Detection of spikes and faults in vibration trend data

An apparatus for monitoring machines comprising:

• at least one sensor for sensing physical parameters produced by the machines and for producing a series of scalar values corresponding to the sensed physical parameters;
• an analyzer connected to the at least one sensor and, for at least one of the sensors, operating to:
• receive the scalar values and produce at least one database based on the scalar values;
• compare each scalar value in the database to a predetermined alarm limit;
• identify each scalar value that exceeds the predetermined alarm as a potential corrected fault;
• for each potential corrected fault, identify a first prior scalar value that is immediately prior to the potential corrected fault, and compare the first prior scalar value to a predetermined warning level;
• for each potential corrected fault, identify first and second trailing scalar values that are immediately after the potential corrected fault, and compare the first and second trailing scalar values to a predetermined advise level;
• and
• for each potential corrected fault, if the first prior scalar value exceeds the predetermined warning level and the first and second trailing scalar values are less than the predetermined advise level, identify the potential corrected fault as an actual corrected fault.

An apparatus for monitoring machines comprising:

• at least one sensor for sensing physical parameters produced by the machines and for producing a series of scalar values corresponding to the sensed physical parameters;
• and an analyzer connected to the at least one sensor and, for each of the sensors, operating to:
• receive the scalar values,
• produce a first database based on the scalar values,
• determine a median value of the scalar values for each sensor, and
• determine a spike level that is a predetermined multiple of the median value;
• and
• a spike filter implemented in the analyzer and, for each sensor, operating to:
• compare the scalar values to the spike level,
• identify a particular scalar value as a potential spike when the particular scalar value differs from the median value by an amount that is equal to or greater than the spike level;
• identify first and second side scalar values, the first side scalar value being immediately before the potential spike and the second side scalar value being after the potential spike,
• compare the first and second side scalar values to the median value,
• identify the potential spike as an actual spike if the first and second side values are within a predetermined range of the median value for the sensor, and
• produce a second database corresponding to the first database with the actual spikes eliminated, whereby a plurality of second databases are produced by the spike filter with one of the second databases being produced for each of the sensors.