运营商学习城市水净化水力动力学和颗粒物运输与物理信息的神经网络
Haochen Li1, Mohamed Shatarah1
1Water Infrastructure Laboratory, Department of Civil and Environmental Engineering, University of Tennessee, Knoxville, TN 37996, USA.
Water research
|January 19, 2024
概括
一个新的复合神经网络 (CPNN) 模型有效地预测了水基础设施的水力动力学和颗粒物运输. 这种基于物理的机器学习方法为改善水系统设计提供了更快,更可靠的预测.
科学领域:
- 环境工程环境工程
- 计算流体动力学 (CFD) 是一种计算流体动力学.
- 科学机器学习 (ML)
背景情况:
- 计算流体动力学 (CFD) 对水基础设施至关重要,但由于高计算成本和技能要求,在实践中受到限制.
- 现有的机器学习模型经常充当"黑子",缺乏物理原理的直接集成.
- 需要有效,准确的替代模型来实时分析和设计水系统.
研究的目的:
- 开发一个复合神经网络 (CPNN) 模型,整合基于物理的机器学习,用于水基础设施分析.
- 为了能够准确地预测清理系统中的流动水力学和颗粒物 (PM) 运输.
- 为工程应用创建一个计算效率高的替代传统的CFD模拟.
主要方法:
- 开发了一个复合神经网络 (CPNN),结合了深度运营者网络 (DeepONet) 编码器和物理信息神经网络 (PINN) 解码器.
- 直接将物理原理纳入CPNN架构,用于过程解决和操作员学习.
- 在不同盆地几何形状和负载条件上训练并验证了CPNN,将预测与CFD模拟进行比较.
主要成果:
- 与CFD模拟相比,CPNN模型实现了显著更高的计算效率 (毫秒).
- 盆地水力动力学的预测准确性显示,在10,000个测试案例中,R2在66.4%的测试中高于0.8,在89.2%的测试案例中高于0.4.
- 在预测颗粒物度方面也观察到类似的强大性能,并获得了对流动模式的几何影响的见解.
结论:
- 开发的CPNN为水基础设施规划和设计提供了一个计算效率高和强大的替代模型.
- 基于物理的机器学习方法可以克服传统的CFD和黑盒机器学习模型的局限性.
- 这项工作是迈向实时,高保真度优化和水基础设施监管的重要一步.
相关概念视频
01:30Typical Model Studies
359
Fluid mechanics model studies often utilize scaled-down systems to predict fluid behavior in full-scale environments, such as river flows, dam spillways, and structures interacting with open surfaces. Maintaining Froude number similarity in river models is crucial, as it replicates surface flow features like wave patterns and velocities.
359
01:12Modeling and Similitude
267
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
267
01:24Rapidly Varying Flow
63
Rapidly varying flow (RVF) in open channels is characterized by abrupt changes in flow depth over a short distance, with the rate of depth change relative to distance often approaching unity. These flows are inherently complex due to their transient and multi-dimensional nature, making exact analysis difficult. However, approximate solutions using simplified models provide valuable insights into their behavior.Key Features of Rapidly Varying FlowRVF is commonly observed in scenarios involving...
63
01:27Laminar Flow
1.1K
Laminar flow represents a smooth, orderly fluid motion where particles move along parallel paths, resulting in minimal mixing between layers. Streamlined particle paths characterize this flow regime and occur under conditions where viscous forces dominate over inertial forces. The distinction between laminar, transitional, and turbulent flow is primarily determined by the Reynolds number, a dimensionless quantity calculated as:
1.1K
01:16Hydraulic Jump: Problem Solving
61
To analyze a hydraulic jump in a rectangular channel with a flow speed of 6 meters per second, follow these steps:Calculate Effective Upstream Velocity:When the downstream gate closes, a hydraulic jump forms, traveling upstream at 2 meters per second. This wave speed combines with the initial channel flow velocity, creating an effective upstream velocity.Identify Flow Velocities Before and After the Hydraulic Jump:Upstream of the hydraulic jump, the effective flow velocity includes both the...
61
01:23Plane Potential Flows
393
Plane potential flows simplify fluid motion by assuming the fluid to be irrotational and incompressible. These characteristics allow these flows to be described by a velocity potential function, ϕ, representing the flow speed in a given direction, and a stream function, ψ, that visualizes the flow path, both governed by Laplace's equation. These parameters help in estimating flow patterns, velocity distributions, and pressure fields around various hydraulic structures.
Uniform...
Uniform...
393