46results about How to "Reduce estimation bias" patented technology

The invention discloses a network service anomaly detection method based on distributed probe monitoring. The method comprises the following steps: 1, a Bayesian network architecture is designed, and the Bayesian network architecture is used to estimate a flow matrix; and 2, the flow matrix is used for network anomaly detection. The first step comprises sub steps: the Bayesian network architecture is designed; the estimation value of the maximum posterior probability of a network parameter is calculated; the initial estimation value of the flow matrix is generated; and the flow matrix is estimated. The second step comprises sub steps: the Bayesian network performs learning through a group of network service flow training data; one flow matrix is generated randomly as a prediction factor for network agendas diagnosis; the network probes are deployed in a distributed mode to different network nodes; and the prediction factor is used for network anomaly detection. According to the network service anomaly detection method based on distributed probe monitoring, through the distributed probe monitoring mode, the network service flow anomalies in the network can be effectively detected and found out, and safe network service transmission can be realized.

The invention discloses a piecewise constant independent component analysis-based method for estimating a harmonic current of a power distribution network. The method comprises the steps of firstly building a piecewise constant model of harmonic current estimation, determining a harmonic current sequence by adopting a correlation coefficient matrix association method of overlapped sections, determining a harmonic current amplitude by utilizing an error minimization method of the overlapped sections, and lowering the need of prior harmonic current data; secondly giving a criterion of power network parameter change; and finally determining a data interval of the power network parameter change through a bisection method, and correcting a harmonic current estimation result during the power network parameter change to improve estimation precision during the power network parameter change. According to the method, harmonic current estimation errors during the power network parameter change can be reduced and the estimation process is free of power network parameter data; and the method is suitable for the harmonic current estimation under the condition that the parameters of the power distribution network are dynamically changed.

The invention discloses a parameter estimation and tracking control method based on a turntable servo system, and belongs to the technical field of parameter identification and electromechanical control. The method comprises analyzing the turntable servo system containing unknown parameters, establishing a mathematical model of the turntable servo system containing unknown parameters according to a mechanism modeling method; updating an adaptive rate by a filtering theory, introducing a performance index function with a forgetting factor, and designing a variable-gain adaptive rate to achieve optimal adaptive parameter estimation by optimizing the performance index function; and achieving position tracking control by using the controller of a sliding-mode control design system using a super-twisting algorithm. The method can achieve the parameter estimation and tracking control of a system and has advantages of (1) reducing a system parameter estimation overshoot and accelerating the convergence of parameter estimation; (2) ensuring that a tracking error converges to zero within finite time; being not required to obtain the derivative information of a sliding-mode variable or not requiring a continuous control law of the sliding mode; eliminating chattering and improving the robustness of the controller.

The invention discloses a target extraction and aggregation method based on a ground surveillance radar system. The method comprises a step of performing false dot filtering processing on target echoinformation, a step of blurring the filtered target echo information to form a two-dimensional image, a step of performing target edge extraction on the two-dimensional image, a step of performing ground object target filtering processing after the target edge extraction, a step of sequentially carrying out velocity aggregation and distance aggregation on target echo information belonging to the same target after the ground object target filtering processing, and a step of performing azimuth aggregation on the target echo information which is subjected to distance aggregation to obtain targetecho information and target parameters belonging to the same target after the azimuth aggregation. According to the invention, the risk that the same target is split into multiple targets when the target extraction is performed is reduced, the risk that targets of different speeds are merged into the same target when the targets are aggregated is reduced, the integrity and accuracy of the target extraction and the accuracy of the target aggregation are improved, and a reliable basis is provided for subsequent track processing.

The invention discloses a signal processing method which comprises the following steps: the strengths of user signals with different arrival angles of a user are determined according to directional vectors of the user signals with different arrival angles of the user, a downlink forming vector corresponding to the user and directional gains of a receiving antenna at different arrival angles; and an azimuth estimation value of the user is determined according to the strengths of the user signals with different arrival angles. Another signal processing method comprises the following steps: the azimuth estimation value of the user is determined; the deviation of the azimuth estimation value of the user is estimated according to the difference between two user signal arrival angles corresponding to a given attenuation threshold; and the azimuth estimation value of the user is corrected by the estimated deviation. The two methods can reduce or even eliminate the deviation between the azimuth estimation value and the true value of the user.

The present invention discloses a sea clutter Pareto distribution parameter estimation method based on a logarithm cumulant. According to the method, firstly a Pareto distribution probability densityexpression of sea clutter is obtained, then two calculation methods of logarithm statistic are used to estimate the Pareto distribution of the sea clutter observed in multiple times, and the estimation of Pareto distribution shape parameters and scale parameters of the sea clutter observed in multiple times is completed. The method concretely comprises the steps of (1) obtaining the Pareto distribution probability density expression of the sea clutter observed in multiple times, (2) calculating a reference value of the logarithm cumulant, (3) obtaining a sea clutter average value sequence, (4)calculating an estimated value of the logarithm cumulant, and (5) extracting parameter information. According to the method, a defect that the parameter estimation of sea clutter Pareto observed in asingle time cannot be carried out in the prior art is overcome, the applicable range of the parameter estimation is expanded, and the accuracy of the parameter estimation is improved.

A neural network interlayer activation value quantification method and apparatus, wherein the method comprises: acquiring an activation value of the previous activation layer of a neural network; obtaining an activation value limit of a current activation layer according to the activation value of the previous activation layer; determining respective activation values of the current activation layer according to the activation value limit of the current activation layer and the activation value of the previous activation layer. Therefore, the activation value information of the previous activation layer can be transmitted to the current activation layer. Compared with the fixedly set limit in the prior art, the technical solution provided by the present invention can dynamically adapt to the change of the activation value according to the activation value of the previous activation layer, thereby reducing the estimated deviation of quantization.

The invention relates to a method for building a path loss model in a short-distance internet things environment, comprising the following steps of: building an original test data optimizing and acquiring scheme of a short-distance internet things environment path loss model; building the path loss model by a novel path loss model building scheme for evaluating short-distance internet things environment path loss characteristic; and evaluating a path loss model parameter by a novel path loss model parameter evaluating scheme for evaluating the short-distance internet things environment path loss characteristic. According to the method, the short-distance internet things environment path loss characteristic can be effectively evaluated, and the path loss evaluating deviation can be obviously reduced.

The invention discloses a signal to noise ratio estimation method under a frequency hopping communication interference condition. The signal to noise ratio estimation method comprises the following steps: at first, carrying out matched filtering on a received signal by a matched filter under the frequency hopping communication interference condition, transmitting the received signal into an energy detector to detect the energy, predicting an interference hop, and demodulating the interference hop; respectively sending the demodulated signal to an LDPC iterative decoder and a signal to noise ratio estimator, iteratively decoding the demodulated signal by the LDPC iterative decoder, sending the iterative decoding information of the signal to the signal to noise ratio estimator, estimating a signal to noise ratio through an estimation algorithm in combination with the demodulated signal information, sending an estimated value to the LDPC iterative decoder to carry out next iterative decoding, repeating the process until reaching an iteration time, and finally, outputting the estimated value. The signal to noise ratio estimation method disclosed by the invention is used for solving the problem of signal to noise ratio estimation under the conditions of high estimation precision, wide signal to noise ratio estimation range, low signal to noise ratio and very small estimation error, and plays a positive role of improving the channel state estimation accuracy during slow frequency hopping communication self-adaptive transmission.

The invention discloses a stable graph structure learning method and device, and relates to the field of machine learning, and the method comprises the steps: obtaining a graph structure implied witha biased relation in each environment through a data-driven mapping method according to a plurality of given biased sampling environments; obtaining the probability that each point in the graph structure is associated with a symbiotic data sample according to the graph structure of each environment and a multi-environment data generation mechanism; initializing a target graph structure, randomly sampling a symbiotic data sample in a plurality of environments, setting conditions for the symbiotic data sample, respectively obtaining a generation probability in each biased environment and a generation probability under an implicit relationship of the target graph structure, and taking the generation probability of the target graph structure as a mean value of the generation probability of each environment; and carrying out loop iteration until a convergence condition is satisfied so as to obtain a stable graph structure. According to the method, estimation deviation of parameters in a graph structure is reduced, and stable performance is kept in a complex and changeable test environment.

The invention discloses a signal processing method which comprises the following steps: the strengths of user signals with different arrival angles of a user are determined according to directional vectors of the user signals with different arrival angles of the user, a downlink forming vector corresponding to the user and directional gains of a receiving antenna at different arrival angles; and an azimuth estimation value of the user is determined according to the strengths of the user signals with different arrival angles. Another signal processing method comprises the following steps: the azimuth estimation value of the user is determined; the deviation of the azimuth estimation value of the user is estimated according to the difference between two user signal arrival angles corresponding to a given attenuation threshold; and the azimuth estimation value of the user is corrected by using the estimated deviation. The two methods provided by the invention can reduce or even eliminate the deviation between the azimuth estimation value and the true value of the user.

The invention discloses a variable bandwidth active disturbance rejection control method, which comprises the following steps of: designing a linear extended state observer with a fixed bandwidth fora first-order controlled object; constructing a variable bandwidth extended state observer on the basis of the linear extended state observer with the fixed bandwidth; establishing an active disturbance rejection control law, and constructing a variable bandwidth active disturbance rejection controller based on the variable bandwidth extended state observer; and adjusting parameters of the variable bandwidth active disturbance rejection controller until a satisfactory control effect is obtained. According to the variable bandwidth active disturbance rejection control method of the invention, the bandwidth of the observer can be adjusted in real time according to the change of the estimation deviation, so as to improve the estimation capability of the observer for time-varying disturbance or sudden disturbance, reduce the estimation deviation, and improve the control precision.