Prediction And Classification Based On Swarm Intelligence Algorithm And Machine Learning

With the rapid development of modern science and technology,artificial intelligence has been widely used in all aspects of people's lives.As the core technology of artificial intelligence,machine learning is facing new challenges in various fields.How to improve the efficiency machine learning for data analysis and application in the new environment and new problems has become a hot issue of global scholars.In this paper,several intelligent optimization algorithms such as Seagull Optimization Algorithm(SOA),Particle Swarm Optimization(PSO)and several machine learning methods such as Support Vector Regression(SVR),Support Vector Machine(SVM)classification,Extreme Learning Machine(ELM)are studied,and these methods are successfully applied to air quality prediction,laser ultrasonic defect detection and concrete brick image recognition.The main achievements are as follows:(1)Considering the impact of some prevention and control measures of COVID-19 on air quality,firstly,the air quality and six main pollution concentrations in four relevant periods including the outbreak period of COVID-19 were evaluated and analyzed by using numerical statistics and Grey Relational Analysis(GRA).Secondly,based on the characteristics of SOA algorithm which only considers the global optimal effect and ignores the individual optimal effect,an improved SOA algorithm is proposed.Combined with SVM model,a hybrid prediction model ISOA-SVR is established.Finally,the proposed ISOA-SVM method is used to predict the Air Quality Index(AQI).The experimental results on two different data show that the proposed ISOA-SVR method has better prediction performance,generalization ability and robustness than other models.(2)Aiming at the problems of high dimension of signal and few samples of data in laser ultrasonic defect detection,a signal feature extraction model based on sparse representation is established.TPSO algorithm based on TDIW inertia weight strategy is proposed.At the same time,by introducing nonlinear contraction factor and more random variables,the IPSO algorithm is proposed,which improves the optimization performance of the algorithm and effectively avoids the problem of falling into local minimum.TPSO-SVM and IPSO-SVM classification models are established and applied to the detection and recognition of laser ultrasonic defect signals.Compared with other classification models,the proposed model has better classification detection performance.(3)Aiming at the problem of concrete brick image classification and recognition,firstly,an image feature extraction method based on the fusion of RGB spatial average pixels and HSV spatial average Gray Level Co-Occurrence Matrix(GLCM)features is proposed,and the extracted fusion features are input into SVM,BPNN and ELM models respectively to realize the automatic classification and recognition of concrete brick image.Through the comparison of the experimental results,the SVM method in the automatic classification and recognition of concrete brick image obtains a higher recognition accuracy.Secondly,a hybrid classification model ISOA-SVM based on ISOA algorithm is proposed.The model parameters of SVM are optimized by using ISOA algorithm.The experimental results on two different concrete brick image samples show that the proposed ISOA-SVM method has better classification and recognition accuracy than SOA-SVM,PSO-SVM,IPSO-SVM,SCA-SVM,WOA-SVM,DESVM and SVM.Finally,the ELM ensemble classification model based on Bagging algorithm is established.The experimental results on two different concrete brick sample sets show that the proposed ELM ensemble model effectively improves the recognition accuracy of single ELM model in concrete brick image classification and recognition.The improvements of several swarm intelligence algorithms and machine learning methods studied in this paper greatly improves the performance of the original algorithm,and has been successfully applied in air quality problems,laser ultrasonic defect detection problems and concrete brick image classification and recognition problems,which has certain practical value.