Learning device, fibrous feedstock recycling apparatus, learning method, and storage medium that stably transports transport target matter

A fibrous feedstock recycling apparatus comprising:

• a web forming section which forms a web from a feedstock containing fibers;
• a web transport section which transports the web formed by the web forming section;
• and
• a control section, wherein
• the web transport section includes
• a first roller which transports the web and a second roller disposed downstream of the first roller in a transport path of the web, and
• a position detection section which detects a position of the web in a direction intersecting the transport path between the first roller and the second roller,
• the control section includes
• a rotation control section which controls at least one of a rotation speed of the first roller and a rotation speed of the second roller,
• a state variable acquisition section which acquires a state variable relating to a state of at least one of the web forming section and the web transport section, and
• a learning section which learns an operational expression for calculating a control value of the rotation control section based on a dataset containing the state variable acquired by the state variable acquisition section and a detection result of the position detection section, and
• the control section controls at least one of the first roller and the second roller using the rotation control section based on the control value calculated based on the operational expression.

A learning method using, as a learning target, a transporting apparatus provided with a first roller which transports web-like or sheet-like transport target matter and a second roller disposed downstream of the first roller in a transport path of the transport target matter, in which the transporting apparatus controls a rotation speed of the first roller and a rotation speed of the second roller, the method comprising:

• detecting a position of the transport target matter in a direction intersecting the transport path between the first roller and the second roller;
• acquiring a state variable which changes a state of the transport target matter;
• and
• learning an operational expression for calculating a control value of the rotation speeds of the first roller and the second roller based on a dataset containing the detected position of the transport target matter between the first roller and the second roller and the state variable.