Method for detecting anomalies in the well and oil reservoir system using artificial intelligence

Abstract

The main objective of the present invention is to enable continuous monitoring and detect anomalous behavior in wells equipped with intelligent completion automatically by means of a method implemented with artificial intelligence. The present invention applies AI techniques to monitor wells in an oil field and has the ability to understand what the usual behavior of each well would be, based on temperature, pressure and flow rate sensors, and then identify by means of a stochastic technique of selection of outliers which a deviation from usual behavior would be. From the outlier detection, it is possible to quantify an anomaly probability and associate the same with a possible event, such as: sensor failure and loss of data, closure of one of the producing intervals, scale deposition, among others.

What is claimed is: 
     
         1 . A method for detecting anomalies in a well and oil reservoir system using artificial intelligence, the method comprising the steps of:
 a) collecting pressure and temperature data from downhole recorders installed in intelligent completions with multiple zones, as well as flow rate data, stored on one or more servers;   b) pre-processing collected data by standardizing time units and interpolation;   c) reducing dimensionality via Principal Component Analysis;   d) performing Artificial Intelligence training by means of a set of instructions based on a stochastic model for detecting outliers;   e) performing inference of the probability of a data sample being related to an anomaly, said sample being the most recent dataset obtained in a pre-defined time interval;   f) storing results, in which, for each outlier detection, a probability is written back to the server for each data sample, which is highlighted for user analysis.   
     
     
         2 . The method according to  claim 1 , wherein, in step (A), the data is stored in the main memory in tabular format. 
     
     
         3 . The Method according to  claim 1 , wherein step (B) standardizes the time units obtained preferably to the format YYYY-MM-DD hh:mm:ss.mmmm+TZ, together with the interpolation of missing data and the interpolation data of daily flow rate data. 
     
     
         4 . The method according to  claim 1 , wherein the principal component analysis identifies signatures in time series and retains the overall variance of the dataset in a reduced dimension space. 
     
     
         5 . The method according to  claim 4 , wherein said principal component analysis preferably uses a technique chosen from: shapelet transforms, bendford correlation, continuous wavelet transform of the Ricker wavelet, coefficients for the fast Fourier transform, Fourier entropy, Friedrich coefficients, Kurtosis, number of peaks, spectral centroid (mean), variance, slope and kurtosis of the absolute spectrum of the Fourier transform and autocorrelation. 
     
     
         6 . The method according to  claim 1 , wherein, in step (D), said training calculates extreme variations and reports them as anomalies in an unsupervised manner. 
     
     
         7 . The method according to  claim 1 , wherein, in step (D), the behavior considered normal is calibrated for each block corresponding to n days and a score is associated with each block that represents the probability of being an anomaly or not. 
     
     
         8 . The method according to  claim 1 , wherein, in step (D), said blocks are preferably generated every 4 days, wherein each of these intervals is a datapoint to be considered. 
     
     
         9 . The method according to  claim 1 , wherein, in step (E), the most recent period is used and, based on the knowledge acquired by the AI with the historical data of each well, there is determined the probability of an anomaly is occurring at this moment, or in the near past, last four days of the period, in each well. 
     
     
         10 . The method according to  claim 1 , wherein said stochastic method is modified with the insertion of physical features determined by means of at least one of pressure delta (δp), temperature delta (δt), derivatives of pressure and temperature, according to the equation:
 ▮ 
 
     
     
         11 . The method according to  claim 1 , wherein said set of instructions generates a probability matrix, wherein the outlier probability of a datapoint is calculated jointly based on the other datapoints not connecting to the same according to the equation: 
       
         
           
             
               
                 
                   p 
                   ⁡ 
                   ( 
                   
                     
                       x 
                       i 
                     
                     ∈ 
                     
                       C 
                       o 
                     
                   
                   ) 
                 
                 = 
                 
                   
                     ∏ 
                     
                       j 
                       ≠ 
                       1 
                     
                   
                   
                     ( 
                     
                       1 
                       - 
                       
                         b 
                         
                           1 
                           ⁢ 
                           j 
                         
                       
                     
                     ) 
                   
                 
               
               ; 
             
           
         
         where b represents the affinity between datapoint i and j in row i. 
       
     
     
         12 . The method according to  claim 1 , wherein, in step (E), a probability greater than 55% is considered an anomaly occurring at the bottom of the well and/or reservoir. 
     
     
         13 . The method according to  claim 12 , wherein scales occur in the range of 55% to 65%; Cross flow between zones of 65 to 75%; Data recording failure occurs above 80% and Spurious valve movement occurs above 80. 
     
     
         14 . The method according to  claim 1 , wherein, in step (F), the method preferably uses a module for writing PI tags.