Based on asynchronous delay tap sampling and artificial neural network statistical machine learning, a novel optical performance monitoring (OPM) technique is proposed. The signal is delay tap sampled to obtain two-dimensional histogram. Then the features of histograms are extracted to train the artificial neural networks. The outputs of trained neural network are used to monitor optical signal impairments. Simulations of optical signal-to-noise ratio, chromatic dispersion and polarization mode dispersion monitoring in 10 Gb/s nonreturn to zero code-on-off keying, 40 Gb/s optical doubinary code and return to zero-differential phase shift keying systems are presented. The simulation results show that the proposed scheme can monitor multiple simultaneous impairments on optical signals of diverse bit rates and formats with high accuracy, from which the monitoring error is less than 5%. The proposed technique is simple, cost-effective and suitable for in-service distributed OPM.