Predictive maintenance for robotic arms using vibration measurements

{
  "publication_number": "US11215535B2",
  "country": "US",
  "kind": "B2",
  "title": "Predictive maintenance for robotic arms using vibration measurements",
  "abstract": "Example implementations described herein involve systems and methods for conducting feature extraction on a plurality of templates associated with vibration sensor data for a moving equipment configured to conduct a plurality of tasks, to generate a predictive maintenance model for the plurality of tasks, the predictive maintenance model configured to provide one or more of fault detection, failure prediction, and remaining useful life (RUL) estimation.",
  "claims": [
    "1. A method, comprising: receiving vibration sensor data for a moving equipment configured to conduct a plurality of tasks; conducting feature extraction on the vibration sensor data to calculate the similarity coefficients between the vibration sensor data and a plurality of templates; inputting the similarity coefficients to a predictive maintenance model for the plurality of tasks, the predictive maintenance model configured to provide one or more of fault detection, failure prediction, and remaining useful life (RUL) estimation; wherein the conducting feature extraction on the vibration sensor data to calculate the similarity coefficients between the vibration sensor data and the plurality of templates comprises: selecting a reference template from the plurality of templates; conducting time-frequency decomposition on the reference templates and the vibration sensor data to generate component signals across a plurality of decomposition levels; and conducting similarity estimation on the component signals for each of the decomposition levels to generate similarity coefficients between the vibration sensor data and the reference template; wherein the plurality of templates are a group of templates selected from a plurality of groups generated from a clustering process, the group of templates selected based on a determination of the group from the plurality of groups having a nearest centroid to the vibration sensor data; wherein the plurality of tasks are identified by the clustering process.",
    "2. The method of claim 1, further comprising, for each of the plurality of groups generated from the clustering process, selecting one or more samples from vibration data samples of the moving equipment operating at normal condition for utilization as the plurality of templates for the each of the plurality of groups.",
    "3. The method of claim 2, wherein the selecting the one or more samples from vibration data samples of the moving equipment operating at normal condition for utilization as the plurality of templates for the each of the plurality of groups is based on a correlation of the one or more samples to other samples in the each of the plurality of groups.",
    "4. The method of claim 1, wherein the selecting the reference template from the plurality of templates comprises calculating a correlation between the vibration sensor data and each of the plurality of templates, and selecting one of the plurality of templates having a highest correlation as the reference template.",
    "5. The method of claim 1, wherein the moving equipment is a robotic arm.",
    "6. A non-transitory computer readable medium, storing instructions for executing a process, the instructions comprising: receiving vibration sensor data for a moving equipment configured to conduct a plurality of tasks; conducting feature extraction on the vibration sensor data to calculate the similarity coefficients between the vibration sensor data and a plurality of templates; inputting the similarity coefficients to a predictive maintenance model for the plurality of tasks, the predictive maintenance model configured to provide one or more of fault detection, failure prediction, and remaining useful life (RUL) estimation; wherein the conducting feature extraction on the vibration sensor data to calculate the similarity coefficients between the vibration sensor data and the plurality of templates comprises: selecting a reference template from the plurality of templates; conducting time-frequency decomposition on the reference templates and the vibration sensor data to generate component signals across a plurality of decomposition levels; and conducting similarity estimation on the component signals for each of the decomposition levels to generate to similarity coefficients between the vibration sensor data and the reference template; wherein the plurality of templates are a group of templates selected from a plurality of groups generated from a clustering process, the group of templates selected based on a determination of the group from the plurality of groups having a nearest centroid to the vibration sensor data; wherein the plurality of tasks are identified by the clustering process.",
    "7. The non-transitory computer readable medium of claim 6, the instructions further comprising, for each of the plurality of groups generated from the clustering process, selecting one or more samples from vibration data samples of the moving equipment operating at normal condition for utilization as the plurality of templates for the each of the plurality of groups.",
    "8. The non-transitory computer readable medium of claim 7, wherein the selecting the one or more samples from vibration data samples of the moving equipment operating at the normal condition for utilization as the plurality of templates for the each of the plurality of groups is based on a correlation of the one or more samples to other samples in the each of the plurality of groups.",
    "9. The non-transitory computer readable medium of claim 6, wherein the selecting the reference template from the plurality of templates comprises calculating a correlation between the vibration sensor data and each of the plurality of templates, and selecting one of the plurality of templates having a highest correlation as the reference template.",
    "10. The non-transitory computer readable medium of claim 6, wherein the moving equipment is a robotic arm.",
    "11. An apparatus communicatively coupled to a moving equipment configured to conduct a plurality of tasks the apparatus comprising: a processor, configured to: receive vibration sensor data for a moving equipment configured to conduct a plurality of tasks; conduct feature extraction on the vibration sensor data to calculate the similarity coefficients between the vibration sensor data and a plurality of templates; input the similarity coefficients to a predictive maintenance model for the plurality of tasks, the predictive maintenance model configured to provide one or more of fault detection, failure prediction, and remaining useful life (RUL) estimation; wherein the processor is configured to conduct the feature extraction on the vibration sensor data to calculate the similarity coefficients between the vibration sensor data and the plurality of templates by: selecting a reference template from the plurality of templates; conducting time-frequency decomposition on the reference templates and the vibration sensor data to generate component signals across a plurality of decomposition levels; and conducting similarity estimation on the component signals for each of the decomposition levels to generate similarity coefficients between the vibration sensor data and the reference template; wherein the plurality of templates are a group of templates selected from a plurality of groups generated from a clustering process, the group of templates selected based on a determination of the group from the plurality of groups having a nearest centroid to the vibration sensor data; wherein the plurality of tasks are identified by the clustering process.",
    "12. The apparatus of claim 11, wherein the processor is configured to, for each of the plurality of groups generated from the clustering process, select one or more samples from vibration data samples of the moving equipment operating at normal condition for utilization as the plurality of templates for the each of the plurality of groups.",
    "13. The apparatus of claim 12, wherein the processor is configured to select the one or more samples from vibration data samples of the moving equipment operating at the normal condition for utilization as the plurality of templates for the each of the plurality of groups based on a correlation of the one or more samples to other samples in the each of the plurality of groups.",
    "14. The apparatus of claim 11, wherein the processor is configured to select the reference template from the plurality of templates by calculating a correlation between the vibration sensor data and each of the plurality of templates, and selecting one of the plurality of templates having a highest correlation as the reference template.",
    "15. The apparatus of claim 11, wherein the moving equipment is a robotic arm."
  ],
  "description_excerpt": "The present disclosure is generally directed to robotic arms, and more specifically, to utilizing vibration measurements in robotic arms to conduct predictive maintenance.\n\nRobotic arms are powered mechanical manipulators that can be controlled or reprogrammed to perform different operation tasks. Usually, its end point can move during the operation, and the motion behavior can be controlled by planning or programming.\n\nFor monitoring the movement and vibration of such equipment with movable mechanisms, a variety of vibration sensors have been utilized for different tasks and applications. Accelerometers are the most commonly used sensor type for vibration measurement in industrial machinery monitoring. When an accelerometer is attached to an object, it measures the acceleration, which is the rate of change of the velocity.\n\nThe vibration of the equipment can be measured by using the mounted vibration sensors like accelerometers. Vibration is defined as an oscillating motion about a position of reference. Some common examples of vibration measurement include monitoring a static object (e.g., electric motor, turbine, bearing), where the reference or equilibrium is stable. More complex examples of vibration measurements could be the monitoring of the motion of moving objects, for example, robotic arms.\n\nVibration signals analysis have been used in the related art for predictive maintenance tasks for industrial equipment, such as rotating machinery as well as equipment with bearings, motor and gearbox.",
  "cpc": [
    "G01M 99/005",
    "B25J 9/161",
    "B25J 9/1674",
    "G01H 1/00",
    "G05B 2219/37351",
    "G05B 2219/37435",
    "G05B 23/0254",
    "G05B 23/0283"
  ],
  "ipc": [
    "G01M 99/00",
    "G05B 23/02"
  ],
  "assignees": [
    "Hitachi Ltd"
  ],
  "inventors": [
    "Wei Huang",
    "Chetan GUPTA",
    "Ahmed Khairy FARAHAT"
  ],
  "filing_date": "2019-11-14",
  "publication_date": "2022-01-04",
  "grant_date": "2022-01-04",
  "priority_date": "2019-11-14",
  "application_number": "US-201916684269-A",
  "family_id": "72826804",
  "cited_by_count": 9,
  "citations": [
    "US20020103626A1",
    "US20070088550A1",
    "US20070280006A1",
    "US20110246123A1",
    "US20210188554A1"
  ]
}