Stability Prediction of Coal Mine Water Disasters Emergency Rescue System Based on Support Vector Machine

In order to effectively improve the stability of the coal mine water disasters emergency rescue system, the establishment 0f the index system to measure the stability standing on the view of mid-control and dynamic analysis, which includes six elements: personnel quality factor、capital factor、management factor、information elements、machine elements and geological environment. As well, aiming at the characteristic of complex and data acquisition difficulty of coal miner rescue system, proposed a forecasting model of the stability of the coal mine water disasters emergency rescue system based on improved particle swarm optimization and support vector machine (IPSO-SVM). Finally, test the validity of the model through the case experiment. Coal mine water disaster is one of the major disasters of the safe and efficient development of coal resources .From a macro point of view, the coal mine water disaster in our country is on the rise; For coal mining enterprises, once the coal mine was flooded by sudden water, ore party are often helpless for water disasters, often delay the best time to rescue, as a result, the coal mine suffered serious economic losses, the miners and their families suffered a great disaster and pain. Therefore, the study of coal mine water disaster emergency rescue system is particularly important. Currently, the study mainly focused on the emergency rescue capability evaluation [2] and rescue logistics management , rarely systematically focused on the coal mine water disaster rescue system. In view of the characteristics of the coal mine water disaster rescue system that the system is complicated, the data acquisition is difficult, we propose a forecasting model of the stability of the coal mine water disasters emergency rescue system based on improved particle swarm optimization and support vector machine. 1. The index system of coal mine water disaster rescue organization system stability Coal mine water disaster rescue is a complex dynamic system involving humanmachine environment, engineering technology, and other elements, in order to improve the reaction efficiency and the quality of crisis management of coal mine, we take management of the stability of coal mine water disaster rescue system to measure the stability of coal mine rescue organization system and to improve the organizational system and to lay a solid foundation for the rescue work. The coal mine water disaster rescue organization system is established according to the thought of system engineering. The system includes personnel qualityB1、capital elementsB2、management factorsB3、information factorsB4、machine elementsB5、geographical environment factorsB6.B1 includes the using proportion talentsC1、the overall educational level of employeesC2 and the ability of employees to continue ascensionC3; B2 includes the proportion of business incomeC4 and the proportion of government investment incomeC5; B3 includes the degree of management systemC6、 popularization of advanced management experienceC7 and the application of information managementC8; B4 includes the flow rate of informationC9 and the update rate of informationC10; B5 includes the update and maintenance rate of system equipmentC11、the perfect degree of the waterproof and drainage systemC12 and the advanced level of relief means and technologyC13; B6 includes the ratio of special equipment and personnelC14、the detection ability of geological environment C15 and the perfect degree of the hydrogeology dataC16. International Conference on Information Sciences, Machinery, Materials and Energy (ICISMME 2015) © 2015. The authors Published by Atlantis Press 340 2. The establishment of prediction model of coal mine water disaster rescue organization system stability There is a vague, non-linear relationship between the mine water disaster rescue organization system stability and affecting factors, the support vector machine with strict theoretical basis has a distinct advantage to solve the problem of nonlinear, high dimension and small sample: less data sample demand, high accuracy and strong adaptability, effective way to avoid the curse of dimensionality, etc. Therefore, the support vector machine water is introduced into the coal mine disaster relief organization system stability prediction, while selecting SVM kernel function and parameter as a breakthrough for in-depth study. 2.1 Basic learning method of support vector machines (SVM) Vapnik [4] proposed the concept of kernel functions about SVM nonlinear classification surface problems, the basic idea is: a pre-determined by non-linear mapping of the input vector x is mapped to a high-dimensional space, then this high-dimensional space constructed optimal classification surface. Set of training data set ( ) { } n i y x i i   , 2 , 1 , , = for the size of n, { } 1 , 1 − + ⊂ y can be transformed into quadratic optimization problem: The constraint conditions: ( ) 1 ≥ + ⋅ b x w y i i n i  , 2 , 1 = (1) Next, seek the minimum value. The solution to this optimization problem is a saddle point of the following Lagrange function: ( ) ( ) [ ] 1 2 1 , , w 1 2 − + ⋅ − = ∑ = b x w y a w a b L i i n