基于神经网络PID的疏浚管道泥浆流速控制

Neural network PID-based slurry flow rate control for dredging pipelines

摘要:

疏浚作业中，泥浆管道内物料的组成、粒径、浓度等随水下地形土质等变化很大，易造成流速波动甚至堵管、爆管等故障，因此泥浆流速稳定控制对泥浆输送的效率和安全具有重要意义；疏浚管道输送系统具有非线性、大时滞和参数时变等特征，传统PID控制方法效果不佳，故此将BP神经网络和传统PID控制算法相结合，并将其应用于泥浆流速控制中。以河海大学管道输送实验平台为对象，采用受控自回归CAR模型描述泥泵变频器频率与管道泥浆流速之间的关系，通过实验和数值处理对模型进行离线辨识；在此基础上通过仿真对比传统PID、单神经元PID和BP-PID的流速控制性能，发现BP-PID控制器的超调量仅为3.8%，响应时间为11s，控制性能较好；最后通过在体积浓度~10%到~30%泥浆范围内，泥浆浓度小幅度和大幅度增减实验，对流速控制方法进行了验证，结果表明在浓度平缓或剧烈波动时，采用BP-PID控制算法的流速控制系统，均能够在保证输送安全的前提下，快速、稳定地达到目标流速，具有较好的自适应控制性能。

Abstract:

In dredging operations, the composition, particle size and concentration of the material in the mud pipeline vary greatly with the underwater topography and soil quality, which may cause flow rate fluctuations and even blockage and bursting of the pipeline. Therefore, the stable control of mud flow rate is of great significance to the efficiency and safety of mud conveying. The dredging pipeline conveying system is characterized by nonlinearity, large time lag and time-varying parameters, and the traditional PID control method is not effective. Therefore, BP neural network and traditional PID control algorithm are combined and applied to the slurry flow rate control. The relationship between mud pump inverter frequency and pipe slurry flow rate is described by a controlled autoregressive CAR model with the pipeline conveying experimental platform of Hohai University, and the model is identified offline through experiments and numerical processing. On this basis, the flow rate control performance of conventional PID, single neuron PID and BP-PID are compared by simulation. It is found that the overshoot of BP-PID controller is only 3.8% and the response time is 11s for better control performance. Finally, the flow rate control method was validated by small and large increases and decreases in mud concentration in the range of ~10% to ~30% volume concentration. The results show that the flow rate control system with BP-PID control algorithm can achieve the target flow rate quickly and stably with good adaptive control performance while ensuring the safety of conveying when the concentration is calm or fluctuating drastically.