The Methane Hydrate With Acidic Gas Forming Condition Prediction Based On The Intelligent Algorithm

Natural gas hydrate is solid complexes which the natural gas and liquid water to form under certain temperature and pressure conditions. The natural gas solid seriously affects the natural gas exploitation, processing and gathering. If we can predict the critical temperature and pressure of the natural gas hydrate formation, by breaking the natural gas hydrate formation conditions, we can better realize the gas mining, processing and gathering.At present, there are four categories of prediction’s method in the research field of the natural gas hydrate’s formation condition.they are the graphical method, the empirical formula method, the phase equilibrium calculation method, the thermodynamic model method. The graphical method and the empirical formula method is simple and convenient calculation, but the error is bigger. The phase equilibrium method is simple, fast, but the reliability and generality is poor. Thermodynamic model method calculate precisely, widely used, but with poor scalability.The wavelet neural network has highly nonlinear mapping and adaptive, self-organizing, self-learning ability, can very good predict nonlinear problem. But wavelet neural network has the problem of slow learning speed and easily trapped in local minima.In this paper, improving wavelet neural network’s learningapproach and joining the momentum item speed up the wavelet neural network learning. Genetic algorithm has the ability of global optimization, this paper uses genetic algorithm to improve the wavelet neural network, and reduce the probability of wavelet neural network into a local minimum value.This article uses the genetic wavelet neural network to predict the hydrate formation conditions.Genetic algorithm has strong global optimization ability, but also widely used in machine learning field, but the genetic algorithm has the defect of prematurity. The cause of premature is generally caused by the uneven distribution of initial population and genetic operators reducing population diversity. In this paper, improving the population initialization method and the replacing strategy, the population initialization individual maximum evenly distributed to the solution space, at the same time, every time the replacing strategy adds some new individual to keep population diversity.Finally, this paper evaluates the experiment of the improved wavelet neural network model, the improved genetic algorithm model and the genetic wavelet neural network model respectively.