Non-Destructive Detection of Fillet Fish Quality Using MQ135  Gas Sensor and Neutrosophic Logic-Enhanced System

M. Y. Shams; M. R. Darwesh; Roheet Bhatnagar; N. S. A. Al-Sattary; A. A. Salama; M. S. Ghoname

Authors

M. Y. Shams Deptartment of Machine Learning, Faculty of Artificial Intelligence, Kafrelsheikh University, Kafrelsheihk 33516, Egypt;
M. R. Darwesh Department of Agriculture Engineering, Faculty of Agriculture, Tanta University, Tanta 31527, Egypt;
Roheet Bhatnagar Department of Computer Science and Engineering Manipal University Jaipur, Rajasthan, India,
N. S. A. Al-Sattary Department of Agriculture Engineering, Faculty of Agriculture, Tanta University, Tanta 31527, Egypt;
A. A. Salama Department of Mathematics and Computer Science, Faculty of Science, Port Said University, Egypt,
M. S. Ghoname Department of Agriculture Engineering, Faculty of Agriculture, Tanta University, Tanta 31527, Egypt;

Keywords:

Adaptive Neuro-Fuzzy Inference System, Neural Networks, Fuzzy Logic, Neutrosophic Logic.

Abstract

This paper demonstrates the feasibility of using an electronic nose to assess fish quality
by analyzing air quality and examining volatile organic compounds (VOCs) alongside physical
variables, with pH, protein content, and VOCs serving as chemical reference points. Artificial
intelligence algorithms were employed to predict quality and calculate regression coefficients.
Using the reference neural network algorithm based on chemical and physical compounds,
regression coefficients (R values) achieved were 0.99, 0.98, and 0.97, respectively. Additionally,
ANFIS (Adaptive Neuro-Fuzzy Inference System) produced R values of 0.99, 0.85, and 0.98. Both
fuzzy logic and ANFIS proved effective for predicting fish quality. Image processing techniques,
including histogram analysis, color mapping, and edge detection, were also applied to assess fish
quality. To enhance the inference process, Neutrosophic Logic-Enhanced Fuzzy Logic Systems were
utilized, addressing uncertainty and imprecision in fish quality assessment. Neutrosophic logic
combines fuzzy logic's partial truth with indeterminacy, represented by three membership
functions: truth, indeterminacy, and falsity. Neutrosophic fuzzy inference integrates steps like
neutrosophication, rule evaluation, aggregation, and defuzzification, ensuring improved
expressiveness and fidelity. For instance, neutrosophic fuzzy rules evaluated fish freshness and
appearance to determine quality ratings such as poor, good, or excellent. This integration enhances
decision-making by accurately modeling complex real-world uncertainties. These methods,
combining electronic nose technology, artificial intelligence, and neutrosophic inference, provide a
robust, non-destructive, and cost-effective approach to detecting spoilage in fillet fish.