84 results about "Model filter" patented technology

Embodiments of the invention can provide systems and methods for initializing dynamic model states using a Kalman or similar type filter. In one embodiment, an adaptable model-based control system for controlling a gas turbine engine is provided. The system can include at least one sensor adapted to obtain dynamic-type information about a current state of the engine. Furthermore, the system can include an engine model adapted to receive information from the sensor, and further adapted to reflect the current state of the engine. In addition, the system can include a model filter adapted to initialize the model with at least a portion of the dynamic-type information, wherein at least one value based at least in part on the dynamic-type information is input to the engine model. Moreover, the model can be further adapted to determine an output from the engine model based at least in part on the at least one value. Further, the system can include a controller adapted to determine an engine control action based at least in part on the output from the engine model, and further adapted to output a control command to implement the engine control action.

A method is disclosed for combining seismic data sets. This method has application in merging data sets of different vintages, merging data sets collected using different acquisition technologies, and merging data sets acquired using different types of sensors, for example merging hydrophone and geophone measurements in ocean bottom seismic data. In one embodiment, a desired data trace is to be determined from a set of measured data traces, and the following steps are applied: (a) model filters are constructed which express the deterministic relationship between the desired data trace and each available measured trace that depends on the desired data trace; (b) the noise properties associated with each measured data trace are determined; (c) a sufficient statistic for the desired data trace is formed by application of an appropriate filter to each measured trace and summing the filter outputs; (d) the sufficient statistic is further processed by a single-input single-output estimator to construct an estimate of the desired data trace from the sufficient statistic.

Computing systems and methods for improving imaging of collected data are disclosed. In one embodiment, a first wavefield is propagated to obtain a first wavefield history; the first wavefield is again propagated to obtain a second wavefield history, wherein the propagation includes integration of one or more Q-effects; a first attenuated traveltime history is estimated based at least in part on the first and second wavefield histories; a first Q-model filter is calculated based at least in part on the first estimated attenuated traveltime; and a first adjusted wavefield is generated based at least in part on application of the first Q-model filter to the first wavefield. In some embodiments, an image is generated based at least on a first adjusted wavefield and a second wavefield.

The invention discloses a combination navigation filtering method of a multi-model underwater vehicle. The navigation filtering method provided by the invention comprises the following steps of firstly establishing a state equation, an observation equation and a noise equation of a SINS/DVL/TAN/MCP combination navigation system according to a underwater vehicle combination navigation system; determining a model set according to a system equation and a noise model; selecting characteristic variable from the combination navigation system, and establishing a bayesian network; and correcting the model switching probability in multi-model estimation by adopting a bayesian network parameter according to a multi-model filtering algorithm structure, and calculating the estimation fusion of a filter in a weight sum manner. The data processing and resolving operations of the combination navigation are finished by a navigation computer according to a filtering model and an algorithm flow of the combination navigation system. The navigation filtering method provided by the invention has the advantages of being capable of improving the filtering accuracy of the combination navigation system under a complicated environment, and strengthening an autonomous navigation positioning property of the underwater vehicle.

The invention relates to a modified variable-structure grid interaction multi-model filtering method, and provides a modified variable-structure grid interaction multi-model filtering method for tracking a hypersonic-speed target of near space. By the modified variable-structure grid interaction multi-model filtering method for tracking the hypersonic-speed target of the near space, the problem that the existing single-model filtering algorithm, the existing fixed-structure interaction type multi-model algorithm and the traditional variable-structure interaction multi-model algorithm cannot be used for quickly tracking a hypersonic-speed maneuvering target of the near space precisely is solved. According to the technical scheme, the modified variable-structure grid interaction multi-model filtering method for tracking the hypersonic-speed target of the near space comprises the following steps of (1) establishing an inertial reference coordinate system and establishing a state equation of maneuvering movement of the target in the inertial reference coordinate system; (2) using a current statistics model of the maneuvering target as a central model and using constant-speed turning models as a left turning model and a right turning model; (3) determining a measurement model of a target tracking system on the basis of the inertial reference coordinate system; and (4) performing state estimation and error covariance matrix fusion. The modified variable-structure grid interaction multi-model filtering method for tracking the hypersonic-speed target of the near space is used for the field of aircrafts.

The invention discloses a combined agricultural robot navigation information fusion method based on rank filtering. Interactive multi-model filtering is employed as an algorithm structure, an initial value of present-moment filtering is calculated based on an output result of each model filter at a previous moment, a rank filter is taken as a sub filter of model filtering steps to carry out filtering, a residual error and a residual error covariance of model filtering are utilized to update model probability, weight fusion of each rank filter estimate is carried out to acquire an interaction output result. Combined navigation data processing and resolution work are accomplished by a navigation computer according to a combined navigation system filtering model and an algorithm flow. The method is advantaged in that filtering precision of the combined navigation system under the complex environment can be improved, and an agricultural robot is made to have advantages of wide application scope and strong interference prevention capability.

The invention provides a filtering method for the multi-strategy maneuvering tracking, and the method comprises the steps: receiving target detection information, employing a current statistical model, introducing a time-varying fading factor, and adjusting the gain in real time according to the motion condition of a target, so as to improve the estimation performance of the current acceleration of the target, and achieve the pre-estimation of the acceleration of the target; carrying out the comprehensive processing of the maneuvering factors of the target, calculating a maneuvering value in acurrent motion state, carrying out the mapping of the maneuvering value to a model set, so as to divide the model set in different maneuvering states and obtain a current output weight of each modelthrough a fuzzy inference system; carrying out the filtering output through an interactive multi-model filtering process, obtaining the current position information and motion information, and predicting and outputting the position information at a next moment. According to the invention, a filtering structure is optimized through the reasonable reduction of a redundant model, the tracking precision is improved, and the method is suitable for a large-range maneuvering tracking of the target.

The invention discloses a space-target positioning method based on multi-model filtering. The space-target positioning method comprises the following steps: firstly, selecting the position and the velocity vector of a space target as state variables, selecting different system noise variance matrixes, and establishing a plurality of models for describing the uncertainty of the orbital maneuver acceleration of the space target; secondly, establishing a plurality of parallel-extension Kalman filtering algorithms for the plurality of models, and respectively carrying out filtering to obtain a plurality of state estimated values; thirdly, calculating corresponding filtering weights according to the conforming degree of all the state estimated values and observed quantities; and finally, calculating the weighted sum of the state estimated values of all the parallel-extension Kalman filtering algorithms according to the filtering weights to obtain navigation filtering results, namely the estimated values of the position and the speed vector of the space target. The space-target positioning method disclosed by the invention has the advantages that the system noise variance matrixes playing a leading role can be selected in a self-adaptive manner, so that the tracking and positioning capabilities of a filter to the maneuver target is enhanced, and the space-target positioning method is favorable for solving the problem that the uncertainty of the maneuver acceleration influences the positioning accuracy of the space target.

The invention discloses a monitoring method for satellite integrity based on assistance of a dynamical model, belonging to the technical field of integrated navigation and fault diagnosis. According to the monitoring method provided by the invention, the dynamical model of a motion carrier is utilized to be combined with the other carried navigation sensors, for realizing the monitoring for the autonomous integrity of a user terminal of a satellite navigation system. According to the monitoring method provided by the invention, a multi-model filtering frame is adopted for realizing the fusion and state estimation for information of an inertia/dynamical model/satellite navigation system; a multi-solution separation method is adopted for realizing the detection and separation for satellite fault; no external equipment is required; the integrity monitoring capacity of satellite navigation system can be promoted; the navigation precision and reliability of the carrier can be ensured; the monitoring method has the characteristics of low cost, strong autonomy, and the like.

The invention relates to a pulp model filtering mould manufacturing method, in particular to a pulp model filtering mould manufacturing method based on the SLS quick-forming theory. The pulp model filtering mould manufacturing method based on the SLS quick-forming theory includes the steps: step 1. the design of filtering vent; step 2. the generation of an STL file; step 3. the processing of STL files: a data processing software named Magics is adopted to process the STL file of pulp model filtering mould to generate a CLI slice file used for quickly forming the pulp model filtering mould; step 4. the conversion of the CLI slice file: the CLI slice file of the pulp model filtering mould is converted to an identifiable file of a quick-forming machine and then the converted file is inputted into the quick-forming machine; step 5. the selection of sintering material, polystyrene powder or nylon powder is selected as sintering material; step 6. the setting of sintering process parameters and manufacturing; step 7. the reinforcement process: resin is painted onto surfaces of prepared prototype parts to reinforce the strength of prepared part to form the filtering mould, so as to enhance the accuracy and efficiency of pulp model filtering mould manufacturing process.

The utility model discloses an automatic alarm refresh method in the network administration system. The method comprises the following steps as the network administrator obtains and keeps all of the alarms from the existing equipment as initializing; the network administration equipment submits the alarms; the network administration judges each alarm message and makes the filtration treatment for each alarm as per the judgment; the filtered alarm is submitted to the network administration and the network administration interface is refreshed. As compared with the existing technology, the utility model filters the repeated and invalid alarm as well as guarantees the synchronization of the server's alarm and the equipment's current alarm. Thus, the alarm will not be lost under any communication conditions, so as to improve the confidence level of the network administration interface; meanwhile, the network administration equipment can report the changed alarm instead of all alarms. Therefore, the communication efficiency of the network administration equipment is effectively improved.

The invention discloses a hull deformation angle estimation method based on interactive multi-model filtering. The method comprises the steps that firstly, a basic model of hull deformation angle estimation is set up, and then interaction models are set up; strapdown inertial navigation systems are set on a bow and a stern respectively, and hull motion information is measured; the hull motion information is processed based on a filter of the interaction model, and the hull deformation angle is estimated. According to the method, the interaction models are set up to cover the dynamic change range of the deformation angle, and the method is suitable for estimating the hull deformation angle under the model parameter uncertainty condition in the dynamic environment.

The invention discloses a real-time passenger flow forecasting method and a forecasting system based on rail transit. Firstly, historical passenger flow data of four different periods of ten-day, week, day and hour are obtained. The historical passenger flow data of four different periods are preprocessed to obtain the time series set. Secondly, the ten-day time model, week time model, day time model and hour time model are established according to the time series set. Ten-day-time model, week-time model, day-time model and hour-time model are used to forecast passenger flow, and the first forecast data set, the second forecast data set, the third forecast data set and the fourth forecast data set are obtained respectively. Then, an interactive multi-model filtering algorithm is used to determine a passenger flow prediction model according to the first prediction data set, the second prediction data set, the third prediction data set and the fourth prediction data set. Finally, the real-time passenger flow of rail transit is forecasted according to the passenger flow forecasting model, which improves the accuracy of real-time passenger flow forecasting based on rail transit.

The invention provides a signal processing method and an RFID reader, relates to the field of radio frequency identification in the Internet of things, and can realize generation of PR-ASK radio frequency signals which conform to the ISO/IEC 18000-6C standard and are ideal in signal quality on the basis of reducing the hardware overhead as far as possible. The signal processing method comprises the steps of receiving a binary original data stream sent by a controller; performing pulse interval encoding (PIE) on the original data stream to obtain a PIE encoded data stream; performing duobinaryencoding on the PIE encoded data stream to obtain a duobinary encoded data stream; performing n-times data interpolation on the duobinary encoded data stream, performing filtering through raised cosine roll-off modeling filter to obtain a first modulation signal; performing waveform adjustment on the first modulation signal by using a cosine waveform template to obtain a second modulation signal;and enabling the second modulation signal to pass through a mixer to obtain a phase reversal amplitude shift keying (PR-ASK) radio frequency signal.

The invention discloses a maneuvering extended target hybrid interactive strong tracking filtering method driven by incomplete measurement data, and the method comprises the following steps: 1, building a hybrid model set, and determining a transition probability matrix of the hybrid model set; 2, calculating an input initial value of the sub-model filter at a k sampling moment after interaction;3, constructing a sub-model filter; 4, updating the model probability of the model at the k sampling moment; 5, calculating state estimation, namely covariance, of the k sampling moment, judging a termination condition, and if the target tracking process is finished, exiting; otherwise, returning to the step 2. The extended target tracking algorithm provided by the invention can effectively adaptto an incomplete measurement environment in actual engineering, has higher tracking precision, and is faster and more accurate in model switching at a target maneuvering moment.

The invention provides a convolutional neural network (CNN) initialization method based on pre-training model filter extraction and relates to the technical field of video processing. The method extracts a filter parameter in a pre-training model by using a minimum entropy loss and a minimum reconstruction error in order to initialize a target task network model and achieve the small and medium-sized network initialization that meets practical application problems. By using the minimum entropy loss and the minimum linear reconstruction method, the method extracts the filter parameter from thepre-training model to initialize the target task network model, does not require that the target task network structure is consistent with the pre-training network structure, flexibly design the network structure of the target task according to practical applications so as to meet the memory overhead and calculation speed requirements in practical application problems.

The invention discloses a security defense method and application for distributed federal learning based on an end-side cloud architecture, and the method comprises the following steps: an edge server receives an updated local model uploaded by a receiving end device, and the updated local model is obtained by training a global model issued by a cloud server by an end device based on private data; the edge server filters the updated local model to obtain a secure local model; and the edge server aggregates the filtered security local model, and uploads the generated edge aggregation model to the cloud server. According to the method, malicious models can be eliminated through a model filtering algorithm, the malicious models are aggregated into a security model, and indirect poisoning attacks (such as label flipping attacks aiming at a data set) aiming at a global model are continuously defended online.

The invention discloses an MEMS (micro-electromechanical system) gyroscope random error modeling filtering method based on modified EMD (empirical mode decomposition), which comprises extracting an IMF (intrinsic mode function) component from an original signal through the EMD algorithm; modeling based on the extracted IMF component; subjecting a model formed by modeling to Kalman filtering, and compensating MEMS gyroscope random errors in real time. Measurement precision of a MEMS gyroscope can be improved.