Modifying a device based on an annotated time series of sensor readings

Patent metadata

Field	Value
Publication number	US-10565035-B2
Application number	US-201816215800-A
Country	US
Kind code	B2
Filing date	Dec 11, 2018
Priority date	Oct 10, 2016
Publication date	Feb 18, 2020
Grant date	Feb 18, 2020

Abstract

A method, system, and/or computer program product modify a hardware device based on a time series of data. One or more processors standardize a time series of data received from sensors that are monitoring a hardware device. The processor(s) establish time ranges before, during and after each event. The processor(s) determine which events represented by the time series of data are significant by comparing means and trends of time sub-series corresponding to the time ranges before, during, and after each event, and then generate a modified time series of data by reducing a number of significant events described by the time series of data, which is used to modify the hardware device.

First claim

What is claimed is:

1. A processor-implemented method comprising:
standardizing, by one or more processors, a time series of data received from sensors that are monitoring a hardware device;
establishing, by one or more processors, time ranges before, during and after each event;
determining, by one or more processors, which events represented by the time series of data are caused by an anomalous condition in the hardware device by comparing means and trends of time sub-series corresponding to the time ranges before, during and after each event, wherein the anomalous condition causes a disruption in the time series of data that lingers after the event is detected by the sensors;
generating, by one or more processors, a modified time series of data by reducing a number of significant events described by the time series of data; and
modifying, by one or more processors, the hardware device based on the modified time series of data.

2. The processor-implemented method of claim 1 , further comprising:
generating, by one or more processors, a modified time series data graph from the modified time series of data; and
applying, by one or more processors, a time series smoother to the modified time series data graph, wherein the time series smoother is derived by applying an autoregressive integrated moving average (ARIMA) algorithm to the modified time series data graph.

3. The processor-implemented method of claim 1 , wherein the hardware device is a manufacturing device, the processor-implemented method further comprising:
generating, by one or more processors, a modified time series data graph from the modified time series of data, wherein a common feature in each event depicted in the modified time series data graph causes the manufacturing device to introduce a defect into a physical product that is constructed by the manufacturing device.

4. The processor-implemented method of claim 1 , wherein the hardware device is a computer, the processor-implemented method further comprising:
generating, by one or more processors, a modified time series data graph from the modified time series of data, wherein a common feature in each event depicted in the modified time series data graph is a result of a hardware defect in the computer.

5. The processor-implemented method of claim 1 , wherein the hardware device is a computer, the processor-implemented method further comprising:
generating, by one or more processors, a modified time series data graph from the modified time series of data, wherein a common feature in each event depicted in the modified time series data graph is a result of a software defect in the computer.

6. The processor-implemented method of claim 1 , wherein the hardware device is a storage device, the processor-implemented method further comprising:
generating, by one or more processors, a modified time series data graph from the modified time series of data, wherein a common feature in each event depicted in the modified time series data graph is caused by a defect in a controller of the storage device.

7. The processor-implemented method of claim 1 , wherein the hardware device is a pressure vessel, wherein the time series of data is a first time series of data from the sensors that are monitoring the pressure vessel during a first time period, and wherein the processor-implemented method further comprises:
comparing, by one or more processors, the first time series of data to a second time series of data, wherein the second time series of data is from the sensors monitoring the pressure vessel during a second time period;
determining, by one or more processors, that the first time series of data and the second time series of data describe values of average pressures in the pressure vessel that match within a first statistical limit;
determining, by one or more processors, that the first time series of data and the second time series of data describe values of pressure increases in the pressure vessel that match within a second statistical limit; and
in response to determining that the first time series of data and the second time series of data describe values of average pressures in the pressure vessel that match within a first statistical limit, and in response to determining that the first time series of data and the second time series of data describe values of pressure increases in the pressure vessel that match within a second statistical limit, determining, by one or more processors, that a cause of a spike in pressure in the pressure vessel is being caused by an event that is described by the first time series of data and the second time series of data.

8. The processor-implemented method of claim 1 , wherein the time series of data is a first time series of data from the sensors that are monitoring the hardware device during a first time period, and wherein the processor-implemented method further comprises:
comparing, by one or more processors, the first time series of data to a second time series of data and a third time series of data, wherein the second time series of data is from the sensors monitoring the hardware device during a second time period, and wherein the third time series of data is from the sensors monitoring the hardware device during a third time period;
determining, by one or more processors, that the first time series of data and the second time series of data share more event tags than the first time series of data and the third series of data; and
in response to determining that the first time series of data and the second time series of data share more event tags than the first time series of data and the third series of data, determining that events described in shared event tags in the first time series of data and the second time series of data caused a fault in the hardware device, wherein the fault in the hardware device caused the disruption in the first time series of data.

9. The processor-implemented method of claim 1 , wherein the hardware device is a pressure vessel, and wherein the processor-implemented method further comprises:
adjusting a pressure relief valve on the pressure vessel based on the modified time series of data, wherein adjusting the pressure relief valve causes the pressure relief valve to open while pressure in the pressure vessel is increased after a spike in pressure in the pressure vessel.

10. A computer program product comprising one or more computer readable storage mediums, and program instructions stored on at least one of the one or more computer readable storage mediums, the stored program instructions comprising:
program instructions to standardize a time series of data received from sensors that are monitoring a hardware device, wherein the time series of data is a first time series of data from the sensors that are monitoring the hardware device during a first time period;
program instructions to establish time ranges before, during and after each event;
program instructions to determine which events represented by the time series of data are significant by comparing means and trends of time sub-series corresponding to the time ranges before, during and after each event;
program instructions to generate a modified time series of data by reducing a number of significant events described by the time series of data;
program instructions to compare the first time series of data to a second time series of data and a third time series of data, wherein the second time series of data is from the sensors monitoring the hardware device during a second time period, and wherein the third time series of data is from the sensors monitoring the hardware device during a third time period;
program instructions to determine t

Assignees

• IBM

Inventors

• Rope Daniel J
• Wills Graham J

Classifications

• G06F9/4887

  involving deadlines, e.g. rate based, periodic · CPC title

• G06F9/542 Primary

  Event management; Broadcasting; Multicasting; Notifications · CPC title

• G06F2201/86

  Event-based monitoring · CPC title

• G06F11/0751

  Error or fault detection not based on redundancy (power supply failures G06F1/30; network fault management H04L41/06) · CPC title

• G06F11/0706

  the processing taking place on a specific hardware platform or in a specific software environment · CPC title