This paper proposes a behavior trajectory restoration algorithm for observation sequence state missing problem, which leeds to terminal trajectory restoration inaccurately. The algorithm utilizes base station layout’s spatial correlation and revises the partial probability of the solution process of the Hidden Markov Models(HMM) to restore the track sequence without considering the missing observation states. Performance analysis and simulation results show that the greater the degree of state propensity is, the higher the success rate of trajectory restoration is. When the degree of state propensity is 0.8, the success rate of trajectory restoration is about 90 percent.
In the single Low Earth Orbit(LEO) satellite localization system, grid search algorithm requires large amount of computation in the precision of search and Taylor-series linearization algorithm exists localization convergence problem. In view of this, a new method is proposed which is combined of grid search and Taylor-series linearization algorithm. Using grid search method to achieve the rough location and taking the location result as the initial value of the Taylor-series linearization, the high-precision position of target can be got. The constrained Cramer-Rao Lower Bound(CRLB) for the source localization estimation is derived with the constraints of altitude. Simulation results are provided to analyze the effects of various factors on the location precision. The analysis results show that the localization from frequency measurement by the single LEO satellite can effectively achieve the target location on both sides below the satellite.
For solving the problem of tracking space target that has unknown maneuver under glint noise, an Adaptive Robust Unscented Particle Filtering(ARUPF) algorithm is proposed in this paper. Adaptive Robust Unscented Kalman Filtering(ARUKF) algorithm is designed by embedding adaptive robust filtering technique into Unscented Kalman filtering(UKF). ARUPF is developed by using ARUKF to generate the importance density function in Particle Filtering(PF). Combined with transient tracking model, ARUPF is applied for space maneuvering target autonomous tracking. Experimental result shows that the proposed algorithm improves the tracking accuracy and robustness, contrasting to the existing filtering algorithms.
In the research of space handwriting recognition technology based on 3D accelerometer, a recognition method based on time-frequency fusion feature is proposed. From accelerometer data, it extracts the Short-time Energy(STE) feature. The hybrid feature which combines Wavelet Packet Decomposition with Fast Fourier Transform(WPD+FFT) are extracted, then the above two categories features are fused together and the Principal Component Analysis(PCA) is employed to reduce the dimension of the fusion feature. Supported Vector Machine(SVM) is used in recognition. Experimental results show that the proposed method can improve the performance of 3D space handwriting recognition system.
This paper proposes a modified Perceptual Evaluation of Speech Quality(PESQ) algorithm to improve the performance of objective speech quality evaluation. Syllable stability detection and Unvoiced/Voiced/Silence(UVS) classification are used in the proposed method. Parameters of stability of frames and syllables distortions are used to describe the effect to hearing perception, and these parameters are different to variant speech segments, especially to unvoiced, voiced and silence segments. Experimental results show that the proposed algorithm for PESQ is able to improve correlation results for narrowband speech and can help to improve other speech quality evaluation.
This paper deeply investigates the homograph in Uyghur language and classifies them according to the different features of homograph, disambiguates the first type of homograph according to the mapping relation between the part of speech and pronunciation, disambiguates the second type of homograph according to vowel weakening when suffix attaches to a stem, and optimal pronunciation mapping method is used to disambiguate the third type of homograph by extracting the contextual features of homograph. Log-likelihood ratio is used to select keywords and keyword selection experiment of different window size is also conducted. Experimental result shows that the homograph disambiguation performance of can be got to 20.9% error rate through the research idea of this paper.