“Significant breakthroughs have been made in the field of identifying the dynamic models of robotic arms. The research team proposed an innovative semi parametric dynamic model identification method for the precise identification of ultra redundant robotic arm dynamic models. This method combines iterative optimization and neural network compensation, effectively improving the accuracy and efficiency of parameter identification. By optimizing the regression matrix condition number through genetic algorithm to generate excitation trajectories, a joint nonlinear friction model was established, further improving the physical feasibility of the model. The experimental results show that compared with traditional algorithms, this method significantly reduces the root mean square value of joint identification torque residuals, verifying its effectiveness and superiority. This study not only provides strong support for precise control of ultra redundant robotic arms, but also opens up new directions for research in the field of dynamic model identification of robotic arms.”