Onboard driver, vehicle and fleet data mining

• US 7,715,961 B1
• Filed: 04/28/2005
• Issued: 05/11/2010
• Est. Priority Date: 04/28/2004
• Status: Active Grant

• Current Assignee: Agnik LLC
• Original Assignee: Agnik LLC
• Inventors: Kargupta, Hillol
• Primary Examiner(s): Tran; Khoi
• Assistant Examiner(s): Marc; McDieunel
• Application Number:
  US11/116,889
• Time in Patent Office: 1,839 Days
• Field of Search: 701/33, 701/1, 701/29, 701/32, 701/35, 701/36, 701/45, 701/70, 701/96, 701/102, 701/123, 701/201, 701/301, 701/701, 439/34, 439/133, 439/134, 342/357.09, 342/357.1, 303/152, 307/10.1, 455/552.1, 340/438
• US Class Current: 701/33
• CPC Class Codes: G06Q 10/08 Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0251 Targeted advertisements
• CPC Class Codes: G07C 5/008 communicating information to a remotely located station transmission systems for measured values G08C

Abstract

The method and system use onboard data stream mining for extracting data patterns from data continuously generated by different components of a vehicle. The system stores the data patterns in an onboard micro database and discards the data. The system uses a resource-constrained, small, lightweight onboard data stream management processor, with onboard data stream mining, an onboard micro database, and a privacy-preserving communication module, which periodically and upon request communicates stored data patterns to a remote control center. The control center uses the data patterns to characterize the typical and unusual vehicle health, driving and fleet behavior.

Claims

1. A vehicle fleet monitoring system comprising:

a sensor data bus connected to vehicle components,

vehicle and driver data collected from the sensor data bus,

an onboard data stream mining and management module,

computing patterns the vehicle defined by the following data types used for representing statistical models generated onboard;

a data structure defining mathematical functions of the form F;

Xⁿ→

Y, where F denotes the function, Xⁿ denotes the domain of the function, n denotes the number of dimensions of the domain, and Y is the range of the function, and the domain of the function is defined by the vehicle, driver, emissions and fleet data,

a data structure defining a collection of mathematical matrices A[t] for t=1, 2, . . . m,

where each cell of the t-th matrix A[i][j][t] is a function Fi,j of the vehicle, driver, emissions and fleet data, and

a data structure defining graphical relationship among different parameters in the vehicle, driver, emissions, and fleet data,

computing “

patterns”

onboard the vehicle defined by the following data types used for representing statistical models generated onboard;

a data structure defining mathematical functions of the form F;

Xⁿ→

Y, where F denotes the function, Xⁿ denotes the domain of the function, n denotes the number of dimensions of the domain, and Y is the range of the function, and the domain of the function is defined by the vehicle, driver, emissions and fleet data,

a data structure defining a collection of mathematical matrices A[t] for t=1, 2, . . . , m, where each cell of the t-th matrix A[i][j][t] is a function Fi,j of the vehicle, driver, emissions and fleet data, and

a data structure defining graphical relationship among different parameters in the vehicle, driver, emissions, and fleet data,

an onboard data stream mining within the onboard data stream mining and management module for receiving the data collected from the sensor data bus,

an onboard micro database,

wherein data collected from the sensor data bus is sent to the onboard data stream mining,

wherein data patterns are selected and sent to the onboard micro database,

and wherein remaining data are thrown away,

a remote control center,

an onboard communication module for managing communication between the remote control center and the onboard micro database and

wherein periodically, or upon queries from the remote control center, the data patterns stored in the onboard micro database are communicated to the remote control center, and

queries and responses are communicated between the remote control center and the onboard micro database.

2. The apparatus of claim 1, further comprising a privacy management module connected to the onboard communication module.

3. The apparatus of claim 1, further comprising a power management module connected to the onboard communication module.

4. The apparatus of claim 1, further comprising an onboard graphic user interface for local user interaction.

5. The apparatus of claim 4, wherein a local user communicates queries and responses directly to the onboard micro database.

6. The apparatus of claim 1, wherein the onboard data stream management system runs on a PDA, cell phone or embedded device.

7. The apparatus of claim 1, wherein actions are triggered by predetermined interesting data and patterns.

8. The apparatus of claim 1, wherein the remote control center allows remote management and mining of vehicle data streams from multiple vehicles, interactive statistical data analysis and visualization of driving characteristics.

9. The apparatus of claim 1, wherein distributed data mining algorithms are used for mining the collected data.

10. A vehicle fleet monitoring method comprising:

collecting vehicle, driver and fleet data with a sensor data bus,

transferring the collected data from the sensor data bus to an onboard data stream management system within an onboard data stream mining and management module,

onboard data stream mining the collected data,

computing patterns the vehicle defined by the following data types used for representing statistical models generated onboard;

a data structure defining mathematical functions of the form F;

Xⁿ→

Y, where F denotes the function, Xⁿ denotes the domain of the function, n denotes the number of dimensions of the domain, and Y is the range of the function, and the domain of the function is defined by the vehicle, driver, emissions and fleet data,

a data structure defining a collection of mathematical matrices A[t] for t=1, 2, . . . m, where each cell of the t-th matrix A[i][j][t] is a function Fi,j of the vehicle, driver, emissions and fleet data, and

a data structure defining graphical relationship among different parameters in the vehicle, driver, emissions, and fleet data,

computing “

patterns”

onboard the vehicle defined by the following data types used for representing statistical models generated onboard;

a data structure defining mathematical functions of the form F;

Xⁿ→

Y, where F denotes the function, Xⁿ denotes the domain of the function, n denotes the number of dimensions of the domain, and Y is the range of the function, and the domain of the function is defined by the vehicle, driver, emissions and fleet data,

a data structure defining a collection of mathematical matrices A[t] for t=1, 2, . . . , m, where each cell of the t-th matrix A[i][j][t] is a function F_i,j of the vehicle, driver, emissions and fleet data, and

a data structure defining graphical relationship among different parameters in the vehicle, driver, emissions, and fleet data,

extracting data patterns from the transferred data with the onboard data stream mining,

throwing away the data after the extracting the data patterns,

sending the data patterns from the onboard data stream mining to an onboard micro database,

storing the data patterns in the onboard micro databases, and

periodically or upon queries from a remote control center sending the stored data patterns from the onboard micro database through an onboard communication module to the remote control center.

11. The method of claim 10, wherein the data stream management system selects data patterns from the data collected from the sensor data bus for directly sending the selected data patterns to the onboard micro database and the remaining collected data is sent to the data stream mining.

12. The method of claim 11, wherein data stream mining separates from the remaining collected data into second interesting data and patterns for sending to the onboard micro database and for throwing away data.

13. The method of claim 10, further comprising managing privacy in the onboard communication module with an onboard privacy management module.

14. The method of claim 10, further comprising managing power with an onboard power management module.

15. The method of claim 10, further comprising providing an onboard graphic user interface for local user interaction and communicating queries and responses directly from a local user to the onboard micro database.

16. The method of claim 10, wherein the onboard data stream management system runs on a PDA, cell phone or embedded device.

17. The method of claim 10, further comprising actions triggered by predetermined interesting data and patterns.

18. The method of claim 10, wherein the mining the data collected from the sensor data bus uses distributed data mining algorithms.

21. The method of claim 10, further comprising of the following algorithms:

fast detection of changes and computation of correlation, covariance, inner product, and distance matrices onboard the vehicle and sending the results to the remote server over wired or wireless network,

a collection of vehicle health tests for monitoring the health of the vehicle and driver behavior and sending the results to the remote server over wired or wireless network,

distributed Euclidean distance, covariance, and inner product matrices computation algorithms in a fleet environment involving the vehicles and the remote servers and sending the results to the remote server over wired or wireless network,

distributed principal component analysis (PCA) in a fleet environment involving the vehicles and the remote servers and sending the results to the remote server over wired or wireless network,

onboard principal component analysis (PCA) of the vehicle and driver data and sending the results to the remote server over wired or wireless network,

onboard randomized projection of the vehicle and driver data and sending the results to the remote server over wired or wireless network,

clustering, predictive function learning, link analysis and outlier detection onboard the vehicle and sending the results to the remote server over a wired or wireless network,

algorithms for modeling fuel consumption behavior as a function of different vehicle and driving parameters and sending the results to the remote server over wired or wireless network, and

time series analysis and data stream mining algorithms for identifying the special types of driving behavior segments based on the vehicle and driver data and sending the results to the remote server over wired or wireless network.

19. A method of using data from operations of a vehicle comprising:

connecting vehicle components to a sensor bus,

supplying vehicle data to the sensor bus,

providing sensor data streams from the sensor bus to an onboard data stream management system,

separating generated data patterns from the data streams by the data stream management system,

computing patterns the vehicle defined by the following data types used for representing statistical models generated onboard;

a data structure defining mathematical functions of the form F;

Xⁿ→

Y, where F denotes the function, Xⁿ denotes the domain of the function, n denotes the number of dimensions of the domain, and Y is the range of the function, and the domain of the function is defined by the vehicle, driver, emissions and fleet data,

a data structure defining a collection of mathematical matrices A[t] for t=1, 2, . . . m,

where each cell of the t-th matrix A[i][j][t] is a function Fi,j of the vehicle, driver, emissions and fleet data, and

a data structure defining graphical relationship among different parameters in the vehicle, driver, emissions, and fleet data,

computing “

patterns”

onboard the vehicle defined by the following data types used for representing statistical models generated onboard;

a data structure defining mathematical functions of the form F;

Xⁿ→

Y, where F denotes the function, Xⁿ denotes the domain of the function, n denotes the number of dimensions of the domain, and Y is the range of the function, and the domain of the function is defined by the vehicle, driver, emissions and fleet data,

a data structure defining a collection of mathematical matrices A[t] for t=1, 2, . . . , m, where each cell of the t-th matrix A[i][j][t] is a function Fi,j of the vehicle, driver, emissions and fleet data, and

a data structure defining graphical relationship among different parameters in the vehicle, driver, emissions, and fleet data,

providing the separated data patterns to an onboard micro database,

storing the separated data patterns in the onboard micro database,

throwing away the data,

receiving requests from a remote control center,

directing the request to the onboard micro database,

responding to the requests by providing the stored data patterns from the onboard micro database to the remote control center, and

periodically providing the stored data patterns from the onboard database to the remote control center.

20. The method of claim 19, further comprising querying the onboard micro database from an onboard control center and responding to the request from the onboard control center by providing the data patterns from the onboard micro database.