US 5,471,381 A · Intelligent servomechanism controller

Abstract

A controller for a servomechanism (such as a computer disk drive) includes both a conventional dumb controller and a neural network controller working in conjunction with one another. In one embodiment, the neural network controller and dumb controller operate in a quasi-series configuration, with the neural network controller receiving and processing the output of the dumb controller to produce a servomechanism control signal. In another embodiment, the neural network controller and dumb controller operate in a quasi-parallel configuration, with the outputs of the neural network controller and dumb controller being combined to produce an intelligent servomechanism control signal. In yet another embodiment, the neural network controller and dumb controller operate in a quasi-series configuration during the recall phase, following the learning phase in which the neural network controller is trained to develop an indirect performance model of the serial combination of the dumb controller and servomechanism. In still another embodiment, the neural network controller and dumb controller operate in a quasi-series configuration during the recall phase, while a model neural network is used to develop a direct performance model during the learning phase of the serial combination of the dumb controller and servomechanism which is then used to train the neural network controller for its operation during the recall phase.

Claims52

Dependent claims

A neural network controller for a plant which has a desired plant performance parameter represented by a reference signal and which has an actual plant performance parameter represented by an actual performance signal, said neural network controller comprising:

a first input port for inputting a controller weights adjust signal;
a second input port for inputting a dumb control signal;
neural network means coupled to said first and second input ports for receiving said inputted controller weights adjust signal and dumb control signal, and for processing said inputted dumb control signal in accordance with said inputted controller weights adjust signal to generate a plant control signal;
an output port for outputting said plant control signal;
and
input signal combiner means coupled to said first input port for receiving and combining a reference signal and an actual performance signal and for providing in accordance therewith a first performance error signal, wherein said reference signal represents a desired plant performance parameter, said actual performance signal represents an actual plant performance parameter and said first performance error signal represents a first difference between said desired plant performance parameter and said actual plant performance parameter.

3 more elements

control neural network means for receiving a reference signal and a controller weights adjust signal, and for providing in accordance therewith an intelligent control signal, wherein said reference signal represents a desired plant performance parameter;
model neural network means coupled to said control neural network means for receiving said intelligent control signal and a performance model error signal, and for providing in accordance therewith a system performance model signal, wherein said performance model error signal represents a difference between said system performance model signal and an actual performance signal, said system performance model signal represents a performance model of a system which includes a dumb controller and a plant, and said actual performance signal represents an actual plant performance parameter of said plant;
control signal combiner means coupled to said control neural network means and said model neural network means for receiving and combining said reference signal and said system performance model signal, and for providing in accordance therewith an error feedback signal;
and
model signal combiner means coupled to said model neural network means for receiving and combining said system performance model signal and said actual performance signal, and for providing in accordance therewith said performance model error signal.

2 more elements

A method for controlling a plant with a neural network controller, wherein said plant has a desired plant performance parameter represented by a reference signal and has an actual plant performance parameter represented by an actual performance signal, said control method comprising the steps of:

inputting a plurality of network input signals to a neural network, including a controller weights adjust signal and a reference signal which represents a desired plant performance parameter;
outputting an intelligent control signal from said neural network in accordance with said plurality of network input signals;
inputting to and combining in a control signal combiner a plurality of controller signals, including said intelligent control signal and a dumb control signal;
outputting a plant control signal from said control signal combiner in accordance with said inputted and combined plurality of controller signals;
inputting and combining a plurality of input signals, including said reference signal and an actual performance signal which represents an actual plant performance parameter;
and
outputting a performance error signal in accordance with said combined plurality of input signals, wherein said performance error signal represents a difference between said desired plant performance parameter and said actual plant performance parameter.

4 more elements

A method for simulating with a neural network an inverse performance model of a system which includes a dumb controller and a plant, wherein said plant has a desired plant performance parameter represented by a reference signal and has an actual plant performance parameter represented by an actual performance signal, said simulation method comprising the steps of:

inputting a plurality of network input signals to a neural network, including an actual performance signal and a performance model error signal;
outputting a system inverse performance model signal from said neural network in accordance with said plurality of network input signals, wherein said actual performance signal represents an actual plant performance parameter of a plant within a system which includes a plant and a dumb controller, and said performance model error signal represents a difference between said system inverse performance model signal and a reference signal, said system inverse performance model signal represents an inverse performance model of said system, and said reference signal represents a desired plant performance parameter of said plant;
combining said system inverse performance model signal and said reference signal;
and
outputting said performance model error signal in accordance with said combined system inverse performance model signal and reference signal.

2 more elements

inputting a plurality of network input signals to a control neural network, including a controller weights adjust signal and a reference signal which represents a desired plant performance parameter;
outputting an intelligent control signal from said control neural network in accordance with said plurality of network input signals;
inputting said intelligent control signal and a performance model error signal to a model neural network;
outputting a system performance model signal from said model neural network in accordance with said intelligent control signal and said performance model error signal, wherein said performance model error signal represents a difference between said system performance model signal and an actual performance signal, said system performance model signal represents a performance model of a system which includes a dumb controller and a plant, and said actual performance signal represents an actual plant performance parameter of said plant;
combining said reference signal and said system performance model signal;
outputting an error feedback signal in accordance with said combined reference signal and system performance model signal;
combining said system performance model signal and said actual performance signal;
and
outputting said performance model error signal in accordance with said combined system performance model signal and actual performance signal.

6 more elements

Dependencies are inferred from the claim text.

Intelligent servomechanism controller

Life of this patent

Intelligent servomechanism controller

Life of this patent

Abstract

Claims52