Grotto temple rock mass fracture development control method and device based on a parallel hybrid network

Abstract

The invention discloses a grotto temple rock mass fracture development control method and device based on a parallel hybrid network, and the method comprises the steps: enabling a local convolution module and a global circulation module to serve as a parallel network structure, and enabling multivariate time sequence data with the size of a time window to serve as input; extracting the time features from each univariate sequence in the multivariate time sequence, and constructing time pattern representation vectors with different time scales; enabling the self-attention module to perform the modeling on each univariate sequence by using a sparse self-attention mechanism, and capturing a dependency relationship among different sequences; wherein the linear autoregression module adds linear components for prediction, and enables output to respond to scale change of input; summing the output of the self-attention module and the output of the linear autoregression module, outputting a prediction result of rock mass fracture development, and performing instability detection. The device comprises a processor and a memory. According to the method, time patterns of different time scales are accurately captured, and a self-attention mechanism is introduced to model a complex dependency relationship among different sequences in multivariate time series data.

Description

technical field [0001] The present invention relates to the field of prevention and control of rock mass fissures in Shikusi, in particular to a method and device for controlling fissure development in Shikusi rock mass based on a parallel hybrid network. The analysis and utilization of historical data of environmental factors are used to predict the development trend of cracks in the cave temples in the future, and then realize the effective protection of the cultural relics of the cave temples. Background technique [0002] my country's cave temples are widely distributed, large in scale, and complete in system, integrating architecture, sculpture, murals, calligraphy and other arts, fully embodying the aesthetic pursuit, value concept, and cultural spirit of the Chinese nation. However, the cracks in the key rock mass of the Grotto Temple will continue to develop under the influence of multiple factors such as natural factors, rock mass structural force, and human activit...

AI Technical Summary

Problems solved by technology

[0006] 1. When predicting time series data, the sequence data will be calculated moment by moment. The prediction at the current moment depends on the previous prediction results, and iterative prediction cannot be accelerated in parallel;

[0007] 2. You can only use information from the previous moment

[0009] However, the current traditional machine learning methods are still basically blank in the prediction of rock mass fracture development, and other statistical methods are also difficult to model this problem, and there are currently no relevant data sets in this field at home and abroad. One Difficulty in Fracture Prediction

Images