81 results about "State covariance" patented technology

A target tracking method uses sensor(s) producing target signals subject to positional and/or angular bias, which are updated with sensor bias estimates to produce updated target-representative signals. Time propagation produces time-updated target states and sensor positional and angular biases. The Jacobian of the state dynamics of a target model produces the state transition matrix for extended Kalman filtering. Target state vector and bias covariances of the sensor are time propagated. The Kalman measurement residual is computed to produce state corrections, which are added to the time updated filter states to thereby produce (i) target state updates and (ii) sensor positional and angular bias updates. The covariance of a state vector comprising target states and sensor positional and angular biases is propagated, producing measurement updated state covariance including (i) target position and velocity measurement covariance updates and (ii) the sensor positional and angular bias measurement covariance updates.

A computerized method generates unbiased estimates of parameters such as position, velocity, and acceleration of a vehicle. The method assigns initial vehicle states. Unbiased vehicle state estimates are obtained using a model incorporating a unique parameterization of the vehicle motion and state dynamics, and sensor registration bias states. Sensor measurements of the vehicle are used to iteratively update the vehicle states and covariances. Previous vehicle state and covariance, are propagated to produce current state and covariance estimates. If needed, a new vehicle model is adopted. Sensor registration biases are estimated to generate corrected measurement information. A measurement matrix and a measurement residual is determined. The Kalman gain matrix is determined, and used with measurement residual for state update. A new current state is generated. State covariance update is determined the same way, and used to generate a new current state covariance. These steps are repeated, producing unbiased vehicle parameter estimates.

The invention discloses a track method before locomotive weak target detection based on multimode grain filtering and data association, which belongs to the field of radar data processing. The method aims at solving locomotive weak target detection and track problems under low signal-to-noise ratio conditions, provides target track information while achieving the locomotive weak target detection, and effectively eliminates the target leakage problem caused by the low signal-to-noise ratio and big locomotive performance. The method mainly comprises the following steps of (1) using the multimode grain filtering to obtain target states and state covariance estimation of each time; and (2) redefining the results of the multimode grain filtering to be a measuration value, and using a state estimation-track data association to give out the target track information. The track method overcomes limitations of a track method before locomotive weak target detection based on multimode grain filtering, guarantees continuity of target tracks by effectively reducing false dismissal probability, simultaneously has the advantages of being simple in structure, easy to achieve hardware and the like, and has strong engineering application value and popularization prospects.

The invention discloses an integrated parameter estimation method for attitude and orbit of a spatial non-cooperative target based on depth learning. The parameter estimation algorithm models the kinematics and dynamics of the space non-cooperative targets by dual vector quaternion, and designs the corresponding state BP neural network parameter estimation algorithm and state covariance matrix convolution neural network parameter estimation algorithm on this basis. The attitude and orbit parameters of space non-cooperative targets are estimated integrally by using dual vector quaternion, and the attitude and orbit coupling effect of space non-cooperative targets is considered. At the same time, BP neural network with single hidden layer and convolution neural network with double hidden layers are designed to estimate the parameters of spatial non-cooperative targets under the condition of measurement failure, so that the parameter estimation algorithm has strong robustness to the spatial environment.

The invention provides a multi-source monitoring data fusion method and device. The method includes the steps that a unit area in which a target is located is positioned, and static weight of a plurality of monitoring sources corresponding to the unit area is acquired; according to the real-time quality monitoring result, a system state covariance estimated value and the last-period fusion effect, the dynamic weight corresponding to the unit area is acquired; according to the static weight and the dynamic weight, comprehensive weight corresponding to the unit area is acquired; multi-source monitoring data are fused in the unit area according to the comprehensive weight. According to the technical scheme, real-time states of all monitoring sensors and difference conditions between a single-source data result and a multi-source data fusing result are taken into consideration, changes of precision of radar are effectively tracked in combination with a fused algorithm of the static weight and the dynamic weight, and the real-time performance is guaranteed.

The invention relates to an RFID and UKF based method for rapidly tracking an indoor target. A motion model is established according to the actual motion characteristics of the target, and an RFID measurement model is established according to the RFID measurement mechanism; a target motion characteristic is predicated according to the established motion model, and a target state predication value is obtained; a target state modification value is calculated by a UKF according to an original target measurement value and a target measurement predication value, then, the target state predication value is modified, and the target state estimation value is obtained; a target state covariance estimation value is calculated, and the error between the target state estimation value and a target state true value are measured; and meanwhile, a system self-adaption parameter is updated in real time according to the target state predication value, further, the motion model is updated, the process is repeated, and the target state predication value of the next time point is predicated according to the updated motion model, so that the indoor target is rapidly tracked, and the tracking accuracy is improved.

The invention discloses an integrity monitoring method for a slowly growing ramp fault of integrated navigation. For a noise-level slowly growing ramp fault in a conventional integrated navigation system, the method comprises constructing recursive exponential integrity detection statistics by means of a process measurement matrix of satellite navigation-and-inertial navigation-integrated navigation extended Kalman filtering and a system state covariance matrix, setting a recurrence time interval and comparing the detection statistics with a detection threshold under the premise of meeting therequirement on a false alarm rate and a missed alarm rate in navigation operation, and ultimately fulfilling integrity monitoring and sending alarm information to a user in time. The integrity monitoring method for the slowly growing ramp fault of integrated navigation, proposed by the invention, is applicable to single-satellite navigation systems, multi-constellation navigation systems and integrated navigation systems, can realize the integrity of complete robustness of an integrated navigation receiver under a complex electromagnetic environment, and has high theoretical value in integrated navigation receiver designs and engineering application value.

The invention provides an improved active section trajectory estimation algorithm. The method comprises the steps of calculating a thrust acceleration template according to the initial mass of a missile target, the air injection speeds of all stages of engines, the shutdown time and the like; Widening the thrust direction angle into a state vector component, and establishing a fine parameterized dynamic model by utilizing the widened state vector and a thrust acceleration template; Establishing an early warning satellite detection model according to the position vectors of the missile target and the two early warning satellites in the earth fixed connection coordinate system and the coordinate transformation matrix from the earth fixed connection coordinate system to each early warning satellite orbit coordinate system; Estimating a missile target initial state vector and an initial state covariance matrix according to the first three groups of equivalent detection positions; Describing a thrust direction angle by utilizing a first-order Markov process, and constructing a process noise matrix adaptive to thrust direction angle change; And performing filtering processing by using aUKF filtering algorithm, and estimating the motion state of the missile target. The method can improve the precision of trajectory estimation of the ballistic missile detected by the early warning satellite.

The invention provides a binocular VIO implementation method based on a variational Bayesian adaptive algorithm, and the method comprises the steps: obtaining an image through a binocular camera, carrying out the feature point extraction of the image, and adding the extracted feature points into a globally maintained map container; carrying out IMU state prediction through the IMU data; when a newframe of image is received, adding the frame of image into the state vector to augment the state vector and the state covariance; judging whether feature points exist or whether the camera needs to be deleted, and if yes, performing filtering fusion; if yes, performing UT transformation based on an observation model of the binocular camera, and calculating a Jacobian matrix corresponding to the observation model; superposing the plurality of Jacobian matrixes, and performing null-space projection to obtain a final standard observation equation; and applying variational Bayesian estimation tothe obtained standard observation equation, and updating the state of the VIO system. The time-varying situation of system observation noise can be well processed, and the robustness is improved whilethe precision is improved.

The invention discloses a fuzzy model particle filtering method, device and equipment and a storage medium. The method comprises the steps that a T-S fuzzy model corresponding to a tracking target isconstructed; by using a preset strong tracking particle filtering algorithm, identifying a rear part parameter of the T-S fuzzy model to obtain a state updating value and a state covariance estimationvalue; through a preset fuzzy C regression clustering algorithm, identifying a front part parameter membership function of the T-S fuzzy model to obtain a front part parameter membership value; usingthe state updating value, the state covariance estimation value and the front part parameter membership value to update the T-S fuzzy model. Compared with the prior art, the method has the advantagesthat the tracking performance is better, and the target can still be effectively and accurately tracked when the tracked target suddenly changes directions or the dynamic prior information of the target is inaccurate.

The invention discloses an adaptive extended Kalman tracking method and device and a storage medium, and belongs to the technical field of satellite navigation. The method comprises that a carrier-to-noise ratio is estimated accurately in real time, a state noise covariance matrix and an observation noise covariance matrix at present time are determined by utilizing a pre-established nonlinear mapping relation according to the carrier-to-noise ratio at present time, further a system state vector and state covariance matrix of present time is obtained according to the state noise covariance matrix and an observation noise covariance matrix at present time, adjustment amount of carrier NCO and code NCO are generated according a system state vector of the present time, the local carrier waveand local pseudo random code are adjusted according to the adjustment amount of carrier NCO and code NCO, and a satellite signal is tracked in real time. Thus, a loop parameter is adjusted adaptivelyby using the carrier-to-noise ratio to match different scenes and signal environments, and the satellite signal tracking precision and sensitivity are improved greatly.

The invention discloses an interactive T-S fuzzy semantic model estimation method and system and computer readable storage medium, which are used for target tracking and solve the problem that the robustness and the accuracy of a research result are reduced because an existing filtering algorithm is difficult to meet the requirement of system performance, including defining different semantic fuzzy sets in a T-S fuzzy model by adopting different language values; setting a probability conversion method among the semantic fuzzy sets; carrying out fuzzy interaction on the initial state of the target to obtain mixed initial state estimation and mixed initial state covariance; carrying out hybrid initial state estimation and hybrid initial state covariance nonlinear filtering processing to obtain an updated state and an updated state covariance; performing calculation and updating on the precursor parameters of the T-S fuzzy model to obtain updated precursor parameters; calculating a standardized model probability; obtaining state estimation and covariance estimation of the target; and the motion state of the target is estimated, so that the robustness and accuracy of the research result are improved.

The embodiment of the invention provides a terminal positioning method and device, equipment and a medium, and the method relates to a navigation positioning technology, can be applied to lane-level positioning of an electronic map, and comprises the following steps: determining terminal state information through observation information, and screening out first screening observation information; setting the terminal state information as initial state information of a target filter, performing time updating on the target filter to obtain a prediction state parameter and a prediction state covariance matrix, screening out second screening observation information used for constructing a measurement updating matrix from the first screening observation information, and performing measurement updating on the target filter according to a measurement variance matrix constructed by the observation error factors, a measurement updating matrix, the second screening observation information and a prediction state covariance matrix, updating the prediction state parameters into target state parameters, and determining a terminal positioning result according to the target state parameters. According to the invention, the positioning accuracy of the terminal can be improved.

The invention relates to a navigation data filtering method, a navigation data filtering apparatus, a computer device and a storage medium. The method comprises the following steps: establishing and initializing an inertial system parameter error model, and setting the time serial number of a next filtering cycle as an assigned value; obtaining a state quantity predication value and a state quantity covariance prediction value with the time serial number being the assigned value; obtaining a measured value estimation value and a measured value covariance estimation value with the time serial number being the assigned value and a state quantity and measured value cross-covariance; calculating a Gaussian kernel coefficient; obtaining a state quantity posterior estimation value with the timeserial number being the assigned value; updating the state quantity covariance with the time serial number being the assigned value; compensating a parameter; and setting an assigned value to be valueof (the assigned value +1), and returning to the obtaining of the state quantity prediction value with the time serial number being the assigned value and the state covariance prediction value with the time serial number being the assigned value. The method in the embodiment of the invention is enforced to realize the capture of non-Gaussian noise high-order terms, improve the error estimation accuracy and improve the positioning precision.

The invention relates to a method for rapidly tracking an indoor target. The method comprises the steps as follows: establishing a movement model according to actual movement characteristics of the target, and establishing an RFID (radio frequency identification) measurement model according to an RFID measurement mechanism; predicting movement characteristics of the target according to the established movement model to obtain a target state prediction value; computing a target state correction value by an extended kalman filter according to an original target measurement value and a target measurement prediction value, and then correcting the target state prediction value to obtain a target state estimation value; computing a target state covariance estimation value, and measuring an error between the target state estimation value and a target state true value; and simultaneously, updating self-adaptive parameters of a system in real time according to the target state estimation value, then realizing updating of the movement model, repeating the process, and predicting a target state prediction value of a next time point by the updated movement model. Therefore, the rapid tracking of the indoor target is realized, and the tracking accuracy is improved.

The invention relates to a high-performance speech recognition co-processor and a method thereof for realizing co-processing. The high-performance speech recognition co-processor comprises a storage module and an output probability calculation module; wherein the storage module is used for storing a characteristic vector, a model state mean value vector, a model state covariance vector and a probability matrix obtained by calculation in a calculation process. The output probability calculation module comprises a mahalanobis distance calculation module and a logarithm domain addition calculation module, wherein the mahalanobis distance calculation module calculates a mahalanobis distance according to the data of the characteristic vector, the model state mean value vector and the model state covariance vector. The logarithm domain addition calculation module carries out logarithm domain addition calculation according to a calculation result of the mahalanobis distance calculation module, writes the probability matrix obtained by calculation into the storage modules for storage and outputs the probability matrix to an external processor. The co-processor and the method thereof for realizing co-processing, which are provided by the invention, can carry out output probability matrix calculation based on an HMM speech recognition algorithm, aim at the application of an embedded type speech recognition system and have the advantages of performance increase, cost reduction and power consumption reduction.

The invention provides a robust filtering method aiming at parameter uncertainty and observation loss. The robust filtering method comprises the following steps: 1, describing a question through a mathematical model, and modeling of norm-bounded parameter uncertainty and observation loss; 2, giving a filter model of undetermined parameters according to the mathematical model established in the step 1; 3, performing augmentation on an original state equation, solving a state covariance matrix after the augmentation, and constructing a novel state vector; 4, searching for an estimation error covariance of the original state vector according to the state covariance matrix after the augmentation in the step 3; and 5, setting the filter parameters, so that a minimum value is obtained at the upper bound of the estimation error covariance obtained in the step 4. Therefore, minimization of the upper bound of the state estimation error covariance is realized under the condition that a filter has the norm-bounded parameter uncertainty and observation loss.

The invention discloses a spacecraft attitude determination method based on central error entropy criterion unscented Kalman filtering, and the method comprises the steps: building a nonlinear system for spacecraft attitude determination according to spacecraft measurement data and a spacecraft attitude dynamic model; according to the state and the state covariance of the spacecraft at the previous moment, generating a plurality of Sigma points by using a preset sampling mode, establishing a time updating transfer formula, and obtaining a one-step prediction state estimated value and a one-step prediction state covariance of the spacecraft at the current moment; according to the one-step prediction state estimation value and the one-step prediction state covariance, generating a plurality of Sigma points by using a preset sampling mode, and obtaining a one-step prediction value, an auto-covariance and a cross-covariance of the measurement output quantity of the spacecraft; and establishing a linearized regression equation of the spacecraft state based on a center error entropy criterion, determining a cost function of center error entropy criterion filtering, and obtaining the state and the state covariance of the spacecraft at the current moment. According to the method, the attitude estimation precision and robustness during non-Gaussian noise processing can be improved.

The invention discloses a single-cell EKF (extended Kalman filter) and UKF (unscented Kalman filter) estimation comparing method. The single-cell EKF and UKF estimation comparing method includes placing an EKF estimation block diagram and a UKF estimation block diagram of a single lithium cell in the same simulation environment, discharging for 3000 s by 1C pulse discharging current, and measuring estimation results of EKF and UKF with the same initial state covariance value and the initial measurement covariance value; placing EKF estimation block diagrams and UKF estimation block diagrams of a lithium cell pack in the same simulation environment, discharging for 1400 s by constant pulse discharging current, and measuring estimation results of EKF and UKF with the same initial state covariance value and the initial measurement covariance value. The single-cell EKF and UKF estimation comparing method can overcome the defects of short service life, poor safety and reliability and the like of the prior art and has the advantages of long service life and high safety and reliability.

The invention relates to a temperature measuring method, in particular to a verification table temperature measuring method based on Kalman filtering. The method includes the steps that parameters of a Kalman filter are initialized; according to a previous state noise value and the optimal value, a current-moment numerical value is predicted, and Kalman gain is calculated; an innovation sequence is added and used for performing Kalman processing on data; an existing state numerical value is estimated according to a previous state numerical value, next state covariance is calculated, related data are updated, and the Kalman filtering optimal value is returned. The innovation sequence is added into Kalman filtering, and the anti-interference capacity of a large verification table temperature measuring circuit is improved for an industry site; Kalman filtering is very good in smoothness, and water temperatures of a verification table water inlet tube, a water return tube and a water tank can be accurately measured in real time; Kalman filtering is introduced into verification table temperature measuring of the heat metering industry, and the method has certain demonstration meaning and has a certain demonstration effect.