Data processing: generic control systems or specific application – Generic control system – apparatus or process – Optimization or adaptive control
Reexamination Certificate
2000-03-02
2003-04-15
Von Buhr, Maria N. (Department: 2125)
Data processing: generic control systems or specific application
Generic control system, apparatus or process
Optimization or adaptive control
C700S108000, C700S065000
Reexamination Certificate
active
06549815
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method and an apparatus for optimizing a control module for controlling a controlled system, and a cooperative method for optimization in a method for optimizing a characteristic of the control module.
2. Description of the Related Art
In the past, optimal values of a characteristic of a control module (namely, parameter values for deciding input-output relationship of the control module) to control a controlled system were determined by experiment at the stages of design or setting before shipment, so that users of a product comprising a controlled system were assumed and the users' characteristics (preference, technique, personality, and use) could be met.
However, with the diversity and advancement of recent technology, the conventional method of deciding optimal values of a characteristic of the control module by experiment brings about difficulty for optimizing the control module, and requires a lot of time. Since personal characteristics or preferences vary from one person to another, the conventional control method cannot provide a characteristic of products which satisfy all users. To solve the problem, there is proposed an evolutionary control system that comprises a basic control module which determines a control output of a controlled system based on predetermined input information, a compensation control module which determines a compensation value for the output of the basic control module, and an evolutionary control module which determines the input-output relationship of the compensation control module by a hereditary algorithm according to the user's intention. The system learns as a teacher data the input-output relationship through which the compensation control module is optimized by an optimal control module.
The system optimizes the input-output relationship of the compensation control module by the evolutionary control module under user's direct instruction as the user's intention, reflecting the compensation control module, which enables each user to obtain a preferred characteristic of the control module in accordance with his feelings. The use of a hereditary algorithm realizes a simple and short-time optimization.
However, the evolutionary control system had the problem that since the input-output relationship of the compensation control module was optimized by the evolutionary control module and the compensation control module was supposed to learn the optimized input-output relationship, it took some time for learning and it also took some time for the optimized characteristic to be reflected to the controlled variable. The system also had the problem that operation was complex.
The object of the present invention is, by solving the problems in the prior art, to reflect the optimized characteristic to the controlled system in a short period of time and to provide a method and an apparatus for optimizing an overall characteristic which simplify the operation.
Moreover, a characteristic optimizing method in prior art has the problem that with regard to a controlled system, a fuel injector, since a plurality of characteristics-fuel consumption performance and drivability performance-are independently optimized, when fuel consumption performance is improved, drivability performance becomes lower, and vice versa, which is a trade-off of optimization of both characteristics. When a plurality of control modules required for controlling a device or controlled system are used, output of manipulated variables of the device and the characteristics of each control module are optimized independently, and thus the same problems as above are raised.
An object of a cooperative method for optimization in a method for optimizing a characteristic is to solve the disadvantage of the prior art and to provide a method for cooperation in order to optimize a plurality of characteristics despite the presence of a trade-off.
SUMMARY OF THE INVENTION
The present invention provides a method for optimizing overall characteristics of a device. The method controls performance of a device, which performance is controlled essentially by at least one control module having an input-output relationship regulated by control parameters. In an embodiment, the method comprises the steps of: (a) preselecting values of the control parameters and activating the device; (b) on-line changing values of the control parameters within predetermined ranges under predetermined coding rules; (c) on-line evaluating the performance of the device based on signals indicative of the performance; (d) on-line selecting values of the control parameters based on the evaluation outcome; and (e) repeating steps (b) through (d) while operating the device until desired performance of the device is demonstrated, wherein the at least one control module is optimized. In the above, in another embodiment, at least two control modules control performance of the device, and optimization by steps (a) through (d) is conducted on each control module in repetitive sequence. In the above, “on-line” means operation on a real-time basis.
In the above, in an embodiment, the device is a control module for controlling another device.
According to the present invention, the problems described above can be resolved.
The present invention can include the following embodiments:
The method optimizes control parameters in a regular control module (i.e., a control module essentially required for controlling a device or system), using an optimization method directly, that determines an output associated with a manipulated variable of a controlled system based on predetermined input information. In the above, the device can be another control module.
The method is the optimization method that uses heuristics.
The method provides an optimal control module for performing said optimization, and after finishing optimal process in said optimal control module, updates the control parameters of the regular control module to the optimized control parameters. The method, in addition, learns the control parameters optimized to the regular control module.
The method provides said regular control module with a control module for executing control and a control module for learning, and after said control module for learning has learned the optimized control parameters, switches said control module for executing control for said control module for learning.
The method provides an optimal control module for outputting the control parameters of the regular control module based on predetermined input information, and optimizes the control parameters of the regular control module by optimizing said optimal control module.
The method uses an algorithm, when the regular control module changes at least part of the control parameters, that can predict influence to other control parameters by the change, uses an algorithm, when the regular control module changes at least part of the control parameters, that can predict influence to the output of the control module by the change, and uses an algorithm that has a linear input-output relation.
The method, as said optimization method, uses an evolutionary calculation method, an adjacent search method and/or an enforced learning method. The evolutionary calculation method includes, for example, a hereditary algorithm, an evolutionary strategy or an evolutionary programming. The adjacent search method includes, for example, an simulated annealing, a hill climbing, a random walk, and a TABU search. The enforced learning method includes, for example, a Q learning or a classifier system.
Said regular control module is the control module which outputs the manipulated variable of the controlled system based on predetermined input information, and inputs a manipulated variable for a user and outputs a manipulated variable for the controlled system.
Said controlled system is a means for controlling performance of a final controlled system, and said final controlled system is a motor, a prime mover, or a combination
Knobbe Martens Olson & Bear LLP
Von Buhr Maria N.
Yamaha Hatsudoki Kabushiki Kaisha
LandOfFree
Method and apparatus for optimizing overall characteristics... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for optimizing overall characteristics..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for optimizing overall characteristics... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3080735