FIELD: physics.
SUBSTANCE: invention relates to production and/or technical processes and process control, and more specifically to technical field of production/technical simulation and/or optimization of model/control parameters. Method includes obtaining (S1) a fully or partially acausal modular parameterized model of a production and/or technical process, comprising at least one physical sub-model and at least a neural network sub-model, including one or more parameters of the parameterized model. Method also comprises generating (S2) a system of differential equations based on a parameterized model and simulation (S3) of dynamic characteristics of one or more states of production and/or technical process in time based on system of differential equations. Method also includes application (S4) of automatic differentiation in reverse mode with respect to system of differential equations during simulation of production and/or technical process to generate an estimate, which is a calculation of a model of a production and/or technical process. Method also includes updating (S5) of at least one parameter of the process model in a fully or partially acausal modular parameterized process model for a production and/or a technical process based on a calculated estimate using a gradient descent or ascent procedure for use when creating control signals that control the operation of the production and/or technical process.
EFFECT: enabling adaptation of the technical model of production and/or technical processes, as well as improved control of production and/or technical processes.
16 cl, 11 dwg
