Recent research has increasingly focused on non-intrusive fault diagnosis of servo system components using rotor speed signals. This paper introduces a gearbox fault detection method for drive chains based on motor speed signals, suitable for variable speed operation. In this innovative alternative, a dual-domain separation deconvolution algorithm is proposed. First, mathematical modeling and analytical investigation are conducted on the characteristic behaviors of gear faults, specifically focusing on tooth breakage and wear conditions. Second, gear fault features in speed signals are reconstructed under dynamic speed conditions via the proposed dual-domain separation deconvoluction method. Then, gear health status is determined by comparing the extracted fault features with the known transmission ratio parameters. Finally, experimental results demonstrate the effectiveness of the proposed method for gear fault diagnosis under variable-speed conditions. The proposed fault diagnosis strategy has significant application potential in servo safety monitoring field, particularly in scenarios where vibration sensor installation is impractical.

We are gradually moving into a realm where sensors, processors, memory and wireless transceivers would be seamlessly integrated together in the physical world and form a wireless sensor network. Such networks pose new challenges in data processing and transmission due to the characteristic of limited communication bandwidth and other resource constraints of sensor networks. To reduce the cost of storage, transmission and processing of time series data generated by sensor nodes, the need for more compact representations of time series data is compelling. Although a large number of data compression algorithms have been proposed to reduce data volume, their offline characteristic or super-linear time complexity prevents them from being applied directly on time series data generated by sensor nodes. Motivated by these observations, we propose an optimal online algorithm GDPLA for constructing a disconnected piecewise linear approximation of a time series which guarantees that the vertical distance between each real data point and the corresponding fit line is less than or equal to ". GDPLA not only generates the minimum number of segments to approximate a time series with precision guarantee, but also only requires linear time O(n) bounded by a constant coefficient 6, where unit 1 denotes the time complexity of comparing the slopes of two lines. The low cost characteristic of our method makes it a proper choice for resource-constrained WSNs. Extensive experiments on two real data sets have been conducted to demonstrate the superior compression performance of our method.

The rapid development in processor, memory, and radio technology have contributed to the furtherance of decentralized sensor networks of small, inexpensive nodes that are capable of sensing, computation, and wireless communication. Due to the characteristic of limited communication bandwidth and other resource constraints of sensor networks, an important and practical demand is to compress time series data generated by sensor nodes with precision guarantee in an online manner. Although a large number of data compression algorithms have been proposed to reduce data volume, their offline characteristic or super-linear time complexity prevents them from being applied directly on time series data generated by sensor nodes. To remedy the deficiencies of previous methods, we propose an optimal online algorithm GDPLA for constructing a disconnected piecewise linear approximation representation of a time series which guarantees that the vertical distance between each real data point and the corresponding fit line is less than or equal to ε. GDPLA not only generates the minimum number of segments to approximate a time series with precision guarantee, but also only requires linear time O(n) bounded by a constant coefficient 6, where unit 1 denotes the time complexity of comparing the slopes of two lines. The low cost characteristic of our method makes it the popular choice for resource-constrained sensor networks. Extensive experiments on a real dataset have been conducted to demonstrate the superior compression performance of our approach.

In wireless sensor networks, congestion not only leads to packet loss, but also increases delays and lowers network throughput with a lot of energy wastage due to retransmissions. Therefore an effective solution should be proposed to mitigate congestion to increase energy efficiency and prolong the lifetime of network. Traditional solutions work in an open-loop fashion. In this paper, we propose a novel decentralized and weighted fairness guaranteed congestion control protocol (WFCC). WFCC introduces node weight to reflect the importance of each node, and uses the ratio of packet inter-arrival time to packet service time as congestion metric. Based on node weight and congestion metric, WFCC divides time axis into period sequences, and uses closed-loop control method to mitigate congestion by adjusting the incoming data rate of each node periodically. Importantly, WFCC introduces a weighted fairness metric and gives its lower bound for the first time, i.e., 1 - (10c/9)2where 0 <; c <; 0.2 is a constant. The results of simulation show that the weighted fairness of WFCC achieves 95% on the average, which is much better than the existing rate-based congestion control protocol. Moreover, WFCC achieves 50% and 19% gains in network throughput and weighted fairness on the average, compared with PCCP that is the state-of-art rate control based congestion control protocol.

In wireless sensor networks, congestion not only leads to buffer overflowing, but also increases delays and lowers network throughput with a lot of wasted energy due to retransmissions. Therefore an effective solution should be proposed to avoid congestion to increase energy efficiency and prolong the lifetime of network. Traditional solutions work in an open-loop fashion and hence fail to adapt to the change of network after network deployment. In this paper, we propose a novel decentralized rate control based congestion avoidance protocol (RCCAP), which is an adaptive rate control algorithm based on the combination of discrete proportional-integral-derivative (PID) control method and single neuron. The main idea of RCCAP is to ensure that the buffer length of each sensor node is as close as possible to an ideal length by a feedback control loop that adaptively calibrates the total rate of data packets entering in each sensor node periodically according to the information collected by the node recently. In addition, we prove the stability of system based on Lyapunov stability theory. We also prove that global weighted fairness metrics fM= 1 - O(M-2), where M is the number of periods, under the condition that the system is stable when time approaches infinity. Our approach has been evaluated on a real wireless sensor network testbed. Detailed experimental results demonstrate that the global weighted fairness of RCCAP achieves 99% on average, which is much stronger than the existing rate-based congestion control protocol. Moreover, RCCAP achieves 52% and 18% gains in network throughput and global weighted fairness over PCCP on average, respectively.

This paper presents a discriminative pruning method of n-gram language model for Chinese word segmentation. To reduce the size of the language model that is used in a Chinese word segmentation system, importance of each bigram is computed in terms of discriminative pruning criterion that is related to the performance loss caused by pruning the bigram. Then we propose a step-by-step growing algorithm to build the language model of desired size. Experimental results show that the discriminative pruning method leads to a much smaller model compared with the model pruned using the state-of-the-art method. At the same Chinese word segmentation F-measure, the number of bigrams in the model can be reduced by up to 90%. Correlation between language model perplexity and word segmentation performance is also discussed.

In Mandarin prosody synthesis by means of hierarchical prosodic structure, the naturalness of the output is reliant largely on the parsing of the prosodic structure. We propose a machine learning approach to improve prosodic structure parsing in cases where full syntax parsing is neglected due to considerations concerning practicality. The novel aspect of our approach is the new attribute in the input vector, which is named connective degree and calculated from the occurrence rate of the punctuation marks between Chinese characters by referring to a large text corpus. The results of experiments show that connective degree yield makes a remarkable contribution to parsing of hierarchical Mandarin prosodic structure.