Intelligent Identification And Evaluation Of Behavior Deviation In Desktop Emergency Drill Based On Text Mining

The identification and evaluation of behavior deviation in desktop emergency drill is an important way for refining and chemical enterprises to find out the weak points of emergency management and improve their emergency capacity.At present,the identification and evaluation of desktop emergency drill behavior deviation in refining and chemical enterprises mostly rely on people’s subjective experience,the results are quite different,and the utilization rate of emergency drill related texts containing a large amount of information law is small.Therefore,in order to ensure the objectivity of drill behavior deviation identification and evaluation and give full play to the role of relevant texts of desktop emergency drill of refining and chemical enterprises,this paper uses text mining technology to extract the information law in relevant texts of desktop emergency drill of refining and chemical enterprises,and mainly carries out the following research:(1)Aiming at the problem of low accuracy of refining and chemical enterprise text segmentation caused by the lack of refining and chemical enterprise user-defined dictionary,a construction method of refining and chemical enterprise user-defined dictionary based on new word recognition algorithm is proposed.By setting the two limiting conditions of mutual information and adjacency entropy,the text related to the emergency drill of refining and chemical enterprises is identified with new words.A total of 1797 commonly used words of refining and chemical enterprises are identified and added into the participle dictionary to complete the construction of self-defined Dictionary of refining and chemical enterprises.Taking the word segmentation of the comprehensive emergency plan of a refining and chemical enterprise as an example,compared with the word segmentation results before and after using the customized Dictionary of the refining and chemical enterprise,the accuracy of word segmentation after use is 97.79%,which is 10.18% higher than that when not used,and the number of fields obtained by text segmentation is reduced by 1789,which is more in line with the actual language expression.(2)Aiming at the problem that the identification and evaluation of desktop emergency drill behavior deviation in refining and chemical enterprises rely too much on people’s subjective experience,resulting in large differences in evaluation results and low utilization rate of text related to desktop emergency drill in refining and chemical enterprises,this paper proposes an intelligent identification and evaluation method of desktop emergency drill behavior deviation based on text similarity.By constructing a text similarity model based on word2 vec and LSTM,the judgment results of text similarity are used to identify the deviation of exercise behavior,and the behavior compliance rate is used to evaluate.The case analysis results using the method proposed in this paper show that this method can effectively identify most of the behavior deviations in the case,and the accuracy,accuracy and F1 value of the case analysis data are 80.56%,45.45% and 58.82% respectively.Compared with other methods,the maximum appreciation is 13.89%,16.88% and 16.71% respectively.(3)In view of the subjectivity of the drill evaluation function in the existing fulllevel scenario construction emergency drill operating system of refining and chemical enterprises,based on the above intelligent identification and evaluation method of desktop emergency drill behavior deviation of refining and chemical enterprises,and based on Python language,this paper designs and develops the emergency drill behavior ability evaluation module to provide data input,behavior deviation identification and behavior deviation evaluation functions,The relevant texts of emergency drill of refining and chemical enterprises are used to the greatest extent,which reduces the impact of people’s subjective experience on the evaluation results.The case analysis results show that the operation results of this module are in line with the objective reality,and can provide ideas for refining and chemical enterprises to find emergency management loopholes.