Data processing: measuring – calibrating – or testing – Measurement system – Performance or efficiency evaluation
Reexamination Certificate
2001-02-16
2003-07-15
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system
Performance or efficiency evaluation
C709S224000, 38
Reexamination Certificate
active
06594618
ABSTRACT:
The present invention relates to a method for monitoring a complex system to provide decisional software tools for the monitoring, the quantified evaluation, the merging, and the anticipation of system performances.
Systems that can be associated with a computerized database, in which the current digital values of a group of digitized data, here called performance indicators, are stored at various time rates, are here considered.
The performance indicator database must be manageable by one of the many available database management systems (DBMS), such as, for example, ORACLE, IP21, etc. The database may be of “real time” type, but it may also be of “static” type with a periodic or interactive updating.
The system whose performances are to be analyzed may for example be an industrial production line in very varied sectors such as for example:
chemistry (product quality for a chemical reactor),
pharmacy (automation of validation tests for drug production),
agribusiness (yeast manufacturing),
automobile industry (quality of parts produced by a stamping process) . . .
For this first application range, the performance indicators may be the digital or qualitative results of technical controls and/or physical measurements regularly performed on the quality of an industrial product, such as the weight, the density, the chemical concentrations, the aspect characteristics (color, transparency, viscosity . . . ), the presence and the intensity of listed manufacturing defects, etc.
The system to be analyzed may also be formed of machine tool parks, of vehicle or airplane fleets, of various industrial installations, in varied sectors such as, for example:
aeronautics (centralized anticipative maintenance of an airplane fleet),
automobile industry (centralized anticipative maintenance of a vehicle fleet),
transport (anticipative maintenance of a subway, train or bus fleet),
electronic components (anticipative maintenance of the machine tool park),
remote control of processing plants (remote maintenance of liquid air micro-plants) . . .
For this second range of applications, the performance indicators may be the numerical or qualitative results of technical controls, periodic inspections, physical or mechanical measurements, failure corrections, technical overhauls, and more generally any operation performed regularly or exceptionally to ensure the good working order, the reliability, and the availability of the different operational units of the monitored fleet or machine park. The performance indicators will thus for example make an inventory of the frequencies of the types of failures or technical incidents, of the effective digital values of critical operational characteristics, physical indicators of operation normality, etc.
A third range of applications concerns the computerized monitoring of the quality of services on telecommunication networks. Let us mention, as examples, the service quality monitoring and prediction on mobile telephone networks, and the multi-node monitoring and prediction of the communication performances on Internet or Intranet computer networks.
For this third range of applications, the performance indicators may be the digital or qualitative values of performance measurements provided by physical sensors and real-time computer systems, which systems are generally managed by the part manufacturers of a mobile telephone system, or by the integrators in charge of the installation of a communication network. The performance indicators will thus record, for example, the frequencies of various types of failures, alarms or technical incidents, the effective digital values of critical operational characteristics such as information output rates, geographically sorted telephone call rates, localized rates of unserved calls, physical indicators of localized operation normality, etc.
In all these systems, in the present state of the art, a great number of indicators are monitored, the evolution of each indicator is examined, and statistics are attempted to be made on the values of these indicators. However, this requires the permanent intervention of an engineer in front of his computer and of the database to select the processing techniques, trigger and evaluate the results. For example, in the above-mentioned case of the monitoring of a telephone network, to maintain a permanent check on the sound working of the system, it may be desired, for a large city such as Paris, to monitor up to 1200 geographic areas. In each geographic area, same ten distinct indicators will for example be monitored, such as traffic indicators, indicators of the number of unprocessed calls, indicators of problem calls, indicators of interrupted calls . . . Reading each of these ten indicators every 5 minutes leads to having to analyze every five minutes the new values of close to 12,000 distinct indicators (which amounts to approximately 144,000 new values per hour). Thus, in practice, in current systems, a trained data system engineer can at the very most directly monitor only a small number of geographic areas.
Thus, either the number of monitored indicators, or the update frequency of these indicators is necessarily reduced, which results in a less precise and more difficult prediction of possible defects resulting from a system degradation. One of the major difficulties, which prevents the computation of efficient syntheses for groups of indicators, is especially linked to the fact that the various performance indicators are often measured in different physical units, varying in very different ranges of values. Further, for each indicator, the optimal values may be high values or low values, or values close to a set point.
Thus, an object of the present invention is to facilitate the intelligent monitoring of the performances of a system, especially to predict the possible degradation of these performances, in a simple and rapid manner, while enabling the use of a great number of indicators, possibly of totally different natures.
A second object of the present invention is to facilitate the intelligent and fast monitoring of very large databases which may include up to hundreds of thousands of performance indicators, by automatically extracting as small a number as possible of essential performance indicators, from which all other performance indicators can be automatically recalculated with a good precision.
A third object of the present invention is to compute, starting from any group of observed indicators, a new performance indicator merging in a self-adaptive and rigorous manner the performance information brought by each of the indicators of the considered group, to provide synthetic and precise information about the performances of a smaller or larger subset of the monitored system.
To achieve these objects, the present invention provides a method for monitoring a system based on a set of k performance indicators, Xj(t), each of which is defined at successive times t, j being an integer varying between 1 and k. This method includes, for each indicator Xj, the steps of performing an observation of a sequence of s values, s being an integer, of said indicator and reordering this sequence into a reference list ordered by increasing values; and determining the relative rank, R[Xj(t)], in said reference list of any new value Xj(t) of said indicator, this relative rank being equal to the rank of the new value divided by the number s.
According to an embodiment of the present invention, from each relative rank R[Xj(t)] a score Sj(t) expressed in the form of a percentage is determined, in such a way that said score becomes smaller whenever the performance becomes better.
According to an embodiment of the present invention, the score is expressed by formula Sj(t)=100{1−R[Xj(t)]/s} provided the indicator increases as the performance improves.
According to an embodiment of the present invention, it is further provided a method to determine from each relative rank R[Xj(t)] an information rate wj(t) proportional to the square root o
Charioui Mohamed
Hoff Marc S.
McDermott & Will & Emery
Miriad Technologies
LandOfFree
System monitoring method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System monitoring method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System monitoring method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3099095