SELF-CORRECTING FEEDBACK CONTROL SYSTEM

• US 3,851,157 A
• Filed: 07/09/1973
• Issued: 11/26/1974
• Est. Priority Date: 07/09/1973
• Status: Expired due to Term

• Current Assignee: Stanley H. Ellis
• Original Assignee: Stanley H. Ellis
• Inventors: Ellis, Stanley H.
• Primary Examiner(s): Botz, Eugene G.
• Application Number:
  US05/377,867
• Time in Patent Office: 505 Days
• Field of Search: 235/150.1,151.3,151.31,153AE 60/39.28R,39.28T 244/77M,77R
• US Class Current: 700/45
• CPC Class Codes: F02C 9/00 Controlling gas-turbine plants; Controlling fuel supply in air- breathing jet-propulsion plants controlling air intakes F02C7/057; controlling turbines F01D; controlling compressors F04D27/00; controlling in general G05
• CPC Class Codes: G05B 9/02 electric
• CPC Class Codes: Y02T 50/60 Efficient propulsion technologies, e.g. for aircraft

Abstract

A self-correcting control system for a controlled device such as an engine is disclosed in which the outputs from a plurality of combinations of sensors measuring selected system parameters produce multiple estimates of one or more variables of operation of the control system, the controlled variables. The multiple estimates are combined into a weighted average of the controlled variables, and individual estimates which differ by more than a specified amount from the weighted average are eliminated from consideration. After elimination of the invalid estimates, a revised weighted average of the controlled variables is computed. The revised average estimate of each controlled variable is then compared with a signal indicative of the commanded value of the controlled variable, and any error is used to regulate actuators which in turn control the controlled device. Predetermined weighting factors are assigned to individual estimates of each controlled variable to reflect both the accuracy of the feedback sensors and the ability of the feedback parameters being sensed to form a good estimate of the controlled variables. Individual feedback sensor combinations whose estimates of the controlled variable have been eliminated are continuously monitored, and whenever the eliminated sensor combination produces an estimate of a controlled variable which falls within the prescribed boundaries, the estimate is reconnected. Examples are provided for an engine control with four feedback sensors, all nonordered combinations of the sensors providing six estimates of each of two controlled variables, with two actuators being controlled in response to the error in the variables.

Claims

1. In a control system for a controlled device having a pair of actuators for regulating the controlled device, said actuators being responsive to first and second variables of operation of said controlled device, said controlled device having connected therewith a plurality of sensors each of which is adapted to produce a signal which is indicative of a condition of operation of said controlled device and which is a function of at least one of the variables of operation of said controlled device, the improvement which consists of producing from said sensor signals an estimate of said one variable of operation comprising means for scaling each of said sensor signals as a function of its relationship to said one variable of operation, means for generating from a plurality of pairs of said scaled sensor signals a plurality of estimate signals indicative of the value of said one variable of operation, means for producing from said estimate signals an average signal indicative of the average of said plurality of estimate signals, means responsive to the difference between each of said estimate signals and said average signal for setting to zero any of said estimate signals which are outside a predetermined range, and means for producing a revised average signal of said one variable in response to the remaining estimate signals.

2. A control system as in claim 1 and including means for producing for each of said estimate signals a weighting signal, means for modifying each said estimate signal in accordance with its respective weighting signal to produce a plurality of weighted estimate signals, means for summing said weighting signals to produce a summed weighting signal, and means for producing from said weighted estimate signals and said summed weighting signals a weighted average signal which is indicative of the weighted average of said one variable.

3. A control system as in claim 2 and including means responsive to the difference between each said estimate signal and said average signal for setting to zero the respective weighting signals associated with each said estimate signal which is set to zero, means for summing the remaining weighting signals which have not been set to zero to produce therefrom a revised summed weighting signal, and means for modifying said revised average signal in response to said revised summed weighting signal to produce therefrom a revised weighted average signal for said one variable.

4. A control system as in claim 1 and including means for reinstating each estimate signal which has been set to zero when the estimate signal differs from said average signal by less than a predetermined amount.

5. A control system as in claim 1 and including means for producing a reference signal indicative of the desired value of said variable of operation, means for comparing said revised mean signal with said reference signal to produce therefrom a control signal proportional to the difference therebetween, and means for supplying said control signal to at least one of said actuators to thereby null said control signal.

6. A control system as in claim 2 in which said means for producing a weighting signal and said means for modifying each said estimate signal to produce a weighted estimate signal includes a dual potentiometer connected with each of said estimate signals.

7. In a self-correcting control system for a controlled device having a plurality of actuators for regulating a plurality of variables of operation of said controlled device, the improvement of producing a signal indicative of the value of one of said variables of operation from a plurality of sensor signals each of which is indicative of a condition of operation of said controlled device and is a function of said variable of operation comprising means for scaling each of said sensor signals as a function of its relationship with said variable of operation, means for producing from selected combinations of said scaled sensor signals a plurality of independent estimate signals each of which representative of the value of said variable of operation, first summing means for summing said plurality of estimate signals and producing therefrom an average signal indicative of the average of said plurality of estimate signals, means for comparing each of said estimate signals with said average signal to produce a difference signal, a permitted error signal, means for comparing each said difference signal with said permitted error signal and producing an editing signal for each said difference signal when said difference signal exceeds said permitted error signal, means responsive to the occurrence of an editing signal for setting the respective estimate signal to zero, and second summing meanS for summing the remaining estimate signals which have not been set to zero and producing therefrom a revised average signal indicative of the estimated value of said variable of operation.

8. A control system as in claim 7 and including means for varying the magnitude of said permitted error signal when said revised average signal is changing at a rate above a preselected rate of change.

9. A control system as in claim 7 and including means for producing for each of said estimate signals a weighting signal, means for modifying each said estimate signal in accordance with its respective weighting signal to produce a plurality of weighted estimate signals, means for summing said weighting signals to produce a summed weighting signal, and means for producing from said weighted estimate signals and said summed weighting signals a weighted average signal which is indicative of the weighted average of said one variable.

10. A control system as in claim 9 and including means responsive to the difference between each said estimate signal and said average signal for setting to zero the respective weighting signals associated with each said estimate signal which is set to zero, means for summing the remaining weighing signals which have not been set to zero to produce therefrom a revised summed weighting signal, and means for modifying said revised average signal in response to said revised summed weighting signal to produce therefrom a revised weighted average signal for said one variable.

11. A control system as in claim 9 and including means for dividing the permitted error signal by the weighting signal produced for each said estimate signal to produce a revised permitted error signal for each said estimate signal, and means for comparing the difference signal for each estimate signal with the respective revised permitted error signal to produce each of the said validity signals.

12. A control system as in claim 7 in which said scaling means includes a variable potentiometer connected with each said sensor signal.

13. In a self-correcting control system for a controlled device having a plurality of actuators for regulating a plurality of variables of operation of said controlled device, the improvement of producing a signal indicative of the value of one of said variables of operation from a plurality of sensor signals each of which is indicative of a condition of operation of said controlled device and is a function of said variable of operation comprising the steps of scaling each of said sensor signals as a function of its relationship with said variable of operation, generating from selected combinations of said scaled sensor signals a plurality of independent estimate signals each of which is representative of the value of said variable of operation, summing said plurality of estimate signals and producing therefrom an average signal indicative of the average of said plurality of estimate signals, comparing each of said estimate signals with said average signal and setting to zero any of the estimate signals which differ from said average signal by more than a preselected amount, and summing the remaining estimate signals to produce a revised average signal indicative of the estimated value said one variable of operation.

14. A self-correcting control system as in claim 13 and including the steps of generating a weighting signal for each of said estimate signals, modifying each of said estimate signals in response to its respective weighting signal to produce a plurality of weighted estimate signals, summing said weighting signals to produce a summed weighting signal, summing said weighted estimate signals to produce a summed weighted estimate signal, and dividing said summed weighted estimate signal by said summed weighting signal to produce a weighted average signal for said one variable.

15. A self-correcting control system as in claim 14 and inclUding the steps of setting to zero the weighting signal corresponding to each estimate signal which has been sent to zero, summing the remaining weighting signals to produce a revised summed weighting signal, and dividing the revised average signal by said revised summed weighting signal to produce therefrom a revised weighted average signal indicative of the estimated value of said one variable.

16. A method for estimating the value of a variable of operation in a process comprising the steps of producing a plurality of signals each of which is indicative of a condition of operation of said process and is related to said variable of operation, generating from a plurality of combinations of said signals a plurality of independent estimates of the value of said variable of operation, assigning to each said independent estimate a weighting factor to produce a plurality of independent weighted estimates of the value of said variable of operation, summing said plurality of weighted estimates and said weighting factors to produce therefrom a weighted average estimate of the value of said variable of operation, setting of zero the weighted estimates of the value of the variable of operation which deviate from said weighted average estimate by more than a predetermined amount, setting to zero the weighting factors associated with the said weighted estimates of the value of the variable of operation which are set to zero, and producing from the weighted estimates of the value of the variable of operation which are not set to zero and from the weighting factors associated therewith a revised weighted average estimate of the value of said variable of operation.