The increase in certain aspects of non-linear system output with the addition of noise to the system input is called stochastic resonance. The possibility of producing and using stochastic resonance to enhance low-level components of signals was investigated. Both a two-component model signal and actual vibration signals were used. The signal analysis system constructed consisted of a multi-component signal, a random or noise component added to the signal and a threshold for signal recognition. The parts of the noise plus original signal augmented signal that exceeded a threshold formed a new signal that was characterised in terms of the magnitudes of signal components. The effects of added noise level on the new signal were ascertained. Stochastic resonance was demonstrated as enhancement of low-level components of the model and actual vibration signals with the effect being dependent on the level of added noise.