Title:
CONTROL SYSTEM USING NEURAL NETWORK
Document Type and Number:
Japanese Patent JP3478855
Kind Code:
B2
Abstract:
PURPOSE: To provide a combine control system using a neural network.
CONSTITUTION: The neural network constituted of a computer 64 is given learning using a common data set to function as the common model of a harvest process and next given learning using extension data set including the common data set featured by setting a local condition input to be a 'zero value' and a data set concerning an individual local condition featured by setting the local condition input to be a 'non-zero value'. When the local condition input is the 'zero value', a learned neural network functions as the common model, and when the local condition input coincides with the code of one data set among local data sets included in learning data, the learned neural network functions as a local model corresponding to it. The learned neural network is combined with a sensor, an actuator, a computer 62 for control and information transmission and a user interface to make a combine control system.
Inventors:
James William Hall
Application Number:
JP22610593A
Publication Date:
December 15, 2003
Filing Date:
September 10, 1993
Export Citation:
Assignee:
DEERE AND COMPANY
International Classes:
G06F15/18; A01D41/127; G05B13/02; G06G7/60; G06N3/00; G06N99/00; (IPC1-7): G06G7/60; G06N3/00
Domestic Patent References:
| JP1232464A | ||||
| JP421153A |
Attorney, Agent or Firm:
Kazuo Shamoto (4 outside)