公路隧道风井送排式纵向通风系统智能控制研究

发布时间：2018-05-01 18:54

本文选题：公路隧道 + 送排式纵向通风　；参考：《北京交通大学》2015年博士论文

【摘要】：公路隧道作为封闭的地下空间,正常运营时需要通风来稀释和排出机动车尾排污染物,为驾乘人员提供安全、舒适和卫生的行车环境;火灾时需要通风来控制烟气流向并提供相应的新风,为驾乘和救援人员提供有利的逃生与救援环境。在我国,长度大于5000m的高速公路隧道通常采用通风井送排式+射流风机的组合纵向通风方式,其通风效能的发挥需要射流风机群与送排风机的联合作用,正常运营和火灾工况时如何控制通风系统的运转,这成为制约长大公路隧道建设与运营的一项重要课题。本文依托张涿高速公路分水岭特长公路隧道工程,应用流体力学和空气动力学理论,通过模型试验、数值模拟和现场测试的综合研究手段,结合人工智能技术,对公路隧道竖井送排式纵向通风系统性能、火灾条件下洞内温度及烟气浓度特征、公路隧道运营期正常运营及火灾工况通风系统智能优化控制等展开深入研究。取得主要成果如下:(1)依据流体动力学及相似模型试验理论,研制加工了分水岭特长隧道单斜井送排式通风系统物理模型试验台,并对模型试验系统基本性能进行了测试,验证了模型试验台的可靠性,采用该物理模型进行了竖井送排式机械通风系统影响试验、分水岭隧道正常运营工况风机梯次关停试验、隧道火灾工况烟气模拟试验等,为后续系统优化控制提供理论基础。(2)基于人工智能与现代优化理论,提出了隧道运营交通量预测的粒子群优化与高斯过程回归耦合(PSO-GPR)模型,对隧道短期交通量预测平均相对误差仅为1.96%,最大不到5%,该模型可以应用于隧道运营期间短期交通量的预测,为隧道运营期通风系统前馈控制提供实时交通量数据。(3)在交通量预测的基础上,基于分水岭隧道正常运营工况三维数值模拟结果,以洞内中轴线1.8m高处纵截面CO最高浓度为控制目标,建立了隧道运营期正常运营工况下通风系统优化控制的进化支持向量回归模型,利用该模型可以实现不同行车速度和不同控制指标情况下通风系统的优化控制,并采用现场实测数据对优化控制模型进行了验证。(4)基于分水岭特长隧道运营期火灾工况下的三维数值模拟,引入进化高斯过程回归算法,提出了隧道运营期火灾工况下通风系统优化的有约束多目标规划模型,即火灾发生时,在保证烟气不回流的前提下,尽量减少火源下游30米1.8米高处烟气的温度和浓度。(5)依据现代优化理论,建立了基于罚函数法的火灾工况下隧道通风系统优化的PSO与BP神经网络耦合求解模型,并与火灾数值模拟相结合,实现了火灾工况下隧道通风系统的有约束多目标规划,为火灾工况时通风系统的智能控制提出了新的解决方法。
[Abstract]:As a closed underground space, highway tunnel needs ventilation to dilute and discharge vehicle exhaust pollutants in normal operation to provide safe, comfortable and hygienic driving environment for drivers. In case of fire, ventilation is needed to control the flow of smoke and provide the corresponding fresh air to provide a favorable escape and rescue environment for drivers and rescuers. In our country, the combined longitudinal ventilation of jet fan is usually used in freeway tunnel with length more than 5000m, and its ventilation efficiency needs the combined action of jet fan group and exhaust fan. How to control the operation of ventilation system in normal operation and fire condition has become an important problem that restricts the construction and operation of long road tunnel. Based on the Zhangzhuo Expressway Dianshui Extra-long Highway Tunnel Project, this paper applies the theories of fluid dynamics and aerodynamics, and combines the artificial intelligence technology with the comprehensive research means of model test, numerical simulation and field test. The performance of vertical ventilation system of highway tunnel shaft, the characteristics of temperature and smoke concentration in tunnel under fire conditions, the intelligent optimization control of ventilation system in normal operation period and fire condition of highway tunnel are studied. The main results are as follows: (1) based on the theory of fluid dynamics and similar model test, the physical model test bench of single inclined shaft feeding ventilation system for long and long tunnel is developed, and the basic performance of the model test system is tested. The reliability of the model test-bed is verified. The physical model is used to carry out the influence test on the mechanical ventilation system of the shaft feeding and discharging type, the wind turbine trapezoidal shutdown test under the normal operating condition of the watershed tunnel, the smoke simulation test under the tunnel fire condition, etc. (2) based on artificial intelligence and modern optimization theory, a PSO -GPRA model for forecasting traffic volume of tunnel operation is proposed, which is based on the regression coupling between particle swarm optimization (PSO) and Gao Si process. The average relative error of short-term traffic volume prediction is only 1.96, and the maximum is less than 5. The model can be used to predict short-term traffic volume during tunnel operation. To provide real-time traffic volume data for feedforward control of ventilation system in tunnel operation period. Based on traffic volume prediction, 3D numerical simulation results of normal operating conditions of watershed tunnel are presented. With the maximum concentration of CO in the longitudinal section of 1.8 m high in the central axis of the tunnel as the control target, an evolutionary support vector regression model for optimal control of ventilation system under normal operating conditions during tunnel operation is established. The model can be used to realize the optimal control of ventilation system with different driving speed and different control index. The optimized control model is verified by the field measured data. Based on the 3D numerical simulation under the fire condition of the watershed super long tunnel, an evolutionary Gao Si process regression algorithm is introduced. A constrained multi-objective programming model for optimization of ventilation system under fire condition in tunnel operation period is proposed, that is, when the fire occurs, the flue gas is not reflux. According to modern optimization theory, a coupling model of PSO and BP neural network for optimization of tunnel ventilation system under fire condition is established based on penalty function method. Combined with numerical simulation of fire, the constrained multi-objective programming of tunnel ventilation system under fire condition is realized, which provides a new solution for intelligent control of ventilation system under fire condition.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U453.5

【相似文献】