Fault diagnostics and prognostics based on distance fault classifiers

• US 7,188,482 B2
• Filed: 07/29/2005
• Issued: 03/13/2007
• Est. Priority Date: 08/27/2004
• Status: Expired due to Fees

• Current Assignee: Carrier Corporation (Carrier Global Corporation)
• Original Assignee: Carrier Corporation (Carrier Global Corporation)
• Inventors: Sadegh, Payman, Farzad, Mohsen
• Primary Examiner(s): NORMAN, MARC E
• Application Number:
  US11/192,595
• Publication Number:
  US 20060042277A1
• Time in Patent Office: 592 Days
• Field of Search: 621/25, 621/26, 621/27, 621/29, 621/30, 621/31, 165/11.1, 165/11.2, 236/94, 702/183, 702/184, 702/185
• US Class Current: 62/126
• CPC Class Codes: F24F 11/30 for purposes related to the operation of the system, e.g. for safety or monitoring
• CPC Class Codes: F24F 11/32 Responding to malfunctions or emergencies
• CPC Class Codes: F24F 11/38 Failure diagnosis
• CPC Class Codes: F24F 11/39 Monitoring filter performance
• CPC Class Codes: F24F 2120/10 Occupancy
• CPC Class Codes: F25B 13/00 Compression machines, plants or systems, with reversible cycle defrosting cycles F25B47/02
• CPC Class Codes: F25B 49/005 of safety devices F25B49/02 and F25B49/04 take precedence
• CPC Class Codes: G01K 15/00 Testing or calibrating of thermometers
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• CPC Class Codes: G05B 23/0254 based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks

Abstract

The present invention is directed to a mathematical approach to detect faults by reconciling known data driven techniques with a physical understanding of the HVAC system and providing a direct linkage between model parameters and physical system quantities to arrive at classification rules that are easy to interpret, calibrate and implement. The fault modes of interest are low system refrigerant charge and air filter plugging. System data from standard sensors is analyzed under no-fault and full-fault conditions. The data is screened to uncover patterns though which the faults of interest manifest in sensor data and the patterns are analyzed and combined with available physical system information to develop an underlying principle that links failures to measured sensor responses. These principles are then translated into online algorithms for failure detection.

Claims

1. A method of diagnosing at least one system fault comprising the steps of:

(a) identifying sensors available within a system, wherein identified sensors generate sensor data relating to system operating characteristics;

(b) analyzing the sensor data to identify patterns indicative of at least one failure mode, wherein the sensor data is analyzed under a full-fault condition and under a no-fault condition;

(c) determining a fault relationship based upon identified patterns;

(d) comparing at least one system operating characteristic to the fault relationship to identify a system fault; and

(e) generating at least one fault code indicative of the at least one failure mode when a system fault is identified in step (d).

2. The method of diagnosing at least one system fault as recited in claim 1, wherein step (b) further includes the steps of:

(f) generating a full-fault data set under the full-fault condition and a no-fault data set under the no-fault condition from each of the identified sensors;

(g) comparing the full-fault data set from each of the identified sensors with the no-fault data set from the same identified sensor,

(h) identifying a maximum separation between the full-fault data set and the no-fault data set for each identified sensor;

(i) selecting the identified sensor associatcd with a greatest maximum separation between the full-fault data set and the no-fault data set associated with the identified sensor; and

(j) identifying the at least one system characteristic of step (d) based upon the selected identified sensor of step (i).

3. The method of diagnosing at least one system fault as recited in claim 1, wherein step (c) further includes the steps of:

(k) determining a threshold value associated with the at least one system characteristic;

(l) monitoring an actual value associated with the at least one system characteristic; and

wherein step (e) further includes generating the at least one fault code based upon a comparison of the actual value and the threshold value.

4. The method of diagnosing at least one system fault as recited in claim 3, wherein step (k) further includes the steps of:

(m) calculating a distance between the full-fault data set and the no-fault data set for the selected identified sensor associated with the maximum separation; and

(n) maximizing the distance based upon filtering parameters and variable sub-selection.

5. The method of diagnosing at least one system fault as recited in claim 3, wherein step (e) further includes the step of:

(o) identifying at least one other system characteristic not monitored by any of the identified sensors, wherein at least one other system characteristic is indicative of the at least one system failure mode associated with the at least one fault code.

6. The method of diagnosing at least one system fault as recited in claim 5, wherein step (e) further includes the steps of:

(p) calculating an avenge value associated the at least one other system characteristic for a given sampling time; and

(q) monitoring an actual value associated with the at least one other system characteristic.

7. The method of diagnosing at least one system fault as recited in claim 6, wherein step (e) further includes generating the at least one fault code when the actual value associated with the at least one system characteristic exceeds the threshold value and the actual value associated with the at least one other system characteristic exceeds the average value.

8. The method of diagnosing at least one system fault as recited in claim 7, wherein the at least one system characteristic is associated with a degree of actuation of at least one control valve within with the system.

9. The method of diagnosing at least one system fault as recited in claim 6, wherein step (e) further includes generating the at least one fault code when the actual value associated with the at least one system characteristic falls below the threshold value and the actual value associated with the at least one other system characteristic exceeds the avenge value.

10. The method of diagnosing at least one system fault as recited in claim 9, wherein the at least one system characteristic is a suction pressure associated with a compressor.

11. The method of diagnosing at least one system fault as recited in claim 1, wherein step (c) further includes the steps of:

(r) analyzing the identified patterns; and

(s) combining the idernified patterns with available physical system information to determine the fault relationship.

12. The method of diagnosing at least one system fault as recited in claim 1, further ineJuding the steps of:

(t) determining a threshold value associated with the at least one system operation characteristic of step (d);

(u) estimating a second system operating characteristic; and

(v) measuring a third system operating characteristic, wherein the threshold value and the second system operating characteristic are calculated and estimated respectively based upon the third system operating characteristic.

13. The method of diagnosing at least one system fault as recited in claim 12, wherein the second system operating characteristic is indicative of the at least one system failure mode associated with the at least one fault code.

14. The method of diagnosing at least one system fault as recited in claim 13 wherein step (e) further includes the step of:

(v) calculating an avenge value associated with the second system operating characteristic for a given sampling time; and

(w) monitoring an actual value of the second system operating characteristic.

15. The method of diagnosing at least one system fault as recited in claim 14, wherein step (e) further includes generating the at least one fault code when the actual value associated with the at least one system operating characteristic exceeds the threshold value and the actual value of the second system operating characteristic exceeds the average value.

16. The method of diagnosing as least one system fault as recited in claim 1, wherein the system is a vapor compression system.

17. The method of diagnosing as least one system fault as recited in claim 16, wherein the identified sensors in the vapor compression system include at least one pressure sensor, at least one temperature sensor, and at least one valve sensor.

18. A vapor compression system comprising:

at least one indoor unit including at least one indoor sensor and at least one indoor valve;

at least one outdoor unit in communication with the at least one indoor unit and including at least one outdoor sensor and at least one outdoor valve;

a control device for controlling the at least one indoor unit and the at least one outdoor unit, wherein the control device further includes;

a controller that collects and interprets data from the at least one indoor sensor and the at least one outdoor sensor and generates at least one fault code based on the data, wherein the at least one fault code is generated based upon analysis of the data under a full-fault condition and under a no-fault condition.

19. The system of claim 18, wherein the controller combines the data with available system information from the at least one indoor unit and the at least one outdoor unit to generate the at least one fault code.

20. The system of claim 18, wherein the controller receives a plurality of data inputs associated with a plurality of system operating characteristics, calculates an indicator based upon the plurality of system operating charactedstics, and generates the at least one fault code based upon a comparison of the indicator to a threshold.