| 摘要 |
PROBLEM TO BE SOLVED: To improve bedrock property predicting accuracy by computing the change pattern of boring data of a hydraulic drill, and separately from this, obtaining an average change pattern through actual boring as input information and actual bedrock properties as output information. SOLUTION: Machine data and the like during boring are measured and computed in a sub-system 1. Neural network with an average change pattern through actual boring and a frequency characteristic as input information and with actual bedrock property is output information is separately constructed. In the neural network, weighting computation is performed to each data of input layers from the information of computed average change to predict the properties of the bedrock. In a sub-system 2, sound and vibration data during boring is measured to predict the properties of the bedrock in a boring plate to be good, normal, bad, or the like by fuzzy reasoning, and after input to a computer, the grade of the bedrock is predicted. Further in a sub-system 3, a core is extracted if necessary to predict the properties of the bedrock. |
