AI Algorithms – Smart Optimization for Production
This project applies Artificial Intelligence (AI) to a real-world industrial problem: optimizing how biscuits of different shapes and sizes are placed on a dough roll to minimize waste and maximize production value. Instead of manually testing every possible combination, the AI algorithm learns to make better placements through iterative improvements.
How AI Was Used
AI techniques were used to solve this optimization problem intelligently. Two algorithms were implemented:
- Local Search : starts from a random layout and gradually improves it by learning which biscuit placements increase total value.
- Constraint Satisfaction Problem (CSP) : explores all possible placements that respect conditions like no overlapping, integer positions, and limited defects, then uses heuristics to find better solutions.
These methods simulate how AI systems make intelligent decisions : exploring, evaluating, and improving solutions automatically to reach near-optimal outcomes.
Results
The AI algorithms were evaluated on total value (optimization quality) and execution time. The improved CSP model, enhanced with heuristics, outperformed the original version and the faster Local Search.
| Algorithm | Total Value | Execution Time | Remarks |
|---|---|---|---|
| Old CSP | 715.0 | 0.23s | Baseline model — faster but less optimal. |
| Improved CSP (Heuristics) | 752.0 | 0.44s | Higher total value after AI-based heuristic tuning. |
| Local Search | 703.0 | Very fast | Quick and efficient — ideal when time is key. |
Summary: AI improved production efficiency by finding smarter biscuit arrangements automatically. The Improved CSP achieved the best overall result, reaching a higher total value (752.0), showing how AI-driven optimization can outperform traditional methods.