The Research And Implementation Of Foreign Invasion Monitoring In Hydraulic Pump Station

With the promotion of national agriculture irrigation system informatization, automated monitoring and management to rivers, reservoirs, canals and water has put forward higher requirements, the gate unattended and intelligent monitoring is one of the important parts.The topic of this paper is about safe operation of large pumps in gate monitoring system, to prevent large foreign matter into the danger area, a water intrusion detection system has been designed and developed based on sonar technology, in order to reduce the cost to the level of civil, a large number of technical improvements have been proposed, and ultimately it gives the principle of system and test results. The main contents and works as follows: 1. Study the basic theory of ultrasonic rangefinder and water dropping fish finder, design a sonar detection system based on the principle, including the control circuit, power supply circuit, a transmitting circuit, a receiving circuit and reception switching circuit, and make underwater acoustic transducer work in transceiver way. 2. Test the method that using the pulse reflection time to determine the distance of intrusion object, we found that reflected pulse contained strong signal interference, and circuit identification is much difficult, so our mind changed from the method of pulse reflection time determined distance to the steering waters invaded echo detection. That is to collect the continuous ultrasonic echo signal, and then extract the echo characteristics to compare. 3. Use the envelope signal as an initial characteristic of the echo, extract the amplitude and wavelet transform coefficients of the envelope signal as the final feature of the echo signal by a large number of observation and analysis, and use categories separability criterion to verify the validity of the feature amount classification. Use common K- nearest neighbor method and Probabilistic neural network training method as identification methods. After the extraction of echo envelope curves, using wavelet transform to extract characteristic points of the curve, and then take the feature points to the Probabilistic neural network training tests.Finally, in the actual environment test, show that the scheme can more effectively, distance to determine the error is within 3%, as the main target, intrusion detection can accurately distinguish whether there is a large objects invasion of the waters around the pump, recognition rate reached 99%, to solve the problem of safe operation of large pumping station, has a certain theoretical significance and practical value.