Papers by Keyword: Data Fusion

Abstract: This article compares the efficiency of vehicle trajectory analysis methods based on data fusion from multiple cameras, monitoring the same area from different views under the condition having detection errors, which causes incorrectly localized and, in some cases, undetected vehicle during the movement. The experiment used the simulation of detection and localization of vehicle moving in straight, curved, zigzag and arbitrary trajectories, with localization errors and multi-level loss of data. By comparing Kalman-filter-based method and Linear-interpolation-based method for analyzing and reconstructing vehicle trajectory, the result shows that the data loss robustness of Kalman-filter-based method is higher than that of Linear-interpolation-based method, with data loss around 97% 97% and 90% for straight, curved and zigzag trajectories respectively. However, for arbitrary trajectory, the Linear-interpolation-based method is better than Kalman-filter-based method in all levels of data loss. In conclusion, Kalman-filter-based method is effective in the case of unchanged or slight transition of direction, while Linear-interpolation-based method is effective in the case of sudden transition of direction.

Abstract: Multiscale surfaces enriched with different scales of features are becoming widely used in many fields such as MEMS and optics industries. Due to the complexity of the multiple scale geometries, a single sensor or a single measurement can hardly obtain the holistic information of these surfaces. Multiple measurement method can solve this problem while it will introduce a lot of challenges for multiscale data fusion for the measurement results. This paper presents a framework of a data fusion algorithm for precision measurement of multiscale surfaces. The method makes use of iterative closest point based method to precisely register the datasets obtained in multiple measurements, and a Gaussian zero-order regression filter is used to separate the geometric features in different scales. Hence, the datasets are fused based on an edge intensity data fusion algorithm within the same wavelength. Finally, the fused datasets of different wavelengths were merged and replaced to the corresponding area in the large scale measurement to form a new surface with holistic multiscale information. The effectiveness of the proposed method has been verified on a v-grooved surface through a series of simulation experiments

Abstract: Traditional research on sensor fault are usually confined to fault space locating, however, it’s very necessary to determine the time that the fault occurrence for subsequent data processing and to guarantee the normal operation of the monitoring system. Therefore, this paper proposes a method of sensors fault time locating. First of all, use Kalman filter to process the sensor data, then define the support level of sliding window correlation to sample correlation, combining with multi sensors data fusion, so as to identify the accurate time point of fault. Combine the deflection sensors data from Caiyuanba bridge, simulating four common faults. The results show that the error of located time is less than the width of the sliding window.

Abstract: Due to the characters of large space and uneven distributing of the temperature and humidity which always exist in the warehouse, the reliability and accuracy of data are influenced when using the wireless sensor network to collect the environmental parameters which are large redundancy and errors. According to the above-mentioned characters, a self-adaptive weighted algorithm based on multi-sensor data fusion was presented. The simulation results show that, the compute of the method is simple, it needs without any prior knowledge of sensor to give the fusion value with least variance, and therefore the proposed method improves the accuracy of measured data.

Abstract: The output of pressure sensor is affected by non-objection parameters in its application, its defect is the sensitivity to temperature. The method of the data fusion from the two sensors based on BP network can eliminate the side effect of temperatures and improve the accuracy and reliability for the pressure sensors. This method based on Levenberg-Marquardt algorithm of BP network has not only a simple network structure, but also a quick learning rate, showing a better prospect.

Abstract: In order to improve the accuracy, range and instantaneity of high temperature measurement, a CCD (Charge Coupled Device) image temperature measurement system was designed based on multi-sensor truncated mean weighed fusion. Firstly, CCD temperature measurement range was expanded by 6 light filters. Then 15 colorimetric-sensors calculated the surface temperature of experimental molten pool. Finally, the more accurate fusion estimate of temperature value was acquired by using truncated mean weighted factors. The experiment results demonstrate that system have faster processing speed in the measurement of 800°C. The errors were kept within ±2%. This method is effective in improvement of system accuracy and instantaneity.

Abstract: The fault diagnosis of electrical control system of certain type mine sweeping vehicle is difficult due to its complex structure and advanced technique. So in the multi-sensor failure diagnosis process, as a result of various reasons, such as the existence of measurement noise, diagnosis knowledge incomplete and so on, it makes the fault diagnosis uncertainty and affects the reliability and the accuracy of the diagnosis result. This article according to the analysis of electrical control system's fault characteristic of the mine sweeping plough’s, proposes a technique based on data fusion fault diagnosis method. The diagnosis process is divided into the sub system and the system-level, the subsystem uses the BP neural network to classify the fault mode, the system-level uses the D-S evidence theory carries on the comprehensive decision judgment for the whole system's fault. Application shows if some sub-neural network diagnosis has error, using D-S evidence theory fusion can effectively improve the accuracy of diagnosis.

Abstract: In this paper, the correlation between fused data and original data, the measured soil and the precipitation data over Huaihe river basin by exploring the inversion of soil moisture from the time and space based on the method of multi-source remote sensing data fusion has been studied. In order to fuse the AMSR-E data which is all-day and all-weather and can penetrate the earth surface to some extent, with the MODIS data that can reflect the surface condition and temperature characteristics, the method of wavelet fusion was carried out in MATLAB. The conclusions of this study are listed as follows: (1) the inversion result of the fused data based on AMSE-E and MODIS is much better than a single remote sensing data inversion; (2) the fused data based on AMSE-E and MODIS is sensitive to soil moisture change trend when the seasons alternated every year, especially in the spring, summer and autumn.

Abstract: Data fusion technology is widely used in data processing due to its characteristic of less transfer data. However, with the continuous application of wireless sensor networks, it raises higher demands for information integrity and privacy, data fusion faces new challenges. The paper focuses on issues related to the integration of wireless sensor network security data, analyzes its attack types of surface degrees and the need for security protection.

Abstract: Energy consumption problem and redundant data processing have become cause for concern in wireless sensor networks. In the article, a parameter estimation method based on temporal-spatial data fusion technology is proposed to reduce huge amounts of data transmission, reduce the node energy consumption, and increase network life cycle. Because of the layout of sensors, monitoring values will be different at different times in different places. Traditionally, though arithmetic average method has some anti-interference ability, there are many shortcomings in the accuracy of measurement. The proposed method can get drastically more accurate and reliable monitoring results in improving quality control of cold chain.