Method and system for connected advanced flare analytics

1. A method for monitoring and controlling flare operations of an industrial plant using a machine deep learning-based self-adaptive industrial automation system comprising:

(a) obtaining, from at least one camera, a sequence of real time images of the flare operation and obtaining data from a network of the industrial plant as to at least one flare parameter;

(b) applying, an optical flow algorithm on the sequence of real time images to separate the background information from useful information to generate motion information;

(c) inputting the real time images and the motion information to a machine deep learning configurable system module;

(d) analyzing the data by the machine deep learning configurable system module using machine learning models and algorithms to assign pixels of the images to categories selected from smoke, flame, and steam;

(e) generating a semantic segmentation map at a pixel level that categorizes segmented image frames indicating the smoke, flame, and steam based on the motion information of the pixels between frames of the sequence of real time images of the flare operation;

(f) displaying the results of the generated segmentation map by the machine deep learning configurable system module; and

(g) issuing a notice when the percentage of pixels in a specific category falls outside a predetermined range.

2. The method of claim 1 wherein the at least one camera comprises at least one video camera and at least one thermal camera.

3. The method of claim 1 further comprising adjusting at least one parameter of the real-time operation of the flare operation of the industrial facility based on the results of the analyzing data by the machine deep learning configurable system module.

4. The method of claim 1 , wherein the machine learning configurable system module comprises a Fully Convolution Neural Network architecture or a variant thereof.

5. The method of claim 1 further comprising obtaining data from a cloud the data relating to a second or more flare operation of a second industrial plant.

6. The method of claim 1 , further comprising generating the machine learning configurable system module comprising:

training the machine learning models and algorithms using a first set of data from the obtained data and from historical data stored in a database; and

validating the machine learning models and algorithms using a second set of data from the obtained data and from historical data stored in the database.

7. The method of claim 1 , further comprising:

displaying flare operations of the industrial plant using the machine learning-based self-adaptive industrial automation system; and

dynamically displaying images of the flare operations processed by a machine deep learning configurable subsystem wherein the pixels of camera images are categorized as smoke, steam, or flame.

8. The method of claim 7 further comprising:

dynamically displaying a count of missed events and/or mitigated events;

dynamically displaying a representation of the percentage of pixels in each category; and

dynamically displaying a graphical log of events.

9. The method of claim 8 wherein the processed pixels of the camera images of the flare operations, the count of missed events and/or mitigated events, the representation of the percentage of pixels in each category and the graphical log of events are displayed using a graphical user interface.

10. The method of claim 9 wherein the graphical user interface displays the pixels of the camera images of the flare operations on a display in three separate segments categorized as the steam, smoke, or flame.

11. The method of claim 9 wherein the graphical user interface displays the images to a user at a stationary monitor.

12. The method of claim 9 wherein the graphical user interface displays the images to a user at mobile device display.

13. The method of claim 7 further wherein the camera images of the flare operations is a live feed from at least one video camera.

14. The method of claim 7 further wherein the camera of the flare operations is a live feed from at least one thermal camera.

15. A system for monitoring and controlling flare operations of an industrial plant using a machine deep learning-based self-adaptive configurable subsystem comprising:

a) an analytics system comprising a processor comprising:

an optical flow subsystem;

a data augmentation subsystem;

the machine deep learning based self-adaptive configurable subsystem;

b) a camera hub comprising at least one camera and a processor configured to:

receive a sequence of real time images of the flare operation from the at least one camera and obtain data from a network of the industrial plant as to at least one flare parameter,

apply, an optical flow algorithm on the sequence of real time images to separate the background information from useful information to generate motion information, and

transmit the real time images and the motion information to the analytics system; and

c) a display configured to receiving the images from the analytics system and display the images generated by the machine deep learning based self-adaptive configurable subsystem wherein pixels of the images are categorized and displayed as smoke, steam, or flame, and

wherein the analytics system is configured to generate a semantic segmentation map at a pixel level that categorizes segmented image frames indicating the smoke, flame, and steam based on the motion information of the pixels between frames of the sequence of real time images of the flare operation.

16. The system of claim 15 wherein the analytic system receives data for at least one parameter of the operation of the industrial plant from an industrial plant network.

17. The system of claim 16 , wherein the at least one camera comprises at least one video camera and at least one thermal camera each camera providing a sequence of real time images of the flare operation to the optical flow subsystem.

18. The system of claim 17 , wherein the data augmentation subsystem comprises:

a Fully Convolution Neural Network architecture or a variant thereof arranged to receive the real time images from the video and thermal cameras and the at least one plant parameter and augment the data into a form compatible for use by the machine deep learning based self-adaptive configurable subsystem; and

transmit the augmented data to the deep learning based self-adaptive configurable subsystem.

19. The system of claim 18 , wherein the machine deep learning based self-adaptive configurable subsystem includes at least one deep learning model that receives the augmented data from the augmentation subsystem that uses the deep learning model to analyze the augmented data and assigning categories to the pixels of the real time images, the pixels selected from smoke, flame, and steam.

20. The system of claim 15 wherein the analytic system further obtains data from a cloud the data relating to a second or more flare operation of a second industrial plant.