As the acquisition of downhole vibration signals develops towards high-frequency acquisition,the data volume required to be stored and transmitted by the downhole vibration acquisition module is gradually increas-ing.In order to solve the difficult problems of downhole data storage and uploading and provide early warning for the operating status of downhole drilling tool,the compressed sensing theory and support vector machine(SVM)model were integrated into the downhole vibration signal storage,transmission and downhole drilling tool status warning.An atomic number adaptive sparse dictionary building method was studied to use a small number of sparse features to express a complete signal.An observation matrix was built to project the original signal onto a low di-mensional space to achieve signal compression.The improved cuckoo search(ICS)was applied for parameter opti-mization of the SVM model,and the trained ICS-SVM model achieved drilling tool status warning.The application results show that the compressed sensing technology can compress downhole vibration data to 12%,with a data re-construction error of 0.177 2,and the success rate of ICS-SVM model for drilling tool status warning reaches 98%.The research results have achieved the goal of alleviating the pressure of storing and uploading downhole vibration data,which helps working personnel better carry out real-time drilling operations and status warnings.
维普
万方数据