Method and system for predictive diagnosing of system reliability problems and/or system failure in a physical system

A method for the predictive diagnosing of system reliability problems and/or system failure in a physical system through the predictive identification of errant fluctuations in one or more operating parameters of said physical system, comprising:

• determining the values of at least one kind of operating parameter of said physical system by direct or indirect measurements of the values of said operating parameters at a plurality of time intervals;
• forming a data stream by arraying said parameter values in sequence;
• dividing said data stream into one or more sets;
• associating said sets into one or more series,
• processing at least one of said sets by determining the w value of said set by Formula 2 of
• w=(square root of n.sub.g)(M-m.sub.g)/s.sub.g
• wherein n.sub.g is the number of said sets in said set's series, m.sub.s is the mean of said parameter values within said set, s.sub.g is an at-large-group standard deviation, and M is a parent mean or a normalized mean, wherein said at-large-group standard deviation is determined by Formula 3 of
• s.sub.g =square root of [(d.sub.gt.sup.2 -[d.sub.gc ].sup.2 /n.sub.g)/(n.sub.g -1)]
• wherein d.sub.gt.sup.2 is the cumulative squares of the individual d.sub.g values of said sets in said set's series, wherein d.sub.gc is the cumulative totals of the individual d.sub.g values of said sets in said set's series, wherein d.sub.g is the difference between m.sub.s and the mean of the m.sub.s values of said sets in said set's series, and wherein n.sub.g is as defined above in Formula 2, and
• determining for said set whether an N0, Nx or N1 condition exists, wherein an N0 condition exists if w is less than "nx" wherein "nx is the w value consistent with x% probability of no significant difference between M and m.sub.s on the w Table or the computational equivalent of the w Table for n.sub.g number of parameter values, wherein an Nx condition exists if w is greater than nx but less than ny wherein ny is the w value consistent with y% probability of no significant difference between M and m.sub.s on the w Table for n.sub.g number of parameter values and wherein an N1 condition exists if w is greater than ny, wherein x and y are numbers and x is numerically greater than y;
• and
• wherein said set's series encompasses a data stream of at least three values.

A system for the predictive diagnosing of system reliability problems and/or system failure in a physical system through the predictive identification of errant fluctuations in one or more operating parameters of said physical system, comprising:

• a means for determining the values of at least one kind of operating parameter of said physical system by direct or indirect measurements of the values of said operating parameters at a plurality of time intervals;
• a means for forming a data stream by arraying said parameter values in sequence;
• a means for dividing said data stream into one or more sets each containing a plurality of said parameter values and an equal number of said parameter values;
• a means for associating said sets into one or more series,
• a means for processing a plurality of said sets separately by Formula 2.1 of
• w=(square root of n.sub.g)(M-m.sub.s)/(s.sub.g or s.sub.p)
• wherein n.sub.g is the number of said sets in said set's series, m.sub.s is the mean of said parameter values within said set, s.sub.g is an at-large-group standard deviation, s.sub.p is a parent-group standard deviation and M is a parent-group mean or a normalized mean, wherein said at-large-group standard deviation is determined by Formula 3 of
• s.sub.g -square root of [(d.sub.gt.sup.2 -[d.sub.gc ].sup.2 /n.sub.g)/(n.sub.g -1)]
• wherein d.sub.gt.sup.2 is the cumulative squares of the individual d.sub.g values of said sets in said set's series, wherein d.sub.gc is the cumulative totals of the individual d.sub.g values of said sets in said set's series, wherein d.sub.g is the difference between m.sub.s and the mean of the m.sub.s values of said sets in said set's series, and wherein n.sub.g is as defined above in Formula 2.1, and
• a means for determining for said set whether an N0, Nx or N1 condition exists, wherein an N0 condition exists if w is less than "nx" wherein "nx is the w value consistent with x% probability of no significant difference between M and m.sub.s on the w Table or via a computational equivalent of the w Table for n.sub.Q number of parameter values, wherein an Nx condition exists if w is greater than nx but less than ny wherein ny is the w value consistent with y% probability of no significant difference between M and m.sub.s on the w Table or via a computational equivalent of the w Table for n.sub.Q number of parameter values, and wherein an N1 condition exists if w is greater than ny, wherein x and y are numbers and x is numerically greater than y and wherein n.sub.Q number of parameter values is n.sub.g number of parameter values when s.sub.g is used in Formula 2.1 and wherein n.sub.Q number of parameter values is greater than 120 parameter values when s.sub.p is used in Formula 2.1, and
• wherein said set's series encompasses a section of said data stream comprising at least three sets each including at least one of said parameter values when s.sub.g is used in Formula 2.1 and wherein said set's series encompasses a section of said data stream comprising at least one set including at least one of said parameter values when s.sub.p is used in Formula 2.1.