Parameters Optimization of Adaptive Cashew Shelling Cutter Based on BP Neural Network and Genetic Algorithm

Yun-Fei Fu, Jie Gong, Hui Huang, Yijun Liu, De-ming Zhu, Pengfei ZhaoAmerican 2015 · DOI

Abstract

This study aims to determine the optimal operational parameters for an adaptive cashew shelling cutter to maximize the whole-kernel rate, a critical metric for industrial profitability. By leveraging a hybrid approach of Back-Propagation (BP) Neural Networks and Genetic Algorithms (GA), the researchers modeled the non-linear relationship between mechanical settings and shelling outcomes. The optimized parameters—a cutter distance of 8.37 mm, spring pre-pressure of 141.56 N, and scraper velocity of 0.57 m/s—resulted in a whole-kernel rate of 0.0533 kg/min. Validation tests confirmed the model's accuracy, with error margins ranging from 2.25% to 13.7%.

Introduction

Cashew nuts are a high-value agricultural product, yet their mechanical processing is notoriously difficult. Unlike more uniform seeds, cashews possess a tough, kidney-shaped shell and contain Cashew Nut Shell Liquid (CNSL), a caustic substance that poses health risks to manual laborers. The primary challenge in mechanical shelling is "size variance": machines with fixed settings often crush smaller nuts or fail to crack larger ones, leading to high breakage rates and reduced market value.

To address this, researchers developed an adaptive shelling cutter designed to adjust to varying nut sizes. This study focuses on optimizing the machine's core parameters to maximize the whole-kernel rate, ensuring that the kernels remain intact and commercially valuable.

The Adaptive Shelling Mechanism

The proposed system features a spring-loaded upper blade assembly that allows for vertical movement, accommodating the natural size distribution of the nuts. The efficiency of this mechanism depends on three primary variables: the distance between the upper and lower cutters ($d$), the pre-pressure applied by the spring ($F_p$), and the velocity of the scraper ($v$) that moves the nuts through the assembly.

The mechanism of the adaptive cashew shelling cutter, highlighting the adjustable distance (d), spring pressure (Fp), and scraper velocity (v).

Optimization Methodology: BPNN and GA

Because the physics of shell-cracking is highly non-linear and influenced by biological variability, standard mathematical models often fall short. The researchers employed a two-stage computational strategy:

  1. BP Neural Network (BPNN): A neural network was trained on experimental data to act as a "digital twin" of the shelling process. It learned to predict the whole-kernel rate based on the three input variables ($d, F_p, v$).
  2. Genetic Algorithm (GA): Once the BPNN provided a reliable predictive model, the GA was used to search for the global maximum. It treated parameter sets as "chromosomes," evolving them through generations of selection, crossover, and mutation to find the peak performance settings.

Additionally, the cutter utilizes a specialized V-shaped "aided knife" designed to apply outward force on the shell, accelerating crack propagation and helping the kernel pop out without damage.

Detailed view of the cutter components: (1) Aided knife, (2) Main knife, and (3) Cutter body.

Experimental Results and Validation

The study was based on 60 physical shelling tests using "Class B" cashew nuts. The nuts were pre-treated by boiling for 30 minutes and air-drying for 48 hours to optimize shell brittleness. After training the BPNN on 50 data groups and validating with 10, the GA identified the following optimal parameters:

  • Distance between cutters ($d$): 8.37 mm
  • Spring pre-pressure ($F_p$): 141.56 N
  • Scraper velocity ($v$): 0.57 m/s

Under these conditions, the predicted whole-kernel rate reached 0.0533 kg/min. To verify these results, five additional physical trials were conducted. The actual performance closely matched the AI's predictions, with an error margin between 2.25% and 13.7%, which is considered highly successful given the inherent unpredictability of biological materials.

Conclusion and Future Work

The integration of AI-driven optimization with adaptive mechanical design provides a robust solution for the cashew processing industry. While the identified parameters serve as an excellent reference, the authors note that settings may need adjustment based on the specific moisture content and origin of different nut batches.

Future research will focus on developing automated feeding mechanisms to ensure consistent nut orientation and integrating real-time sensors that allow the machine to adjust its own pressure and speed dynamically during operation.

Summary updated

Abstract

The aim of this study is to determine the optimal parameters of the adaptive cashew shelling cutter. To meet the requirements of cashew nut processing enterprises, this study takes the whole-kernel rate as the optimization objective. Let the three main parameters of the adaptive cashew shelling cutter: The distance between upper and lower cutters, the pre-pressure of the spring and the velocity of the scraper be design variables. The BP neural network and genetic algorithm is used to find out the optimal parameters based on the limited shelling test data of cashew nuts. The optimal test result and its corresponding parameters are obtained by executing the BP neural network and genetic algorithm. The optimal whole-kernel rate is 0.0533 kg min 1 and the optimal parameters are the distance between upper and lower cutters of 8.37 mm, the prepressure of the spring of 141.56 N and the velocity of the scraper of 0.57 m sec 1 . To evaluate the accuracy of the predicted value, the shelling tests under the optimal parameters are carried out 5 times. The test results show that the error between the predicted value and actual value is in the range of 2.25 to 13.7%.
BibTeX
@inproceedings{fu2015parameters,
  author = {Yun-Fei Fu and Jie Gong and Hui Huang and Yijun Liu and De-ming Zhu and Pengfei Zhao},
  title = {{Parameters Optimization of Adaptive Cashew Shelling Cutter Based on BP Neural Network and Genetic Algorithm}},
  booktitle = {American},
  year = {2015},
  doi = {10.3844/ajeassp.2015.648.658},
}
Figure