Studying On The Optimum Selecting Of Additives And Formula Of Gasoline Engine Oil On The Bases Of Intelligence Technical

Selecting the most appropriate base oil and additives according to the demand of the manufactured oil is the first problem to be solved before studying the lubricating oil.But there are no uniform and standard identification in the classify of the lubricating oil, and in every classification of lubricating oil, the different level has different asks on the base oil and additives.At present , there are a large kind of base oil and additives , and the most important task to choice the additives is selecting the variety and quality accurately , and choosing the amount of additives scientifically .At the same time , the effect and match actions of every kind of additives should be considered.In the general researching of lubricating oil , besides of analyzing physical chemistry characteristic of base oil and additives ,it is very important to use experiential method and do a lot of experiment . But these search are very blindness and exhaust a lot of time and workload.So the key question is how to summarize the experience or half experience disciplinarians which can direct or offer clue to the experiment , so that the workload and the blindness to the research work can be reduced .While dealing with the experiment result and finding out the math relationship between the parameters related to the target value .these math relationship generally are not linearity, whit high yawp ,and has many factorial. Moreover the problems of asymmetry of the data sample are exist. So ,if the supposes are based on linearity by the means of math and experiment ,it is hardly to ensure that the developed lubricate oil can believable.This dissertation aim at the characteristics of developing the lubricate oil, combine the computer analysis technical with the traditional lubricate developing skill. Also it first use these intelligence technical such as the Artificial intelligence(AI), Expert system, Fuzzy math, Pattern recognition, Artificial neural network(ANN), Genetic algorithm into the developing of the lubricate oil in the selection of additives, forecast of performance, formula technical . It established a suit of intelligence analysis method to the developing of lubricate oil, and this innovates the traditional lubricate oil developing skill, saving the cost, increase the efficiency and reliability.In the establishing of knowledge base of the Additives selecting expert system, according to the characteristic of additives, on the base of rule-based knowledge expressing system, it use object-oriented method to express the knowledge. The programs are simple , easy to maintenance, and the data bases are convenient to be called. While establishing the Additives selecting expert system, it combine the data base with the knowledge base ,storing all kind of the property of base oil and additives , choosing the appropriate base oil and additives .In the designing of reasoning machine of the Additives selecting expert system, combining the Fuzzy based reasoning with the Certainty Factor based reasoning , positive reasoning with the reverse reasoning . In selecting the kind of additives, it use Certainty Factor based Fuzzy reasoning . In the selecting of a certain additives , it use Fuzzy based reasoning. In the resolution of the conflict of the conclusion, it use Uncertainty reasoning with limit, Uncertainty reasoning with adding power, Fuzzy reasoning with Certainty Factor etc. This make the conclusion more integrity and trustiness.This dissertation using ANN technology , simulates and forecasts the performance of many kind of additives, formula between each kind of additives. As the problem of local minims in ANN, it using the GA combined with ANN, optimizes the structure and weight of the ANN. This ensures the conclusion more validity and reliability , the net more abroad , the seeking more global.This dissertation first using BP NN, RBF NN, forecasts the formula of each kind of complex additives and complete formula of lubricate oil. While using NN to analyze the lubricate additives and formulas, it adopts Momentum ,Variable Learning Rate , Conjugate Gradient, Newton, Levenberg Marquardt(LM) Algorithm to increases the study precision . The system can calculate the adding amount of each kind of additives according to the knowledge base and data base, so as to come out the optimum formula.This dissertation discards dependence on the linearity of traditional data analysis . It designs the model for the lubricate developing by the method of AI, Expert system and ANN. This can save a lot of manpower, material resources and financial. The aim is to offer a new , simple, fast method for the developing of lubricate oil, so as to make a new way for the lubricate oil developing and manufactory . The practice has proved that it is viable to pick up disciplinarian from a lot of data andexperience by the method of AI and ANN, to find out the half experience and experience disciplinarian of lubricate oil by computer instead of human .This new research will has a great future in the time of that large amount of data base and science information can be shared be the internet.