最新刊期
卷 48 ， 期 8 ， 2020
- 摘要：Dynamic vision sensor (DVS) shows significant advantages on low computational latency, low memory usage and high dynamic range by utilizing the event-driven principle to extract features from moving objects. Current research shows that DVS-based neural networks improve object detection speed obviously. However, the sample sets required by such neural networks mainly rely on specific DVS cameras while lacking efficient generation methods for the sample sets. It limits the application and development of those neural networks. According to the principle of DVS, this paper presents a DVS sample set modeling method based on frame images, in which the sample set can be generated by encoding and normalizing the address-event (AE) data after being trigged by dynamic differential comparisons and logical judgments. The experimental results for modeling the MNIST and CIFAR-10 sample sets show that, the sample set modeled by the proposed method is basically matched with the real DVS cameras. Compared with traditional frame image sample sets, this method can significantly reduce the memory usage. The sample set generated by the proposed modeling method has also been verified by training and testing a typical convolutional neural network.关键词：dynamic vision sensor;sample set modeling;event-driven;memory optimization76|132|1更新时间：2025-07-16
- 摘要：This paper proposes a group behavior recognition framework with "key persons" as the core and Gated Fusing Unit (GFU) for feature fusion. Its aim is to solve the following two problems: 1) Group behavior information is redundant, focusing on key person behavior characteristics, ignoring the influence of unrelated personels on group behavior. 2)The internal interaction relationship is complex within group, GFU is used to effectively model interaction feature centered around the key characters and it is temporally evolved into the group characteristics via LSTM processing. Finally, the group behavior category is classified with Softmax. The algorithm achieves an average recognition rate of 86.7% on the volleyball dataset.关键词：group behavior recognition;Key person modeling;interaction feature modeling;gated fusion unit75|62|3更新时间：2025-07-16
- 摘要：Aiming at the problem that the target is easy to lose in the complex scene such as low resolution, occlusion, the interference of similar objects, this paper proposes a visual tracking algorithm based on second-order pooling network. Most of the existing methods use the first-order pooling network, which makes the difference between similar targets insufficient. In this paper, based on the VGG16 network structure, the last first-order pooling layer is replaced by the second-order covariance pooling layer, and then the network is retrained on ImageNet and CUB200-2011 image data sets. In order to reduce the computational burden, only the fourth convolution feature of the pre-training network is extracted as the appearance representation of the target. Finally, the extracted features are combined with the existing correlation filtering algorithm. The experimental results show that the algorithm achieves excellent performance in tracking accuracy and success rate.关键词：visual tracking;second-order pooling network;deep features;correlation filter39|60|4更新时间：2025-07-16
- 摘要：Sparse cross array that operating in the 90~94GHz frequency band was designed for the handheld millimeter-wave human body security inspection. The time-domain correlation algorithm and the modified range migration algorithm (RMA) were chosen as image reconstruction methods. Due to the existence of spherical wave expansion approximation and complex interpolation processes in the deriving of the modified RMA algorithm, a high-precision FFT-based imaging algorithm without spherical wave expansion and complex interpolations was proposed. The electromagnetic simulation software was used to build target model obtaining raw echo data, the resolution test and noise robustness verification experiments were carried out. The azimuth resolution can achieve 5mm, which meets the system design requirements, and verifies the effectiveness of the proposed algorithm. The computational efficiency of the proposed algorithm is better than that of the time-domain correlation algorithm and the noise robustness is better than the modified RMA algorithm. It can be concluded that the proposed algorithm is a better choice in real-time imaging scenarios of the handheld millimeter-wave human body security inspection.关键词：sparse planner array;handheld human body security inspection;the modified RMA algorithm;FFT-based algorithm;noise robustness43|38|1更新时间：2025-07-16
- 摘要：In order to improve the spectral efficiency and symbol error rate (SER) of three-dimensional orthogonal frequency division multiplexing (OFDM) system, a simple method for designing four dimensional signal constellations and corresponding OFDM system model based on designed constellations are proposed in this paper. In proposed system, the incoming information bits are mapped through four dimensional signal mapper to each subcarrier of OFDM signals which is represented by two complex number. OFDM signal in the frequency domain is modulated into the time domain using the two-dimensional inverse discrete Fourier transform (IDFT). In addition, both designed 8 points and 16 points four dimensional constellations structure have a certain regular. Simulation results show that the proposed system has better SER performance compared with the corresponding three-dimensional OFDM systems, and the spectral efficiency can also be increased by one third.关键词：four-dimensional signal constellation;orthogonal frequency division multiplexing(OFDM);two-dimensional inverse discrete Fourier transform(IDFT);spectral efficiency;symbol error rate(SER)41|38|2更新时间：2025-07-16
- 摘要：Aiming at the problem that traditional search method in visual tracking is not efficient and the global optimization is hard to be solved, as the global optimization ability of quantum genetic algorithm, we put forward a visual tracking method by using quantum genetic algorithm as the search strategy. In the framework of quantum genetic algorithm, regard the pixel positions as the individuals in the population, and extract the color histogram as characteristics. The individual fitness are calculated by taking similarity measure as the objective function. We find out the maximum similarity and output its homologous position, to finish the tracking. The experimental results show that the algorithm has obvious advantages in fast speed, occlusion and non-rigid deformation, and the tracking speed is fast.关键词：visual tracking;quantum genetic algorithm;color feature53|54|2更新时间：2025-07-16
- 摘要：Conventional Android malware detection method can easily be evaded. In this study, we propose a detection method of Android malicious code based on short-term memory network (LSTM), which makes malware more difficult to evade from detection. In this method, a program analysis framework that combines static and dynamic analysis is proposed at first to get the permission information, protection information and behavior information. Secondly, entrenched features such as ability features and behavior features are extracted from the information that provided by the program analysis framework. With the entrenched features, we design a malware detection method based on LSTM model to distinguish benign applications from the malicious ones. Experimental results demonstrate that our approach is more effective and robust in Android malware detection than the state-of-the-art methods.关键词：android malware;static analysis;dynamic analysis;deep learning;LSTM116|62|4更新时间：2025-07-16
- 摘要：Existing saliency detection methods can not suppress the background effectively and detect the salient object accurately in complex background, a method of superpixel content-aware priors based multi-scale Bayesian saliency detection is proposed. Firstly, the image containing object is segmented into multi-scale superpixel maps, then the content-aware priors of contrast priors, center position priors, and boundary connected background priors are introduced on each scale to calculate the salient object values on a single scale; Secondly, the content-aware priors values of the various scales generate a rough saliency map; Thirdly, the rough saliency map value is used as the prior probability, and the likelihood is calculated according to the color histogram and the convex hull center, using the multi-scale Bayesian model to obtain the final salient object; Finally, three public data sets, five evaluation indicators, and seven existing methods are used for comparative experiments. The experiments show that the method has better performance in the detection of salient objects.关键词：saliency;multi-scale;content-aware prior;boundary connectivity;Bayesian model37|46|1更新时间：2025-07-16
- 摘要：A generic ambiguous optimistic fair exchange (AOFE) scheme, a variant of OFE, is proposed by Huang et al. The AOFE scheme prevents signature verifiers from convincing anybody about the authorship of a partial signature generated by the signer. However, the AOFE scheme cannot be directly applied to an actual user environment. A generic AOFE scheme and an instantiation of the generic construction in an identity-based user environment were proposed in this paper. In the generic construction of identity-based AOFE (ID-AOFE), the tag-based encryption and zero-knowledge proof algorithms in Huang et al.'s AOFE was removed and the non-interactive witness indistinguishable proof algorithms extracting the hided witness via keys was employed. Furthermore, we summarized and redefined the security of the ID-AOFE scheme. Then, an ID-AOFE security model was defined based on the Huang et al.'s AOFE security model and the selective identity security model. Under the selective identity security model of ID-AOFE, the fairness of our scheme is reduced to the securities of several classical cryptographic primitives. In addition, this paper discussed the message interaction model of the ID-AOFE scheme, and analyzed the mechanism of dispute resolution.关键词：optimistic fair exchange;signature exchange;identity-based cryptography;non-interactive witness indistinguishable proof;fairness;signer ambiguity;standard model57|31|1更新时间：2025-07-16
- 摘要：Aiming at the impact of unstructured and rotational variability of three-dimensional information in point cloud on prediction results, a feature-supervised three-dimensional information encoding and decoding convolution deep learning network is proposed. The network is composed of feature monitoring coding and decoding modules in series. In the first part of the module, a three-dimensional convolution module is used in the form of hourglass structure to realize the coding and decoding of the feature map. In the second part, the residual blocks of different parameters are connected in parallel to realize the monitoring and fusion of feature maps. Feature monitored coding and decoding modules can build networks with different depths in series according to the size of data sets. At the same time, according to the data resolution, modules parameters can be set to realize feature learning from rough to fine, and ultimately obtain the best network. The experiment of ITOP database shows that the network achieves the end-to-end deep learning of three-dimensional information, significantly improves the system performance and has higher precision accuracy.关键词：computer vision;depth image;pose estimation;deep learning;3D-CNN17|32|3更新时间：2025-07-16
- 摘要：In an electric power distribution grid using wireless communication access, IDS is used to decide system the intrusive event through analyzing the network transmission data. In this paper, to improve the detection accuracy, a deep learning theory is studied for the IDS in the wireless communication network of a power distribution grid. The proposed Recurrent Neural Network (RNN) model is composed of Gated Recurrent Unit (GRU), Multi-Layer Perceptron (MLP) and Softmax. The experimental results on the attack testing baseline demonstrate the effectiveness of the IDS defenses. In the KDD99 test data, its negative error rate and accuracy are with 0.06% and 96.43%,and in the NSL-KDD test data, those statistics are 0.86% with 99.33%, respectively.关键词：electric distribution network;wireless network;intrusion detection;deep learning;recurrent neural network(RNN)374|121|5更新时间：2025-07-16
- 摘要：The existing outsourcing algorithms for modular exponentiations has some problems such as low security, low probability of validation and wrong calculation results. a CP-ABE (Ciphertext-Policy Attribute-Based Encryption) scheme with verifiable outsourced encryption and decryption is proposed with the help of the improved algorithm for secure outsourcing of modular exponentiations. The scheme outsources attribute-related key subitems, outsources half of the computing tasks of shared ciphertext subitems, and verifies all the outsourced results. Theoretical and experimental analysis show that compared with the existing related schemes, both in key generation and encryption, the computing overhead of both authority centers and user clients of the proposed scheme has been significantly reduced. Security analysis shows that the proposed scheme can defend against chosen plaintext attacks.关键词：ciphertext sharing;outsourced encryption;outsourced decryption;attribute-based encryption41|41|2更新时间：2025-07-16
- 摘要：The security of the wireless access network of electric power grids is critical for power grid productions. However, the control data anomalies are difficult to be detected in a fast and effective manner, due to the high dimension of the control protocol data in IEC 60870-5-104 protocol, as well as the dynamics on the quality of wireless channels. To this end, this paper proposes an anomaly traffic detector (ATD) for the wireless network of power girds based on multi-resolution low rank (MRLR) model. Firstly, the ATD is designed with the MRLR for the protocol, to regularize and reduce the security feature dimensions. Secondly, it utilizes the improved recursive feature selection and focused classification algorithms for accurate data anomaly detection. The results demonstrate the accuracy for the classification on data anomalies, and the performance for the dimensionality reduction.关键词：wireless access networks for power grids;IEC 60870-5-104;multi-resolution low rank model;feature selection;multi-resolution analytics;anomaly traffic detection207|109|5更新时间：2025-07-16
- 摘要：In the complex application environment of the internet of things, tag collision in RFID (radio frequency identification) technology has become one of the core research hotspots in recent years, which is the main reason for the high delay of tag identification and communication cost. To solve the problems above, a MRCT (multi-reader cooperative tree-based algorithm) is proposed in this paper. In the first stage of the algorithm, readers delete part of the invalid nodes according to the advantage of manchester code which can identify the collision bits of tags; In the second stage, readers share the prefix and the current broadcast result in each round of broadcast. The tree structure of the readers is continuously optimized by the information readers shared. The optimization strategy is to allocate a queue of nodes’ priority to be broadcasted for each reader. The probability of tags to be broadcasted is calculated by the shared information. The priority of nodes to be broadcasted is adjusted according to its probability, and the invalid nodes are deleted continuously. In the next broadcast, readers select the node with the highest probability to broadcast. Simulation results show that the average tag recognition efficiency of this algorithm is 59.9%, and MRCT has advantage in total time slots.关键词：RFID(radio frequency identification);tag recognition;multi-reader;binary tree;anti-collision47|38|0更新时间：2025-07-16
- 摘要：With the development of technology, information security has been greatly challenged. The Physical Unclonable Function (PUF) circuit is a new type of key generation circuit. Resistive Random Access Memory (RRAM) can provide a source of physical random entropy, which makes the PUF physically impossible be attacked. However, in the RRAM-based PUF design, the RRAM delay unit’s test response pair (Challenge Response Pair, CRP) efficiency is not high enough. In this paper, a PUF structure based on RRAM delay unit is proposed. The delay unit outputs the resistance of the RRAM to the inverter to form a delay of the pulse. Finally, the determiner determines that the two pulses are in sequence and is coded as "0" and "1", this is the output bit of the PUF. Based on the RRAM delay unit, this paper designs 8-bit, 16-bit, 32-bit, 64-bit PUF. These PUFs greatly improve the efficiency of PUF RRAM cells under the premise of ensuring good randomness, stability and uniqueness. The experimental results show that the design can effectively improve the efficiency of RRAM, so that PUF can prevent external attacks better.关键词：physical unclonable function;resistive memory;CRP efficiency19|40|2更新时间：2025-07-16
- 摘要：In the Wi-Fi passive target tracking technology, it is difficult to extract the target's reflected path signals due to direct path signals and noise effects. A passive target tracking technology based on Wi-Fi multi-dimensional parameter feature is proposed. This technology uses serial interference cancellation instead of all-zero initialization to complete multiple paths initialization including Arrival of Angle (AoA), Time of Flight (ToF) and Doppler Frequency Shifts (DFS). And the Frequency Domain Space Alternating Generalized Expectation-maximization (FD-SAGE) algorithm is improved to make up for traditional algorithms shortcomings such as slow convergence speed and noise impact. In addition, a hybrid data association method based on the least-cost multi-path network is used to solve the problem of unable to effectively associate paths when there are different numbers of paths at different times. Meanwhile, the method uses the optimal match in a fixed time window as the correlation data at a certain time, which reduces the irreversible error caused by a subsequent association failure caused by the association error. Experimental results show that this paper can achieve an average tracking accuracy of 1.3m in a complex indoor environment.关键词：channel state information;joint estimation of multidimensional parameters;hybrid data association;passive target tracking88|52|1更新时间：2025-07-16
- 摘要：We study the method of family activity recognition based on audio and propose a capsule neural network recognition model based on additive margin. In view of the drawbacks of the traditional capsule neural network objective function only with the output capsule mode length as the constraint, this paper adds a Transition layer to the capsule neural network structure from the perspective of geometry and uses the Transition layer to rebase the capsule unit spatial relationship to the one-dimensional. Then, using the additive margin Softmax as the objective function, the change of similar features is small, and the difference of non-similar features is used as the optimization strategy to construct the objective function based on the capsule vector space relationship to improve model classification ability. Finally, test this method by classified identified for audio events for family activities. Selecting Detection and Classification of Acoustic Scenes and Events (DCASE) 2018 Challenge Task 5 as a dataset for classifier construction and testing, with a final average F1 score of 92.3%, which is superior to other mainstream methods.关键词：classification of acoustic events;family activity recognition;capsule network;additive margin softmax45|38|2更新时间：2025-07-16
- 摘要：The global convergence is one of the important features of an intelligent algorithm. In this paper, we take the initiative to handle the convergence of Grey Wolf Optimization (GWO) with Markov chain. Firstly, a Markov chain model of the GWO algorithm is established through defining the state transition sequence of a gray wolf population. Analyzing the properties of the Markov chain proves that it is homogeneous finite. Secondly, based on the convergence criteria of random search algorithms, the global convergence of the GWO algorithm is verified via analyzing the final state transition sequence of the grey wolf population. Finally, simulation studies on typical testing functions, shifting functions and rotating functions are carried out comparing with a few typical swarm intelligent algorithms. The experimental results show that the GWO algorithm has excellent performance on the global convergence, the computational time and precision of optimization.关键词：Grey Wolf Optimization;grey wolf population state space;Markov chain;state transition probability;convergence criteria116|70|13更新时间：2025-07-16
- 摘要：Relaxed dominance relation based evolutionary algorithms for many-objective optimization (MaOEAs) are widely concerned because they effectively improve the ability to identify solutions, however, the most of these algorithms show poor versatility in solving different objectives of optimization problems. To address this issue, this paper proposes an adaptive dominance criterion based evolutionary algorithm for many-objective optimization (ADCEA). Firstly, the ADC designs an adaptive niche method based on the angle information and the number of objectives between adjacent solutions in the objective space, and combines the convergence indicator information to achieve non-dominated sorting of candidate solutions. Then, in order to further enhance the diversity of the population, the reference vector population technique is introduced in the environment selection. Finally, a reasonable fitness function is constructed, and the non-dominated solution set with better convergence and diversity is selected according to the fitness value. The experimental results demonstrate that the proposed method improves the versatility in solving different objectives of optimization problems, and achieves significant effects in balancing convergence and diversity.关键词：relaxed dominance relations;many-objective evolutionary algorithms;versatility;adaptive dominance criterion37|61|2更新时间：2025-07-16
- 摘要：Due to the low computational complexity and acceptable ability in reducing musical noise effect, the decision-directed (DD) approach is widely used for estimating the a priori signal-noise-ratio (SNR) in many speech enhancement systems. However, the DD approach suffers from the problem of time delay and the performance is very sensitive to the fixed smoothing factor. Firstly, the performance of DD approach in musical noise reduction as well as speech distortion attenuation are analyzed using actual speech and noise data, and the boundary values of smoothing factors are presented in view of the analyzed results. Then, a novel algorithm is proposed, in which two DD approaches with different smoothing factors are probabilistically combined in an attempt to put together the best properties of them. The contribution of either DD approach to the combination is automatically adjusted in accordance with the speech absence probability, which can be computed using the complex Gaussian model and soft decision technique. Experiments are carried out in different noise and input SNR conditions, and the results demonstrate that the proposed algorithm can significantly outperform the popular methods for estimating the a priori SNR.关键词：speech enhancement;decision-directed algorithm;smoothing factor;probability combination44|49|1更新时间：2025-07-16
- 摘要：A model that incorporates semantic information can alleviate the sparse data and cold start problems in a music recommendation system. Current models learn the global relations between users and music by minimizing prediction score error. However, they ignore the fine-grained differences between the implicit features of users and music. Current models also extract semantic features by unsupervised learning that is irrelevant to recommending, leading to inaccurate recommendations. We propose a metric-ranking-learning recommendation model that incorporates content representation (CAMRL). This model is a probabilistic graphical model that optimizes a personalized ranking and explores user music preferences through metric learning at both global and fine-grained levels. To solve the cold start recommendation problem, a supervised learning strategy in relation to the recommendation task is proposed to train the model of content semantic feature exaction. The results of trials using the KKBOX and MIGU datasets show that the proposed model significantly improves cold start music recommendations when compared with other algorithms; it is also more robust when using sparse datasets.关键词：content representation;metric learning;ranking learning;music recommendation18|43|4更新时间：2025-07-16
- 摘要：Aiming at the problem of long test time and affecting test efficiency in integrated circuit testing process, a hierarchical dynamic adjustment method for integrated circuit testing process was proposed. The Bayesian probability model was established by counting the test failure rate of each test type and each test vector in the sample integrated circuit, and the testing process is hierarchically adjusted according to the probability of their hit fault. With the test progressed, the test data was collected continuously, the test failure rate of test type and vector were updated dynamically, and their loading order was adjusted synchronously. The experimental results showed that the proposed method could significantly reduce test time and improve test efficiency. Furthermore, the proposed algorithm is completely based on software, no additional hardware overhead, and can be compatible with the traditional integrated circuit testing process.关键词：adaptive test;test type reordering;test pattern reordering;Bayesian statistics;Poisson distribution80|46|6更新时间：2025-07-16
- 摘要：The nondestructive and high-resolution near-field distribution imaging of microwave and millimeter wave chips is very important for the function and failure analysis of RF chips. This experiment is based on the unique quantum system of diamond NV (nitrogen-vacancy) color center. A diamond sample with a diameter of about 14 μm is selected and adhered to the tapered tip of a 20 μm fiber. A high-resolution, non-destructive and miniaturized probe is prepared by analyzing the ground state spin evolution law of NV color center in the change of microwave field, and the all optical method is used. The whole field distribution on the chip surface is obtained by imaging. In this paper, the near-field distribution image of GaN high electron mobility transistor is given. The ODMR spectrum and Rabi spectrum are fitted, and the imaging results are analyzed. This system has the advantages of high efficiency, high resolution, high sensitivity and low near-field interference. It is expected to provide a new scheme for the application of high integration microwave circuit fault diagnosis, antenna radiation profile, microwave integrated circuit electromagnetic compatibility test, etc.关键词：nitrogen vacancy(NV)center in diamond;optical detection magnetic resonance;microwave field imaging;high resolution87|28|3更新时间：2025-07-16
- 摘要：This paper investigates the synergistic effect of Total Ionizing Dose (TID) -Single Event Transient (SET) of the bipolar voltage comparator LM311 under high and low level output bias conditions. The results showed that the SET of LM311 after the cumulative total dose under high level output bias condition was significantly inhibited, mainly due to the decrease of transient pulse width, and the SET amplitude was also slightly reduced. The SET of LM311 after accumulating the total dose under the condition of low-level output bias has a promoting effect, which mainly manifests as the transient pulse width increases and the SET amplitude increases. Compared with the high-level output bias condition, the LM311 is not sensitive to SET under the low-level output bias condition. It was found that the interface trap charges and oxide trap charges induced by TID are the root cause of the TID-SET synergistic inhibitory effect.关键词：total ionization dose;single-particle transient;synergistic effect;bipolar voltage comparator51|41|2更新时间：2025-07-16
- 摘要：This paper presents an independently adjustable dual-band bandpass filter. The filter is mainly composed of a pair of varactor-loaded parallel coupling short circuits, a pair of λ/2 varactor-loaded open resonators, and a common input/output component. The varactor diode and the parallel coupling short circuits are connected in parallel to form a resonator which produces the first passband. The λ/2 varactor-loaded open resonators are used to load the varactors and produce the second passband. By adjusting the bias voltage on the varactors to change the capacitance value of the varactors, the resonant frequency of the resonators can be shifted and thus the operating frequency of the filter can be adjusted. Both the simulation and testing results show that, the center frequencies of the first and second passbands range from 0.2 to 0.4GHz and from 1.2 to 1.42GHz, respectively, and their adjustment percentages are 66.5% and 15.4%, respectively. The simulation results are well consistent with the testing results, proving that our filter design method is effective. The designed filter can be applied to the cheat prevention and control system.关键词：independently adjustable;dual-band;bandpass filter;cheat prevention and control system21|31|1更新时间：2025-07-16
- 摘要：Traditional principal component analysis, relative principal component analysis and other multivariate statistical methods based on threshold to do the fault diagnosis. Since multivariate statistical method is an equivalent representation of the original space, it does not add any amount of information, making it difficult to diagnose minor faults. And the original space is reduced dimensionally into the principal component space and the residual space, so that the tiny information cannot be fully expressed. Deep learning has been successfully applied in pattern recognition. However, multilevel networks of deep learning represent linear combinations of details but do not have explanatory. Only with the result of training without theoretical basis, it is difficult to carry out mechanistic analysis. This paper presents a fault diagnosis method which combines PCA thought and deep learning thought. Based on the original PCA, this paper first expands and then reduces the dimension, making the inexplicit information in the original space fully expressed and interpreted. Theoretical and simulation experiments show that this method can judge the minor faults which cannot be detected by traditional PCA, improve the detection rate of fault detection and have interpretability.关键词：multilevel high-dimensional;principal component analysis;projection frame;fault diagnosis93|120|5更新时间：2025-07-16
- 摘要：In recent years, extensible processors have been increasingly applied in various embedded systems. The use of custom instructions around the extensible processors can guarantee flexibility while still meeting the high performance and low power requirements of embedded applications. Automatic custom instruction identification is one of the key issues involved in the design of extensible processors. In view of the application fields and development trends of extensible processors, this survey introduces the research progress of automatic identification of custom instructions in recent years. On this basis, there are four key issues involved in the automatic identification of custom instructions: intermediate representation generation, custom instructions enumeration, custom instructions selection and code transformation. This survey analyzes the advantages and difficulties of previous methods, and presents the summary of the applications of extensible processor in terms of different application fields. Finally, the trends and possible research directions of automatic identification of custom instructions is presented.关键词：extensible processor;custom instruction identification;custom instruction enumeration;custom instruction selection76|74|0更新时间：2025-07-16
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0