US Patent No. 10,925,192

USING PREDICTIVE ANALYTICS IN ELECTROCHEMICAL AND ELECTROMECHANICAL SYSTEMS


Patent No. 10,925,192
Issue Date February 16, 2021
Title Using Predictive Analytics In Electrochemical And Electromechanical Systems
Inventorship Craig Campiano, Oakland, CA (US)
Steven De Gryze, San Francisco, CA (US)
Rosa Catala-Luque, San Francisco, CA (US)
Assignee Modius Inc., San Francisco, CA (US)

Claim of US Patent No. 10,925,192

1. A method of operating a computer room temperature control facility, including steps ofcollecting past data with respect to control settings and sensor readings in a control facility, and maintaining said collected past data into a first database operating on a first control computer;
collecting metadata, external to said control facility, said metadata having an effect on load on said control facility, and maintaining said metadata into a second database operating on a second control computer;
communicating said data from said first control computer to said second control computer;
generating correlated data, in response to said data and said metadata, with respect to at least one time series;
dividing said correlated data into a plurality of training sets, each of said training sets including a sequence of said time series correlated data;
for each of said training sets, dividing said training set into a first portion and a second portion;
for each of said training sets, selecting a set of tuning parameters that provide a model of said second portion in response to said first portion;
collecting present data with respect to said control settings and sensor readings, maintaining said present data in said first database, and communicating said present data from said first control computer to said second control computer;
collecting present metadata, external to said control facility, said present metadata having an effect on load on said control facility, and maintaining said metadata into said second database;
operating said model on said second control computer, and presenting outputs from said model to one or more users at a logically remote facility;
receiving commands from said one or more users with respect to said control facility;
directing, by said second control computer, said control facility to perform steps having effect on its operation, in response to said commands;
collecting new data with respect to said control settings and sensor readings in said control facility, maintaining said new data in said first database, and communicating said new data from said first control computer to said second control computer;
collecting new metadata, external to said control facility, said new metadata having an effect on load on said control facility, and maintaining said new metadata into said second database;
updating said model on said second control computer in response to said new data and said new metadata;
repeating said steps of operating said model; receiving commands; directing said control facility to perform steps; collecting new data; collecting new metadata; and updating said model; until terminated by one or more users.