19results about How to "Achieve robustness" patented technology

This invention belongs to the field of nonlinear data classification technology and discloses a nonlinear data classification method based on deep transfer learning and visual recognition capabilities. This invention transforms high-dimensional numerical features into structured images, enabling convolutional neural networks to leverage their powerful visual recognition capabilities to handle tabular data classification tasks. This image-based representation naturally reveals the implicit nonlinear relationships within the data. By further integrating deep transfer learning, this method effectively alleviates the data sparsity problem in small-sample scenarios and improves generalization performance through a series of training optimization strategies. The framework proposed in this invention achieves an effective connection between the numerical feature space and the image-based deep learning architecture, providing theoretical innovation and practical value for complex nonlinear data classification tasks.

The application discloses a data-driven optimal control method for a grid-connected inverter. The method comprises the following steps: constructing a DeePC optimization controller of the grid-connected inverter on the grid side; acquiring a data matrix of the grid-connected inverter in a preset historical period and an initial trajectory after a preset adjacent period, inputting the controller, setting parameters for different control targets and types of the grid-connected inverter, outputting an optimal input data sequence after processing, generating a sinusoidal pulse width modulation wave signal and acting on the grid-connected inverter, and realizing data-driven optimal control. The method can automatically perceive and adapt to the dynamic characteristics of the actual power grid through real-time data-driven predictive control, can present or enhance the control behavior of the grid-constructing or grid-following inverter by modifying the cost parameters, can realize multi-mode flexible, optimal and coordinated control, has excellent dynamic response and steady-state performance, is beneficial to safe and stable operation, and improves the ability to adapt to complex and variable operating conditions and power grid strength.

This disclosure provides an ammonia injection method and apparatus based on a denitrification system. The method includes: acquiring current boiler parameters of a coal-fired boiler; collecting nitrogen oxide (NOx) parameters at the denitrification inlet of the denitrification system as a first NOx parameter; collecting NOx parameters at the denitrification outlet of the denitrification system as a second NOx parameter; collecting NOx parameters at the total exhaust outlet of the denitrification system as a third NOx parameter; preprocessing the current boiler parameters, the first NOx parameter, the second NOx parameter, and the third NOx parameter respectively; determining the current required ammonia quantity for the denitrification system based on at least one of the preprocessed current boiler parameters, the first NOx parameter, the second NOx parameter, and the third NOx parameter; and injecting ammonia into the denitrification system according to the current required ammonia quantity. This method ensures that the calculated ammonia injection quantity is significantly more accurate.

The invention discloses an API path aggregation identification method and device and a storage medium, and relates to the technical field of electronic digital data processing, and the method comprises the steps: carrying out the preliminary aggregation identification of API path data based on the path structure features of the API path data; if the preliminary aggregation identification is not successful, performing deep aggregation identification on the API path data based on word segmentation features and / or statistical features of the API path data; and generating an aggregation identification result of the API path data. According to the method and the device, the API path aggregation identification accuracy is improved.

The invention relates to the technical field of computer vision and digital image processing, and discloses an adhesion object image segmentation method and system based on an improved watershed algorithm, a computer readable storage medium and a computer program product. Reconstructing and normalizing the distance map through morphology; extracting a foreground mark from the normalized distance map; based on the foreground mark and the background mark of the binary image, performing mark-controlled watershed transformation on the modified gradient image to obtain an initial segmentation result; and carrying out contour extraction and shape fitting on each region in the initial segmentation result, and outputting a final segmentation contour and corresponding morphological parameters. According to the method, robust and accurate segmentation of an adhesion object is realized through optimized preprocessing and mark extraction strategies in combination with shape fitting of post-processing, over-segmentation is effectively inhibited, and a smooth contour better conforming to a real form is obtained.

The application discloses a sample model training method and system based on data driving, relates to the technical field of model training, and comprises the following steps: obtaining an initial sample set and inputting the initial sample set and a noise vector into a generator together to generate an initial prediction sample; constructing a composite feature tensor based on the initial sample, monitoring the difference between the prediction sample and a reference label in a feature space of a discriminator and the change trajectory of a total loss of the generator, and outputting a training termination condition tensor; feeding back the tensor to the generator as a condition constraint to update the prediction sample, and inputting the prediction sample and the initial sample set into the discriminator together to perform adversarial training, wherein the total loss of the generator is a weighted sum of an adversarial loss and a regularization loss, and network parameters are updated in reverse propagation according to the total loss; in the process of continuous updating of the network parameters, feature resampling is triggered to dynamically correct training data, and a final generator network is output when the difference between two consecutive rounds of training reaches a convergence threshold, so that the method realizes closed-loop correction of adaptive determination of training convergence.

This invention provides a calibration-free path reconstruction method for buried cables based on Topo-NeRF and active sensing. Addressing the issues of traditional magnetic field inversion relying on calibration and being prone to mismatch in environments with strong interference such as reinforced concrete, the method preprocesses raw data to obtain a spatiotemporal feature matrix. This matrix undergoes feature extraction, spatial alignment, and drift compensation to eliminate extrinsic parameter calibration, resulting in multi-source fused data. This multi-source fused data is combined to characterize the cable's magnetic field using a continuous three-dimensional topological manifold. An implicit neural radiation field is constructed and coupled with differentiable electromagnetic rendering. The cable body and branch nodes are separated within a persistently cohomologically constrained topological bottleneck layer. The initial reconstruction results are combined with conditional diffusion and ground-penetrating radar dielectric priors to enhance low signal-to-noise ratio weak fields, ensuring consistency between the reconstruction results and the underground physical structure. Based on reconstruction uncertainties, topological entropy is calculated, and online acquisition trajectories are planned to form a closed-loop active sensing system. Finally, a three-dimensional path model is output, enabling rapid detection of underground cables.

This invention discloses a reverse logistics network design method for addressing the inherent uncertainties of construction machinery, comprising: Step S1: determining the endogenous and exogenous uncertainties and the construction machinery recycling and remanufacturing objectives that need to be considered in formulating a reverse logistics network construction machinery recycling and remanufacturing plan; Step S2: constructing a mathematical optimization model for maximizing the total profit of the reverse logistics network, determining the influence relationship between decision variables and endogenous uncertainties, and constructing a decision-dependent uncertainty set; Step S3: obtaining the final mixed-integer linear programming model, solving it using Python and the operations research optimization software Gurobi, and obtaining the optimal profit value and the optimal solution for the recycling and remanufacturing decision variables; Step S4: determining the influence of uncertainties on the pricing strategy, expected profit, and recycling objectives of each node center in the reverse logistics network based on sensitivity analysis. This method can resist the disturbance effects caused by uncertainty and achieve robustness in the recycling decisions of the main enterprise.

The invention provides a micro-nano satellite multi-constraint fine control method based on actuator dynamics, and belongs to the field of spacecraft control, and the method comprises the steps: considering the electromechanical characteristics of a reaction wheel, and building a micro-nano satellite attitude deep coupling model containing the dynamic characteristics of the reaction wheel; aiming at reaction wheel axle friction and counter electromotive force interference, designing a double-interference observer to accurately estimate the reaction wheel axle friction and counter electromotive force interference; designing a nonlinear state-dependent obstacle function to convert a constraint state into a tracking control variable according to the star rotating speed and reaction wheel current physical constraint; and a composite controller is constructed by combining a double-interference observer and a nonlinear state-dependent barrier function, and the design of the high-precision controller of the micro-nano satellite is completed. According to the method, the problem of various physical constraints in micro-nano satellite control can be solved, the interference of the reaction wheel is compensated, and the method has the characteristics of high reliability and high accuracy.

A synchronization system and method incorporates an absolute location sensor, such as a position sensor, and a relative location sensor, such as a Hall sensor, of an electric shift lever system. The synchronization system includes: a shift lever sensor configured to detect a manipulation signal of a shift lever when a vehicle traveling mode is switched; an electric motor configured to switch shift stages according to manipulation of the shift lever; a Hall sensor attached to the electric motor and configured to detect a relative angle of rotation; a position sensor configured to detect an absolute location of the electric motor; and a controller configured to receive signals generated from the shift lever sensor, the Hall sensor, and the position sensor, and control the electric motor such that a shift stage according to manipulation of the shift lever is switched based on the received signals.

The application discloses a channel optimization method for configuring an electroencephalogram monitoring device, and belongs to the technical field of biomedical signal processing and artificial intelligence. The method obtains high-density multi-channel electroencephalogram data and a depression risk label of a subject group, pre-processes and segments the electroencephalogram data, extracts multi-domain features such as time domain, frequency domain, nonlinear entropy value and inter-channel connectivity of each physical channel, and trains a machine learning classification model in a cross-validation framework. Further, a channel-level grouping permutation importance strategy is used to calculate channel contribution, and a minimum optimal channel combination is determined based on contribution ranking and a performance threshold to generate an algorithm deployment package containing a channel mask index and a lightweight classification model. The application significantly reduces the number of electroencephalogram channels, reduces hardware complexity, power consumption and computing load under the premise of ensuring the performance of depression risk assessment, and is suitable for engineering deployment of low-density electroencephalogram monitoring devices.

This invention discloses a stepped steer-by-wire redundancy control method and system for steering actuator failure, applicable to intelligent electric vehicles with independent steering and drive for all four wheels. When steering actuator failure is detected, the location of the failed wheel is determined and its steering angle is locked. Based on the state of the remaining healthy steering wheels, the maximum yaw moment they can provide is estimated in real time using a tire model. The total required yaw moment is calculated based on driver input and vehicle status. By comparing the two values, a compensation mode is automatically selected: if the steering force is sufficient, a pure steering compensation mode is entered, optimizing only the steering angle of the healthy wheels; if the steering force is insufficient, a hybrid compensation mode is entered, first controlling the healthy wheels to operate in the optimal efficiency state, then calculating the missing steering force step size and compensating through differential torque. This invention pre-sets differentiated strategies for multi-wheel failure scenarios, achieving stepped safety protection from single-wheel to multi-wheel failure, improving vehicle safety, and is low-cost and robust.

The invention discloses a context anaphora resolution method and system based on dialogue structure perception, and relates to the field of natural language processing, and the method comprises the steps: obtaining multiple rounds of dialogue corpora, storing dialogue entities in a shared memory pool and a private memory pool respectively, clearly distinguishing entities with effective individual cognition and global common-known entities, and obtaining an entity with effective individual cognition and a global common-known entity; therefore, anaphora ambiguity caused by confusion of knowledge ranges of different speakers is eliminated; when anaphora resolution is carried out, through multi-layer structure signal fusion such as a significance recursion updating mechanism driven by a dialogue structure, semantic role consistency judgment and round correlation attenuation modeling, an interpretable cognitive compatibility scoring system is established, and an entity selection mechanism jointly constrained by multi-dimensional information is formed. Therefore, by introducing a dialogue structure perceived double-domain entity memory modeling mechanism, pronoun forepointing accuracy and robustness in complex scenes such as long dialogue, cross-speaker and multi-round interaction are effectively improved.

The invention relates to the technical field of data processing, in particular to an exhibition and display interaction logic processing method and system and a storage medium. Comprising the steps of collecting a user interaction request and a historical access log to form an initial execution path set; constructing a branch model, and identifying a high-frequency condition judgment sequence; if the sequence is not matched with the operation data, optimizing to obtain an adjusted condition judgment rule; extracting real-time features of the user interaction request from the operation data to generate feature vector representation, and traversing the adjusted rule to determine an optimal execution path; comparing the optimal path with the initial path set to obtain a path optimization index and generate a resource allocation adjustment scheme; when the response delay is reduced, the condition judgment rule is updated to form a final interaction processing mechanism; follow-up interaction requests are processed accordingly, and the execution efficiency and the dynamic adaptation capacity are evaluated. The problems that an existing exhibition and display system is fixed in interaction logic and difficult to adaptively optimize are solved, and dynamic optimization and efficient response of exhibition and display interaction logic are achieved.

The application discloses an energy management method for a new energy three-electricity system of an engineering vehicle, and relates to the technical field of new energy three-electricity systems.The method comprises the following steps: S100, analyzing the driving requirements of the engineering vehicle; S200, intelligently deciding the power; and S300, executing in real time based on feedforward.The application constructs a three-layer closed-loop control architecture of accurate requirement perception, intelligent power decision and safe feedforward execution, and constructs an energy management method with intelligent decision capability and strong execution force.The method not only eliminates the risk of engine stalling due to the superposition of electric-hydraulic load from the root by feedforward constraint based on real-time hydraulic power, but also realizes intelligent scheduling and fine management of battery and fuel energy by means of fixed electric quantity following and regenerative braking recovery strategy.Meanwhile, the safety execution is ensured by battery power boundary protection and anti-integration fan mechanism, so as to comprehensively guarantee the stability and service life of the three-electricity system.

The application discloses a context reference resolution method and system based on dialogue structure perception, and relates to the field of natural language processing. The method comprises the following steps: acquiring multi-round dialogue corpus, storing dialogue entities in a shared memory pool and a private memory pool respectively, and distinguishing entities that are only valid for individual cognition from globally known entities, so as to eliminate reference ambiguity caused by confusion of knowledge ranges of different speakers; when performing reference resolution, a multi-layer structure signal fusion is performed through a dialogue structure driven saliency recursive updating mechanism, semantic role consistency determination and round correlation decay modeling, an interpretable cognitive compatibility scoring system is established, and an entity selection mechanism jointly constrained by multi-dimensional information is formed. Thus, through the introduction of the double-domain entity memory modeling mechanism based on dialogue structure perception, the accuracy and robustness of the antecedent before the pronoun in complex scenarios such as long dialogue, cross-speaker and multi-round interaction are effectively improved.

The application discloses a human posture recognition system based on a wearable sensor and belongs to the technical field of sensor measurement. The human posture recognition system collects foot bottom pressure through a pressure sensor module, filters and denoises and performs smooth processing through a signal processing module, and wirelessly transmits sensor signals to an upper computer through a wireless transmission module. The upper computer adopts a hidden Markov algorithm and a random forest fusion algorithm to recognize postures. The system comprises a pair of flexible pressure sensor arrays, the sensor arrays are worn on the foot bottom to collect foot bottom pressure signals, the signal processing module is completed by the upper computer, the wireless transmission module selects a Bluetooth module, and the upper computer performs real-time recognition after training the model. The system improves the recognition rate, is high in adaptability, and is suitable for medical treatment, human-computer interaction, sports training and other fields.

The invention discloses a multi-modal sensor environment perception fusion prediction method based on a neural network, and relates to the technical field of computer-aided prediction. The method comprises the following steps: collecting multi-channel data and a calibration reference value of a sensor array; preprocessing and constructing a data set; constructing a fusion neural network including frequency domain modeling, a linear trend and a nonlinear residual branch; designing a joint loss function fusing data fitting and a physical prior regular term, and introducing an adaptive learning mechanism of regularization weight to carry out joint training; and finally, performing fusion prediction on real-time sensor data by using the trained model. According to the method, frequency domain global modeling and trend separation are unified through a three-branch explicit decomposition architecture, the precision, robustness, cross-working-condition consistency and real-time processing efficiency of multi-modal sensor fusion prediction in a complex industrial internet environment are effectively improved in combination with an adaptive training mechanism, and deployment on an edge platform with limited resources is facilitated.

The application relates to the technical field of adaptive testing, and particularly provides a WebUI adaptive testing system and method based on multi-agent cooperation and multi-modal perception, which comprises a multi-modal perception subsystem, an agent cooperative testing subsystem and an assertion and verification subsystem. The method comprises the following steps: adjusting element positioning strategies through a continuous interaction feedback mechanism for a multi-dimensional interface feature set, and outputting an element descriptor set; deploying a multi-agent cooperative decision unit comprising an interface understanding agent, a test logic agent and a data agent; outputting a semantic test script formed through a cooperative fusion mechanism; performing pixel-level difference calculation, identifying visual inconsistency areas through multi-scale image comparison; performing business-level logic difference, analyzing the consistency of interface element states and business rules, and outputting a hierarchical verification report. The application can maintain high coverage, low maintenance cost and reliable defect positioning capability of the test in an environment where pages frequently evolve.