Research On Out-of-order Data Processing Approaches For Logistics

With the popularity of modern information technology applications, the entire logistics system is realizing electronic information gradually, logistics data processing is also being gradually community awareness. The main purpose of data processing is to extract the useful information from the original out-of-order data stream to users to complete a series of prediction.In this thesis, we analyze the out-of-order data streams, then set up a three-level data processing framework and a matching model, thus dynamically processing data and output the prediction results. The main contribution of this thesis is as follow:(1)Due to the difficulty of getting the association rules over out-of-order streams for logistic data, a new improved BP algorithm based on dynamic adjustment is proposed. We firstly use a dynamic adaptive structural adjustment mechanism to change the network training structure according to the environmental requirements, which can automatically remove invalid training node, and optimize the iterative training process. Secondly, we adjust three factors during the learning process to speed up the learning response, and to enhance the stability of the network.(2) Due to the explosive increment of data in logistic data, it is a challenging task to analyze and extract meaningful data for users. Data needs to be timely operated because of the time sensitivity, so it faces enormous pressure in storage and computing. To deal with the problem that it is hard to achieve valuable information from out-of-order streams over logistic data in short time, a model-matching algorithm based on improved BP(Back Propagation) is proposed. In the algorithm, the matching model is set dynamically. Information is extracted for users according to the order of arriving time. Furthermore, the neuron selecting mechanism is optimized and the algorithm parameters are automatically adjusted in the process of learning and matching. Accordingly, the response speed of learning is accelerated and the time of matching is reduced.(3) Focusing on the difficulty of achieving associated information from sensors located in logistic supply chain. To this end, a TLBP(Three-level framework based on Improved Back Propagation) is proposed. The real-time data streams are parallelly processed according to the arriving time in input layer. Then, we proposed an improved mechanism of multi-dimension learning factor to lower the learning error and higher convergence rate.