Research On Workers' Activity Recognition And Management Based On Intelligent Mobile Terminal For Data Acquisition

Analyzing and tracking workers' activity in a timely and effective manner is significant in safety,quality and efficiency management of construction site.The traditional monitoring approach of direct observation wastes human resources and is vulnerable to the subjectivity of researchers.Automated data acquisition has a clear advantage in tracking and monitoring labor.In order to make up the flaw of insufficient data dimension from single sensing technology and the limited system flexibility from integrated sensing technology and instability of image sensing technology affected by illumination,smartphones were utilized as data acquisition tools in this research.We established a model for construction activity recognition based on smartphone.The smartphone which can be used as a single-node integrated platform of multiple sensor-based technologies,can satisfy the dimensional demands of data acquisition and adapt to a complex construction environment.In this paper,the construction activities of slab reinforcing bar are taken as the research object.Firstly,eight types of activities are determined according to the construction process and craft of reinforcing bar.They are standing,walking,squatting,cleaning up the template,fetching and placing a rebar,locating the rebar,banding the rebar,and placing concrete pads.Then the characteristics of human body parts in the process of activities are analyzed to determine the data acquisition location.Before collecting data,smartphones were fixed on subject's right wrist and upper right leg with armbands to collect acceleration and angle data when imitating construction workers' activities.The Orion-CC app was used to extract the acquired data stored in smartphones.Since two smartphones are used to acquire data,there is an acquisition time difference between two devices,which will affect the accuracy of activity identification.In such cases,the paper utilizes a preprocessing technique to control the time difference and then extracts the domain features including mean,standard variation,IQR,skewness and covariance.The CART algorithm of a decision tree was adopted to build a classification training model.The effectiveness of the model was evaluated and verified through cross-validation.The final results showed that the average classification accuracy of individual samples reached 95.45% and the prediction accuracy reached 92.98%.For overall samples,the accuracy of classification was up to 89.85% and the accuracy of prediction was 94.91%.The recognition results of the model indicated that the integration of a decision tree algorithm and smartphones could be used to classify the complex activities of construction workers.Based on that,this paper takes the construction efficiency management as an example to analyze the application of this model in the field of construction management.An evaluation method of construction efficiency is established,which takes "effective work efficiency" as evaluation index,"time-ratio model" and "the figure of construction process" as auxiliary management tools.The rationality and effectiveness of this method are proved in theory.