Evaluation Research on the Crane Failure Prediction based on Machine Learning

Abstract

It is difficult to spot failures in port machinery and equipment, and maintaining such systems is even more complex. Maintenance such modifications in a reasonable time is a tough challenge since each change might have an endless number of test cases run. It's critical to have a risk assessment of the impact of such maintenance fixes. In the software engineering community, there has been a considerable amount of study on failure prediction. Regrettably, there is little evidence of their application in day-to-day software development in port machinery and equipment. In this paper, we propose an unsupervised machine learning (k-means clustering) method for categorising cranes for maintenance and use a machine learning pipeline to solve the classification of crane failure data. The crane's maintenance decision data demonstrates the method's effectiveness. It was demonstrated that the Linear Support Vector Machine could give a superior classification accuracy of crane maintenance prediction with a 100 percent accuracy in train set and 94.5 percent accuracy in test set.