Systems and methods for determining disaggregated energy consumption based on limited energy billing data

Abstract

Various embodiments of the present disclosure can include systems, methods, and non-transitory computer readable media configured to train a Bayesian network model based on a given set of data. Information associated with a user can be received. The information can include aggregated energy consumption data at one or more low frequency time intervals. At least a portion of the information can be inputted into the Bayesian network model. A plurality of energy consumption values for a plurality of energy consumption sources associated with the user can be inferred based on inputting the at least the portion of the information into the Bayesian network model.

What is claimed is: 
     
         1 . A method of determining energy consumption, comprising:
 training a network model, wherein the trained network model represents probabilistic relationships and dependencies between (i) a plurality of user features and (ii) a plurality of external features that are independent of or agnostic to the user features; and   using the trained network model to infer or estimate disaggregated energy consumption values for a plurality of energy consumption sources for a plurality of customers, based on limited amounts of data received at one or more low frequency intervals.   
     
     
         2 . The method of  claim 1 , wherein the plurality of user features include household or business properties, the plurality of external features include external conditions, and wherein the probabilistic relationships and dependencies are defined in the trained network model between (i) the household or business properties, (ii) the external conditions, (iii) the energy consumption sources, (iv) disaggregated energy consumption, and (v) aggregated energy consumption. 
     
     
         3 . The method of  claim 2 , wherein the limited amounts of data relate to at least one of the external conditions, the household or business properties, or the energy consumption sources. 
     
     
         4 . The method of  claim 1 , wherein the network model comprises a Bayesian network model. 
     
     
         5 . The method of  claim 4 , wherein the Bayesian network model utilizes a directed acyclic graph (DAG) to represent the user features and the external features. 
     
     
         6 . The method of  claim 4 , wherein each user feature or external feature is represented by a node in the Bayesian network model. 
     
     
         7 . The method of  claim 6 , wherein training the network model comprises using at least one of a structured learning process or a parameter learning process to train the network model. 
     
     
         8 . The method of  claim 7 , wherein the structured learning process comprises determining, developing or specifying which nodes within the network model are influenced by or dependent upon other nodes, as well as which nodes are connected, and directions of each connection between different nodes. 
     
     
         9 . The method of  claim 7 , wherein the parameter learning process comprises determining or calculating a set of probabilities based on counting of a number of instances or occurrences of correlations between different features or sets of features. 
     
     
         10 . The method of  claim 7 , wherein the structured learning process comprises initially training the network model using a set of baseline data for establishing classification accuracy. 
     
     
         11 . The method of  claim 10 , wherein the set of baseline data comprises data acquired from one or more residential energy consumption surveys. 
     
     
         12 . The method of  claim 11 , wherein the one or more residential energy consumption surveys comprise at least one binary response question or open-ended question. 
     
     
         13 . The method of  claim 11 , wherein the set of baseline data further comprises data that is inferred from partial or limited responses to the one or more residential energy consumption surveys. 
     
     
         14 . The method of  claim 1 , wherein the disaggregated energy consumption values for the plurality of energy consumption sources are inferred or estimated by applying a maximum a posteriori estimation process to the network model. 
     
     
         15 . The method of  claim 1 , wherein the one or more low frequency intervals comprise daily, weekly, monthly, or yearly intervals. 
     
     
         16 . The method of  claim 1 , wherein the plurality of external features comprise geophysical and/or building characteristics. 
     
     
         17 . The method of  claim 16 , wherein the geophysical and/or building characteristics include at least one of a heating degree day (HDD) metric, a cooling degree day (CDD) metric, a year-made metric, a building type metric, a locational metric, or a climate metric. 
     
     
         18 . The method of  claim 1 , wherein at least some of the plurality of different energy consumption sources are associated with an electric appliance or a gas appliance. 
     
     
         19 . The method of  claim 1 , wherein the limited amounts of data include one or more inputs from one or more of the plurality of customers, wherein the one or more inputs comprise a housing type input, a square-footage input, a rent-or-own input, an occupant amount input, a fridge amount input, an air-conditioning input, and/or a heating input. 
     
     
         20 . The method of  claim 1 , further comprising: dynamically updating the trained network model, based on changes to the limited amounts of data received at the one or more low frequency intervals.