System and method for generating maintenance actions of a vehicle based on trained machine learning of monitored vehicle data

Patent metadata

Field	Value
Publication number	US-11720862-B2
Application number	US-201916294221-A
Country	US
Kind code	B2
Filing date	Mar 6, 2019
Priority date	Mar 6, 2019
Publication date	Aug 8, 2023
Grant date	Aug 8, 2023

Abstract

Systems and methods for providing a disposal recommendation for a vehicle are provided. The method includes generating a profile for a first vehicle, the first vehicle being an income generating vehicle and the first vehicle profile including at least age, mileage, and location of the first vehicle, receiving monitoring data of the first vehicle, and receiving sales data for second vehicles, the second vehicles corresponding to the profile. The method may further include generating predicted depreciation data of the first vehicle, based on the sales data and the monitoring data, generating a predicted profit data from the first vehicle, and providing a recommended disposal time for the first vehicle, based on the predicted depreciation data and the predicted profit data.

First claim

What is claimed is:

1. A computer implemented method for generating maintenance actions of a vehicle based on trained machine learning of monitored vehicle data, the computer implemented method comprising:
collecting, by an onboard diagnostic data scanner of an internal monitoring system of the vehicle, onboard diagnostic data comprising: speed, acceleration and deceleration of the vehicle, as well as location of the vehicle and a failure of brakes of the vehicle;
detecting, by a plurality of cameras of an external monitoring system of the vehicle, a vehicle collision indicating that the vehicle has been hit by other vehicles or objects from: a back of the vehicle, a front of the vehicle, or a side of the vehicle;
detecting, by the plurality of cameras of the external monitoring system, visual data indicating a presence of a bump on a road;
correlating, by a plurality of sensors of the external monitoring system, the acceleration of the vehicle with the location of the vehicle, and with sound data and the visual data observed from the plurality of cameras;
correlating, by the plurality of sensors of the external monitoring system, the visual data of the bump with a vertical vibration of the vehicle as it passes over the bump;
correlating, by the plurality of sensors of the external monitoring system, the vertical vibration of the vehicle with the speed of the vehicle while the vehicle is passing over the bump;
storing, by the internal and external monitoring systems, monitoring data from: the speed, the collision, the failure of the brakes and the deceleration of the vehicle, indicating that a significant vehicle-related event has occurred;
transmitting, via a mobile network, the monitoring data of the significant vehicle-related event when a signal strength of the mobile network is greater than a predetermined signal strength of a long-term evolution connection;
establishing, an overall wear-and-tear score of the vehicle based on a weighted average of respective wear-and-tear scores of the brakes and other individual components of the vehicle, based on the monitoring data and based on the vehicle reaching a predetermined number of braking events resulting in the deceleration of the vehicle above a certain limit;
training, by one or more processors, a neural network, by:
consecutively selecting and processing a number of features and layers specific to a model of the neural network according to: the monitoring data, the overall wear-and-tear score of the vehicle and repairs done to the vehicle, to improve performance of the neural network model;
rating the neural network model based on an average error generated by the neural network model according to the number of consecutively selected and processed features and layers specific to the neural network model;
detecting, by the neural network model using the rating, that the average error is above a threshold error value;
modifying, by the neural network model, the number of the features and the number of the layers specific to the neural network model in response to detecting that the average error is above the threshold error value until the average error drops below the threshold error value;

suggesting, by the trained neural network model, maintenance actions for the vehicle based on the overall wear-and-tear score for the vehicle and the repairs done to the vehicle.

2. The computer implemented method of claim 1 , further comprising:
generating predicted maintenance costs for the vehicle.

3. The computer implemented method of claim 2 , wherein the predicted maintenance costs further comprise:
predicted cost of cleaning the vehicle;
predicted cost of operating the vehicle; and
predicted cost of repairing the vehicle.

4. The computer implemented method of claim 1 , further comprising:
generating a profile for a user, the profile for the user comprising a history of respective monitoring data associated with vehicles driven by the user.

5. The computer implemented method of claim 4 , further comprising:
generating a first contract between the user and an owner for the vehicle, the first contract including a first price for using the vehicle.

6. The computer implemented method of claim 5 , further comprising:
generating a second contract between the user and the owner based on the first contract and the history of the respective monitoring data.

7. The computer implemented method of claim 6 , wherein the second contract comprises a second price for using the vehicle.

8. The computer implemented method of claim 1 , wherein the monitoring data comprises the acceleration for the vehicle.

9. The computer implemented method of claim 1 , wherein the monitoring data comprises the onboard diagnostics data.

10. The computer implemented method of claim 1 , further comprising:
generating predicted depreciation data of the vehicle, based on sales data and the monitoring data.

11. The computer implemented method of claim 1 , further comprising:
providing a recommended disposal time for the vehicle, based on the overall wear-and-tear score.

12. The computer implemented method of claim 11 , wherein providing a recommended disposal time comprises:
evaluating a marginal effective profit value;
evaluating a running average of the marginal effective profit value;
establishing a target range; and
recommending disposal of the vehicle when the marginal effective profit value for the vehicle and the running average of the marginal effective profit are within the target range.

13. The computer implemented method of claim 1 , further comprising:
generating a profile for the vehicle, the vehicle being an income generating vehicle and the profile comprising at least age, mileage, and the location of the vehicle.

14. The computer implemented method of claim 1 , further comprising:
generating and plotting on an interface, a graph of the overall wear-and-tear score, wherein the graph includes one or more of a bar graph and a chart graph.

15. The computer implemented method of claim 1 , further comprising:
generating predicted profit data from the vehicle.

16. A system for generating maintenance actions of a vehicle based on trained machine learning of monitored vehicle data, the system comprising:
internal and external monitoring systems of the vehicle, configured to:
store monitoring data from: speed, collision, failure of brakes and deceleration of the vehicle, indicating that a significant vehicle-related event has occurred;

an onboard diagnostic data scanner of the internal monitoring system of the vehicle, configured to:
collect onboard diagnostic data comprising: acceleration, the speed, and the deceleration of the vehicle, as well as location of the vehicle and the failure of the brakes of the vehicle;

a plurality of cameras of the external monitoring system of the vehicle, configured to:
detect the vehicle collision indicating that the vehicle has been hit by other vehicles or objects from: a back of the vehicle, a front of the vehicle, or a side of the vehicle, and
detect visual data indicating a presence of a bump on a road;

a plurality of sensors of the external monitoring system, configured to:
correlate the acceleration of the vehicle with the location of the vehicle, and with sound data and the visual data observed from the plurality of cameras,
correlate the visual data of the bump with a vertical vibration of the vehicle as it passes over the bump,
correlate the vertical vibration of the vehicle with the speed of the vehicle while the vehicle is passing over the bump,

a mobile network config

Assignees

• Capital One Services Llc

Inventors

• Dagley Geoffrey
• Tang Qiaochu
• Hooshmand Habeeb
• Price Micah
• Duckworth Staevan
• Wylie Stephen
• Hoover Jason

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06Q10/20 Primary

  Administration of product repair or maintenance · CPC title

• G06Q30/0206 Primary

  Price or cost determination based on market factors · CPC title

• G06N20/20

  Ensemble learning · CPC title