Integrating machine learning into control systems for industrial facilities

Abstract

Methods, systems, apparatus and computer program products for implementing machine learning within control systems are disclosed. An industrial facility setting slate can be received from a machine learning system and a determination can be made as to whether to adopt the settings in the industrial facility setting slate. The machine learning model can be a neural network, e.g., a deep neural network, that has been trained, e.g., using reinforcement learning to predict a data setting slate that is predicted to optimize an efficiency of a data center.

1 . A method comprising:
 receiving, from a machine learning system, an industrial facility setting slate that the machine learning system predicts will optimize an efficiency of an industrial facility, wherein the industrial facility settings slate defines a respective setting for each of a plurality of industrial facility controls;   determining whether industrial facility settings defined by the industrial facility setting slate can safely be adopted by the industrial facility; and   in response to determining that the industrial facility settings defined by the industrial facility setting slate can safely be adopted, adopting the industrial facility settings defined by the industrial facility setting slate.   
     
     
         2 . The method of  claim 1 , further comprising:
 in response to determining that the industrial facility settings cannot safely be adopted by the industrial facility, adopting settings provided by a default control system for the industrial facility.   
     
     
         3 . The method of  claim 2 , wherein the default control system is a rule-based control system. 
     
     
         4 . The method of  claim 1 , wherein determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted comprises:
 determining whether each of the industrial facility settings defined by the industrial facility setting slate falls within an acceptable range for the industrial facility setting.   
     
     
         5 . The method of  claim 1 , wherein determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted further comprises:
 determining whether predictions received from the machine learning system have become unstable.   
     
     
         6 . The method of  claim 5 , wherein determining whether predictions received from the machine learning system have become unstable comprises:
 determining, for each of the industrial facility controls, whether a rate of change of recently predicted settings for the industrial facility control has satisfied a threshold.   
     
     
         7 . The method of  claim 5 , wherein determining whether predictions received from the machine learning system have become unstable comprises:
 determining, for each of the industrial facility controls, whether a variance of recently predicted settings for the industrial facility control has satisfied a threshold.   
     
     
         8 . The method of  claim 1 , further comprising:
 prior to adopting the industrial facility settings, receiving state data characterizing a current state of the industrial facility; and   wherein determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted comprises:
 determining whether the current state of the industrial facility is suitable for adopting the industrial facility settings. 
   
     
     
         9 . The method of  claim 8 , wherein determining whether the current state of the industrial facility is suitable for adopting the industrial facility settings comprises determining whether any sensor readings by a sensor identified in the state data fall outside of an acceptable range for the sensor. 
     
     
         10 . The method of  claim 1 , further comprising:
 determining that no communications have been received from the machine learning system for more than a threshold amount of time; and   in response, controlling the industrial facility using a default control system for the industrial facility.   
     
     
         11 . The method of  claim 1 , further comprising:
 sending state data characterizing an updated state of the industrial facility to the machine learning system after the industrial facility settings have been adopted for use in generating a new predicted data setting slate.   
     
     
         12 . The method of  claim 1 , wherein the machine learning system includes a machine learning model that is a neural network. 
     
     
         13 . The method of  claim 12 , wherein the machine learning model is a deep neural network. 
     
     
         14 . The method of  claim 12 , wherein the neural network has been trained using reinforcement learning based on measured or calculated efficiency of the industrial facility. 
     
     
         15 . A system comprising:
 one or more computers; and   one or more storage devices storing instructions that are operable, when executed on one or more computers, to cause the one or more computers to perform operations comprising:   receiving, from a machine learning system, an industrial facility setting slate that the machine learning system predicts will optimize an efficiency of an industrial facility, wherein the industrial facility settings slate defines a respective setting for each of a plurality of industrial facility controls;   determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted by the industrial facility; and   in response to determining that the industrial facility settings defined by the industrial facility setting slate can safely be adopted, adopting the industrial facility settings defined by the industrial facility setting slate.   
     
     
         16 . The system of  claim 15 , the operations further comprising:
 in response to determining that the industrial facility settings cannot safely be adopted by the industrial facility, adopting settings provided by a default control system for the industrial facility.   
     
     
         17 . The system of  claim 15 , wherein determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted further comprises:
 determining whether predictions received from the machine learning system have become unstable.   
     
     
         18 . A computer program product comprising instructions that are executable by a processing device and upon such execution cause the processing device to perform operations of:
 receiving, from a machine learning system, an industrial facility setting slate that the machine learning system predicts will optimize an efficiency of an industrial facility, wherein the industrial facility settings slate defines a respective setting for each of a plurality of industrial facility controls;   determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted by the industrial facility; and   in response to determining that the industrial facility settings defined by the industrial facility setting slate can safely be adopted, adopting the industrial facility settings defined by the industrial facility setting slate.   
     
     
         19 . The computer program product of  claim 18 , the operations further comprising:
 in response to determining that the industrial facility settings cannot safely be adopted by the industrial facility, adopting settings provided by a default control system for the industrial facility.   
     
     
         20 . A device for controlling physical infrastructure in an industrial facility, the device comprising:
 a controller that performs operations comprising:
 receiving, from a machine learning system, an industrial facility setting slate that the machine learning system predicts will optimize an efficiency of an industrial facility, wherein the industrial facility settings slate defines a respective setting for each of a plurality of industrial facility controls; 
 determining whether the industrial facility settings defined by the industrial facility setting slate can safely be adopted by the industrial facility; and 
 in response to determining that the industrial facility settings defined by the industrial facility setting slate can safely be adopted, adopting the industrial facility settings defined by the industrial facility setting slate.