46 results about "Neural control" patented technology

The invention discloses an industrial AI assistant cross-modal interaction method based on a dynamic knowledge graph, and the method comprises the following steps: collecting texts, voices, images and multi-source sensor data in an industrial system, carrying out the modal recognition, feature extraction and time alignment, and generating an event feature set and a state feature set with timestamps. Modeling a time dependency relationship between event types by constructing a multivariable Hawkes model, and outputting an event trigger sequence and trigger strength; and in combination with a neural controlled differential equation model, guiding the state to evolve along with time and jump at a specific moment to form a state evolution trajectory. Performing fusion coding on the event and the state, constructing a dynamic knowledge graph with a causal structure and semantic continuity, and generating interactive output based on context reasoning; and after system feedback is received, the triggering strength and the state track are updated, and continuous evolution of the knowledge graph and reverse optimization of model parameters are achieved.

The invention relates to the technical field of neural engineering and brain-computer interfaces, in particular to a closed-loop multi-modal nerve stimulation system and method based on time interference, and the system comprises a multi-modal stimulation module which is used for integrating electrical stimulation, magnetic stimulation and optical genetic stimulation, and generating a time interference field domain; the neural state sensing module is used for collecting real-time electroencephalogram signals, blood oxygen concentration and neural metabolite level data; the neural control center is used for fusing neural state data and stimulation parameters, and dynamically adjusting time interference frequency and stimulation intensity through an adaptive algorithm; the time sequence cooperation engine predicts a neural response time phase based on a deep learning model, and optimizes a stimulation time sequence and a mode switching strategy; and the visual interaction platform is used for rendering the nerve activation thermodynamic diagram and the stimulation parameter adjustment curve in real time. Therefore, the problems of adjustment strategy solidification, low adjustment precision, insufficient energy conversion efficiency and the like in the prior art are solved.

The invention discloses a power control method and device for an industrial forklift charger, and the method comprises the steps: obtaining industrial forklift charging demand data and LLC resonant topology characteristic parameters, and constructing an LLC resonant converter; on the basis of the working frequency range and the power density requirement, through a probability learning algorithm and a random agent model, a random agent model for multi-objective optimization of magnetic element parameters is formed, and a high-frequency low-loss magnetic element is obtained; a high-frequency low-loss magnetic element is integrated into an LLC resonant converter, a random neural control barrier function algorithm is realized, and dynamic power control is performed on a charging process; based on the real-time data, a data-driven hybrid prediction control strategy is realized, and a charging parameter optimization instruction is generated; and in combination with the battery charge and discharge characteristic data and the battery aging state evaluation model, battery state evaluation and charging parameter adaptive adjustment are realized. The problems that a traditional charger is insufficient in control precision, low in efficiency, short in battery life and the like in a complex industrial environment are solved.

The invention discloses a scoliosis neuromuscular rehabilitation system based on a brain-computer interface and digital twinning, and the system comprises a multi-mode sensing module which is used for synchronously collecting a neurophysiological signal, a muscle electrical activity signal and a motion posture signal of a user; the intelligent processing and simulation module is used for carrying out fusion processing on the multi-modal signals, decoding the motion intention of the user and carrying out real-time simulation prediction on a biomechanical effect generated by the motion intention based on a personalized digital twinborn model of the user; and the decision and feedback module is used for generating and executing a personalized feedback instruction according to the decoding result and the simulation prediction result of the motion intention. The invention belongs to the technical field of medical rehabilitation engineering and biomedical engineering, and particularly provides a neuromuscular function remodeling method which is capable of directly providing a scoliosis patient with biomechanical accuracy, real-time interactivity and individual adaptability from a neural control source by constructing an integrated rehabilitation system.

The invention discloses an SCR denitration ammonia escape and NOx intelligent control method based on machine learning. The method comprises the following steps that boiler operation data are collected and preprocessed; constructing a feature input tensor set; inputting the feature input tensor into a prediction model based on a neural control differential equation, and outputting a predicted NOx and ammonia escape concentration sequence in combination with a Koopman observable space and an energy gating mechanism; constructing a candidate ammonia injection distribution set meeting the valve constraint based on the prediction result; multi-objective optimization is carried out under emission limitation and ammonia injection stability constraint, and an optimal ammonia injection distribution vector is generated; projecting to a feasible region barrier set and then generating an ammonia spraying execution instruction; and collecting feedback data for incremental training and model adaptive updating. The ammonia spraying control precision and the denitration efficiency are improved, and intelligent prediction of ammonia escape and self-adaptive control of NOx are achieved.

The invention discloses a multi-dimensional network intrusion behavior intelligent identification method based on deep learning. The method comprises the following steps: collecting network multi-dimensional data and generating a standardized network event set and a network control path input set; establishing a neural controlled differential equation model, and generating a continuous time context representation set through hidden state evolution; establishing a neuro-hox process identification model, performing intensity function modeling, and generating an event intensity prediction sequence set and an identification intermediate representation set; forming an intrusion behavior decision rule set and a reasoning configuration set through joint training; new network multi-dimensional data are collected, the reasoning configuration set operation model is loaded, and a new event intensity prediction sequence set and a new candidate trigger time set are output; and generating a network intrusion behavior recognition result set in combination with the intrusion behavior decision rule set. According to the method, time modeling and logical reasoning are fused, and high-precision intrusion identification is realized.

The invention discloses a carbon footprint real-time evaluation system based on the Internet of Things, and the system comprises the steps: collecting the high-frequency energy consumption and carbon emission activity data of industrial production equipment and environment monitoring equipment in real time, and forming a data sequence with a timestamp; performing anomaly detection, denoising and interpolation preprocessing on the data sequence; extracting high-dimensional continuous time features of the data sequence; predicting a carbon emission sequence based on a neural controlled differential equation model; time-varying carbon emission factors corresponding to the enterprise area and the power grid nodes are obtained in real time; and coupling the carbon emission sequence and the time-varying carbon emission factor sequence in a continuous time domain, calculating a real-time carbon footprint, and performing dynamic monitoring and visual display. According to the invention, high-precision real-time prediction and monitoring of the carbon footprint data are realized, and the real-time performance and accuracy of evaluation are improved.

The application belongs to the field of industrial control network security, and discloses an attack detection method for multi-element sensor data in an incomplete SCADA scene. An interpolation network based on a neural controlled differential equation is used to repair original samples under different missing degrees; a window self-adaptive mechanism is used to solve the problem of unreasonable sample window setting in the preprocessing process without prior knowledge. An adaptive mask mechanism is proposed to guide the detection model from the input side, enhance the understanding of the dependency between time and sensor nodes, and provide a certain degree of interpretability for the model from the mechanism definition angle. In combination with adversarial training, the application defects of the original diffusion in the SCADA anomaly detection field are optimized. Compared with other filling methods, the diffusion step is greatly compressed, and the training and detection efficiency is significantly improved. Full consideration is given to the application requirements in actual production, and the multi-element sensor data anomaly detection in the heterogeneous SCADA scene can be supported without prior knowledge.

The invention provides an intelligent fatigue recognition method and device, electronic equipment and a storage medium, and the method comprises the steps: evaluating a fatigue index of a target user through two dimensions, i.e., a target conduction velocity and a target fractal dimension, the target conduction velocity being capable of reflecting a muscle fiber function, the target fractal dimension being capable of reflecting neural control complexity, and the target fractal dimension being capable of reflecting neural control complexity; one-sidedness of single index evaluation is avoided, and muscle fatigue and central fatigue can be distinguished more accurately; through multi-channel and multi-time-scale analysis, signal characteristics of different muscle areas and different contraction rhythms are covered, and errors of a single channel and a time scale are reduced. Furthermore, the peripheral fatigue index and the central fatigue index of the target user are calculated in combination with the baseline conduction velocity and the baseline fractal dimension obtained based on the steady state surface electromyogram signal of the target user, and the method can adapt to the muscle basic states of different users, so that the evaluation result is more personalized and higher in accuracy.

The invention discloses an intelligent sensor cooperative supervision method and system based on the Internet of Things. The method comprises the following steps: collecting heterogeneous original monitoring data of various intelligent sensors and carrying out unified time sequence alignment; constructing a unified semantic mapping relationship to obtain standardized data; constructing a sensor state representation model based on a neural controlled differential equation, and obtaining a continuous hidden state vector; integrating newly obtained data in real time to update continuous hidden state vectors and calculating fusion feature representation; the current cooperation state is discriminated in real time through an anomaly detection discriminator; and when an exception is detected, starting a cooperative event disposal strategy, generating a supervision linkage disposal instruction, and executing closed-loop supervision. According to the invention, the real-time collaborative supervision of heterogeneous intelligent sensor data is realized, and the efficiency and accuracy of cross-platform data collaborative supervision are improved.

The invention discloses a primitive milling track generation method based on path learning, which comprises the following steps of: acquiring primitive feature data and historical milling track data, and respectively preprocessing the primitive feature data and the historical milling track data; inputting to a primitive feature coding module of the improved NCDE model, coding, setting an initial hidden state, and driving time evolution calculation; inputting to a multi-head output module for decoding, constructing a multi-target joint loss function and iteratively updating; inputting to an improved NCDE inference model, and carrying out coding and hidden state evolution decoding; weighting calculation is carried out, and a candidate milling track sequence corresponding to the highest value is screened out; and the optimal milling track sequence is converted into a track instruction file, and a numerical control machine tool is driven to execute milling machining. According to the method, the improved neural controlled differential equation and primitive feature learning are fused, intelligent optimization of the milling track is achieved, and the method has the advantages of being high in efficiency, long in cutter service life and excellent in quality.

This invention discloses an AI-based modeling method for assessing the health status of integrated power supply boxes, comprising the following steps: S1, constructing a multivariate time series; S2, modeling the multivariate time series trajectory using a neural control differential equation model to obtain the latent space trajectory; S3, performing dynamic modal decomposition on the latent space trajectory to construct a modal health space; S4, generating an enhanced training sample set using a manifold interpolation hybrid method in the modal health space; S5, inputting the hybrid modal vectors from the enhanced training sample set into a health classification module; S6, inputting the latent space trajectory into a temporal semantic playback module to generate the final predicted state trajectory; S7, comparing the final predicted state trajectory with the actual historical state trajectory, and generating an anomaly marker and performing a state rollback operation when the semantic difference exceeds a set tolerance threshold. This invention integrates neural differential modeling, interpolation enhancement, and temporal semantic playback to achieve accurate assessment of the health status of integrated power supply boxes.

Systems, devices, and methods for treating drug-refractory epilepsy are disclosed. [Solution] In one embodiment, a method for treating epilepsy is disclosed, comprising detecting electrophysiological signals of a subject using a first electrode array coupled to a first intravascular carrier. The method further comprises analyzing the electrophysiological signals using a neuromodulator electrically coupled to the first electrode array, and stimulating a target site in the subject's body using a second electrode array coupled to a second intravascular carrier implanted in a portion of a body blood vessel above the base of the subject's skull.

This invention provides an advanced AGI (Awareness, Intelligence, and Knowledge) mind guidance system and method that solves problems such as complex consciousness content analysis. It includes: S1: Neural control decoding; S2: Developmental gap analysis; S3: Strategy generation and ethical review; S4: Multi-channel dynamic guidance; S5: Performance retrospection and evolution. This invention has advantages such as good consciousness content analysis effect and high level of intelligence.

The invention discloses a control method, system and equipment for anesthesia prognosis optimization and a storage medium, and relates to the technical field of biomedical engineering. The cognitive function evaluation data, the brain structure image data and the brain function image data of the patient before the operation are calculated, and a cognitive expression index and a nerve image index are obtained; calculating the neural image index and the cognitive performance index to obtain a cognitive reserve index; determining the risk level of the target patient according to the cognitive reserve index, and inputting the physiological data of the patient and the risk level into a technology library for query to obtain a core regulation and control technology; adjusting the stimulation parameters of the core regulation and control technology according to the cognitive reserve index, and generating a noninvasive nerve regulation and control scheme according to the core regulation and control technology, the adjusted stimulation parameters and the stimulation target; and sending the non-invasive nerve regulation scheme to non-invasive nerve regulation equipment to control the non-invasive nerve regulation equipment to intervene the target patient before the operation. By implementing the technical scheme provided by the invention, the stress ability of the brain of the patient to surgical anesthesia is improved.

The embodiment of the application discloses a traffic speed prediction method and device based on self-attention neural control, relates to the technical field of intelligent transportation, and can improve the precision of traffic speed prediction while maintaining low calculation complexity. The application comprises: obtaining historical data of traffic speed, and generating an original input sequence; establishing an AB-NCDE model, and performing a forward propagation process on the original input sequence; establishing a loss function and performing a backward propagation process; and using the trained AB-NCDE model to predict traffic speed. The scheme is used for traffic speed prediction.

The application discloses a kind of based on digital twinning linear control chassis energy consumption simulation method and system, including following steps: S1, constructs digital twinning model, generates state sequence by collecting and processing operation data;S2, using neural control differential equation extraction fusion state;S3, based on fourier neural operator simulation calculation power sequence;S4, through long-term memory network prediction trend, total energy consumption is generated using Simpson integral;S5, disturbance control parameter is generated disturbance energy consumption sequence using SASP algorithm;S6, compare two kinds of energy consumption, and control parameter is optimized using PPO algorithm;S7, according to the total energy consumption of optimization result regeneration, and with actual energy consumption error analysis, update digital twinning model.The application fuses neural control differential equation, fourier neural operator etc., with the advantages of high modeling precision, energy consumption prediction accuracy and high simulation efficiency.

The application discloses an AI pest situation identification and early warning method and system based on multispectral imaging. The method uses a multispectral imaging unit to obtain multi-band images of pests, and synchronously obtains time, point, weather, and crop growth period information. Through correction, registration, and band reliability evaluation, standardized multispectral images are obtained. Based on a background spectrum dictionary and sparse reconstruction residual, the pest area is extracted and the adhered pests are separated. Combined with the prior key parts, the spectral shape part joint feature is constructed. The collaborative network with a neural controlled differential equation spectral line coding branch and a part hypergraph attention branch is used to complete pest identification. Then, the unknown pest is identified in combination with the prototype memory library, and a pest situation spatio-temporal hypergraph is constructed to realize risk early warning. The method can improve the pest identification accuracy, unknown pest identification ability, and pest situation early warning accuracy in complex scenes.

This invention proposes a multi-scale cortical-muscle coupling network analysis method based on ordinal patterns. First, the invention uses multivariate variational mode decomposition to decompose brain electromyography (EMG) signals at the same frequency scale. Then, it uses a multivariate binary temporal partitioning transfer network method to calculate the coupling strength and coupling adjacency matrix of the multi-component signals, constructing a cortical-muscle coupling network. Complex network parameters are then used to analyze the characteristics of this coupling network. This method has advantages in representing the characteristics of information transmission between the cortex and muscle and their variations across different grip force patterns. It also demonstrates the hierarchical nature of causal links between brain EMG signals, confirming the method's potential in deconstructing the intrinsic connections between the cortex and muscles during upper limb movement and the mechanisms of neural control of muscle movement.

The application discloses a world action model enhancement method and device fusing structured prior and continuous dynamics, wherein the method takes human videos in a first view and a third view as input, constructs scene-event graphs and structured priors such as causality, counterfactual constraints and execution constraints; under weak labeling conditions, performs spectrum analysis on local time sequence features of actions, and jointly encodes the time sequence features and space-time features into latent actions; further, continuous time latent dynamics modeling is introduced into a unified world action model through a neural control differential equation, and joint reasoning is performed on video state changes and latent action recovery in combination with task semantics, structured priors and counterfactual conditions. Finally, long-time rolling error accumulation is inhibited through real observation rewriting, action block generation and asynchronous closed-loop execution, so that the feasibility of weakly labeled video action learning, the rationality of action generation and the stability of closed-loop control are improved.

The invention discloses a signal processing method, a brain-computer system, electronic equipment and a storage medium. The method comprises the following steps: acquiring a tested target neural signal; decoding the target neural signal by adopting a pre-trained decoder to obtain a decoding result; according to the decoding result, the controlled peripheral is controlled to continuously move towards an expected contact target outside the preset area until the controlled peripheral makes expected contact with the expected contact target, the controlled peripheral is a tested expected control target, and after the controlled peripheral makes contact with the area boundary of the preset area, the controlled peripheral moves towards the expected contact target according to the target motion attribute, and the controlled peripheral is controlled to be in expected contact with the expected contact target. The target motion attribute is the speed and direction when the controlled peripheral makes contact with the region boundary. The technical problem that a feedback control strategy in the prior art needs to continuously acquire neural signals to continuously iteratively control a cursor to move to an expected contact target in small steps, so that cursor control does not conform to a tested movement habit, namely does not conform to a neural control principle, is solved.

The application discloses a standing-up and walking test method and system based on inertial sensors and electromyography technology, relates to the technical field of gait assessment and rehabilitation detection, and comprises the following steps: data acquisition is completed by using inertial sensors and electromyography technology; through a first algorithm, differences between left and right support phases and swing phases are comprehensively evaluated to determine whether the gait is symmetrical and stable; and through a second algorithm, muscle coordination of a patient is evaluated. The method realizes synchronous data fusion of a behavior layer and a neural control layer, lays a data foundation for subsequent construction of a precise and multi-dimensional gait and electromyography evaluation model, realizes joint evaluation of two core indexes of gait symmetry and stability, improves detection capability of neurological gait disorders, provides reliable criteria for individualized rehabilitation intervention, realizes modeling and difference measurement of timing control characteristics, and improves recognition capability of neuromuscular control disorders.

The invention discloses a regional power grid cooperation method based on knowledge graph driving, and the method comprises the following steps: collecting regional power grid data, and generating a multi-source heterogeneous power grid data set; constructing a regional power grid knowledge graph, and initializing a federal heterogeneous graph neural network and graph neural controlled differential equation combined framework; executing federal heterogeneous graph neural network coding, and generating local embedding and mode weight decoupling parameters and local coding parameters; modeling a running diagram neural controlled differential equation, and generating a continuous time state trajectory and a kinetic parameter set; performing federal aggregation on a client set result to generate global model parameters; issuing global model parameters and updating a local model; and generating a regional power grid coordination result set based on the knowledge graph data, the control path input and the updated model parameters. According to the method, the cross-regional power grid collaborative prediction precision and the calculation efficiency are remarkably improved, and the operation safety and the scheduling reliability are effectively guaranteed.

The application discloses a public safety crowd flow early warning method and system fusing space-time residual learning, relates to collecting video sensing data, passing count data, wireless residence migration data and scene context data of a target public area, and performing time synchronization and space mapping; a region-channel directed topological graph is constructed according to a building plan, a BIM model, a CAD drawing or an electronic map, node features and edge features are extracted, and a node-edge joint space-time state tensor is generated; baseline flow prediction results are obtained based on trend period decomposition and neural controlled differential equations, and residuals are modeled based on fractional order gated time series convolution and hypergraph attention propagation, crowd state prediction values in a future prediction window are generated, risk indexes are calculated, and early warning grades, risk areas and disposal suggestions are output. The application can improve the accuracy and adaptability of crowd flow early warning in complex public scenarios.

The invention relates to a non-linear control system-oriented micro neural control barrier function security verification method, which supports more efficient verification of NCBF of micro monotonic activation functions such as tanh in a framework for verifying NCBF by a CROWN-based method, and is suitable for analysis of a continuous time system because the NCBF is micro. The derivative interval is not indirectly derived through linear relaxation of the activation value, but the real extreme value of the sigma '(z) is directly solved on the pre-activation interval, and the derivative boundary is tighter. As the derivative boundary is more accurate, back propagation accumulative relaxation is significantly reduced, and the final Jacobian interval is closer to a real value. Compared with a traditional SMT / MIP / CROWN method, the method has the advantages that acceleration is achieved by several times while strict safety is kept, and near-real-time deployment is facilitated.

The invention discloses a ground source heat pump invisible level fault diagnosis method based on fault evolution process modeling, which comprises the following steps: firstly, establishing a fault evolution model based on discrete level fault data to capture a nonlinear change rule of an equipment state among different fault levels; and then, complementing the complete fault evolution data by using a data generation technology so as to make up for the deficiency of training data and enhance the adaptability of the model to different fault levels. And finally, a fault diagnosis model is trained on the basis, the diagnosis capability of the model on unknown-grade faults is improved, and the stability and generalization capability of the intelligent fault diagnosis method in actual industrial application are improved. The time sequence variational auto-encoder potential space linear modeling method based on the neural controlled differential equation is provided for solving the problems of high time-varying property and high concealment of early grade faults of a ground source heat pump system in a shallow geothermal energy development and utilization system, and the blank of the unknown grade fault diagnosis technology of the ground source heat pump system is filled.

The invention discloses an automatic driving high-risk test scene generation method and device and a storage medium. The method comprises the steps that key interaction track fragments before collision are extracted from historical collision data; constructing a continuous encoder by using a neural controlled differential equation, and mapping the discrete trajectory into a context vector representing continuous dynamics; by taking the vector as a condition, diversified potential codes are generated in a hidden space through a conditional continuous normalization stream, and the diversified potential codes are restored into a physically consistent track through a decoder combined with a gating cycle unit; calculating the minimum collision time and post-intrusion time of the generated trajectory to quantify the risk; and finally, by taking a single accident as a seed, synthesizing high-risk scene variants related to the single accident in batches through coding, generating and screening processes, and realizing directional efficient amplification of the test scene library. According to the method, the problems of low fidelity of the generated scene and weak risk pertinence of the existing method are solved, and the efficiency and coverage rate of the security test are remarkably improved.

The application provides a nerve stimulation system, comprising a stimulation device, the stimulation device comprising a signal generating circuit, a control module, a signal collecting circuit and a multiplexed electrode, the stimulation device being configured to be implanted in vivo, to generate a stimulation signal and to apply the stimulation signal to a target nerve through the multiplexed electrode; the control module being configured to perform a closed-loop titration procedure: the signal generating circuit applies a titration stimulation signal to the target nerve according to a preset mode, in which the signal intensity of the titration stimulation signal starts from an initial threshold value, stepwise increases to a peak value, and then gradually decreases to an end threshold value after the peak value is maintained for a period of time or a plurality of stimulation cycles; the signal collecting circuit is capable of collecting an evoked potential corresponding to the application of the titration stimulation signal to the target nerve; and when an evoked potential period meeting a target threshold value is selected, a stimulation parameter with the lowest signal intensity in the evoked potential period is selected as a parameter of a treatment stimulation signal, which is used for the subsequent application of the treatment stimulation signal.

The invention provides an original traffic risk grading handover method based on anomaly detection. The method specifically comprises the following steps: S1, collecting original traffic, and constructing a confidence coefficient structure vector containing a probability component, a stability component, a data quality component, a similarity component and a time continuity component; and S2, generating a flow control sequence, a confidence control sequence and an abnormal trajectory control sequence in a continuous time window, and keeping consistent in position dimension. And S3, inputting the three types of control sequences into a neural controlled differential equation, and introducing a step length adjustment strategy and a position adaptive control kernel to obtain an abnormal power trajectory code. And S4, the track code and the confidence coefficient structure vector are sent into an ordinal number regression model, and a sample weight and a time sequence regular term are added. And S5, generating a risk level according to a weighted loss and monotonicity constraint learning level threshold. And S6, transferring the information to different processing units according to a grade result. According to the invention, the risk identification and classification stability in a continuous time scene is improved.

The embodiment of the application provides a kind of based on neural control hierarchical hybrid planning articulated vehicle motion planning method.Application in vehicle engineering technical field, the method is by obtaining the starting pose, target pose and environmental obstacle information of vehicle;The starting pose, target pose and environmental obstacle information of vehicle are input into trained neural guide planner and are analyzed and handled, obtain first path planning result, neural guide planner includes: environment encoder and planning network;According to the feasibility verification condition of pre-established, the segmented verification result of first path planning result is verified, and the segmented verification result is obtained;According to segmented verification result, the kinematic planning controller constructed is used to optimize and adjust first path planning result, and the final path planning result is obtained, and the final path planning result includes vehicle pose information and corresponding control instruction.The method improves the planning efficiency and path feasibility, enhances the robustness and practicality under complex working conditions.