The image reconstructed by ECT image reconstruction algorithm has the problems of unclear edge and low imaging accuracy. It is easy to misjudge the interface of different mediums, and the accuracy of cross-section water content calculation depends on the accuracy of binary segmentation of the reconstructed image. Therefore, by analyzing the shortcomings of the global threshold segmentation algorithm, we propose a local threshold segmentation algorithm combined with simulated annealing algorithm. In order to verify the effectiveness, the reconstructed images of laminar flow, central flow, annular flow are segmented by local optimal threshold, and the cross-section water content is calculated. By comparing with the global threshold segmentation algorithm, it is proved that the local threshold segmentation algorithm has a relatively higher accuracy in calculating the section water content. In addition, considering the nonlinear error of ECT system, an image threshold segmentation algorithm considering the nonlinear error of ECT is adopted by combining with the local threshold segmentation algorithm. Comparing the global threshold segmentation algorithm and local threshold segmentation algorithm without considering nonlinear error, shows that the local threshold segmentation algorithm considering ECT nonlinear error has a relatively higher accuracy in the calculation of cross-section water content.

The flow pattern recognition technology based on capacitive tomography (ECT) system is widely used in dynamic measurement of various industrial processes due to its non-invasive and visual advantages. To improve the flow pattern identification accuracy of ECT system, recognition algorithm based on neural network model is widely used. But the ECT system can’t get a lot of ECT system measured samples, resulting in low accuracy, low model generalization. To solve this problem, this paper adopted the dataset amplification method based on deep convolution generative adversarial network (DCGAN), and used the images reconstructed by the preprocessed LSSVM method to build a flow pattern image generation model based on DCGAN for experiments. Gan-train and GAN-test were used to evaluate the generated images. After that, the amplified dataset is used for flow pattern identification based on YOLOv3. The experimental results show that the DCGAN method can effectively amplify the original data set and generate images with high accuracy and richness. Together with the target detection algorithm of YOLOv3, the flow pattern recognition accuracy of ECT system can be improved to a great extent.

For a long time, China’s energy structure has been dominated by coal-fired power generation. The smoke and dust generated by coal-fired power generation are the main atmospheric particulate pollutants, causing serious harm to the environment and human health. It is very important to monitor the speed of soot emission accurately to prevent environmental pollution effectively. In view of the advantages of electrostatic sensor with high sensitivity, not affected by flue dust accumulation, simple structure and low cost, this paper adopts probe electrostatic sensor to measure dust flow parameters in power plant flue. By acquiring the electrostatic signal which contains rich information of two-phase flow, the dust flow velocity is monitored by the spatial filtering method based on Hilbert-Huang transform. Through constructing pneumatic conveying experiment platform measuring system, the static characteristic of flow velocity is tested. The experimental results show that the electrostatic method used to detect the smoke velocity data has higher reliability, measurement error is smaller, the smoke velocity detection technology has significant improvement.