本文建立了一种基于仿生技术的榨菜脆性检测方法。模拟了人类专家对榨菜脆性感官评定的过程，设计制作了由仿生咀嚼装置、仿牙周膜压力传感器、仿耳膜声音传感器和测试软件等组成的仿生脆性检测系统，使用该系统对五种榨菜样品进行了脆性测试，提取脆裂力信号和脆裂声音信号特征值，并与脆性感官评分进行相关性分析和多元回归分析，建立榨菜脆性预测方程。结果表明：仿生技术检测方法测得的最大值、峰谷差、平均差和幅值差与感官评分达到了极显著的相关性，相关系数分别为0.997、0.993、0.975、0.968；榨菜脆性预测方程的预测平均相对误差小于10%，RSD值小于2.0%，且T检验的显著性水平大于0.05，预测值与实测值无明显差异。本方法准确、客观、快速，可以替代人类专家进行榨菜脆性检测。

A method was established for the determination of brittleness of mustard, based on the bionic technique. The process by which human experts conduct sensory evaluations of the brittleness of mustard was simulated, and a bionic system consisting of a bionic chewing device, a bionic periodontal ligament pressure sensor, a bionic eardrum sound sensor, and test software to measure the brittleness was designed and manufactured. This system was used to test the brittleness of five different mustard samples, and the characteristic values of the embrittling force signals and embrittling sound signals were determined and used for correlation analysis and multiple regression analysis of the scores of sensory evaluation on brittleness, to establish a prediction model for the brittleness of mustard. The results showed significant correlations between sensory evaluation scores and the values measured by the bionic technique, including the maximum value, peak-trough difference, mean difference, and amplitude difference, with correlation coefficients of 0.997, 0.993, 0.975, and 0.968, respectively. The average relative error between the prediction equation and sensory evaluation was less than 10%, and the relative standard deviation was less than 2%. Besiseds, the significance level was greater than 0.05 by t-tests. There were no significant differences between the predicted values and measured values. This method was accurate, objective, and rapid, might be used for replacing human experts in the analysis of mustard brittleness.

重庆市科委基础与前沿研究项目(cstc2013jcyjA0098)；涪陵区科委科技项目( FLKJ,2013ABB2079)；2013年重庆高校创新团队建设计划资助项目(KJTD201322)；三峡库区特色作物工程研究中心资助项目。