The Design Of Snowmelt Flash Flood Warning System

Our country has a vast territory, east、northwest snow spread, snowmelt flash floodsoccurred in the context of global warming, the frequency gradually increased. Snowmelt flashfloods monitoring process is severely limited in the complexity of the environmental factorsand uncertainty, the experimental monitoring conditions scarcity and relatively weak researchbase, so far snowmelt flash flood forecasting is still in its infancy, thus the design ofsnowmelt flash flood warning system is necessity and urgency. In response to the currentsituation,to design warning system to forecast trunk river snowmelt flash flood flow based onregional snow melt conditions.Subjects design ideas: To arranged the snowmelt site location in the region using of“from the point to the surface, from the surface to the area”to planning the zone.According tothiessen polygon method to calculate average snow melt total of the zone in certain period oftime. And using hydrological station to monitor the snowmelt flash floods actual river flowvalue in real time. To obtain a set of sample data constituting by the average snow melt totalof the zone, actual river flow and ambient temperature. And making the sample data importedinto snowmelt flash flood warning system model base on BP neural network to trainingparameter. Completing the construction of the snowmelt mountain torrents flow forecastingmodel. Through measuring the average snow melt total of the zone and ambient temperatureimporte into just completed the BP neural network to complete on the the snowmelt rivertorrential flow value of real-time forecast. Comparing with the actual river flood flowmonitored by hydrological station. If the prediction error isn’t within the acceptable errorrange, again the area measured to the total melting snow, the actual river flow as well as theambient temperature as the sample data imported into BP neural network having been built totraining network parameters. controlling the snowmelt flash flood river flow prediction errorwithin a certain range by repeated training-forecast-re-training.Subject specific research work as follows:(1) the design of snowmelt measuringinstrument. the major role to measuring snow melt thickness and ambient temperature, Themain design work including: analysis of the ultrasonic measuring principle of snowthickness and determine the measurement program. In hardware design, mainly to thecompletion of the selection of ultrasonic sensors, ARM microprocessor core devices,and todesign the ultrasonic transmitter and receiver circuit, the signal processing circuit, a datacommunication module and the storage module, and a subsidiary circuit. In software design,mainly to complete the program of the driver of above the circuit module and the wholefunctional system. In order to improve the snow melting the measurement accuracy of themeasuring instrument, add the temperature compensation algorithm, anti-pulse interferenceaverage filtering algorithm and linear compensation algorithm to system program.(2) Thecalculation of the total of the regional average snow melt. First, according to the idea of“from the point to the surface, from the surface to the area” to layout area snowmelt site,and according to thiessen polygon method and calculated the total regional average snow melt.Using the polygon split triangle principle to thiessen polygon area calculations.(3) Flashflood warning system modeling based on BP neural network Snowmelt. First complete set ofcharacteristic parameters of the BP neural network input output, network layers, hidden nodesand activation function. Then importing the measured sample data into the network totraining and getting a set of optimal weights, threshold to complete model. Making the thetotal regional average snow melt and ambient temperature as an input factor import into themodel having been established to predict the actual snowmelt flash floods flow value of river.