COMPUTERIZED TECHNIQUES FOR MONITORING AND ASSESSING REAL-TIME AND FUTURE OPERATIONAL AND HEALTH STATUSES OF PHYSICAL COMPONENTS , EQUIPMENT, AND/OR STRUCTURES USING MACHINE LEARNING BASED MODELS
Various automated techniques are disclosed herein for monitoring and assessing real-time and future operational and health statuses of real-world assets (e.g., physical devices, components, equipment, structures, buildings, machines, infrastructure, piping systems, etc.). A monitoring system incorporates the use of intelligent Monitoring Devices for monitoring pipe systems and/or other infrastructure for leaks/issues using IoT devices and machine learning. One or more Monitoring Device(s) are attached to specific physical equipment/structure(s) to be monitored A Monitoring Device performs comprehensive field data gathering. The collected field data is used to train a customized ML model, which is stored locally in that Monitoring Device. The Monitoring Device monitors current conditions of the pipe using various sensors, and uses its locally stored customized, trained model to perform real-time, edge-based analysis of the monitored data to identify possible issues in real-time and/or to predict future maintenance/service needs without relying on continuous cloud connectivity.
1 . A fluid monitoring system, comprising:
a plurality of sensors configured to detect fluid data associated with a fluid flowing through a first pipe system, the plurality of sensors comprising a combination of two or more of: a MEMS sensor, an accelerometer, gyroscope, ultrasound sensors, and temperature sensors;
wherein a first set of sensors of the plurality of sensors is configured or designed to be mounted to a first pipe or conduit of the first pipe system;
wherein at least some of the plurality of sensors is configured to detect fluid data in an x-axis, y-axis, and/or z-axis, and wherein changes over times in each axis are used to train models to determine normal or abnormal conditions;
a wired or wireless communication interface;
at least one processor, the at least one processor being operable to execute a plurality of instructions for:
receiving the fluid data;
determining whether the fluid data is indicative of a normal condition or an abnormal condition; and
upon determining the fluid data is indicative of an abnormal condition, at least one of: (i) causing a flow control valve coupled to the first pipe system to adjust; and (ii) transmitting to at least one remote device, via the communication interface, at least one of the fluid data and a notification relating to the fluid data.
2 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
executing a field data collection procedure for model training by configuring the fluid monitoring system to enter a data collection mode;
causing cycling of the flow control valve of the first pipe system through different flow positions to induce various flow rates of fluid through the first pipe system; and
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system.
3 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
causing uploading of the collected field measurement data to a PipeX Server System for model training;
training a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data; and
deploying the first trained model to the fluid monitoring system.
4 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
initiating training of a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data;
storing a digital representation of the first trained model at the fluid monitoring system; and
analyzing, at the fluid monitoring system and using the stored digital representation of first trained model, the fluid data to determine whether the fluid data is indicative of a normal condition or an abnormal condition.
5 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
initiating training of a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data;
storing a digital representation of the first trained model at the fluid monitoring system; and
analyzing, using the stored digital representation of first trained model, the fluid data to determine whether the fluid data is indicative of a normal condition or an abnormal condition, wherein the analyzing is performed without requiring access to cloud connectivity.
6 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model; and
analyzing, the fluid data using the first customized machine learning-based inference model to determine whether the fluid data is indicative of a normal condition or an abnormal condition.
7 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model;
analyzing, the fluid data using the first customized machine learning-based inference model to determine whether the fluid data is indicative of a normal condition or an abnormal condition; and
generating and transmitting a first alert notification upon detecting conditions indicative of a predicted abnormal condition of the first pipe system.
8 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
initiating training of a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data;
storing a digital representation of the first trained model at the fluid monitoring system; and
generating predictions relating to an operational state or health status of the first pipe system using the stored digital representation of first trained model and the fluid data.
9 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model; and
generating predictions relating to an operational state or health status of the first pipe system using the stored digital representation of first trained model and the fluid data.
10 . The system of claim 1 , further comprising:
a first computing system configured to run a PipeX software application configured or designed to facilitate operation of the fluid monitoring system;
the system being operable to cause the at least one processor to execute additional instructions for:
communicating with the PipeX software application; and
utilizing the first computing system to facilitate communication between the fluid monitoring system and a PipeX Server System.
11 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model;
generating predictions relating to an operational state or health status of the first pipe system using the stored digital representation of first trained model and the fluid data; and
initiating updating of the first inference model in response to detected prediction inaccuracies.
12 . The system of claim 1 being operable to cause the at least one processor to execute additional instructions for:
actively evaluating a mounting integrity of the first set of sensors to the first pipe or conduit by utilizing temperature differential analysis;
determining whether mounting integrity of the first set of sensors to the first pipe or conduit is indicative of improper sensor attachment to the first pipe or conduit; and
generating and transmitting a first alert notification in response to detecting conditions indicative of improper sensor attachment to the first pipe or conduit.
13 . The system of claim 1 , further comprising:
a first temperature sensor and a second temperature sensor;
the system being further operable to cause the at least one processor to execute additional instructions for:
using the first temperature sensor to measure a temperature of the first pipe or conduit;
using the second temperature sensor to measure a temperature of an ambient environment surrounding the first pipe or conduit;
performing a comparative analysis of the first and second temperatures to detect discrepancies indicative of improper sensor attachment to the first pipe or conduit; and
initiating, in response to detecting conditions indicative of improper sensor attachment to the first pipe or conduit, a first action for facilitating adjustment of the mounting of the first set of sensors to the first pipe or conduit.
14 . A method for monitoring fluid flow in a first pipe system, the method being implemented in a fluid monitoring system comprising a plurality of sensors configured to detect fluid data associated with a fluid flowing through the first pipe system; the plurality of sensors comprising a combination of two or more of: a MEMS sensor, an accelerometer, gyroscope, ultrasound sensors, and temperature sensors; wherein a first set of sensors of the plurality of sensors is configured or designed to be mounted to a first pipe or conduit of the first pipe system; the fluid monitoring system further comprising: a wired or wireless communication interface, and at least one processor;
the method comprising causing at least one processor to execute a plurality of instructions for:
detecting, using at least one of the plurality of sensors, fluid data in an x-axis, y-axis, and/or z-axis, wherein changes over times in each axis are used to train models to determine normal or abnormal conditions;
receiving the fluid data;
determining whether the fluid data is indicative of a normal condition or an abnormal condition; and
upon determining the fluid data is indicative of an abnormal condition, at least one of: (i) causing a flow control valve coupled to the first pipe system to adjust; and (ii) transmitting to at least one remote device, via the communication interface, at least one of the fluid data and a notification relating to the fluid data.
15 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
executing a field data collection procedure for model training by configuring the fluid monitoring system to enter a data collection mode;
causing cycling of the flow control valve of the first pipe system through different flow positions to induce various flow rates of fluid through the first pipe system; and
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system.
16 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
causing uploading of the collected field measurement data to a PipeX Server system for model training;
training a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data; and
deploying the first trained model to the fluid monitoring system.
17 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
initiating training of a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data;
storing a digital representation of the first trained model at the fluid monitoring system; and
analyzing, at the fluid monitoring system and using the stored digital representation of first trained model, the fluid data to determine whether the fluid data is indicative of a normal condition or an abnormal condition.
18 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
initiating training of a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data;
storing a digital representation of the first trained model at the fluid monitoring system; and
analyzing, using the stored digital representation of first trained model, the fluid data to determine whether the fluid data is indicative of a normal condition or an abnormal condition, wherein the analyzing is performed without requiring access to cloud connectivity.
19 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model; and
analyzing, the fluid data using the first customized machine learning-based inference model to determine whether the fluid data is indicative of a normal condition or an abnormal condition.
20 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model;
analyzing, the fluid data using the first customized machine learning-based inference model to determine whether the fluid data is indicative of a normal condition or an abnormal condition; and
generating and transmitting a first alert notification upon detecting conditions indicative of a predicted abnormal condition of the first pipe system.
21 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
collecting field measurement data corresponding to a plurality of different flow rates of the fluid through the first pipe system;
initiating training of a first customized machine learning-based inference model for the fluid monitoring system using the collected field measurement data;
storing a digital representation of the first trained model at the fluid monitoring system; and
generating predictions relating to an operational state or health status of the first pipe system using the stored digital representation of first trained model and the fluid data.
22 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model; and
generating predictions relating to an operational state or health status of the first pipe system using the stored digital representation of first trained model and the fluid data.
23 . The method of claim 14 , further comprising causing the at least one processor to execute additional instructions for:
communicating with a first computing system configured to run a PipeX software application configured or designed to facilitate operation of the fluid monitoring system; and
utilizing the first computing system to facilitate communication between the fluid monitoring system and a PipeX Server system.
24 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
storing at a first memory of the fluid monitoring system a first customized machine learning-based inference model;
generating predictions relating to an operational state or health status of the first pipe system using the stored digital representation of first trained model and the fluid data; and
initiating updating of the first inference model in response to detected prediction inaccuracies.
25 . The method of claim 14 further comprising causing the at least one processor to execute additional instructions for:
actively evaluating a mounting integrity of the first set of sensors to the first pipe or conduit by utilizing temperature differential analysis;
determining whether mounting integrity of the first set of sensors to the first pipe or conduit is indicative of improper sensor attachment to the first pipe or conduit; and
generating and transmitting a first alert notification in response to detecting conditions indicative of improper sensor attachment to the first pipe or conduit.
26 . The method of claim 14 :
wherein the fluid monitoring system further comprises a first temperature sensor and a second temperature sensor;
the method further comprising causing the at least one processor to execute additional instructions for:
using the first temperature sensor to measure a temperature of the first pipe or conduit;
using the second temperature sensor to measure a temperature of an ambient environment surrounding the first pipe or conduit;
performing a comparative analysis of the first and second temperatures to detect discrepancies indicative of improper sensor attachment to the first pipe or conduit; and
initiating, in response to detecting conditions indicative of improper sensor attachment to the first pipe or conduit, a first action for facilitating adjustment of the mounting of the first set of sensors to the first pipe or conduit.