Industrial WinCC Shearer Motor Records and Leakage Controlled Six Level Load Pattern Analysis
Description
This dataset supports the article “Intelligent Recognition of Shearer Cutting Patterns: A CPSO Optimized SVM Strategy”. The deposit contains 7,048 timestamped WinCC records acquired during a real industrial shearer operating campaign, 2,823 curated observations, and an exact mapping from every curated observation to its source raw row. Six commanded load classes represent 0%, 20%, 40%, 60%, 80%, and 100% of rated load. The 640 constant speed observations were used exclusively for model development. The complete sinusoidal profile (n = 723) and sawtooth profile (n = 1,460) were reserved for field evaluation. The deposit includes raw and curated data tables, provenance audits, analysis roles, sample level predictions, aggregate performance metrics, uncertainty estimates, repeated optimizer runs, statistical tests, publication figures, and executable Python code. All records belong to one equipment campaign. Current and voltage are retained in the source logger scales because traceable engineering unit calibration metadata were not preserved. The dataset does not establish transfer across machines or mines or prospective operational performance.