Artificial Intelligence-Assisted Linear Electromagnetic Inertia System for Industrial Microgrid Stability During Network Disturbances

Abstract

Short voltage interruptions and deep voltage sags can trip variable-speed drives before conventional standby generation becomes effective. This paper develops a loss-aware conceptual design for a linear electromagnetic inertia system (LEIS) intended only for sub-second-to-few-second ride-through of selected industrial loads. The selected actuator is a permanent-magnet linear synchronous machine (PMLSM) with a guided translator, a bidirectional converter, and a constrained kinetic-energy reserve. The term electromagnetic inertia is used as a system name and not as an assertion that magnetic-field energy, mechanical inertia, and power-system inertia are physically interchangeable. A unified electromechanical model includes copper, core, eddy-current, guide-friction, converter, and auxiliary losses, together with translator travel and current limits.

The distinguishing contribution is the coordinated combination of a travel-limited linear kinetic store, one reversible PMLSM/converter path, a safety-gated disturbance predictor, and explicit ride-through energy and stopping-distance constraints for industrial microgrids. Artificial intelligence is restricted to anticipatory supervision; deterministic protection retains authority and initiates support if the predictor is late, unavailable, or uncertain. A 50 kW, 2 s analytical sizing example shows that losses and reserve margins materially increase the required kinetic energy and translator speed. No experimental performance or prediction-accuracy claim is made. The paper specifies the data, metrics, latency budget, edge cases, and hardware-in-the-loop tests required before industrial deployment.