Method and apparatus for distinguishing between deployment events and non-deployment events in an SIR system

• US 5,583,771 A
• Filed: 08/04/1994
• Issued: 12/10/1996
• Est. Priority Date: 08/04/1994
• Status: Expired due to Term

• Current Assignee: Delphi Technologies, Inc. (BorgWarner Incorporated)
• Original Assignee: Delco Electronics Corporation (General Motors Company)
• Inventors: Kiselewich, Stephen J., Long, James F., Brumbach, Rex P. Jr., Garcia, Porfirio Jr., Turner, Douglas D., Lynch, David D.
• Primary Examiner(s): PARK, COLLIN
• Application Number:
  US08/285,673
• Time in Patent Office: 859 Days
• Field of Search: 364/424.05, 364/148, 364/149, 364/150, 364/151, 395/20, 395/21, 395/23, 395/905, 180/271, 180/282, 280/734, 280/735, 280/728.1, 307/10.1
• US Class Current: 701/36
• CPC Class Codes: B60R 21/0132 responsive to vehicle motion parameters , e.g. to vehicle longitudinal or transversal deceleration or speed value
• CPC Class Codes: B60R 21/01338 using vector analysis
• CPC Class Codes: G06N 3/048 Activation functions
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: Y10S 706/905 Vehicle or aerospace

Abstract

A pattern recognition system is utilized in a supplementary inflatable restraint (SIR) system to distinguish between deployment and non-deployment events. The pattern recognition system preferably includes dedicated hardware or a microprocessor programmed to perform a neural network simulation utilizing crash data in the form of vehicle acceleration data. Training and trial vectors are generated from the data to train and, subsequently, test the neural network.

Claims

1. A method for distinguishing between deployment and non-deployment events from crash signals containing crash data in an SIR system for a motor vehicle, the method comprising the steps of:

generating pairs of training vectors from the crash data, each pair of training vectors including a vector having input values and a vector having a known output value associated with either a deployment event or a non-deployment event;

inputting the pairs of training vectors to a trainable pattern recognition system;

recursively adjusting the pattern recognition system to converge to a configuration of the pattern recognition system to obtain a trained pattern recognition system for matching an input vector with either a deployment event or a non-deployment event;

generating at least one trial input vector from the crash data, the at least one trial input vector corresponding to either a deployment event or a non-deployment event; and

inputting the at least one trial input vector to the trained pattern recognition system to generate an output signal representation of one of the events corresponding to the input vector.

2. The method of claim 1 wherein the crash data is vehicle acceleration data.

3. The method of claim 1 wherein the crash data is vehicle acceleration data and wherein the step of generating pairs of training vectors includes the step of converting the acceleration data to another format.

4. The method as claimed in claim 1 wherein the pattern recognition system includes a microprocessor programmed to perform a neural network simulation.

5. The method as claimed in claim 1 wherein the pattern recognition system includes dedicated hardware to perform pattern recognition.

6. The method as claimed in claim 1 wherein the pattern recognition system includes a microprocessor programmed to perform a back-propagation analysis.

7. The method as claimed in claim 1 wherein the trained pattern recognition system includes a neural network having an input layer, a hidden layer and an output layer.

8. The method as claimed in claim 7 wherein the input layer includes a plurality of neurodes and wherein the method further comprises the step of identifying a range defined by a minimum and a maximum for each of the neurodes.

9. The method as claimed in claim 8 wherein the step of recursively adjusting includes the step of adding new data to the input layer and clipping each of the neurodes of the input layer to fall within their respective ranges to obtain a clipped input layer.

10. The method as claimed in claim 1 wherein the step of generating pairs of training vectors includes the step of sampling the crash data at a sampling rate to obtain sampled crash data.

11. Apparatus for distinguishing between deployment and non-deployment events from crash data in an SIR system for a motor vehicle, the apparatus comprising:

means for generating pairs of training vectors from the crash data, each pair of training vectors including a vector having input values and a vector having a known output value associated with either a deployment event or a non-deployment event; and

a pattern recognition system for receiving the pairs of training vectors and for recursively adjusting to a configuration of the pattern recognition system to obtain a trained pattern recognition system for matching an input vector with either a deployment event or a non-deployment event;

the generating means also generating at least one trial input vector from the crash data, the at least one trial input vector corresponding to either a deployment event or a non-deployment event;

the at least one trial input vector being received by the trained pattern recognition system to generate an output signal representation of one of the events corresponding to the input vector.

12. The apparatus as claimed in claim 11 wherein the crash data is vehicle acceleration data.

13. The apparatus as claimed in claim 11 wherein the crash data is vehicle acceleration data and wherein the generating means includes converting means for converting the acceleration data to another format.

14. The apparatus as claimed in claim 11 wherein the pattern recognition system includes a microcomputer programmed to perform a neural network simulation.

15. The apparatus as claimed in claim 11 wherein the pattern recognition system includes hardware dedicated to performing pattern recognition.

16. The apparatus as claimed in claim 11 wherein the pattern recognition system includes a microprocessor programmed to perform a back-propagation analysis.

17. The apparatus as claimed in claim 11 wherein the trained pattern recognition system includes a neural network having an input layer, a hidden layer and an output layer.

18. The apparatus as claimed in claim 17 wherein the input layer includes a plurality of neurodes and wherein the apparatus further comprises means for identifying a range defined by a minimum and a maximum for each of the neurodes.

19. The apparatus as claimed in claim 18 wherein the trained pattern recognition system includes means for adding new data to the input layer and means for clipping each of the neurodes of the input layer to fall within their respective ranges to obtain a clipped input layer.

20. The apparatus as claimed in claim 11 wherein the generating means includes sampling means for sampling the crash data at a sampling rate to obtain sample crash data.