Improving the ability to detect small leaks to prevent more severe accidents plays an extremely important role in the safe operation of pipelines. To tackle the issue of low diagnostic accuracy associated with single sensors for detecting small leaks, a multi-source multi-modal feature fusion method for gas pipeline leak detection was proposed. First, the collected data from multiple sensors were transformed into two-dimensional time-frequency images for input into the feature extraction network. Then, the dual-information fusion (DIF) module was introduced, incorporating the attention mechanism and multi-scale feature fusion to enhance the model’s feature expression capability, and fully interact with the multi-modal features. Secondly, the channel split multiscale convolution (CSMC) module was designed to accommodate the diversity of input data and improve the model’s generalization capability. The DIF and CSMC modules were cascaded and fused to produce the classification results through the fully connected layer. Finally, the effectiveness of the proposed method was assessed using pipeline leak data collected in the laboratory. The experimental results demonstrate that the proposed multi-modal deep learning model can effectively identify the small leak state in pipelines, exhibiting superior diagnostic performance when compared to the current mainstream image classification models.

Natural gas pipelines are an important mode of transportation. How to quickly and accurately detect abnormal signals that occur during pipeline operation is crucial to judging the pipeline status. In this paper, an abnormal signal detection method based on improved YOLOv7 (YOLO-PD) was proposed to address the low detection accuracy of abnormal signals and the dependence on manual signal segmentation scale setting in pipeline operation. Firstly, a new refined channel attention (RCA) module was proposed to enhance network feature extraction. Secondly, a MaxPooling with RCA module was designed to strengthen network feature extraction and fusion capabilities. Furthermore, the atrous spatial pyramid pooling with RCA module was used to expand the model’s receptive field and enhance the core information features, thereby improving the detection accuracy of the model for multi-scale objects. Subsequently, the SIoU loss function was utilized to expedite network convergence and optimize the model training process. The experimental results demonstrate that YOLO-PD exhibits high accuracy and speed, with an mAP@0.5 of 0.996 and a frame rate of 65. In comparison to other prevalent target detection algorithms, the method proposed in this paper can effectively identify abnormal signals in pipelines and achieve higher detection accuracy.

This study presents a novel sensor that combines air pressure and temperature measurement using dual-interference microfiber resonators. The sensor leverages intermode interference and the resonant ring resonance effects within the microfiber loop structure, enabling simultaneous demodulation of temperature and air pressure. We have developed a mathematical model that is closely monitored alongside the reflectance spectral structure. The sensor exhibits two distinct linear sensitivities for pressure and two different linear sensitivities for temperature, as revealed by our testing results. Specifically, microfiber mode interference demonstrates a temperature and pressure sensitivity of 160 pm/°C and 167/kPa, respectively, while the ring resonance effect yields a temperature and pressure sensitivity of 132 pm/°C and 62 pm/kPa, respectively. The maximum temperature resolution stands at 0.025°C, and the pressure resolution is 0.023 kPa. These findings are supported by the concept of a linear connection, which serves as the foundation of the transmission matrix and enables concurrent discrimination of air pressure and temperature. This sensor offers advantages due to its straightforward structure, high extinction ratio, superior electrical safety, and effective protection against electromagnetic interference.

In the field of structural health monitoring, it has been pursued to achieve large-range and high-sensitivity strain and vibration measurements. This paper proposes a sensor based on a microfiber coupler (OMC) and a Sagnac ring to achieve large-range and high-sensitivity strain and vibration measurements. The sensor takes advantage of the high sensitivity of the evanescent field around the OMC and the high tensile resistance of the Sagnac ring. This paper analyzes the strain and vibration sensing mechanism, simulates the mode distribution in the OMC, and presents an experimental setup. Experimental results demonstrate that the sensor can achieve strain measurement with a sensitivity of -9.11 pm/με in the range of 0-4500 με and vibration measurement at 0.5 Hz-5 kHz. The proposed sensor provides potential technical support for the fields of structural health monitoring and fault diagnosis.

Rapid and accurate estimation of the state of health of lithium-ion batteries is of great significance. This paper aims to address two issues faced when applying deep learning methods to estimate the health status of lithium-ion batteries: high data quality requirements and poor model generalizability. This paper proposes a comparison-transfer learning approach with cyclic synchronization to estimate the state of health of lithium-ion batteries. First, a cyclic synchronization method based on the Bezier curve fitting algorithm is introduced to synchronize the data obtained at different charge–discharge cycles of the lithium-ion battery, facilitating input to the model. Second, a comparison-transfer network using the Pearson correlation coefficient is proposed to transfer knowledge from the source dataset to predict the target dataset under different environmental temperatures. By training a pre-trained model on the source dataset and utilizing the correlation coefficient to analyze the similarity between the source and target datasets, the accumulated knowledge in the source dataset can be effectively utilized to enhance prediction performance on the target dataset. In the experiments, the proposed method is validated using the lithium-ion battery aging public datasets. The experimental results demonstrate that the proposed approach achieves superior prediction performance in the case of small-sample sizes, exhibiting higher accuracy and stability compared to traditional deep learning methods.