宁夏测控闸门测流精度试验及流量预测模型

Flow Measurement Accuracy Test and Flow Prediction Model for Measurement and Control Gates in Ningxia

测控一体化闸门作为一种新型测控设备,通过测箱内分层布置的超声波探头进行流量计算,但当水流流态紊乱或流态快速转变时,会出现个别探头无法准确获取路径流速,采用流速面积法进行流量计算时会产生较大误差,目前尚缺少一种对测箱内流量进行率定的方法。针对宁夏引黄灌区广泛使用的箱涵式测控一体化闸门开展了大量室内测箱过流试验,分析了淹没水深和探头遮挡对测流精度的影响。基于450组过流试验样本数据和反向传播(back propagation,BP)神经网络及其优化算法,建立了可自适应流态变化的测控一体化闸门流量预测模型,并对预测模型的有效性和精确性进行了验证。通过与测箱内层流速计算结果进行对比,探讨了优化神经网络模型在测控一体化闸门流量率定中的应用效果。结果表明:测箱淹没水深不应小于10 cm,为提高设备抗干扰能力,探头层数应至少满足8层要求;相较于其他3种优化算法,经麻雀搜索算法(sparrow search algorithm,SSA)或粒子群算法(particle swarm algorithm,PSO)优化的BP模型具有更高的预测精度和更集中的误差分布;综合分析预测精度、误差分布和运行时间3个方面,对预测精度有较高要求时可选用SSA-BP模型,对运行时间有较高要求可选用PSO-BP模型。该预测方法可为宁夏引黄灌区测控一体化闸门的流速校准和异常点剔除提供参考。

The integrated measurement and control gate,as a new type of measurement and control equipment,calculates the flow rate through ultrasonic probes arranged in layers inside the measurement box.However,when the flow pattern is turbulent or the flow pattern is rapidly changing,the individual probe will not be able to accurately obtain the path flow velocity,and the flow velocity area method will produce large errors when used for flow calculation.There is a lack of a method to rate the flow in the measurement box.A large number of indoor measuring box overflow tests were conducted on the box culvert type integrated measurement and control gate widely used in the Ningxia Yellow Diversion Irrigation Area.The effects of submerged water depth and probe shading on flow measurement accuracy were analyzed.Based on 450 sets of overflow test sample data and BP neural network and its optimization algorithm,a integrated measurement and control gate flow prediction model with adaptive flow state change capability was established,and the effectiveness and accuracy of the prediction model were verified.The results show that the submerged depth of the measuring box should not be less than 10 cm.To improve the anti-interference ability of the equipment,the number of probe layers should meet the requirement of at least 8 layers.Compared to the other three optimization algorithms,the back propagation(BP)model optimized by the sparrow search algorithm(SSA)or the particle swarm algorithm(PSO)has a higher prediction accuracy and a more concentrated distribution of the error.Comprehensively analyzing the three aspects of prediction accuracy,error distribution and running time,SSA-BP model can be selected when there is a high demand for prediction accuracy,and PSO-BP model can be selected when there is a high demand for running time.The prediction method can provide a reference for flow rate calibration and anomaly rejection of integrated gate for measurement and control in Ningxia Yellow Diversion Irrigation Area.