The Application Of Distance Discriminant Method In State Recognition Of Seawall Safety Monitoring

Seawall play an important role in defensing against storm attacks, it’s safetydirectly related to the safety of people’s lives and properties. Seawall generallybuilt of local materials and, in the long process of development constantlyheightened and reinforced, so it’s structure is quite complex. Seawall attacked byrainstorm and waves frequently, it’s working condition change rapidly. It’s anchallenge to control the safety of seawall. Due to rapid development of economy,dike safety becoming more and more important, so that people give more attentionto seawall safety monitoring. Based on monitoring data, seawall state evaluationresult provides an important reference for the seawall management staff toguarantee seawall safety operation, and to analyze and control running state.According to the actual situation of seawall monitoring, in this thesis weanalyze index system of seawall state recognition, and construct seawall workingstate recognition quota set by tide, rainfall and osmotic pressure. We evaluateseawall working state by two kinds of discriminant methods. One is using tidemarkand osmotic pressure to build Mahalanobis distance state recognition; the other isusing weighted Mahalanobis distance to achieve state recognition. In realizeweighted Mahalanobis distance discriminant process, we set up three evaluationquota sets by tide, rainfall and osmotic pressure base on characteristic analyzing ofeach quota.The status of seawall is divided into four grades according to safety gradedefinition related to levee engineering. Ddifferent indices representing differentmeanings， so we use different ways to divide each quota into four levels. Tidallevel classification mainly based on tidal characteristic levels. Rainfall gradesdivision based on the24hours rainfall. Seepage pressure grades division is notspecified, we combined with the confidence level to devide into fours levels.State variables reflects the status changes of seawall, it’s monitoring data contains some corresponding trend. In order to improve the accuracy ofdiscriminant, the osmotic pressure trend removed. All the quotas are standardizedand calculated in separate time, which means designated data form the same timeevery day into one group, and calculated data of different time respectively.Weight analysis is an important part of the weighted Mahalanobis distancediscriminant process. We introduce the common weights analysis methods, and useentropy method and AHP method to analyze quotas weights under three evaluationquota sets based on the actual situation of the seawall.According to the seawall seepage characteristics of cyclical, we establishharmonic quantity regression model to predict the next24hours osmotic pressuredata. Tidal level use forecast value by tide station, and rainfall according toforecast value by the meteorological department. And eventually realize predictionon the future state of seawall discrimination by using the weight matrix combinesubjective with objective.In order to introduce the specific implementation of seawall state evaluationand verify its rationality, this paper takes Pudong seawall safety monitoring data oftidal level, rainfall and seepage pressure as an example to explain. Also weconstruct a short-term forecast of future status in24hours. The results show thatusing distance discriminant method to realize seawall working state evaluation andprediction can correctly reflect the seawall actual condition, has a goodpracticability and veracity.