Application Of PSO In Cement Based SEM Image Recognition

Cement and its base materials are indispensable materials in the development of society,they are very important in the field of construction.The evaluation of cement performance is determined by its strength,so the cement model is determined by the strength of cement.Many people engaged in computer related work began to try to solve the problem of cement strength by computer.Neural network is a new type of intelligent information processing system developed in the process of dealing with the problem of human brain.Based on the neural network classifier,the classification and recognition of cement based material scanning images are improved.In this paper,the main work of this study is completed:Through the experiment of process standards,the process of cement hydration for 12 hours,1 days,36 hours,2 days,60 hours,3 days a large number of SEM images of cement micro morphology,established the image database,realize the unified management of image.The image of cement micro morphology is mainly processed by image enhancement.At the same time,particle swarm optimization(PSO)algorithm and chaotic particle swarm optimization(PSO)algorithm are used to optimize the image respectively.In order to verify the superiority of the chaotic particle swarm optimization algorithm,the accuracy and efficiency of the chaotic particle swarm optimization(CPSO)algorithm are tested by the test function before the image segmentation.Then,4 kinds of optimization algorithms(CPSO,PSO,PSO+ and PSO+)are used to optimize and contrast the maximum between class variance method.In order to select the best segmentation image,to provide a preprocessing library for the latter research.Based on the feature data extracted from the experiment of cement hydration,the BP neural network,PSO neural network and SMPSO neural network are used to classify and identify the cement image.The experiment proved that chaos initialization and chaos disturbance with PSO in comparison with traditional PSO,chaotic particle swarm algorithm can improve the convergence speed and global search ability of particle swarm algorithm,thealgorithm can efficiently obtain the optimal solution of the problem.