Prediction system, prediction method, and program

Abstract

In order to enable more accurate prediction of an anomaly of a drain pump provided in an air conditioner, a prediction system includes an air conditioner including a drain pump and a controller, and the controller acquires data of a number of revolutions of the drain pump or a current value of the drain pump, and outputs a prediction result indicating that an anomaly of the drain pump is predicted, based on a change in data for a predetermined period in the data.

1 . A prediction system comprising:
 an air conditioner including a drain pump; and   a controller,   wherein the controller is configured to:
 acquire data of a number of revolutions of the drain pump or a current value of the drain pump; and 
 output a prediction result indicating that an anomaly of the drain pump is predicted, based on a change in data for a predetermined period in the data. 
   
     
     
         2 . The prediction system as claimed in  claim 1 , wherein the controller outputs the prediction result based on a representative value of the data for the predetermined period. 
     
     
         3 . The prediction system as claimed in  claim 1 , wherein the controller predicts the anomaly of the drain pump based on an average value of data for a first predetermined period in the data and an average value of data for a second predetermined period in the data, the second predetermined period being shorter than the first predetermined period. 
     
     
         4 . The prediction system as claimed in  claim 3 , wherein the controller outputs the prediction result indicating the anomaly of the drain pump when a divergence between the average value of the data for the first predetermined period and the average value of the data for the second predetermined period exceeds a threshold. 
     
     
         5 . The prediction system as claimed in  claim 1 , wherein the controller predicts the anomaly of the drain pump based on environmental data, the environmental data including temperature data or humidity data. 
     
     
         6 . The prediction system as claimed in  claim 1 , wherein when the air conditioner or the drain pump is not operating, the controller predicts the anomaly of the drain pump, using, instead of data for a period during which the air conditioner or the drain pump is not operating in the data, data before the period in the data. 
     
     
         7 . The prediction system as claimed in  claim 6 , wherein the controller determines the period during which the drain pump is not operating, based on environmental data, the environmental data including temperature data or humidity data. 
     
     
         8 . The prediction system as claimed in  claim 1 , wherein the controller determines, from environmental data, whether drain water is generated, the environmental data including temperature data or humidity data, and predicts the anomaly of the drain pump, excluding data for a period during which the drain water is not generated in the data of the number of revolutions of the drain pump or the current value of the drain pump. 
     
     
         9 . The prediction system as claimed in  claim 8 , wherein the controller further uses information indicating whether the air conditioner is operating in a predetermined mode to determine whether the drain water is generated. 
     
     
         10 . The prediction system as claimed in  claim 1 ,
 wherein the prediction system comprises an edge device configured to collect the data from the air conditioner, and   wherein the controller acquires data obtained by the air conditioner or the edge device averaging the data.   
     
     
         11 . The prediction system as claimed in  claim 1 , wherein the controller predicts the anomaly of the drain pump by using a learned prediction model obtained by performing machine learning using data when the drain pump is in a normal state in the data and data when the drain pump is in an anomaly state in the data as training data. 
     
     
         12 . The prediction system as claimed in  claim 1 , wherein the controller predicts the anomaly of the drain pump by using a learned prediction model obtained by performing machine learning using an image representing data when the drain pump is in a normal state in the data and an image representing data when the drain pump is an anomaly state in the data as training data. 
     
     
         13 . A prediction method in a prediction system including an air conditioner that includes a drain pump; and a controller, the prediction method comprising:
 acquiring, by the controller, data of a number of revolutions of the drain pump or a current value of the drain pump; and   outputting, by the controller, a prediction result indicating that an anomaly of the drain pump is predicted, based on a change in data for a predetermined period in the data.   
     
     
         14 . A non-transitory computer-readable recording medium having stored therein a program for causing a computer to perform a process in a prediction system including an air conditioner and a controller, the air conditioner including a drain pump, and the process including:
 acquiring data of a number of revolutions of the drain pump or a current value of the drain pump; and   outputting a prediction result indicating that an anomaly of the drain pump is predicted, based on a change in data for a predetermined period in the data.   
     
     
         15 . The prediction system as claimed in  claim 1 , wherein the controller generates a learned prediction model by performing machine learning using data when the drain pump is in a normal state in the data and data when the drain pump is in an anomaly state in the data as training data, and predicts the anomaly of the drain pump by using the generated learned prediction model.