Chemical Fault Diagnosis Modeling Optimization Based on Machine Learning

Abstract

With the increasing complexity of modern chemical production processes, chemical operating conditions have become more and more diversified, which has improved the production efficiency of chemical companies and increased the probability of chemical production failures. With the expansion of the chemical industry’s scale of development, traditional manual diagnosis methods have been difficult to detect complex fault information. Therefore, the introduction of modern technology has become the key to chemical fault diagnosis. This paper studies chemical fault diagnosis based on machine learning, uses SVM classifier to perform chemical fault diagnosis, and then optimizes the parameters of SVM classifier through three methods of GS, GA and PSO machine learning algorithms. It is found that the diagnosis accuracy rate after GS algorithm optimization is 99.1273%, GA algorithm optimized diagnosis accuracy rate was 99.3548%, PSO algorithm optimized diagnosis accuracy rate was 99.0626%, but from the perspective of fault diagnosis running time, GS algorithm optimization running time is shorter, so GS algorithm is used to optimize SVM classification the parameters of the device more effectively improve the efficiency of diagnosis.