Research On Outlier Detection Algorithm For Express Logistics

With the rapid development of electronic commerce, China's express delivery logistics industry has entered the era of rapid development. The traditional single, small-scale logistic model has been unable to meet the development needs of modern logistics. In this context, China's modern logistics model is expressing to the joint distribution trend. However, the joint distribution of the large-scale collection contained distribution method will inevitably lead to the occurrence of abnormal distribution problems. Therefore, it is one of the key problems in modern express logistics model's development that how to effectively detect the abnormal distribution problem.Aiming at the outlier detection methods that at home and abroad cannot be directly applied to the express logistics data, we proposed two outlier detection methods that can be respectively applied in the distribution of the general cargo and the special items in this paper. The main work and contribution of this dissertation are summarized as follows:1. Aiming at the characteristics of general cargo's express logistics that until the cargo is sent to a distribution center, its distribution status will be updated, and an outlier detection method for express logistics of general cargo is proposed by considering the amount of goods in distribution centers, the maximum route transport capacity, the storage capacity of the distribution center, the constraints of distribution time and other distribution factors.2. The trajectory of special items' distribution with time, coordinates, speed, directions and other property characteristics compared to the traditional trajectory, therefor, the existing trajectory outlier detection algorithm cannot be directly applied to the express logistics. And the express logistics abnormalities also include distribution time delay and destination error and other aspects. Therefore, an outlier detection method that is suitable for the express logistics of special items, is proposed in this paper by combining the outlier detection method base on distance with the method based on angle.3. In order to test the efficiency of the proposed algorithms, we have built an express logistics system simulation model by taking orders, transportation, warehousing, and distribution four main functional elements of the logistics system. The model is used to simulate the operation of the whole logistics activities, and provide the test platform and simulation data.The experimental results on random data and model data shows that the proposed algorithms can effectively detect anomalies appear in the express logistics distribution process, and they are more efficient and accurate than the classic TRAOD algorithm. Therefore, the effectiveness of the proposed algorithms has been proven, and they are very suitable for the outlier detection of express logistics.