79results about How to "Faster convergence time" patented technology

Data stored in non-volatile storage is decoded using iterative probabilistic decoding. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding sensed states of a set of non-volatile storage elements. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. Simulated annealing using an adjustable temperature parameter based on a level of error in the data read from the system can be performed to assist the iterative decoding process. The simulated annealing can introduce randomness, as noise for example, into the metric based decoding process. Moreover, knowledge of the device characteristics can be used to guide the simulated annealing process rather than introducing absolute randomness. The introduction of a degree of randomness adds flexibility during the iterative decoding that permits possible faster convergence times and convergence in situations where data may otherwise be uncorrectable.

The invention discloses a PID controller parameter setting algorithm based on an improved PSO (particle swarm optimization) algorithm, and the algorithm comprises the following steps: 1, initializing the algorithm parameters; 2, switching to an iterative loop, and carrying out the updating of the position and speed of each particle; 3, randomly searching a new position in the neighborhood of a current position; 4, calculating the adaptability difference between two positions, and judging whether to accept the new position or not through a simulated annealing mechanism when the adaptability of the new position is inferior to the adaptability of an original position but is superior to the adaptability of a global optimal position; 5, updating the global optimal position of a population, carrying out the natural selection operation, carrying out the arrangement of all particles according to the adaptability values, and employing the information of a part of better particles to replace the information of the other half particles; 6, judging whether to stop the iteration or not; 7, outputting PID controller parameters or executing step 2 again. The method can achieve the automatic setting of control parameters, irons out a defect that a conventional PSO algorithm is very liable to be caught in local optimization, achieves the complementation of the simulated annealing operation and a natural selection strategy, improves the convergence precision of the algorithm under the condition that the number of convergence times of the algorithm is guaranteed, is higher in robustness and precision, and enables the PID controller to generate a more excellent control effect.

The present invention proposes a method of providing information of the echo path of a speech connection in a Packet Data Network, said method comprising the steps of requesting delay information by means for triggering echo removal, said delay information being obtained by signaling a non-speech measurement data unit to said Packet Data Network by an echo estimation means, and measuring the delay of the echo of said measurement data unit by said echo estimation means; replying the echo delay of said measurement data unit from said echo estimation means; generating delay information including at least said echo delay of said measurement data unit; and adjusting a shift register being related to an echo removing device by the use of said delay information such that the performance of said echo removing device is optimized in view of its echo removing capability.

In an ADPLL circuit, on the basis of a gain of a digitally controlled oscillator estimated when a loop gain of a certain value is set in the loop filter and on the basis of a device parameter of the digitally controlled oscillator, the DCO gain estimation unit estimates a gain of the digitally controlled oscillator when a loop gain of another value is set in the loop filter.

The invention requests a dual-link redundancy control method for protection of an Ethernet for plant automation (EPA) switch, wherein the invention relates to the industrial ethernet on-site control field. According to the invention, in the network, a dual-link loop network is formed between EPA switches by connection of two links that are redundant mutually; on the basis of analyses on the links, a strong link between the two links is selected as a working link and the other link is selected as a standby link; the network is trimmed as a loop-free topological structure; transmission of a token that is capable of being elected as main switch authority is carried out and main switch replacement is periodically carried out on the switches; meanwhile, state detection is carried out on the work link and the standby link as well as a plurality of link faults can be processed, so that reliability, stability and robustness of the network are improved.

The invention discloses a real-time ionosphere modeling and monitoring method based on a regional CORS (Continuous Operational Reference System). The method comprises the steps of: receiving originaldual-frequency observation data and a navigation message of a regional CORS station, rejecting a gross error, and then calculating a satellite position, an elevation angle and latitudes and longitudesof a puncture point; employing the Hatch filtering method and employing a carrier smooth pseudo-range observation value for differencing to obtain a TEC (Total Electron Contents) observation value without a geometric ionosphere; employing a spherical ion function to simulate the TEC distribution of the regional ionosphere, taking the model parameters and the satellite and the receiver differential code deviation as parameters to be estimated, constructing observation vectors by employing the observation data of all CORS stations of the single epoch to construct an observation equation; establishing a Kalman filter to perform real-time filtering processing for the parameters to be estimated, and separating from hardware delay to obtain an ionosphere TEC model; and finally, employing a model to calculate the grid point ionosphere delay, and storing and broadcasting the grid point ionosphere delay to regional users in a grid form. After an ionosphere product is obtained, a regional single-frequency PPP user location precision is obviously improved, and the dual-frequency PPP user convergence time is significantly improved.

The invention discloses a broadband anti-interference system based on deep reinforcement learning and an anti-interference method. A group of transmitter and receiver pairs are considered as a communication user; when a user performs communication, a plurality of jammers perform malicious interference on the user, the user intelligently selects a communication frequency from a wide frequency bandby utilizing spectrum sensing information, the user decision process is modeled into a Markov decision process, and a user frequency decision is optimized to maximize the throughput of the user. According to the anti-interference method, a layered deep reinforcement learning algorithm is designed, then a two-dimensional window is controlled according to a frequency band decision, a frequency spectrum waterfall plot of a corresponding frequency band is selected, and a frequency selection network is designed for learning to obtain an optimal frequency decision. The broadband anti-interference system is complete in model and reasonable and effective in design algorithm, and compared with a traditional deep reinforcement learning anti-interference algorithm, iteration time and calculation complexity are effectively reduced while an excellent anti-interference effect is guaranteed.

The invention belongs to the technical field of artificial intelligence. The invention discloses a multi-agent cooperation information processing method and system, a storage medium and an intelligentterminal. Each agent reserves own information trace in the environment when taking the next behavior in the environment, searches surrounding information trace when other agents reach the state in the environment, and adds the information trace into the neural network for training; the grouping model finds a better cooperation strategy among the multiple agents, and the grouping model predicts the optimal grouping of the multiple agents at the next moment by using the grouping relationship among the multiple agents; when one round of training through the G model is finished each time, the loss function formula of each agent is regarded as fitness, the loss value mean value of each round of intelligent agent trajectory is counted, and the loss values of all agent trajectories of each roundare summarized. According to the method, the team learning efficiency of the multiple agents is improved, and the multiple agents can better complete tasks through team cooperation.

The invention relates to an improved particle swarm algorithm for automatic optimization of control law parameters of an unmanned aircraft. The particle swarm algorithm is based on an iterative optimized algorithm; control parameters are initialized to be a group of random solutions; and optimal values are found by iteration; the improved particle swarm algorithm is characterized by comprising the following steps of: in the particle swarm algorithm, introducing a crossover operator, selecting a plurality of particles in which the optimal position value of single particle is positioned in the middle to carry out random two-two crossover; generating offspring particles with the same number; replacing parent particles with offspring particles; selecting crossover time according to convergent algebra of the standard particle swarm algorithm, namely, 17 to 20 generations for pitching and yawing channel and 7 to 10 generations for rolling channel; and selecting algebra of disturbing start as follows: starting from twelfth generation for the pitching and yawing channel and starting from fourth generation for the rolling channel. The improved particle swarm algorithm expands the range of understanding the space, has total searching capacity; the obtained parameters can meet given performance indexes; and the algorithm can save design time and has application value of engineering.

The invention discloses a sparse reference station network zenith troposphere delay modeling method considering the elevation difference. Conventional local area network enhancement PPP (precise point position) modeling methods do not consider effects caused by the elevation difference on the interpolation accuracy and depend on a certain amount of reference stations. According to the method, the relation between the elevation difference and zenith troposphere delay is considered, reference station zenith troposphere wet delay values are calculated through non-ionosphere combined precise point filtering, zenith troposphere dry delay values are calculated through a model, the dry (wet) delay values are divided into an elevation strong correlation part and an elevation weak correlation part according to the relation between the troposphere delay values and elevation factors, and the sparse reference station network zenith troposphere delay modeling method considering the elevation difference is provided on the basis. The method considers the effects caused by the elevation difference on the troposphere delay values, is applicable to sparse reference station local area networks and is low in dependence on the number of the reference stations.

A technique for reconstructing a mask pattern corresponding to a photo-mask using a target mask pattern (which excludes defects) and an image of at least a portion of the photo-mask is described. This image may be an optical inspection image of the photo-mask that is determined using inspection optics which includes an optical path, and the reconstructed mask pattern may include additional spatial frequencies than the image. Furthermore, the reconstructed mask pattern may be reconstructed based on a characteristic of the optical path (such as an optical bandwidth of the optical path) using a constrained inverse optical calculation in which there are a finite number of discrete feature widths allowed in the reconstructed mask pattern, and where a given feature has a constant feature width. Consequently, the features in the reconstructed mask pattern may each have the constant feature width, such as an average critical dimension of the reconstructed mask pattern.

The present invention relates to electronic systems for use in high demand or mission critical environments, including power stations, sub-stations, roadside transportation, rail, and industrial applications. The present invention provides a) increased resilience of mission critical communications infrastructure b) ability to maintain multiple independent RSTP domains over HSR ring d) method of using a hidden VLAN as an RSTP domain ID c) a method of creating multiple redundant protocols domains over HSR ring.

A system for adaptively generating a sidelobe null in a radar transmit antenna pattern by positioning a small air vehicle along the radial of the sidelobe to be suppressed. The air vehicle is fitted with a receiver and antenna facing the radar, as well as a GPS device for maintaining the designated position. The vehicle further includes a communication link to the processor of the main radar transmitter to form a closed loop that enables adjustment of the attenuators and phase shifters of the auxiliary channel(s) to suppress signals transmitted in the sidelobe to be nulled. The com link may be replaced by a suitable transponder.

An embodiment of the invention discloses an interchanger stacking system, a slave unit, a switching chip and a method for processing a protocol message. In the scheme, an ENCAP module which is a hardware module in the switching chip is independently used for packaging a protocol message which is about to be processed by a CPU of main equipment. Compared with a method of packaging the protocol message by forwarding the protocol message to a CPU of a slave unit in the prior art, the method not only consumes shorter time in a packaging process, but also simplifies a forwarding process, only the switching chip is utilized, the hardware forwarding speed is high, protocol convergence time of the stacking system is shortened greatly, the protocol message processing efficiency is improved, and the packet loss amount in a protocol switching process is reduced. In addition, DCB information, which is up-supplied to a CPU of a main unit, of the protocol message is packaged at one step in a RCPU head, and the information, which is up-supplied to the CPU of the main unit, of the protocol message received from a device is consistent to the information, which is up-supplied to the CPU of a stand-alone interchanger when the interchanger is in a stand-alone state.

The invention provides a single channel communication signal and interference blind separation method based on mutation particle swarm particle filtering (MPSOPF-SCBSS), and belongs to the communication signal processing and communication anti-interference field. In view of the single channel communication signal and interference blind separation problem in the condition of large interference-to-signal ratio, small number of particles and low signal-to-noise ratio, the method adopts Bayesian filtering estimation for communication signal and interference separation. According to the method, a single channel disturbed communication signal state space model is firstly built, the single channel interference and signal blind separation problem is converted into joint estimation of communication symbols and unknown system parameters, and joint estimation is realized through mutation particle swarm optimization particle filtering to achieve the blind separation purpose; through a velocity-displacement update function and a Metropolis-Hastings mutation rule in the particle swarm optimization, the needed particle number in the sequential estimation process is reduced and diversity of the particle set can be ensured. In addition, the provided mutation particle swarm optimization rule reduces demands on the signal-to-noise ratio of communication signals in the single channel disturbed condition.

The invention relates to a network intrusion detection method based on a CNN-GRU hierarchical neural network, and the method comprises the steps: capturing a network flow data package, namely, a to-be-classified data package, through Wreshark software; performing data packet marking, preprocessing and data cleaning on the to-be-classified data packets, analyzing the data packets into decimal data, and converting the decimal data into a 40 * 40 single-channel grey-scale map to obtain a sample complete set; dividing the sample complete set into a training set and a test set, taking the single-channel grey-scale map matrix as an input vector, and establishing a CNN-GRU hierarchical neural network classification model through the training set; and after model training is completed, transmitting data of the test set are into the model, the model predicts the input data according to parameters obtained through training, and unknown network traffic is classified to judge whether the unknown network traffic is attack traffic. Experimental results show that the accuracy of the method for classifying the normal traffic and the attack traffic reaches 99.92%.

The invention relates to a robust adaptive autopilot control algorithm for a navigation simulator. Aiming at various ship models in navigation simulator training, a nonlinear mathematic model for ship motion is obtained based on half theoretical calculation and half experimental result, and the nonlinear mathematic model considers the interferences of wind, wave, current, and the like and relates to a low-speed shallow-water model. The designed autopilot algorithm is based on a closed-loop gain shaping simple robust control algorithm and is a new ship autopilot robust adaptive control algorithm having robustness and adaptability by combining a neurotic network direct controlling training method. The robust adaptive autopilot control algorithm has adaptability by considering the robustness of a ship motion control algorithm as well as the complexity of ship navigation and solves the problem that the prior algorithm has either robustness or adaptability and can not have both.

The invention discloses a real-time precise point positioning method based on the satellite clock and R-orbit error mixed estimation. A real-time satellite clock error estimation terminal and a mobile station client terminal are adopted. The method includes that the real-time satellite clock error estimation terminal estimates the satellite clock errors in real time through broadcast Ephesus and regional CORS continuous observation data and transmits to a mobile station through a mobile network rapidly; the mobile station client terminal receives GNSS observation data and acquires and decodes satellite clock error data through the mobile network, and the precise point positioning is performed after data p reprocessing. By the aid of the method, the positioning precision and convergence time of precise point positioning are improved and increased significantly on the basis of the IGS real-time data flow method, the small-scale CORS application is supported, and the limitation that satellite orbit determination requires a large range of ground stations is overcome; a user of the mobile station only receives the real-time satellite clock error correcting data instead of precise orbit data, and the network communication load is reduced.

The invention discloses a processing method and device of a link state. The method comprises the following steps: receiving a change message of the link state, wherein, the change message of the link state indicates change of the link state of a user network interface (UNI) port of a provider edge (PE) device of a terminal; packaging an remote defect indication (RDI) message according to the change message of the link state, wherein, the RDI message is used for informing the PE device at an opposite terminal of starting or closing the UNI port of the PE device at the opposite terminal; and sending the RDI message to the PE device at the opposite terminal. In the invention, port state synchronization can be realized so as to enhance user experiences and accelerate the convergence speed of customer service.

A method and apparatus are provided for reducing, and preferably substantially eliminating, data-pattern autocorrelations found in digital communication systems. The method employed is referred to as Data Subset Selection (DSS) and is implemented in the form of DSS engine. Autocorrelations in the data-pattern can cause many digital adaptive systems to converge to an incorrect solution. For example, the LMS method, which is often used in adaptive filtering applications, can converge to an incorrect set of filter coefficients in the presence of data-pattern autocorrelations. Digital timing recovery methods are also susceptible. Other impairments that result from data-pattern autocorrelations include increased convergence time and increased steady-state chatter. DSS reduces, and preferably substantially eliminates, data-pattern autocorrelations by selecting a subset of the data stream that either demonstrates smaller autocorrelations or no autocorrelations, thus improving the performance of the aforementioned digital adaptive systems.

The invention relates to a smart phone ionized layer error correction method and device based on regional CORS. The method comprises the steps of obtaining a double-frequency pseudo-range observation value, a double-frequency carrier phase observation value and a first navigation message in a regional CORS station data stream through a server, obtaining an ionized layer smooth observation value and a first puncture point geocentric longitude and latitude, fitting a regional VTEC formula by combining a low-order spherical harmonic function model, and obtaining an observation equation; resolving the observation equation by using a Kalman filter to obtain regional ionospheric model parameters; enabling a smart phone terminal to obtain regional ionospheric model parameters from the server, analyze an observed value to be corrected according to the obtained GNSS original observed value, analyze geocentric longitude and latitude of a second puncture point according to the obtained second navigation message, determine ionospheric delay correction number according to the geocentric longitude and latitude of the second puncture point and the regional ionospheric model parameters, and correct the observed value to be corrected; and performing positioning analysis by adopting a Kalman filtering method based on the modified observation value, and obtaining a positioning result, so that the real-time positioning precision is improved.

The invention discloses a modeling method for an optical proximity correction process model. The method comprises the steps of measuring original data and line width roughness of a semiconductor process layer; and determining weighted values of all critical sizes according to the line width roughness of the measured value of each critical size, a datum line width roughness at design rule checking position of the semiconductor process layer and a datum weighted value at the design rule checking position of the semiconductor process layer. By introducing the line width roughness to determine the weighted values of all the critical sizes, to-be-measured data can be reduced; some unstable factors and random factors of process, measuring machines and measuring processes can be filtered; fitting time and convergence time of the process model establishment can be shortened; and simultaneously accuracy and reliable stability of the process model can be guaranteed.

Abstract of Disclosure An auto gain control (AGC) algorithm which is suitable for a digital processing system. The AGC algorithm includes an adaptive stage and a sleeping and tuning stage. In the adaptive stage, determining if an strength of a first received signal is within a first range and is larger than a second threshold value, if no, measuring strength for sequentially received signals by adaptively adjusting the gain of the AGC algorithm until the strength of one of the sequentially received signals is within the first range and is larger than the second threshold value. In the sleeping and tuning stage, the AGC algorithm stops tuning the gain for a first period, and then measuring strength for sequentially received signals by iteratively adjusting the gain of the AGC algorithm until the strength of the other one of the sequentially received signals is within a second range. When the strength is within the second range, setting the gain according to the energy and locking the gain for remaining sequentially received signals.

The invention discloses a wireless sensor network topology optimization method based on asynchronous deep reinforcement learning. The method comprises the following steps: a step of generating an initialized wireless sensor network topology structure by utilizing a rule of a scale-free network model; a step of compressing the topological structure of the wireless sensor network; a step of initializing an asynchronous deep reinforcement learning model; a training and testing stage; a training stage, wherein firstly, a wireless sensor network topological structure is serialized, and a row vectoris used for representing the wireless sensor network topological structure; then, respectively inputting the row vectors of the network topology structure into different local network training models; secondly, the local network training model comprises two neural network models, namely an action selection strategy network and a strategy evaluation network; a test stage wherein the global networktraining model performs test evaluation on the test data set; a step of repeating the steps 1, 2, 3 and 4 till the maximum number of iterations is reached.

The invention discloses a micro grid power allocation method based on a distributed finite-time controller, comprising the following steps: processing the reference value and actual value of the active power of a micro grid through a nonlinear integral controller to get reference values of active power utilization rate of all power supplies; establishing a communication network connecting all the power supplies inside the micro grid; and controlling all the power supplies through a distributed finite-time controller to make the actual values of active power utilization rate of the all power supplies approach the reference values. By using the method, the active power distribution capacity of the micro grid is improved, the interference of the malfunction of the communication network to the stable operation of the micro grid is eliminated, the influence of the traditional micro grid control center on the reliability of the micro grid is eliminated, the convergence of active power output by the distributed power supplies is speeded up, and the dynamic performance of the micro grid is improved.

The invention discloses a method and equipment for quickly determining an OSPF (Open Shortest Path First) interface role. The method comprises the following step: when an OSPF interface is in Waiting state, network equipment triggers a Best Priority event, transfers an OSPF interface state from the Waiting state to a DR (Designated Router) state or a BDR (Backup Designated Router) state and sends a HELLO message which carries DR or BDR information to other network equipment. In the invention, time overhead for a neighbor establishing process is saved, and longer dead timer time is allowed to be configured.

The invention provides an SCMA multi-user detection method based on weight factor message passing, belongs to the field of signal detection of wireless communication systems, and relates to a sparse code multi-address access (SCMA) uplink communication system multi-user detection method which is high in convergence speed, excellent in performance and low in complexity and can be used for solving the multi-user detection problem in a 5G mobile communication alternative multi-address access technology. According to the method, the initial condition probabilities of all superposition constellation points are changed by introducing weight factors alpha on the basis of a message passing algorithm (MPA) to enable the constellation points near a receiving signal to be capable of obtaining higher initial condition probabilities, therefore, the convergence time of the message passing iteration process is shortened, the iteration result is more accurate, and good comprise between the performance and the complexity of an SCMA system is achieved. The method can be applied to SCMA system-based multi-user detection in next generation mobile communication.

The invention discloses a method for leading a photo mask deviation magnification coefficient into optical proximity effect modeling, and during collection of line width data for a basic process of optical proximity effect, a process model is built, a weight value of experienced line width data required by the process model is correlated with the photo mask deviation magnification coefficient of a graphics, so that a plurality of processes and unstable factors in measurement can be filtered, and the fitting and convergence time for building the process model can be greatly accelerated, meanwhile, the accuracy and reliable stability of the process model can be ensured.