Research On The Method Of Miners’ Unsafe Behavior Identification Based On Ontology

The unsafe behavior of coal miners is the main cause of coal safety accidents.Therefore,from the behavior of miners’ point of view,the management and improvement of coal safety production is one of the fundamental ways to solve mines’ safety problems.With the development of Internet of things technology and artificial intelligence,lots of researches have been consider on humans’ behavior understanding and judgment,and some achievements have been achieved.However,the general behavior understanding method is more suitable for simple behavior understanding in fixed scenes.If it is directly applied to the understanding of unsafe behavior of miners in complex underground environment,there are lots of problems like complex modeling.At the same time,after acquiring behavior and behavior context information,due to the lack of extraction,representation and utilization of domain knowledge,it is difficult to automatically identify unsafe behaviors of miners.Based on this,this paper is put forward a method of behavior understanding and judgment based on ontology and combining data and knowledge,which can not only solve the problem of behavior understanding in complex environment,but also automatically judge the unsafe behaviors of miners,and form a structured knowledge of unsafe behaviors of miners,which is of great significance to the improvement of mine safety.Firstly,aiming at the problem that the knowledge of unsafe behaviors in mines is diverse and heterogeneous,the total amount of scientific data in coal mines is increasing,and it is difficult to manually integrate the knowledge,a knowledge extraction algorithm for the field of unsafe behaviors in coal mines is proposed.The algorithm combines a variety of word vector information,based on the deep attention mechanism to jointly learn a variety of knowledge extraction tasks,and solves the phenomenon of one word polysemy or multi word synonymy in the field of coal mine safety related data because it contains coal mine geographic information and a large number of proper nouns.Experiments show that compared with the best method at present,the F1 value of the model is improved by 1.9 percentage.At the same time,the addition of parallelism and the omission of decoding layer improve the training speed of 33 K.Secondly,based on the underground intelligent space platform,with ontology as the carrier,the connection of miner behavior data and knowledge is realized.Firstly,in the aspect of ontology construction,combined with the above knowledge extraction model,the method and basic framework of semi-automatic ontology construction are proposed.Then,in the domain scope of ontology,the behavior information of miner is represented in four fields: environment,miner state,space-time and behavior.By comparing the advantages and disadvantages of ontology storage tools,choose relational database as the storage tool to management knowledge in ontology.Then,the ontology reasoning method based on CART and SWRL is proposed to realize the understanding and judgment of miner’s behavior at the semantic level.Firstly,the behavior data is through CART classification tree to generate atomic behavior rules automatically.Then,the atomic behavior and other behavior data are fused by the SWRL rule description language to complete the more complex task of miner behavior understanding.Finally,through the definition of unsafe behavior ontology class,the judgment of unsafe behavior of miners is completed.Finally,based on the above research content,this paper puts forward a method of knowledge driven and data-driven behavior understanding and identification for the unsafe behavior of miners in the underground intelligent space,and proves the effectiveness of the scheme through the sub module experiment and the overall experiment.There are 38 figures,17 tables and 78 references in this paper.