Enhancing Navigation Aids Status Recognition Based on Extended Space Forest and Time Domain Features Fusion

Navigation aids can help ships find the correct waterway and indicate dangerous areas in the channel, ensuring safe passage for vessels. To address the problem of sparse feature numbers in collected navigation aids data, this paper proposes a model called TDES-RF, which is based on extended space forest and the fusion of time domain features. Four feature generating operators in the extended space forest are used to expand the original feature set, generating different expansion spaces for each base classifier. In addition, time domain features fusion captures dynamic changes and time correlation of data, enriching feature information for more accurate classification. The results show that the accuracy of navigation aids state recognition using random forest as the base model is only 79.88%. After adopting extended space forest and time domain features fusion separately, the accuracy of state recognition improves by 0.12% and 3.33% compared to the base model respectively. Finally, combining these two methods, our model achieves the highest accuracy among all state recognition models at 84.17%.