Mechanical exaptation of existing drive signals for sustainable sensorless diagnostics in extrusion machinery
Year: 2026
Type: article
Abstract: Existing extrusion drives continuously generate electrical quantities for power transmission, control, protection, and energy accounting. This study proposes informational mechanical exaptation as a conceptual engineering framework and operational admissibility guideline for deciding when retained drive signals may acquire an additional diagnostic function. The contribution is not a new sensor, classifier, or physical principle, but a mechanics-based decision capability that separates defensible signal co-option from ordinary monitoring and unsupported post hoc signal reuse. The framework requires an existing signal source, a genuinely additional function, preservation of the original machine function, a physically defensible condition-signal pathway, task-specific diagnostic validation, and evidence of lifecycle relevance before deployment claims are made. The mechanical basis is expressed through an electromechanical model linking screw resistance, shaft torque, angular velocity, phase current, active power, and non-stationary signal structure. A 7.5 kW single-screw wheat extruder served as the demonstration system. Previously validated current-based continuous-wavelet-transform scalograms distinguished new and worn screw states with 92.3% accuracy. A second dataset comprising normal extrusion and a documented clogging event tested whether the same retained signal could support disturbance diagnosis. At 500 rpm, measured active power corresponded to input-power-equivalent torque indicator values differing by only 2.27%. Across non-overlapping 20 s windows, univariate effects were small, all Mann-Whitney tests were non-significant, and multivariate separation was weak. The admissibility test therefore accepted screw-wear inference within its validated operating domain but rejected scalar-power clogging inference. The study demonstrates a procedure for qualifying an additional diagnostic function in a specific machine context; broader sustainability and industrial applicability require further validation, avoided-intervention evidence, and lifecycle assessment.
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7203891991