Study On Automatic Verification System For Electric Meter Based On Machine Vision

Electric power meter, whose accuracy is of significance, is an important tool for people to detect and monitor system instrument, therefore enterprises need to calibrate the instrument periodically resulting in a lot of work. As the calibration that depends on manual work is inefficient and easy to lead to mistake, people do research on kinds of automation verification systems. Now, however, the existing systems are either single type of calibration, high cost; or lower test efficiency, such as the calibration of high-precision pointer instrument, people need to repeatedly adjust the image capture device, which cost a lot of time. Therefore we need to study a verification system which is not only efficient but also generic.In order to calibrate the electricity meters automatically, this dissertation has studied an automatic verification system for electric meter based on machine vision, which can automatically recognition meter reading such as the analog voltmeter, ammeter and desktop digital display meters. The dissertation has built a machine vision-based power meter calibration hardware platform, which capture an image b y using a CMOS smart camera instead of the human eyes, using a electric platform to drive the intelligent camera moves to the position where can capture images clearly. In order to ensure a stable detection environment, set the lighting source and hood on the hardware platform.This dissertation focuses on the meter readings recognition algorithm in the calibration process, and do research on the meter readings algorithm according to the type of instrument. For precision pointer instrument using a readings recognition method based on a look-up table, solving the problem of the existing system which change the position of the camera iteratively. For the digital display instrument, using the character identification algorithm based on BP neural network. By studying the image processing and pattern recognition technology, the automatically recognition of different font formats numbers of digital instrument are realized.This dissertation has selected the Lab VIEW software to develop an electric power meter verification system software based on machine vision, system software uses a modular approach, including precision platform control module, standard source control module, instrument calibration module, calibration data storage module and other modules. The results of experimental have demonstrated the reliability of the software and the dissertation analyzes the error sources. Experimental results demonstrate that the design of the system can meet the requirements of practical application with some prospects.