Magnetic Conveyor Actuation Using Neural Network Setpoint Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Regulating the movement of objects using magnetic conveyor devices is complex and costly due to the need for precise control of magnetic forces, requiring significant computing power and time.
Innovation Solution
A method and device utilizing an artificial neural network to determine setpoint values for actuators based on actual and setpoint values for pose, torque, and force, allowing for efficient activation of actuators, such as permanent magnets or electromagnets, to generate magnetic fields that move objects with reduced computing effort and faster control cycles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex calculation methods are used to determine actuator setpoint values, then precision of control is improved, but computing time and computing power requirements increase
Solution Approach 1:
A neural network model is trained in advance with simulation data to learn the mapping between desired device poses and required actuator setpoint values. During actual operation, the pre-trained model directly outputs setpoint values without requiring complex real-time calculations, thus achieving both high precision and fast response.
Solution Approach 2:
The complex physical system is replaced by a simplified computational model (neural network) that has been trained to replicate the system's behavior. This virtual copy enables fast inference without needing to solve complex physical equations in real-time.
2Measurement precision
If complex calculation methods are used to determine actuator setpoint values, then control accuracy is improved, but device complexity increases
Solution Approach 1:
Complex mechanical/computational calculation systems are replaced by a trained neural network model. The model performs inference through simple forward propagation operations rather than solving complex equations, significantly reducing computational complexity while maintaining control accuracy.
3Productivity
If traditional control methods are used, then system robustness is maintained, but productivity and control speed decrease
Solution Approach 1:
The neural network model is trained offline with extensive simulation data covering various operating conditions and disturbances. This preliminary training enables the model to generalize well to unseen situations, maintaining robustness while achieving fast real-time control during operation.
Solution Approach 2:
The system uses actual pose measurements as feedback to continuously determine appropriate actuator setpoints. The closed-loop control structure with the trained model maintains system robustness by adapting to actual system state while achieving fast response times.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution enables quick and efficient determination of setpoint values for actuator activation, reducing computational complexity and enabling precise control of object movement with minimal computing power, thus improving control quality and reducing costs.
Implementation Method 1
The trained model, for example, an artificial neural network, maps the actual values for the pose, the setpoint values for the torque, and the setpoint values for the force efficiently on a particular setpoint value for the actuators
Implementation Method 2
Magnetic forces may be used to cause objects, which are influenceable with the aid of magnetic force action, to float or move
Implementation Method 3
The at least one actuator preferably includes at least one permanent magnet, in particular a Halbach array, by which a magnetic field is generatable to influence a movement of the device
Implementation Method 4
the at least one actuator includes an electromagnetic element by which a magnetic field is generatable to move the device
Implementation Method 5
A magnetic field for moving the device is preferably generatable by a superposition of magnetic fields of the at least one permanent magnet of a plurality of different actuators with magnetic fields of at least one electromagnet or permanent magnet at the device
Data Source
AI summary
A device and method for activating a conveyor device. An actual value for a pose of a device movable by the conveyor device by a magnetic force action is received. Depending on the actual value for the pose, as a function of a setpoint value for a torque, as a function of a setpoint value for a force, and as a function of a model, a setpoint value for the activation of at least one actuator of the conveyor device is determined. The model is trained to determine setpoint values for the activation of the at least one actuator as a function of actual values for poses of the device and as a function of setpoint values for torques and setpoint values for forces, using which the device is to be moved.