基於振動信號之非破壞檢測於鳳梨質地量化指標研究

A Study of Nondestructive Testing Based on Vibration Data for Indexes of Pineapple Texture

葉勁輝(Yeh Chin-Hui)

指導教授 ： 林章生 ； 許志仲

國立屏東科技大學/工學院/車輛工程系所/碩士(2021年)

https://doi.org/10.6346/NPUST202100444

摘要

過去鳳梨的質地分級方法是根據農民經驗作為參考依據，因此判斷基準會因作業員的不同而有所差異。本研究係使用非破壞檢測之方法，進行鳳梨質地的估測。本文藉由振動平台的激勵信號，以及量測鳳梨的響應信號所計算得到的頻率響應函數分析鳳梨之動態特性，並訂定鳳梨質地之判別閥值，作為鼓聲果及肉聲果的分類依據。吾人基於混淆矩陣及馬修斯相關係數 (MCC)開發演算法進行評估，確認鳳梨質地之判別閥值。ROC曲線及其曲線下面積 (AUC)確認分類模型的鑑別力，建立鳳梨質地量化指標之技術。吾人利用主成分分析 (PCA)及線性判別分析 (LDA)對鳳梨數據矩陣進行降維，以及其特徵擷取加以識別鳳梨質地。由研究結果得知，鳳梨質地經由MCC判別與農民判別的結果有相當的一致性，並且LDA能有效地區分鳳梨質地。本研究提出的技術可應用於鳳梨質地識別，減少人為的誤判及成本，進而有效提升判別鳳梨質地的準確性。

關鍵字

鳳梨質地評估 ； 非破壞檢測 ； 頻率響應函數 ； 線性判別分析

並列摘要

In the past, pineapple texture grading was depended on farmer’s experience as a reference, so the criteria for judgment will vary from operator to operator. This study employs non-destructive testing methods to estimate the texture of pineapples. Based on the excitation signal of the vibration platform and the frequency response function calculated by measuring the response signal of the pineapple, we analyze the dynamic characteristics of the pineapple, and determine the threshold value of the pineapple texture as the classification basis for the hollow sound and solid sound fruits. Based on confusion matrix and Matthews Correlation Coefficient (MCC), the proposed algorithm is employed to confirm the discriminant threshold of pineapple texture. ROC curve and its area under the curve (AUC) confirm the discriminative power of the classification model, and establish a technique to quantify the texture of pineapple. Principal component analysis (PCA) and linear discriminant analysis (LDA) reduce the dimensionality of the pineapple data matrix, and extract its features to identify the texture of the pineapple. According to the research results, the pineapple texture has a certain consistency between the results of MCC discrimination and farmer discrimination, and LDA can effectively distinguish the pineapples texture. The proposed technique in this study can be applied to recognition of pineapple texture, reduction of human misjudgment and cost, and effective improvement of the accuracy of pineapple texture discrimination.

並列關鍵字

pineapple texture evaluation ； nondestructive testing ； frequency response function ； linear discriminant analysis

參考文獻

[1] Abbaszadeh et al, 2011, “Assessment of Watermelon Quality using Vibration Spectra,” Communications in Computer and Information Science Vol. 241, pp. 21-29.

[2] Abbaszadeh, R., Moosavian, A., 2013, “An Intelligent Procedure for Watermelon Ripeness Detection Based on Vibration Signals,” J Food Sci Technol, Vol. 52, pp. 1075-1081.

[3] Ballabio, D., Grisoni, F., and Todeschini, R., 2018, “Multivariate Comparison of Classification Performance Measures,” Chemometrics and Intelligent Laboratory Systems, Vol. 174, pp. 33-44.

[4] Bekkar, M., Djemaa, H. K., and Alitouche, T. A., 2013, “Evaluation Measures for Models Assessment over Imbalanced Data Sets,” Journal of Information Engineering and Applications, Vol. 3, No. 10, pp.27-39.

[5] Boughorbel, S., Jarray, F., and EI-Anbari, M., 2017, “Optimal Classifier for Imbalanced Data using Matthews Correlation Coefficient Metric,” PLOS ONE., Vol. 12, No. 6, pp. 1-17.