Diamond sawblade is an efficient tool to building renovation or demolition. Concrete used in construction is a typical composite material with random distribution, which is difficult to accurately identify and predict even under the same processing conditions, and tool life of diamond sawblade is difficult to control.

In this paper, by cutting out single component of the hard and soft aggregate separately from concrete, the single component and concrete experiments were carried out to understand the sawing characteristics of different components. The wavelet decomposition was used to analyze the characteristic of each frequency band of the different components sawing force and vibration signals, and the sensitive frequency bands after correlation coefficient and energy ratio variation of each wavelet layer were extracted to judge the bluntness status of sawblade. By taking the Root-Mean-Square (RMS) value, the energy ratio of d2 and d5 wavelet layers and the standard deviation of sawing force and vibration signal as the characteristic values of the sawblade, a neural network optimized by bat algorithm was established to analyze the concrete processing signals and predict the working state of the sawblade. Evidence theory was adopted to combine the prediction results of sawing force and vibration samples to increase the overall prediction accuracy and reliability. The test sample showed that this method can correct inconsistent individual sensor predictions while being as close to the actual status value as possible. It provides an effective tool life prediction way of the diamond sawblade and a theoretical method for the monitoring of non-metallic materials with inhomogeneous components.

This content is only available via PDF.
You do not currently have access to this content.