact condition,and the recognition effects of different regularization methods and different response points are compared.The effects of the regularization parameters of the sparse regularization method on the load recognition results are discussed,and the regularization parameters are optimized.The noise immunity performance of different regularization methods is analyzed and compared under different noise levels.The computational efficiency of different regularization methods is compared,and the relationship between computation time and coefficients of sparse regularization methods is analyzed.The effects of different measurement point distances on the impact load identification results are also explored.（3）The identification of the impact loads acting on the restrained damping noise reduction plate of high-speed trains is carried out,and the single impact as well as continuous impact loads are reconstructed,and the effects of different impact conditions and different response points on the identification results are compared and analyzed;and the effects of different hammer head materials on the identification of impact loads are analyzed and discussed.In addition,the inverse method and regularization algorithm are used to reconstruct the hanging loads of the undercar equipment of high-speed trains,and the recognition effects of different regularization methods,different undercar equipment and different noise levels are explored.The results show that the sparse regularization method has higher recognition accuracy and stability for impact loads compared with several other regularization methods,and the selection of appropriate regularization parameters will improve the recognition accuracy of the sparse regularization method,and its computational efficiency decreases gradually with the increase of coefficients.The sparse regularization method is also robust to the reconstruction of impact loads under high noise level.In a certain range,the accuracy and stability of the overall impact load recognition keep improving as the distance between the response point and the impact point keeps increasing.The sparse regularization method can still achieve high-precision reconstruction of the impact load acting on the wheel constrained damping noise reduction plate,which verifies the generality of the method.the TSVD regularization method has the highest accuracy and strongest noise resistance for the identification of the hanging load of the three under-vehicle devices.