Aiming at the long-term cumulative error inherent in pedestrian indoor inertial positioning filed, that error is mainly due to the low signal-to-noise ratio of the sensor output signal quality, the temperature drift of the gyro and the accuracy of error estimation. This paper proposes a new optimization method for array distributed MEMS-IMU: this method performs filtering and noise reduction optimization processing on inertial sensor data; The effect of temperature on the gyroscope is reduced by matrix-optimized layout, and distributed temperature compensation is performed for eight IMUs. We used MEMS-IMU worn on the foot finishing the data acquisition. Then improved a novel Pearson coefficient particle filtering method to finally complete the information fusion and positioning process in a blind environment (no beacon auxiliary information) high-precision personal large span (long time span, large distance span). The indoor positioning test results in the No. 6 Office Building of National Defense Science and Technology Park in Beijing Institute of Technology verify that the method has a horizontal error of only 6.23m ( TTD≈0.52%) during the horizontal span positioning which the total distance is about 1200m; In terms of vertical large-span positioning accuracy: the height error is only 4.56m ( TTD≈7.6%) during the positioning process of 68 minutes and 35 seconds (including intermediate stop). Compared with other multi-IMU personal positioning optimization methods, it has the advantages of high sensor data quality, small gyro temperature influence, good system error estimation accuracy and long-term long-distance positioning results. It provides a good and reliable theoretical reference for this field or extension applications.

In order to resolve the problem that the sample of image for internal detection of DN100 buried gas pipeline microleakage is single and difficult to identify, a recognition method of microleakage image of the pipeline internal detection robot is proposed. First, nongenerative data augmentation is used to expand the microleakage images of gas pipelines. Secondly, a generative data augmentation network, Deep Convolutional Wasserstein Generative Adversarial Networks (DCWGANs), is designed to generate microleakage images with different features for detection in the pipeline of gas pipelines to achieve sample diversity of microleakage images of gas pipelines. Then, a bi-directional feature pyramid network (BiFPN) is introduced into You Only Look Once (YOLOv5) to retain more deep feature information by adding cross-scale connecting lines in the feature fusion structure; finally, a small target detection layer is constructed in YOLOv5 so that more shallow feature information can be retained to achieve small-scale leak point recognition. The experimental results show that the precision of this method for microleak identification is 95.04%, the recall rate is 94.86%, the mAP value is 96.31%, and the minimum size of identifiable leaks is 1 mm.

Aiming at the inherent problems of heading divergence and error accumulation in indoor inertial positioning of pedestrian field, we proposed a new indoor positioning error correction method for pedestrian multi-motions recognition. This method is aimed at pedestrians’ 7 common indoor movements. Periodically divide the motion data of the accelerometer in sensitive-axis-direction (vertical geodetic direction) of MEMS-IMU worn on the pedestrian’s waist. And acquire the feature vector through feature extraction by the hybrid-orders fraction domain transformation (HFDT). The effective features with high identification degree under the different optimal order transformation are mixed and matched. Then these are sent to each sub-classifier to complete the classification process by the dichotomy, and finish the processes of subsequent machine learning. The error correction of the heading angle and positioning is carried out combining with the improved HDE algorithm, the innovative floor constraints method and the motion states transition correction technique. The final indoor positioning experiment results show that the classification accuracy of this motion detection method compared with the traditional method, the average classification accuracy reaches up to 97%, and reduces the computing requirements for hardware. By using the floor constraint method, the vertical height difference can achieve the effect of complete reset of the origin. The trajectory lines of the movement in the same floor are well displayed. At the same time, the best horizontal error positioning result is only 1.72m during the total travel distance: 412.40m (TTD≈0.42%).