80results about How to "Robust" patented technology

The invention belongs to the technical field of industrial process control, and discloses an internet-of-vehicles cooperative vehicle fleet mixed triggering formation control method under network limitation, which comprises the following steps: constructing a longitudinal dynamic model and a formation topological structure of cooperative vehicles to describe a relative relationship between the vehicles, establishing an environment limitation model, and constructing a formation model; the nonlinear state of the cooperative vehicle is observed and estimated through the mixed trigger communication condition; a hybrid triggering mechanism is constructed by using a network quality of service QoS mapping function and a self-adaptive threshold value, so that the problem of complexity explosion caused by error derivation is avoided; and finally, constructing a consistency controller model of the collaborative vehicles, and realizing stable tracking and formation collaboration of the motorcade in a dynamic network environment. According to the invention, the vehicle can still keep efficient and smooth communication and control under the conditions of bandwidth limitation, delay change or data packet loss and the like, the network utilization rate is improved, and stable control of a motorcade is realized under various communication topologies.

An intelligent compensation method for coupling errors of a dynamic gravimeter comprises the following steps: selecting typical environment interference influencing the dynamic gravimeter in a complex measurement environment, and analyzing an influence mechanism of each typical environment interference on the measurement accuracy of the dynamic gravimeter; according to the influence mechanism under each typical environment interference, analyzing a coupling error influence mechanism between the typical environment interferences; based on the coupling error influence mechanism, constructing a coupling error comprehensive compensation model; adding the coupling effect loss into a data-driven loss function, and defining a model coefficient constrained by physical information; and performing model training by using the training data, and calculating the total loss of the verification set on the obtained model to obtain a final coupling error comprehensive compensation model. According to the invention, the coupling error of the dynamic gravimeter in a complex environment is compensated based on coupling error mechanism analysis and a physical constraint long-short-term memory model, and the measurement precision and environmental adaptability of the dynamic gravimeter under complex multi-factor interference conditions are improved.

The application discloses a kind of battery core temperature online estimation and prediction method, system, equipment and storage medium based on mixed TSDM-AMBO-GRU, applied to lithium ion battery core temperature estimation and prediction field, comprising: obtaining battery operating data, and utilizing TSDM to create synthesis of battery operating data, as training set with battery operating data together;Training set is input to BO-GRU-AM framework, and the battery core temperature prediction model based on BO-GRU-AM is obtained by training;Input test set to battery core temperature prediction model, and obtain battery core temperature prediction result.The application not only can effectively overcome the problem caused by uneven distribution of temperature sensors and limited data in lithium ion battery system, and can realize real-time, efficient and accurate battery core temperature prediction under various temperatures and complex conditions, provide reliable fault warning and real-time monitoring support for BMS in practical application.

This application discloses a method, system, and storage medium for controlling the charging power of an electric vehicle. The method includes: acquiring the target charging completion time and target charging amount set by the user, as well as the current battery charge; calculating the remaining charge required and the remaining available charging time, wherein the remaining available charging time deducts the reserved buffer time, the derating time in the high charge range, and the safety margin time based on the battery state; determining the target average charging power for the next cycle based on the ratio of the remaining charge required to the remaining available charging time; and comparing the actual charge with the expected charge after the time cycle of each charging process. If there is a deviation, the target power for the next cycle is dynamically corrected bidirectionally based on the magnitude of the deviation and the urgency of the remaining time, ensuring that the corrected power does not exceed the safety limits of the battery and the charging pile. This application achieves precise matching between the charging process and the user's time needs, improving charging efficiency and user experience.

The invention discloses an image watermarking algorithm based on Kyber fully homomorphic encryption, which is characterized in that an anti-quantum Kyber lattice cryptographic algorithm is combined with a Patchwork watermarking algorithm, so that the exchangeability (first encryption and then embedding or first embedding and then encryption) of an encryption and watermark embedding sequence is realized, and dual-mode watermark extraction of a ciphertext domain and a plaintext domain is supported. According to the method, the addition homomorphic characteristic of Kyber is utilized, it is ensured that the operation sequence does not affect ciphertext generation and watermark extraction, the safety defect of a traditional algorithm in the quantum environment is overcome, and the method is suitable for scenes of digital copyright management, medical data sharing and the like and has the advantages of quantum attack resistance, high robustness, outstanding privacy protection capacity and the like.

This invention discloses a preventative voltage control method for grid-connected energy storage based on voltage sensitivity, comprising the following steps: updating the voltage sensitivity matrix online based on real-time operating data of the distribution network and predicting future photovoltaic output curves; determining whether there is a voltage limit exceeding situation at future nodes based on the voltage sensitivity matrix; if there is a voltage limit exceeding situation at future nodes, calculating the active power required by the grid-connected inverter based on the voltage sensitivity matrix and photovoltaic output, and generating a charging curve; adaptively adjusting the charging curve based on the real-time state of charge of the energy storage, and charging the grid-connected inverter before the voltage limit exceeds the limit. The beneficial effects of this invention are: changing the voltage support mode of the energy storage module in the grid-connected energy storage system from passive response to active prevention; when the system's reactive power capacity is insufficient, pre-charging the energy storage allows for sufficient safety margin for the system voltage during critical future periods and maximizes the absorption of photovoltaic power, reducing curtailment.

The invention discloses a sludge age self-adaptive regulation and control method and system based on dissolved oxygen trend recognition. Aiming at the problem of impact response lag caused by difficulty in COD on-line measurement, the method comprises the following steps: under the condition of stable water inlet and aeration, filtering and sliding window feature extraction are performed on an on-line DO signal to obtain a regression slope representing the change of DO along with time, and COD rising impact is reliably identified in combination with an ORP persistence criterion. A mechanism and machine learning fusion model is adopted to determine an identification threshold value and a target sludge age SRT, the sludge discharge amount is rapidly adjusted after impact triggering, the treatment capacity is synchronously matched, and aeration is used for assisting in maintaining DO stable. The system integrates an online monitoring module, a DO recognizable interval judgment module, a feature calculation module, a load recognition module, an SRT decision module and a sludge discharge execution module. According to the method, rapid impact recognition and closed-loop regulation and control can be realized without an online COD instrument, and the impact resistance and the operation stability of the process are remarkably improved.

The application discloses a kind of active control method, device and car of automobile interior noise combined with virtual sensing.The method can include: for the interior noise of vehicle at different speeds, by reference signal, physical monitoring signal and virtual error signal respectively construct multiple groups of auxiliary filter containing optimal control filter information and observation filter containing the transfer function between physical monitoring point and virtual error point;By mean method, construct composite observation filter;At different speeds, based on least mean square estimation error matching mechanism, by reference signal, physical monitoring signal, auxiliary filter, composite observation filter, active noise control is carried out at noise reduction target position.The present application can significantly reduce the noise at ear at different speeds while the arrangement position of error microphone does not affect the normal activities of passengers, improve the comfort of riding.

The present application relates to a kind of angle flicker noise preprocessing methods based on adaptive amplitude suppression, comprising the following steps: obtaining the angle measurement data sequence of current batch, based on the target position estimation value and speed prediction value of last time, the target position prediction value of current time is calculated;Deviation calculation is carried out to angle measurement data sequence and target position prediction value, and deviation sequence is obtained;Adaptive amplitude suppression processing is implemented to the deviation sequence, specifically including: setting amplitude threshold, identifying large value component and small value component in deviation sequence;Large value component is scaled nonlinearly, and its amplitude is compressed to the level close to small value component, while keeping small value component unchanged;After amplitude suppression processing, the deviation sequence is added to target position prediction value, and the modified measurement sequence is obtained;Mean filtering is carried out to the modified measurement sequence, and the preprocessing result is output, and the preprocessing result is input to tracking filter, and is used to update target state estimation.

The motion quality evaluation method based on motion perception and double-flow space-time graph convolution relates to the motion quality evaluation field, solves the problems that the existing rehabilitation motion quality evaluation method is difficult to accurately model the cooperation between joints, the fixed joint grouping strategy is not flexible enough, the model has insufficient recognition of the detail differences of complex rehabilitation motions, and the reliability of the evaluation result is affected, a dynamic joint grouping strategy based on motion amplitude driving is provided, a double-flow STGCN architecture is designed, the position and orientation information of the joints are processed respectively and then fused, a two-stage self-attention module SADG is developed, the inter-group time sequence mode and the intra-group spatial relationship are modeled in stages, and multi-granularity feature interaction fusion is realized. The attention mechanism introduces a constraint guided by the motion amplitude, so that the attention weight is more interpretable and has physical meaning.

The application discloses a kind of based on three-dimensional voxelization map small unmanned aerial vehicle fine inspection path planning method, system, setting and medium, it is related to unmanned aerial vehicle control technical field, the method includes: S1: three-dimensional voxelization modeling is carried out to inspection object and environment, and safety inflation processing is carried out to obstacle;S2: the original inspection view point set is clustered and clustered, and the view point in clustering cluster is sorted according to S type path order;S3: the three-dimensional A* algorithm of optimization is used to generate collision-free polyline path;S4: the line path generated in step S3 is smoothed using line-of-sight algorithm, removes redundant node, generates smooth flight trajectory;S5: the path point after smoothing is converted from local model coordinates to WGS84 latitude and longitude coordinates using decoupling fitting strategy based on control point, generates the GPS waypoint sequence executable by unmanned aerial vehicle.The method provided by the application can solve the problems of unsafe path planning during inspection, low inspection point traversal efficiency, non-smooth flight path and the like in the prior art.

This invention discloses a fault detection and fault-tolerant control method for roll-to-roll systems, belonging to the field of industrial printing control. This method solves the problem of registration error propagation caused by actuator failure. Its technical solution is based on a coupled dynamics model established by fully decoupled proportional-derivative control, and designs a fault detection index based on the 2-norm to identify and locate faulty units in real time. Then, upstream and downstream collaborative fault-tolerant control is implemented: adaptive feedback adjustment combined with gain scheduling is used for the upstream faulty unit, and control law reconstruction that removes the dependence on coupling compensation for the faulty unit is used for the downstream faulty unit to block the error propagation path. This method can effectively suppress error propagation under fault conditions and improve the system control accuracy and robustness.

The application provides a self-adaptive adversarial training method for dynamic visual cabinet recognition, comprising the following steps: S1, initializing target network parameters or initializing target network pre-training configuration, obtaining X clean correctly recognized and not disturbed in a dynamic visual cabinet; S2, generating an adaptive adjustment attack parameter vector theta by using a heuristic differential evolution algorithm of a strategy generator according to the robustness of the target network; S3, inputting the attack parameter vector theta into an adversarial sample generator, generating an adversarial sample by adding disturbance in the X clean ; S4, inputting the X clean and the adversarial sample into the target network for training, and setting a training target function; S5, repeating steps S2-S4 until the maximum iteration number is reached, and obtaining a recognition model for dynamic visual cabinet recognition. The application can improve the adversarial attack capability of the recognition model in the dynamic visual cabinet.

The application discloses a personalized picture retrieval method in a federated learning scenario. It includes: 1) loading a public picture-text pair dataset to generate a model training dataset; 2) based on the text information in the dataset in step 1), the tokenizer word segmentation tool is used to obtain the mapping relationship between words and Token positions; 3) based on the picture information in the dataset in step 1), the picture tensor is obtained after preprocessing; 4) a unified model is constructed based on the pre-training model CLIP; 5) the unified model is trained using the text and image expressions obtained in steps 2) and 3); 6) image retrieval is performed using the unified model trained in step 5), and pictures consistent with the query text are obtained. The application designs a private retrieval database for the federated learning client, integrates client-specific information into the pictures and text, and solves the problems of retrieval accuracy decline and slow convergence caused by the non-independent and identically distributed characteristics of data in the federated learning scenario.

The invention discloses a multi-source single-output green light beacon laser for space application, which is characterized in that a light source part comprises two butterfly-shaped packaging laser devices which are cold backups for each other, the butterfly-shaped packaging laser devices are solid laser devices packaged in butterfly-shaped shells, each butterfly-shaped packaging laser device comprises a semiconductor pumping chip, a shaping optical fiber, a laser crystal Nd: YVO4 and a frequency doubling crystal PPLN, and green light of 532nm is generated; in the filtering and beam combining light path part, 532nm green light emitted by each light source is respectively provided with a half-wave plate for adjusting the polarization direction, is combined into the same light path through a polarization splitting prism, is filtered by a light filter, enters a multimode optical fiber through an optical fiber collimator, and is divided into two paths through an optical fiber coupler; one path is used for monitoring the output power and health state of the laser in real time, and the other path is used as main beacon light. The problem of performance reduction caused by the overheating phenomenon of the light-emitting unit is effectively solved, high-power output can be kept for a long time, and meanwhile the overall light weight and flexibility of the laser are considered.

The application discloses a kind of LDPC multi-bit quantization hardware decoding method suitable for high-speed communication, it is related to the field of communication physical layer digital signal transmission, including the steps 1) reset decoding module, from channel obtains the information sequence to be decoded and carries out suitable bit quantization extension;2) to meet the engineering demand of super high speed, low resource consumption, low latency, the application adopts a kind of full parallel decoding structure, the full parallel decoding structure can realize all check nodes update simultaneously and provide to all variable nodes.3) check node update is carried out by 6 parallel ways, the normalization factor used in the application check node update is obtained by adaptive mode;4) the check node information obtained in step 3 is as the input of variable node update module, variable node update is carried out by 6 parallel ways, the input in the application variable node update module is improved by expanding the number of bit quantization to improve the decoding ability of decoder.

The invention provides a track irregularity signal compression and reconstruction method and system based on a compressed sensing theory, and mainly relates to the technical field of railway infrastructure health monitoring and big data processing. According to the method, the sampling accuracy in the engineering field is mainly improved, redundant information is reduced, the system bottleneck of storage and transmission is improved, a measurement matrix irrelevant to a sparse transformation base is introduced at a signal acquisition end, and a compression observation value far lower than the Nyquist rate is directly obtained. And then, an original track irregularity signal is reconstructed from a small number of observation values with high precision by solving a norm minimization problem at a data processing end. According to the method, the front-end data sampling rate, the hardware load and the data transmission and storage requirements are greatly reduced, meanwhile, it is guaranteed that the reconstructed signal meets the engineering analysis precision, and a core technical scheme is provided for a new-generation efficient and low-cost track detection system, real-time train-structure system dynamic analysis and the like.

The invention discloses a multi-state node conversion method and system based on a block chain, and relates to the technical field of node conversion, and the method comprises the following steps: constructing a node steady state recognition model, continuously monitoring a trigger condition generated in a node operation process, calculating a change slope of the trigger condition, and calculating a node steady state recognition model; comparing the change slope with the threshold difference of the previous moment, and outputting the current transient confidence result of the node; and establishing an adaptive filtering window based on a transient confidence result, dynamically adjusting a boundary in the adaptive filtering window, weakening a low-confidence trigger signal, and only outputting an effective trigger signal subjected to boundary screening. According to the invention, through a steady state identification and dynamic filtering mechanism, the accuracy and anti-jitter capability of node state judgment are improved; and in combination with aggregation buffering, load prediction and probability convergence control, high-credibility execution of state switching and resource controllability are realized, and the stability, intelligence and traceability of node management are enhanced.

The invention discloses a high-robustness lossless high-capacity audio watermark embedding method based on deep learning, and relates to the technical field of digital watermarking. The method comprises the following steps of audio-watermark information preprocessing, watermark information embedding and watermark embedding detection and judgment. According to the method, the original audio and the watermark information to be embedded are correspondingly preprocessed, so that the quality of the original audio and the watermark information is effectively improved, the subsequent embedding process is more efficient and accurate, then the preprocessed audio and watermark information are input into the audio embedding network to output the watermark audio, and the watermark embedding efficiency is improved. According to the method, high-capacity watermark information is embedded, audio distortion in the embedding process is effectively avoided, the quality of the embedded audio is guaranteed, finally, the stability and the anti-jamming capability of the embedded watermark can be effectively detected by introducing adversarial training attack simulation, the audio distortion is reduced to the maximum extent while the watermark capacity is guaranteed, and the watermark quality is improved. And the anti-interference capability of the watermark is improved.

The present application provides a kind of sparse gradient enhanced edge preserving image smoothing method, comprising: introducing structure fidelity term on the basis of existing global optimization model containing data fidelity term, in structure fidelity term, the gradient of input image is preprocessed using the way of combining mapping function and bilateral filtering;And introduce sparse gradient enhancement term on the basis of structure fidelity term, in sparse gradient enhancement term, L2 norm and L1 norm are used to constrain the gradient of output image to enhance sparsity, L1 norm is iteratively solved by combining subgradient method, fast Fourier transform and alternating direction multiplier method, or L p norm is used to constrain the gradient of output image, and L p Norm is converted to L1 norm for solving by iterative reweighting method, and preconditioned conjugate gradient method is used to improve calculation efficiency in solving process.The present application can retain complete semantic information while showing sparse smoothing effect when processing simple and complex non-texture images and texture images.

The application provides a medical image representation learning method and system based on multi-granularity world modeling. The method first enhances the radiograph image into first and second enhanced views with spatial overlap, and inputs a visual transformer to extract basic patch features; then constructs multi-granularity anatomical representation through hierarchical aggregation; then drives the world model to perform anatomical structure modeling, anatomical layout modeling and domain change perception modeling tasks, infers the relative spatial coordinates across views by using the overlap ratio of the overlapping area features, and modulates the input features by using the granularity perception enhancement parameters; finally, the model parameters are optimized based on a joint loss function. The application can solve the technical problems of the prior art, such as lack of unified modeling of multi-level anatomical semantics of the radiograph image, insufficient cross-view spatial layout reasoning capability, and difficulty in maintaining anatomical consistency under domain change.

The application discloses a kind of off-site computing power dynamic collaborative optimization methods based on multi-agent reinforcement learning, including the following steps: constructing resource topology diagram;Based on resource topology diagram, get node-level state feature vector and system-level state feature vector;Multi-agent environment is constructed;Local observation, global information and agent action set are input into improved CTDE model, output policy network parameters and value network parameters and construct training batch;Based on training batch, get converged policy network parameters and converged value network parameters;Get execution result;Obtain dynamically updated policy network parameters and dynamically updated value network parameters, realize the dynamic collaborative optimization of task acceptance, resource allocation, task migration, replica start-stop and bandwidth matching.

The application belongs to the field of industrial safety monitoring, and discloses a foreign matter border crossing detection method, computer equipment and a storage medium. The method comprises the following steps: acquiring a current video frame; inputting the current video frame into a warning line segmentation model and a foreign matter detection model respectively to determine a binary mask of the current video frame and a dynamic detection frame for foreign matter; determining at least one warning boundary point set of the current video frame based on the binary mask; performing linear regression fitting on each warning boundary point set to obtain a corresponding candidate fitting warning line, and generating a corresponding candidate warning area based on each candidate fitting warning line; selecting a target warning area with the highest overlap degree with a current calibration warning area from each candidate warning area; if the overlap degree is less than a preset threshold, regarding the target warning area as a new calibration warning area; and determining whether there is foreign matter border crossing in the current video frame based on the dynamic detection frame and the new calibration warning area. The application can improve the accuracy of foreign matter border crossing detection.

This invention belongs to the field of computer vision technology, specifically relating to a skin lesion image segmentation method based on a Transformer dual-branch model. The method constructs and trains a Transformer dual-branch model, inputting the image to be processed into the trained Transformer dual-branch model to obtain the segmentation result. The Transformer dual-branch model includes a main branch network, an auxiliary branch network, and an information aggregation module. This invention proposes a novel skin lesion image segmentation method that addresses the shortcomings of traditional deep learning methods in extracting global contextual information. It utilizes an efficient multi-scale visual Transformer as an encoder to extract more powerful and robust features. Simultaneously, it introduces low-level feature modules and high-level feature fusion modules to effectively improve the network's feature learning ability and segmentation performance.

The invention discloses a low-power-consumption high-speed two-stage dynamic comparator for an analog-to-digital converter, relates to the technical field of integrated circuits, and aims to solve the problems that a traditional dynamic comparator is high in power consumption, limited in speed and uncontrollable in balance between offset voltage and performance. The comparator comprises a two-stage dynamic structure of a pre-amplification stage and a latch stage, wherein the pre-amplification stage takes a PMOS transistor as an input tube and is matched with an NMOS switching tube and a PMOS active load tube to realize preliminary amplification of an input differential signal; the latch stage adopts a PMOS latch structure instead of a traditional NMOS latch structure. And through two paths of delay controllable clock signals (clkb1 and clkb2), the latch stage is controlled to be activated in a delayed manner after the preamplifier stage works for a preset time, and meanwhile, a current source of the preamplifier stage is closed when the latch stage is activated. By adjusting latch level delay time, offset voltage, power consumption and speed can be flexibly balanced, performance requirements of different analog-to-digital converters are met, and the analog-to-digital converter is suitable for scenes such as portable equipment, communication systems and the like with high requirements on energy efficiency and speed.

This invention provides a water hammer prediction system and device. The invention relates to the field of pipeline fluid transport safety technology. The water hammer prediction system is configured to: acquire static topology and physical attribute information and multi-dimensional real-time operating condition data of the fluid transport system, and pre-calibrate a dynamic simulation model based on the information and data; when a water hammer triggering event is detected, invoke the pre-calibrated dynamic simulation model to perform an initial prediction to generate an initial prediction result, and issue a pre-control command to the first-level protection device based on the initial prediction result; acquire the measured impact index at the first-level protection device and compare it with the initial prediction result to determine the prediction error; and, based on the prediction error, correct the dynamic simulation model in real time, and drive the corrected dynamic simulation model to perform cascade control of the next-level protection device.

The application discloses a kind of multi-level man-machine collaborative data labeling method and system based on task ambiguity evaluation, first constructs and trains ambiguity prediction model, utilizes lightweight ambiguity prediction model, predicts the cognitive ambiguity score and logic ambiguity score of each data instance;Four-stage labeling resource pool containing low-performance large language model, high-performance large language model, ordinary crowd sourcing labeler and expert labeler is constructed;According to the predicted two-dimensional ambiguity score, the data instance is automatically routed to the optimal labeling resource by dynamic scheduling strategy, for the task with high cognitive ambiguity, the upgrading mechanism combining crowd sourcing voting and expert arbitration is used, and the final labeled data set is output.The application can intelligently match labeling cost and ability according to the intrinsic properties of the task, significantly reduce the labeling cost and improve the labeling efficiency under the premise of ensuring the data labeling quality, solve the problem of high cost of traditional crowd sourcing and unstable quality of single model labeling.

The present application relates to the field of aerospace TT&C and artificial intelligence cross technology, and in particular to a federated learning driven edge node multi-axis collaborative optimization method and system, aiming at the defects of large transmission pressure, poor data security, weak multi-axis collaboration ability, single point failure and the like of traditional TT&C antenna centralized architecture, the system is composed of a central aggregation node, an edge LCU control node and an antenna multi-axis execution mechanism. The system adopts a horizontal federated learning mode, the edge node locally collects and stores original working condition data, and only encrypts and uploads model parameters after completing model training; the central node completes parameter aggregation through a multi-dimensional dynamic weight algorithm, generates a global model and iterates. The present application can optimize antenna multi-axis action timing, tracking accuracy and load distribution, has advantages of data security, low transmission cost, strong robustness, easy expansion and the like, and is suitable for various TT&C antenna cluster scenes.

The invention provides a medical image representation learning method and system based on multi-granularity world modeling. The method comprises the following steps: firstly, enhancing a ray image into a first enhanced view and a second enhanced view which are spatially overlapped, and inputting the first enhanced view and the second enhanced view into a visual converter to extract basic patch features; then multi-granularity anatomical representation is constructed through hierarchical aggregation; then driving the world model to execute anatomical structure modeling, anatomical layout modeling and domain change perception modeling tasks, inferring cross-view relative space coordinates by using an overlapping proportion of overlapping region features, and modulating input features by using granularity perception enhancement parameters; and finally, optimizing model parameters based on the joint loss function. According to the method, the technical problems that unified modeling of multi-level anatomical semantics of the ray image is lacked, the cross-view spatial layout reasoning capability is insufficient and the anatomical consistency is difficult to maintain under the domain change in the prior art can be solved.