【摘要】 【目标】针对高寒地区公路隧道水消防电伴热系统普遍存在温度控制智能化程度低、系统能耗大的问题，提出一种基于深度学习算法的隧道水消防电伴热智能降耗方法。【方法】采用本地控制和远程优化决策两级控制模型。首先，在本地控制级，根据能量守恒定律与热平衡方程获得具有大惯性特性的电伴热系统数学模型来设计模糊PID控制器，使其能够自适应地调整控制参数以此实现温度的精准调控；在远程优化决策级，基于一维卷积神经网络(1DCNN)和XGBoost模型构建了以环境温度、电伴热带功率、设定温度、电伴热系统实际温度等关键参数与加热效率之间复杂非线性的映射模型，通过遗传优化算法对温度设定值进行单目标优化，以保证系统始终在最高加热效率与最优能效状态下运行。【结果】与深度置信网络模型相比，1DCNN-XGBoost组合预测模型对加热效率预测具有更好的预测精度。基于河北省某高速隧道实测数据验证，采用本研究方法相对于传统方式平均能耗减少约40%。【结论】1DCNN-XGBoost组合预测模型与遗传优化算法的降耗方法能显著降低隧道水消防电伴热系统能耗，提升温度控制的智能化程度。本研究为高寒地区公路隧道水消防伴热系统提供了有效的节能控制方案。更多还原

【Abstract】 [Objective]Aiming at the common problems of low intelligent temperature control and high energy consumption in water fire-fighting electric heat tracing system for highway tunnel in alpine regions, this study proposes a deep learning-based intelligent energy conservation method for tunnel water fire-fighting electric heat tracing. [Method]The proposed method employed two-tier control architecture comprising local control and remote optimization decision-making. First, at the local control tier, a mathematical model of electric heat tracing system exhibiting large-inertia characteristics was derived based on the law of energy conservation and thermal balance equations. This model was used to design the fuzzy PID controller capable of adaptively adjusting control parameters to achieve precise temperature regulation. At the remote optimization decision-making tier, a complex nonlinear mapping model between key parameters(e.g., ambient temperature, electric heat tracing power, temperature setpoint, actual heat tracing system temperature) and heating efficiency was constructed by using one-dimensional convolutional neural networks(1DCNN) and XGBoost. A genetic optimization algorithm was then applied to perform single-objective optimization on temperature setpoint, ensuring system consistently operating in conditions of highest heating efficiency and optimal energy performance. [Result]Compared with deep belief network model, 1DCNN-XGBoost combined prediction model demonstrates superior prediction accuracy for heating efficiency. The validation based on measured data from a highway tunnel in Hebei Province indicates that the proposed method achieves an average reduction in energy consumption of approximately 40% compared with traditional approaches. [Conclusion]1DCNN-XGBoost combined prediction model coupled with the genetic optimization algorithm substantially reduces energy consumption in tunnel water fire-fighting electric heat tracing system, and significantly enhances the intelligence of temperature control.The findings provide effective energy-saving control scheme for alpine tunnel water fire-fighting electric heating tracing system.更多还原