Machine learning apparatus and method learning predicted life of power device, and life prediction apparatus and motor driving apparatus including machine learning apparatus

1. A machine learning apparatus that learns a predicted life of a power device of a motor driving apparatus that converts DC power into AC power by a switching operation of the power device to supply the AC power to an AC motor, the machine learning apparatus comprising:

a state observation processor to sample a state variable indicative of a state of the motor driving apparatus, the state variable having an effect on the life of the motor driving apparatus, the state variable being one or more of: data regarding a number of times of switching of the power device, data regarding a junction temperature of the power device, and data regarding at least one of: presence and absence of a failure of the power device;

wherein data regarding the number of times of switching of the power device is obtained by the state observation processor in response to receiving a switching command for ON/OFF driving of the power device generated by the motor driving apparatus;

wherein data regarding at least one of presence and absence of the failure of the power device is obtained by the state observation processor in response to receiving the switching command for the power device generated by the switching operation of the motor driving apparatus and AC power output from the power device, the AC power output being detected by a current detector;

wherein data regarding the junction temperature of the power device is obtained by the state observation processor based on: a temperature of a fin cooling the power device detected by a first temperature sensor, ambient air temperature of the motor driving apparatus detected by a second temperature sensor, a value of current supplied to the AC motor from the motor driving apparatus and detected by the current detector, and a value of voltage applied to the AC motor from the motor driving apparatus,

the state observation processor being in communication with the first temperature sensor and the second temperature sensor to receive the measured temperature of the fin cooling the power device and the ambient air temperature,

the state observation processor being in communication with the current detector to receive the AC power output and/or the current supplied to the AC motor detected by the current detector; and

a learning processor in communication with the state observation processor, the learning processor to learn the predicted life of the power device based on the state variable;

wherein the learning processor:

determines a function for computing the predicted life of the power device, each function being associated with a reward obtainable by the learning processor,

determines the function corresponding to a maximum value of the reward, and

changes one or more operating and/or environmental parameters of the motor driving apparatus based on the predicted life of the power device.

2. The machine learning apparatus according to claim 1 , wherein the learning processor is configured to:

compute a reward based on:

the number of times of switching of the power device; and

at least one of: the presence and absence of the failure of the power device; and

updates, based on the state variable and the reward, the function for computing the predicted life of the power device;

wherein the reward is decreased when at least one of:

the number of times of switching of the power device exceeds a specified number of times; and

a failure of the power device occurs under a state in which the number of times of switching of the power device does not exceed the specified number of times.

3. The machine learning apparatus according to claim 1 , wherein the learning processor is configured to perform:

an error computation to compute an error on the basis of the state variable.

4. The machine learning apparatus according to claim 1 , wherein the learning processor is configured to learn the predicted life of the power device in accordance with a training data set acquired for a plurality of motor driving apparatuses, the training data set comprising the state variable as an input and the predicted life of the power device as a result.

5. A life prediction apparatus for a power device in the motor driving apparatus, the life prediction apparatus comprising the machine learning apparatus according to claim 1 , wherein the life prediction apparatus further comprises:

a decision-making processor in communication with the learning processor, the decision-making processor computing the predicted life of the power device, on the basis of the function function updated by the learning processor at an end of learning, in response to an input of a present state variable.

6. The machine learning apparatus according to claim 2 , wherein the learning processor updates the function for computing the predicted life of the power device on the basis of the state variable and the reward in accordance with a neural network model.

7. The life prediction apparatus according to claim 5 , further comprising:

a notification processor in communication with the decision-making processor, wherein the notification processor:

receives the predicted life of the power device from the decision-making processor; and

provides the predicted life computed by the decision-making processor as an output.

8. The life prediction apparatus according to claim 5 ,

wherein the learning processor is configured to re-learn and update the predicted life of the power device in accordance with an additional training data set, the additional training data set comprising the state variable as an input and the predicted life of the power device as a result.

9. A motor driving apparatus comprising:

the life prediction apparatus according to claim 5 ;

a power supply in communication with the life prediction apparatus, the power supply converting DC power into AC power by a switching operation of the power device to supply the AC power to the AC motor;

wherein failure of the power device corresponds to a state when a value of current flowing into the AC motor is close to zero when the switching command has been provided to the power device, or when the value of current flowing into the AC motor does not correspond to an expected value of current based on the switching command provided to the power device.

10. The life prediction apparatus according to claim 7 , wherein the notification processor further provides to an operator:

information for prompting replacement of the power device or replacement of the motor driving apparatus on the basis of the predicted life computed by the decision-making processor.

11. The motor driving apparatus according to claim 9 , further comprising:

a temperature calculation processor in communication with the state observation processor, the temperature calculation processor calculating the junction temperature of the power device on the basis of the temperature of the fin cooling the power device, the ambient air temperature of the motor driving apparatus, the value of current supplied to the AC motor from the motor driving apparatus, and the value of voltage applied to the AC motor from the motor driving apparatus.

12. The motor driving apparatus according to claim 9 , further comprising:

a temperature sensor measuring the junction temperature of the power device.

13. A machine learning method that learns a predicted life of a power device of a motor driving apparatus and converts DC power into AC power by a switching operation of the power device to supply the AC power to an AC motor, the machine learning method comprising:

sampling a state variable indicative of a state of the motor driving apparatus, the state variable being one or more of:

data regarding the number of times of switching of the power device,

data regarding junction temperature of the power device, and

data regarding at least one of:

presence and absence of a failure of the power device;

wherein data regarding the number of times of switching of the power device is obtained in response to a switching command for ON/OFF driving of the power device generated by the motor driving apparatus;

wherein data regarding at least one of presence and absence of the failure of the power device is obtained in response to receiving the switching command for the power device generated by the switching operation of the motor driving apparatus and AC power output from the power device detected by a current detector; and

wherein data regarding a junction temperature of the power device is obtained based on: a temperature of a fin cooling the power device detected by a first temperature sensor, ambient air temperature of the motor driving apparatus detected by a second temperature sensor, a value of current supplied to the AC motor from the motor driving apparatus and detected by the current detector, and a value of voltage applied to the AC motor from the motor driving apparatus;

learning the predicted life of the power device in accordance with a training data set generated based on the state variable , the learning comprising:

determining a function for computing the predicted life of the power device, each function being associated with an obtainable reward

determining the function corresponding to a maximum value of reward; and

changing one or more operating and/or environmental parameters of the motor driving apparatus based on the predicted life of the power device.