| Healthy forest ecosystem became a significant target in managing of forest resource,and forest health evaluation is the key act to achieve this target. Currently, water conservation forest health evaluation research in China is in a stage of development, although the certain achievements have been obtained, but it is lack of a recognized evaluation index system and evaluation model of water conservation. In this paper, we rearch about the water conservation forest in Mengjiagang forest farm of Heilongjiang province, and explore the theory, model and method of water conservation forest health evaluation.This paper mainly focus on four aspects:evaluation index system of water conservation forest health, the water conservation forest health evaluation model, the initial evaluation of water conservation forest health, and the sustained evaluation of water conservation forest health.(1) The evaluation index system of water conservation forest healthIn the water conservation forest health evaluation domain, the quantity of evaluation index is too much and too general, some of that have bad problem of operation and it is difficult to get or evaluate the sample to measure and 2 kinds of inventory data, also lack of unified standards. So, we need to understand the concept of forest ecosystem health connotation in accurately and deeply, identify scientifically and reasonably the key factors that affect forest health, take the VOR model as a theoretical basis, base on analyzing the characteristics of vigor, organization structure, and restoring force for evaluation objects, follow science, maneuverability, guidance, independence, stability and cost control principles of establishing this six items, building science, easy to operate and adaptive hierarchical evaluation index system of forest health.(2) The water conservation forest health evaluation model based on RS-MNNThis paper make the rough set RS combine with module neural network MNN by analysising the advantages and disadvantages of the current typical method of forest health evaluation model, propose water conservation forest health evaluation model based on the RS-MNN. In the model, the related principle and method of rough set is used to achieve forest health assessment results much objectively on the basis of the value of evaluation indexes. Module neural network has strong nonlinear mapping, learning ability and high efficiency when handle multiple input sample data, so the health evaluation system which is able to support the continuous evaluation health of research object is established.(3) The health evaluation of water conservation forest in Mengjiagang forest farmThe construction of the water conservation forest health decision table belons the initial evaluation stage, the main task of this stage is to obtain the evaluation index of the data value, calculate the weight of evaluation index, and get the health status of research object by using the weighted average method. At present, the common index weight calculation method based on human judgment, partially based on objectively calculate, this is hard to ensure objectivity of evaluation results and reliability. This paper adopted the attribute importance of rough set method, starting from the index data itself, rather than relying on evaluation of experts in the field, thus we can get more objective evaluation index weight. At the same time,before calculate evaluation index, we can use attributes reduction unnecessary attributes of the evaluation index system base on the attribute reduction of rough set, to obtained a more scientific, reasonable, objective and accurate evaluation index system. At last, based on the reduction of collection evaluation index system and index weight, get the health status of research object, finally combine the evaluation index system and the health index, then construct the forest health decision table.(4) The building of sustained health evaluation system of water conservation forestThis paper select directly part of factors to replace its corresponding evaluation index, in order to reduce artificial disturbance by man-made analysis and calculation parameter values, so the BP neural network can’t meet the demand of practical application. In this paper, the multiple BP neural networks as classifier to build neural network module and set up sustained health evaluation system. In order to eliminate the disadvantage of trapped in local minimum when the BP neural network training, we introduce the artificial colony ABC algorithm, combine with traditional BP learning algorithm, using the improved module based on ABC-BP neural network to build the sustained health evaluation system of water conservation forest, and make continuous health evaluation to Mengjiagang forest farm of Heilongjiang province, verify the effectiveness of the sustained health evaluation system in a forest health evaluation practice.In this paper, firstly, the water conservation forest of Mengjiagang forest farm is taken as the research object. The evaluation index system of water conservation forest health based on VOR is established, and then the water conservation forest health assessment model based on RS-MNN is studied. Aiming at its sustained evaluation part, a new building method based on ABC-MNN is proposed, and then the trained sustained evaluation system is applied to evaluate the health of water conservation forest, at last, the accurate and reliable evaluation result is obtained, and the effectiveness of the sustained evaluation system is verified at the same time. |
