Abstract: In view that the daily operation cost of offshore drilling is quite high, one of the important research subjects is how to make use of big data and artificial intelligence technologies to improve the rate of penetration (ROP), so as to shorten the operation cycle and reduce the operation cost. In this paper, the big data information at drilling sites was collected, and mud logging data, wireline logging data and drilling fluid property were put into the neural network to calculate the initial predicted ROP. Then, the real-time global optimal solution was calculated by means of the optimization algorithm, and the real-time ROP optimization model based on machine learning method and optimization algorithm was established. Finally, the model was embedded into the visual system to guide the field operation, so as to realize ROP improvement. This technology was applied to the real-time drilling process by taking Well A of South China Sea PY Oilfield as the test well. The practice shows that this technology can not only improve ROP effectively, but is of reference significance to the digital development of oil fields.