Dissertation > Industrial Technology > Oil and gas industry > Oil, natural gas processing industry > Synthetic oil > Extraction of oil from other raw materials

BP network optimization based on genetic algorithm optimization of the biodiesel process

Author ZhouXiaoFei
Tutor SuYouYong
School Kunming University of Science and Technology
Course Agricultural Biological Environmental and Energy Engineering
Keywords biodiesel artificial neural network BP network conversion rate genetic algorithm
Type Master's thesis
Year 2011
Downloads 16
Quotes 0
Download Dissertation

In the process of biodiesel production, there are a variety of factors which have an impact on the biodiesel conversion. Various parameters are used to look for the optimum condition in the process of biodiesel production, and then the orthogonal test is applied to gain the optimal condition. The orthogonal test is a scientific method reaching and dealing with multi-factors test and reasonable arrangement test. It is based on the principles of mathematical statistic, and the optimum condition is selected scientifically by using the orthogonal form in a lot of test sites. Compared with the traditional test methods, the orthogonal experiment method has the high test efficiency, but the process takes a lot of time and energy. Establishing artificial neural network model can simulate orthogonal test of various horizontal spacing and get more satisfactory results by using the neural networks orthogonal experimental data. Only the establishment of network prediction model can accurately predict the best conditions with a BP neural network in the application of biodiesel preparation technology, it can effectively avoid complicated large artificially experiment and data analysis, saving manpower, improving the efficiency. In the preparation of biodiesel process, because different reaction conditions have different influences on the conversion of biodiesel, it is very difficult to adopt an accurate mathematical model which can influence the conversion of biodiesel by using the traditional method. BP network can extract experimental data from mathematical model, it is based on test data without prior formula of form, and it gets a mathematical model which reacts with the internal law of experimental data after "training" by using the mathematical model, then by using the model to reasoning.The conversion of biodiesel depends on the conversion ratio of methanol to oil, reaction temperature, and catalyst concentration and so on. This study was made to prepare biodiesel from Jatropha Curcasl oil by transesterification and using KOH as a catalyst. The optimum conditions it received through single factor experiment and orthogonal test are as follows: catalyst concentration is 0.9%, reaction temperature is 85℃, reaction time is 1 h and ratio of alcohol to oil is 6.5:1. The BP algorithm can be used to establish the technique of artificial neural network of biodiesel conversion forecasting model by using experimental data as samples. It would get some parameters suitable for biodiesel conversion which are related to artificial neural network model such as reaction temperature, reaction time, catalyst dosage, ratio of alcohol to oil and so on. It will accurately predict biodiesel conversion rate. Because BP algorithm may make the test data into the local minimum, a genetic algorithm is adopted to looking for biodiesel conversion the optimal value of artificial neural network model based on the BP algorithm. Then using the optimal process conditions has been verified experimental process condition: catalyst concentration is 1.2%, reaction temperature is 80℃, reaction time is 1.5 h and ratio of alcohol to oil is 6:1. Five parallel tests have been verified on that basis, the results showed that the conversion was 98.82% under this condition, the predictive value was 96.57%, and the error was 2.25%.Results of study show that the simulation results and experimental results had tallied, which explained that the genetic algorithm was used to optimize the BP neural network for biodiesel preparation process optimization was feasible and practical.

Related Dissertations
More Dissertations