Machine learning device, numerical control system, and machine learning method

A machine learning device for performing machine learning on a numerical control device which analyses a machining program, and when a first command including a corner portion, composed of two blocks in the machining program, generates a second command in which the two blocks are replaced with m or more blocks (m is a natural number) which is more than the two blocks, the machine learning device comprising:

• a hardware processor;
• and
• a non-transitory storage that stores a program, which when executed causes the hardware processor to:
• acquire state information including the first command, coordinate values of each block in the m or more blocks, and location information of a machining path which is obtained by performing the machining program by the numerical control device;
• output to the numerical control device, action information including adjustment information about the coordinate values of each block in the m or more blocks included in the state information;
• output a reward value in reinforcement learning based on an inward turning amount in the corner portion which is calculated by using the first command included in the state information and the location information of the machining path, and based on the machining time;
• update a value function based on the value of the reward outputted from the reward output unit, the state information and the action information;
• based on the updated value function, create and output adjustment information about coordinate values of each block in the m or more blocks to the numerical control device;
• and
• controlling the machining path based on the output adjustment information.