Pipe break prediction based on evolutionary data-driven methods with brief recorded data

Xu, Qiang; Chen, Qiuwen; Li, Weifeng; Ma, Jinfeng

doi:10.1016/j.ress.2011.03.010

Cited by 53 publications

(30 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Regarding the statistical models, the K-means clustering approach was applied to separate the data into a number of specified clusters based on pipe diameter, length, and age. Other variables were excluded as using the pipes' attributes helps to obtain models with greater statistical significance [5,53]. To determinate the optimal number of clusters, the Davies-Bouldin criterion was applied.…”

Section: Statistical Modelsmentioning

confidence: 99%

Comparison of Statistical and Machine Learning Models for Pipe Failure Modeling in Water Distribution Networks

Giraldo-González

Rodríguez

2020

Water

View full text Add to dashboard Cite

The application of statistical and Machine Learning models plays a critical role in planning and decision support processes for efficient and reliable Water Distribution Network (WDN) management. Failure models can provide valuable information for prioritizing system rehabilitation even in data scarcity scenarios, such as developing countries. Few studies have analyzed the performance of more than two models, and examples of case studies in developing countries are insufficient. This study compares various statistical and Machine Learning models to provide useful information to practitioners for the selection of a suitable pipe failure model according to information availability and network characteristics. Three statistical models (i.e., Linear, Poisson, and Evolutionary Polynomial Regressions) were used for failure prediction in groups of pipes. Machine Learning approaches, particularly Gradient-Boosted Tree (GBT), Bayes, Support Vector Machines and Artificial Neuronal Networks (ANNs), were compared in predicting individual pipe failure rates. The proposed approach was applied to a WDN in Bogotá (Colombia). The statistical models showed an acceptable performance (R2 between 0.695 and 0.927), but the Poisson Regression was the most suitable for predicting failures in pipes with lower failure rates. Regarding Machine Learning models, Bayes and ANNs exhibited low performance in the prediction of pipe failure condition. The GBT approach had the best performing classifier.

show abstract

Section: Statistical Modelsmentioning

confidence: 99%

Comparison of Statistical and Machine Learning Models for Pipe Failure Modeling in Water Distribution Networks

Giraldo-González

Rodríguez

2020

Water

View full text Add to dashboard Cite

show abstract

“…This technique has been applied to various fields such as pipe break prediction for water distribution systems (Xu et al, 2011), soil moisture content estimation (Elshorbagy and ElBaroudy, 2009) and so on. In this work, it was used to model the relationship between surface area of scale samples and different iron oxides content.…”

Section: Evolutionary Polynomial Regressionmentioning

confidence: 99%

Morphological and physicochemical characteristics of iron corrosion scales formed under different water source histories in a drinking water distribution system

et al. 2012

View full text Add to dashboard Cite

“…New models are incorporated with selected existing models and the new population of models is carried on to the next iteration where the algorithm repeats itself. A few examples where genetic programming has been used to create models addressing many different problems include the evolution of natural laws (Iba 2008, Schmidt andLipson 2009), computer vulnerability testing (Kayacik et al 2011), prediction of longitudinal dispersion coefficients in streams (Azamathulla and Ghani 2011), protein binding sites (Bains et al 2004), interpretation of microbial flow cytometric data (Davey and Davey 2011), embedding and decoding of digital watermarks (Usman et al 2011), synthesis of polymorphic combinational circuits (Gajda and Sekanina 2011), pipe break prediction modeling (Xu et al 2011), estimation of daily pan evaporation (Shiri and Kisi 2011), software engineering predictive modeling (Afzal and Torkar 2011), modular neural network programming (Tsai and Lin 2011), selfreproducing machines (Zykov et al 2005), and steel beam load capacity prediction models (Gandomi et al 2011).…”

Section: Motivationmentioning

confidence: 99%

Integration of Reaction Kinetics Theory and Gene Expression Programming to Infer Reaction Mechanism

White

Srivastava

2017

Applications of Evolutionary Computation

View full text Add to dashboard Cite

Mechanistic mathematical models of biological systems have been used to describe biological phenomena, including human disease, in the hope that one day these models may be used to better understand diseases, as well as to develop and optimize therapeutic strategies. Evolutionary algorithms, such as genetic programming, may be used to symbolically regress mathematical models describing chemical and biochemical species for which kinetic data are available. However, current evolutionary algorithms are restricted to the formulation of simple or approximate models due to the computational cost of evolving mechanistic models for more complex systems.It was hypothesized that chemical reaction kinetic theory could be used to sufficiently reduce the model search space for an evolutionary algorithm such that it would be possible to infer mechanistic mathematical models of complex biologicalinteractions. An evolutionary algorithm capable of formulating mass action kinetic models of biological systems from time series data sets was developed for a system of nspecies using heuristics from chemical reaction kinetic theory and a gene expression programming (GEP) based approach.Jason Robert White -University of Connecticut, 2014The resulting algorithm was then successfully validated on a general model of viral dynamics that accounted for six pathways relating the change in viral template, viral genome, and viral structural protein concentrations over time.The algorithm was applied to generate cohort-specific models of HIV dynamics from a clinical data set. HIV-1 infection models were defined as sets of two ordinary differential equations describing the change in CD4 + T-cell and HIV-1 concentrations over time. The evolved models were used to generate hypotheses regarding treatment effectiveness and the potential for viral rebound in three cohorts of HIV-1 positive individuals receiving different Highly Active Antiretroviral Therapy (HAART) regimens.It was hypothesized by the algorithm that HAART was effective in stopping HIV-1 propagation in two of the three cohorts studied. In the other cohort, it was hypothesized that HIV-1 continued to propagate and that there was the potential for viral rebound.The result of this work was the development of an algorithm that can be used for the generation of complex mechanistic biological models based upon kinetic data with potential uses in fields ranging from biomedical to biotechnological.

show abstract

Pipe break prediction based on evolutionary data-driven methods with brief recorded data

Cited by 53 publications

References 24 publications

Comparison of Statistical and Machine Learning Models for Pipe Failure Modeling in Water Distribution Networks

Comparison of Statistical and Machine Learning Models for Pipe Failure Modeling in Water Distribution Networks

Morphological and physicochemical characteristics of iron corrosion scales formed under different water source histories in a drinking water distribution system

Integration of Reaction Kinetics Theory and Gene Expression Programming to Infer Reaction Mechanism

Contact Info

Product

Resources

About