Data processing: measuring – calibrating – or testing – Measurement system – Performance or efficiency evaluation
Reexamination Certificate
1998-09-15
2001-02-20
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system
Performance or efficiency evaluation
C702S183000, C702S185000, C714S005110, C714S040000
Reexamination Certificate
active
06192325
ABSTRACT:
TECHNICAL FIELD
The present invention relates generally to predictive maintenance of machines. More particularly, it relates to establishing and executing predictive maintenance programs for machines.
BACKGROUND
A highly specialized art, commonly referred to as “predictive maintenance”, has evolved in an effort to predict when a machine will fail or requires maintenance so that corrective measures can be taken on an “as needed” basis. In a typical predictive maintenance program, periodic inspections of the machinery are made in order to assess the current operating condition of the machine. Data collected during the periodic inspections is typically stored and trended over time as an additional measure of the machine's operational health. The periodic inspections often involve multiple technologies including oil sampling and analysis, infrared thermography analysis, and vibrational analysis. For example, vibrational analysis is usually accomplished with the aid of a portable instrument which senses and processes vibration generated by the machine. These portable instruments, which are often referred to as data collectors or data analyzers, typically include a vibration transducer attached to what is essentially a highly specialized hand-held computer. The maintenance technician places the vibration transducer against a predefined test point of the machine. In a typical application, the resultant machine vibration signal produced by the transducer is provided to the data collector where the data is processed (and perhaps analyzed) according to predefined conditions and stored for later downloading to a machine database which has been previously set up on a host computer. The host computer analyzes the vibration data for faults or other anomalous conditions and machine data is stored in the database.
Machines within a facility are typically monitored according to a route which is generated by a maintenance technician and programmed into the data collector with the aid of the host computer. The route typically includes a list of machines, measurement test points, and setup conditions for each test point. There are usually many machines in the route with many test points on each machine, and for each test point there may be specified a vibration frequency range to be analyzed, a type of analysis to be performed, a particular type or set of data to be measured and stored, and similar other data collection and analysis parameters. In response to commands from the user, the hand held instrument prompts the user for the identity of the machine and the test point to be monitored, and it automatically sets up the instrument, for example, to accept the specified frequency range for the test point, perform the specified analysis and store the specified type or set of data. A Fast Fourier Transform analysis may be performed on a pre-selected frequency range of the data and all or part of the resulting frequency spectrum may be stored and displayed. As the user progresses through the many machines and the corresponding test points, he collects and stores measured data which is subsequently transferred to the host computer for long term storage and further analysis.
In a typical predictive maintenance program as described above, the user or maintenance technician defines each test point for each machine, including the location for each test point, analysis parameters sets including the type and quantity of data to take for each test point, the type of analysis to be performed on the data, frequency ranges, alarm levels, and the like. The locations of the test points and the parameters or settings chosen for each test point vary depending on the type, size and combinations of machines. For example, a motor driving a fan may have different test points and settings than the same motor driving a pump. Defining a predictive maintenance program of this type for such a large number of different machines requires a great deal of expertise on the part of the maintenance technician. Such expertise is typically gained only through years of experience and/or extensive training including training on how to select a probe/sensor, which parameters to measure for a particular machine, locations corresponding to the most data rich points for measuring particular parameters, frequency ranges, analysis alarm limits, and other such information. The maintenance technician must define the measurements points, analysis parameter sets, and alarm limit sets, and he must integrate these items into a useable database setup. Further complicating the maintenance technician's task is the fact that knowledge and expertise gained through the experience of engineers and technicians is too often not shared among users and others skilled in the art. More common is the situation where expertise is lost as a result of job changes and the like. Even when expertise is documented in user's manuals or lab notebooks, a great deal of studying and understanding is required in order for the information to be put to proper use.
The difficulty and complexity of providing the settings for each measurement point may be more fully appreciated by considering the settings disclosed for measurement point examples in the following detailed description. While a relatively small number of examples are discussed herein, it will be understood that a typical predictive maintenance database will have hundreds of measurement points with each measurement point typically having multiple settings. In predictive maintenance databases involving vibration, each measurement point will typically have many settings, often 50 or more.
Clearly, the proficiency and skill required to define an acceptable predictive maintenance program for a machine is beyond the abilities of a layperson. Because of the complexity involved and the expertise needed to properly set up an adequate predictive maintenance program or model, data technicians have been known to simply use the default settings provided by vendors of data collection/analysis instruments.
SUMMARY
What is needed, therefore, is a simplified method and apparatus for creating a predictive maintenance database. The solution should include the integration of machine database setup with the use of multiple measurement technologies, and it should be capable of pooling the scattered expertise of predictive maintenance modelers and significantly reducing the level of skill and expertise needed to establish an adequate predictive maintenance program.
With regard to the foregoing and other objects, the invention in one aspect provides a method for establishing a predictive maintenance database that defines information needed to monitor components in accordance with a predictive maintenance plan. In the method, a component type corresponding to a particular component to be monitored is identified, and physical characteristic information is also identified which corresponds to the identified component type and the particular component to be monitored. A knowledge base in the computer defines relationships between monitoring practices, component types and physical characteristic information for component types, and an inference engine operates in part on the knowledge base to construct predictive maintenance databases. Information for a predictive maintenance database is constructed for each component to be monitored using the inference engine operating on the knowledge base, the selected component type and the selected physical characteristic information.
In accordance with the above method, the database information for the predictive maintenance database may be constructed in a number of ways based on a number of factors. For example, the database information may be constructed to correspond in part to a defined type of data to be measured for the identified component type and/or defined measurement points on the component from which data will be measured. An analysis parameter set may also be defined by the constructing step, including setup parameters for use by a data collection instrument to collect da
Glumac Miodrag
Hilemon Christopher G.
Piety Kenneth R.
Reeves Todd W.
Rich Michael D.
Bui Bryan
CSI Technology, Inc.
Hoff Marc S.
Luedeka Neely & Graham P.C.
LandOfFree
Method and apparatus for establishing a predictive... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for establishing a predictive..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for establishing a predictive... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2596490