Based On ACO To Optimize The SVM Research In Logging Lithological Identification

With the depth and harder detection of the oil and gas exploration, using the method of reservoir prediction can't distinguish the position of oil and gas well in complicated regions.Whereas it can be solved well utilizing the lithology. The lithology identification is a typical process of high-dimensional and nonlinear pattern recognition. Support Vector Machine(SVM) is a new pattern recognition method developed on the basis of statistiacal learning theory, which represents many unique advantages on solving the problems of small sample.Nonlinearity and high dimensional pattern recognition. It's also provided available methods for lithologic analysis using the well log data.In this paper, the theory of ant colony algorithm and the theory of SVM are briefly analyzed; secondly, the selection of SVM parameters directly affects the classification performance of SVM; therefore, the introduction of the several support vector machine parameter searching optimal method, such as grid search method and bilinear search method,exhaustive method, genetic algorithm, particle swarm optimization(PSO), exhaustion method,genetic algorithm, particle swarm optimization. The new ant colony algorithm, which is the effective combination of the cross validation and ant colony algorithm, is introduced and the new ant colony optimization algorithm is compared with the ant colony optimization support vector machine, that the former not only shortens the time of optimizing SVM, but also improves the accuracy of classification, finally, for the problem of large scale data classification, the traditional SVM is a lot of problems. Therefore, the traditional support vector machine is improved recently. The support vector machine and traditional SVM are improved after training of lithology log data, the experimental results are compared with the results of the results, for large scale data classification, nearest neighbor support vector machine has some advantages.