Research On A Rapid Detection System For Microorganism In Food

Food is the material foundation which human beings rely on for existence in activities of life. It is also vital energy for human beings. The relationship between food hygiene and people's health is extremely close. Both of the Total Bacterial count and Coliform Bacterial number are the most principal standards in the food detection. At present, it takes a lot of time to use the traditional counting method—Culture Counting Method and it will also consume the high cost, that will not catch up with the rapid pace of life. Especially, the complicated step is needed in the process of the traditional Coliform Bacterial detection, which needs cultivate counting time after time with the expensive culture medium. It wasted a lot of manpower and resource and it will not catch up with the increasingly fast life rhythm. At present, other various methods of detection at home and abroad were born, but they all have respective disadvantages that prevented their expanding and they have not completely replaced the traditional Culture Counting Method. Therefore, our laboratory researched on the food microbial rapid detection system by using the digital micro-image identification technology which is based on Machine Vision and Artificial Neural Network. Before this dissertation, my laboratory has developed a rapid detection system which can identify and count Total Bacteria. However, because of the pre-manufacture handling systemic and the pre-manufacture methodological defects, the detection system has some counting error. Therefore, this dissertation firstly researched on the pre-manufacture handling system and the pre-manufacture method of the detection system. Subsequently, this dissertation researched on the other rapid detection method which can count the other important standard—Coliform Bacterial number. This dissertation mainly carried out following work, and has got related conclusions as follows:1. Basing on the Total Bacterial count rapid detection system, my research mainly improved the preparation system and looked for the appropriate preparation method to make the detection results more accurate. At the same time, we made every steps'operation specification quantified in order to popularize easily.This dissertation researched on the once centrifugal settling separation method and the twice centrifugal settling separation method, and the conclusion was showed that the twice centrifugal settling separation's result was better than the other one, and than this dissertation ascertained the most appropriative centrifugalization parameters: use 8000r/min to centrifuge 5min firstly, than use 11000r/min to centrifuge 5 min again. The results showed that the twice centrifugal settling separation method can make our fungi succus, which has different concentrations among our demanded detection range, condense to five microlitre and deposit to the bottom of the centrifuge tube, what is a surprise that the state is steady.In the research of the attractive function and the attractive function between the strepta-vidin and the bacteria, we found that using a microscopic slide which had 100ug/ml strepta-vidin covered on its face to centrifugate in the centrifuge tube can deeply depressed the centrifuge condition. Moreover, the state was steady. But the strepta-vidin's price is too high to apply extensively, that will confine its extensive application.Basing on the detection system's experimental requirements, this dissertation designed the microscopic slides for this detection system. The slide was made of polyethylene and its integral size is length×width=2.5×8cm, the integral thickness was 1.5mm. A repeatly knocked-down roundlet whose diameter is 8 millimeter was cut from the microscopic slide. At the roundlet centre a 1×4.5 square millimeter foveolae gutter was processed, and the processing deep is 0.2 millimeter. This foveolae gutter is the detection area. In this way, the demanded collective detection pictures added up to one hundred.2. This dissertation also researched on the rapid detection methods of Coliform Bacteria. Firstly, six kinds of food specimen (sausages, vegetable & fruit juice, non-fermented bean products, puffed food, candy, baking food) were selected to do substantive detection experiments in order to discuss the Coliform Bacterial rapid detection method's feasibility. Total Bacterial count and Bacilus'number were detected by the rapid detection system and the Coliform Bacterial corresponding MPN value was detected by the traditional lactose bile salt fermented method. Than basing on the results the dissertation respectively founded the Binary Linear Regression Model and the Error Back Propagation Neural Network Model to forecast the Coliform Bacterial number through the Total Bacterial and Bacilus'value which can be detected easily by this food microorganism rapid detection system. Compared with the Bacilus'Error Back Propagation Neural Network Model's forecasting accuracy and the predicting errors, when the foodstuff is judged whether regular or not, the Linear Regession Model's accuracy arrived at about 100%, which is much higher than the Error Back Propagation Neural Network Model's. So we can use it to judge the Coliform Bacterial number in the foodstuff whether regular or not. However, the Error Back Propagation Neural Network's predicting accuracy is much higher than the Linear Regession Model's when they are used to predict the off-grade foodstuffs'Coliform Bacterial number. Its accuracy can reach 90%, and the average predicting error is 0.01314, therefore the trained Error Back Propagation Neural Network can be used to predicte the off-grade products'Coliform Bacterial number of the off-grade products.After that, the dissertation utilized Visual C++ to fabricate the detection operating system, which could realize many functions: select food category, input Total Bacterial and the Bacilus'number, judge the foodstuff whether regular or not, compute and output Coliform Bacterial value, et al.There is no similar research on rapid detection of Coliform Bacteria in food like this dissertation at present. The dissertation is an initial experiment. Though there are many defects in some aspects, the results are satisfing. Total Bacterial and Coliform Bacterial detection results are the most important standards of estimating food whether can leave factories or not. Considering the system's particular merit——swiftness, it will have a good market if this system is used as a food precheck method. We should make further improvement in the designed software humanization, the operational process simplification, and try to apply the system in the market.