Building automation system with edge processing diversity

Abstract

A rooftop unit includes a housing, air conditioning components coupled to the housing, and circuitry enclosed within and/or coupled to the housing and programmed to execute a control logic for the air conditioning components, an expression-based event processing logic, and a machine learning algorithm.

What is claimed is: 
     
         1 . A rooftop unit comprising:
 a housing;   air conditioning components coupled to the housing; and   circuitry enclosed within and/or coupled to the housing and programmed to execute a control logic for the air conditioning components, an expression-based event processing logic, and a machine learning algorithm.   
     
     
         2 . The rooftop unit of  claim 1 , wherein the expression-based event processing logic performs pattern recognition for data received by the circuitry from the air conditioning components or one or more external data sources. 
     
     
         3 . The rooftop unit of  claim 1 , wherein the machine learning algorithm is based on a machine learning model trained at a cloud system remote from the rooftop unit. 
     
     
         4 . The rooftop unit of  claim 3 , wherein the machine learning algorithm comprises a modified version of the machine learning model trained at the cloud system that is configured to execute on more limited processing resources of the circuitry relative to the cloud system. 
     
     
         5 . The rooftop unit of  claim 1 , wherein the expression-based event processing logic diagnoses occurring fault conditions and the machine learning algorithm predicts future fault conditions. 
     
     
         6 . The rooftop unit of  claim 1 , wherein the circuitry is programmed to modify the expression-based event processing logic in response to remote updates received at the circuitry. 
     
     
         7 . The rooftop unit of  claim 1 , wherein the expression-based event processing logic and the machine learning algorithm have a combined memory footprint of less than 256 MB. 
     
     
         8 . The rooftop unit of  claim 1 , wherein the circuitry receives a first data set from the air conditioning components and a second data set from an external sensor, wherein the machine learning algorithm uses the first data set and the second data set as inputs. 
     
     
         9 . The rooftop unit of  claim 8 , wherein the external sensor is an indoor air quality sensor. 
     
     
         10 . The rooftop unit of  claim 1 , wherein:
 the circuitry is configured to established communications with a cloud system; and   the control logic for the air conditioning components, the expression-based event processing logic, and the machine learning algorithm are functional during interruptions of the communications with the cloud system.   
     
     
         11 . The rooftop unit of  claim 1 , wherein the control logic for the air conditioning components is a native control logic, and wherein the expression-based event processing logic comprises a supplement or modification of the native control logic. 
     
     
         12 . The rooftop unit of  claim 11 , wherein the supplement or modification of the native control logic is received from a cloud system or another computing system external from the rooftop unit via a network connection. 
     
     
         13 . The rooftop unit of  claim 12 , wherein the supplement or modification of the native control logic is received after installation of the rooftop unit while the rooftop unit is connected to the network connection and operational. 
     
     
         14 . The rooftop unit of  claim 1 , wherein the control logic comprises a first set of one or more fault detection and/or diagnostics rules, and wherein the expression-based event processing logic comprises a second set of one or more fault detection and/or diagnostics rules that supplement or modify the first set of one or more fault detection and/or diagnostics rules, the second set of one or more fault detection and/or diagnostics rules received from a cloud system or another source remote from the rooftop unit and defined according to an expression-based language. 
     
     
         15 . A unit of building equipment comprising:
 a mechanical component controllable to affect a condition of a building; and   circuitry packaged with the mechanical component and programmed to execute a control logic for the heating, ventilation, or cooling component, an expression-based event processing logic, and a machine learning algorithm.   
     
     
         16 . The unit of building equipment of  claim 15 , wherein the expression-based event processing logic performs pattern recognition for data received by the circuitry from the mechanical component or one or more external data sources. 
     
     
         17 . The unit of building equipment of  claim 15 , wherein the machine learning algorithm is based on a machine learning model trained at a cloud system remote from the unit of building equipment. 
     
     
         18 . The unit of building equipment of  claim 17 , wherein the machine learning algorithm comprises a modified version of the machine learning model trained at the cloud system that is configured to execute on more limited processing resources of the circuitry relative to the cloud system. 
     
     
         19 . The unit of building equipment of  claim 15 , wherein the expression-based event processing logic diagnoses occurring fault conditions and the machine learning algorithm predicts future fault conditions. 
     
     
         20 . The unit of building equipment of  claim 15 , wherein the circuitry is programmed to modify the expression-based event processing logic in response to remote updates received at the circuitry. 
     
     
         21 . The unit of building equipment of  claim 15 , wherein the expression-based event processing logic and the machine learning algorithm have a combined memory footprint of less than 256 MB. 
     
     
         22 . The unit of building equipment of  claim 15 , wherein the circuitry receives a first data set from the mechanical component and a second data set from an external sensor, wherein the machine learning algorithm uses the first data set and the second data set as inputs. 
     
     
         23 . The unit of building equipment of  claim 22 , wherein the external sensor is an indoor air quality sensor. 
     
     
         24 . The unit of building equipment of  claim 15 , wherein:
 the circuitry is configured to established communications with a cloud system; and   the control logic for the mechanical component, the expression-based event processing logic, and the machine learning algorithm are functional during interruptions of the communications with the cloud system.   
     
     
         25 . The unit of building equipment of  claim 15 , wherein the control logic for the heating, ventilation, or cooling component is a native control logic, and wherein the expression-based event processing logic comprises a supplement or modification of the native control logic. 
     
     
         26 . The unit of building equipment of  claim 25 , wherein the supplement or modification of the native control logic is received from a cloud system or another computing system external from the unit via a network connection. 
     
     
         27 . The unit of building equipment of  claim 26 , wherein the supplement or modification of the native control logic is received after installation of the unit while the unit is connected to the network connection and operational. 
     
     
         28 . The unit of building equipment of  claim 15 , wherein the control logic comprises a first set of one or more fault detection and/or diagnostics rules, and wherein the expression-based event processing logic comprises a second set of one or more fault detection and/or diagnostics rules that supplement or modify the first set of one or more fault detection and/or diagnostics rules, the second set of one or more fault detection and/or diagnostics rules received from a cloud system or another source remote from the unit and defined according to an expression-based language. 
     
     
         29 . A system comprising:
 a unit of building equipment comprising:
 a heating, ventilation, or cooling component; 
 onboard circuitry configured to execute a control logic for the heating, ventilation, or cooling component, an expression-based event processing logic, and a machine learning algorithm; and 
   a cloud system communicably connectable to the onboard circuitry and comprising circuitry configured to:
 transmit an expression to the onboard circuitry for use by the expression-based event processing logic; and 
 transmit a machine learning model to the onboard circuitry for use by the machine learning algorithm. 
   
     
     
         30 . The system of  claim 29 , wherein the expression-based event processing logic performs pattern recognition for data received at the onboard circuitry from the heating, ventilation, or cooling component or an external data source. 
     
     
         31 . The system of  claim 29 , wherein the cloud system is configured to generate the machine learning model by training a neural network on a training data set comprising historical data from at least one of the unit of building equipment or other building equipment and generating the machine learning model to transmit to the onboard circuitry using the neural network. 
     
     
         32 . The system of  claim 29 , wherein the expression-based event processing logic diagnoses occurring fault conditions and the machine learning algorithm predicts future fault conditions. 
     
     
         33 . The system of  claim 29 , wherein the expression-based event processing logic and the machine learning algorithm have a combined memory footprint of less than 256 MB. 
     
     
         34 . The system of  claim 29 , further comprising a plurality of external data sources providing a plurality of data streams to the onboard circuitry, wherein the expression-based event processing logic and the machine learning algorithm are adapted to use the plurality of data streams as inputs. 
     
     
         35 . The system of  claim 29 , wherein the control logic, the expression-based event processing logic, and the machine learning algorithm are fully functional during interruptions of a connection between the onboard circuitry and the cloud system. 
     
     
         36 . A method comprising:
 providing a package comprising a heating, ventilation, or cooling component and onboard circuitry;   executing, by the onboard circuitry, control logic to control the heating, ventilation, or cooling component;   executing, by the onboard circuitry, an expression-based event processing logic; and   executing, by the onboard circuitry, a machine learning algorithm.   
     
     
         37 . The method of  claim 36 , wherein the method further comprises:
 training, by a computing system remote from the onboard circuitry, a neural network on a training data set comprises historical data from at least one of the heating, ventilation, or cooling component or other heating, ventilation, or cooling components; and   generating the machine learning algorithm to transmit to the onboard circuitry using the neural network.   
     
     
         38 . The method of  claim 36 , wherein executing the expression-based event processing logic provides recognition of patterns in data received at the onboard circuitry from the heating, ventilation, or cooling component or another data source. 
     
     
         39 . The method of  claim 36 , wherein executing the expression-based event processing logic diagnoses an occurring fault condition and wherein executing the machine learning algorithm predicts a future fault condition. 
     
     
         40 . The method of  claim 36 , further comprising receiving, at the onboard circuitry and from a cloud system, a set of expressions and a machine learning model;
 wherein executing the expression-based event processing logic comprises using the set of expressions and executing the machine learning model comprises using the machine learning model.