Data Mining Tools to Improve Buried Water Infrastructure Decision-Making

随着基础设施的使用寿命接近尾声，加拿大各地的地下水基础设施正在恶化。虽然对基础设施的组成部分进行补救是可行的，但确定最佳补救行动和在水基础设施的不同部分之间优先考虑投资战略是极具挑战性的。由于供水和下水管道的失败率受到许多外部因素的影响，因此在决策方面面临挑战。许多现有的方法依赖于非线性回归，这种方法受到方法中隐含的假设和约束的高度约束。本研究将采用具有不同架构类型的新数据挖掘模型，以评估其提供处理能力的能力，这些处理能力能够执行大量并行处理计算，并且不受非线性回归中存在的限制性假设的约束。

Buried water infrastructure throughout Canada is deteriorating as the infrastructure approaches the end of its useful life. While remediation of components of the infrastructure is feasible, identifying the best remedial actions and prioritizing investment strategies between different portions of the water infrastructure, is extremely challenging. The challenges in decision-making arise since the rates of failure of water and sewer pipes are influenced by numerous external factors. Many existing approaches have relied upon nonlinear regression, an approach which is highly constrained by assumptions and constraints implicit in the methodology. This research will employ new data mining models with different types of architecture, to assess their capabilities to provide processing capabilities which are capable of performing massive parallel processing computations and are not constrained by the limiting assumptions existing in nonlinear regression.