Application Research Of Improved Ant Colony Clustering Algorithm In Forest Fire Forecasting

In this paper, it introduces an Improved Character-base Ant Colony Algorithm,for the shortcomings of local optimum exist in K-means algorithm.The ant colony algorithm using the principle of positive feedback,also an essentially parallel algorithms, therefore it has a strong ability to find better solutions, and not easy to fall into local optimum, which can effectively compensate for local optimum defects exist in K-means algorithm. ICACA algorithm proposed in this paper is basic Classical Ant Colony Algorithm, and introduce ant species classification and sensory perception, which makes the ant colony to find the optimal solution of the maximum possible and effective to avoid the possibility of local optimum. Using this improved ant colony algorithm to optimize the K-means algorithm, establish the improved ant colony clustering model and simulation experiments show that the algorithm can effectively prevent stagnation and the phenomenon of local optimum in the process of clustering in K-means. It can better reach the the objective of global optimum and optimize the overall performance of clustering..The improved algorithm is applied to the prediction of forest fires, due to the large data dimension,large data values, and complex calculations in forest fires datas. as the K-means algorithm implementation process simple, the algorithm's time complexity is low, it is the preferred method of forest fire forecasting. The improved ant colony clustering algorithm can achieve a good balance between time complexity and clustering accuracy, is more suitable for analysis of forest fire forecasting. Finally through the establishment of mining models of forest fires data,then clustering the forest fire data, to achieve the purpose of data classification, so as to provide scientific, reliable and objective basis to prediction of forest fires.