93results about How to "Short convergence time" patented technology

The invention discloses a distributed finite time tracking control method for a multi-robot system in view of interference and model uncertainty, and relates to a control method for a multi-robot system. In order to solve the problems of poor robustness of the existing control method for the multi-robot control system and heavy overall communication burden of the multi-robot system, the method comprises the steps of establishing a dynamical model FORMULA (shown in the specification) of a pilot robot and a dynamical model FORMULA (shown in the specification) of each following robot in the multi-robot system, and calculating a weighted adjacent matrix A and a Laplacian matrix in a directed graph theory of the multi-robot system; and then designing a distributed finite time tracking control law FORMULA (shown in the specification) of the multi-robot system, so that each following robot follows the pilot robot with a dynamic time change locus within finite time and finite time tracking control of the multi-robot system is completed. The method is applicable in the field of control of multi-robot systems.

The invention relates to a method for estimating pavement peak attachment coefficients of a distributed driving electric vehicle. The method comprises the following steps of: acquiring a driving force signal, a longitudinal vehicle speed signal, a vehicle-mounted longitudinal acceleration sensor signal, and a wheel speed signal; estimating slip rate and utilization attachment coefficients in real time respectively by using the longitudinal dynamics and tire dynamics of the vehicle; transmitting an estimated slip rate value to a simplified tire model, so as to obtain utilization attachment coefficient estimation values of a group of different pavements; and comparing the utilization attachment coefficient estimation values with the utilization attachment coefficients of pavements, selecting a value with the minimum error, and solving a peak value by using the simplified tire model, so as to obtain the peak attachment coefficients. The method is simple in solving process, quick in operation, short in convergence time, high in robustness, and applicable to real-time estimation of the peak attachment coefficients of the various pavements in the running process of the vehicle.

The invention relates to a formation satellite configuration containment control method, in particular, a formation satellite finite-time configuration containment control method. The objective of the invention is to solve the problem of low robustness of an existing multiple-satellite system formation control method and the problem of incapability of completely adapting to a practical application environment which is caused by a situation that inter-satellite communication topology is an undirected graph in the existing multiple-satellite system formation control method. The formation satellite finite-time configuration containment control method includes the following steps that: a relative orbit dynamic model of formation satellites i of a satellite formation system and relative reference points is established according to an established relative motion dynamic equation of reference satellites and accompanying satellites, and is simplified as an expression described in the descriptions; a weighted adjacent matrix A and a Laplacian matrix in the graph theory of a directed graph in the satellite formation system are provided according to the formation types of the formation satellites i; and a distributed finite-time configuration containment control laws of the multi-dynamic-pilot-satellite satellite formation system are designed, and therefore, each following satellite can achieve at a configuration convex hull formed by pilot satellites in finite time, and formation satellite finite-time configuration containment control can be realized. The formation satellite finite-time configuration containment control method of the invention is applicable to the control field of formation satellite configuration.

A signal processor includes: a first adaptive filter that takes a first signal as input and generates a first pseudo signal; a first subtractor that subtracts the first pseudo signal from a second signal to supply a first differential signal as output; a second adaptive filter that takes the first signal as input to generate a second pseudo signal; a second subtractor that subtracts the second pseudo signal from the second signal to supply a second differential signal as output; a first step size control circuit that generates a first step size used in updating the first adaptive filter in accordance with the relation between the second pseudo signal and the second differential signal; and a second step size control circuit that generates a second step size used in updating the second adaptive filter in accordance with the relation between the first signal and the second signal.

The invention discloses a global fusion positioning method based on self-adaptive Monte Carlo and feature matching, which comprises the following steps of: extracting map line features in a grid map and storing the map line features in a map feature database; extracting the scan line features of a non-second frame of a laser radar scan map, and initializing the particle set pose of the self-adaptive Monte Carlo by adopting the estimated pose calculated by the first frame of the laser radar scan map; updating the pose and weight of the particles by adopting odometer data, a motion model, a laser radar scan map and a measurement model for a second frame of laser radar scan map; updating the initial pose of a part of particles in a particle set by adopting the estimated pose calculated by thelaser radar scan map for the remaining frames of the laser radar scan map; updating the weight of corresponding particles; and meanwhile, updating the pose and weight of the rest particles in the particle set by adopting the odometer data, the motion model, the laser radar scan map and the measurement model, and finally combining the two particle sets updated with the pose and the weight into a new particle set.

Embodiments of the invention provide a method and a device for estimating parameters of mixture Gaussian distribution. According to the method, when mixture Gaussian is used for modeling, all initial component parameters and initial mixed coefficients of a mixture Gaussian probability density function are estimated according to a clustering algorithm, then all the initial component parameters and initial mixed coefficients are taken as initial values of an expectation maximization (EM) algorithm or a greedy EM algorithm, and iterative operation is performed through the EM algorithm or the greedy EM algorithm so as to obtain all final component parameters and final mixed coefficients of the mixture Gaussian probability density function. Through adoption of the method, the clustering algorithm provides a coarse estimation of the parameters of the mixture Gaussian probability density function; on the basis of the coarse estimation of the parameters, the EM algorithm improves the precision of parameter estimation, so that combination of the clustering algorithm and EM algorithm guarantees a likelihood function to converge to global maximum point, the convergence time of the likelihood function is shortened and the implementation complexity is lowered.

The invention discloses two Lagrange navigation constellations and a construction method thereof. One of the two navigation constellations is formed by three Lagrange navigation stars, the three Lagrange navigation stars are respectively positioned in three period orbits, and the three period orbits comprise a first period orbit constructed nearby an earth-moon system translation point L2, a second period orbit constructed nearby an earth-moon system translation point L4 and a third period orbit constructed nearby an earth-moon system translation point L5; and the other one of the two navigation constellations is formed by four Lagrange navigation stars, the four Lagrange navigation stars are respectively positioned in four period orbits, and the four period orbits comprise the above three period orbits, and also comprise a fourth period orbit constructed nearby the earth-moon system translation point L1. The two Lagrange navigation constellations can realize seamless coverage of the moon space in order to provide continuous navigation information for a moon detector.

The invention relates to a method for compensating three-phase unbalanced currents on the basis of an SVG (static var generator). The method includes determining whether load currents needs to be compensated or not via three-phase unbalanced judging links by the aid of load current signals which are used as control input; acquiring frequencies and phases of power grid voltages via adaptive frequency locking and phase locking links when the load currents need to be compensated; filtering harmonic current components in the load currents by the aid of discrete second-order generalized integrators; computing negative-sequence current values required to be converted for various phases if the three-phase unbalanced currents are in balanced states; transmitting the negative-sequence current values to control instruction computation modules in SVB devices; driving IGBT (insulated gate bipolar transistors) to act; generating currents equal to fundamental negative-sequence components in the load currents so that the three-phase unbalanced currents can be in the balanced states. The negative-sequence current values are used as reference current signals. The directions of the currents are opposite to the directions of the fundamental negative-sequence components in the load currents. The method has the advantages that the three-phase unbalanced currents can be dynamically compensated, and accordingly the stability and the reliability of power grids can be improved; low-pass filters can be omitted when negative-sequence currents are extracted, and accordingly the method is low in computational burden and high in response speed and tracking precision.

The invention provides a medicament module pharmacokinetic property and toxicity predicting method based on a capsule network. After a comprehensive module fingerprint and a module descriptor are constructed and early-period preparing operation for establishing model is performed, a low-grade characteristic content of a molecule is extracted from an upper-layer low-grade characterized through convolutional or restricted Boltzmann machine operation; then a capsule network method is used for abstracting the high-grade characteristic of the molecule in a lower-layer high-grade characteristic; a relation between the high-grade characteristic and an active label is fit through a dynamic routing algorithm, thereby predicting the pharmacokinetic property and the toxicity class of an unknown smallmolecule. The method does not require collection of large scale datasets for training, optimization is performed on input through end-to-end and furthermore automatic dimension reduction is realized.A coupling coefficient is updated through iterating a dynamic routing process. The dynamic routing conveys all characteristics of an upper-layer capsule to a random lower-layer capsule, thereby greatly reserving a hierarchical position relation. The method realizes a better predicting effect than that of a traditional machine learning method.

Provided is an apparatus and method for separating music and voice using an independent component analysis method for a two-dimensional forward network. The apparatus of separating music and voice can separate voice signal and a music signal, each of which are independently recorded, from a mixed signal, in a short convergence time by using the independent component analysis method, which estimates a signal mixing process according to a difference in record positions of sensors. Thus, users can easily select accompaniment from their own compact discs (CDs), digital video discs (DVDs), or audio cassette tapes, or FM radio, and listen to music of improved quality in real time. Accordingly, the users can just enjoy the music or sing along. Furthermore, since the independent component analysis method in the apparatus of separating music and voice is simple and time taken to perform the method is not long, the method can be easily used in a digital signal processor (DSP) chip, a microprocessor, or the like.

The invention discloses a digital background correction method applicable to a pipelined analog-to-digital converter; by adding pseudorandom quantity, the information of sampling capacitance mismatch and harmonic coefficients of an operation amplifier can be obtained to further correct the main output of the pipelined analog-to-digital converter. The method has the advantages of short convergence time and high error estimation precision, and the performance of the analog-to-digital converter is greatly improved.

The invention discloses a torque impulse balance control method of an AC synchronous motor system. The torque impulse balance control method comprises the following steps of firstly, setting a dynamic rotational speed threshold and a target rotational speed of a motor in advance, acquiring an actual rotational speed of the motor, and calculating the sizes of a real-time electromagnetic torque and a load torque according to the current of a motor armature, the position of a rotor and a speed signal; secondly, judging whether an absolute value of the difference between the actual rotational speed and the target rotational speed is greater than the dynamic rotational speed threshold, set in advance, of the motor or not; and finally, if the absolute value of the difference is greater than or equal to the rotational speed threshold, allowing the system to work in a torque impulse balance control mode so that the rotational speed of the motor can be restrained after being subjected to one-time speed reduction and one-time speed rise, wherein the rotational speed convergence time is the shortest, and the rotational speed dynamic ripple is minimum. By the method, the rotational speed convergence time is the shortest, the rotational speed dynamic ripple is minimum, thus, the rotational speed convergence time and the rotational speed dynamic ripple of an arbitrary load during abrupt changing cannot be affected by proportional integration (PI) parameters of a rotational speed ring, and a speed regulation system has the optimal dynamic performance.

The invention discloses a MPPT strategy of a partial overshadowing photovoltaic system based on ant colony optimization and variable step size disturbance observation algorithms. According to a MPPT method, the problem of searching a global maximum power point when multiple peak values occur in the photovoltaic system under a partial overshadowing condition is solved. When a proper situation appears, the early stage of photovoltaic output is tracked by virtue of an ant colony algorithm, so that the phenomenon that the trapping of a system in a partial maximum power point by a disturbance observation algorithm is avoided, and meanwhile, the convergence speed is increased. After the secondary iteration of the ant colony algorithm ends, the algorithm is stopped, the later stage of the photovoltaic output is tracked by virtue of the variable step size disturbance observation algorithm, and the steady-state oscillation of the system caused by the ant colony algorithm is avoided by virtue of the robustness of the variable step size disturbance observation algorithm. Finally, the global maximum power point of the partial overshadowing photovoltaic system can be rapidly, accurately and stably tracked.

The invention discloses a software vulnerability detection method based on short system call sequences. The method is used for detecting and positioning software vulnerability. The method comprises the steps of building the function call chain CS of a tested system, storing the function call chain CS in a function stack table, and denoting the index of the function call chain as CSV; acquiring a system call sequence under the normal behavior, and taking the sequences from each invoke node i to n-1 nodes after the invoke node as normal short sequences; acquiring unknown short sequences under unknown behaviors in the similar way; adding call deviation values of corresponding elements of each unknown short sequence and the corresponding normal short sequence together to obtain sequence deviation, and obtaining the minimal sequence deviation value; taking the number of unmatched elements as Hamming distance, and finding out minimal Hamming distance; finding out the invoke node with the largest call deviation value in the unknown short sequence which does not exceed a threshold value most recently when the largest value of the minimal Hamming distance and the largest value of the minimal sequence deviation both exceed a certain threshold value, and finding out the vulnerability position according to the CSV value. By the adoption of the method, detection efficiency is improved.

In a wireless transmitter and receiver, a background calibration type analog-to-digital converter generally occupies a large area because of the phase compensating capacity of an op-amp included in a reference analog-to-digital conversion unit. Further, the calibration type analog-to-digital converter generally requires a sample and hold circuit to exclude influence of parasitic capacitance of wirings, thereby increasing power consumption. Digital calibration is performed by using, as a signal for calibration, an input signal of a digital-to-analog converter in a transmitter circuit of the wireless transmitter and receiver and inputting an output signal from the digital-to-analog converter to the analog-to-digital converter in the receiver circuit.

The invention discloses a control method for a hybrid excitation generator. The integral value of the average value of the output current of a rectification circuit is enabled to be equal to the integral value of the load current in the control process through reasonable design of the change rate of the excitation current at each moment when the load of the hybrid excitation generator suddenly increases or decreases, and the instantaneous value of the average value of the current output by a rectifying circuit at the end of the process is equal to the instantaneous value of the load current at the sudden change moment, so that the output voltage can be converged only through one adjustment process, the load sudden change process is free from overshoot or down-regulation, the convergence time is shortest, the problems that the dynamic curve of the charging current and the output voltage controlled by the existing generator voltage based on charging and discharging energy control is influenced by PID parameters, the optimization cannot be realized, and multiple adjustment processes exist are solved.

Provided are a vibrating motor with long life wherein vibration of the vibrating motor can be converged instantaneously, and an electronic device. Upon receiving an input signal from the operating section (5) of a touch panel, an electronic device (1) feeds a drive current (A) to a vibrating motor (7) and gives a sense of operation by vibrating the vibrating motor, wherein a brake current (B) is fed in the reverse direction after the vibrating motor (7) is driven by feeding the drive current (A) thereto and the vibrating motor (7) is damped thus converging vibration of the vibrating motor instantaneously.

The invention discloses a visual navigation method based on deep reinforcement learning and direction estimation. The method comprises the following steps: 1) generating an AI2-THOR simulation platform offline data set; 2) extracting image features of the RGB-D image of the offline data set; 3) constructing an A3C deep reinforcement learning model, and taking the features in the step 2) as the input of the model to train the model; 4) migrating the model trained in the step 3) according to the real scene, performing fine tuning on the model to improve the generalization ability, and finally applying the fine-tuned model to the visual navigation of the real scene. According to the method, a three-dimensional geometric method is used, the position and the direction of navigation to the target object are calculated through a direction estimation method, direction features are provided for a deep reinforcement learning model, the model can converge more quickly, the generalization ability is higher, and accurate visual navigation is achieved.

The invention relates to a DDoS attack detection method in an SDN environment based on a Renyi entropy and a BiGRU algorithm, and belongs to the technical field of network information security. Firstly, an OpenFlow switch is utilized to collect flow table information, six feature vectors are extracted from the collected data to serve as feature vectors of attack detection, then Renyi entropy is applied to carry out abnormal flow detection, and detection is divided into a normal result and an abnormal result. And if the flow is detected to be abnormal, attack detection is carried out on the traffic by applying a BiGRU (Bi-gated recurrent unit, BiGRU) algorithm. According to the invention, the Renyi entropy comprises multiple types of entropies, so that more complex conditions can be dealt with in the initial detection process. According to the invention, the BiGRU model is applied to solve the problem of gradient disappearance or gradient explosion of a traditional RNN along with increase of the sequence length, and bidirectional propagation is achieved; the output layer has complete past and future information of each point in the input sequence; the training parameters are few, the convergence time is short, and the detection rate of the DDoS attack in the SDN network is improved.