Method for predicting indoor movement trajectory data based on HMM model

Abstract

The invention belongs to the field of indoor movement trajectory management and prediction, and specifically provides a method for predicting indoor movement trajectory data based on an HMM model, which is used for realizing prediction based on HMM trajectory data in an indoor space. The method provided by the invention comprises the following steps: firstly, based on a historical trajectory sequence, taking the minimum Euclidean distance between consecutive non-repetitive points of the historical trajectory sequence as a side length of a mesh unit, performing hierarchical meshing on an indoorspace model to generate a mesh space; based on the mesh space, projecting the historical trajectory data to generate a mesh sequence, and preprocessing the mesh sequence to generate a historical meshtrajectory database; then, performing clustering on the basis of a DBSCAN algorithm to generate a clustering information database, and constructing the HMM model according to an clustering information table; and finally, based on the trained HMM model, using a Viterbi algorithm to perform prediction. The invention provides an indoor trajectory data position prediction method, which effectively improves the accuracy of the predicted trajectory data, and further optimizes the performance of an indoor trajectory data management system.

Description

technical field [0001] The invention belongs to the field of indoor trajectory management and prediction, and in particular relates to indoor trajectory prediction under big data, specifically a method for predicting indoor trajectory data based on an HMM model. Background technique [0002] With the rapid development of wireless interconnection technology, location-based services have been widely used; location-based services are divided into outdoor location services and indoor location services, and positioning technology is one of the core technologies of location services. In large indoor environments, such as shopping malls, hospitals, airports, etc., users have an increasing demand for location-based services. With the increasingly sophisticated indoor positioning technology of Bluetooth and WIFI, and the popularization of smart terminals, a large amount of trajectory data of moving objects has been collected; how to reasonably analyze, efficiently store, and accurate...

AI Technical Summary

Problems solved by technology

The indoor trajectory position prediction method based on the HMM model is to divide the space into disjoint regions and use the region to represent the trajectory points in order to simplify the trajectory data and improve the speed of prediction calculations; however, the classical HMM model has hidden state discontinuity, In addition, there are floor attributes in the indoor space, so that there are the same position coordinates between different floors, resulting in prediction failure

At present, the method to solve the transition probability of the discontinuous hidden state is 0 is to use the historical data multiple times in gradients during the establishment of the HMM, extract the trajectory points at periodic intervals, and calculate the state transition matrix together with the continuous trajectory points to obtain the discontinuous The two state transition probabilities; however, this algorithm requires multiple visits to historical data. When the amount of historical data storage is accumulated, the amount of calculation becomes larger, and the model overfitting occurs

The state stay problem refers to the situation that among the first N trajectory points, there are multiple consecutive points belonging to the same hidden state, and the rotation probability of the state transition probability matrix is ​​0, which leads to the failure of prediction; Obtain the experience value so that the probability of model rotation is not 0; since this method requires artificial settings, the prediction progress on different trajectory data is quite different

Images