Research On Fault Diagnosis Of Ballast Water System Based On Optimized Support Vector Machine

The intellectualization and automation of ship has become the development trend of ship industry,and the research of intelligent fault diagnosis method of ship related system has become an important hotspot.In recent years,support vector machine technology has made important achievements in various industries,and it is very valuable to introduce it into the field of intelligent ship fault diagnosis.The main functions of ballast water system in ships are:adjusting the draft of the ship,balancing the ship's longitudinal and transverse direction,and maintaining the proper height of the stability center;reducing the deformation of the ship and reducing the vibration of the ship's hull.In this paper,the ballast water system of a container ship simulator developed by Dalian Maritime University is taken as the research object.According to the characteristics of common faults of ballast water system,a fault diagnosis method of ballast water system based on optimized support vector machine is proposed.In this paper,the structure,working process and common fault characteristics of ballast water system are studied and analyzed.Aiming at four common system faults,fault diagnosis monitoring points are set to collect system operation data.Then,the fault diagnosis method based on support vector machine theory is studied deeply.In order to improve the classification ability of support vector machine,aiming at the problem that it is difficult to select the penalty factor and kernel function parameters,an improved artificial bee colony algorithm is proposed to optimize the relevant parameters of support vector machine.In order to solve the problems of slow convergence speed and easy to fall into local optimal solution,the global search factor is introduced into the search formula of the original algorithm.And the optimized SVM is verified by using UCI data set,which proves its good classification performance.In the process of working condition monitoring of ballast water system,there are many state variables that need to be measured and there is a certain correlation between them.If the collected sample data is directly input into SVM for fault diagnosis,it will lead to a long time of fault diagnosis and low efficiency of diagnosis.Aiming at this problem,a method of sample data reduction and fault detection using PCA is proposed.Finally,the method is verified by MATLAB simulation with the collected data of ballast water system.Firstly,PCA is used to detect the fault of ballast water system.Then two groups of comparative experiments are carried out.The first experiment is to optimize the fault diagnosis of SVM based on the IABC algorithm.The experimental results show that compared with GS-SVM and ABC-SVM,IABC-SVM can obtain higher classification accuracy.And in parameter optimization,the improved IABC algorithm is faster than the traditional ABC algorithm because of the introduction of global search factor,and can jump out of the local optimal solution.The second experiment is IABC-SVM fault diagnosis based on PCA fault feature extraction.The first six principal components are extracted from the feature parameters of the sample data using PCA as the input of IABC-SVM.Experimental results show that,compared with the diagnosis results of sample data without dimensionality reduction,PCA can remove redundant information in the sample feature information,maximize the amount of compressed information,effectively reduce the feature dimension,and improve the diagnostic efficiency of the SVM classifier.Simulation experiments prove that the fault diagnosis method of the ballast water system proposed in this paper can effectively detect and identify the faults of the ballast water system,with high accuracy and diagnostic efficiency.