Incident prediction system

• US 10,229,361 B2
• Filed: 06/18/2015
• Issued: 03/12/2019
• Est. Priority Date: 06/18/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chandramohan, Sarika, Dayama, Pankaj S., Gupta, Saurabh, Weitner, Mark A.
• Primary Examiner(s): Nilsson, Eric
• Application Number:
  US14/742,765
• Publication Number:
  US 20160371597A1
• Time in Patent Office: 1,363 Days
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 10/00 Administration; Management
• CPC Class Codes: G06Q 50/265 Personal security, identity or safety

Abstract

The present invention may be a method, computer program product, and system for predicting future safety incidents. The embodiments may receive a first set of incident data about a first safety incident. The first set of data may include a first set of factors and a first severity. The embodiments may create a safety incident model based on the first set of data by correlating the first set of factors with the first severity. The embodiments may receive a second set of incident data, where the second set of data may include a second set of factors. The embodiments may determine a predicted severity of the second set of incident data based on the safety incident model and the second set of factors. The embodiments may display one or more factors from one of the first and second set of factors, having the highest predicted incident severity.

Claims

1. A method for predicting a severity level for an incident, the method comprising:

receiving a first set of incident data about a first safety incident, wherein the first set of incident data includes a first set of factors and a severity of the first safety incident, wherein receiving the first set of incident data about the first safety incident further comprises receiving the first set of incident data from a first user and a second user, and wherein the first set of incident data includes written reports of safety incidents, process variables and operating conditions;

preparing the first set of incident data by determining two factors of the first set of factors that are statistically similar, and retaining only one of the two statistically similar factors in the first set of incident data, wherein preparing the first set of incident data comprises removing irrelevant factors of the first set of factors based on known relationships received from a third user, and wherein preparing the first set of incident data comprises extracting semantic concepts from the written reports;

creating a safety incident model based on the first set of incident data by correlating the first set of factors with the severity of the first safety incident, wherein creating the safety incident model further comprises allocating a higher weight to incident data of the first set of incident data received from the first user over incident data of the first set of incident data received from the second user, wherein correlating the first set of factors correlates the extracted semantic concepts to process variables and operating conditions;

receiving a second set of incident data, wherein the second set of incident data includes a second set of factors from a plurality of measurement devices located around an industrial plant, wherein the second set of factors comprise process variables and operating conditions;

determining a predicted severity of the second set of incident data based on the safety incident model and the second set of factors from the second set of incident data; and

enabling a corrective action that would lower the predicted severity of the second set of factors.

2. The method of claim 1 further comprising:

receiving a third set of incident data about a third safety incident, wherein the third set of incident data includes a third set of factors and a severity of the third safety incident, wherein the first set of factors and the third set of factors are substantially similar, and wherein the severity of the third safety incident is greater than the severity of the first safety incident.

3. The method of claim 2, wherein creating the safety incident model is further based on the third set of incident data by correlating the third set of factors with the severity of the third safety incident;

and

wherein determining the predicted severity of the second set of incident data based on the safety incident model and the second set of factors from the second set of incident data further comprises assigning a weight to the severity of the third incident that is greater than a weight assigned to the severity of the first incident.

4. The method of claim 1, further comprising preparing the first set of incident data prior to creating the safety incident model based on the first set of incident data, wherein preparing the first set of incident data comprises removing irrelevant factors of the first set of factors based on known relationships received from a user.

5. The method of claim 1, wherein the predictive model is a decision tree model.

6. A computer program product for predicting a severity level for an incident, the computer program product comprising:

one or more non-transitory computer-readable storage devices and program instructions stored on at least one of the one or more tangible storage devices, the program instructions comprising;

program instructions to receive a first set of incident data about a first safety incident, wherein the first set of incident data includes a first set of factors and a severity of the first safety incident, wherein the program instructions to receive the first set of incident data about the first safety incident further comprises program instructions to receive the first set of incident data from a first user and a second user, and wherein the first set of incident data includes written reports of safety incidents, process variables and operating conditions;

program instructions to prepare the first set of incident data by determining two factors of the first set of factors that are statistically similar, and retaining only one of the two statistically similar factors in the first set of incident data, wherein the program instructions to prepare the first set of incident data comprise program instructions to remove irrelevant factors of the first set of factors based on known relationships received from a third user, and wherein the program instructions to prepare the first set of incident data comprises the program instructions to extract semantic concepts from the written reports;

program instructions to create a safety incident model based on the first set of incident data by correlating the first set of factors with the severity of the first safety incident, wherein the program instructions to create the safety incident model further comprises program instructions to allocate a higher weight to incident data of the first set of incident data received from the first user over incident data of the first set of incident data received from the second user, wherein program instructions to correlate the first set of factors correlates the extracted semantic concepts to process variables and operating conditions;

program instructions to receive a second set of incident data, wherein the second set of incident data includes a second set of factors from a plurality of measurement devices located around an industrial plant, wherein the second set of factors comprise process variables and operating conditions;

program instructions to determine a predicted severity of the second set of incident data based on the safety incident model and the second set of factors from the second set of incident data; and

program instructions to enable a corrective action display a process change that would lower the predicted severity of the second set of factors.

7. The computer program product of claim 6 further comprising:

the program instructions to receive a third set of incident data about a third safety incident, wherein the third set of incident data includes a third set of factors and a severity of the third safety incident, and wherein the first set of factors and the third set of factors are substantially similar, and wherein the severity of the third safety incident is greater than the severity of the first safety incident.

8. The computer program product of claim 7, wherein the program instructions to create the safety incident model is further based on the third set of incident data by correlating the third set of factors with the severity of the third safety incident;

and

wherein the program instructions to determine the predicted severity of the second set of incident data based on the safety incident model and the second set of factors from the second set of incident data further comprises assigning a weight to the severity of the third incident that is greater than a weight assigned to the severity of the first incident.

9. The computer program product of claim 6, further comprising program instructions to prepare the first set of incident data prior to creating the safety incident model based on the first set of incident data, wherein the program instructions to prepare the first set of incident data comprises program instructions to remove irrelevant factors of the first set of factors based on known relationships received from a user.

10. The computer program product of claim 6, wherein the predictive model is a decision tree model.

11. A computer system for predicting a severity level for an incident, the computer system comprising:

one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage devices, and program instructions stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, the program instructions comprising;

program instructions to receive a first set of incident data about a first safety incident, wherein the first set of incident data includes a first set of factors and a severity of the first safety incident, wherein the program instructions to receive the first set of incident data about the first safety incident further comprises program instructions to receive the first set of incident data from a first user and a second user, and wherein the first set of incident data includes written reports of safety incidents, process variables and operating conditions;

program instructions to prepare the first set of incident data by determining two factors of the first set of factors that are statistically similar, and retaining only one of the two statistically similar factors in the first set of incident data, wherein the program instructions to prepare the first set of incident data comprise program instructions to remove irrelevant factors of the first set of factors based on known relationships received from a third user, and wherein the program instructions to prepare the first set of incident data comprises the program instructions to extract semantic concepts from the written reports;

program instructions to create a safety incident model based on the first set of incident data by correlating the first set of factors with the severity of the first safety incident, wherein the program instructions to create the safety incident model further comprises program instructions to allocate a higher weight to incident data of the first set of incident data received from the first user over incident data of the first set of incident data received from the second user, wherein program instructions to correlate the first set of factors correlates the extracted semantic concepts to process variables and operating conditions;

program instructions to receive a second set of incident data, wherein the second set of incident data includes a second set of factors from a plurality of measurement devices located around an industrial plant, wherein the second set of factors comprise process variables and operating conditions;

program instructions to determine a predicted severity of the second set of incident data based on the safety incident model and the second set of factors from the second set of incident data; and

program instructions to enable a corrective action that would lower the predicted severity of the second set of factors.

12. The computer system of claim 11 further comprising:

the program instructions to receive a third set of incident data about a third safety incident, wherein the third set of incident data includes a third set of factors and a severity of the third safety incident, and wherein the first set of factors and the third set of factors are substantially similar, and wherein the severity of the third safety incident is greater than the severity of the first safety incident.

13. The computer system of claim 12, wherein the program instructions to create the the safety incident model is further based on the third set of incident data by correlating the third set of factors with the severity of the third safety incident;

and

wherein the program instructions to determine the predicted severity of the second set of incident data based on the safety incident model and the second set of factors from the second set of incident data, wherein the severity of the third incident has greater weight towards the predicted severity determined by the safety incident model than the severity of the first incident.

14. The computer system of claim 11, further comprising program instructions to prepare the first set of incident data prior to creating the safety incident model based on the first set of incident data, wherein the program instructions to prepare the first set of incident data comprises program instructions to remove irrelevant factors of the first set of factors based on known relationships received from a user.