Integrated controlling system

• US 6,021,369 A
• Filed: 06/27/1997
• Issued: 02/01/2000
• Est. Priority Date: 06/27/1996
• Status: Expired due to Fees

• Current Assignee: Yamaha Hatsudoki Kabushiki Kaisha (Yamaha Corporation)
• Original Assignee: Yamaha Hatsudoki Kabushiki Kaisha (Yamaha Corporation)
• Inventors: Yamaguchi, Masashi, Kamihira, Ichikai
• Primary Examiner(s): Wolfe, Willis R.
• Assistant Examiner(s): Vo, Hieu T.
• Application Number:
  US08/884,210
• Time in Patent Office: 949 Days
• Field of Search: 701/102, 701/106, 701/57-60, 706/13-15, 706/20, 706/25, 706/31, 364/148.01, 364/149, 364/151, 364/152
• US Class Current: 701/102
• CPC Class Codes: F02D 41/1401 characterised by the control or regulation method F02D41/1473, F02D41/1477 take precedence
• CPC Class Codes: F02D 41/1404 Fuzzy logic control
• CPC Class Codes: F02D 41/1405 Neural network control
• CPC Class Codes: F02D 41/1406 with use of a optimisation method, e.g. iteration
• CPC Class Codes: F02D 41/2429 Methods of calibrating or learning
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: G05B 13/029 using neural networks and expert systems

Abstract

An integrated control for a subject such as an engine installed in a vehicle or vessel is conducted by the steps of: determining the characteristics of a user and/or using conditions; and changing characteristics of a control system of a subject in accordance with the determined characteristics. Normally, the control system includes: a reflection hierarchy for outputting a basic output; a learning hierarchy for learning and operation; and an evolutionary-adaptation hierarchy for selecting the most adaptable module. The subject is "trained" to suit the characteristics of the user and/or the using conditions.

Claims

1. A method for modifying a causative signal outputted from a control unit of a machine which is operable by the causative signal, based on an evaluation signal of operation of the machine, said control unit outputting the causative signal when receiving indicative signals indicating operation of the machine, said control unit comprising:

(i) a base value output unit to output a base value of a causative signal when receiving indicative signals;

(ii) an evaluation unit programmed to output an evaluation signal of operation of the machine when receiving pre-selected signals associated with operation of the machine; and

(iii) an evolution unit comprising an evolution neural network programmed to output and add a compensation value to the base value of the causative signal when receiving indicative signals, said evolution neural network provided with a predetermined genetic algorithm using, as genes, coupling coefficients regulating the relationship between output and input of the evolution neural network to select coupling coefficients based upon an evaluation signal of operation of the machine;

wherein the base value output unit and the evolution unit are placed in parallel to receive indicative signals and output causative signals to be combined,

said method comprising the steps of;

(a) activating the base value output unit to output a base value of the causative signal;

(b) activating the evolution neural network of the evolution unit to output and add a compensation value to the base value of the causative signal to operate the machine;

(c) activating the evaluation unit to output an evaluation signal of operation of the machine and input the evaluation signal into the evolution unit; and

(d) selecting coupling coefficients by the predetermined genetic algorithm using coupling coefficients as genes based upon the evaluation signal, while the machine is operated, thereby outputting and adding a compensation value to the base value of the causative signal to operate the machine by using the evolution unit with the selected coupling coefficients.

2. A method according to claim 1, wherein the control unit further comprises a learning unit comprising a neural network or cerebellar model arithmetic computer (CMAC) programmed to copy the output and input relationship of the evolution neural network, said method further comprising the steps of:

(e) after step (d), copying the output and input relationship of the evolution neural network by the neural network of the learning unit or CMAC; and

(f) using the neural network or CMAC to output and add a second compensation value to the base value of the causative signal upstream of the evolution unit and downstream of the base value output unit to operate the machine, when the neural network or CMCA receives indicative signals.

3. A method according to claim 2, wherein the learning unit further comprises a second neural network or CMAC having the same function as the first-mentioned neural network or CMAC, said method further comprising the steps of:

(g) after step (f), copying the output and input relationship of the sum of the evolution neural network and the first-mentioned neural network or CMAC by using the second neural network or CMAC of the learning unit;

(h) switching between the first-mentioned neural network or CMAC and the second neural network or CMAC to output and add a second compensation value to the base value of the causative signal upstream of the evolution unit and downstream of the base value output unit to operate the machine; and

(i) repeating steps (e) through (h), wherein the first-mentioned neural network or CMAC and the second neural network or CMAC are interchanged with each repetition.

4. A method according to claim 2, wherein the evolution neural network is activated at predetermined intervals.

5. A method according to claim 1, wherein the evolution unit further comprises at least another evolution neural network, each outputting a compensation value when receiving indicative signals at least one of which is different from that received by the first evolution neural network, wherein output of the at least another evolution neural network and output of the first evolution neural network are combined as output of the evolution unit.

6. A method according to claim 1, wherein the evaluation signal is derived from the pre-selected signals by using a predetermined neural network or a predetermined fuzzy rule.

7. A method for modifying a causative signal outputted from a control unit of a machine which is operable by the causative signal, based on an evaluation signal of operation of the machine, said control unit outputting the causative signal when receiving indicative signals indicating operation of the machine, said control unit comprising:

(i) a base value output unit to output a base value of a causative signal when receiving indicative signals;

(ii) an evaluation unit programmed to output an evaluation signal of operation of the machine when receiving pre-selected signals associated with operation of the machine; and

(iii) an evolution unit programmed to output and add a compensation value to the base value of the causative signal when receiving indicative signals, said evolution unit provided with a predetermined genetic algorithm using, as genes, control parameters regulating the relationship between output and input of the evolution unit to select control parameters based upon an evaluation signal of operation of the machine;

wherein the base value output unit and the evolution unit are placed in parallel to receive indicative signals and output causative signals to be combined,

said method comprising the steps of;

(a) activating the base value output unit to output a base value of the causative signal;

(b) activating the evolution unit to output and add a compensation value to the base value of the causative signal to operate the machine;

(c) activating the evaluation unit to output an evaluation signal of operation of the machine and input the evaluation signal into the evolution unit; and

(d) selecting control parameters by the predetermined genetic algorithm using control parameters as genes based upon the evaluation signal, while the machine is operated, thereby outputting and adding a compensation value to the base value of the causative signal to operate the machine by using the evolution unit with the selected control parameter.