Neural Machine Models for Electric Machine Control Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for modeling the operation of machine tools with complex control devices are expensive and complicated, making it difficult to obtain accurate information about the control devices used in these machines.
Innovation Solution
A method using artificial neural networks to train models based on temporal series of measured values from machine operations, with the option of extending trained models through transfer learning to adapt to similar machines, allowing for the derivation of control variables and simulation of machine operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning methods are used to reproduce operation of the electric machine controlled with the control device, then accurate information about the control device can be obtained, but the training of the model becomes very expensive and complicated
Solution Approach 1:
The patent creates a virtual copy (digital twin) of the physical machine and its control device through neural network modeling. Instead of directly analyzing the complex control device, a replicated virtual model is trained to mimic the control device's behavior, making it easier to extract information while maintaining accuracy.
Solution Approach 2:
The neural network model serves as an intermediary between the observable machine operations and the hidden control device logic. Rather than directly accessing the complex control device internals, the model acts as a mediator that learns the input-output relationships and enables indirect analysis of control device behavior.
2Measurement precision
If a model is trained on measured values from one machine, then accurate modeling of that specific machine is achieved, but the model cannot be easily applied to similar machines without extensive retraining
Solution Approach 1:
The patent designs the neural network model with universal applicability across multiple machines of the same type. By training on aggregated data from multiple machines and incorporating transfer learning capabilities, the model can be deployed to individual machines with minimal retraining, serving multiple functions across different physical instances.
Solution Approach 2:
The model allows for parameter adaptation when moving between different machines. Instead of complete retraining, the system adjusts specific parameters and weights of the neural network to accommodate variations between machines, enabling flexible deployment across similar equipment while maintaining accuracy.
Data Source
AI summary
In a method for training a model for an electric machine controlled by a control device, a temporal series of measured values that describe an operating variable of the electric machine is received by a training system. An untrained model embodied as an artificial neural network is then trained with the received measured values to produce a trained model. Control variables that describe the control device are determined with the trained model. The training system then receives a temporal series of measured values of a further electric machine that is different from the electric machine and controlled by a further control device. The trained model is then trained further with the computing facility using measured values of the further electric machine to produce a further trained model. The trained model and the further trained model is outputted via a second interface of the training system.