Motor Control Device Using Neural Network Feed-Forward Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motor control devices face challenges in accurately controlling the position and velocity of controlled objects due to static friction and complex control systems, leading to inefficiencies in position and velocity deviations in machine tools and robots.
Innovation Solution
A motor control device that includes a position deviation computation portion, position feedback control, velocity deviation computation, velocity feedback control, and a velocity feed-forward control portion with a learning weight mechanism to adjust command currents based on detected deviations, using a perceptron or neural network configuration to optimize control signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a position reference model and velocity reference model are used for control, then control accuracy can be maintained in simple systems, but the system becomes extremely difficult to apply to controlled objects with static friction or complex nonlinear characteristics
Solution Approach 1:
The patent transforms the control approach by changing from model-based control to neural network-based control. The neural network learns optimal control parameters through training, adapting to the specific characteristics of the controlled object including static friction and nonlinear velocity characteristics, without requiring explicit mathematical models of the system dynamics
Solution Approach 2:
The patent replaces the traditional mechanical control system based on differential equations and mathematical models with an intelligent control system using neural networks. This substitution allows the system to handle complex nonlinear characteristics and static friction that are difficult to model mathematically, while maintaining control accuracy through learned parameters
2Device complexity
If only velocity feedback control is used, then the control system is simple, but position deviation and velocity deviation that significantly impact accuracy cannot be efficiently reduced
Solution Approach 1:
The patent implements dual feedback control by computing both position deviation (difference between command position and actual position) and velocity deviation (difference between command velocity and actual velocity). Both deviations are fed back to the neural network controller, which uses this information to generate appropriate current commands, efficiently reducing both position and velocity deviations simultaneously
Solution Approach 2:
The patent adds the velocity deviation dimension to the traditional position-based feedback control. By computing and utilizing velocity deviation in addition to position deviation, the control system gains an additional dimension of information that significantly improves accuracy in controlling both position and velocity, addressing the limitations of single-dimensional feedback
3Device complexity
If velocity feed-forward control receives only velocity command value, then the control structure is simple, but position deviation and velocity deviation cannot be efficiently reduced
Solution Approach 1:
The patent enhances the velocity feed-forward control portion to serve multiple functions: it receives not only the velocity command value but also the position command value and computed position deviation. This multi-functional feed-forward controller generates a velocity feed-forward output signal that compensates for both velocity and position errors, improving overall control accuracy without significantly increasing structural complexity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In a motor control device, a velocity feed-forward control portion (24) includes a velocity-side acceleration input portion that outputs received high-order command acceleration as a velocity-side acceleration output; a velocity-side velocity input portion that outputs a received high-order command velocity as a velocity-side velocity output; velocity-side boundary-velocity input portions which are prepared so as to respectively correspond to boundary velocities, and to output velocity-side boundary velocity outputs from the velocity-side boundary-velocity input portions corresponding to the high-order command velocity, the boundary velocities being velocities at boundaries of preset adjacent velocity ranges obtained by dividing a limited velocity range; a velocity-side first weight learning portion that changes velocity-side first learning weights in accordance with a velocity deviation, the velocity-side first learning weights respectively corresponding to velocity-side first outputs; and a velocity-side output portion that outputs, as a second tentative command current, a value obtained by summing velocity-side first multiplication values.