MADDPG-based magnetotelluric data inversion method

• Main Authors: ZHANG BO, LING HONG, YIN CHANGCHUN, LIU YUNHE, REN XIUYAN, SU YANG, WANG LUYUAN
• Format: Patent
• Language: Chinese; English
• Published: 20.01.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a magnetotelluric data inversion method based on MADDPG, and the method comprises the following steps: obtaining observation electromagnetic data, and extracting observation geoelectric parameter data participating in inversion from the observation electromagnetic data; defining an electromagnetic data inversion problem as a reinforcement learning problem, including: constructing a single agent for each type of geoelectric parameters, regarding an environment state of an agent exploration space as predicted geoelectric parameter data obtained by forward modeling of a geologic model, regarding an output decision action of each agent as an update quantity of each type of geoelectric parameter data, and determining the geoelectric parameters of the geologic model; constructing a reward value according to a fitting difference between the predicted geoelectric parameter data and the observed geoelectric parameter data; based on the defined reinforcement learning problem, the MADDPG algorith