Self-adaptive step length point cloud alignment method applied to automatic alignment of parts

A self-adaptive step length point cloud alignment method applied to automatic alignment of parts comprises the steps that mathematical models are established for part CAD three-dimensional point cloud, actual three-dimensional point cloud, target poses and actual poses, and parts are made to achieve more accurate part pose transformation through a machine learning algorithm; and a learning rule delta is adopted, and adaptive adjustment is performed on the step length Delta W, and therefore, target output can be prevented from falling into a locally optimal solution, and a result is stable, anda globally optimal solution can be obtained. The method has the advantages that the weight coefficient is modified in the negative gradient direction to reduce the error, the modification amount of each step is controlled according to the optimal value of the current distance, and the purposes of adjusting the learning rate in real time and accelerating convergence of target function solving areachieved. The problems tha