Magnetic disk device, control method, and computer program product

A magnetic disk device comprising:

• a magnetic disk;
• an arm;
• a support member provided at a distal end of the arm;
• a magnetic head provided on the support member and configured to access the magnetic disk;
• a first actuator configured to rotate the arm;
• a second actuator configured to drive the support member;
• and
• one or more processors configured to implement a controller configured to perform seek control of the magnetic head, wherein
• the controller includes:
• a feedback controller configured to control a first command value for driving the first actuator such that an error between a position of the magnetic head and a target position is reduced;
• and
• an iterative learning controller configured to receive an input value based on the error and output a second command value for driving the second actuator by iterative learning,
• the feedback controller includes
• an observer configured to estimate a position of the magnetic head as a state, and output an estimated value of the position of the magnetic head, and
• the input value is a difference between a subtraction value and the estimated value, the subtraction value being obtained by subtracting a value corresponding to an amount of movement of the second actuator from the error.

A control method executed by a magnetic disk device,

• the magnetic disk device including:
• a magnetic disk;
• an arm;
• a support member provided at a distal end of the arm;
• a magnetic head provided on the support member and configured to access the magnetic disk;
• a first actuator configured to rotate the arm;
• a second actuator configured to drive the support member;
• and
• a controller configured to perform seek control of the magnetic head,
• the method comprising:
• by the controller,
• controlling a first command value for driving the first actuator such that an error between a position of the magnetic head and a target position is reduced;
• and
• receiving an input value based on the error and outputting a second command value for driving the second actuator by iterative learning, wherein the iterative learning includes:
• storing, in a memory, a plurality of pieces of learning data output from a learning filter in a plurality of sampling times of the seek control, the learning filter being a filter for the input value that has been input, and is a multi-rate phase delay filter configured to approximate an inverse characteristic of a closed-loop characteristic of the controller, and
• integrating values based on the plurality of pieces of learning data stored in the memory, to output the second command value.