Abstract
Systems and methods are provided for determining and distinguishing buried objects using Artificial Intelligence (AI). In an exemplary embodiment, electromagnetic data related to underground utilities and communication systems is collected and provided to a Deep Learning model to build a training set. The Deep Learning model may be trained based on collected sets of Training Data, testing data, and/or user predefined classifiers. The Deep Learning model may use thresholds to determine if a set of data falls within a specific class. Classes may include gas, electric, water, cable, communications lines, or other buried utility and communication classes. Electromagnetic data collected may include multi-frequency measurements, phase measurements, signal strength measurements, and other related measurements. Data may be collected from locators, Sondes, transmitting and receiving antennas, inductive clamps, electrical clips, and satellite systems such as GPS, and other sources. Determined class data may organized and displayed to a user.
Claims
1. A system for determining and distinguishing buried objects using Artificial Intelligence (AI) comprising: a receiving element for collecting at least one of multifrequency electromagnetic signal data and communication signal data (“collected data”); an input element for allowing a user to input one or more predefined classifiers; a processor for combining at least a portion of the collected data with at least one predefined classifier, wherein the processor outputs Training Data; at least one Neural Network for processing the Training Data using Deep Learning performed by Artificial Intelligence (AI), and classifying the collected data based on a predicted probability; and an output element for presenting the classification data to a user.
2. The system of claim 1, wherein the receiving element comprises at least one of a locator, Sonde, transmitting antenna, receiving antenna, transceiver, inductive clamp, electrical clip, or a satellite system.
3. The system of claim 1, wherein Training Data further includes imaging data collected from a camera or imaging element.
4. The system of claim 1, wherein Training Data further includes sensor data.
5. The system of claim 1, wherein Training Data further includes mapping data.
6. The system of claim 5, wherein mapping data includes at least one of depth or orientation data.
7. The system of claim 1, wherein Training Data further includes fiber optic data.
8. The system of claim 1, wherein Training Data further includes one or more of image data, current and/or voltage data, even and odd harmonics data, active and/or passive signal data, and spatial relationship data.
9. The system of claim 1, wherein Training Data further includes one or more of phase data and phase difference data.
10. The system of claim 1, wherein Training Data further includes other data.
11. The system of claim 10, wherein other data comprises one or more of observed data, user classification data, and ground truth data.
12. The system of claim 11, wherein ground truth data comprises one or more of ownership data, manufacturer data, connection data, utility box or junction data, and obstacle data.
13. The system of claim 1, wherein Training Data may be processed and classified in real time, or stored and post-processed in the Cloud.
14. The system of claim 1, wherein classifying the collected data comprises determining at least one of a utility type, electrical characteristics, connection type, asset type, manufacturer type, ownership type, location type, direction type, right of way type, or damaged asset type.
15. The system of claim 1, wherein the output element comprises one or more of a visual display, a speaker or other sound producing element, and a vibration or other tactile producing element.
16. A computer implemented method for determining and distinguishing buried objects using Artificial Intelligence (AI) comprising: collecting at least one of multifrequency electromagnetic signal data and communication signal data (“collected data”) from a plurality of sources; using the collected data alone or in combination with user predefined classifiers as Training Data; providing the Training Data to at least one Neural Network; using at least one Neural Network for processing the Training Data using Deep Learning performed by Artificial Intelligence (AI) and classifying the collected data based on a predicted probability; and organizing and presenting the classified data to a user.
17. The method of claim 16, wherein collecting the at least one of multifrequency electromagnetic signal data and communication signal data comprises receiving the data from at least one of a locator, Sonde, transmitting antenna, receiving antenna, transceiver, inductive clamp, electrical clip, or a satellite system.
18. The method of claim 16, wherein Training Data may further include imaging data collected from a camera or imaging element.
19. The method of claim 16, wherein Training Data further includes sensor data.
20. The method of claim 16, wherein Training Data further includes mapping data.
21. The method of claim 20, wherein mapping data includes at least one of depth or orientation data.
22. The method of claim 16, wherein Training Data further includes fiber optic data.
23. The method of claim 16, wherein Training Data further includes one or more of image data, current and/or voltage data, even and odd harmonics data, active and/or passive signal data, and spatial relationship data.
24. The method of claim 16, wherein Training Data further includes one or more of phase data, phase difference data, ground penetrating radar (GPR) data, acoustic data, and tomography data.
25. The method of claim 16, wherein Training Data further includes other data.
26. The method of claim 25, wherein other data comprises one or more of observed data, user classification data, ground truth data, physics model data, and ground return current data.
27. The method of claim 26, wherein ground truth data comprises one or more of ownership data, manufacturer data, connection data, utility box or junction data, and obstacle data.
28. The method of claim 16, wherein Training Data may be processed and classified in real time, or stored and post-processed in the cloud.
29. The method of claim 16, wherein classifying the collected data comprises determining at least one of a utility type, electrical characteristics type, connection type, asset type, manufacturer type, ownership type, location type, direction type, right of way type, or damaged asset type.
30. The method of claim 16, wherein presenting classified data to a user comprises an output element including one or more of a visual display, a speaker or other sound producing element, and a vibration or other tactile producing element.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0043] FIG. 1 is an illustration of different methods for collecting multifrequency electromagnetic data from buried objects associated with utilities or communication systems, as known in the prior art.
[0044] FIG. 2 is an illustration of an embodiment of a method of using Deep Learning/artificial intelligence to recognize patterns and make predictions related to underground utilities, in accordance with certain aspects of the present invention.
[0045] FIG. 3 is an illustration of an embodiment of a system, including a service worker using a portable locator and a Sonde to collect electromagnetic frequency data or other data from underground or buried assets, in accordance with certain aspects of the present invention.
[0046] FIG. 4 is an illustration of an embodiment of a system, including a vehicle equipped with a locator to collect electromagnetic frequency data or other data from underground or buried assets, in accordance with certain aspects of the present invention.
[0047] FIG. 5 is an illustration of an embodiment of a method of providing Training Data to a Neural Network to use Deep Learning/artificial intelligence to recognize patterns and make predictions related to underground utilities, in accordance with certain aspects of the present invention.
[0048] FIG. 6 is an illustration of an embodiment of a method of using Artificial Intelligence (AI) to classify collected data based on a predicted probability, and to test the accuracy of the prediction, as known in the prior art.
[0049] FIG. 7 is an illustration of an embodiment of a data base structure for using Artificial Intelligence (AI) to recognize patterns, in accordance with certain aspects of the present invention.
[0050] FIG. 8 is an illustration of an embodiment of a chart showing various types of collected and other data as Training Data for Deep Learning in a Neural Network that uses Artificial Intelligence (AI), in accordance with certain aspects of the present invention.
DETAILED DESCRIPTION OF EMBODIMENTS
[0051] It is noted that as used herein, the term “exemplary” means “serving as an example, instance, or illustration.” Any aspect, detail, function, implementation, and/or embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects and/or embodiments.
Example Embodiments
[0052] FIG. 1 illustrates details of an exemplary embodiment of different methods 100 for collecting multifrequency electromagnetic data from buried objects associated with utilities 150. Methods 100 may include collecting data from various apparatus with the ability to receive, measure, or sense single or multifrequency electromagnetic data from under or above ground sources. The various apparatus may include one or more of the following: GPS or other satellite systems 110, utility or other locator systems 120, equipment with one or more transmitters, receivers, or transceivers 130, Sonde equipment 140, and many other types of utility sensing equipment well known by those skilled in the art.
[0053] FIG. 2 illustrates details of an exemplary method 200 of using Deep Learning/Artificial Intelligence (AI) to recognize patterns and make predictions related to underground utilities. The method starts at block 210 collecting data and proceeds to block 220 where a Training Data Base, also known as a Data Suite or Training Data Suite, is assembled. The method then proceeds to block 230 where Deep Learning is used to train a Neural Network using Artificial Intelligence. Finally, the method proceeds to block 240 where AI estimates the probability that underground or buried objects or assets are specific types of equipment or utilities, or have other characteristics and specifics, including but not limited to current and/or voltage data, even and odd harmonics data, active and/or passive signal data, and spatial relationship data. It would be understood by one or ordinary skill in the art that there is an almost endless amount of characteristics and specifics related to utility equipment and assets.
[0054] FIG. 3 illustrates details of an exemplary embodiment 300 of a system including a service worker 310 using a portable locator 320, and a Sonde 330 located underground 340 to collect single or multifrequency electromagnetic data from an underground or buried utility asset 350. In some embodiments, system 300 may include one or more cameras 360 which could be attached to, or integral with portable locator 320, or could be located separately.
[0055] FIG. 4 illustrates details of an exemplary embodiment 400 of a system including a vehicle 410 equipped with an omni-directional antenna 420, a locator 430, a dodecahedron antenna 440, and a Sonde 450 located underground 460, used to collect single or multifrequency electromagnetic data from an underground or buried utility asset 470. In some embodiments, system 400 may include one or more cameras 480 which could be attached to, or integral with vehicle 410, 415 or a mount 417 connected to the vehicle 410 and/or bumper 415, or could be located separately.
[0056] FIG. 5 illustrates details of an exemplary embodiment 500 of a method of providing Training Data to a Neural Network to use Deep Learning/artificial intelligence to recognize patterns and make predictions related to underground utilities. Multifrequency Electromagnetic Data 510 may be collected from multiple sources, any Predefined Classifier(s) 520 may be inputted or entered by a user, and both may be combined in block 530. Other data such as image data, harmonics data, etc. may also be combined in block 530. The Combined data is also known as a Data Suite. Data combined at 530 becomes available to be used as Training Data, also known as a Training Data Suite, at block 540. The Training Data 540 is then provided to one or more Neural Networks 550 which use Deep Learning to predict one or more data classes 560 for the underground or buried assets related to utility and communication systems. Artificial Intelligence (AI) is used to provide a probability that specific assets have specific characteristics, have relationships between other assets, and fall into one or more classification or categories by using the Training Data to recognize patterns.
[0057] FIG. 6 illustrates details of an exemplary embodiment 600 of a method of providing test data to a Deep Learning system that uses Artificial Intelligence (AI) to check the accuracy of determined predictions, as known in the prior art. The method starts by Collecting Data 610. This step is followed by Splitting the Data 620 into Test Data 630 and Training Data 640. In decision block 650 it is determined whether the Training Data 640 is continuous (YES), or non-continuous (NO). If the answer is YES, the method proceeds to block 660 for Regression Testing; if the answer is NO, the method proceeds to block 670 to determine a Data Type (e.g. electromagnetic data, video data, user inputted classifications or categories, etc.). In block 680 the Trained Model is determined using AI based on the Training Data provided, and in block 690 the accuracy of the Trained Model is tested using the Test Data 630.
[0058] FIG. 7 illustrates details of an exemplary embodiment 700 of a database structure for using Artificial Intelligence (AI) to recognize patterns, as known in the prior art. The database structure includes a System/Environment 710, Deep Learning 720 which includes a Processor 730, Working Memory 740, and Non-Volatile Memory 750. In some embodiments, Experience Store Code and Target Q Code may be optionally stored in Non-Volatile Memory 750 in order to facilitate the use of a second or subsequent Neural Network. The Experience Data and Q Code would be provided to a first Neural Network to generate target values for training a second or subsequent Neural Network. In Block 760 Action Data is provided to the System/Environment 710 which outputs State Data 770. Removable Memory 780 is provided, as well as a Parameter Memory 790. Weights of one or more Neural Networks 792 are provided to Deep Learning 720.
[0059] FIG. 8 illustrates details of an exemplary embodiment 800 of a chart showing various types of collected and other data as Training Data for Deep Learning in a Neural Network that uses Artificial Intelligence (AI). Collected Data 805 may include Multifrequency Electromagnetic Data 810, Imaging Data 815, Mapping Data 820 which may include Depth and/or Orientation Data, Current and/or Voltage Data 825, Harmonics Data 830 including Even and /or Odd Harmonics Data, Active and/or Passive Signal Data 835, Spatial Relationship Data 840, Fiber Optic Data 845, Phase Data 850 which may include Single Phase or Multiphase Data, Phase Difference Data 855, Ground Penetrating Radar Data (GPR) 856, Acoustic Data 857, Tomography Data 858, and Magnetic Gradiometry Data 859. It is contemplated that additional types of Collected Data 805 related to utilities and communication systems could also be used, and would be apparent to those skilled in the art. Training Suite Data 860, which may include Collected Data 805, may also include Other Data 865. Other Data 865 may include one or more of the following: Observed Data 870, User Classification Data 875, and Ground Truth Data 880. It is contemplated that additional types of Other Data 865 related to utilities and communication systems could also be used, and would be apparent to those skilled in the art. Some examples of such data are paint marks including previous paint on the ground, pipeline markers, overhead utilities/powerlines, construction techniques uses, e.g. trenchfill (conductivity and magnetic permeability), local ground conductivity, type of equipment in operation on the grid. For instance, equipment could include horizontal drilling equipment that generally runs generally straight between a drill “in” pit and a drill “out pit. There are of course innumerable types of equipment that could be operating on the grid at any given time, these are well known in the art. Collected can include data collected walking and/or by vehicle including air collected data such as data collected by a drone. Collected data can be used separately or combined from multiple sources.
[0060] The scope of the invention is not intended to be limited to the aspects shown herein but are to be accorded the full scope consistent with the disclosures herein and their equivalents, wherein reference to an element in the singular is not intended to mean “one and only one” unless specifically so stated, but rather “one or more.” Unless specifically stated otherwise, the term “some” refers to one or more. A phrase referring to “at least one of” a list of items refers to any combination of those items, including single members. As an example, “at least one of: a, b, or c” is intended to cover: a; b; c; a and b; a and c; b and c; and a, b and c.
[0061] The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use embodiments of the present invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the disclosures herein and in the appended drawings.
