Machine learning apparatus, air conditioning system, and machine learning method

A machine learning apparatus for learning at least one pair of:

• a pair of temperatures that are a temperature of cooling water supplied by a cooling water pump and a temperature of cold water supplied by a cold water pump;
• or a pair of flowrates that are a flowrate of the cooling water supplied by the cooling water pump and a flowrate of the cold water supplied by the cold water pump, in an air conditioning system including a water-cooled chiller, a cooling water pump configured to supply cooling water for cooling refrigerant by performing heat exchange in the water-cooled chiller, a cooling tower configured to bring the cooling water transferred from the water-cooled chiller into contact with outside air to cool the cooling water, an air-conditioning apparatus, and a cold water pump configured supply the cold water cooled by the refrigerant in the heat exchange performed in the water-cooled chiller to the air-conditioning apparatus,
• wherein the machine learning apparatus comprises:
• a memory;
• and
• a processor coupled to the memory and configured to:
• obtain state variables including an operation condition of the cooling tower, an operation condition of the air-conditioning apparatus, and a load of the air-conditioning apparatus;
• perform learning by associating the state variables with the at least one pair;
• and
• calculate a reward, based on a total value of energy consumptions of the cooling tower, the water-cooled chiller, the cooling water pump, the cold water pump, and the air-conditioning apparatus,
• wherein the processor is configured to:
• perform the learning by using the calculated reward, and
• control, based on the calculated reward, the air conditioning system to reduce power consumption.

A machine learning apparatus for learning at least one of a temperature or a flowrate of a thermal medium supplied by a pump, in an air conditioning system including a heat source apparatus configured to heat or cool the thermal medium, the pump configured to discharge the thermal medium heated or cooled by the heat source apparatus, and a heat exchanger configured to perform heat exchange between passing air and the thermal medium discharged from the pump, wherein the air-conditioning apparatus blows the air having passed through the heat exchanger to a target space,

• wherein the machine learning apparatus comprises:
• a memory;
• and
• a processor coupled to the memory and configured to:
• obtain state variables including an operation condition of the heat source apparatus, an operation condition of the air-conditioning apparatus, and a load of the air-conditioning apparatus;
• perform learning by associating the state variables with at least one of the temperature or the flowrate of the thermal medium;
• and calculate a reward, based on a total value of energy consumptions of the heat source apparatus, the air-conditioning apparatus, and the pump,
• wherein the processor is configured to:
• perform the learning by using the calculated reward, and
• control, based on the calculated reward, the air conditioning system to reduce power consumption.