NEWS
Data Mining for Maintenance Inspection


Makoto KIKUCHI
Senior Researcher,
Facilities Management Systems,
Transport Information Technology Div.


  Previously, the maintenance information relating to facilities has been managed manually in the field of railroad. However, in recent years there has been wider use of information systems to realize computerization of equipment ledgers and drawings, support of inspection planning and control of the history of inspection. These systems expedite the creation of database of the information hitherto manually written into inspection ledgers and charts. Therefore, it may be possible to review methods and periods of inspection and support construction work. Fig. 1 shows a diagram of the information system.

fig.1
Fig. 1 Diagram of maintenance inspection system

  In the field of the reliability engineering, the Weibull distribution is easily applicable to the failure rate and life of facilities. The downtime density function f(t) of the Weibull distribution is normally expressed as

()(^)(t^)m-1exp[|(t^)m]

where is the geometry parameter to decide the curve profile, and is the scale parameter to decide the time unit.

  In the above expression, the geometry parameter shows changes in the failure rate over time. Fig. 2 shows the geometry parameter for an equipment item estimated by analyzing inspection data to determine the changes in the failure rate over time and applying the Weibull distribution.

fig.2
Fig. 2 Changes in the failure rate over time

  As to the analysis of the inspection data, the "existence time analysis," one method of statistical analysis, is also used. Fig. 3 shows that the existence rate (probability of unfailed state) of equipment decreases over time under four different environments. We expect that inspection planning and replacement planning will be evaluated and optimized, if the failure rate of an equipment item can be estimated for each factor, such as the environment of installation, equipment type, manufacturer, operating condition, etc.

fig.3
Fig. 3 Reliability of facilities by existence time analysis