Prognostic processor system for real-time failure analysis of line replaceable units

• US 7,509,537 B1
• Filed: 02/02/2006
• Issued: 03/24/2009
• Est. Priority Date: 02/02/2006
• Status: Active Grant

• Current Assignee: Rockwell Collins, Inc. (Raytheon Technologies Corporation d/b/a RTX)
• Original Assignee: Rockwell Collins, Inc. (Raytheon Technologies Corporation d/b/a RTX)
• Inventors: Jensen, David W., Marek, James A.
• Primary Examiner(s): Baderman; Scott T.
• Assistant Examiner(s): Truong; Loan
• Application Number:
  US11/345,704
• Time in Patent Office: 1,146 Days
• Field of Search: 714/47, 714/48
• US Class Current: 714/47.2
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms
• CPC Class Codes: G06F 11/008 Reliability or availability analysis

Abstract

A prognostic processor for predicting machine failure in avionics electronics comprises prognostic capabilities in a single integrated circuit, with a processor, volatile and non-volatile memory, clock, on-chip and off-chip sensors and transducers, A/D converters, a common I/O interface adapted to be employed in a network of similar prognostic processors, and predictive Failure Analysis (FA) model software, which may be distributed throughout the network. The FA software employs a log file history, with the log file history storing data collected by the prognostic processor, real-time execution of a predictive model, with the ability to update the FA model with data from field failures. The prognostic processor network supports hierarchical processing to work with multiple prognostic processors. The prognostic processor system is applicable to FA monitoring of a wide range of avionics electronic equipment, in particular, Line Replacement Units (LRUs).

Claims

1. A computer controlled method for predicting machine failure employing a prognostics processor chip comprising a processor, memory, clock and I/O, components and devices comprising on-chip and off-chip sensors operatively interacting with the I/O;

the I/O sending and receiving data from the sensors; and

a program stored in the memory and running the processor, the program containing a failure analysis model for predicting machine failure of a machine being monitored by the prognostics processor chip, the method comprising the steps of;

collecting sensor data from the sensors of the prognostics processor by sampling the sensors through the I/O;

processing the sensor data with the processor of the prognostics processor;

storing the processed sensor data in a historic log in the memory of the prognostics processor, wherein the memory comprises non-volatile memory, and the historic log maintains a cumulative collection of sensor data over time;

executing a failure analysis model in the program, the failure analysis model employing the sensor data from the historic log,

operatively attaching to the prognostic processor an external equipment monitor external to the prognostic processor, the external equipment monitor separate and apart from the machine being monitored by the prognostic processor;

creating a plurality of prognostic processors for a plurality of machines to be monitored for failure analysis;

connecting each of the plurality of prognostic processors to the machines being monitored for failure analysis;

establishing a network of prognostic processors from the plurality of prognostic processors, said network connected along a system bus;

notifying the machine that a prognostic processor is monitoring that a failure is likely;

issuing queries with the external equipment monitor to the prognostic processor to read registers in the prognostic processor and obtain the historic log data;

forming the network of prognostic processors so that the I/O of the prognostic processors share a common standard;

dividing tasks amongst the network of prognostic processors, wherein, the control and state messages of the system data collected comprise Control messages that will allow the external equipment monitor to control and monitor the prognostic processor and Status messages that return the prognostic processor results to the external equipment requests, the model predicts likely failure of a machine being monitored by the prognostics processor, collecting system data for the prognostics processor by querying the I/O for system data, the system data comprising state and control messages, and storing the processed system data in the historic log in the memory of the prognostics processor, and the network interacts with the external equipment monitor external to the machines being monitored, and the failure analysis model in the program of the prognostic processors indicates when at least one of the machines is likely to fail.

2. The method according to claim 1, further comprising the steps of:

connecting the plurality of prognostic processors to create a hierarchical network, with a first prognostic processor communicating with the external equipment monitor, and the first prognostic processor controlling at least a second prognostic processor through the external equipment monitor.

3. The method according to claim 1, further comprising the step of:

storing in said non-volatile memory past field failures of machines of the kind being monitored by the prognostics processor;

wherein,

said sensor data is selected from the group of system characteristics of the machine being monitored consisting of;

voltage, current, resistance, power, peak power, thermal parameters, temperature, electronic life, vibration, electrical and mechanical contacts, noise, resets of the machine being monitored, the number of times a switch is depressed on the machine being monitored, detection of biological agents, detection of chemical agents, power glitches, power-on hours of the machine being monitored, accumulative vibration, electronic life, power glitches, acceleration, corrosion, mechanical stress or strain; and

further comprising the step of,

calculating in said failure analysis model when the probability of a component in a machine being monitored will fail over a finite time period, and,

calculating for said failure analysis model a threshold for the probability of failure, based on the sensor data and the past field failure data.

4. The method according to claim 1, further comprising the steps of:

calculating in the failure analysis model when the probability of a component in a machine being monitored will fail over a finite time period; and

,

calculating for the failure analysis model a threshold for the probability of failure.