ISSN 
2024 Vol. 53, No. 1
2024, 53(1): 1-7.
doi: 10.12178/1001-0548.2022384
Abstract:
In this paper, the electromagnetic (EM) scattering characteristics of inhomogeneous plasma-coated target are studied by volumetric shooting and bouncing rays (VSBR) method. The paper focuses on the propagation characteristics of EM waves in medium and the analysis methods of scattering. In order to improve the ray tracing efficiency in plasma medium, an iterative raytracing acceleration technique is proposed to solve the problem that the amount of transmitted and reflected rays increases rapidly. Some simulations show the volumetric shooting and bouncing rays method can calculate the scattering characteristics of target coated with inhomogeneous plasma more quickly and accurately compared with the traditional shooting and bouncing rays method. Furthermore, the influence of plasma sheath on the radar cross section of Apollo re-entry capsule is studied with the proposed method, and it is verified that the plasma has a certain reduction effect on the target radar cross section.
In this paper, the electromagnetic (EM) scattering characteristics of inhomogeneous plasma-coated target are studied by volumetric shooting and bouncing rays (VSBR) method. The paper focuses on the propagation characteristics of EM waves in medium and the analysis methods of scattering. In order to improve the ray tracing efficiency in plasma medium, an iterative raytracing acceleration technique is proposed to solve the problem that the amount of transmitted and reflected rays increases rapidly. Some simulations show the volumetric shooting and bouncing rays method can calculate the scattering characteristics of target coated with inhomogeneous plasma more quickly and accurately compared with the traditional shooting and bouncing rays method. Furthermore, the influence of plasma sheath on the radar cross section of Apollo re-entry capsule is studied with the proposed method, and it is verified that the plasma has a certain reduction effect on the target radar cross section.
2024, 53(1): 8-13.
doi: 10.12178/1001-0548.2023054
Abstract:
In this paper, in order to solve the problem of low efficiency of traditional wideband high-power amplifier, a new type of resistive reactance continuous B/J power amplifier mode is adopted to expand the output load impedance space of transistor with high efficiency. As a result, the drain stage output efficiency of broadband power amplifier is improved. Then, a balanced power amplifier (PA) based on 0.25 μm gate-length GaN HEMT is proposed. The PA combines LC matching network and Chebyshev impedance converter to realize broadband input and output impedance matching of GaN HEMT device, and utilizes a 3 dB Lange coupler to achieve broadband balanced power combination. Under the operating condition of continuous wave (CW), the proposed balanced PA has an output power greater than 100 W, drain efficiency greater than 45%, power gain greater than 9 dB, and anti-load mismatch better than 5:1 in the frequency range of 2 GHz to 6 GHz.
In this paper, in order to solve the problem of low efficiency of traditional wideband high-power amplifier, a new type of resistive reactance continuous B/J power amplifier mode is adopted to expand the output load impedance space of transistor with high efficiency. As a result, the drain stage output efficiency of broadband power amplifier is improved. Then, a balanced power amplifier (PA) based on 0.25 μm gate-length GaN HEMT is proposed. The PA combines LC matching network and Chebyshev impedance converter to realize broadband input and output impedance matching of GaN HEMT device, and utilizes a 3 dB Lange coupler to achieve broadband balanced power combination. Under the operating condition of continuous wave (CW), the proposed balanced PA has an output power greater than 100 W, drain efficiency greater than 45%, power gain greater than 9 dB, and anti-load mismatch better than 5:1 in the frequency range of 2 GHz to 6 GHz.
2024, 53(1): 14-20.
doi: 10.12178/1001-0548.2022389
Abstract:
Graphene-based electromagnetic wave absorbing materials have received great attention. In this paper, cavity cubic few-layer graphene was synthesized by solid-phase pyrolysis using sodium chloride as the template and phthalocyanine as the carbon source. In the preparation process, the sodium chloride cubic template was prepared by the antisolvent method and the homogeneous mixing of carbon source and template was achieved simultaneously. The experimental results show that the products obtained by pyrolysis at 700 °C can achieve an effective absorption bandwidth of 6.7 GHz with a filler loading of only 4 wt% and a coating matching thickness of 2.5 mm.
Graphene-based electromagnetic wave absorbing materials have received great attention. In this paper, cavity cubic few-layer graphene was synthesized by solid-phase pyrolysis using sodium chloride as the template and phthalocyanine as the carbon source. In the preparation process, the sodium chloride cubic template was prepared by the antisolvent method and the homogeneous mixing of carbon source and template was achieved simultaneously. The experimental results show that the products obtained by pyrolysis at 700 °C can achieve an effective absorption bandwidth of 6.7 GHz with a filler loading of only 4 wt% and a coating matching thickness of 2.5 mm.
2024, 53(1): 21-28.
doi: 10.12178/1001-0548.2023017
Abstract:
In this paper, a high-sensitivity integrated humidity and pressure sensor chip is designed. The pressure sensing unit is based on the silicon on insulator (SOI) and serpentine resistors structure. The sensitivity of the sensor at room temperature is 0.026 mV/kPa with the pressure of 3 kPa to 129 kPa, which is consistent with the finite element simulation. The thermal sensitivity shift reaches 0.004‰ FS/℃ and the thermal zero shift is 0.25% FS/℃. The humidity sensing unit adopts interdigital electrodes (IDT) and capacitive structure. The hydrophobic group introduced by fluorinated polyimide (PI) is used as the humidity sensitive film. The design of Mo resistance heating structure speeds up the dehumidification process of the sensor and shortens the dehumidification time by nearly 32%. In the humidity range of 10%RH to 90%RH, the sensitivity of the humidity sensor with fluorinated PI reaches 0.121 pF/%RH, slightly lower than the fluorine-free sensor. The humidity hysteresis of the sensor with fluorinated PI is reduced by 16% compared with the sensor without fluorinated PI. The capacitance-humidity curve is an exponential distribution, the correlation coefficient R2=0.996. The test results show that the humidity sensing unit and the pressure sensing unit have good independent performance.
In this paper, a high-sensitivity integrated humidity and pressure sensor chip is designed. The pressure sensing unit is based on the silicon on insulator (SOI) and serpentine resistors structure. The sensitivity of the sensor at room temperature is 0.026 mV/kPa with the pressure of 3 kPa to 129 kPa, which is consistent with the finite element simulation. The thermal sensitivity shift reaches 0.004‰ FS/℃ and the thermal zero shift is 0.25% FS/℃. The humidity sensing unit adopts interdigital electrodes (IDT) and capacitive structure. The hydrophobic group introduced by fluorinated polyimide (PI) is used as the humidity sensitive film. The design of Mo resistance heating structure speeds up the dehumidification process of the sensor and shortens the dehumidification time by nearly 32%. In the humidity range of 10%RH to 90%RH, the sensitivity of the humidity sensor with fluorinated PI reaches 0.121 pF/%RH, slightly lower than the fluorine-free sensor. The humidity hysteresis of the sensor with fluorinated PI is reduced by 16% compared with the sensor without fluorinated PI. The capacitance-humidity curve is an exponential distribution, the correlation coefficient R2=0.996. The test results show that the humidity sensing unit and the pressure sensing unit have good independent performance.
2024, 53(1): 29-39.
doi: 10.12178/1001-0548.2022376
Abstract:
Vehicle Edge Computing (VEC), combining mobile edge computing (MEC) with the Internet of Vehicles (IoV) technology, offloads vehicle tasks to the edge of the network to solve the problem of limited computing power at the vehicle terminal. In order to overcome the difficulty of on-board task scheduling due to the sudden increase in the number of tasks and provide a low-latency service environment, the vehicle tasks are divided into three types of main tasks by using improved Analytic Hierarchy Process (AHP) according to the dynamic correlation change criteria of the selected five feature parameters, and the joint modeling of resource allocation is carried out based on three kinds of offloading decisions. Then, the constraints of the modeling are eliminated by using scheduling algorithm and penalty function, and the obtained substitution value is taken as the input for the following deep learning algorithm. Finally, a distributed offloading network based on deep learning is proposed to effectively reduce the energy consumption and delay of VEC system. The simulation results show that the proposed offloading scheme is more stable than traditional deep learning offloading scheme and has better environmental adaptability with its less average task processing delay and energy consumption.
Vehicle Edge Computing (VEC), combining mobile edge computing (MEC) with the Internet of Vehicles (IoV) technology, offloads vehicle tasks to the edge of the network to solve the problem of limited computing power at the vehicle terminal. In order to overcome the difficulty of on-board task scheduling due to the sudden increase in the number of tasks and provide a low-latency service environment, the vehicle tasks are divided into three types of main tasks by using improved Analytic Hierarchy Process (AHP) according to the dynamic correlation change criteria of the selected five feature parameters, and the joint modeling of resource allocation is carried out based on three kinds of offloading decisions. Then, the constraints of the modeling are eliminated by using scheduling algorithm and penalty function, and the obtained substitution value is taken as the input for the following deep learning algorithm. Finally, a distributed offloading network based on deep learning is proposed to effectively reduce the energy consumption and delay of VEC system. The simulation results show that the proposed offloading scheme is more stable than traditional deep learning offloading scheme and has better environmental adaptability with its less average task processing delay and energy consumption.
2024, 53(1): 40-49.
doi: 10.12178/1001-0548.2022400
Abstract:
In view of the challenges that the modern electromagnetic spectrum equipment or terminals has such characteristics as network, agility, multi-function, adaptive, complex variety of types and working in complex environments, etc., in this paper, main research is focused on the structure of electromagnetic signal processing and analysis in complex electromagnetic environments. A new architecture is proposed, which combines artificial intelligence and is suitable for cognitive processing and analysis of electromagnetic signals in complex electromagnetic environments. In this framework, electromagnetic signal analysis and processing are divided into four levels: signal parameter estimation, signal type classification, radiation source identification, signal feature database and knowledge map construction[1 -3 ], The framework can be used to analyze and process closed electromagnetic signal set and open electromagnetic signal set; In addition, the identification of unknown emitter, the intelligent identification of cross-modal emitter and the individual identification of electromagnetic emitter are proposed in this framework. The verification test on the measured data set shows that these algorithms are effective and feasible in closed set and open set electromagnetic signal analysis.
In view of the challenges that the modern electromagnetic spectrum equipment or terminals has such characteristics as network, agility, multi-function, adaptive, complex variety of types and working in complex environments, etc., in this paper, main research is focused on the structure of electromagnetic signal processing and analysis in complex electromagnetic environments. A new architecture is proposed, which combines artificial intelligence and is suitable for cognitive processing and analysis of electromagnetic signals in complex electromagnetic environments. In this framework, electromagnetic signal analysis and processing are divided into four levels: signal parameter estimation, signal type classification, radiation source identification, signal feature database and knowledge map construction[
2024, 53(1): 50-59.
doi: 10.12178/1001-0548.2022285
Abstract:
In this paper, in the distributed Reconfigurable Intelligence Surface (RIS) assisted multi-user millimeter wave (mmWave) system, the deep reinforcement learning (DRL) theory is used to learn and adjust transmit beamforming matrix at the base station and phase shift matrix at the RIS, and jointly optimize the transmit beamforming matrix and phase shift matrix to maximize the weighted sum-rate. Specifically, in the discrete action space, we first design the power codebook and the phase codebook, and propose the Deep Q Network(DQN) algorithm to optimize the beamforming matrix and phase shift matrix; then, in the continuous action space, the Twin Delayed Deep Deterministic (TD3) policy gradient algorithm is used to optimize the beamforming matrix and phase shift matrix. The weighted sum-rates of the system in discrete action space and continuous action space with different number of codebook bits are compare through simulation. In addition, compared with the traditional convex optimization algorithm and the zero-forcing precoding with a random PBF algorithm, the sum-rate performance of DRL algorithm is significantly improved, and the sum-rate of the continuous TD3 algorithm exceeds the convex optimization algorithm by 23.89%, and the performance of the discrete DQN algorithm exceeds the traditional convex optimization algorithm when the number of codebook bits is 4.
In this paper, in the distributed Reconfigurable Intelligence Surface (RIS) assisted multi-user millimeter wave (mmWave) system, the deep reinforcement learning (DRL) theory is used to learn and adjust transmit beamforming matrix at the base station and phase shift matrix at the RIS, and jointly optimize the transmit beamforming matrix and phase shift matrix to maximize the weighted sum-rate. Specifically, in the discrete action space, we first design the power codebook and the phase codebook, and propose the Deep Q Network(DQN) algorithm to optimize the beamforming matrix and phase shift matrix; then, in the continuous action space, the Twin Delayed Deep Deterministic (TD3) policy gradient algorithm is used to optimize the beamforming matrix and phase shift matrix. The weighted sum-rates of the system in discrete action space and continuous action space with different number of codebook bits are compare through simulation. In addition, compared with the traditional convex optimization algorithm and the zero-forcing precoding with a random PBF algorithm, the sum-rate performance of DRL algorithm is significantly improved, and the sum-rate of the continuous TD3 algorithm exceeds the convex optimization algorithm by 23.89%, and the performance of the discrete DQN algorithm exceeds the traditional convex optimization algorithm when the number of codebook bits is 4.
2024, 53(1): 60-66.
doi: 10.12178/1001-0548.2022401
Abstract:
The key of gesture recognition based on skeleton sequence is how to fuse spatio-temporal information and extract discriminate features. This paper proposes a key point focusing module. Through the global context modeling and the convolution method not limited to the fixed form, the network can span multiple frames and irrelevant key points, adaptively aggregate key point information closely related to gesture actions in the global scope, and extract the spatio-temporal characteristics of gesture. Experiments on Chalearn2013 and SHREC datasets show that the accuracy of our proposed method can reach 94.88% and 95.23%, and the method outperforms state-of-the-art methods. In addition, the method has better stability in dealing with noisy data and dynamic gestures.
The key of gesture recognition based on skeleton sequence is how to fuse spatio-temporal information and extract discriminate features. This paper proposes a key point focusing module. Through the global context modeling and the convolution method not limited to the fixed form, the network can span multiple frames and irrelevant key points, adaptively aggregate key point information closely related to gesture actions in the global scope, and extract the spatio-temporal characteristics of gesture. Experiments on Chalearn2013 and SHREC datasets show that the accuracy of our proposed method can reach 94.88% and 95.23%, and the method outperforms state-of-the-art methods. In addition, the method has better stability in dealing with noisy data and dynamic gestures.
2024, 53(1): 67-75.
doi: 10.12178/1001-0548.2022307
Abstract:
Aiming at the problems of insufficient extraction of users’ main interest preferences in session-based recommender algorithms based on graph neural networks, a Session-Based Recommender Method Based on Interest Attention Network (SR-IAN) is proposed. First, the graph neural network is used to obtain the context transformation relationships between the items, and the graph embedding vectors of the items are obtained; Secondly, the graph embedding vector input into the interest attention network to extract the user’s main interest preferences; Then the graph embedding vectors of the items are weighted by the attention layer; Finally, the click probability values of the candidate items are obtained through the prediction layer and sorted. The proposed algorithm model was verified by experiments on three public datasets Diginetica, Retailrocket and Tmall, which showed an improvement of 0.942%, 1.183% and 2.977% compared with the baseline model on MRR@20. Besides, the time complexity of the model is reduced, which verifies the effectiveness and high efficiency of the proposed method.
Aiming at the problems of insufficient extraction of users’ main interest preferences in session-based recommender algorithms based on graph neural networks, a Session-Based Recommender Method Based on Interest Attention Network (SR-IAN) is proposed. First, the graph neural network is used to obtain the context transformation relationships between the items, and the graph embedding vectors of the items are obtained; Secondly, the graph embedding vector input into the interest attention network to extract the user’s main interest preferences; Then the graph embedding vectors of the items are weighted by the attention layer; Finally, the click probability values of the candidate items are obtained through the prediction layer and sorted. The proposed algorithm model was verified by experiments on three public datasets Diginetica, Retailrocket and Tmall, which showed an improvement of 0.942%, 1.183% and 2.977% compared with the baseline model on MRR@20. Besides, the time complexity of the model is reduced, which verifies the effectiveness and high efficiency of the proposed method.
2024, 53(1): 76-83.
doi: 10.12178/1001-0548.2022373
Abstract:
Automatic recognition of metal coating defects has significant value in realistic applications. As deep learning makes breakthrough in surface defect segmentation for a variety of materials, most of deep convolutional neural network segmentation models are trained in an end-to-end manner. However, it is difficult to exploit prior knowledge about metal coating defects in end-to-end deep learning and adapt to the variable scale of the defects and the limited training data. This paper proposes a defect segmentation algorithm based on prior knowledge about metal coating defects to unify U-Net, a deep learning segmentation model for automatic metal coating defect recognition. This anomaly segmentation is based on Hue channel distribution and edge response. Being trained in a knowledge driven manner, the model can exclude outliers from training data and effectively avoid over-fitting. On a metal coating defect image dataset with four defect types, including crack, blister, rusting and flaking, the proposed method achieves 81.24% mIoU, which is advantageous over end-to-end deep learning. The experiment shows that knowledge-driven model can boost the performance of deep learning models in metal coating defect segmentation.
Automatic recognition of metal coating defects has significant value in realistic applications. As deep learning makes breakthrough in surface defect segmentation for a variety of materials, most of deep convolutional neural network segmentation models are trained in an end-to-end manner. However, it is difficult to exploit prior knowledge about metal coating defects in end-to-end deep learning and adapt to the variable scale of the defects and the limited training data. This paper proposes a defect segmentation algorithm based on prior knowledge about metal coating defects to unify U-Net, a deep learning segmentation model for automatic metal coating defect recognition. This anomaly segmentation is based on Hue channel distribution and edge response. Being trained in a knowledge driven manner, the model can exclude outliers from training data and effectively avoid over-fitting. On a metal coating defect image dataset with four defect types, including crack, blister, rusting and flaking, the proposed method achieves 81.24% mIoU, which is advantageous over end-to-end deep learning. The experiment shows that knowledge-driven model can boost the performance of deep learning models in metal coating defect segmentation.
2024, 53(1): 84-91.
doi: 10.12178/1001-0548.2022380
Abstract:
The classification and recognition of emergency braking behavior based on electroencephalography (EEG) is a key issue in the development of human-centered intelligent assisted driving systems. In order to realize the classification and recognition of emergency braking and normal driving behaviors during driving, a feature representation method based on Phase Locking Value (PLV) was proposed to construct functional brain networks, the feature parameters of significant differences are determined via statistical analysis of the network feature parameters, and the spatial features of EEG were extracted through Log-Euclidean distance. Combined with machine learning algorithm, emergency braking and normal driving behavior are classified and recognized. The results show that the accuracy of emergency braking and normal driving for 17 participants is higher than 84%, and the highest accuracy rate reaches 95.7%, and the analysis of functional brain network results show that in the process of two driving behaviors, the interaction between brain regions involves the whole brain area, and in the emergency braking process, the interaction between brain regions mainly occurs in the frontal-central-temporal lobe area, which is consistent with the brain focusing more on judgment and decision-making under emergency braking. The results of this paper have certain reference value for understanding the dependence between the driver’s corresponding brain zones during driving, especially during emergency braking, and for developing intelligent assisted driving systems to identify emergency braking intentions in advance during driving.
The classification and recognition of emergency braking behavior based on electroencephalography (EEG) is a key issue in the development of human-centered intelligent assisted driving systems. In order to realize the classification and recognition of emergency braking and normal driving behaviors during driving, a feature representation method based on Phase Locking Value (PLV) was proposed to construct functional brain networks, the feature parameters of significant differences are determined via statistical analysis of the network feature parameters, and the spatial features of EEG were extracted through Log-Euclidean distance. Combined with machine learning algorithm, emergency braking and normal driving behavior are classified and recognized. The results show that the accuracy of emergency braking and normal driving for 17 participants is higher than 84%, and the highest accuracy rate reaches 95.7%, and the analysis of functional brain network results show that in the process of two driving behaviors, the interaction between brain regions involves the whole brain area, and in the emergency braking process, the interaction between brain regions mainly occurs in the frontal-central-temporal lobe area, which is consistent with the brain focusing more on judgment and decision-making under emergency braking. The results of this paper have certain reference value for understanding the dependence between the driver’s corresponding brain zones during driving, especially during emergency braking, and for developing intelligent assisted driving systems to identify emergency braking intentions in advance during driving.
2024, 53(1): 92-101.
doi: 10.12178/1001-0548.2022377
Abstract:
Spatial information network is a kind of network with high-speed and periodically running nodes. With the increasing number of low Earth orbit satellites, the topology of spatial information networks is highly dynamic, and the problem of network topology survivability optimization will be of great research significance. Considering the visibility of satellite networking, the connectivity of satellite nodes, and the number of communication links in the entire network, a network topology optimization model satisfying multiple constraints is constructed to minimize the end-to-end delay among satellite nodes in the network, and then an optimized simulated annealing algorithm is proposed to solve the model. In the simulated annealing process, the network flow algorithm is innovatively proposed to solve the neighborhood. The experimental results show that the simulated annealing hybrid neighborhood algorithm is significantly better than the simulated annealing random neighborhood algorithm.
Spatial information network is a kind of network with high-speed and periodically running nodes. With the increasing number of low Earth orbit satellites, the topology of spatial information networks is highly dynamic, and the problem of network topology survivability optimization will be of great research significance. Considering the visibility of satellite networking, the connectivity of satellite nodes, and the number of communication links in the entire network, a network topology optimization model satisfying multiple constraints is constructed to minimize the end-to-end delay among satellite nodes in the network, and then an optimized simulated annealing algorithm is proposed to solve the model. In the simulated annealing process, the network flow algorithm is innovatively proposed to solve the neighborhood. The experimental results show that the simulated annealing hybrid neighborhood algorithm is significantly better than the simulated annealing random neighborhood algorithm.
2024, 53(1): 102-109.
doi: 10.12178/1001-0548.2022128
Abstract:
In order to solve overfitting of modeling in noisy circumstance, a novel cost function with corresponding training algorithm is proposed for wavelet networks based on sampling theory. Since such an algorithm can use sample distributions and errors respectively to train input and output weights, learning efficiencies of wavelet networks are improved greatly. The theories and experiments show that this novel cost function can ensure generalizations of wavelet networks. Simultaneously, the new algorithm can converge globally and is robust to noise varying.
In order to solve overfitting of modeling in noisy circumstance, a novel cost function with corresponding training algorithm is proposed for wavelet networks based on sampling theory. Since such an algorithm can use sample distributions and errors respectively to train input and output weights, learning efficiencies of wavelet networks are improved greatly. The theories and experiments show that this novel cost function can ensure generalizations of wavelet networks. Simultaneously, the new algorithm can converge globally and is robust to noise varying.
2024, 53(1): 110-118.
doi: 10.12178/1001-0548.2022340
Abstract:
A Multi-head Quantum Self-Attention Predict Network (MQSAPN) is designed in hybrid manner, which could be used in time-series forecasting. MQSAPN comprises two components, one is the Multi-head Quantum Self-Attention (MQSA) model, and the other is the predicting Variational Quantum Circuits (pVQC). When fed with sequential inputs, the MQSA firstly computes the key, query, and value vectors corresponding to all time steps through the variational circuits, and then according to exist studies, the attention is estimated via Gaussian function. With residual link on input and multi-head features, the output of MQSA were pushed to pVQC part, which was encoded into quantum circuit again, and the prediction would be ultimately calculated out by measurements on observables. The prediction results of MQSAPN numerical experiments on atmospheric variables indicate the effectiveness of quantum self-attention, by comparison with the results of a data-reuploading VQC model with almost same amount of parameters. The accuracy of predicting is close to classic multi-head transformer model and LSTM net. To be noted, as input time window extends or the more features are adopted, the number of parameters of pVQC will also increases correspondingly, which makes the pVQC part become the bottleneck of the whole model due to ‘barren plateau’ problems during training process.
A Multi-head Quantum Self-Attention Predict Network (MQSAPN) is designed in hybrid manner, which could be used in time-series forecasting. MQSAPN comprises two components, one is the Multi-head Quantum Self-Attention (MQSA) model, and the other is the predicting Variational Quantum Circuits (pVQC). When fed with sequential inputs, the MQSA firstly computes the key, query, and value vectors corresponding to all time steps through the variational circuits, and then according to exist studies, the attention is estimated via Gaussian function. With residual link on input and multi-head features, the output of MQSA were pushed to pVQC part, which was encoded into quantum circuit again, and the prediction would be ultimately calculated out by measurements on observables. The prediction results of MQSAPN numerical experiments on atmospheric variables indicate the effectiveness of quantum self-attention, by comparison with the results of a data-reuploading VQC model with almost same amount of parameters. The accuracy of predicting is close to classic multi-head transformer model and LSTM net. To be noted, as input time window extends or the more features are adopted, the number of parameters of pVQC will also increases correspondingly, which makes the pVQC part become the bottleneck of the whole model due to ‘barren plateau’ problems during training process.
2024, 53(1): 129-143.
doi: 10.12178/1001-0548.2023022
Abstract:
As one of the keys of economic growth, the development of cities is driven by city’s own factors and the interaction among cities. In this paper, from the perspective of economic connectivity and based on the hypothesis of the gravity model, an indicator f naming “the intensity of the external influence field” is constructed as a metric of the influence of each city from the other cities. For each city, the total value F of its f, as well as the Zipf’s exponent α of the distribution of f, is strongly and positively correlated to the city’s economy size, and more than 50% on the variation of city’s economy can be described by the changes of F and α, revealing the strong impact of city’s external economic environment on the development of city’s economy. Furthermore, from the spatial autocorrelation analysis of these external influence field indicators, we observed the remarkable heterogeneity on the trend of local spatial autocorrelations of cities at different levels, namely, the high-level cities tend to strengthen their local spatial autocorrelations but the lower-level cities tend to weaken the local spatial autocorrelations. By comparing the variation trend of local spatial autocorrelations of cities in different administrative regions, it is found that administrative regions with small population and small economic size tend to concentrate on the development of a few cities. These findings indicate that, as the indicators without any direct relationship to city’s own economy, F and α can efficiently separate the external economic influences and other influences on city to dig out the hidden patterns in city development, which is helpful in mining of the relationship between city development and the interaction of cities.
As one of the keys of economic growth, the development of cities is driven by city’s own factors and the interaction among cities. In this paper, from the perspective of economic connectivity and based on the hypothesis of the gravity model, an indicator f naming “the intensity of the external influence field” is constructed as a metric of the influence of each city from the other cities. For each city, the total value F of its f, as well as the Zipf’s exponent α of the distribution of f, is strongly and positively correlated to the city’s economy size, and more than 50% on the variation of city’s economy can be described by the changes of F and α, revealing the strong impact of city’s external economic environment on the development of city’s economy. Furthermore, from the spatial autocorrelation analysis of these external influence field indicators, we observed the remarkable heterogeneity on the trend of local spatial autocorrelations of cities at different levels, namely, the high-level cities tend to strengthen their local spatial autocorrelations but the lower-level cities tend to weaken the local spatial autocorrelations. By comparing the variation trend of local spatial autocorrelations of cities in different administrative regions, it is found that administrative regions with small population and small economic size tend to concentrate on the development of a few cities. These findings indicate that, as the indicators without any direct relationship to city’s own economy, F and α can efficiently separate the external economic influences and other influences on city to dig out the hidden patterns in city development, which is helpful in mining of the relationship between city development and the interaction of cities.
2024, 53(1): 144-148.
doi: 10.12178/1001-0548.2022346
Abstract:
With the help of the space-efficient quantum Karatsuba algorithm for multiplication, the quantum implementation for the\begin{document}$ 8 \times 8 $\end{document} ![]()
![]()
\begin{document}$ 8 \times 8 $\end{document} ![]()
![]()
With the help of the space-efficient quantum Karatsuba algorithm for multiplication, the quantum implementation for the
2024, 53(1): 149-154.
doi: 10.12178/1001-0548.2022344
Abstract:
In today’s era of big data, it is difficult for traditional naive Bayesian algorithms to efficiently and accurately deal with the complexity and uncertainty of big data. Based on the traditional Naive Bayes algorithm, this paper proposes an efficient and accurate quantum fuzzy Bayesian classification algorithm. First, the “fuzzy set theory + naive Bayes theory” is cross-integrated, the fuzzy prior probability and fuzzy conditional probability are defined, and the naive Bayes is extended to fuzzy naive Bayes to construct a fuzzy Bayes model; Secondly, a quantum fuzzy naive Bayesian classification algorithm is investigated and implemented by quantizing fuzzy data sets (encoding to quantum states) and designing quantum circuits. Finally, the algorithm proposed in this paper is applied to the iris dataset. Simulation experiments show that the proposed classification algorithm has higher classification efficiency and accuracy compared with the traditional Naive Bayesian classification algorithm.
In today’s era of big data, it is difficult for traditional naive Bayesian algorithms to efficiently and accurately deal with the complexity and uncertainty of big data. Based on the traditional Naive Bayes algorithm, this paper proposes an efficient and accurate quantum fuzzy Bayesian classification algorithm. First, the “fuzzy set theory + naive Bayes theory” is cross-integrated, the fuzzy prior probability and fuzzy conditional probability are defined, and the naive Bayes is extended to fuzzy naive Bayes to construct a fuzzy Bayes model; Secondly, a quantum fuzzy naive Bayesian classification algorithm is investigated and implemented by quantizing fuzzy data sets (encoding to quantum states) and designing quantum circuits. Finally, the algorithm proposed in this paper is applied to the iris dataset. Simulation experiments show that the proposed classification algorithm has higher classification efficiency and accuracy compared with the traditional Naive Bayesian classification algorithm.
2024, 53(1): 155-160.
doi: 10.12178/1001-0548.2022233
Abstract:
One of the key problems in reversible logic synthesis is optimizing the reversible circuits, and the focus of research is on how to decompose advanced reversible gates into basic reversible gates more efficiently. To improve the decomposition and optimization efficiency of Multiple Control Target (MCT) gates, an optimal decomposition method of MCT gates is proposed in the paper, along with an MCT decomposition template which correctness is verified. Based on this template, the corresponding decomposition and optimization algorithm is given. Using the optimal decomposition template, the algorithm classifies the Toffoli circuits decomposed by MCT gates and decomposes them into NCV circuits. The time complexity of the algorithm is O(m), which is better than O(m2) for the conventional algorithm. Experiments on MCT gates with benchmark reversible circuits for control bits m∈{3,10} show the effectiveness of the algorithm’s optimization and decomposition.
One of the key problems in reversible logic synthesis is optimizing the reversible circuits, and the focus of research is on how to decompose advanced reversible gates into basic reversible gates more efficiently. To improve the decomposition and optimization efficiency of Multiple Control Target (MCT) gates, an optimal decomposition method of MCT gates is proposed in the paper, along with an MCT decomposition template which correctness is verified. Based on this template, the corresponding decomposition and optimization algorithm is given. Using the optimal decomposition template, the algorithm classifies the Toffoli circuits decomposed by MCT gates and decomposes them into NCV circuits. The time complexity of the algorithm is O(m), which is better than O(m2) for the conventional algorithm. Experiments on MCT gates with benchmark reversible circuits for control bits m∈{3,10} show the effectiveness of the algorithm’s optimization and decomposition.
