83results about How to "Strong robustness" patented technology

The invention provides a distributed steer-by-wire vehicle stability control method. The method comprises the steps that a distributed steer-by-wire vehicle dynamics model, a two-degree-of-freedom reference model and a four-wheel steering angle distribution model are established; in the design of the yaw stability controller, a hybrid sensitivity # imgabs0 # robust control method is adopted, the uncertainty of the system is fully considered, and the robustness of the system can be considered while the target tracking performance is ensured.

The invention provides a prediction method for the remaining service life of a rolling bearing based on transfer learning, and belongs to the technical field of artificial intelligence and mechanical engineering. Rolling bearing vibration signals are obtained, and a data set is constructed and divided into a source domain and a target domain; constructing a prediction model of the remaining service life of the rolling bearing; the model comprises the following steps: obtaining low-dimensional features and model weights by using a coding layer; the coding layer comprises a multi-head self-attention sub-layer used for obtaining a relationship between features and a model weight; the feedforward mechanism is used for obtaining low-dimensional mapping; predicting the residual life of the bearing by using the feature representation output by the last coding layer through the full connection layer; training the prediction model of the remaining service life of the rolling bearing: calculating the difference between a source domain and a target domain by using the data set, and constructing a loss item; weight parameters are added to the loss item, the weight parameters and the root-mean-square error loss item are added for back propagation, and the model weight is updated. And the generalization ability of predicting the bearing life by the model is improved.

The invention discloses a method and device for extracting a centerline of a quasi-cylinder by using a point cloud, and the method comprises the steps: obtaining a quasi-circular target of which the centerline is to be extracted, and carrying out the three-dimensional point cloud scanning, and obtaining the three-dimensional point cloud data; point clouds of a target in the three-dimensional point cloud data are cut at equal intervals in the direction of the center line, and a plurality of annular point clouds with noise are formed; projecting the annular point cloud to a two-dimensional plane in a direction perpendicular to the center line; calculating distances from all point positions in the annular two-dimensional image to external points on the projection image; the probability of occurrence of each distance is calculated, and the position corresponding to the pixel when the probability is the maximum value is obtained as the plane center point of the annular point cloud corresponding to the two-dimensional image; associating z values of the annular point cloud during cutting in the direction of the center line to form three-dimensional coordinates of the center point; and connecting the three-dimensional coordinates of all the central points of the equally-spaced cutting annular point clouds to obtain the central line of the cylinder-like body. According to the invention, the extraction precision of the centerline of the quasi-cylinder and the extraction reliability of the centerline are improved.

The invention discloses a power transmission and distribution power grid robust optimization unit scheduling method based on the whole frequency response process, and relates to the technical field of power systems. The method comprises the following steps: constructing a whole-process frequency security constraint around three links of inertial response, primary frequency modulation and secondary frequency modulation, establishing a power transmission and distribution network two-stage robust unit combinatorial optimization model considering the whole frequency response process by taking min-max-min as a structure, and minimizing the start-stop cost, the power generation cost and the standby cost of the primary frequency modulation and the secondary frequency modulation as a target; and converting the two-stage robust unit combinatorial optimization model of the power transmission and distribution network into a mixed integer second-order cone programming model by adopting a column sum constraint generation algorithm and a strong duality theory, and carrying out iterative solution. According to the method, the frequency regulation capability of the power transmission and distribution network can be effectively improved, the operation cost is reduced, the robustness and efficiency of a scheduling scheme are improved, and safe, stable and economical operation of the power transmission and distribution network under the output uncertainty condition of new energy such as wind and light is ensured.

The invention discloses a weak supervision defogging method based on uncertainty driving, and the method employs a weak supervision defogging network of a U-Net architecture, and the weak supervision defogging network comprises a regional dynamic attention module RDAM, a lower sampling layer, a multi-scale feature fusion module MSFFM, and an upper sampling layer. The RDAM is used for processing an input image or the output of a previous layer of RDAM, constructing a dynamic convolutional layer and obtaining a regional dynamic attention feature map through the dynamic convolutional layer; the down-sampling layer is used for down-sampling the output of the RDAM and reducing the spatial resolution; the MSFFM is used for processing input features through a window multi-head self-attention mechanism so as to capture a long-distance dependency relationship between pixels and performing multi-scale feature extraction and fusion; the upper sampling layer is used for recovering the feature map output by the MSFFM to the resolution ratio of the feature map output by the MSFFM of the upper layer and performing feature splicing with the feature map output by the MSFFM of the upper layer; and enabling the up-sampled feature map to sequentially pass through a 3 * 3 convolutional layer, a ReLU activation function and a 1 * 1 convolutional layer, and finally outputting a defogged image.

The invention discloses a biomedicine named entity recognition method based on multi-corpus aggregation and large model corpus expansion. The method mainly comprises the steps of multi-corpus aggregation based on BERT-BiLSTM-CRF and corpus expansion based on a large language model. According to the BERT-BiLSTM-CRF-based multi-corpus aggregation, firstly, text data of a plurality of biomedical corpora are collected and screened, corpus preprocessing and aggregation operation are carried out through a BIO annotation format, and an aggregation strategy comprises direct aggregation and an aggregation mode for marking corpus sources. Corpus expansion based on a large language model aggregates corpora through a cue word mechanism, and performs model training in combination with a LoRA technology to generate expanded corpus data. And the expanded corpus is re-processed and then re-aggregated with the original corpus, and finally input into the BERT-BiLSTM-CRF model for named entity recognition. Through the combination of multi-corpus aggregation and corpus expansion, the method can significantly improve the efficiency and accuracy of named entity recognition, and enhance the generalization ability of the model.

The invention discloses a brain-controlled motion intention recognition method based on global optimization ANFIS, which utilizes an ANFIS model to perform brain-controlled intention classification decision, introduces a global spatio-temporal expression module, fully considers global spatio-temporal information of multi-input EEG data, introduces a rule attention module, and allows the model to recognize the most active intention-related fuzzy rule through an attention mechanism. And important intention features are selected to carry out accurate EEG motion intention classification decision. According to the method, the interpretability characteristic of the fuzzy inference system and the learning ability of the adaptive network are comprehensively utilized, system parameters can be changed according to priori knowledge, the extracted intention characteristics are explained layer by layer, and the final decision output is closer to the real output. According to the method, the performance of a brain-controlled intention recognition model applied to the field of brain-computer interaction is effectively improved, and the method can be suitable for assisting brain-controlled motion intention classification decision making based on EEG under a lower limb rehabilitation task and has high robustness and high efficiency.

The invention relates to the technical field of optical fiber wireless communication and mobile forward transmission, in particular to a digital analog optical fiber wireless forward transmission transmitting end parameter self-adaptive optimization method and system based on reinforcement learning. Decomposing the normalized wireless signal into digital parts of all levels and a final analog residual error part through multi-level modulation, and transmitting the digital parts and the final analog residual error part in a time division multiplexing manner; a receiving end represents inter-symbol interference caused by link nonlinear distortion and bandwidth limitation, and the signal-to-noise ratio of a recovery signal is calculated based on the error vector amplitude. Parameters of a transmitting end are optimized and modeled into a Markov decision process, states comprise rounding factors, scaling factors, constellation forming factors and pre-equalization filter tap coefficients of all levels, and actions are discrete increments of the parameters; a reinforcement learning agent is constructed by using a dual deep action value network, and training is updated in combination with experience playback and a target network. The reward function comprises a threshold satisfaction item and a signal-to-noise ratio change item; parameters are interactively updated in the training stage, and online self-adaptive adjustment is performed in the reasoning stage.

The invention discloses an intelligent glasses double-target speech enhancement method and system based on a multi-microphone array, terminal equipment and a medium, and relates to the technical field of speech enhancement. The method is applied to the intelligent glasses, and comprises the following steps: collecting noise-containing multichannel original time domain voice signals through a multi-microphone array of the intelligent glasses, and performing time-frequency domain conversion to obtain a frequency spectrum set; frequency domain complex spectrum amplitude-phase characteristics of each channel are extracted, a local time sequence context is introduced, and multi-channel space acoustic characteristics of amplitude-phase decoupling are constructed; and regulating and controlling the double-tone-area complex spectrum through pre-trained deep neural network modeling to obtain a target complex spectrum, and performing inverse time-frequency domain conversion and reconstruction to obtain an enhanced independent wearer voice signal and a preset direction target voice signal. According to the method, double-target voice parallel enhancement and non-target interference suppression are realized, the power consumption is low, the real-time performance is high, the method adapts to the harsh performance constraint of the end side of the intelligent glasses, and the robustness of voice enhancement in a complex noise scene is excellent.

The invention discloses a code abstract generation method and system based on hierarchical context awareness, and belongs to the field of natural language processing and software engineering. The method comprises the following steps: generating hierarchical annotations for a code warehouse based on a function call graph, firstly extracting a call relationship in the graph and eliminating a ring structure, and then classifying leaf nodes and generating annotations; a topological sequence from bottom to top is adopted in the core step, the importance score of a child node is calculated for each father node, child node annotations, class function descriptions and code bodies of the child nodes are fused, and a large language model is input to generate annotations till the whole graph is covered. The method is suitable for an automatic document generation scene of a complex software project, and through a technical path combining function call graph analysis, child node importance measurement and class function description, the accuracy and semantic consistency of code annotation are improved; the method can be widely applied to the fields of software maintenance, code understanding, intelligent development tools, automatic document generation of enterprise-level code warehouses and the like.

The invention relates to a dual three-phase permanent magnet synchronous motor model-free prediction repetitive control method and system based on a double-subspace virtual vector, and the method comprises the steps: decomposing the vector of a motor into a fundamental wave subspace and a harmonic wave subspace which are orthogonal to each other based on an obtained operation parameter, and constructing an independent super-local model in the two subspaces; constructing a linear expansion state observer in the fundamental wave subspace, and constructing a repetitive expansion state observer based on repetitive control in the harmonic wave subspace; respectively synthesizing an independent virtual voltage vector set and a decoupling virtual voltage vector set in the two subspaces; and calculating reference voltage vectors of the two subspaces, selecting an optimal virtual voltage vector and an optimal decoupling virtual voltage vector, respectively calculating optimal duty ratios, synthesizing control signals applied to each bridge arm of the inverter, and driving the dual three-phase permanent magnet synchronous motor. According to the method, model-free control of double sub-spaces is realized, model parameters of a motor system of an algorithm are controlled, and better robustness is shown when the parameters are mismatched.

The application relates to a low-complexity adaptive beamforming method based on phase apodization cross-correlation, and belongs to the technical field of ultrasonic imaging. The method comprises the following steps: focusing and Hilbert transforming a sampling echo signal to obtain a complex echo signal; constructing multiple groups of amplitude apodization functions with different side lobe levels and peak response, weighting the echo signal, calculating and comparing the estimated variances of the apodized signals; constructing multiple groups of complementary phase apodization function pairs, phase apodizing the apodized signal with the minimum estimated variance, and calculating the real part similarity coefficient matrix of the phase-apodized signal pair; summing and averaging the multiple groups of thresholded similarity coefficient matrices and performing two-dimensional mean filtering; and multiplying the sum result of the apodized signal with the minimum variance estimation and the filtered similarity coefficient matrix to obtain the output of a beamformer and form an image. The application has the advantages of low calculation complexity, high robustness, etc., and can significantly improve the resolution and contrast of an ultrasonic image.

This invention discloses a damage identification method for beam structures based on the frequency contour band method, comprising the following steps: (1) obtaining a panoramic view of the damage frequency; (2) considering the influence of noise and error, obtaining a measured frequency range covering the true frequency of the damaged beam; (3) taking n-order measured frequency ranges and marking them all on the panoramic view of the damage frequency to obtain damage frequency contour bands, each order of damage frequency contour bands representing all combinations of damage locations and damage degrees corresponding to that frequency; (4) projecting the damage frequency contour bands onto the xOy plane to obtain the intersection area of ​​n contour bands, all points within the intersection area are used to indicate possible damage locations and damage degrees; (5) calculating the center point of the intersection area, using the horizontal and vertical coordinates of the center point to represent the location and degree of damage. This invention considers the influence of environmental noise and test errors, enabling the frequency-based damage identification method to achieve a certain degree of noise immunity and exhibiting good robustness.

The application discloses a risk-aware event-triggered saturation T-S fuzzy system reliable H∞ control method, which comprises the following steps: constructing a global networked T-S fuzzy system coupled with actuator efficiency attenuation failure and input saturation constraint to construct a comprehensive operation risk index; constructing a drift and saturation-aware dynamic memory event trigger mechanism according to the comprehensive operation risk index; constructing a matching type saturation sector constraint model; constructing a non-PDC fuzzy fault-tolerant H∞ controller according to the global networked T-S fuzzy system combined with the drift and saturation-aware dynamic memory event trigger mechanism; and realizing risk-aware event-triggered saturation T-S fuzzy system reliable H∞ control. The application solves the technical problems of the prior art, such as insufficient flexibility of the event trigger mechanism, low precision of the regional performance configuration, strong saturation processing conservativeness, and difficulty in balancing communication efficiency and control performance.

The application relates to a method and system for identifying infrastructure damage based on an improved MCMC method. The method comprises the following steps: acquiring a mode of a test cantilever beam; forming a population according to a Markov chain obtained from a standard MH algorithm according to the measured mode; iteratively including a normal period with only differential evolution and a competition period with competition after the normal period; obtaining two groups of correction vectors after mutation, crossover and selection in each iteration in the competition period, combining the two groups of correction vectors with correction vectors in the normal period, and recording two groups of complete iteration Markov chain parameter optimization schemes as Y W and Y L ; calculating the average value of the corrected parameters after convergence as Y W and Y L , learning Y W from the winner and Y L from the loser, and taking Y W as the optimal correction scheme according to comparison with the existing measured mode. By adopting the scheme, the calculation efficiency is high, the precision is higher, and the robustness of the system is stronger.

The invention discloses a small target detection method based on a time-space domain recursive network. The method comprises the following steps: step 1, preprocessing image data as input data; 2, constructing a spatial-temporal feature extraction backbone network, inputting the input data into the backbone network, and generating feature maps of four stages; 3, constructing a time domain recursion module, and generating a fused spatial-temporal feature based on the feature map of the third stage; and step 4, constructing a lightweight detection head, constructing multi-scale features based on the spatial-temporal features and the feature maps of the other three stages, and outputting a classification result and bounding box coordinates through the lightweight detection head. The method has high detection precision, and the spatial-temporal feature fusion effectively enhances a target signal and inhibits background interference. The method is suitable for a static platform and a motion platform, does not need complex inter-frame registration, and has no strict constraint on a target motion mode; and meanwhile, the algorithm complexity is low, a fixed weight convolution kernel or a lightweight network can be adopted, and hardware implementation is facilitated.

The embodiment of the present invention discloses a model generation, image segmentation method, system, electronic device and storage medium. The model generation method may include: obtaining a sample image, an enhanced image and a segmentation label of the sample image, and using the sample image and the segmentation label, and the enhanced image and the segmentation label as a group of training samples respectively; training the original segmentation model based on multiple groups of training samples to obtain an image segmentation model; wherein the enhanced image includes a first enhanced image generated according to the sample image and the image template, and / or a second enhanced image generated according to the target feature map of the sample image and the target feature map template, the image template includes pixel points with a first value, the target feature map template includes pixel points with a second value, and the first value and the second value are less than the preset value. The technical solution of the embodiment of the present invention is to obtain an image segmentation model with strong robustness based on sample image training under a finite data distribution.

This invention relates to the field of physical oceanography, specifically to a method for rapidly identifying the upper and lower boundaries of the ocean thermocline. The method utilizes trigonometric functions as basis functions to construct a 10th-order trigonometric polynomial, and obtains the fitting parameters and results using the least squares method. Then, the first derivative of the trigonometric polynomial is calculated. The maximum value and its location, as well as all minimum values ​​and their locations, are determined. From the location of the maximum value, the two locations of the first minimum values ​​less than 1 / 10 of the maximum value are found on both sides. This yields the upper and lower boundaries of the ocean thermocline depth. This invention eliminates the subjectivity of parameter selection by eliminating the need for manually setting thresholds; it also avoids complex transformation and construction processes, resulting in a clear algorithm structure and high computational efficiency. It can operate stably in different sea areas and under various data precision conditions, significantly improving the automation, robustness, and computational efficiency of thermocline identification.

This invention discloses a method and system for synchronous inversion of multiple atmospheric parameters, belonging to the field of atmospheric laser remote sensing technology. The probe light and the reference light are obtained by splitting laser pulses; the probe light is expanded, reflected, and then perpendicularly incident into the atmosphere, where it is scattered by the atmosphere to obtain the echo light. That is, all parameters originate from the same measurement and the same spectrum, resulting in perfect spatiotemporal consistency of the data, efficient measurement methods, and accurate measurement results. The temperature and pressure parameters in the atmosphere are calculated using the normalized scattering spectrum, and then aerosol and wind speed parameters are obtained, achieving true synchronous inversion of the four key parameters of temperature, pressure, wind speed, and aerosols for the first time. Furthermore, in applications such as meteorological forecasting, achieving synchronous measurement of atmospheric temperature, pressure, wind speed, and aerosols can save costs and improve efficiency, which has significant practical implications.

The present application relates to a kind of multi-unmanned aerial vehicle narrow airspace cooperative path planning method based on distributed optimization, comprising the following steps: establishing the three-dimensional kinematic model of unmanned aerial vehicle, and defining narrow airspace boundary constraint;According to task demand, generate reference path for each unmanned aerial vehicle;Design distributed optimization controller based on local information;With reference path as tracking reference, use distributed optimization controller to solve control input vector in real time, to control the actual flight path of unmanned aerial vehicle.The method of the present application can realize completely distributed control, eliminate single point failure risk, and can ensure that unmanned aerial vehicle is collision-free and maintains desired formation, realizes speed consistency, improves cooperative efficiency.

This invention discloses a nighttime glass inspection system and method based on visible light and near-infrared images, belonging to the field of glass visual inspection technology. The nighttime glass inspection method based on visible light and near-infrared images includes: acquiring a near-infrared image and a visible light image of the same glass at night; processing the near-infrared image to obtain near-infrared reflectance and near-infrared illumination components, and processing the visible light image to obtain visible light reflectance and visible light illumination components; receiving the visible light reflectance and near-infrared reflectance components to obtain a near-infrared enhancement feature set; receiving the near-infrared illumination component, visible light illumination component, and near-infrared enhancement feature set to obtain glass image fusion features for the same glass at night; and convolving the glass image fusion features to obtain a glass mask image. This invention can extract visible light and near-infrared features of glass at night, obtaining high-quality nighttime glass surface inspection results.

This invention relates to the fields of artificial intelligence and medical imaging technology, specifically to a method and system for generating radiology reports based on temporal reasoning and evolutionary analysis. The method includes: acquiring the patient's current medical images, current examination reports, historical medical images, and historical examination reports; encoding the features of the current and historical medical images using a visual encoding network, then inputting these features into a temporal feature extraction network to obtain temporal visual features; comparing and aligning the temporal visual features with the corresponding text in the examination reports to obtain visual text features; and inputting the visual text features into a fine-tuned large language model to generate the report text. This invention, by performing temporal modeling on images from multiple patient examinations, can effectively capture the evolutionary patterns of diseases in images at multiple time points, thereby generating a diagnostic description with temporal consistency.

The present application belongs to the field of information technology, and particularly relates to a coal mine methane emission tracing method based on multi-isotope analysis and associated gas coordination. By fusing multi-isotope characteristics and associated gas ratio analysis, a multi-dimensional tracing system for coal mine regional methane emission is constructed, which significantly improves the accuracy and reliability of methane source identification. Compared with the traditional tracing method relying on single isotope or concentration statistical method, the present application can effectively distinguish various emission types such as coal mine gas, coking plant, coal-fired and agricultural sources in a complex mixed source environment, and realize high-precision cause analysis of multi-source methane emission. The present application realizes the spatio-temporal matching of isotope characteristics and transport path, and can quantitatively identify the influence of high-altitude and external source input on ground concentration.

The invention provides a font recognition method, device and equipment based on self-adaptive preprocessing and a medium, and the method comprises the steps: carrying out the self-adaptive preprocessing of a collected font image, the preprocessing comprises dynamic noise suppression, illumination unevenness correction and sub-pixel-level geometric correction, and obtaining a corrected character image; performing multi-dimensional feature extraction on the corrected character image, and constructing a fusion feature vector including local detail features, global style features and geometric correction features; and performing dynamic weight matching based on the feature contribution degree evaluation model, calculating the similarity between the to-be-matched character and the candidate fonts, and outputting the recognized font type according to the similarity result, thereby ensuring the font recognition accuracy.

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.

This application relates to a functional safety system and a method for using a functional safety system. The functional safety system includes a service module, a monitoring module, and a processing module. The service module includes at least one sub-service module. Each sub-service module is configured to send an error report to the module performing the monitoring operation in the event of an error, perform the monitoring operation based on a monitoring priority, and perform the processing operation based on a processing priority. The monitoring module is configured to perform the monitoring operation and the processing operation based on the processing priority. The processing module is configured to perform the processing operation. This functional safety system incorporates the service module into the functional safety monitoring service, enabling a higher level of functional safety mechanism.

The invention provides a mechanical arm trajectory tracking control method fusing model prediction and sliding mode control. The mechanical arm trajectory tracking control method comprises the steps that a discrete sliding mode controller is constructed, and a discrete sliding mode surface based on trajectory tracking errors is designed; calculating a sliding mode control law; constructing a dynamics prediction model based on an echo state network, and performing online learning and updating by taking historical state data and a sliding mode control law of the mechanical arm as input so as to predict a state track of the mechanical arm in a future time domain; constructing a model prediction controller embedded with sliding mode control, and introducing a prediction state and a sliding mode control law into an optimization objective function of the model prediction controller; and under the model prediction controller, an optimization problem is converted into a quadratic programming problem to be solved, a smooth optimization control quantity meeting physical constraints is obtained, and the control quantity acts on the mechanical arm system. The high-frequency buffeting problem of sliding mode control is effectively solved, and meanwhile high-precision trajectory tracking of the mechanical arm under strong nonlinear interference is guaranteed.

This invention discloses a control method for an underwater electro-manipulator based on model predictive control. The method includes: establishing a nonlinear dynamic model of the underwater electro-manipulator; establishing a predictive model of the underwater electro-manipulator using model predictive control; inputting the angular displacement and angular velocity of each joint of the underwater electro-manipulator into the predictive model in real time, and outputting the driving torque of each joint of the underwater electro-manipulator in real time, thereby controlling the continuous operation of the underwater electro-manipulator and realizing real-time control of the underwater electro-manipulator. This invention improves the control accuracy of the manipulator in underwater scenarios by utilizing model predictive control. It can promptly compensate for model mismatch when dealing with time-varying disturbances, thus exhibiting good robustness for nonlinear underwater manipulator systems under actual constraints such as torque, angular velocity, and angular acceleration. Specifically, it achieves relatively smooth control output at the procedural level under underwater transient disturbances, demonstrating good anti-disturbance performance.

The application provides a battery pack heating method and device, electronic equipment and storage medium, comprising: obtaining the external state parameter of each battery monomer; determining the target heating power of the battery pack based on the external state parameters of all battery monomers; for each battery monomer, determining the internal state parameter of the battery monomer based on the external state parameter of the battery monomer; determining the pulse parameter of the battery pack based on the target heating power and the internal state parameters of all battery monomers; wherein the pulse parameter comprises a reference pulse current; determining the actual pulse current of each battery monomer based on the internal state parameters of all battery monomers and the reference pulse current of the battery pack; and heating each battery monomer respectively based on the actual pulse current of each battery monomer and the pulse parameter of the battery pack. The application provides a technical solution, which realizes efficient, safe and balanced heating of the battery.