Fuel Oil Plants Screening Model Based On BP Neural Network

With the reform and opening developing in China, the economic level will increase with the global economy integration, energy demand will continue to increase, so expanding biodiesel industry is the main strategy to slove the energy shortage.Biodiesel is a kind of biofule, which is a non-toxic, cleaner burning, renewable fuels from vegetable oils or animal fats, and other agricultural products from China. Because of the need of plant materials and energy plants, it is very important for agriculture and forestry development. The rich energy plants resource in China is the development base for biodieselArtificial neural networks are current methods of computer software to create a dedicated neural network chips, include other methods such as VLSI Implementation (digital simulation, and digital-analog hybrid optoelectronic Internet), dedicated neural network parallel computing system, Neural Network Computer optical realization and biological achievement. BP network is the hot topic of neural network research since the mid-1980s It has an unique non-linear mapping capability in various engineering fields for a wide variety of applications.In the present study, we established a complete database of fuel oil plants iaccording to the plant characteristics, analysis of the fuel, the chemical and physical properties and ete.. Moreover, taking the international biodieseI standard as input indicators, the screening model of fuel oil plants was optimized and set up using the BP-based neural network, the model can evaluate the performance of fuel oil palnts as the biodiesel materials.The raw data preproeessing, and through different network structure, Algorithm under different conditions was established by the neural network model comparison, TRAINGDM for neural network algorithm was identified.In this paper, Biodiesel plants database of raw data pretreatment, and through different network structure, Algorithm under different conditions (Deltalin, Deltalog, Deltatan. Errsurf, Initff, Traingdm) was established by the neural network model to be integrated, identify neural network algorithm for TRA1NGDM. Better predicted results were obtained under such conditions, and provided the main program code with MATLAB neural network toolbox, 350 groups of data were divided into 250 groups of training data, 50 groups of certification, and 50 groups of test. Among them, training groups were used in order to establish network, the certification ones for the promotion capacity of the network, and the test ones for further test of the accuracy and practicality of the neural network. The pretreatment of training groups increased the network training speed and accuracy of the forecast. After that, the training process error curve continued to decline, and the error reached below 10^-4alter 97 times training. The network can guarantee good generalization and better convergence capabilities. In order to achieve better user interactive function, to provide an intuitive and convenient platform for user, the Biodiesel plant screening model window was set up with the application of Matlab and delphi automation technology. The model can solve the problems of how to screen the potential fuel oil plants from the rich resource in China. The model is built for screening fuel oil plants for the fast time, so it has important value of theory reference and application in the field ofbiodiesel.