roceed as C(t) ∞ tD, where the fractal dimentsion D is 0.43≤D＜1.0. As the fractring process progresses, the system's state initially changes from ordered to disordered (fractal dimension D decreases form aobut 1 to abiout 0.48) and then changes back to ordered (fractal dimentsion increases form 0.48 to about 0.91). Corresponding to each evolutionary process of the system's states, AE event patterns such as the AE event rate, ae count rate, and amplitude in rock fracturing processes were recognized using neural network techniques. AE event patterns at 8-10 succedding time points were predicted using thecorresponding models. AE event patterns in rock microfractring processes are effectively described by the neural dynamic model NN(n,h,1). The models so obtained are applicable for extrapolatd recognition of AE event patterns with adequate accuracy. An improved learning algorthm is proposed to train the networks with generally improved performace of the models.

Using acousticemission data obtained from laboratory bouble torsion tests, we have analysed the fractal natureof aseries of 29 granited microfracturing processes in time. The data represent a wide variety of timescales, stress environments (increasing load with a constant displacement rate, relaxation, creep), soaking nditions [air, water, dodecyl trimethyl ammonium bormide (DTAB), poylethelene oxide (PEO)], and materialanisotroy. We fine that the timedistribution of rock microfraturing displays fractal and multifractal properties, In some cases, it hasasinglefractal of a multifractal structure. In other casese, it changes form a single fractal structure into a multifractal structure as the system eveoles dynamically. We suggest that the heterogeneity ofthe rock, the distribution of jojints of weak planes, the stress level, and the nature of the microfracturing mechanism lead to these multifractal properties. Whatever the fractal structure of the system, al lowerfractal dimentsion is generally produced at near-failure of the rockduee to an increased clusteing. This resultconcerning the fractal-dimension decrease is consistent with the conclusion drawn from the spatialdistribution of rock microfracturing. Thererore, from the vantage oint of observation of thime distribution of rock microfracturing, the decrease of the fractal dimension has a potential use as a rock failure predictor.

地下硐室围岩的稳定性评介是个极其复杂、知识匮乏的问题。首交将数据挖掘方法应用到地下硐室围岩稳定性判别知识的自学心中。鉴于现有数据挖掘方法未能考虑负属性的挖掘，提出了可能考虑负属怀的一种新的数据挖掘方法，它从硐室岩稳定性的实例数据中挖掘出知识、并将得到的知识输入专家系统，进行不确定的推理，对下硐室围岩的稳定性进行合理的判别。

针对岩爆预测的问题，提出了基于支持向量机的预测方法，通过对影响岩爆因素的分析，然后运用支持向量机理论建立岩爆预测的支持向量机模型，结果表明，基于支持向量机的岩爆预测方法具有较高的准确率，该方法是科学可行的，具有广泛的应前景。