Study On Intelligent Runoff Sampling System Based On Neural Network And LoRa Wireless Network

Runoff sampling system has been developed for many years as an important part of runoff plot facilities.The emergence of runoff sampling system solves the problem that the researchers need to go to the runoff plot facilities in the test site to extract the runoff,which is time-consuming and laborious,Nor can timely and effective collection of runoff information and other issues in the rainwater process.With the development of Internet of things,big data,artificial intelligence and sensor technology,runoff sampling system has become more intelligent.This thesis mainly around the runoff sampling system solutions,introduces some design case,the previous also reference some design methods,sums up many of the problems,after repeated experiments and plan changes,finally realizes the runoff gathering intelligence and visualization of gathering information,greatly reduce the workload of soil and water erosion,the researchers.The intelligent runoff sampling system adopts a large amount of electronic control technology to avoid the efficiency and precision problems caused by the traditional mechanical fluid control unit.Due to the relatively weak output signal of turbine flowmeter,the problem of data discontinuity occurred in the previous experiment.By improving the signal detection circuit,the flow signal could be captured normally.Under the same sampling rule,after several sampling tests,it was found that the sample size of the collected bottle was significantly different.It was found that the reason was that the gas-liquid two-phase flow appeared at the front of the acquisition pump pipe in the sampling device,which affected the measurement data of the flow measurement sensor.In order to find out the change rule,the original measurement data should be collected,and then the measurement error was mathematically modeled.A set of gas-liquid two-phase flow experimental scheme and experimental process were designed.In the actual situation,there are many factors affecting the measurement errors,which are complex and non-linear,and it is difficult to model from the theoretical analysis.Therefore,the method of experimental modeling is adopted to establish the relationship between input and output based on the collected experimental data.As one of the most commonly used classification algorithms,BP neural network is easy to transplant to the control system because of its low complexity.The final experiment shows that,compared with the case without algorithm,the maximum absolute error of the flow pattern of the circumfluence gas-liquid two-phase flow is reduced from 48.5% to 18.23%,and the sampling accuracy of the sampling device is greatly improved.The function of runoff collection system in runoff communities at home and abroad is yet to be improved,and the level of intelligence is relatively low.Most of them still use wired method for data transmission,and their transmission efficiency is greatly affected by severe lightning and rainfall and complex geomorphic vegetation environment.In order to solve the problem of transmission efficiency and transmission distance,this article is based on distance radio(LoRa)technology,designed a kind of intelligent runoff plot runoff collecting system,can realize the field data wireless transmission under complex environment,effectively avoid the because of cable transmission mode brings a series of problems,at the same time also can reduce the system cost and power consumption,which can realize runoff real-time transmission of information.The system adopts STM32 microprocessor of ARM company as the core controller,which is responsible for the intelligent control of the whole sampling system.The system is divided into device layer,transport layer,platform layer and application layer.The equipment layer is composed of 22 runoff collection devices in the runoff plot,which can remotely set sampling rules and dynamically monitor and timely collect runoff.At the same time,each device can store runoff samples after multiple rainfall,which solves the problem that researchers have to go to the field to collect samples every time.Small size and convenient installation greatly reduce the overall time of the project.The transport layer is responsible for the real-time transmission of data from all field runoff sampling devices,gathering data to the central node and transferring it to the platform layer via TCP/IP network.The platform layer adopts the OneNET cloud platform,which is responsible for the management and storage of the data of the whole runoff collection system,which can be used for subsequent research and analysis.The application layer provides users with a visual data monitoring interface.Meanwhile,the sampling rules of the field runoff sampling device can be modified remotely anytime and anywhere,so as to meet the needs of different researchers and provide them with Web and APP applications.Through field experiments,the stability and practicability of the system are verified by the final experimental data,which provides a strong support for soil erosion research.