Symptom Reliability and Hazard for Systems Condition Monitoring

Decomposition Of The Symptom Observation Matrix And Grey Forecasting In Vibration Condition Monitoring Of Machines

“…In this way, it is just ready to undertake the go/do not go diagnostic decision at the end of the bearing diagnostic test. This can be done by an operator or automatically, when applying the concept and calculation of symptom reliability and the symptom limit value S l (Cempel et al 2000;Natke, 2002;Cempel 1991).…”

“…So, we should determine the symptom limit value S l which enables us to do this safely. This limit value can base on some experimental practice, some standards (e.g., ISO), or it can be assessed by the new concept of symptom reliability R(S) (Natke, 1997;Cempel, 2000). Figure 7 shows this possibility of assessing the symptom limit value S l (the bottom-right panel), combined with another possibility of optimizing the dimension of the primary symptom observation space.…”

“…The application of singular value decomposition (SVD) to this set of diagnostic data enables us to observe the evolution of a few generalized faults of the diagnosed machine, starting from the fault of maximal severity. Applying next the concept of symptom reliability (Cempel et al 2000;Natke et al 1997;Cempel, 1991) to the so extracted generalized fault symptoms, one can calculate the symptom limit value S l , the basis for any diagnostic decision. However, the loadings of machines by production processes (or the environment) are not constant, so that the resulting symptom readings may have some disturbances influencing the assessment of the machine condition.…”

The Symptom Observation Matrix for Monitoring and Diagnostics

“…E the observation matrix contains the lifetime-dependent symptom vectors, which permits information condensation via the SVD, E the SVD results in generalized fault/damage indicators, E in order to perform, for example, reliability investigations, one can relate one limit value [9] to each generalized fault/damage instead of relating a limit value to each symptom as in other techniques for monitoring and diagnosis, E the matrix elements of W (equation (33)) serve as symptom assurance criteria (SAC), E the non-zero SVs, ordered by their magnitudes, are fault/damage ranking indices, E the magnitudes of the SVs describe fault/damage intensities, E the small SVs are most sensitive to system modi"cations (and to measuring distortions), E the (directions of the) left singular vectors characterize the system modi"cations during lifetime of the system. However, the right singular vectors are more sensitive than the left singular vectors with respect to modi"cations.…”

“…If no mathematical model of the system or its part with the detected modi"cation is available, reliability investigations [9] have to be performed based on the symptoms due to the test forces.…”