Artificial Intelligence Method For Calculating Mobility Of Parallel Mechanism

As the most basic parameter of the mechanism,mobility plays an important role in kinematics analysis,drive configuration and control of the mechanism.The research on the mobility of mechanism can be divided into the number of mobility and the attributes of mobility.The calculating of the parallel mechanism mobility,which is a complex space mechanism,is always a difficult point in the study of mechanism.Based on artificial intelligence method,this dissertation completed the calculation of the parallel mechanism's mobility through sample feature extraction,data preprocessing,algorithm model learning and model evaluation test of symmetric parallel mechanism.This dissertation's man content is as follows:By analyzing the mobility of the parallel mechanism calculated by the screw theory,the mechanism characteristics that affect the mobility of the parallel mechanism are studied and converted into mathematical expressions.After analyzing the constraint relation between and within the branches of parallel mechanism,two characteristic matrices of symmetric parallel mechanism are defined,and the sample collection is carried out in combination with the existing configuration comprehensive literatureXGboost algorithm is used to analyze the mobility of the mechanism of parallel mechanism.By preprocessing the sample data,establishing the algorithm model and setting up the experimental environment for training the mobility algorithm model of parallel mechanism,the accuracy of the algorithm model was finally obtained through parameter adjustment and optimization,which is 90.76%,and the test set was used for verification.Recurrent neural network algorithm and convolution neural network algorithm are used to calculate the mobility of parallel mechanism.First,the sample data was preprocessed,then the data was merged and filled,and then the samples were processed out of order to prevent over-fitting in the algorithm training.Finally,the established experimental environment was used for model learning and testing.The recurrent neural network model accuracy is 96.41%,and the convolution neural network's accuracy is 95.93%,and then using the corresponding test set to verify the validity of the algorithm model.The analysis shows that the deep learning method is more accurate and effective than the machine learning methodThrough comparative analysis of three algorithms in terms of feature data processing,algorithm time complexity,algorithm space complexity and model training accuracy,it is concluded that the recurrent neural network algorithm performs well in feature processing interpretation,time and space complexity and model training accuracy.Using the artificial intelligence method to analyze the mobility of the parallel mechanism improves the calculation efficiency,and a new idea for the synthesis of parallel mechanism configuration is proposed.