Development of Failure Prediction Models for Subsea Blowout Preventers Using Data Analytics and AI

2021 ◽  
Author(s):  
Rodrigo Chamusca Machado ◽  
Fabbio Leite ◽  
Cristiano Xavier ◽  
Alberto Albuquerque ◽  
Samuel Lima ◽  
...  
Keyword(s):  
Machine Learning ◽  
Real Time ◽  
Data Analytics ◽  
Historical Data ◽  
Learning Algorithms ◽  
Failure Prediction ◽  
Machine Learning Algorithms ◽  
Time Data ◽  
Real Time Data ◽  
Operational Data

Objectives/Scope This paper presents how a brazilian Drilling Contractor and a startup built a partnership to optimize the maintenance window of subsea blowout preventers (BOPs) using condition-based maintenance (CBM). It showcases examples of insights about the operational conditions of its components, which were obtained by applying machine learning techniques in real time and historic, structured or unstructured, data. Methods, Procedures, Process From unstructured and structured historical data, which are generated daily from BOP operations, a knowledge bank was built and used to develop normal functioning models. This has been possible even without real-time data, as it has been tested with large sets of operational data collected from event log text files. Software retrieves the data from Event Loggers and creates structured database, comprising analog variables, warnings, alarms and system information. Using machine learning algorithms, the historical data is then used to develop normal behavior modeling for the target components. Thereby, it is possible to use the event logger or real time data to identify abnormal operation moments and detect failure patterns. Critical situations are immediately transmitted to the RTOC (Real-time Operations Center) and management team, while less critical alerts are recorded in the system for further investigation. Results, Observations, Conclusions During the implementation period, Drilling Contractor was able to identify a BOP failure using the detection algorithms and used 100% of the information generated by the system and reports to efficiently plan for equipment maintenance. The system has also been intensively used for incident investigation, helping to identify root causes through data analytics and retro-feeding the machine learning algorithms for future automated failure predictions. This development is expected to significantly reduce the risk of BOP retrieval during the operation for corrective maintenance, increased staff efficiency in maintenance activities, reducing the risk of downtime and improving the scope of maintenance during operational windows, and finally reduction in the cost of spare parts replacementduring maintenance without impact on operational safety. Novel/Additive Information For the near future, the plan is to integrate the system with the Computerized Maintenance Management System (CMMS), checking for historical maintenance, overdue maintenance, certifications, at the same place and time that we are getting real-time operational data and insights. Using real-time data as input, we expect to expand the failure prediction application for other BOP parts (such as regulators, shuttle valves, SPMs (Submounted Plate valves), etc) and increase the applicability for other critical equipment on the rig.

Stuck Pipe Early Detection on Extended Reach Wells Using Ensemble Method of Machine Learning

2021 ◽  
Author(s):  
Rushad Ravilievich Rakhimov ◽  
Oleg Valerievich Zhdaneev ◽  
Konstantin Nikolaevich Frolov ◽  
Maxim Pavlovich Babich
Keyword(s):  
Machine Learning ◽  
Real Time ◽  
Historical Data ◽  
Learning Algorithm ◽  
Ensemble Method ◽  
Machine Learning Algorithm ◽  
Time Data ◽  
Machine Learning Model ◽  
Real Time Data ◽  
The Ensemble Method

Abstract The ultimate objective of this paper is to describe the experience of using a machine learning model prepared by the ensemble method to prevent stuck pipe events during well construction process on extended reach wells. The tasks performed include collecting, analyzing and cleaning historical data, selecting and preparing a machine learning model, testing it on real-time data by means of desktop application. The idea is to display the solution at the rig floor, allowing Driller to quickly take actions for prevention of stuck pipe event. Historical data mining and analysis were performed using software for remote monitoring. Preparation, labelling and cleaning of historical and real-time data were executed using programmable scripts and big data techniques. The machine learning algorithm was developed using the ensemble method, which allows to combine several models to improve the final result. On the field of interest, the most common type of stuck pipe are solids induced pack offs. They occur due to insufficient hole cleaning from drilled cuttings and wellbore collapse due to rocks instability. Stuck pipe prevention on extended reach drilling (ERD) wells requires holistic approach meanwhile final role is assigned to the driller. Due to continuously exceeding ERD envelope and increased workloads on both personnel and drilling equipment, the effectiveness of preventing accidents is deteriorating. This leads to severe consequences: Bottom Hole Assembly lost in hole, the necessity to re-drill the bore and eventually to increased Non-Productive Time (NPT). Developed application based on ensemble machine learning algorithm shows prediction accuracy above 94%. Reacting on alarms, driller can quickly take measures to prevent downhole accidents during well construction of ERD wells.

Emergent Advancements in M.tb Detection: An Artificial Intelligence Approach (Preprint)

2021 ◽  
Author(s):  
Jasleen Kaur ◽  
Shruti Kapoor ◽  
Maninder Singh ◽  
Parvinderjit Singh Kohli ◽  
Urvinder Singh ◽  
...  
Keyword(s):  
Machine Learning ◽  
Early Detection ◽  
Real Time ◽  
Automatic Detection ◽  
Machine Learning Algorithms ◽  
Real Time Monitoring ◽  
Time Data ◽  
The Real ◽  
Handheld Device ◽  
Real Time Data

BACKGROUND Infectious diseases are the major cause of mortality across the globe. Tuberculosis is one such infectious disease which is in the top 10 deaths causing diseases in developing as well as developed countries. The biosensors have emerged as a promising approach to attain the early detection of the pathogenic infection with accuracy and precision. However, the main challenge with biosensors is real time data monitoring preferentially reversible and label free measurements of certain analytes. Integration of biosensor and Artificial Intelligence (AI) approach would enable better acquisition of patient’s data in real time manner enabling automatic detection and monitoring of Mycobacterium tuberculosis (M.tb.) at an early stage. Here we propose a biosensor based smart handheld device that can be designed for automatic detection and real time monitoring of M.tb from varied analytic sources including DNA, proteins and biochemical metabolites. The collected data would be continuously transferred to the connected cloud integrated with AI based clinical decision support systems (CDSS) which may consist of the machine learning based analysis model useful in studying the patterns of disease infestation, progression, early detection and treatment. The proposed system may get deployed in different collaborating centres for validation and collecting the real time data. OBJECTIVE To propose a biosensor based smart handheld device that can be designed for automatic detection and real time monitoring of M.tb from varied analytic sources including DNA, proteins and biochemical metabolites. METHODS The Major challenges for control and early detection of the Mycobacterium tuberculosis were studied based upon the literature survey. Based upon the observed challenges, the biosensor based smart handheld device has been proposed for automatic detection and real time monitoring of M.tb from varied analytic sources including DNA, proteins and biochemical metabolites. RESULTS In this viewpoint, we propose an application based novel approach of combining AI based machine learning algorithms on the real time data collected with the use of biosensor technology which can serve as a point of care system for early diagnosis of the disease which would be low cost, simple, responsive, measurable, can diagnose and distinguish between active and passive cases, include single patient visits, cause considerable inconvenience, can evaluate the cough sample, require minimum material aid and experienced staff, and is user-friendly. CONCLUSIONS In this viewpoint, we propose an application based novel approach of combining AI based machine learning algorithms on the real time data collected with the use of biosensor technology which can serve as a point of care system for early diagnosis of the disease which would be low cost, simple, responsive, measurable, can diagnose and distinguish between active and passive cases, include single patient visits, cause considerable inconvenience, can evaluate the cough sample, require minimum material aid and experienced staff, and is user-friendly.

Covid-19 Pandemic: Data Analysis and Forecasting using Machine Learning Algorithms (Preprint)

2020 ◽  
Author(s):  
Sohini Sengupta ◽  
Sareeta Mugde
Keyword(s):  
Machine Learning ◽  
Time Series ◽  
Real Time ◽  
Machine Learning Algorithms ◽  
Feature Engineering ◽  
Logarithmic Scale ◽  
Time Data ◽  
Transfer Data ◽  
Real Time Data ◽  
Current Scenario

BACKGROUND India reported its first Covid-19 case on 30th Jan 2020 with no practically no significant rise noticed in the number of cases in the month of February but March2020 onwards there has been a huge escalation as has been the case with like many other countries the world over. This research paper analyses COVID -19 data initially at a global level and then drills down to the scenario obtained in India. Data is gathered from multiple data sources- several authentic government websites. Variables such as gender, geographical location, age etc. have been represented using Python and Data Visualization techniques. Getting insights on Trend pattern and time series analysis will bring more clarity to the current scenario as analysis is totally on real-time data(till 19th June). Time Series Analysis and other pattern-recognition techniques are deployed to bring more clarity to the current scenario as analysis is totally based on real-time data(till 19th June,2020) Finally we will use some machine learning algorithms and perform predictive analytics for the near future scenario. We are using a sigmoid model to give an estimate of the day on which we can expect the number of active cases to reach its peak and also when the curve will start to flatten. Strength of Sigmoid model lies in providing a count of date –this is unique feature of analysis in this paper. We are also using certain feature engineering techniques to transfer data into logarithmic scale for better comparison removing any data extremities or outliers. Certain feature engineering techniques have been used to transfer data into logarithmic scale as is affords better comparison removing any data extremities or outliers. Based on the predictions of the short-term interval, our model can be tuned to forecast long time intervals. Needless to mention there are a lot of factors responsible for the cases to come in the upcoming days. One factor being extent of adherence to the rules and restriction imposed by the Government by the citizens of the country. OBJECTIVE Prediction of the number of positive covid cases in the next few months . METHODS Machine Learning Model - Clustering Sigmoid Model RESULTS The model predicts maximum active cases at 258846. The curve flattens by day 154 i.e. 25th September and after that the curve goes down and the number of active cases eventually will decrease. CONCLUSIONS There are a lot of research works going on with respect to vaccines, economic dealings, precautions and reduction of Covid-19 cases. However currently we are at a mid-Covid situation. India along with many other countries are still witnessing upsurge in the number of cases at alarming rates on a daily basis. We have not yet reached the peak. Therefore cuff learning and downward growth are also yet to happen. Each day comes out with fresh information and large amount of data. Also there are many other predictive models using machine learning that beyond the scope of this paper. However at the end of the day it is only the precautionary measures we as responsible citizens can take that will help to flatten the curve. We can all join hands together and maintain all rules and regulations strictly. Maintaining social distancing, taking the lockdown seriously is the only key. This study is based on real time data and will be useful for certain key stakeholders like government officials, healthcare workers to prepare a combat plan along with stringent measures. Also the study will help mathematicians and statisticians to predict outbreak numbers more accurately.

Machine Learning for Real-time Data Reduction in Cloud of Things

2020 ◽  
Author(s):  
Atheer Alahmed ◽  
Amal Alrasheedi ◽  
Maha Alharbi ◽  
Norah Alrebdi ◽  
Marwan Aleasa ◽  
...  
Keyword(s):  
Machine Learning ◽  
Real Time ◽  
Data Reduction ◽  
Time Data ◽  
Real Time Data

scholarly journals An LSTM-Based Method Considering History and Real-Time Data for Passenger Flow Prediction

2020 ◽  
Vol 10 (11) ◽  
pp. 3788 ◽  
Author(s):  
Qi Ouyang ◽  
Yongbo Lv ◽  
Jihui Ma ◽  
Jing Li
Keyword(s):  
Feature Extraction ◽  
Real Time ◽  
Short Term Memory ◽  
Historical Data ◽  
Time Interval ◽  
Information Coding ◽  
Time Data ◽  
Passenger Flow ◽  
Flow Prediction ◽  
Real Time Data

With the development of big data and deep learning, bus passenger flow prediction considering real-time data becomes possible. Real-time traffic flow prediction helps to grasp real-time passenger flow dynamics, provide early warning for a sudden passenger flow and data support for real-time bus plan changes, and improve the stability of urban transportation systems. To solve the problem of passenger flow prediction considering real-time data, this paper proposes a novel passenger flow prediction network model based on long short-term memory (LSTM) networks. The model includes four parts: feature extraction based on Xgboost model, information coding based on historical data, information coding based on real-time data, and decoding based on a multi-layer neural network. In the feature extraction part, the data dimension is increased by fusing bus data and points of interest to improve the number of parameters and model accuracy. In the historical information coding part, we use the date as the index in the LSTM structure to encode historical data and provide relevant information for prediction; in the real-time data coding part, the daily half-hour time interval is used as the index to encode real-time data and provide real-time prediction information; in the decoding part, the passenger flow data for the next two 30 min interval outputs by decoding all the information. To our best knowledge, it is the first time to real-time information has been taken into consideration in passenger flow prediction based on LSTM. The proposed model can achieve better accuracy compared to the LSTM and other baseline methods.

Sign in / Sign up

Export Citation Format

Share Document