Soft-Computing-Based Estimation of a Static Load for an Overhead Crane

Area of Science:

• Engineering
• Artificial Intelligence
• Industrial Automation

Background:

• Payload weight detection is crucial for crane condition monitoring and automation.
• Traditional load cells can be costly or impossible to install on hoisting equipment.
• Indirect methods using existing sensors offer a viable alternative for payload estimation.

Purpose of the Study:

• To estimate static payload weight indirectly using crane girder strain and trolley position.
• To apply soft-computing techniques for deriving nonlinear input-output relationships.
• To compare the performance of different data-driven identification methods.

Main Methods:

• Utilized grammar-guided genetic programming with sparse regression.
• Employed multi-gene genetic programming.
• Applied subtractive fuzzy clustering with the least squares algorithm.
• Used experimental data from a laboratory overhead crane.

Main Results:

• Multi-gene genetic programming and grammar-guided genetic programming with sparse regression showed similar, high accuracy.
• Both genetic programming variants outperformed subtractive fuzzy clustering.
• The novel grammar-guided approach yielded a more parsimonious model with lower execution time.

Conclusions:

• Soft-computing techniques, particularly genetic programming variants, effectively estimate crane payload weight indirectly.
• These data-driven methods provide accurate and efficient alternatives to traditional load cells.
• The developed models offer practical solutions for crane automation and condition monitoring.