US-12620832-B2 - Systems and methods for monitoring power systems

US12620832B2US 12620832 B2US12620832 B2US 12620832B2US-12620832-B2

Abstract

Systems and methods for monitoring apparatus(es) and equipment using a sensor cluster are described. The systems and methods can be used to automatically repair or otherwise address actual and predicted failure modes of the apparatus(es), which include electrical systems, power systems, energy storage systems, and other systems.

Inventors

Brenton Parr Munson
Francis Mitchell Toglia
Jose Maria Guadalupe Gomez

Assignees

FLUID POWER AI, LLC

Dates

Publication Date: 20260505
Application Date: 20230627

Claims (16)

1 . A system for monitoring an apparatus, the system comprising: a sensor subsystem comprising a voltage sensor, a current sensor, and a temperature sensor; a housing surrounding the sensor subsystem; a mounting interface coupled to the housing and configured to couple to the apparatus, such that the sensor subsystem can detect signals from an output of at least one of a battery, a generator, a motor, and an inverter of the apparatus, if the apparatus comprises the battery, the generator, the motor, or the inverter; a signal conditioning and communications subsystem coupled to the sensor subsystem within the housing and configured to receive a voltage signal stream, a current signal stream, and a temperature signal stream from the sensor subsystem; and a processing subsystem coupled to the signal conditioning and communications subsystem and comprising a neural processing unit (NPU) comprising 1 trillions of operations per second (TOPS) capability, comprising self-attention time-series transformer architecture with an encoder block comprising multi-head attention subarchitecture, and omitting a decoder block, the processing subsystem contained within the housing and comprising non-transitory media storing instructions that, when executed, perform operations for: receiving data derived from the voltage signal stream, the current signal stream, and the temperature signal stream; performing a set of transformation operations upon said data; identifying a set of unique signatures corresponding to states of a set of subcomponents of the apparatus, from the set of transformation operations; and returning an analysis comprising a recommended action for maintaining performance of the apparatus, based upon the set of unique signatures.
2 . The system of claim 1 , wherein the sensor subsystem further comprises a demand sensor structured to detect rotation of a rotating component of the apparatus.
3 . The system of claim 1 , wherein the NPU comprises energy use performance of less than 1 picojoule per operation.
4 . The system of claim 1 , wherein the apparatus comprises a solar energy system.
5 . The system of claim 4 , wherein the set of subcomponents comprises a solar panel component, an inverter component, an energy storage component, an electrical panel component, and an electric meter component.
6 . The system of claim 1 , wherein the apparatus comprises an electric vehicle, and wherein the set of subcomponents comprises an energy management system component, an electric vehicle battery component, an inverter component, an electric motor component, a drivetrain component, and a regenerative breaking system component.
7 . The system of claim 1 , wherein the apparatus comprises an electric vehicle charger, and wherein the set of subcomponents comprises an alternating current (AC) supply component, a safety interlock component, a rectifier components, a power control unit component, a direct current (DC) converter component, a battery monitor component, a battery management component, a control area network (CAN) bus control and authentication component, or a vehicle-to-grid charging component.
8 . A system for monitoring an apparatus, the system comprising: a voltage sensor and a current sensor; a housing surrounding the voltage sensor and the current sensor; a mounting interface coupled to the housing and configured to couple the system to the apparatus such that the voltage sensor and the current sensor can detect signals from an output of a battery of the apparatus without directly contacting the battery; a signal conditioning and communications subsystem coupled to the voltage sensor and the current sensor within the housing and configured to receive a voltage signal stream from the voltage sensor and a current signal stream from the current sensor; and a processing subsystem coupled to the signal conditioning and communications subsystem and comprising a neural processing unit (NPU) comprising 1 trillions of operations per second (TOPS) capability, comprising self-attention time-series transformer architecture with an encoder block comprising multi-head attention subarchitecture, and omitting a decoder block, the processing subsystem contained within the housing and comprising on-chip self-attention time-series transformer architecture for processing the voltage signal stream and the current signal stream.
9 . The system of claim 8 , wherein the processing subsystem comprises non-transitory media storing instructions that, when executed, perform operations for: receiving data derived from the voltage signal stream and the current signal stream; performing a set of transformation operations upon said data; identifying a set of unique signatures corresponding to states of a set of subcomponents of the apparatus, from the set of transformation operations; and returning an analysis comprising a recommended action for improving or maintaining proper performance of the apparatus, based upon the set of unique signatures.
10 . The system of claim 8 , wherein the apparatus comprises at least one of: a renewable energy system, an electric vehicle, and an electric vehicle charger.
11 . A method for monitoring an apparatus, the method comprising: providing a mounting interface between a sensor subsystem coupled to a signal processing subsystem, and the apparatus, the sensor subsystem comprising a voltage sensor, a current sensor, and a temperature sensor, and wherein providing the mounting interface comprises mounting the sensor subsystem to the apparatus, such that the voltage sensor, the current sensor, and the temperature sensor can detect signals from an output of a battery of the apparatus without directly contacting the battery; sampling a voltage signal stream, a current signal stream, and a temperature signal stream generated from the sensor subsystem during operation of the apparatus; performing a set of transformation operations upon the voltage signal stream, the current signal stream, and the temperature signal stream with a neural processing unit (NPU) comprising 1 trillions of operations per second (TOPS) capability, wherein the set of transformation operations comprises operations applied by self-attention time-series transformer architecture with an encoder block comprising multi-head attention subarchitecture, and wherein the self-attention time-series transformer architecture omits a decoder block; identifying a set of unique signatures corresponding to faults of a set of subcomponents of the apparatus from the set of transformation operations; and returning an analysis comprising a recommended action for maintaining performance of the apparatus, based upon the set of unique signatures.
12 . The method of claim 11 , wherein the apparatus comprises a renewable energy system, and wherein the set of subcomponents comprises a component of the renewable energy system.
13 . The method of claim 12 , further comprising executing the recommended action, wherein the recommended action comprises guiding a drone for inspection of the component of the renewable energy system, in response to a fault of the component of the renewable energy system.
14 . The method of claim 11 , wherein the apparatus comprises an electric vehicle charger, and wherein the set of subcomponents comprises an alternating current (AC) supply component, a safety interlock component, a rectifier components, a power control unit component, a direct current (DC) converter component, a battery monitor component, a battery management component, a control area network (CAN) bus control and authentication component, or a vehicle-to-grid charging component.
15 . The method of claim 14 , further comprising executing the recommended action, wherein the recommended action comprises guiding an electric vehicle to an alternative properly-functioning electric vehicle charger, in response to a fault of at least one of the set of subcomponents.
16 . The method of claim 11 , further comprising executing the recommended action, wherein executing the recommended action comprises: transmitting information from the analysis, for observation through a mixed-reality device; receiving an input from a user of mixed reality device, the input configured to respond to at least one of a set of faults of the apparatus indicated in the analysis; and executing instructions for addressing the at least one fault of the set of faults, based upon the input, wherein the apparatus is positioned remote from the user.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of U.S. Provisional Application No. 63/483,744 filed on 7 Feb. 2023 and U.S. Provisional Application No. 63/497,280 filed on 20 Apr. 2023, which are each incorporated in its entirety herein by this reference. TECHNICAL FIELD This invention relates generally to fields related to electrical power system monitoring, and more specifically to new and useful systems and methods for monitoring electrical power system events at subcomponent and global levels, with generation of analyses to improve system maintenance and operation. BACKGROUND The commercial utilization of and ongoing costs of maintaining power systems is greatly impacted by proper maintenance practices that include regular service and replacement of system components. Furthermore, the technical domain knowledge required to properly troubleshoot and diagnose these complex systems is not readily available to the majority of power system maintainers and operators that rely on these systems to operate smoothly, with little or no downtime. The importance of proper operation will continue to grow, due to rapid increases in energy demand, and expanded options for providing energy through renewable sources and other sources. Furthermore, the current state-of-the-art in monitoring and predicting failure in power systems is limited by the need for vast amounts of data, compute-intensive training, hard-to-scale classification models, and equipment-specific limitations. With the rise of the industrial industry of things (IIOT), demand for improved maintenance and monitoring has rapidly increased. However, existing solutions are expensive to implement, do not scale in terms of data management, and are often machine specific. The high opportunity cost of equipment downtime and the lack of resources to properly diagnose power systems and related equipment produces reactive maintenance practices where equipment is utilized in non-ideal states for long periods of time before finally failing, often catastrophically. Without the proper tools and domain expertise, technicians are typically under pressure to replace components in order to return systems to service with minimal downtime, once there is a failure event, but do not resolve the root cause of failure. This leads to accelerated system wear that results in an endless costly battle to keep systems in production. Consequences of equipment downtime and costs of repair have also been amplified significantly by supply chain issues resulting from past and current world events. Unexpected downtime and poor system efficiency thus have associated costs that can be prevented or reduced with better monitoring, forecasting and troubleshooting systems. Current solutions for full system monitoring that can diagnose subcomponent issues in various types of power systems use many distributed sensors and custom algorithms. Implementing these options requires application-specific domain expertise, and the resources to design, deploy and maintain power systems is typically a non-optimal solution for utilities and operators, creating issues that affect all involved parties, down to the end consumer. Thus, there is a need to create new, scalable, and useful systems and methods for evaluating power system events at subcomponent and global levels, with generation of analyses to improve system operation and maintenance effectively lowering total cost of ownership and maximize production output. BRIEF DESCRIPTION OF THE FIGURES FIG. 1A depicts an embodiment of a system for apparatus monitoring. FIG. 1B depicts aspects of an embodiment of a system for apparatus monitoring. FIG. 1C depicts an example of an embodiment of a system for apparatus monitoring. FIG. 2 depicts an embodiment of a method for apparatus monitoring. FIG. 3 depicts an example of attention-based model architecture for monitoring apparatuses. FIG. 4 depicts flow of various use cases associated with an embodiment of a system for apparatus monitoring. FIG. 5 depicts a variation of a flow of the method for apparatus monitoring. FIG. 6 depicts a variation of a portion of a method for executing actions, in coordination with AR/MR/VR devices. FIG. 7 depicts an example of an interface for an operator of interest in apparatus monitoring and remote addressing of apparatus faults, through AR/MR/VR interfaces. FIG. 8 depicts example computing architecture of a system for apparatus monitoring. INCORPORATION BY REFERENCE All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference in their entireties for all purposes and to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference. Furthermore, where a range of values is provided, it is understood that each intervening value, between the upper and lower limit of that range and any other stated or interve