Study On The Safety Risk Assessment Method Of Mountain Tunnel During Construction

It is extremely important to carry out safety risk management during the construction of mountain tunnels. However, existing research approaches are seriously lagging behind the engineering applications. For the ongoing excavated tunnels, there are various influence factors leading to low reliability, poor accuracy and weak practical applicability for risk assessment results. These factors include subjectivity in determining index weights and difficulty in obtaining the certain, precise and complete evaluation indexes. Because of this, the existing risk assessment techniques fail to meet the requirements of standardization, informatization and security management for tunnel constructions. Therefore, in this paper, a new risk assessment method was established by combining case-based reasoning, geological prediction and rough set theory. This novel method had been applied in engineering practice and the feasibility had also been verified.Taking into account of the paper length and research time, evaluation of loss caused by risk events, risk control, etc., are not involved in this paper. This paper focuses on the probability of occurrence of typical tunnel risk events. Based on the reviews of current researches, the paper consists of four parts. Part one is about the problems of traditional rough sets theory in tunnel risk assessment and the improvements of attribute simplification method of the condition information entropy. Part two constructs the assessment index system based on case-based reasoning. Part three concerns the indicative effect of ahead geological forecasting results on the typical geological units and applies the representative geophysical prospecting results in the evaluation indexes. The last part establishes the risk assessment method by combining of the above rough sets theory, case-based reasoning and ahead geological prediction.a) About application research of rough setsFor sample sets of practical tunnel collapse cases, based on the dependency principle of attributes, the attribute importance degrees and weights were calculated before and after attribute simplification. It was discovered that the simplification based on the dependency of attributes was excessive in risk assessment of tunnels. Both extra-reduction and low reliability of the results were due to the granularity representation of knowledge and the restriction of decision information table. There was another possible problem, i.e., too much attribute simplification results could lead to choose impossibility and the attribute importance degrees and weights could not even be determined. Based on the situation, this paper proposed a method of determining the attribute importance degrees and weights upon the information entropy. It was a new attribute simplification method as well as the related software. This method had been applied in practical tunnel collapse cases and the results were good enough. This method could filter the risk influence factors, abandon insignificant/extra factors, calculate the objective weights of every factor and solve the problem that weights were assigned completely by subjectivity.b) About structure of assessment index systemBased on past tunnel incidents and technical regulations, the risk carrying environment and risk-caused factors were set up by level analysis method for typical risk incidents, such as collapse and inrush disasters. After the integration and conclusion of relevant information and factors, the assessment index system was built for typical risk incidents, and this laid a foundation for rough sets in case-choosing and evaluation procedure.c) About ahead geological prediction researchGeological monitoring and prediction system were founded based on geological investigations, tests of excavated section and synthesized geophysical detections of unexcavated section. The characteristics of the dynamic response were studied for the seismic wave and electromagnetic wave. By comparison between detection results and real situation of revealed surrounding rocks, the indicative effect of active dynamic response parameters to typical geological body was dug out, and introduced into the risk assessment index. Quantifiable indexes were developed for rough sets theory and tunnel security evaluation. The risk forecasting was realized for the unexcavated section with certain distance in front of the tunnel face.d) About risk assessment method The main risk carrying environment and geological prediction results were analyzed for to-be evaluated tunnel, then cases with similar attribute characters were chosen from the case library. After that, the attribute importance degrees, index weights and attribute simplification of similar accident cases were calculated based on rough sets theory, which optimized evaluation index selection indirectly. Finally, optimized evaluation indexes and weights were applied to fuzzy comprehensive evaluation model so that the comprehensive assessment of risk events probability was achieved for the to-be evaluated tunnels.