Abstract: Provided is a device for performing simulation for work schedule of a factory includes an experiment design module for generating experiment design data by selecting one or more factors from a plurality of factors included in reference information as target factors, and setting levels of the selected target factors, the reference information being an input value to a model that serves as a target of a what-if simulation experiment, and an experiment execution module for outputting an experiment execution result by executing the what-if simulation experiment for the model in parallel by using an executor including a plurality of execution engines based on the experiment design data.

Abstract: A manufacturing system state image preprocessing device in accordance with one embodiment of the present disclosure includes: a basic image generator configured to generate a basic image that visually represents a manufacturing system state based on manufacturing system state data, wherein the basic image includes a Gantt chart; and a conditioned image generator configured to generate one or more conditioned images in which one or more sets of conditions have been applied to the basic image, wherein an output image including at least one of the basic image and the one or more conditioned images is provided as input to an artificial neural network module.

Abstract: Provided is a dispatching method in a factory based on reinforcement learning. The dispatching method in a factory based on reinforcement learning may comprise: constructing a Markov decision process (MDP) for dispatching actions of a dispatcher in the factory and resulting rewards and states of the factory; performing learning by applying reinforcement learning (RL) to the constructed MDP; and as a result of said RL, selecting a job that maximizes a weighted sum of a plurality of scored dispatching rules.

Abstract: Provided is a method for resource planning in a factory based on simulations. The method for resource planning may comprise: modeling factory resources as capacity buckets; allocating a plurality of demands to the modeled capacity buckets; and, constructing factory resource planning by performing capacity bucket simulations (CBSs) based on the factory resources to which the plurality of demands are allocated.

Abstract: Provided is a dispatching method in a factory based on reinforcement learning. The dispatching method in a factory based on reinforcement learning may comprise: constructing a Markov decision process (MDP) for dispatching actions of a dispatcher in the factory and resulting rewards and states of the factory; performing learning by applying reinforcement learning (RL) to the constructed MDP; and as a result of said RL, selecting a job that maximizes a weighted sum of a plurality of scored dispatching rules.