Machine learning synthesis of formation evaluation data

A method comprising:

• receiving historical drilling mechanics data and historical formation evaluation data from one or more wells;
• training a machine learning model using the historical drilling mechanics data and the historical formation evaluation data to generate a trained machine learning model, wherein the trained machine learning model includes a generative adversarial network (GAN) machine learning model, wherein the GAN machine learning model includes a generator network and a discriminator network, wherein the generator network receives random noise as an input and generates synthetic formation evaluation data and the discriminator network receives the synthetic formation evaluation data generated by the generator network and the historical formation evaluation data as having been filtered via a filtering operation or preprocessed via a preprocessing operation;
• receiving drilling mechanics data collected using one or more surface sensors while drilling a well;
• synthesizing formation evaluation data for the well from the drilling mechanics data collected using the one or more surface sensors by processing the drilling mechanics data collected using the one or more surface sensors with the trained machine learning model;
• and
• controlling drilling of the well using the synthesized formation evaluation data by steering a drill, varying one or more operational parameters during drilling, or optimizing drilling parameter settings based upon the synthesized formation evaluation data.

A method comprising:

• receiving historical formation evaluation data collected from one or more similar wells to a well for which formation evaluation data is to be synthesized;
• training a generative adversarial network (GAN) machine learning model using the historical formation evaluation data to generate a trained GAN machine learning model, wherein the GAN machine learning model includes a generator network and a discriminator network, wherein the generator network receives random noise as an input and generates synthetic formation evaluation data and the discriminator network receives the synthetic formation evaluation data generated by the generator network and the historical formation evaluation data as having been filtered via a filtering operation or preprocessed via a preprocessing operation;
• synthesizing the formation evaluation data for the well using the trained GAN machine learning model;
• and
• controlling drilling of the well using the synthesized formation evaluation data by steering a drill, varying one or more operational parameters during drilling, or optimizing drilling parameter settings based upon the synthesized formation evaluation data.