System and method for generating alerts through multi-variate data assessment

• US 6,216,066 B1
• Filed: 07/01/1998
• Issued: 04/10/2001
• Est. Priority Date: 07/01/1998
• Status: Expired due to Term

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Goebel, Kai Frank, Doel, David Lacey
• Primary Examiner(s): Cuchlinski, Jr., William A.
• Assistant Examiner(s): ARTHUR JEANGLAUDE, GERTRUDE
• Application Number:
  US09/108,359
• Time in Patent Office: 1,014 Days
• Field of Search: 701/14, 701/15, 701/16, 701/29, 701/33, 701/35, 701/99, 701/100, 702/179, 702/185, 706/900, 340/438, 340/439, 340/425.5, 340/426, 73/112, 73/117.1, 244/158 R
• US Class Current: 701/29
• CPC Class Codes: G05B 23/0232 based on qualitative trend analysis, e.g. system evolution
• CPC Class Codes: G05B 23/0262 Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred
• CPC Class Codes: G06F 18/23213 with fixed number of clusters, e.g. K-means clustering
• CPC Class Codes: G06F 18/2433 Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
• CPC Class Codes: G07C 3/00 Registering or indicating the condition or the working of machines or other apparatus, other than vehicles

Abstract

This invention provides a system and method for generating alerts through multi-variate assessment. A normalizer normalizes data obtained from a process. A classifier classifies the correlated data in a multi-dimensional space defined for the variables in the process. The normalized data are classified into a normal cluster indicative of normal operating conditions and at least one alert cluster each indicative of alert conditions. An alert verifier verifies correlated data classified as an alert condition. In another embodiment of this invention, a tracker is used for addressing drifting data sets that arise in the process over time which allows the adaptation of classifiers to the shifted state. In a third embodiment, the multi-variate generator is used in parallel with a trend performance analysis tool to validate alerts generated therefrom.

Claims

1. A system for generating an alert from data obtained from a process, comprising:

a normalizer for normalizing the data;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of alert conditions in the process; and

an alert verifier for verifying data classified as an alert condition.

2. The system according to claim 1, wherein said normalizer comprises means for scaling the data according to normalization parameters.

3. The system according to claim 1, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

4. The system according to claim 3, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

5. The system according to claim 4, wherein the shape of the normal cluster and the at least one alert cluster in the multi-dimensional variable space is an irregularly shaped hyper-ellipsoid.

6. The system according to claim 1, wherein the classification of normalized data into the normal cluster and the at least one alert cluster is based on the degree of confidence that the data belong to the normal cluster and the at least one alert cluster.

7. The system according to claim 1, wherein the alert condition is representative of at least one of a failure, a fault condition, or symptoms of an incipient failure condition.

8. The system according to claim 1, wherein said alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

9. The system according to claim 8, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

10. A system for generating an alert from data obtained from an aircraft engine process, comprising:

a normalizer for normalizing data obtained from said aircraft engine process;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

an alert verifier for verifying data classified as an alert condition.

11. The system according to claim 10, wherein said normalizer comprises means for scaling the data according to normalization parameters.

12. The system according to claim 10, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

13. The system according to claim 12, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

14. The system according to claim 13, wherein the shape of the normal cluster and the at least one alert cluster in the multi-dimensional variable space is an irregularly shaped hyper-ellipsoid.

15. The system according to claim 10, wherein the classification of normalized data into the normal cluster and the at least one alert cluster is based on the degree of confidence that the data belong to the normal cluster and the at least one alert cluster.

16. The system according to claim 10, wherein the alert condition is representative of at least one of a failure, a fault condition, or symptoms of an incipient failure condition.

17. The system according to claim 10, wherein the alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

18. The system according to claim 17, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

19. The system according to claim 10, wherein the multi-dimensional variable space comprises engine variables comprising delta exhaust gas temperature, fuel flow, core speed, and exhaust gas temperature divergence.

20. A method for generating an alert from data obtained from a process, comprising:

normalizing the data;

classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of an alert condition in the process; and

verifying the data classified as an alert condition.

21. The method according to claim 20, wherein said normalizing comprises scaling the data according to normalization parameters.

22. The method according to claim 20, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

23. The method according to claim 22, wherein boundaries between the normal cluster and the alert cluster are non crisp.

24. The method according to claim 23, wherein the shape of the normal cluster and the at least one alert cluster in the multi-dimensional variable space is an irregularly shaped hyper-ellipsoid.

25. The method according to claim 20, wherein the classification of normalized data into the normal cluster and the at least one cluster is based on the degree of confidence that the data belongs to the normal cluster and the at least one cluster.

26. The method according to claim 20, wherein the alert condition is representative of at least one of a failure, a fault condition, or symptoms of an incipient failure condition.

27. The method according to claim 20, wherein said verifying an alert comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

28. The method according to claim 27, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been issued.

29. A method for generating an alert from data obtained from an aircraft engine process, comprising:

normalizing the data obtained from said aircraft engine process;

classifying the normalized data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

verifying data classified as an alert condition.

30. The method according to claim 29, wherein said normalizing comprises scaling the data according to normalization parameters.

31. The method according to claim 30, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

32. The method according to claim 31, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

33. The method according to claim 32, wherein the shape of the normal cluster and the at least one alert cluster in the multi-dimensional variable space is an irregularly shaped hyper-ellipsoid.

34. The method according to claim 29, wherein the classification of normalized data into the normal cluster and the at least one alert cluster is based on the degree of confidence that the data belong to the normal cluster and the at least one alert cluster.

35. The method according to claim 29, wherein the alert condition is representative of at least one of a failure, a fault condition, or symptoms of an incipient failure condition.

36. The method according to claim 29, wherein said verifying comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

37. The method according to claim 36, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been generated.

38. The method according to claim 29, wherein the multi-dimensional variable space comprises engine variables comprising delta exhaust gas temperature, fuel flow, core speed, and exhaust gas temperature divergence.

39. A system for generating an alert from data obtained from a process, comprising:

a normalizer for normalizing the data;

a tracker for tracking the normalized data for drifts that arise over time in the process;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of an alert condition in the process; and

an alert verifier for verifying data classified as an alert condition.

40. The system according to claim 39, wherein said normalizer comprises means for scaling the data according to normalization parameters.

41. The system according to claim 39, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

42. The system according to claim 41, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

43. The system according to claim 39, wherein said alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

44. The system according to claim 43, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

45. The system according to claim 39, wherein the tracker comprises means for adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

46. The system according to claim 45, wherein the classifier classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

47. A system for generating an alert from data obtained from an aircraft engine process, comprising:

a normalizer for normalizing data obtained from said aircraft engine process;

a tracker for tracking the normalized data for drifts that arise over time in the aircraft engine process;

a classifier for classifying the data in a multi-dimensional space defined for variables in the aircraft engine process, the data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

an alert verifier for verifying data classified as an alert condition.

48. The system according to claim 47, wherein said normalizer comprises means for scaling the data according to normalization parameters.

49. The system according to claim 47, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

50. The system according to claim 49, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

51. The system according to claim 47, wherein the alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

52. The system according to claim 51, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

53. The system according to claim 47, wherein the multi-dimensional variable space comprises engine variables comprising delta exhaust gas temperature, fuel flow, core speed, and exhaust gas temperature divergence.

54. The system according to claim 47, wherein the tracker comprises means for adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

55. The system according to claim 54, wherein the classifier classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

56. A method for generating an alert from data obtained from a process, comprising:

normalizing the data;

tracking the normalized data for drifts that arise over time in the process;

classifying the data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of an alert condition in the process; and

verifying the data classified as an alert condition.

57. The method according to claim 56, wherein said normalizing comprises scaling the data according to normalization parameters.

58. The method according to claim 56, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

59. The method according to claim 58, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

60. The method according to claim 56, wherein said verifying an alert comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

61. The method according to claim 60, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been issued.

62. The method according to claim 56, wherein said tracking comprises adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

63. The method according to claim 62, wherein said classifying classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

64. A method for generating an alert from data obtained from an aircraft engine process, comprising:

normalizing data obtained from said aircraft engine process;

tracking the normalized data for drifts that arise over time in the aircraft engine process;

classifying the data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

verifying data classified as an alert condition.

65. The method according to claim 64, wherein said normalizing comprises scaling the data according to normalization parameters.

66. The method according to claim 64, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

67. The method according to claim 66, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

68. The method according to claim 64, wherein said verifying comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

69. The method according to claim 68, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been generated.

70. The method according to claim 64, wherein the multi-dimensional variable space comprises engine variables comprising delta exhaust gas temperature, fuel flow, core speed, and exhaust gas temperature divergence.

71. The method according to claim 64, wherein said tracking comprises adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

72. The method according to claim 71, wherein said classifying classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

73. A system for validating an alert generated from a trend performance analysis tool used to monitor data obtained from a process, comprising:

a normalizer for normalizing the data monitored by the trend performance analysis tool;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of alert conditions in the process; and

an alert verifier for verifying data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

74. The system according to claim 73, wherein said normalizer comprises means for scaling the data according to normalization parameters.

75. The system according to claim 73, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

76. The system according to claim 75, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

77. The system according to claim 73, wherein the classification of normalized data into the normal cluster and the at least one alert cluster is based on the degree of confidence that the data belong to the normal cluster and the at least one alert cluster.

78. The system according to claim 73, wherein said alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

79. The system according to claim 78, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

80. A system for validating an alert generated from a trend performance analysis tool used to monitor data obtained from an aircraft engine process, comprising:

a normalizer for normalizing data obtained from said aircraft engine process and monitored by the trend performance analysis tool;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

an alert verifier for verifying data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

81. The system according to claim 80, wherein said normalizer comprises means for scaling the data according to normalization parameters.

82. The system according to claim 80, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

83. The system according to claim 82, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

84. The system according to claim 80, wherein the classification of normalized data into the normal cluster and the at least one alert cluster is based on the degree of confidence that the data belong to the normal cluster and the at least one alert cluster.

85. The system according to claim 80, wherein the alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

86. The system according to claim 85, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

87. A method for validating an alert generated from a trend performance analysis tool used to monitor data obtained from a process, comprising:

normalizing the data monitored by the trend performance analysis tool;

classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of an alert condition in the process; and

verifying the data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

88. The method according to claim 87, wherein said normalizing comprises scaling the data according to normalization parameters.

89. The method according to claim 87, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

90. The method according to claim 89, wherein boundaries between the normal cluster and the alert cluster are non crisp.

91. The method according to claim 87, wherein the classification of normalized data into the normal cluster and the at least one cluster is based on the degree of confidence that the data belongs to the normal cluster and the at least one cluster.

92. The method according to claim 87, wherein said verifying an alert comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

93. The method according to claim 92, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been issued.

94. A method for validating an alert generated from a trend performance analysis tool used to monitor data obtained from an aircraft engine process, comprising:

normalizing the data obtained from said aircraft engine process and monitored by the trend performance analysis tool;

classifying the normalized data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

verifying data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

95. The method according to claim 94, wherein said normalizing comprises scaling the data according to normalization parameters.

96. The method according to claim 94, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

97. The method according to claim 96, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

98. The method according to claim 94, wherein the classification of normalized data into the normal cluster and the at least one alert cluster is based on the degree of confidence that the data belong to the normal cluster and the at least one alert cluster.

99. The method according to claim 94, wherein said verifying comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

100. The method according to claim 99, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been generated.

101. A system for validating an alert generated from a trend performance analysis tool used to monitor data obtained from a process, comprising:

a normalizer for normalizing the data monitored by the trend performance analysis tool;

a tracker for tracking the normalized data for drifts that arise over time in the process;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundarY between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of alert conditions in the process; and

an alert verifier for verifying data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

102. The system according to claim 101, wherein said normalizer comprises means for scaling the data according to normalization parameters.

103. The system according to claim 101, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

104. The system according to claim 103, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

105. The system according to claim 101, wherein said alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

106. The system according to claim 105, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

107. The system according to claim 101, wherein the tracker comprises means for adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

108. The system according to claim 107, wherein the classifier classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

109. A system for validating an alert generated from a trend performance analysis tool used to monitor data obtained from an aircraft engine process, comprising:

a normalizer for normalizing data obtained from said aircraft engine process and monitored by the trend performance analysis tool;

a tracker for tracking the normalized data for drifts that arise over time in the process;

a classifier for classifying the normalized data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

an alert verifier for verifying data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

110. The system according to claim 109, wherein said normalizer comprises means for scaling the data according to normalization parameters.

111. The system according to claim 109, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

112. The system according to claim 111, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

113. The system according to claim 109, wherein the alert verifier comprises means for generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

114. The system according to claim 113, wherein said alert verifier verifies an alert condition after a predetermined number of vigilance level flags have been generated.

115. The system according to claim 109, wherein the tracker comprises means for adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

116. The system according to claim 115, wherein the classifier classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

117. A method for validating an alert generated from a trend performance analysis tool used to monitor data obtained from a process, comprising:

normalizing the data monitored by the trend performance analysis tool;

tracking the normalized data for drifts that arise over time in the process;

classifying the normalized data in a multi-dimensional space defined for variables in the process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for the process and the at least one alert cluster is indicative of an alert condition in the process; and

verifying the data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

118. The method according to claim 117, wherein said normalizing comprises scaling the data according to normalization parameters.

119. The method according to claim 117, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

120. The method according to claim 119, wherein boundaries between the normal cluster and the alert cluster are non crisp.

121. The method according to claim 117, wherein said verifying an alert comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

122. The method according to claim 121, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been issued.

123. The method according to claim 117, wherein said tracking comprises adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

124. The method according to claim 123, wherein said classifying classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.

125. A method for validating an alert generated from a trend performance analysis tool used to monitor data obtained from an aircraft engine process, comprising:

normalizing the data obtained from said aircraft engine process and monitored by the trend performance analysis tool;

tracking the normalized data for drifts that arise over time in the process;

classifying the normalized data in a multi-dimensional space defined for variables in the aircraft engine process, the normalized data being classified into a normal cluster and at least one alert cluster defined in the multi-dimensional variable space, wherein each cluster has a boundary between each other and the normal cluster is indicative of normal operating conditions for said aircraft engine process and the at least one alert cluster is indicative of an alert condition for said aircraft engine process; and

verifying data classified as an alert condition, wherein the verified alert condition is an indication that the alert generated from the trend performance analysis tool is valid.

126. The method according to claim 125, wherein said normalizing comprises scaling the data according to normalization parameters.

127. The method according to claim 125, wherein the normal cluster and the at least one alert cluster are of non-uniform and non-linear degrading size and shape.

128. The method according to claim 127, wherein boundaries between the normal cluster and the at least one alert cluster are non crisp.

129. The method according to claim 125, wherein said verifying comprises generating a vigilance level flag after a suspicious data reading, the vigilance level flag increasing as the number of consecutive suspicious data reading increases.

130. The method according to claim 129, wherein said verifying verifies an alert condition after a predetermined number of vigilance level flags have been generated.

131. The method according to claim 125, wherein said tracking comprises adapting the locations and shapes of the normal cluster and the at least one cluster in the multi-dimensional variable space in accordance with the drifting data.

132. The method according to claim 131, wherein said classifying classifies incoming data in accordance with the adapted locations and shapes of the normal cluster and the at least one alert cluster.