Electric Rotating Machine Simulation from Rating Plate Data

The patent applies preliminary action by pre-training a neural network function using comprehensive measurement data from multiple machines before actual use. The training phase (performed once) collects idle measurements, rotating measurements under no load, and rotating measurements under load to establish accurate parameter relationships. During actual operation, the pre-trained function requires only simple input data (shaft height, number of pole pairs, rating plate specifications) to generate accurate simulation models, eliminating the need for repeated expensive measurements.

The patent uses copying by creating a virtual model (simulation model) of the electric rotating machine through the trained neural network function. Instead of performing physical measurements on each machine, the system copies the behavior characteristics from training data to generate accurate parameter estimates. The trained function acts as a digital twin that replicates the machine's electrical and mechanical behavior based on minimal input data.

The patent performs comprehensive measurements during the preliminary training phase to establish accurate parameter relationships. The neural network is trained once using extensive measurement data, and this pre-acquired knowledge is then rapidly applied to determine operating behavior for any machine of the same type. This eliminates the need for time-consuming measurements during each operational assessment.

The patent replaces the mechanical measurement system with an information processing system. Instead of physically connecting sensors and measurement equipment to the machine, the system uses a trained neural network function that processes electrical input data (shaft height, pole pairs, rating plate specifications) to generate accurate operating behavior parameters. This substitution dramatically reduces measurement time while maintaining reliability.

The patent replaces simple interpolation methods with an advanced information processing approach using a trained neural network function. The system maintains ease of operation by requiring only basic input data (shaft height, number of pole pairs, rating plate specifications) while achieving high measurement precision through the trained function that processes these inputs to generate accurate simulation model parameters and operating behavior data.

The patent transforms the approach from direct physical measurement to parameter-based calculation. The trained neural network function takes basic geometric and rating parameters as input and transforms them into accurate operating behavior parameters (efficiency, power factor, torque, speed) through learned relationships from training data, achieving both simplicity and accuracy.

The patent creates a complete information set by copying behavioral characteristics from training data to the target machine through the trained neural network function. The function generates comprehensive operating behavior information (efficiency, power factor, torque, speed characteristics) by replicating the relationships learned during training, providing complete specifications without requiring complex measurement systems.

The patent replaces complex measurement systems with an information processing system that uses the trained neural network function. The system processes basic input parameters through the trained function to generate complete detailed specifications, eliminating the need for expensive sensors and measurement equipment while achieving full information completeness.