36 results about "Model filter" patented technology

The invention discloses a curve model filtering-based high-speed wire rotational flow pool liquid level correction control method and system, and relates to the technical field of steel production, and the method comprises the following steps: collecting a liquid level signal, and constructing a fixed curve model according to a process stage; a staged deviation zone is set according to the model, deviation is compared and calculated in real time, and abnormity judgment is carried out; performing previous normal value substitution, model prediction value substitution or triggering process alarm on the abnormal section according to the duration to obtain a substitution sequence; and performing composite filtering on the alternative sequence, and outputting a corrected liquid level for closed-loop control and visualization. Through four-stage fixed curve modeling and staged deviation band judgment, graded substitution is triggered by a pull-back amount, a phase drift amount and a logic consistency rate, and compound filtering closed-loop control is performed on a substitution sequence, so that accurate anomaly detection, low false alarm, continuous curve and stable control are realized.

The invention relates to the technical field of modeling prediction, and discloses a DSP integrated degradation and LSTM model optical module life prediction method, and the method comprises the steps: enabling a multi-parameter coupling degradation model and a Kalman filtering algorithm to be embedded into an optical module DSP at a calculation level through a DSP embedded mechanism model filtering LSTM network time sequence calibration collaborative architecture, and achieving the prediction of the life of an optical module. The in-situ, real-time and safe evaluation of the life state is realized, and the traditional mode of relying on external computing resources is changed; on the model level, the interpretability of a physical mechanism and the adaptive capacity of data driving are fused, fundamental degradation dynamics is described through a nonlinear coupling model, and a specific degradation rule is learned from individual historical data by using an LSTM network; finally, real-time, accurate and self-adaptive evaluation and prediction of the service life state of the optical module are realized.

The invention provides an unmanned cluster battle decision-making method and device. The method comprises the following steps: acquiring current state information and own resource information of a plurality of enemy target individuals in a cluster battle; based on the current state information of the plurality of enemy target individuals, estimating the state information of the plurality of enemy target individuals in the next time period by using an interactive multi-model filtering algorithm; inputting the state information of the plurality of enemy target individuals in the next time period and the resource information of the enemy target individuals into a constructed multi-target optimization model under a multi-constraint condition; solving the multi-target optimization model under the multi-constraint condition based on a multi-target optimization algorithm to obtain output target distribution information; and generating an own strike decision in different decision spaces according to the target distribution information. By adopting the scheme, the prediction of the enemy target individual is realized only by using the information observed in the confrontation, and the attack decision is obtained by comprehensively considering various constraint conditions in the battle, so that the method is more suitable for an actual confrontation scene.

The invention discloses a fault reasoning path optimization method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining associated data when an elevator has a fault, the associated data comprises a fault log and sensor data of the elevator, and the fault log comprises at least one current fault mode; extracting an association relationship between the current fault modes from the associated data through a preset large language model; updating a preset fault map according to the association relationship, and obtaining a fault link comprising a current fault mode from the updated fault map; and inputting the fault link and the sensor data into a preset reasoning model, so that the reasoning model screens out a fault reasoning path from the fault link. The incidence relation between the fault modes is extracted in combination with sensor data, the interaction effect of multiple factors in an elevator system can be considered, the situation that the reasoning ability of a reasoning model is limited in a complex fault scene is avoided, the fault map can adapt to the current environment, and the fault reasoning accuracy is improved.

The invention is applied to the technical field of Internet of Things equipment monitoring, and particularly discloses an Internet of Things equipment operation monitoring method and system based on artificial intelligence, and the method comprises the following steps: S1, data collection; s2, data preprocessing; s3, constructing a model; s4, monitoring and predicting; and S5, early warning and decision making. According to the Internet of Things equipment operation monitoring method and system based on artificial intelligence, the monitoring model based on the attention mechanism is constructed, so that the model can automatically focus key parameters which have relatively great influence on the operation state of the Internet of Things equipment, and thus the monitoring model filters invalid data interference; real-time monitoring data of the Internet of Things equipment are input into the monitoring model, the monitoring model is used for analyzing operation parameters of the Internet of Things equipment, accurate identification of fault features of the Internet of Things equipment and advanced pre-judgment of potential hazards are achieved, and output results of three states are output according to parameter data of the equipment according to output results of the monitoring model.

The invention provides a verification code generation method and device and a readable storage medium, and the method comprises the steps: generating an initial confrontation sample based on an original verification code sample; analyzing the input and output relation of the original verification code sample and the initial confrontation sample in the target black box model, and constructing a covariance matrix; performing iterative optimization on the initial adversarial sample according to the covariance matrix to obtain an optimized adversarial sample; and determining the optimized adversarial sample as a finally generated verification code. According to the method, the confrontation sample is generated based on target black box model feature extraction of the covariance matrix in combination with model filtering optimization, so that the user can easily identify the verification code, the identification difficulty of an artificial intelligence cracking model is increased, and the security and the user experience are both considered.

This invention provides a method for filtering air quality forecast results based on sliding optimal matching, relating to the fields of environmental air quality forecasting and data science technology. The method includes: acquiring multiple predicted air quality index data obtained from various air quality prediction models; acquiring multiple measured air quality index data at multiple times within a past first preset time period; setting a time window; filtering target air quality prediction models from among the multiple air quality prediction models; and obtaining air quality index forecast data. According to this invention, predicted air quality index data and measured air quality index data can be acquired, and error statistics can be performed based on the time window to filter target air quality prediction models. The sliding optimal matching algorithm enables real-time evaluation and automatic filtering of model performance, overcoming the technical bottlenecks of traditional manual operation and static integration, improving the real-time performance and efficiency of air quality prediction model filtering, and providing high-precision and high-efficiency forecast support.

The present invention discloses a computer-implemented method, a computer system, and a computer programming product, wherein the computer-implemented method uses dirty training data to clean a training data set of a neural network. The technique begins by accessing a labeled training data set, the labeled training data set including relatively dirty labeled data elements. The labeled training data set is divided into a first subset A and a second subset B. The process includes looping between the first subsets A and B, the process includes generating refined model-filtered subsets of the first subsets A and B to provide a cleaned data set. Each refined model-filtered subset can increase cleanliness and increase the number of elements.

The utility model discloses a filter device for feed additive production in the related technical field of filter devices, which comprises a filter cartridge and an upper sealing cover, a filter screen is arranged in the filter cartridge, the upper end of the filter screen is fixedly connected with a lower sliding rod, the upper end of the lower sliding rod is fixedly connected with a lifting matching plate, and the lifting matching plate is fixedly connected with the upper sealing cover. An upper sliding rod is fixedly arranged at the upper end of the lifting matching plate, an upper sliding matching seat is arranged on the outer side of the upper sliding rod, a lower sliding matching seat is arranged on the outer side of the lower sliding rod, a lifting matching sliding groove is formed in the lifting matching plate, and the middle of the lifting matching sliding groove is of an arc-shaped structure. The filter screen of a cone structure is arranged in the filter cylinder, the impurity containing groove of a turnover structure is formed in the outer side of the lower portion of the filter screen, filtered impurities can be effectively collected, meanwhile, the filter screen can do reciprocating lifting movement under the action of the lifting matching plate, and the probability that the impurities block meshes is further reduced.

The invention provides a model prediction control strategy and a modulation method for a three-phase four-wire three-level inverter, and the control strategy comprises the steps: building a mathematical model according to the working principle of the inverter, carrying out the discretization of the model through employing a forward Euler method, calculating the predicted values of the output voltage at the next two moments, and carrying out the prediction of the output voltage at the next two moments; an inverter side bridge arm voltage reference value at the current moment is obtained by combining an output voltage cost function, the inverter side bridge arm voltage reference value is input into a modulation module after passing through a wave trap, a driving signal for driving a switch tube is obtained through a modulation method of the modulation module, and a model filtering inductance value is obtained through a real-time filtering inductance parameter identification method with low-pass filtering; according to the modulation method, dual-modulation wave carrier modulation or sine pulse width modulation is selected by judging the fluctuation amplitude of the neutral-point voltage on the direct-current side of the inverter and the balance capacity of the neutral-point voltage of each phase and comprehensively considering the neutral-point voltage balance and the system efficiency on the direct-current side in combination with a fuzzy control method. According to the invention, the multi-objective optimization control of the output voltage quality and the system efficiency when the inverter is provided with the balanced / unbalanced load can be realized.

In accordance with an example embodiment of the present invention there is at least one method and apparatus to perform: obtaining at least one convolutional cross-component model filter for a sample set, where the sample set is reconstructed samples for two channels of an image; applying at least one filter to the reconstructed sample set of the first channel; applying at least one convolutional cross-component model filter using an output of at least one filter for a reconstructed sample set of the first channel as an input; applying at least one filter to an output of the at least one convolutional cross-component model filter; and applying a cross-component filter to the reconstructed sample set of the first channel, or to the output of at least one filter for the reconstructed sample set of the first channel, to obtain a correction for the output of the at least one filter using the output of the at least one convolutional cross-component model filter as input, as visible in FIG. 13.

The embodiment of the invention belongs to the field of data analysis, and relates to a front-end plug-in implementation method based on artificial intelligence supply chain data analysis, which comprises the following steps: receiving supply chain data from a plurality of service systems and artificial intelligence analysis results corresponding to the supply chain data through a network interface; analyzing confidence coefficient information and a prediction interval of the artificial intelligence analysis result; generating a differential patch message for the standardized event stream and maintaining version vector and sequence control; a comprehensive scoring function is formed by the data type adaptation degree, the equipment side frame rate, the memory probe, the interaction complexity and the readability index; screening, drilling, linkage and backtracking interaction events are modeled into a state machine with priority and backspacing rules, and cross-view linkage and concurrent operation are arbitrated. The invention further provides a front-end plug-in implementation system based on artificial intelligence supply chain data analysis. According to the technical scheme, total redrawing and invalid transmission can be reduced, and the end-side real-time performance and the reading credibility are remarkably improved.

Some embodiments disclose an obstacle detection method and system applied to automatic charging equipment, electronic equipment and a medium. The method comprises the following steps: acquiring a depth image of an automatic charging scene by using a depth camera to form a point cloud; performing multi-level filtering on the formed point cloud to filter out environment point cloud; wherein the multi-layer filtering comprises the following steps: filtering point clouds according to a cuboid range of a working space of the mechanical arm; filtering the vehicle point cloud based on the vehicle pose of the charging port; filtering the point cloud of the mechanical arm; and analyzing the filtered point cloud to determine whether an effective obstacle exists or not. According to the method, the depth image is scaled to generate point clouds, so that the number of the point clouds is greatly reduced; through a three-layer filtering strategy of working space body filtering of a mechanical arm, vehicle point cloud filtering and mechanical arm model filtering, all environment point clouds in a scene are filtered, finally, residual point clouds are mapped into a two-dimensional mask image, and whether an effective obstacle exists in the mask image or not is judged by utilizing a connected domain analysis technology and setting an area threshold value.

A method of decoding video data performed by an electronic device is provided. The method receives the video data and determines a block unit from a current frame included in the video data. The method further determines a plurality of luma reconstructed samples in a luma block of the block unit based on the video data and determines a prediction model filter of a prediction model mode for a chroma block of the block unit based on the video data. The method then determines a prediction model filter of a prediction model mode for a chroma block of the block unit based on the video data and reconstruct the chroma block of the block unit by applying the plurality of luma square values and the plurality of luma gradient values to the prediction model filter.

The invention relates to the field of dexterous hands of humanoid robots, in particular to a robot dexterous hand control method and device based on interference identification and a medium, and the method comprises the steps that S1, a hand shape sequence is obtained based on task information analysis; s2, generating a corresponding joint state for each hand shape in the hand shape sequence, and obtaining a joint state vector corresponding to each hand shape; s3, all two adjacent hand shapes in the hand shape sequence form a hand shape group; s4, taking the joint state vectors of the two hand shapes of each hand shape group as boundary constraint conditions, and combining interference identification model filtering to generate a plurality of joint transition state vectors; s5, splicing all the joint state vectors and the joint transition state vectors in sequence to obtain a control vector of the dexterous hand; and S6, controlling each joint action of the dexterous hand based on the obtained control vector. Compared with the prior art, the method has the advantages that complex space state calculation is not needed, and the like.

A method comprising: obtaining a first set of reconstructed samples of a first color channel of an image; obtaining a second set of reconstructed samples of a second color channel of the image; providing the first set of reconstructed samples and the second set of reconstructed samples as input to a cross-component model filter derivation process; applying cross-component filter coefficients obtained from the cross-component filter derivation process in a cross-component filter provided with the first set of reconstructed samples and the second set of reconstructed samples as inputs to obtain an additional input for reconstructing the second set of samples; applying a first adaptive in-loop filter for the first set of reconstructed samples; and applying a second adaptive in-loop filter during a filtering stage of the second set of reconstructed samples.

The invention relates to the field of vehicle positioning, and discloses an unmanned aerial vehicle to-ground vehicle fine positioning and dynamic positioning optimization method based on elevation data. Obtaining a ray analytic expression of a camera sight line vector in an NED coordinate system through Beidou positioning and attitude information of a photoelectric pod of the unmanned aerial vehicle and a pixel position of a target; importing a digital elevation, and carrying out iterative height search in a line-of-sight vector direction by using a dichotomy to obtain target longitude and latitude coarse positioning; and solving the intersection point of the terrain plane and the sight line vector by using a local tangent plane method according to the coarse positioning to obtain fine positioning. The invention further provides a terrain adaptive interactive multi-model filtering TA-IMM algorithm, vehicle motion is described by adopting different motion models, and model state vectors are finally fused and output based on terrain fluctuation characteristics of a target neighborhood, adaptive observation noise and motion model transition probability. According to the method, digital elevation data are fully utilized, and unmanned aerial vehicle-to-ground vehicle fine positioning and dynamic positioning with lightweight hardware and high positioning precision are realized on the whole.

The utility model belongs to the technical field of aviation test devices, and discloses a movable filter element bubble point test device. By means of the traditional mechanical design, 360-degree rotation of the filter element is achieved, meanwhile, multi-model filter element tests of an airplane are achieved through the replaceable airtight connector, and the device is a movable filter element bubble point test device, improves work efficiency and reduces work difficulty. The oil filter cleaning device is simple and practical in structure, has very high practicability, rapidness, accuracy and safety, is mainly applied to oil filter cleaning of certain aircraft types, and is connected with a ground joint at the bottom of an engine compartment before a decomposition part of the engine compartment of each aircraft is started; and residual hydraulic oil in the cabin is discharged without turning over the oil filter device, so that the working efficiency is enhanced, the working difficulty is reduced, and the oil filter device has been applied to an airplane electromechanical workshop.

This invention discloses a method and system for tracking the trajectory of a LiDAR and camera based on bidirectional correlation. The method includes acquiring first sensing data from the LiDAR and second sensing data from the camera, and performing spatiotemporal registration and coordinate unification; filtering the first sensing data to obtain a first predicted state sequence of the target; completing the state of trajectory interruptions in the second sensing data to obtain a second supplementary state sequence of the target; making a correlation decision for the same target based on the first predicted state sequence and the second supplementary state sequence; and using a multi-model filtering algorithm to fuse and update the target state based on the correlation decision result, outputting a continuous moving trajectory of the target ahead. This invention synchronously registers LiDAR and camera data, fuses radar-based predicted trajectories and vision-based completed trajectories for bidirectional correlation, and finally outputs a continuous target trajectory after multi-model filtering.

The invention discloses a precise VC standard micro-vibration environment simulation method based on ground-based feedforward. A passive vibration isolation unit is used for supporting the platform, a foundation vibration sensor is used for sensing foundation vibration, and an actuator is used for outputting feed-forward control force and serving as an active excitation source at the same time; 2) designing a feed-forward controller according to the system model, and calculating to obtain a feed-forward control signal by using a foundation vibration signal; 3) based on a VC vibration level standard, quantizing into a discrete frequency spectrum, and reconstructing a time domain vibration excitation signal conforming to the frequency spectrum characteristic; and 4) obtaining an inverse model filter according to the system model, obtaining a micro-vibration excitation control signal through the filter, combining the micro-vibration excitation signal with the feedforward control signal to obtain a final control signal, and converting the control signal into a corresponding output acting force by using an actuator. The micro-vibration environment with the micro-vibration smaller than the VC level of the ground can be achieved.

The utility model discloses a universal test fixture for filters of multiple models. The universal test fixture comprises a test platform, a clamping assembly and a moving assembly, a plurality of interfaces are arranged on the periphery of the test platform in an array manner; the clamping assembly comprises a base, a fixed seat and a rotating rod; one end of the rotating rod is fixedly connected with a clamping plate, and the other end is fixedly provided with a rotating round table; the moving assembly comprises a sliding rail, a sliding seat and a rotating seat; the rotating seat is arranged below the base; by arranging the slide rail on the test platform, the filter can move back and forth on the test platform to be close to the position where the interface is needed, and by arranging the rotating seat, the filter is driven to rotate, so that the interface position on the filter corresponds to the interface position on the test platform at a more appropriate angle, the addition of leads is reduced, and the test efficiency is improved. The problems that when the position of a wiring port is far, a plurality of leads are needed for guiding, so that wiring is complex, and follow-up adjustment or troubleshooting is difficult are solved.

The invention provides a railway inertia precision keeping method and system based on double-ring quaternion modeling, and belongs to the technical field of train operation control. According to the method, an observation difference model is constructed through acceleration original observation provided by an accelerometer and theoretical acceleration provided by an inner ring model filter, so that optimal estimation of an error quaternion is realized, and a zero offset noise error and a rotation error of an inertial device are compensated through fusion calculation; the double-ring cross coupling of the quaternion vector in the traditional combined positioning model and the inner ring model is realized, the railway inertia precision keeping method based on double-ring quaternion coupling enhancement is established, and a better inertia device precision keeping effect is realized.

The invention relates to an active and passive combined space-based moving target positioning method based on feature constraint, and the method comprises the steps: S1, obtaining a measurement value of a target position, and the measurement value comprises an azimuth angle sequence, a pitch angle sequence and a distance sequence; s2, performing segmented constant acceleration process filtering on the azimuth angle sequence and the pitch angle sequence; s3, obtaining a coordinate value of the remote sensor under the earth-fixed system, and calculating to obtain a three-dimensional position measurement value of the target under the earth-fixed system; s4, obtaining a positioning correction result in the error elliptical plane through coordinate transformation and height constraint correction; s5, carrying out constant acceleration motion model filtering on the positioning correction result of the error elliptical plane; and step S6, repeatedly executing the step S4 and the step S5 until convergence, and obtaining a final error correction result. According to the invention, the positioning precision in an active and passive integrated detection scene is improved.

The invention discloses a multi-UUV (Unmanned Underwater Vehicle) pure azimuth target state estimation method and system adopting a cross positioning improved IMM (Inertial Model Modeling) and a medium, and belongs to the field of target tracking of unmanned underwater vehicles. Comprising the steps of initializing parameters, and obtaining state information and model probability in real time; dynamically judging whether the IMM algorithm has model competition or not by calculating a normalized probability entropy value representing the model competition degree; when the entropy value exceeds a set threshold value, a cross positioning correction mechanism is triggered, namely, pure azimuth observation information of a plurality of UUVs is fused, a more reliable target correction state and covariance are obtained through calculation, and the target correction state and covariance are used as new input of each model filter, so that estimation performance reduction caused by model competition is effectively inhibited; and if the competition does not occur, executing a standard IMM process. According to the method, the dynamic judgment and correction mechanism is introduced, so that the state estimation precision and the algorithm robustness of the maneuvering target under the multi-UUV pure azimuth observation scene are remarkably improved.

The utility model relates to the technical field of pneumatics, in particular to a model-level compressor shell which comprises a machine shell and a filter screen clamping block, and the filter screen clamping block can be inserted into the machine shell through a filter screen opening. A spring is arranged on the filter screen clamping block, a clamping block is arranged at the left end of the spring, and a filter element is arranged in the middle of the filter screen clamping block; a clamping block baffle is arranged at one end of the clamping block, and when the filter screen clamping block can be inserted into the machine shell through the filter screen opening, the clamping block baffle is exposed outside. In addition, a clamping block positioning groove is formed in the clamping block baffle, and a semicircular positioning block is arranged in the machine shell. Impurities in air are filtered and removed through the filter screen clamping block, the precise structure in the compressor is protected, the service life of the compressor is prolonged, and the working efficiency of the compressor is improved. And the filter screen clamping block is easy to disassemble and assemble through a spring and other mechanisms, and is convenient to install and clean.

The invention relates to a maneuvering extended target tracking method based on a golden section self-adaptive grid, which comprises the following steps of: firstly, representing a model set for tracking in a grid form, and dividing the grid into an internal grid and a peripheral grid; then generating a local weight and an overall weight of the model set, arranging the local weight and the overall weight with the two models in the internal grid according to a golden section proportion, and calculating a distance between the local weight and the overall weight to reflect a difference between the local weight and the overall weight and a target real motion mode matching degree so as to serve as a basis for model parameter updating; and finally, according to a result obtained by filtering the updated model set, updating a transition probability matrix, and according to a model filtering residual error with the maximum model probability, judging the maneuvering degree of a target so as to determine whether each module of the algorithm is executed or not, thereby improving the efficiency of the algorithm. The method provided by the invention can effectively solve the problem that the tracking precision of an extended target tracking method in a strong maneuvering scene is reduced.

The application relates to the technical field of modeling prediction, and discloses a light module life prediction method of a DSP integrated degradation and an LSTM model. Through a cooperative architecture of a DSP embedded mechanism model filtering an LSTM network time sequence calibration, a multi-parameter coupling degradation model and a Kalman filtering algorithm are embedded into a light module DSP at a calculation level, in-situ, real-time and safe evaluation of a life state is realized, and a traditional mode depending on external calculation resources is changed; at a model level, the explainability of a physical mechanism and the adaptive ability of data driving are fused, fundamental degradation dynamics are described through a nonlinear coupling model, and specificity degradation rules are learned from individual historical data through an LSTM network; and finally, real-time, accurate and adaptive light module life state evaluation and prediction are realized.

The invention discloses an optimal junction termination structure screening method based on machine learning. The method comprises the following steps: S1, training an intelligent design model of different junction termination technical structure parameters based on machine learning; s2, constructing a model filter; s3, performing first-stage joint optimization; s4, performing two-stage special optimization; and S5, screening an optimal junction termination structure. According to the method, multi-model collaborative design is achieved, the optimal termination structure meeting the requirements of a designer can be rapidly obtained, compared with traditional design depending on experience by the designer, the method has the advantages of being high in speed, high in accuracy and the like, the design efficiency of the designer can be improved, the design time can be saved, and the error compared with TCAD simulation is small.

The embodiment of the invention discloses an AI local knowledge base construction method and system based on hybrid retrieval and closed-loop optimization, and the method and system are used in an enterprise and aim at guaranteeing data safety and improving retrieval accuracy and efficiency. The method comprises the steps that intelligent blocking and vectorization storage are conducted on an original document; screening most relevant knowledge fragments by adopting a hybrid retrieval mechanism in which semantic vector retrieval and keyword retrieval are parallel and combining a reordering model; generating answers by using a local large language model and marking reference sources to ensure traceability; and performing continuous closed-loop optimization on the retrieval model and the knowledge base based on user feedback. The method is operated in a completely privatized deployment environment, has high retrieval accuracy, strong data security and adaptive evolution capability, and effectively overcomes the defects of traditional keyword retrieval and a general RAG scheme in the aspects of semantic comprehension, complex document processing and continuous optimization.

A method comprising: obtaining a first set of reconstructed samples of a first color channel of an image; obtaining a second set of reconstructed samples of a second color channel of the image; applying a first adaptive in-loop filter for the first set of reconstructed samples; applying a second adaptive in-loop filter for the second set of reconstructed samples; providing the first set of reconstructed samples and the second set of reconstructed samples as input to a convolutional cross-component model filter; applying the convolutional cross-component model (CCCM) filter to the first and second set of reconstructed samples to obtain an additional input for reconstructing the second set of samples; applying a cross-component adaptive in-loop filter for outputs of the first adaptive in-loop filter and the second adaptive in-loop filter; and providing an output of the cross-component adaptive in-loop filter to the output of the second adaptive in-loop filter as an additional correction for the second set of samples.