33 results about "Unscented transform" patented technology

The invention discloses a power generator dynamic state estimation method based on unscented transformation strong tracking filtering. The power generator dynamic state estimation method comprises two steps, namely, prediction and filtering, wherein in the prediction step, sigma point sampling is carried out through the symmetrical sampling strategy according to a filtering mean value and a filtering covariance matrix at a previous moment, a predicted measurement calculation value is calculated, a residual equation is obtained, and a fading factor is introduced to correct a predicted covariance matrix; in the filtering step, a gain matrix is adjusted in online mode, and estimate values of the power angle and electrical angle speed of a power generator in the electromechanical transient process are obtained after correction. Compared with unscented Kalman filtering and strong tracking filtering, the power generator dynamic state estimation method provided by the invention is high in tracking speed, tracking precision and noise robustness.

The invention relates to a state estimation method based on high-order unscented Kalman filtering. A high-order unscented Kalman filter is used for finishing the state estimation task in the target tracking process. According to the state estimation method based on the high-order unscented Kalman filtering, the state estimation task in the target tracking process is finished by the high-order unscented Kalman filter. In the target tracking process, the state equation and the measurement equation of target tracking are established; a sigma point required for the target tracking filter is obtained by high-order unscented transformation, and the weight of the sigma point is calculated; and the state estimation is obtained by iterating the sigma point and the weight of the sigma point to realize the real-time tracking of the target. The tracking precision of the state estimation method is higher than those of the existing target tracking methods based on other filters, a proper performance parameter k is selected to further improve the precision of the proposed high-order unscented Kalman filtering (UKF) target tracking method, and the high-precision real-time tracking to the target is realized. The state estimation method disclosed by the invention is applied to the technical field of the target tracking.

The invention relates to a self-contained navigation method for measurement of an earth ultraviolet sensor, which comprises the following steps of: initializing: utilizing satellite injection information to supply a state initial value and a covariance initial value outlier of a satellite at a starting moment, accumulating geocentric vector information and earth apparent radius information obtained by the ultraviolet sensor for a specified period of time, and then utilizing an outlier eliminating algorithm to eliminate the outlier in the information; batch processing: performing a least square method on the geocentric vector information and the earth apparent radius information after the outlier is eliminated, and calculating a state estimation Xk and a covariance estimation Pk of the satellite at an ending moment of observation; unscented transformation: utilizing the unscented transformation to realize the time updating of the state and the covariance, obtaining a state forecast X(k+1) and covariance forecast P(k+1) of the satellite at the next ending moment of observation; and recursion: continuously accumulating the geocentric vector information and the earth apparent radius information for a period of time, and repeating the steps and performing continuous recursion and calculation, thereby continuously obtaining the estimation of the state of the satellite. According to the method, the precision and the stability of a navigation system are increased.

The invention provides a smoothing constraint unscented Kalman filtering method. The smoothing constraint unscented Kalman filtering method comprises the following steps: 1, acquiring an original prior probability density function of a target state at a current target observation moment according to unscented transformation; 2, calculating a mean value and a variance of an original prior probability through numerical expectation; 3, introducing noise constraint information, and calculating the center of the approximate feasible region to obtain corrected prior probability density; 4, seeking aGaussian distribution mean value and a variance meeting the constraint condition through posteriori iterative optimization, and generating a new Gaussian sigma point meeting the constraint condition;and 5, carrying out weighted calculation on the Gaussian sigma points to complete a filtering process. The smoothing constraint unscented Kalman filtering method has the advantages in the aspects ofaccuracy and robustness, and meanwhile, the real-time performance of the smoothing constraint unscented Kalman filtering method is superior to that of a particle filtering algorithm. Correspondingly,the invention further provides a target tracking method.

The invention discloses a robot pose positioning method and a system, and relates to the field of robot positioning, wherein the method comprises the following steps of: acquiring IMU odometer data asa local reference system, acquiring a pose state vector and a covariance matrix at the previous time according to the local reference system, sampling the pose state vector at the previous time, performing unscented transformation to sampling points, predicting the pose state vector and the covariance matrix at the previous time subjected to unscented transformation by using a system model to obtain a prediction value at the current time, carrying out filtering treatment to the prediction value at the current time according to the actual measurement value to obtain the relative pose measurement value at the current time; after filtering, obtaining the global pose estimation at the current time according to the relative pose measurement value at the current time through coordinate transformation; and carrying out robot pose positioning according to the global pose estimation value at the current time. The unscented Kalman filter algorithm is combined with IMU odometer data and actual measurement values collected by GPS satellites or vision systems to obtain global pose estimation values, which are robust to complex environments and improve positioning precision.

The invention is applied in the technical field of data fusion, and provides a method for performing space registration on a radar and an infrared sensor configured on an identical platform. The method comprises the following steps of: establishing a system state equation and a measurement equation taking polar coordinate positions, rectangular coordinate velocity and polar coordinate registration deviations as state vectors; performing unscented transform (UT) on the measurement, performed at previous moments, of the radar and the infrared sensor to obtain initial values of the system state vectors; and performing unscented Kalman filtering (UKF) registration on measured data according to the system state equation, the measurement equation and the initial values of the state vectors to obtain the space registration deviation of the infrared sensor relative to the radar. By the method, limitations to the conventional space registration of the radar and the infrared sensor are avoided,and registration accuracy is improved.

The invention discloses a constraint multi-model filtering method based on L1 regular unscented transformation, and the method comprises the steps: providing a first constraint for the measurement truncation of a system, and integrating the first constraint and the latest measurement sequence of the system into the prior updating process of the system; smoothing between system observation intervals providing a second constraint of the system, and weighting correction and compensation of the space-time constraint by a fuzzy logic algorithm; approximating a feasible region by a heuristic optimization method, wherein a low-dimensional manifold model is modeled through a constraint auxiliary particle filtering framework, and unscented Kalman smoothing filtering is carried out on a model statebased on L1 regularization.

The invention discloses a method for estimating ammonia coverage and inlet ammonia gas concentration of a diesel engine SCR system. Firstly, a three-order state spatial model of the SCR is establishedaccording to an inner chemical reaction mechanism of the SCR system. Secondly, according to the established system model, the ammonia coverage and the inlet ammonia gas concentration in the SCR system are estimated by means of an SRUKF algorithm. Based on the SRUKF algorithm, through acquiring the Sigma point, traceless deformation is used for processing nonlinear transmission of a mean value anda covariance. No linear neglecting of high-order items is realized, and high estimating precision of the algorithm is obtained. Accuracy in estimating the ammonia coverage and the inlet ammonia gas concentration can be improved. Furthermore a problem of incapability of acquiring the ammonia coverage by means of a physical sensor is settled, and urea injection closed-loop control of the SCR is facilitated. The method can accurately estimate the inlet ammonia gas concentration in the SCR system, thereby reducing use of ammonia gas concentration sensors and reducing use cost.

The embodiment of the invention discloses a voltage stability evaluation method and device considering wind power output. The method comprises the steps of obtaining historical data in a power system,and calculating the mean value and covariance of an input random variable in the historical data; inputting the random variable into a deterministic voltage stability evaluation model by adopting a hypersphere unscented transformation algorithm to obtain a covariance matrix of an input variable and an output response; inputting the covariance matrix into a global sensitivity algorithm model to obtain an important input variable influencing the voltage stability; and inputting important input variables influencing the voltage stability into the deterministic voltage stability analysis model byadopting a symmetric unscented transformation algorithm, and outputting a probability static voltage stability analysis result. According to the method, the unscented transformation algorithm and global sensitivity analysis are combined, so that the speed and precision of probability static stability analysis are improved.

The invention belongs to the technical field of satellite attitude determination, and discloses a satellite sensor and gyroscope combined attitude determination method, a combined attitude determination system and application, and the method comprises the following steps: calculating a satellite attitude at a current moment k through an attitude estimation result of filtering at a k-1 moment and a satellite angular velocity measured by a gyroscope; performing unscented transformation on the mean value of the satellite attitude errors to obtain a sampling point; constructing a nonlinear system equation, performing nonlinear state updating on the sampling point, and performing cholesky updating to obtain a state prediction mean value and a variance square root; constructing a linear measurement equation, and performing linear measurement updating according to the predicted attitude error mean value; using Kalman filtering to fuse observation information and a predicted value to estimate an attitude error; and compensating the attitude estimation result by using the attitude error. According to the method, linear measurement updating is used for replacing nonlinear measurement updating of the system in combination with an application scene determined by the attitude, so that the calculation amount of the SRUKF is reduced, and the filtering efficiency can be improved.

The invention discloses a phase noise compensation method, module and system of a CO-FBMC system, and belongs to the field of optical communication. The compensation method comprises the following steps: performing first unscented transformation on posterior estimation of phase noise at a previous moment to obtain an initial Sigma point set, approximating a priori estimation value of the phase noise at the current moment according to the initial Sigma point set, performing second unscented transformation on the priori estimation value to obtain a multiplied Sigma point set, substituting the multiplied Sigma point set into a nonlinear equation for iteration, and obtaining a multiplied Sigma point set; and solving to obtain an estimated mean value and a covariance of the FBMC symbol at the current moment, calculating a Kalman gain matrix according to the estimated mean value and the covariance, and obtaining a posteriori estimated value of the phase noise at the current moment according to the Kalman gain matrix. The invention also provides an unscented Kalman filter compensation module and a CO-FBMC system. According to the method, a nonlinear equation does not need to be approximated to a linear equation, and the method has higher phase noise tolerance under the same precision. Meanwhile, a Jacobian matrix does not need to be solved, and the calculation amount is greatly reduced.

The invention discloses a dynamic state estimation algorithm for a power distribution network, which comprises the following steps of: firstly, controlling the distribution of Sigma sampling points by adaptively adjusting a proportional correction factor of unscented transformation and introducing a fading factor to adjust Kalman filtering gain on line so as to improve the quality of an importance density function, and then, realizing state estimation by using a particle filtering theory. The improved unscented Kalman filter (UKF) algorithm is combined with the particle filter (PF) algorithm to form a power distribution network state estimation algorithm with stronger robustness and higher filtering precision, and the number of sample particles is reduced to achieve state estimation with higher precision requirements, so that the algorithm time consumption is reduced; when the system has bad measurement data, the bad measurement data can be quickly sensed, and the predicted covariance is corrected through the introduced fading factor, so that the filtering gain is adjusted, and the robustness of the algorithm is enhanced.

The invention discloses a maneuvering target tracking method and device and a computer readable storage medium, and the method comprises the steps: constructing a minimum fuzzy error entropy criterion based on a fuzzy membership degree of an index representing the influence of different error samples on a state estimation result; obtaining state and prior estimation of the corresponding covariance by using an unscented transformation framework, and observing and prior estimation of the corresponding covariance; constructing a linear regression model based on state and prior estimation of corresponding covariance, observation and prior estimation of corresponding covariance, and obtaining error information through system reconstruction; based on a minimum fuzzy error entropy criterion and error information, a posteriori estimate of the state and the corresponding covariance is calculated. Through the implementation of the method, the fuzzy membership degree is introduced to adaptively adjust the weight and derive the minimum fuzzy error entropy unscented filtering model according to different error sample conditions, and a nonlinear system under a complex non-Gaussian noise signal shows good estimation performance, has a good target tracking effect and shows relatively high stability.