Cultivated Land Early Warning System Based On SVM

With the living and developing of human beings, we need the land as the place of living, we need the cultivated land to provide us food, clothing and so on. Cultivated land is one of the most precious parts of cultivated land resources. Due to the scarcity of cultivated land as an irreplaceable element of the natural and economic inputs, the non-mobility of spatial location and exclusive characteristic of our land-use process, In the process of land use, ensuring the sustainable use of cultivated land resources and pursuing the sustainable development of economic, social and ecological benefits of cultivated land resources has become a consensus. Population growth, the development of modern industrial and urbanization, desertification, pollution increase, has already resulted in the decrease of the quantity of cultivated land and the quality of cultivated land.This grim situation has already caused serious concern of Chinese government as well as the whole world to China’s food security.China has the most stringent farmland protection system, which requires the concepts, technology and means in other areas to provide reliable information and services during the process of practice. Early warning is an old and often-updated problem, the emergence of early warning provides a new way of thinking about the practice of cultivated land protection system. Early warning of the cultivated land has become the key of spatial decision supporting process. As the complex characteristic of the source data of cultivated land early warning, inextricable link between the factors and the results, uncertainty of reasoning, how to use modern computer technology and how to use information of cultivated land in GIS environment to land early warning is a valuable research direction. Previous research results of our study opens the door of early warning of cultivated land, many results of land-related research give us inspiration and provide us basis of further research. But the existing early warning methods of cultivated land are based on empirical knowledge, simple mathematical assumptions and extreme formalized inference, which cause the over-dependence on empirical knowledge, highly inaccurate and unreliable of the results. Pattern recognition is a rapidly developed information processing method in recent years, which makes the results of early warning more objective and more realistic by using data-driven mechanism to avoid or minimize the influence of human factors in the process of early warning. In this paper, vector machine model is supported for the introduction of village-scale early warning of the cultivated land, and early-warning model of cultivated land is built in the perspective of the modern early-warning methods to achieve the intelligent of early warning of cultivated land to a certain extent and to minimize the influence of small-sample data on self-learning, self-organizing and self-adaptive properties of the model.At first the article analyzes the present situation of early warning research of cultivated land detailedly, sums up in the perspectives of the theoretical research of early warning system, the research and development of economic early warning, the development of disaster warning systems and the development of early warning systems of cultivated land, and analyzes in depth the advantages and disadvantages of these several stages. Subsequently, the connotations and characteristics of early warning system of the villages and towns are discussed, and the targets and goals of early warning system of cultivated land are clarified. Then, the theoretical basis of early warning systems of cultivated land of villages and towns and methods of system theory are introduced, with man-land relationship, the principle of sustainable land use, statistical theory, knowledge discovery and data mining theory, vector machine supporting principle and the principle of particle swarm as the guiding ideology of early warning of cultivated land of villages and towns.In Chapter three, the basic connotation of an early warning of cultivated land is clarified, and the rules and steps of land warning of cultivated land are general introduced. The most important question of early warning of cultivated land is how to measure, to transfer the qualitative description into a quantitative expression. Accordingly the measure issue of early warning of cultivated land is elicited. And then after elaboration on relations of the police situation, the police sources and the warning signs of warning early warning of cultivated land of this "issue-driven" model, three major indicator systems, which conclude the police situation indicator, the police source indicators and warning signs of cultivated land early warning targets, are introduced and highlighted. Around the entire early warning system, this article uses vector machine supporting algorithm as the main algorithm to build the early warning model of cultivated land villages and towns. The model was the reflect of thoughts of model-recognition in early warning, including a clear police intelligence, data collection and clean-up, a warning signs subsystem to identify, a method subsystem of early warning, a the output subsystem of police degrees, in which the warning signs subsystem to identify includes identification of the warning signs and indicators and rule-mining of early warning.The core of early warning systems of cultivated land of towns and villages is identification of warning signs, rule mining of early warning and forecasting of early warning. In view of the three core contents of this article, particle swarm optimization algorithm and decision tree algorithm are introduced into the vector machine supporting system, namely automatic identification algorithm of the warning signs and self-mining algorithm of early warning rules. We take identification of the warning signs and alarm limits indicators as feature extraction and data discrete, and definite and descript its key issues such as multi-classifier of vector machine supporting, generation of initial swarms and so on.Then, this paper takes Yicheng city as an example early warning empirical research of cultivated land of urban villages and towns, focuses on the choice of samples, identification of indicators, data collection and other preliminary work, and then delineates and proves the automatic recognition of warning signs, rule mining of early warning and police warning signs limits of cleaned-up data. Early warning analysis of its 2005,2010 and 2020 situation is done, the police source the results is also analyzed. Finally, we summarize this study, as well as the problems of the research, verify the effectiveness of early warning models of the village land based on the vector machine supporting, as well as provide a new thought and a more scientific method for land early warning.