System and method for energy sample forecasting of HVAC-R systems

A system comprising:

• a heating, venting and air conditioning-refrigeration (HVAC-R) system, wherein the HVAC-R system comprises:
• at least one sensor for detecting actual energy samples of the HVAC-R system;
• and
• a forecasting system communicatively coupled to the HVAC-R system via a communication network, wherein the forecasting system comprises:
• at least one processor;
• and
• a memory communicatively coupled to the at least one processor, wherein the memory comprises an energy sample forecasting module to:
• forecast a first expected energy sample of the HVAC-R system at a first time period by modelling actual energy samples at previous time periods using a statistically-based seasonal-autoregressive integrated moving average (SARIMA) model, wherein the actual energy samples at the previous time periods of the HVAC-R system are obtained from the at least one sensor;
• compare the first expected energy sample with an actual energy sample at the first time period;
• detect an anomaly at the first time period based on non-energy samples of the HVAC-R system and a deviation between the first expected energy sample and the actual energy sample at the first time period, wherein the deviation between the first expected energy sample and the actual energy sample is greater than a dynamic context sensitive threshold, wherein the dynamic context sensitive threshold is a threshold defined based on a location of the HVAC-R system, a time-of-day and thermodynamic parameters of the location;
• and
• forecast an expected energy sample of the HVAC-R system at next time period by modelling a second expected energy sample of the HVAC-R system using the statistically-based SARIMA model upon detecting the anomaly, wherein the second expected energy sample is an energy sample that is forecasted by modelling the actual energy samples at the previous time periods using a physical model.

A non-transitory computer readable medium comprising thereon instruction which when executed by a possessor on a computing device, cause the processor to perform a method comprising:

• forecasting a first expected energy sample of a heating, venting and air conditioning-refrigeration (HVAC-R) system at a first time period by modelling actual energy samples of the HVAC-R system at previous time periods using a statistically-based seasonal-autoregressive integrated moving average (SARIMA) model;
• comparing the first expected energy sample with an actual energy sample at the first time period;
• detecting an anomaly at the first time period based on non-energy samples of the HVAC-R system and a deviation between the first expected energy sample and the actual energy sample at the first time period, wherein the deviation between the first expected energy sample and the actual energy sample is greater than a dynamic context sensitive threshold, wherein the dynamic context sensitive threshold is a threshold defined based on a location of the HVAC-R system, a time-of-day and thermodynamic parameters of the location;
• and
• forecasting an expected energy sample of the HVAC-R system at next time period by modelling a second expected energy sample of the HVAC-R system using the statistically-based SARIMA model upon detecting the anomaly, wherein the second expected energy sample is an energy sample which is forecasted by modelling the actual energy samples at the previous time periods using a physical model.