Research On Express Safety Status Recognition Method And Implementation Of Application System

In the field of modern logistics,the express industry is gaining increasing attention as an important part of it.However,with the continuous growth of the express business,express safety issues have become increasingly prominent.The mailing of contraband in the express and the occurrence of the abnormal status of express parcels not only disrupt social order but also threaten the safety of people’s lives and property.Therefore,efficient monitoring of express safety has become an urgent issue to be addressed.To address this issue,by X-ray imaging technology,sensor technology,and deep learning algorithms,for three typical contraband,namely controlled knives,police apparatus,and explosives,a method for recognizing contraband in express based on transfer learning and residual networks is proposed(TLRes Net18).This method achieves the recognition of contraband without affecting the efficiency of security inspection,thereby refusing the mailing of contraband.In addition,for three common abnormal statuses of express parcels,including dropping,throwing,and kicking,a method for recognizing the abnormal status of express parcels based on CNN-LSTM with channel attention mechanism is proposed(CLSNet).This method can monitor the status of express parcels in real-time,helping to reduce or even eliminate the occurrence of abnormal status in the express.Furthermore,the implementation of the express abnormal status monitoring system can more efficiently monitor the safety status of the express.The main contributions of this thesis are as follows:(1)The research focuses on the method framework for recognizing the safety status of the express.It elaborates on the basic concept of express safety and proposes two technical routes for recognizing contraband and the abnormal status of express parcels.(2)Constructing target domain datasets of three typical contraband and non-contraband by X-ray imaging technology,with the source domain dataset selected from the SIXRay dataset,which is similar to the target domain.The source domain data set is used to train Res Net18 to generate the pre-training model,and the network parameters of the pre-training model are selectively migrated to the target domain.The constructed dataset is used as the input as input of different algorithm models and records the recognition results.The experimental results show that the TLRes Net18 algorithm model performs better and is not limited by the number of samples.(3)Designing a data acquisition terminal to obtain real-time information such as acceleration,geographic location,and time of express parcels during the delivery process.Simulated experiments are conducted in the laboratory using automated equipment to obtain sufficient sample data for constructing the dataset.Necessary preprocessing of the sample data is carried out,including truncating potential abnormal data exceeding the threshold and extracting seven features,including mean,variance,kurtosis,skewness,dynamic range,short-term energy,and zero-crossing rate,after adding time windows to compress the data.To further verify the performance of the proposed CLSNet algorithm model,the model is applied to the constructed dataset and compared with other algorithms.The results show that the CLSNet algorithm model achieves better results,demonstrating its good generalization ability.(4)Building an express abnormal status monitoring system,which consists of four levels: data perception layer,transmission layer,platform layer,and application layer.42 actual mailing experiments are conducted on this system,and the results are analyzed.The results verify the effectiveness of the CLSNet algorithm model in recognizing the status of express parcels in actual mailing.At the same time,the widespread existence of the throwing express parcels phenomenon is also verified.The above-mentioned method for recognizing the safety status of express showed good generalization ability in the experiment,providing technical support for effective safety status recognition during the complex delivery process.These achievements are of great significance for real-time monitoring of express safety status.