Novel autonomous artificially intelligent system to predict pipe leaks

Abstract

Embodiments of the disclosure are directed towards pipe leak prediction systems configured to predict whether a pipe (e.g., a utility pipe carrying some substance such as waster) is likely to leak. The pipe leak prediction system may include one or more predictive models based on one or more machine learning techniques, and a predictive model can be trained using data for the characteristics of various pipes in order to determine the patterns associated with pipes without leaks and the patterns associated with pipes with leaks. A predictive model can be validated, used to construct a confusion matrix, and used to generate insights and inferences associated with the determinant variables used to make the predictions. The predictive model can be applied to data for various pipes in order to predict which of those pipes will leak. Any pipes that are identified as likely to leak can be assigned for further investigation for potential repair or preventative maintenance.

What is claimed is: 
     
         1 . A computer-implemented method for predicting pipe leaks, the method comprising:
 receiving a training dataset including first data items and known leaks associated with respective pipes of a first plurality of pipes, wherein the first data items include characteristics of the respective pipes, wherein the characteristics of the respective pipes include only lengths of the first plurality of pipes, soil resistivity at the first plurality of pipes, pressure rating of the first plurality of pipes, and elevation of the first plurality of pipes;   applying a supervised machine learning technique to generate a predictive model configured to determine a leak prediction of a pipe by training the predictive model based on the first data items associated with respective pipes of the first plurality of pipes;   receiving, a validation dataset including second data items and known leaks associated with respective pipes of a second plurality of pipes, wherein the second data items include characteristics of the respective pipes;   validating the predictive model by at least comparing a set of leak predictions of the pipes of the second plurality of pipes with known leaks of the pipes of the second plurality of pipes to determine an accuracy of the leak predictions of the pipes of the second plurality of pipes;   accessing a pipeline dataset including third data items associated with a third plurality of pipes; and   applying the predictive model to a pipeline dataset to determine leak predictions of respective pipes of a third plurality of pipes.   
     
     
         2 . The computer-implemented method of  claim 1 , wherein the predictive model includes a random forest model. 
     
     
         3 . The computer-implemented method of  claim 1 , wherein the predictive model includes a logistic regression. 
     
     
         4 . The computer-implemented method of  claim 1 , wherein the predictive model includes a naive Bayes model. 
     
     
         5 . The computer-implemented method of  claim 1 , wherein the method further comprises:
 ordering the pipes of the third plurality of pipes based on the determined leak predictions of the pipes of the third plurality of pipes.   
     
     
         6 . The computer-implemented method of  claim 1 , wherein the method further comprises:
 presenting the determined leak predictions of respective pipes of the third plurality of pipes using a display device.   
     
     
         7 . The computer-implemented method of  claim 1 , wherein the training dataset and the validation dataset are generated by splitting a larger dataset, including fourth data items associated with a fourth plurality of pipes, into the training dataset and the validation dataset. 
     
     
         8 . A computing system comprising:
 one or more data stores storing:
 a training dataset including first data items and known leaks associated with respective pipes of a first plurality of pipes, wherein the first data items include characteristics of the respective pipes; 
 a validation dataset including second data items and known leaks associated with respective pipes of a second plurality of pipes, wherein the second data items include characteristics of the respective pipes; 
   a computer processor; and   a computer readable storage medium storing program instructions configured for execution by the computer processor in order to cause the computer processor to:
 receive a training dataset including first data items and known leaks associated with respective pipes of a first plurality of pipes, wherein the first data items include characteristics of the respective pipes, wherein the characteristics of the respective pipes include only lengths of the first plurality of pipes, soil resistivity at the first plurality of pipes, pressure rating of the first plurality of pipes, and elevation of the first plurality of pipes; 
 apply a supervised machine learning technique to generate a predictive model configured to determine a leak prediction of a pipe by training the predictive model based on the first data items associated with respective pipes of the first plurality of pipes; 
 receive, a validation dataset including second data items and known leaks associated with respective pipes of a second plurality of pipes, wherein the second data items include characteristics of the respective pipes; 
 validate the predictive model by at least comparing a set of leak predictions of the pipes of the second plurality of pipes with known leaks of the pipes of the second plurality of pipes to determine an accuracy of the leak predictions of the pipes of the second plurality of pipes; 
 access a pipeline dataset including third data items associated with a third plurality of pipes; and 
 apply the predictive model to a pipeline dataset to determine leak predictions of respective pipes of a third plurality of pipes. 
   
     
     
         9 . The system of  claim 8 , wherein the predictive model includes a random forest model. 
     
     
         10 . The system of  claim 8 , wherein the predictive model includes a logistic regression. 
     
     
         11 . The system of  claim 8 , wherein the predictive model includes a naïve Bayes model. 
     
     
         12 . The system of  claim 8 , wherein the program instructions, when executed by the computer processor, further cause the computer processor to:
 order the pipes of the third plurality of pipes based on the determined leak predictions of the pipes of the third plurality of pipes.   
     
     
         13 . The system of  claim 8 , wherein the program instructions, when executed by the computer processor, further cause the computer processor to:
 present the determined leak predictions of respective pipes of the third plurality of pipes using a display device.   
     
     
         14 . The system of  claim 8 , wherein the training dataset and the validation dataset are generated by splitting a larger dataset, including fourth data items associated with a fourth plurality of pipes, into the training dataset and the validation dataset. 
     
     
         15 . A non-transitory computer-readable medium storing a set of instructions configured for execution by a computer processor in order to cause the computer processor to:
 receive a training dataset including first data items and known leaks associated with respective pipes of a first plurality of pipes, wherein the first data items include characteristics of the respective pipes, wherein the characteristics of the respective pipes include only lengths of the first plurality of pipes, soil resistivity at the first plurality of pipes, pressure rating of the first plurality of pipes, and elevation of the first plurality of pipes;   apply a supervised machine learning technique to generate a predictive model configured to determine a leak prediction of a pipe by training the predictive model based on the first data items associated with respective pipes of the first plurality of pipes;   receive, a validation dataset including second data items and known leaks associated with respective pipes of a second plurality of pipes, wherein the second data items include characteristics of the respective pipes;   validate the predictive model by at least comparing a set of leak predictions of the pipes of the second plurality of pipes with known leaks of the pipes of the second plurality of pipes to determine an accuracy of the leak predictions of the pipes of the second plurality of pipes;   access a pipeline dataset including third data items associated with a third plurality of pipes; and   apply the predictive model to a pipeline dataset to determine leak predictions of respective pipes of a third plurality of pipes.   
     
     
         16 . The non-transitory computer-readable medium of  claim 15 , wherein the predictive model includes a random forest model. 
     
     
         17 . The non-transitory computer-readable medium of  claim 15 , wherein the predictive model includes a logistic regression. 
     
     
         18 . The non-transitory computer-readable medium of  claim 15 , wherein the predictive model includes a naive Bayes model. 
     
     
         19 . The non-transitory computer-readable medium of  claim 15 , wherein the instructions further cause the computer processor to:
 ordering the pipes of the third plurality of pipes based on the determined leak predictions of the pipes of the third plurality of pipes.   
     
     
         20 . The non-transitory computer-readable medium of  claim 15 , wherein the training dataset and the validation dataset are generated by splitting a larger dataset, including fourth data items associated with a fourth plurality of pipes, into the training dataset and the validation dataset.