| Rapid numerical modeling of targets is one of the key issues in eddy current nondestructive testing.The Method of Moments(Mo M)discretizes the continuous equations into algebraic equations,and the algorithm strictly calculates the mutual coupling between each subsystem to ensure the overall minimum error system,so it can well simulate the eddy current nondestructive testing problem.However,when the number of unknowns is large and the target size is large,the Method of Moments will encounter the problem of high computational cost.Therefore,various fast algorithms are proposed to improve the computational efficiency of the Method of Moments.Among these fast algorithms,Adaptive Cross Approximation(ACA)algorithm and Kernel Degeneration(KD)algorithm have pure mathematical characteristics,so they can be easily combined with the code of the Method of Moments.In addition,the efficiency of the algorithm can be further improved through parallel processing.In this paper,the ACA algorithm and KD algorithm are used to accelerate the solution of the Method of Moments for eddy current nondestructive testing.The Stratton-Chu equation without lowfrequency collapse problem is selected.By introducing the hierarchical matrix(H matrix)method,the capacitive coefficient is set,and different H matrix structures are obtained to realize the regulation of the storage and calculation of the algorithm under different circumstances.Among the seven sub matrices generated by the Method of Moments,the Method of Moments is used to accurately solve the non-admissible blocks,and ACA algorithm or KD algorithm is used to jointly compress the admissible blocks,so as to improve the overall storage and calculation efficiency of the impedance matrix.Finally,by analyzing the numerical examples of single turn coil detecting three-dimensional metal ball and finite section coil detecting three-dimensional metal plate,the numerical results of ACA-KD algorithm are compared with the analytical method,semi analytical method,Mo M numerical results and ACA numerical results,so as to test the accuracy and efficiency of ACA-KD algorithm. |
