Fuzzy C-Means Clustering Based Phonetic Tied-Mixture HMM in Speech Recognition

A fuzzy clustering analysis based phonetic tied-mixture HMM (FPTM) was presented to decrease parameter size and improve robustness of parameter training. FPTM was synthesized from state-tied HMMs by a modified fuzzy C-means clustering algorithm. Each Gaussian codebook of FPTM was built from Gaussian components within the same root node in phonetic decision tree. The experimental results on large vocabulary Mandarin speech recognition show that compared with conventional phonetic tied-mixture HMM and state-tied HMM with approximately the same number of Gaussian mixtures, FPTM achieves word error rate reductions by 4.84% and 13.02% respectively. Combining the two schemes of mixing weights pruning and Gaussian centers fuzzy merging, a significantly parameter size reduction was achieved with little impact on recognition accuracy.     

A New Arithmetic Coding System Combining Source Channel Coding and MAP Decoding

A new arithmetic coding system combining source channel coding and maximum a posteriori decoding were proposed. It combines source coding and error correction tasks into one unified process by introducing an adaptive forbidden symbol. The proposed system achieves fixed length code words by adaptively adjusting the probability of the forbidden symbol and adding tail digits of variable length. The corresponding improved MAP decoding metric was derived. The proposed system can improve the performance. Simulations were performed on AWGN channels with various noise levels by using both hard and soft decision with BPSK modulation.The results show its performance is slightly better than that of our adaptive arithmetic error correcting coding system using a forbidden symbol.     

An Adaptive Joint Source/Channel Coding Using Error Correcting Arithmetic Codes

An approximately optimal adaptive arithmetic coding (AC) system using a forbidden symbol (FS) over noisy channels was proposed which allows one to jointly and adaptively design the source decoding and channel correcting in a single process, with superior performance compared with traditional separated techniques. The concept of adaptiveness is applied not only to the source model but also to the amount of coding redundancy. In addition, an improved branch metric computing algorithm and a faster sequential searching algorithm compared with the system proposed by Grangetto were proposed. The proposed system is tested in the case of image transmission over the AWGN channel, and compared with traditional separated system in terms of packet error rate and complexity. Both hard and soft decoding were taken into account.     

Wavelet De-noising of Speech Using Singular Spectrum Analysis for Decomposition Level Selection

The problem of speech enhancement using threshold de-noising in wavelet domain was considered.The appropriate decomposition level is another key factor pertinent to de-noising performance.This paper proposed a new wavelet-based de-noising scheme that can improve the enhancement performance significantly in the presence of additive white Gaussian noise.The proposed algorithm can adaptively select the optimal decomposition level of wavelet transformation according to the characteristics of noisy speech.The experimental results demonstrate that this proposed algorithm outperforms the classical wavelet-based de-noising method and effectively improves the practicability of this kind of techniques.     

智能网联车辆混行交通流中灯语意图识别模型研究

为使混行交通流下智能网联车辆(Connected and Automated Vehicles, CAV)实现对人工驾驶车辆(Human-driven Vehicle, HV)前照灯灯语意图(Vehicle Headlights Intention, VHI) 的识别,弥补车对车(Vehicle to Vehicle, V2V)和鸣笛意图识别技术的不足,更好地与HV交互沟通,提出CAV对HV的VHI识别模型.模型包括：灯光感知、光数据处理、VHI识别3个模块,灯光感知模块通过RGB(Red-Green-Blue, RGB)和HSV(Hue-Saturation-Value, HSV)颜色空间感知前照灯(Vehicle Headlights, VH),采用KLT(Kanade-Lucas-Tomasi Tracking,KLT)和车辆匹配算法定位跟踪发出灯语的HV;光数据处理模块采用光通道增益算法计算光辐射通量变化; VHI识别模块基于双层隐马尔可夫模型(Double-layer Hidden Markov Model,DHMM)辨识VH 闪烁次数和HV行驶状态,实现VHI识别.在3种灯语示意典型场景下的实验结果表明：1 s内 VH感知准确率为96.8%,定位跟踪精度小于1°,VHI识别率为96.6%,满足混行交通环境下 CAV对HV驾驶意图的识别要求,基本保证实时性,为混行交通流中CAV自动驾驶决策提供理论依据.     

智能网联车辆混行交通流中灯语意图识别模型研究

为使混行交通流下智能网联车辆(Connected and Automated Vehicles, CAV)实现对人工驾驶车辆(Human-driven Vehicle, HV)前照灯灯语意图(Vehicle Headlights Intention, VHI) 的识别,弥补车对车(Vehicle to Vehicle, V2V)和鸣笛意图识别技术的不足,更好地与HV交互沟通,提出CAV对HV的VHI识别模型.模型包括：灯光感知、光数据处理、VHI识别3个模块,灯光感知模块通过RGB(Red-Green-Blue, RGB)和HSV(Hue-Saturation-Value, HSV)颜色空间感知前照灯(Vehicle Headlights, VH),采用KLT(Kanade-Lucas-Tomasi Tracking,KLT)和车辆匹配算法定位跟踪发出灯语的HV;光数据处理模块采用光通道增益算法计算光辐射通量变化; VHI识别模块基于双层隐马尔可夫模型(Double-layer Hidden Markov Model,DHMM)辨识VH 闪烁次数和HV行驶状态,实现VHI识别.在3种灯语示意典型场景下的实验结果表明：1 s内 VH感知准确率为96.8%,定位跟踪精度小于1°,VHI识别率为96.6%,满足混行交通环境下 CAV对HV驾驶意图的识别要求,基本保证实时性,为混行交通流中CAV自动驾驶决策提供理论依据.     

A Heuristic Method for Two-sided Assembly Line Balancing Problem

A two-sided assembly line is typically found in plants producing large-sized products. Its advantages over a one-sided line and the difficulties faced in two-sided line balancing problems were discussed. A mathematical model for two-ALB problem was suggested. A modification of the “ranked positional weight” method, namely two-ALB RPW for two-ALB problems was developed. Experiments were carried out to verify the performance of the proposed method and the results show that it is effective in solving two-sided assembly line balancing problems.     

面向自组织平衡的共享电动汽车调度优化方法

针对共享电动汽车运营中出现的站点车辆供需不平衡及人工调度成本过高等问题,提出面向自组织平衡的共享电动汽车调度优化方法.利用共享电动汽车站点选择行为数据,构建用户取车站点选择MNL(Multinominal Logit)模型,分析用户的站点选择行为;构建面向各站点车辆供需自组织平衡的站点间车辆调度优化方法,即通过车辆动态折扣策略引导用户改变默认的最近取还车站点,在运营结束后以运营企业的调度总成本最小为目标,建立人工调度优化模型并应用禁忌搜索算法求解;最后,选取北京市海淀区部分共享电动汽车站点进行算例分析.结果表明：本文提出的优化方法与无动态折扣策略相比,调度总成本降低了 4.5%,人工调度成本降低了21.1%,人工调度任务数减少了8.3%.     

面向自组织平衡的共享电动汽车调度优化方法

针对共享电动汽车运营中出现的站点车辆供需不平衡及人工调度成本过高等问题,提出面向自组织平衡的共享电动汽车调度优化方法.利用共享电动汽车站点选择行为数据,构建用户取车站点选择MNL(Multinominal Logit)模型,分析用户的站点选择行为;构建面向各站点车辆供需自组织平衡的站点间车辆调度优化方法,即通过车辆动态折扣策略引导用户改变默认的最近取还车站点,在运营结束后以运营企业的调度总成本最小为目标,建立人工调度优化模型并应用禁忌搜索算法求解;最后,选取北京市海淀区部分共享电动汽车站点进行算例分析.结果表明：本文提出的优化方法与无动态折扣策略相比,调度总成本降低了 4.5%,人工调度成本降低了21.1%,人工调度任务数减少了8.3%.