Adaptive fuzzy controller that modifies membership functions

• US 5,822,740 A
• Filed: 06/28/1996
• Issued: 10/13/1998
• Est. Priority Date: 06/28/1996
• Status: Expired due to Term

• Current Assignee: Honeywell Incorporated (Honeywell International Inc.)
• Original Assignee: Honeywell Incorporated (Honeywell International Inc.)
• Inventors: Pirovolou, Dimitris K., Woessner, Michael A., Haissig, Christine M.
• Primary Examiner(s): Macdonald, Allen R.
• Assistant Examiner(s): Patel, Jagdish
• Application Number:
  US08/671,997
• Time in Patent Office: 837 Days
• Field of Search: 395/900, 395/1, 395/61, 395/3, 364/160, 364/161, 364/162, 364/163
• US Class Current: 706/3
• CPC Class Codes: F24D 19/1066 for the combination of central heating and domestic hot water
• CPC Class Codes: F24D 19/1084 for air heating systems
• CPC Class Codes: G05B 13/0275 using fuzzy logic only
• CPC Class Codes: Y10S 706/90 Fuzzy logic
• CPC Class Codes: Y10S 706/903 Control

Abstract

A direct adaptive fuzzy controller having an adaptation mechanism that modifies the locations of output membership functions to improve performance of the fuzzy controller. The controller has a fuzzy rule base that has a different or unique output membership function for each fuzzy rule. The adaptation mechanism modifies the location of the output membership functions in response to the performance of the controller system, to continuously improve its performance. The controller is a feedback mechanism that functions not only using measured outputs of the process controlled, but has feedforward compensation that causes the controller to anticipate feedback due to measured disturbances or other parameters of the process. The adaptive fuzzy controller can be used for various processes, including combi-boiler domestic hot water temperature control, automobile cruise control, a residential thermostat, duct air temperature or static pressure control for the air handling unit in a heating ventilation and air conditioning system, and furnace temperature control for curing parts.

Claims

1. A adaptive fuzzy controller for controlling a process, comprising:

an input processing unit;

an adaptive fuzzy control logic unit connected to said input processing unit;

an output processing unit connected to said adaptive fuzzy control logic unit;

a sensor proximate to the process and connected to said input processing unit; and

an actuator, for controlling the process, connected to said output processing unit; and

wherein;

said adaptive fuzzy control logic unit comprises;

an adaptation mechanism connected to said input processing unit; and

a fuzzy control logic unit connected to said adaptive mechanism and to said input and output processing units;

said sensor provides a process parameter measurement to said input processing unit;

said input processing unit has an input for a process parameter setpoint;

said adaptation mechanism;

computes an error measure between the process parameter measurement and the process parameter set point;

determines whether the absolute value of the error measure is less than an adaptation threshold;

does nothing further if the absolute value of the error measure is less than the adaptation threshold;

obtains degrees of use of fuzzy rules and current locations of output rule membership functions from said fuzzy control logic unit;

calculates new locations of the output rule membership functions; and

sends the new locations to said fuzzy control logic unit;

said fuzzy control logic unit comprises;

a fuzzifier connected to said output processing means;

a fuzzy inference engine connected to said fuzzifier and to said adaptation mechanism;

a defuzzifier connected to said inference engine and to said output processing unit;

a collection of input membership functions unit connected to said fuzzifier;

a collection of output membership functions unit connected to said adaptation mechanism and said defuzzifier; and

a fuzzy rule base connected to said fuzzy inference engine, said collection of input membership functions unit and said collection of output membership functions unit;

said sensor provides an environment disturbance parameter measurement to said input processing unit;

said input processing unit has an input for receiving parameters for configuring and adjusting the adaptation fuzzy controller;

the process is hot water temperature control of a boiler;

said sensor comprises;

a temperature sensor; and

a flow rate sensor;

the process parameter measurement is hot water temperature;

the environment disturbance parameter measurement is hot water flow rate; and

said actuator is an actuator of a gas valve which controls a flow of gas to a combustion process of the boiler, for heating hot water.

2. The adaptive fuzzy controller of claim 1, wherein:

the boiler is a combi-boiler; and

the hot water is domestic hot water.

3. An adaptive fuzzy logic controller for controlling a process parameter, having an adaptive fuzzy control logic which comprises:

an adaptation mechanism for;

computing an error measure;

determining whether the absolute value of the error measure is greater than or equal to an adaptation threshold value;

obtaining degrees of membership of fuzzy rules and current locations of output membership functions of the fuzzy rules, and calculating new locations of the output membership functions of the fuzzy rules, only if the absolute value of the error measure is greater than or equal to the adaptation threshold value; and

a fuzzy control logic unit comprising;

a fuzzifier for receiving and transforming the process parameter measurement and the process parameter setpoint into a process parameter measurement signal coded with a first set of shapes and locations of input membership functions and a process parameter setpoint signal coded with a second set of shapes and locations of input membership functions;

a collection of input membership functions for providing said fuzzifier with sets of shapes and locations of input membership functions;

a collection of output membership functions, for providing current locations of output membership functions to said adaptation mechanism and for receiving new locations of the output membership functions;

a fuzzy rule base for receiving sets of shapes and locations of input membership functions from said collection of input membership functions, receiving sets of shapes and locations of output membership functions from said collection of output membership functions, and mapping a combination of input membership functions to each output membership function;

a fuzzy inference engine for receiving the process parameter measurement signal coded with a first set of shapes and locations of input membership functions and the process parameter setpoint signal coded with a second set of shapes and locations of input membership functions, receiving a mapping of a combination of input membership functions to each output membership function, indicating the degree to which each fuzzy rule is used, and providing the degrees of membership of the fuzzy rules to said adaptation mechanism; and

a defuzzifier for receiving sets of shapes and locations of output membership functions from said collection of output membership functions, receiving degrees of membership of the fuzzy rules, and providing command information signals for controlling the process.

4. The adaptive fuzzy logic controller of claim 3 wherein said adaptation mechanism calculates the new locations of the output membership functions according to ##EQU6## where Rj indicates the current locations of the output membership functions, γ

is an adaptation rate which controls how fast Rj changes location, μ

_Rj is the degrees of memberships of the fuzzy rules, J is the error measure which has an absolute value to be driven to less than the adaptation threshold value by the adaptive fuzzy logic controller, and Δ

t is a time cycle between each indication of new locations of the output membership functions.

5. An adaptive fuzzy logic controller comprising:

input processing means, having a plurality of inputs for receiving disturbance, setpoint, process parameter, and other parameter signals, and having at least one output, for transforming normalizing, scaling, filtering, integrating, taking derivatives, and/or analog-to-digital converting the signals into an output as specified by an application for controlling at least one process parameter;

an adaptive fuzzy control logic means, having at least one input connected to the at least one output of said input processing means, and having at least one output, wherein said adaptive fuzzy control logic means comprises;

an adaptation mechanism having a first plurality of inputs for receiving the output of said input processing means, a second plurality of inputs indicating degrees of membership of fuzzy control logic rules in that for each time the adaptive fuzzy logic controller cycles, the degree of membership being the extent that each rule was used and indicating a relationship between the inputs and the outputs of the fuzzy control logic means, a third plurality of inputs indicating current locations of output member functions, and a plurality of outputs indicating new locations of output membership functions; and

a fuzzy control logic having a first plurality of inputs for receiving the output of said input processing means, a second plurality of inputs connected to the plurality of outputs from said adaptation mechanism, a first plurality of outputs connected to the second plurality of inputs of said adaptation mechanism, a second plurality of outputs connected to the third plurality of inputs of said adaptation mechanism, and a third plurality of outputs indicating raw control command information;

wherein said fuzzy control logic comprises;

a fuzzifier having a first plurality of inputs for receiving the output of said input processing means, and a second plurality of inputs indicating shapes and locations and a set of outputs providing a fuzzified output of said input processing means, indicating shapes and locations of the input membership functions, wherein the fuzzified output is taking measured values of the output and transforming them in view of the input membership functions for the outputs;

a collection of input membership functions having a plurality of outputs connected to the second plurality of inputs of said fuzzifier;

a fuzzy rule base having a plurality of outputs for indicating and mapping a combination of input membership functions to each output membership function;

a fuzzy inference engine having a plurality of inputs connected to the plurality of outputs of said fuzzifier, having a second plurality of inputs connected to the plurality of outputs of said fuzzy rule base, and having a plurality of outputs, for indicating the degree to which each fuzzy rule is used;

a collection of output membership functions having a first plurality of outputs indicating the shape and locations of the output membership functions, a second plurality of outputs indicating the current locations of the output membership functions, and a plurality of inputs indicating the new locations of the output membership functions; and

a defuzzifier having first and second pluralities of inputs connected to the outputs of said fuzzy inference engine and of said collection of output membership functions, having a plurality of outputs indicating raw command information signals;

output processing means, having a plurality of inputs connected to the plurality of outputs of said defuzzifier, and a plurality of outputs indicating raw control command information, signals for transforming, normalizing, scaling, filtering, integrating, taking derivatives, and/or digital-to-analog converting raw command information signals into actuating signals; and

a process having a first plurality of inputs connected to the plurality of outputs of output processing means, a second plurality of inputs receiving measured disturbances of the process, and a plurality of outputs indicating at least one property of a parameter or parameters being controlled.

6. The controller of claim 5 wherein:

said adaptation mechanism computes an error measure from the output of said input processing means, and determines whether the absolute value of the error measure is less than an adaptation threshold;

if the error measure is less than the adaptation threshold, then no change to output membership functions is made;

if the error measure is equal to or greater than the adaptation threshold, then degrees of membership of the fuzzy logic rules μ

_Rj and the current locations of output membership functions Rj are received from said fuzzy control logic via the second and third pluralities of inputs of said adaptation mechanism;

new locations Rj'"'"' of the output membership functions Rj are calculated according to ##EQU7## where γ

is a rate of adaptation of the adaptive fuzzy logic controller which indicates rate that the output membership functions change locations, and ##EQU8## where Δ

t is the time difference between the current locations and the new locations of the output membership functions; and

the new locations of output membership functions are passed from the plurality of outputs of said adaptation mechanism.

7. An adaptive fuzzy logic controller for controlling a process parameter of a process, comprising:

an input processing unit;

an adaptive fuzzy control logic, connected to said input processing unit, comprising;

an adaptation mechanism for;

computing an error measure;

determining whether the absolute value of the error measure is greater than or equal to an adaptation threshold value; and

obtaining degrees of membership of fuzzy rules and current locations of output membership functions of the fuzzy rules, and calculating new locations of the output membership functions of the fuzzy rules, only if the absolute value of the error measure is greater than or equal to the adaptation threshold value; and

a fuzzy control logic unit comprising;

a fuzzifier for receiving and transforming a process parameter measurement and a process parameter setpoint into a process parameter measurement signal coded with a first set of shapes and locations of input membership functions and a process parameter setpoint signal coded with a second set of shapes and locations of input membership functions;

a collection of input membership functions for providing said fuzzifier with sets of shapes and locations of input membership functions;

a collection of output membership functions, for providing current locations of output membership functions to said adaptation mechanism and for receiving new locations of the output membership functions;

a fuzzy rule base for receiving sets of shapes and locations of input membership functions from said collection of input membership functions, receiving sets of shapes and locations of output membership functions from said collection of output membership functions, and mapping a combination of input membership functions to each output membership function;

a fuzzy inference engine for receiving the process parameter measurement signal coded with a first set of shapes and locations of input membership functions and the process parameter setpoint signal coded with a second set of shapes and locations of input membership functions, receiving a mapping of a combination of input membership functions to each output membership function, indicating the degree to which each fuzzy rule is used, and providing the degrees of membership of the fuzzy rules to said adaptation mechanism; and

a defuzzifier for receiving sets of shapes and locations of output membership functions from said collection of output membership functions, receiving degrees of membership of the fuzzy rules, and providing command information signals for controlling the process; and

an output processing unit connected to said adaptive fuzzy control unit.

8. The adaptive fuzzy logic controller of claim 7, further comprising:

a sensor connected to said input processing unit; and

an actuator connected to said output processing unit.

9. The adaptive fuzzy logic controller of claim 8, wherein:

the process is temperature control of a furnace;

said sensor comprises a temperature sensor;

the process parameter is furnace temperature; and

said actuator is an actuator of a valve which controls a flow of fuel to the furnace.

10. The adaptive fuzzy logic controller of claim 8, wherein:

the process is a vehicle speed control;

said sensor comprises a speedometer

the process parameter measurement is a speed of the vehicle; and

the actuator is a servo control of a throttle position which controls the speed of the vehicle.

11. The adaptive fuzzy logic controller of claim 8, wherein:

the process is temperature control of a space;

the sensor comprises;

a space temperature sensor; and

an outside air temperature sensor;

the process parameter measurement is space temperature; and

said actuator is a relay of a valve which controls a flow of fuel or electrical power to a space heater or cooler.

12. The adaptive fuzzy logic controller of claim 8, wherein:

the process is duct air temperature control for an air handling unit in an HVAC system;

the sensor comprises;

a duct air temperature sensor; and

an upstream air temperature sensor; and

said actuator is an actuator that controls a water valve position.

13. The adaptive fuzzy logic controller of claim 8, wherein:

the process a duct static pressure control for an air handling unit in an HVAC system;

the sensor is a duct static pressure sensor;

the process parameter measurement is a duct static pressure; and

said actuator is a device which controls air flow.

14. The adaptive fuzzy logic controller of claim 8, wherein:

the process is water temperature control of a boiler;

said sensor comprises;

a temperature sensor; and

a flow rate sensor;

the process parameter measurement is water temperature; and

said actuator is an actuator of a valve which controls a flow of fuel to a combustion process of the boiler, for changing the temperature of water.

15. The adaptive fuzzy logic controller of claim 8, wherein:

the process is a vehicle speed control;

said sensor comprises a vehicle speed indicator;

the process parameter measurement is a speed of the vehicle; and

the actuator affects the speed of the vehicle.

16. The adaptive fuzzy logic controller of claim 8, wherein:

the process is temperature control of a furnace for curing parts;

said sensor comprises a temperature sensor;

the process parameter is part temperature; and

said actuator provides a furnace wall setpoint temperature.

17. The adaptive fuzzy logic controller of claim 8, wherein:

the process is temperature control of a space;

the sensor is a space temperature sensor;

the process parameter measurement is space temperature; and

said actuator is a relay of a valve which controls a flow of water to a convector.

18. The adaptive fuzzy logic controller of claim 8, wherein:

the process is duct air temperature control for an air handling unit in an HVAC system;

the sensor comprises;

a duct air temperature sensor; and

an upstream air temperature sensor; and

said actuator is an actuator that controls a valve position.