<p>Pulse eddy current can be used for corrosion testing of coated pipelines due to its instantaneous high-energy magnetic field and wide frequency spectrum. However, due to the unevenness of the coating layer, the linear mapping method used in traditional pulse eddy current testing cannot eliminate lift-off interference, which seriously affects the accuracy of defect assessment. This paper proposes a multi-characteristic signal (SP,SG) fusion method based on the uncertainty problem of lift-off values, and uses the sparrow search algorithm to optimize the BP neural network (SSA-BP) to find and solve the complex nonlinear relationship between multi-characteristic signals and thickness, thereby eliminating lift-off interference in defect evaluation. This paper provides theoretical analysis and verification of multi-characteristic signals and SSA-BP algorithm, and then compares the detection performance of traditional linear mapping method and multi-characteristic signal fusion method based on SSA-BP algorithm through simulation and experimental results. The results show that the multi-characteristic signal fusion method based on SSA-BP algorithm has better evaluation ability than traditional linear mapping method, and can effectively eliminate the influence of unknown lift-off values. The predictive performance of SSA-BP algorithm is superior to that of Particle Swarm Optimization algorithm to optimize the BP neural network (PSO-BP) and Genetic Algorithm to optimize the BP neural network (GA-BP).This paper contributes to optimizing the application of pulse eddy current testing technology in defect detection of coated pipelines, improving the accuracy and efficiency of detection.</p>