Logistics unmanned aerial vehicle abnormal behavior intelligent identification method based on isolated forest method

Abstract

The invention discloses a logistics unmanned aerial vehicle abnormal behavior intelligent identification method based on an isolated forest method. The logistics unmanned aerial vehicle abnormal behavior intelligent identification method comprises the following specific steps: 1, carrying out outlier calculation and observation on logistics unmanned aerial vehicle flight data; 2, constructing isolated trees according to the input data, combining the single isolated trees with the data features, and constructing a set of isolated forests; 3, calculating the average path length of each isolated tree and the expectation E (h (x)) of the path length, and finally solving the abnormal score of the sample through the E (h(x)); 4, dividing the abnormal data according to the calculation result of the abnormal score; 5, substituting the longitude, latitude, elevation angle, climbing speed and abnormal score data into the model, and evaluating the accuracy. The method has the advantages that intelligent learning and efficient detection of unmanned aerial vehicle abnormal behaviors are achieved, the pressure of high-speed development unmanned aerial vehicle operation on flight safety and public safety can be effectively relieved, and thereby a technical foundation is laid for development of the unmanned aerial vehicle logistics distribution industry.

Description

technical field [0001] The invention relates to the technical field of unmanned aerial vehicles, in particular to an intelligent identification method for abnormal behavior of logistics unmanned aerial vehicles based on the isolated forest method. Background technique [0002] The average annual growth rate of express parcels in my country has reached 10 billion, surpassing developed economies such as the United States, Japan, and Europe for six consecutive years. At present, the single-day express delivery volume in my country exceeds 100 million pieces, and the growing business volume makes it increasingly difficult for traditional delivery methods to meet the growing service demand. With the disappearance of the demographic dividend, logistics companies are beginning to face the current situation of high labor costs and difficult distribution. Using drones to participate in logistics distribution can not only greatly reduce distribution costs, but also improve efficiency...

AI Technical Summary

Problems solved by technology

The traditional low-altitude supervision industry adopts the human-carried loop loop method to detect the abnormal behavior of drones based on the knowledge base and physical model, which has problems such as low efficiency and poor accuracy, and requires the inspectors to have sufficient professionalism and knowledge systems to be effective. Good recognition effect, so poor universality and anti-interference ability

However, the existing data-driven anomaly detection algorithm research is mainly carried out from the two directions of clustering and complex neural network deep learning, which has problems such as poor robustness, weak generalization ability, and high overhead.

Images