关键词: B柱加强板, 过渡特性, 表面硬度, 磁巴克豪森噪声, BP神经网络

Abstract: The conventional method to evaluate the transition characteristics of the hard-to-soft part of B-pillar reinforcement plate used in automobile is to test the local hardness by destructive sampling, which is not convenient and efficient for the parts directly, and is time-consuming and costly. Hence, magnetic Barkhausen noise measurement method is investigated for nondestructive evaluating the transition characteristics to overcome the shortcomings of traditional methods. Firstly, the self-developed multi-functional micro-magnetic instrument is used to carry out calibration experiments. Then, the estimation method of the position where the transition zone begins and ends and length of the transition zone is proposed. Subsequently, a quantitative prediction model of surface hardness in transition region is established. The results show that most of the feature parameters of MBN and TMF can effectively evaluate the transition range of hard-to-soft part. The estimation error for the starting point is ±14 mm, and the relative error of length estimation is less than 10% using parameters of x₁, x₅ and x₁₇. The RMSE of linear characterization model with parameter of x₁₇ is about 12.38 HV10, while BP neural network model established by integrating multiple magnetic parameters has a higher prediction accuracy with a RMSE value less than 6 HV10. Obviously, a micro-magnetic evaluation method for the transition characteristics of the hard-to-soft part of key automotive components is proposed, in which the relationship between feature parameters from magnetic Barkhausen noise and hardness is studied and the quantitative characterization of surface hardness and transition range of hard-to-soft part of B-pillar is realized. The method has broad prospects in engineering application.

Key words: B-pillar, transition characteristics, surface hardness, magnetic Barkhausen noise, BP neural network