DEVICE FOR MEASURING MILL CHARACTERISTIC

Japanese Patent JP3565380

B2

PROBLEM TO BE SOLVED: To make it possible to learn a parameter for a simulation module during the driving of a mill and to predict a change of a transition state from a change of a steady state.
SOLUTION: A data sorting means 10 sorts data into steady state data and transition state data and outputs sorted data and a featured value arithmetic means 11 inputs steadly state I/O data and calculates the feature value of the steady state. A simulation model 12 inputs transition state input data and outputs transition state output data YS based upon a parameter. An error computing element 13 calculates an error (e) to be a difference from the data YT. A neural network 16 inputs the feature value of the steady state and outputs a parameter for the model 12, but when the error (e) is larger than a prescribed value, an error judging means 15 outputs a learning instruction to the neural network 16 so that the error (e) reaches zero.

JP3836028	Roller mill classifier
JPH0636871	[Title of Invention] Vertical crusher
JPH0347543	GRINDING SURFACE OF ROLLING MILL

Shiromaru Isao
Hideki Kidori
Shigeki Murayama

JP1972696A

September 15, 2004

February 06, 1996

Click for automatic bibliography generation  

Chugoku Electric Power Co., Inc.
Ishikawajima Harima Heavy Industries Co., Ltd.

B02C15/00; B02C25/00; G05B13/02; G05B13/04; G06F19/00; G06N3/02; G06Q10/04; G06Q50/00; G06Q50/06; G06F17/00; (IPC1-7): G05B13/04; B02C15/00; B02C25/00; G05B13/02; G06N3/02

JP5001928A
JP7105177A
JP6035893A
JP6266408A
JP7121206A
JP3216778A

Minoru Hotta
Shigeru Nara

Previous Patent: DRIVE CONTROL UNIT FOR TRACTOR

Next Patent: SEPARATION OF 1,1,1,3,3-PENTAFLUOROPROPANE