Acoustic emission signal identification of different rocks based on SE-1DCNN-BLSTM network model

In order to study fracture mechanism of rocks in different brittle mineral contents,this study pro-poses a method to identify the acoustic emission signal released by rock fracture under different brittle miner-al content(BMC),and then determine the content of brittle matter in rock.To understand related interference such as the noises in the acoustic emission signals released by the rock mass rupture,a 1DCNN-BLSTM network model with SE module is constructed in this study.The signal data is processed through the 1DCNN and BLSTM networks to fully extract the time-series correlation features of the signals,the non-correlated features of the local space and the weak periodicity law.Furthermore,the processed signals data is input into the fully connected layers.Finally,softmax function is used to accurately identify the acoustic emission signals released by different rocks,and then determine the content of brittle minerals contained in rocks.Through experimental comparison and analysis,1DCNN-BLSTM model embedded with SE module has good anti-noise performance,and the recognition accuracy can reach more than 90 percent,which is better than the traditional deep network models and provides a new way of thinking for rock acoustic emission re-search.

Identification of acoustic emission signal in aluminum alloys spot welding based on fractal theory

The acoustic emission signal of aluminum alloys spot welding includes the information of forming nugget and is one of the important parameters in the quality control. Due to the nonlinearity of the signals, classic Euclidean geometry can not be applied to depict exactly. The fractal theory is implemented to quantitatively describe the characteristics of the acoustic emission signals. The experiment and calculation results show that the box counting dimension of acoustic emission signal, between 1 and 2, are distinctive from different nugget areas in AC spot welding. It is proved that box counting dimension is an effective characteristic parameter to evaluate spot welding quality. In addition, fractal theory can also be applied in other spot welding parameters, such as voltage, current, electrode force and so on, for the purpose of recognizing the spot welding quality.

Chaos analysis of acoustic emission signals in spot welding process

Based on chaos time series and fractal theory, acoustic emission signals were studied in the process of spot welding. According to calculating 8 welding parameters using phase space reconstruction method, the largest Lyapunov exponents were positive values and chaos characteristics were firstly discovered from acoustic emission signals in spot welding. In order to evaluate acoustic emission signal, Hausdorff dimension is put forward to analyze and estimate chaos characteristics. The experiment and calculation results indicate that the Hausdorff dimension of acoustic emission signal is significantly distinguishable in the nuggets with different welding parameters. This research provides a new method for measuring the resistance spot welding quality.

Noise reduction algorithm of corrosion acoustic emission signal based on short-time fractal dimension enhancement

The general corrosion and local corrosion of Q235 steel were tested by acoustic emission （AE） detecting system under 6% FeCl3.6H2O solution to effectively detect the corrosion acoustic emission signal from complex background noise. The short-time fractal dimension and discrete fractional cosine transform methods are combined to reduce noise. The input SNR is 0-15 dB while corrosion acoustic emission signals being added with white noise, color noise and pink noise respectively. The results show that the output signal-to-noise ratio is improved by up to 8 dB compared with discrete cosine transform and discrete fractional cosine transform. The above-mentioned noise reduction method is of significance for the identification of corrosion induced acoustic emission signals and the evaluation of the metal remaining life.

Combined optical fiber interferometric sensors for the detection of acoustic emission

A type of combined optical fiber interferometric acoustic emission sensor is proposed. The sensor can be independent on the laser source and make light interference by matching the lengths of two arms,so it can be used to monitor the health of large structure. Theoretical analyses indicate that the system can be equivalent to the Michelson interferometer with two optical fiber loop reflectors,and its sensitivity has been remarkably increased because of the decrease of the losses of light energy. PZT is powered by DC regulator to control the operating point of the system,so the system can accurately detect feeble vibration which is generated by ultrasonic waves propagating on the surface of solid. The amplitude and the frequency of feeble vibration signal are obtained by detecting the output light intensity of interferometer and using Fourier transform technique. The results indicate that the system can be used to detect the acoustic emission signals by the frequency characteristics.

A Rub-Impact Recognition Method Based on Improved Convolutional Neural Network

Based on the theory of modal acoustic emission(AE),when the convolutional neural network(CNN)is used to identify rotor rub-impact faults,the training data has a small sample size,and the AE sound segment belongs to a single channel signal with less pixel-level information and strong local correlation.Due to the convolutional pooling operations of CNN,coarse-grained and edge information are lost,and the top-level information dimension in CNN network is low,which can easily lead to overfitting.To solve the above problems,we first propose the use of sound spectrograms and their differential features to construct multi-channel image input features suitable for CNN and fully exploit the intrinsic characteristics of the sound spectra.Then,the traditional CNN network structure is improved,and the outputs of all convolutional layers are connected as one layer constitutes a fused feature that contains information at each layer,and is input into the network’s fully connected layer for classification and identification.Experiments indicate that the improved CNN recognition algorithm has significantly improved recognition rate compared with CNN and dynamical neural network(DNN)algorithms.

Low Speed Bearing Fault Diagnosis Based on EMD-CIIT Histogram Entropy and KFCM Clustering

In view of weak defect signals and large acoustic emission(AE) data in low speed bearing condition monitoring, we propose a bearing fault diagnosis technique based on a combination of empirical mode decomposition(EMD), clear iterative interval threshold(CIIT) and the kernel-based fuzzy c-means(KFCM) eigenvalue extraction. In this technique, we use EMD-CIIT and EMD to complete the noise removal and to extract the intrinsic mode functions(IMFs). Then we select the first three IMFs and calculate their histogram entropies as the main fault features. These features are used for bearing fault classification using KFCM technique. The result shows that the combined EMD-CIIT and KFCM algorithm can accurately identify various bearing faults based on AE signals acquired from a low speed bearing test rig.