Neural network compensation for sensors

• US 5,554,273 A
• Filed: 07/26/1995
• Issued: 09/10/1996
• Est. Priority Date: 07/26/1995
• Status: Expired due to Term

• Current Assignee: Praxair S.T. Technology, Inc. (Linde Plc)
• Original Assignee: Praxair Technology Incorporated (Linde Plc)
• Inventors: Borkman, James D., Mazzarella, Richard B., Demmin, Hollis C.
• Primary Examiner(s): Bell, Bruce F.
• Application Number:
  US08/506,752
• Time in Patent Office: 412 Days
• Field of Search: 204/408, 204/406, 204/403, 204/416, 204/418, 204/431, 204/432, 205/775, 205/782, 205/785, 395/21, 395/24, 128/635
• US Class Current: 205/785
• CPC Class Codes: G01N 27/4163 checking the operation of, or calibrating, the measuring apparatus G01N27/3274, G01N27/4175 and G01N33/0006 take precedence

Abstract

A method for correcting an output from an electrochemical cell sensor, having an output responsive to a concentration of sensed species in a stream, a pressure and a temperature in an interactive non-linear relationship, comprising the steps of providing a pressure sensor for producing an output responsive to a pressure in proximity to the electrochemical sensor; providing a temperature sensor for producing an output responsive to a temperature of said electrochemical sensor; and processing the outputs of the electrochemical sensor, pressure sensor and temperature sensor in a neural network having an output function which compensates said electrochemical sensor for changes in pressure and temperature to indicate a concentration of the sensed species. An apparatus is also provided, for compensating an electrochemical sensing apparatus, comprising an electrochemical sensor, being responsive to a sensed species, and an environmental variable; an environmental variable sensor, being responsive to said environmental variable; and a compensation network for producing a compensated output based on an output of said electrochemical sensor based and an output of said environmental variable sensor, said network comprising a neural network.

Claims

1. A method for correcting an output from an electrochemical cell sensor, having an output responsive to a concentration of sensed species in a stream, a pressure and a temperature in an interactive non-linear relationship, comprising the steps of:

(a) providing an pressure sensor for producing an output responsive to a pressure in proximity to the electrochemical sensor;

(b) providing a temperature sensor for producing an output responsive to a temperature of said electrochemical sensor; and

(c) processing the outputs of the electrochemical sensor, pressure sensor and temperature sensor in a neural network having an output function which compensates said electrochemical sensor for changes in pressure and temperature to indicate a concentration of the sensed species.

2. The method according to claim 1 wherein said neural network is a three layer feedforward network trained with backpropagation of errors.

3. The method according to claim 1, further comprising the step of providing a concentration of sensed species in a stream sensed by said electrochemical sensor.

4. The method according to claim 1, further comprising the step of varying a pressure or temperature while sensing said sensed species.

5. The method according to claim 1, wherein said electrochemical sensor is responsive to an oxygen concentration in said stream.

6. The method according to claim 1, further comprising the step of providing a second pressure sensor for producing an output responsive to a barometric pressure.

7. An apparatus for compensating an electrochemical sensing apparatus, comprising:

(a) an electrochemical sensor, being responsive to a sensed species, and an environmental variable;

(b) an environmental variable sensor, being responsive to said environmental variable; and

(c) a compensation network for producing a compensated output based on an output of said electrochemical sensor and an output of said environmental variable sensor, said network comprising a neural network.

8. An apparatus according to claim 7, wherein said electrochemical sensing apparatus measures a concentration of a gas used for semiconductor fabrication.

9. The apparatus according to claim 7, wherein neural network comprises an input layer, a hidden layer and an output layer in a feedforward arrangement, having inputs to said input layer from said electrochemical sensor and said environmental variable sensor, wherein said electrochemical sensor, and said environmental sensor have a cross responsivity to a system condition, and wherein said electrochemical sensor, and said environmental variable sensor have a non-linear interactivity, weighted connections between said input layer and said hidden layer, weighted connections between said hidden layer and said output layer, and an output from said output layer.

10. The apparatus according to claim 7, further comprising a plurality of environmental variable sensors, said electrochemical sensing cell having a non-linear response to a plurality of environmental variables sensed by said plurality of environmental variable sensors, said neural network compensation network producing a compensated output based on an output of said electrochemical sensor and outputs of said plurality of environmental variable sensors.

11. The apparatus according to claim 10, wherein:

said electrochemical sensing cell is responsive to the sensed component and a plurality of environmental conditions;

said plurality of environmental sensors are each responsive to at least one of said plurality of environmental conditions, said environmental conditions sensed by said environmental sensors including those environmental conditions to which said electrochemical sensing cell is responsive; and

said compensation network compensates the output of said electrochemical cell for said plurality of environmental conditions and linearizes said output of said electrochemical sensing cell.

12. A neural network for temperature and pressure compensating an electrochemical cell sensor responsive to a component of a stream, comprising:

an electrochemical cell sensor input, a temperature sensor input, and a pressure sensor input; and

a neural network compensation network, trained to produce a compensated output representing a stream component concentration from said electrochemical cell sensor input, temperature sensor input and pressure sensor input.

13. The system according to claim 12, wherein said pressure sensor input receives a back pressure sensor input for measuring a pressure in a gaseous medium sensed by said electrochemical cell, further comprising a barometric pressure sensor input, for measuring an environmental pressure, wherein said compensation network receives an input from said back pressure sensor, and compensates said electrochemical sensor cell for the back pressure of the gaseous medium.

14. The system according to claim 12, wherein said neural network compensation network is trained by a back propagation of error method in a training phase.

15. The system according to claim 12, wherein said neural network receives calibration data input from a replaceable electrochemical sensor.

16. The system according to claim 12, further comprising a partial response maximum likelihood predictive system for anticipating an output value after settling.

17. The system according to claim 12, further comprising a humidity sensor input, said compensation network receiving said humidity sensor input.

18. The system according to claim 12, further comprising a thermostatically controlled enclosure for an electrochemical sensor cell.

19. The system according to claim 12, further comprising a controlled pressure enclosure for an electrochemical sensor cell.

20. The system according to claim 12, further comprising a pressure regulator to regulate a back pressure of a stream sensed by an electrochemical sensor cell.