# US 2026/0064089 A1 · Building automation monitoring using artificial intelligence supported by rules and/or analytics > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2026/0064089 A1 - Document ID: US20260064089A1 - Status: Published application - Kind: A1 - Application number: 18/819,211 - Filed: 2024-08-29 - Published: 2026-03-05 - Earliest claimed: 2024-08-29 - Assignees: Siemens Industry, Inc. - Applicants: Siemens Industry, Inc. - Inventors: Robert Hafernik; Tyrell Vaughn; Marcus Boerkei; Stephen Henigman - Web page: https://patentgrep.com/patent/US20260064089A1 ## CPC - G05B 13/027: {using neural networks only} - G05B 15/02: electric - G05B 23/0245: {based on a qualitative model, e.g. rule based; if-then decisions} - G05B 2219/2642: Domotique, domestic, home control, automation, smart house - G05B 2223/02: Indirect monitoring, e.g. monitoring production to detect faults of a system - G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G) ## Abstract For monitoring a building automation system, an artificial intelligence may be used to prioritize, diagnose, and/or provide cost estimates. For ease of interaction, a large language model (LLM) may be used. For operation of the LLM, a prompt requesting assistance is generated. The prompt includes context. Rather than providing the quite voluminous time series data, the context uses the faults and statistics. A processor adds this context to a prompt generated for input to the LLM. Based on the context and prompt, the LLM may rapidly and/or regularly provide accurate diagnosis, prioritization, and/or cost estimates for monitoring building automation. Due to the prompt and context, plain text information assisting in understanding and solving problems in building automation is output without a need for an expert. ## Claims ### Claim 1 A method for monitoring in a building automation system, the method comprising: generating exceptions to rules in operation of the building automation system and information about an aggregation of the operation of the building automation system; adding the exception and information as context to a prompt for analysis, the prompt requesting analysis of the building automation system; querying a large language model with the prompt including the context; and outputting the analysis created by the large language model in response to the prompt. ### Claim 2 (depends on 1) The method of claim 1 wherein generating the exceptions comprises generating a list of faults in the operation of the building automation system, and wherein generating the information comprises generating averages in energy usage and temperature over a period. ### Claim 3 (depends on 1) The method of claim 1 wherein generating the exceptions comprises listing the exceptions provided by a rules engine of a building management system, the exceptions provided by the rules engine from time series data, and wherein adding comprises adding the exceptions as the context without adding the time series data to the context. ### Claim 4 (depends on 1) The method of claim 1 wherein adding further comprises adding a location of the building automation system, acronym definitions, and term explanations as the context. ### Claim 5 (depends on 1) The method of claim 1 wherein adding further comprises adding user comments and/or user priorities in issues as the context. ### Claim 6 (depends on 1) The method of claim 1 wherein adding comprises adding the context to the prompt, the prompt being a pre-configured request for the analysis. ### Claim 7 (depends on 1) The method of claim 1 wherein adding comprises adding the context to the prompt where the analysis is for diagnosis of problems with the building automation system, prioritization of the problems, and cost estimates for the problems. ### Claim 8 (depends on 1) The method of claim 1 wherein adding comprises automatically populating the prompt with the context. ### Claim 9 (depends on 1) The method of claim 1 wherein querying comprises querying automatically on a periodic basis, and wherein outputting comprises displaying the analysis on the periodic basis as plain text. ### Claim 10 (depends on 1) The method of claim 1 wherein generating comprises generating the exceptions and information for a plurality of buildings, wherein adding comprises adding the exceptions and the information for the plurality of buildings as the context, and wherein outputting comprises outputting the analysis with prioritization of the buildings. ### Claim 11 (depends on 10) The method of claim 10 wherein outputting comprises outputting with the prioritization including a summary for each of the buildings in the prioritization. ### Claim 12 A method for monitoring in a building automation system, the method comprising: detecting, by a rules engine executed by a processor for a building management system, abnormal performance of the building automation system; aggregating, by the processor of the building management system, performance statistics of the building automation system; populating context of a prompt for a large language model, the prompt being for monitoring the building automation system, the context being the abnormal performance and the performance statistics; and storing the prompt with the populated context in a queue for querying a large language model. ### Claim 13 (depends on 12) The method of claim 12 wherein detecting comprises detecting faults as the abnormal performance, the faults detected from time series data of the building automation system, and wherein populating the context comprise populating the context with the faults and without the time series data. ### Claim 14 (depends on 12) The method of claim 12 wherein aggregating comprises aggregating the performance statistics for energy usage and connectivity. ### Claim 15 (depends on 12) The method of claim 12 wherein populating comprises further including location of the building automation system in the context. ### Claim 16 (depends on 12) The method of claim 12 wherein populating comprises populating automatically by the processor or another processor. ### Claim 17 A building automation monitoring system comprising: a building management system configured to generate a list of faults in operation of a building automation system and to generate statistics for the operation of the building automation system; a processor configured to populate context in a large language model prompt, the context populated with the list of faults and the statistics; and a memory configured to store the prompt as populated with the context. ### Claim 18 (depends on 17) The building automation monitoring system of claim 17 wherein the processor is configured to execute the building management system, and wherein memory is configured to store the prompt prior to the population as a default request for analysis of the building automation system. ### Claim 19 (depends on 17) The building automation monitoring system of claim 17 further comprising: a server configured to execute a large language model and generate an analysis of the building automation system by the large language model in response to the large language model prompt; and a display configured to display the analysis. ### Claim 20 (depends on 17) The building automation monitoring system of claim 17 wherein the building automation system is a first building automation system of a plurality of building automation systems, and wherein the processor comprises a first server configured to generate the list and the statistics for the plurality of the building automation systems.