One embodiment of the present invention sets forth a technique for generating simulated training data for a physical process. The technique includes receiving, as input to at least one machine learning model, a first simulated image of a first object, wherein the at least one machine learning model includes mappings between simulated images generated from models of physical objects and real-world images of the physical objects. The technique also includes performing, by the at least one machine learning model, one or more operations on the first simulated image to generate a first augmented image of the first object. The technique further includes transmitting the first augmented image to a training pipeline for an additional machine learning model that controls a behavior of the physical process.

When operating a production site, it is preferable to be aware of the degree to which a production plan is improved as a result of performance improvements being reflected in a model which is used to generate the production plan. Provided is a system including a planning section that generates a production plan for controlling a production site during a target interval, using a planning model that has been updated to a newest state; a base planning section that generates a base plan to serve as a base of the production plan, using an unupdated planning model; and a tracking section that tracks improvement in the production plan caused by the update of the planning model, based on the production plan and the base plan.

A method includes receiving, via a first component in a production environment, a sensor measurement corresponding to a second component in the production environment. A first digital twin corresponding to the first component is identified, and a perception algorithm is applied to identify a component type associated with the second component. A second digital twin is selected based on the component type, and a third digital twin is selected that models interactions between the first digital twin and the second digital twin. The third digital twin is used to generate instructions for the first component that allow the first component to interact with the second component. The instructions may then be delivered to the first component.

A digital twin management system manages a virtual model that represents an actual physical system in a virtual space on a real-time basis. To generate an integrated virtual model by adding a second virtual model to a first virtual model, a processor of the digital twin system extracts multiple parts that can be used in common in the first virtual model and the second virtual model, generates multiple integrated virtual models that are candidates of an integrated virtual model by changing the extracted parts that can be used in common, calculates an evaluation of each of the generated integrated virtual models, and outputs configuration information regarding each of the integrated virtual model candidates and an evaluation of the integrated virtual model candidate in association with each other.

A replacement part for replacing an original mechanical machine part having has an original mechanical configuration with original part descriptive data is made by first receiving performance data obtained by monitoring the machine during operation with the original machine part with one or more sensors and then sending the performance data to a modeling server. The modeling server then calculates multiple optimized mechanical configurations of the replacement part with one or more modeling algorithms based on different optimization criteria using at least the original part descriptive data and the received performance data. Then a selection of several performance options representing the multiple mechanical configurations of the replacement part are provided, one of which is selected by a user. Finally a replacement part is made with the final optimized configuration corresponding to the selected performance option or sending optimized part descriptive or construction data with the final optimized configuration.

A method includes generating a queue of a plurality of operations in a sequence recipe, the plurality of operations being associated with substrate processing. The method further includes sorting the plurality of operations in the queue based on a plurality of completion times corresponding to the plurality of operations. The method further includes, for each operation of the plurality of operations in the queue, obtaining a next operation in the queue and setting a virtual time axis to time leap to a corresponding completion time of the next operation until each operation of the plurality of operations in the queue are completed to generate a schedule for the sequence recipe.

An evaluation apparatus includes a storage unit that stores a model modeling a state of a facility provided in a plant, a simulator that adjusts a parameter that is set in the model so that a difference between an actual measurement value based on a process value of the facility in a first state and a first simulate value calculated by using the model is equal to or less than a threshold, and an estimation unit that estimates a first estimated operating point that indicates an operation state of the facility in the first state based on the adjusted parameter.

A method includes receiving a plurality of operations in a sequence recipe. The plurality of operations are associated with processing a plurality of substrates in a substrate processing system. The method further includes identifying a plurality of completion times corresponding to the plurality of operations. Each completion time of the plurality of completion times corresponds to completion of a respective operation of the plurality of operations. The method further includes simulating the plurality of operations by setting a virtual time axis to each of the plurality of completion times to generate a schedule for the sequence recipe. The method further includes causing, based on the schedule, the plurality of substrates to be processed or performance of a corrective action.

Simulation methods for simulating a real control (2) for an industrial process, a plant or a machine shall be able to determine errors occurring in the course of simulation more easily. For this purpose the invention proposes that the simulation system (7) stores intermediate states during the simulation and time-stamps them, wherein a stored intermediate state can be reloaded into the simulation system (7) at a later time and a simulation carried out on the basis thereof. As a result, simulations do not always need to start with the beginning of the control program to be simulated.

One embodiment of the present invention sets forth a technique for generating simulated training data for a physical process. The technique includes receiving, as input to at least one machine learning model, a first simulated image of a first object, wherein the at least one machine learning model includes mappings between simulated images generated from models of physical objects and real-world images of the physical objects. The technique also includes performing, by the at least one machine learning model, one or more operations on the first simulated image to generate a first augmented image of the first object. The technique further includes transmitting the first augmented image to a training pipeline for an additional machine learning model that controls a behavior of the physical process.