Detection And Research On Pork Freshness Based On Rbf Neuronic Network

With the social advancement, the healthy diet is becoming a big issue day by day. Meat food is needed more and more in people's diet. The quality affects common people's health and influences the quality of life. The detection of pork products is particularly important. Freshness is a key factor of meat quality. The research on the meat freshness makes much sense in the point of economic and practice.Currently, the methods to detect pork freshness mainly are traditional, including sensory detection, microorganism detection, chemical method etc. There are some limitations existing in all the detection methods above, such as being easily influenced by observers, the time-wasting, complex process, no real-time detecting. That makes the way just limited to some special departments, not available for normal consumers.The application and combination of the sensor technology, digital image technology, intelligent information technology and multi-data fusion technology provides the theoretical support and methods for pork freshness detection research.Pork putridity is an extremely complex pHysical and chemical changing process. It is very difficult to achieve a comprehensive and accurate judgment on the freshness just by a single data. While good detection results can be achieved by multi-data fusion technology through a non-coherent Micro-parameters method. Some special information appear and change during the pork putridity, as the pork will release the gases such NH3 and H2S, pork color as well as PH value on the cell surface and the number of integral cells are changing during the putridity. The data-changing trend with time is monotonous, which can reflect the degree of pork corruption in some extent. In subject, all the data are collected at the same time and the same place, including the NH3 and H2S, the color values, cells number, PH value, so as to ensure data synchronization and effectiveness. Two sensors were set in each side to reduce the error impact. The data were transported to PC through data collector. The image information were acquired from the pictures taken by the CCD as well as the cell numbers; PH values ware taken by the precision test paper made by Beijing Chemical Manufacturing. The TVBN value is the national standards of pork freshness, the only index to reflect the pork freshness, which is the output of the recognition system;To get the practical data, TVBN data as the teacher signal were achieved by the method of semi-micro determination of the nitrogen. The promoted forward neural network RBF was adopted in designing to prove the prediction function with stronger error tolerance and less difference. At last, the system was designed through connection of MATLAB and Visual C ++.In this paper, the characteristic information of pork freshness, such as PH value, picture information and gas siginal, is collected through gas sensor array, CCD image acquisition techniques, image processing technology. And the freshness identification process gets realized through artificial neural networks based on Multi-Data fusion.A rapid detection system of pork freshness with sensors arranged multi-angled in space, along with data collecting card of USB kind linking to PC, could be a model for a portable detection terminal for household using.