MATLAB-Data-Model

A comprehensive MATLAB repository containing data models, optimization algorithms, and deep learning architectures tailored for textile engineering applications. This project focuses on optimizing cotton blending processes, predicting yarn quality attributes, and performing parameter inversions for spinning systems.

---

📂 Repository Structure

MATLAB-Data-Model/
├── 1. Automatic cotton blending/          # Optimization frameworks for raw material allocation
│   ├── peimian_GA/                        # Genetic Algorithm implementation
│   ├── peimian_NSGAII/                    # Non-dominated Sorting Genetic Algorithm II (Multi-objective)
│   ├── peimian_PSO_GSO/                   # Particle Swarm & Glowworm Swarm Optimization hybrid
│   └── peimian_yuan/                      # Source baseline datasets & scripts
│
├── 2. Inversion of spinning process parameters/  # Parameter inversion using deep architectures
│   ├── CNN_BP/                            # CNN feature extraction mapped to Backpropagation network
│   ├── CNN_LSTM_IGA/                      # CNN-LSTM temporal network optimized by Improved GA
│   ├── LSTM/                              # Standard Long Short-Term Memory recurrent model
│   ├── Multiout_CNN_LSTM_IGA/             # Multi-output CNN-LSTM structure with IGA optimization
│   └── Multiprocess_LSTM/                 # Parallel multi-process LSTM framework
│
└── 3. Yarn quality prediction/            # Machine learning regressors for textile metrics
    ├── mixed_kernel_SVR/                  # Support Vector Regression with custom hybrid kernels
    ├── Multi_kernel_SVR/                  # Multiple Kernel Learning (MKL) for SVR
    ├── PSO_GA_BP/                         # BP Neural Network optimized via PSO and GA
    ├── PSO_SVR/                           # Support Vector Regression optimized via Particle Swarm
    ├── Research_Drawing.m                 # Main data visualization and plotting routine
    └── spider_plot.m                      # Custom radar/spider chart generation script

---

🚀 Core Research Modules

1️⃣ Automatic Cotton Blending

This module provides meta-heuristic optimization algorithms to automate the cotton "peimian" (material blending) process. It balances properties like fiber length, micronaire, and strength while minimizing production costs.

• GA & NSGA-II: Implements single and multi-objective evolutionary strategies to select the optimal combination of raw cotton bales from warehouse inventory.
• PSO-GSO: Combines particle swarm dynamics with glowworm swarm visibility behaviors to bypass local optima in non-linear constraints.

2️⃣ Inversion of Spinning Process Parameters

Reverses the manufacturing pipeline by taking desired yarn properties and inferring the structural spinning process parameters (e.g., draft ratios, spindle speeds, traveler weights).

• CNN-LSTM-IGA: Employs Convolutional Neural Networks for spatial spatial pattern identification from fiber arrays, combined with LSTMs for consecutive drafting zone tracking. The entire pipeline is fine-tuned via an Improved Genetic Algorithm (IGA).
• Multi-output Architectures: Solves complex engineering trade-offs by simultaneously predicting multiple system variables.

3️⃣ Yarn Quality Prediction

Predicts critical mechanical and physical properties of the spun yarn (such as breaking tenacity, evenness CV%, and hairiness) based on raw material inputs and mechanical status data.

• MKL-SVR: Utilizes Multi-Kernel Support Vector Regression to map highly non-linear, heterogeneous fiber properties into higher-dimensional spaces for high-accuracy regression.
• Visualization Tools: Includes native spider_plot.m functions to produce qualitative radar charts comparing predicted properties against experimental baselines.

🛠️ Requirements & Quick Start

• Environment: MATLAB (R2021a or newer recommended).

• Toolboxes Required:

• Global Optimization Toolbox

• Statistics and Machine Learning Toolbox

• Deep Learning Toolbox

• Running an Optimization Model

• Open MATLAB and navigate to the directory of interest:

cd('1. Automatic cotton blending/peimian_NSGAII')

• Execute the main entry file or optimization script to view the Pareto front.
• For data visualization pipelines, execute the data rendering engine:

run('3. Yarn quality prediction/Research_Drawing.m')

🤝 Contribution & Citation

Contributions making these models more robust or adding new optimization layers are welcome. Please open an issue or fork this repository to submit a pull request.