Back to results
US 12,597,332 B2Granted
Cloud-based machine health monitoring
AssigneeAnalog Devices International Unlimited Company
InventorsLewis Malaver, David Redmond
Life of this patent
Estimating the term…
Abstract
Systems and methods for detecting, isolating, and diagnosing machine faults are discussed. An exemplary system includes a network of sensor nodes deployed at machines or machine components to sense information of machine characteristics, and to generate physical or statistical features using the sensed machine characteristics. A cloud-computing device, communicatively coupled to the sensor network, can provide cloud-based services including detecting a machine fault, diagnosing a fault type, or estimating time to machine failure. A user interface associated with a client device can alert a user of the detected fault, such that the user can take preventive actions to minimize machine downtime.
Claims20
A system for monitoring machine health condition, the system comprising:
- a sensor network including sensor nodes configured to sense information of machine characteristics respectively from a plurality of machine parts, the sensor nodes each including:
- a first sensor module configured to acquire first machine data, and based thereon to detect an ON state of a machine part indicating execution of an intended operation, or an OFF state of the machine part indicating non-execution of the intended operation, using a trained machine operational state model;
- and
- a second sensor module, differing from the first sensor module in at least one of sensing modality or performance characteristic, configured to acquire second machine data, and based thereon to generate one or more features, including spectral contents at one or more predetermined frequencies or frequency ranges corresponding to respective one or more fault types;
- a cloud-computing device, communicatively coupled to the sensor network, configured to:
- generate the trained machine operational state model using a clustering algorithm and historical sensor data collected from the plurality of machine parts;
- and
- query a state of a portion of the plurality of machine parts and, conditional upon the queried state, provide one or more cloud-based services to the portion of the plurality of machine parts for detecting an existence of a fault or diagnosing a fault type thereof using the generated one or more features including the spectral contents at the one or more predetermined frequencies or frequency ranges corresponding to the respective one or more fault types, such that the cloud-based services of fault detection or fault type diagnosis are provided when the portion of the plurality of machine parts are in the ON state and not provided when the portion of the plurality of machine parts are in the OFF state;
- and
- a user interface configured to alert a user of the detected fault or the diagnosed fault type.
7 more elements
A method for monitoring machine health condition, the method comprising:
- sensing information of machine characteristics of a plurality of machine parts using networked sensor nodes respectively deployed to the plurality of machine parts, the networked sensor nodes each including a first sensor module configured to acquire first machine data and a second sensor module configured to acquire second machine data, the second sensor module differing from the first sensor module in at least one of sensing modality or performance characteristic;
- training a machine operational state model for the sensor nodes using a clustering algorithm and historical sensor data collected from the plurality of machine parts;
- based on the first machine data, detecting an ON state of a machine part indicating execution of an intended operation, or an OFF state of the machine part indicating non-execution of the intended operation, using the trained machine operational state model;
- based on the second machine data, generating one or more features, including spectral contents at one or more predetermined frequencies or frequency ranges corresponding to respective one or more fault types;
- via a cloud-computing device, querying a state of a portion of the plurality of machine parts and, conditional upon the queried state, providing one or more cloud-based services to the portion of the plurality of machine parts and generating cloud-based analytics including detecting an existence of a fault or diagnosing a fault type thereof using the generated one or more features including the spectral contents at the one or more predetermined frequencies or frequency ranges corresponding to the respective one or more fault types, such that the cloud-based services of fault detection or fault type diagnosis are provided when the portion of the plurality of machine parts are in the ON state and not provided when the portion of the plurality of machine parts are in the OFF state;
- and
- alerting a user of the detected fault or the diagnosed fault type.
4 more elements
A non-transitory machine-readable storage medium that includes instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
- receiving information of machine characteristics of a plurality of machine parts sensed by networked sensor nodes deployed to the plurality of machine parts, the networked sensor nodes each including a first sensor module configured to acquire first machine data and a second sensor module configured to acquire second machine data, the second sensor module differing from the first sensor module in at least one of sensing modality or performance characteristic;
- based on the first machine data, detecting an ON state of a machine part indicating execution of an intended operation, or an OFF state of the machine part indicating non-execution of the intended operation, using a machine operational state model;
- based on the second machine data, generating one or more features, including spectral contents at one or more predetermined frequencies or frequency ranges corresponding to respective one or more fault types;
- querying a state of a portion of the plurality of machine parts and, conditional upon the queried state, providing one or more cloud-based services to the portion of the plurality of machine parts and generating cloud-based analytics including detecting an existence of a fault or diagnosing a fault type thereof using the generated one or more features including the spectral contents at the one or more predetermined frequencies or frequency ranges corresponding to the respective one or more fault types, such that the cloud-based services of fault detection or fault type diagnosis are provided when the portion of the plurality of machine parts are in the ON state and not provided when the portion of the plurality of machine parts are in the OFF state;
- and
- generating an alert of the detected fault or the diagnosed fault type.
3 more elements
Dependencies are inferred from the claim text.