Open Access

ARTICLE

Modeling and Optimization of the Shear Strength of Cassava Starch-Based Adhesives Using Artificial Intelligence Methods

Weixing Zhang, Chunxia He^*

College of Engineering, Nanjing Agricultural University, Nanjing, 210031, China

* Corresponding Author: Chunxia He. Email:

(This article belongs to the Special Issue: Renewable and Biosourced Adhesives-2021)

Journal of Renewable Materials 2022, 10(12), 3263-3283. https://doi.org/10.32604/jrm.2022.020516

Received 28 November 2021; Accepted 15 February 2022; Issue published 14 July 2022

Abstract

With the exponential growth of the computing power, machine learning techniques have been successfully used in various applications. This paper intended to predict and optimize the shear strength of single lap cassava starchbased adhesive joints for comparison with the application of artificial intelligence (AI) methods. The shear strength was firstly determined by the experiment with three independent experimental variables (starch content, NaOH concentration and reaction temperature). The analysis of range (ANORA) and analysis of variance (ANOVA) were applied to investigate the optimal combination and the significance of each factor for the shear strength based on the orthogonal experiment. The performance of all AI models was characterized by mean absolute error (MAE), root mean square error (RMSE) and regression coefficient (R² ) compared with the experimental ones. The GA-optimized ANN model was combined with the genetic algorithm (GA) to find the optimal combination of factors for the finalized optimized cassava starch adhesives (CSA-OP). The physicochemical properties of the cassava starch and CSA-OP were determined by the FTIR, TGA and SEM-EDS, respectively. The results showed that the numerical optimized condition of the GA-optimized ANN model was superior to the orthogonal experimental optimized condition. The sensitivity analysis revealed that the relative importance of variables was consistent with the results from ANOVA. FTIR results showed that there were high hydroxyl groups in cassava starch. TGA results showed that the residue of CSA-OP was higher than the cassava starch. SEM-EDS results showed that both the cassava starch and CSA-OP had abundant carbon and oxygen functional groups. Consequently, the obtained results revealed that the use of AI methods was an adequate approach to model and optimize the experimental variables of the shear strength of single lap cassava starch-based adhesive joints.

---

Keywords

---