The Stand Growth Model Construction Research Based On Timing Analysis And Clustering

China's forest resources management, investigate and monitor in particular, starts early. Basic forest resources monitoring data is rich. A great deal of management information is accumulated. Forest resources information management system setted up is followed by. Substantial information to be stored and provided query. The proliferation of data hides a lot of important information. Simple query and statistics have been unable to meet the needs of forestry. New approach of mining knowledge hidden behind data is required. Data Mining extracts potentially useful information and knowledge from substantial, incomplete,noise,fuzzy,random data which people do not know. Data Mining is an emerging interdisciplinary. It brings together the results of various disciplines which include machine learning, pattern recognition, database, data warehousing, statistic, artificial intelligence and information management system.Cluster Analysis is an important area of Data Mining. It is an important mean and method of data partition. Early versions of the clustering algorithm calculate the distance between the points of the model. It require to scan data many times in order to identify the packet.But Scalable Expectation Maximization Algorithm(Referred to SEM), according to the density of groups to create a clustering,most simply scan source data once. And at any point in the calculation process it can provide reasonable results.The biggest advantage is that it only needs a limited amount of memory.Usually,when the algorithm deal the records,It will create a number of clustering and a??t these cluster centers with the increase of data in order to find a group characteristics that best describes the similar cases. The whole process runs faster and will not consume all available memory so as to avoid machines paralysis.Time series is also an important area of data mining.Time series refers to a series of number got by observing the same phenomenon in different period of time.The prediictly way of time series is achieved by exploring the laws that phenomenal change with time,in the historical statistics of time series.Time series extend the law to the future so as to predict the future of a phenomenon.The taditional analytical method of time series analysis applied in economy is mainly the analytical method of time series in a fixed time,such as Exponential Smoothing method,Moving Average method,Decomposition of the time series and so on.Box and Jenkins proposed an analytical method of time series based on random theory which not only takes the theory of time series analysis to a new level but also promotes the preciseness of prediction.the basic analytical models of time series are ARMA Model and ARIMA Model.Stand growth model is a kind of indirect estimation methods.Based on the different species at different growth conditions of different developmental stages of growth investigate,It uses the mathematical methods such as graphs, charts, formulas or computer programs and other forms of indirect estimates to forecast stand growth and harvest.Stand growth model used to predict forest growth under natural conditions and also used to predict the Operating measures impact on tree growth.It laies a foundation for the optimize development of operating programs and oriented cultivation.Its application provides the necessary conditions and the basic means for new developments in the basic theory of forest management and modern forest management technologies. In this paper Yong'an City Masson Pine survey datas was used to set up a many-shaped tree height growth model, many-shaped diameter growth model and many-shaped volume growth model.Using many law-shaped curve to construct oriented curve, the classification of samples is very important. Because of Stands locate at different geographical and stand growth has great changes with the age trend,this study uses SEM clustering to classify the factor of geographical and Time Series Algorithm to build a tree-oriented high-growth curve. Many curve-shaped curve makes stand growth model toward a more sophisticated and more accurate.With traditional methods,the superficial relations between the dependent and other factors are always emphasized,so the problems such as"mutual prediction"in the factors and"iterative estimate"during the equation groups are existed.All these result in an imprecise estimate value.A good growth model shouldn't include excessive variables.It should include only one dependent(estimate value) and one independent(age) to avoid the"mutual prediction"problem.Using only one model to describe the same biological growth under different habitats,Many law-shaped curve can solve the problem of a variety of growth model.It only uses one-variable(age) and each stand has a growth curve.It reduces the system error effectively.As the rapid development of web information teehnology,it shows many great advantages to build an information management system for use of lnternet range.And now,there are more and more forest units such as companies and administration departments trying or preparing to build their internet forest resources information management systems.Forest resources information management system is the basement of our country's forest infomationization.The research of how to get a good software system architecture with lower development cost,good extent and easy deployment abilities is very important.Innovative point of this article mainly manifested in:Data mining methods be applied in forestry, thus a new way of thinking about the decision-making for forestry operators was provided;Contain only one dependent variable and one independent variable, thus avoiding the problem of inter-prediction;With the use of SEM algorithm, the implementation runs faster and better;using time series analysis,prediction results better. A good software system based on struts2,spring,hibernate architecture with lower development cost , good extent and easy deployment abilities was builded.