Research On The Application Of Soft Sensor Based On BP Neural Network In The Detection Of Breathability

With the developing of China’s economy, more and more people paying attention to health issues, and increasing awareness of self-protection, while smoking have had a significant harm on people’s health, in the tobacco industry, the air permeability of the perforated tipping paper can effectively control cigarette tar content, and determines the taste of cigarettes, in fact, the tipping paper air permeability testing equipment the price is very high, and the detection instruments also need 4S, will bring to the life with lots of inconvenience, therefore, in order to solve this problem,proposed a technology which based on the tipping paper air permeability detection soft sensor, using digital image processing techniques to extract image features. The main work of this thesis as follows:(1) To the problem of the tipping paper air permeability testing tipping paper presents a soft air permeability measurement method based on improved BP neural network, this method is to use strong nonlinear fitting capabilities of BP neural network to establish a mathematical model tipping paper punch holes pixels, background pixels and tipping paper air permeability between the experimental results show that the method of fitting a mathematical model to detect air permeability of the training error and measurement error in the network less than 5% range, reaching the level of practical application.(2) To the problem of detect the air permeability of the tipping paper tipping paper presents a soft air permeability measurement method based on intelligent algorithm to optimize BP neural network, the method using artificial fish swarm algorithm and differential evolution algorithm strong global search capabilities, optimizing BP neural network weights and thresholds, therefore, the network convergence speed is very fast, and the final fitting accuracy of mathematical models in the range of 5%, compared with the improved BP neural network algorithm to fit the model more accurate, Therefore, this method is more suitable for actual production applications.(3) To the problem in the tipping paper air permeability test, designed a tipping paper air permeability detection software, a detailed description of the process of image processing software, including image acquisition, image graying, median filtering, binary image processing. The design of the air permeability detection software are applying in Yunnan Province, a tipping paper mills practical application, and achieved certain results, the software is further perfected.Because of in tobacco industry, the air permeability of perforated tipping paper can effectively control the tar content of cigarettes, and directly determines the taste of the water permeability of the cigarette tipping paper, there is no measurement technique using domestic soft air permeability of the tipping paper for testing The method, therefore, we use soft measurement based on BP neural network technology for tipping paper permeability detection research has important practical value and economic significance, with some innovation.