Search

Object extraction provides the basis for subsequent tasks such as object recognition, and its importance is self-evident. For this purpose, this paper proposes an object extraction algorithm based on saliency prior information. Firstly, the SIFT operator and the oriented edge forest method are used to extract the saliency points and the saliency edges respectively. Then construct a simple small convolutional neural network for the saliency fusion task, and fuse the saliency point and the saliency edge to obtain the saliency fusion map. Then the fusion map is added to the object extraction network structure as a priori information. In this paper, the residual network is used for the object extraction task, and finally the high-quality object extraction work is realized. The algorithm uses IOU parameters and Precision as the quantitative evaluation index, and uses qualitative analysis to comprehensively analyse the experimental results. Experiments show that the proposed algorithm has good results in both image quality and visual, and has good robustness.

As an important part of electronic products, analog circuits are widely used in various fields. However, the fault prediction technology of analog circuit is still in its infancy because of its nonlinear characteristics such as nonlinearity and tolerance. According to the characteristics of small sample nonlinear data in analog circuits, the farther the prediction time distance of IAALO-SVM algorithm is from the initial training sample, the greater the prediction error is. Therefore, this article through the use of analog circuit fault parameters offline database, introduced the choice of time series similarity measure method is similar to database, according to the size of the similarity for failure prediction result, combining the measured data and offline data is put forward based on IAALO - SVM and similarity measure the failure prediction of new scheme to predict the residual service life of main amplifier circuit. The example shows that the new scheme can obtain high prediction accuracy in a long time and has practical application value.

A stair-like generalized predictive control (GPC)strategy based on multiple models switching for a 4-wheel-drive with 4 hub motors is presented in this paper. It ensures that the vehicle can change its motion softly and meet real-time optimal control requirements of systems with jumping parameters. The strong nonlinear dynamics and interaction characteristics of electric vehicle result in large model uncertainties and the conventional robust or parameter adaptive control methods can hardly ensure the stability and accuracy of vehicle motion without delay. A conventional adaptive vehicle model, an assigned initial value vehicle model and multiple fixed models are established to identify the vehicle dynamic characteristic in parallel. Multiple fixed vehicle models can improve the transient performance. Moreover, we ensure the stability of vehicle systems by the conventional adaptive vehicle and further enhance vehicle transient performance by the assigned initial value adaptive vehicle model. The best sub-model is selected as global model according to the switching index, and a stair-like generalized predictive controller is designed for this model in electric vehicle. Due to the GPC method employed in the design, the computation complexity and cost are greatly reduced. As a result, the vehicle delivers superior performance in hazardous driving conditions.

In recent years, the consensus problem is well recognized as a fundamental problem in the cooperation control of multi-agent systems, we study the noise problem of the linear consensus protocol in networked multi-agent systems, and point out that the noise in discrete linear consensus protocol is uncontrollable. A strategy using noise suppression operator to control the noise is put forward, and inference about the reasonable value range of the noise suppression operator is vigorously proved, and sufficient conditions for noise controllable are further presented.

This paper presents a digital hysteresis control method, with which the rectifier can work at a constant switching frequency and the dynamic response of current is fast. And also, the control doesn't need single-phase lock loop. Simulation results validate the proposed method.

This paper presents a topology of bi-directional Z-source converter based on improved Z-source inverter, which can operate with small inductor at wide-range load and realize low capacity voltage stress. Meanwhile, it inherits merits of limitation of inrush-current at startup and the same voltage gain which improved Z-source inverter has. By controlling the additional switch properly, the proposed Z-source converter can eliminate abnormal operation state and work on synchronous rectification mode to achieve dual power flow. The operating principle and parameters design of the proposed converter have been analyzed. Simulation results are presented to verify the validity of analysis for the proposed Z-source converter.

While the society of PRC has drawn near to an aging society, more and more people find it difficult to have enough time and energy to look after their aged parents because of the increasingly heavy living burden and social pressures so that rising concerns are made over the health and security of the aged Chinese throughout the whole society. With the booming of so-called "smart cities" in advanced countries, in PRC, experts tend to realize the necessity to construct more "smart senior citizens' communities", which, based on intelligent theory and information technology, proves to be more convenient, more comfortable, more smart, more sustainable and more suitable to the trend of social development as a brand new "green" practice of modern communities to support the aged. In this thesis, important concepts corresponding to IT-based smart senior citizens' communities are introduced, a new type of management system for smart senior citizens' community service is constructed and the important social significance of constructing the IT-based smart senior citizens' communities is explored.

This paper presents some explorations and studies on image retrieval. Firstly, RGB color space is converted to HSV color space for feature extraction. Then, the texture features are obtained by using wavelet, which are combined with some color features based on wavelet transform. Finally, the multi-features generated by Gaussian Mixture Model (GMM) are employed to an image retrieval algorithm. The experimental results on an image database show the effectiveness and competitive performance of the GMM-based image retrieval algorithm.

At present, to strengthen visualization level of power station facilities plays an important role in improving information technology standard of power industry. Recent visualization systems had exposed the deficient of facility status and environment around cannot be expressed by simple geometry. In order to solve this problem, a solution of image-base 3-D panoramic technology, which can support facility management of power stations perfectly, is discussed in this paper. In this solution, 3-D panoramic technology integrated in 2-D GIS, 3-D virtual scene based on GIS was constructed, which makes visualized management of substation equipments more convenient.

In recent years, millimeter-wavelength radar plays an important role in radar remote sensing field. In this article, application of this technique is introduced at first. Then the observation of non-precipitating and weak precipitation clouds by this type of radar is compared with those obtained by X-Band weather radar. The comparison shows that millimeter-wavelength cloud radar is a better way in both the macroscopic and microscopic investigation of clouds. However, this technique undergoes serious rain attenuation which is also introduced in the article.

Proposing the optimal AVC coordinating parameters and comparing effects of different planning grid in an scientific manner is one of the functional characteristics of smart AVC. In this paper Auxiliary Optimization Parameter(AOP) system is developed in Hebei grid for this purpose. With the AOP system, the traditional AVC system can be developed to smart AVC system‥ The data of the system are taken from EMS. The state estimation is carried out periodically after obtaining real-time data. Taking state estimation results as the basic power flow and simulating AVC operations, the voltage variations and line loss analysis can be calculated for comparision. The structure, hardware configruation and data exchange of the system are illustrated. Finally, several analysis tools of the AOP system are introduced. With the system, it can be evaluated that effects of reducing line loss benefited from the AVC operations. It can propose coordinating priority of generators or on-load regulating transformers. Also it can provide the optimal AVC coordinating parameters and the optimal grid plans, which lead to better voltage control and better layout of reactive power compensation equipments.

Heavy ion irradiation facility dedicated for single event effect ground simulation testing has been developed on Beijing HI-13 tandem accelerator. Two dimensional magnetic scanning method was used to produce a large and uniformly distributed beam. Non uniformity of beam distribution is less than 10% over an area of 100mm×100mm. In order to enhance the output energy of tandem accelerator highly stripped charge state beam has been successfully developed by so-called “magnetic rigidity simulation” technique. Gold ion with energy of 360 MeV and stripped probability of less than 0.01% has been obtained and successfully used in experiments. The beam purity is about 80%. The contamination is caused by ions of other charge state combinations due to close magnetic rigidities.

In this paper, the fragmentation of fuzzy numbers is defined as new sort indicators Based on the Centroid and fragmentation of fuzzy numbers, paper proposes a new ranking method of fuzzy numbers which can overcome the shortcomings of some ranking method to some degree. At last, paper illustrates the practicability of the proposed method by the analysis of examples.

Three-dimensional single-input-multiple-output synthetic aperture radar (SIMO-SAR) imaging method is proposed in this paper based on Capon adaptive beam-forming technology, and we have designed the downward-looking geometry architecture and signal echo model of the imaging system. Imaging algorithm is derived from the model and simulated. The advantages of three-dimensional imaging with SIMO-SAR are shown by the simulation results that it can get the height information of the targets besides the azimuth and range information, thus the targets can be located accurately and viewed in a stereogram. Furthermore, the algorithm can give better three-dimensional resolution.

Target recognition technology is a difficulty in the research field of household service robot. In this paper, connected domains are picked out on the basis of color image segmentation, denoising and graying. The algorithm has improved through shape moment invariants and weighting Euclidean distance. The experiment results show that this algorithm is not sensitive to the light conditions and the perspective of camera. Furthermore, this algorithm has big advantage to distinguish objects with similar color and it is efficient, accurate and robust.

A time-frequency synchronization method using shift-orthogonal constant amplitude zero auto correlation (CAZAC) sequences constructing the synchronization preamble is designed for multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) shortwave systems. The timing synchronization and integral frequency offset is accomplished by using the good correlation properties of training sequence. The simulations results show that the new algorithm works well at low SNR over HF channel. Finally, it provides higher timing synchronization accuracy even in distributed MIMO-OFDM systems with different time offsets.

A method of data connection by a digital simulator and EMS system is provided in the paper, and integration with off-line data and EMS real-time data can be operated automatically. On present situation, almost all of analyses of electric power system adopt off-line data which is much more onerous input in computer by hand, and the result can not accurately describe the actual condition in operation. With the use of standard IEC61970 in our country, EMS system has the uniform data-interface. Applying on-line data to playback of system failure, the contrast of oscilloscope data and simulation data demonstrates that on-line data can not only describe the steady characteristic, but also accurately describe transient characteristic of power system. Therefore, researching on EMS data-interface will greatly improve the accuracy of simulation veracity and it will has great significance.

This paper has a further exploration and study of visual feature extraction. According to the HSV (Hue, Saturation, Value) color space, the work of color feature extraction is finished, the process is as follows: quantifying the color space in non-equal intervals, constructing one dimension feature vector and representing the color feature by cumulative histogram. Similarly, the work of texture feature extraction is obtained by using gray-level co-occurrence matrix (GLCM) or color co-occurrence matrix (CCM). Through the quantification of HSV color space, we combine color features and GLCM as well as CCM separately. Depending on the former, image retrieval based on multi-feature fusion is achieved by using normalized Euclidean distance classifier. Through the image retrieval experiment, indicate that the use of color features and texture based on CCM has obvious advantage.