1993 results about "Time frequency domain" patented technology

The invention relates to the technical field of automation, in particular to an automatic monitoring and optimizing system for a fine chemical production process, which comprises the following steps of: extracting time-frequency domain fusion characteristic quantity of stirring torque power time sequence fluctuation data in real time through an incidence matrix construction module, dynamically inverting a thixotropic index by combining a deep neural network model, and optimizing the stirring torque power time sequence fluctuation data; the problem that a traditional method is difficult to perceive material rheological characteristics in real time is solved. The dynamic coupling analysis module analyzes the material viscosity change rate based on the thixotropic index, fuzzy PID control is adopted to generate a stirring speed adjusting instruction dynamically matched with the viscosity and a jacket temperature compensation value, and the defects of uneven mixing and local overheating caused by lagging adjustment of process parameters are overcome; the multi-target collaborative optimization module locks the mass optimization weight in the viscosity sudden change stage, rapidly stabilizes the reaction condition through a feed-forward compensation algorithm, dynamically balances the stirring power consumption and the heat transfer efficiency based on Pareto frontier search in the steady state stage, and solves the conflict between the mass and the energy efficiency target.

The invention relates to the technical field of vehicle motors, in particular to a vehicle motor stator and rotor life prediction method and system, and the method comprises the steps: obtaining temperature distribution data, vibration spectrum data and current harmonic characteristic data during the operation of a stator and a rotor; performing time-frequency domain fusion analysis based on the temperature distribution data and the vibration spectrum data to obtain thermal coupling degradation characteristics, and obtaining electromagnetic performance degradation characteristics based on the current harmonic characteristic data; training to obtain a life prediction model based on a mapping relation between historical degradation characteristics and a failure threshold value, inputting the thermal coupling degradation characteristics and the electromagnetic performance degradation characteristics into the life prediction model, and outputting a residual life prediction value of the stator and the rotor; and dynamically adjusting operation parameters of the motor according to the residual life prediction value. According to the method, the problems that the life prediction precision is low and the operation strategy cannot be dynamically optimized due to the fact that the collaborative degradation effect of the stator and the rotor of the vehicle motor under the multi-physical field coupling effect is difficult to quantify in real time are effectively solved.

The invention relates to the technical field of wireless signal detection, and discloses a wireless anti-interference signal detection method, device and equipment and a storage medium, and the method comprises the steps: collecting a standardized interference feature vector through a multi-scene wireless signal receiving device; performing interference feature enhancement processing on the standardized interference feature vector to obtain an interference feature sample set; constructing a dual-array cooperative detection system comprising a passive detection array and an active detection array, and performing real-time monitoring to obtain dual-path interference signal monitoring data; performing spatial domain and time-frequency domain dual feature extraction and fusion processing on the dual-path interference signal monitoring data to obtain a comprehensive interference feature; the receiver filtering parameter matrix and the detection threshold vector of the double-array cooperative detection system are dynamically adjusted according to the comprehensive interference characteristics, the method can adapt to changes of different interference environments in a self-adaptive mode, comprehensive coverage of communication spectrums is achieved, and the wireless anti-interference monitoring range and the spectrum sensing capacity are greatly expanded.

The invention relates to a railway traction substation state monitoring method and system, equipment and a medium. The monitoring method comprises the following steps: acquiring real-time monitoring data of the equipment in a railway traction substation; performing data preprocessing on the real-time monitoring data to obtain a preprocessed monitoring data set, performing protocol identification, and converting heterogeneous data in the monitoring data set into structured data according to a preset protocol template library; based on the structured data, time-frequency domain characteristic parameters of the equipment are extracted, and a multi-dimensional characteristic matrix is constructed; inputting the multi-dimensional feature matrix into a pre-trained hybrid diagnosis model, and generating an equipment health degree score and a fault probability value; according to the health degree score and the fault probability value, generating an early warning instruction in combination with a dynamic threshold algorithm; and generating a priority maintenance strategy through a maintenance strategy optimization model based on the early warning instruction and the equipment maintenance resource constraint condition. According to the invention, accurate perception and intelligent decision making of the equipment state are realized in a multi-source heterogeneous data environment.

The invention discloses a GIS partial discharge intelligent diagnosis system and method based on one-dimensional ultrahigh frequency signal analysis, and relates to the technical field of power electrical equipment intelligent monitoring, and the system comprises a signal collection and preprocessing module which is used for collecting ultrahigh frequency signals of GIS equipment and obtaining preprocessed signal data through a dynamic threshold algorithm; the discharge initial judgment module is used for performing multi-dimensional sequential judgment to eliminate interference discharge data so as to obtain effective discharge signal data; the feature extraction module is used for performing time domain kurtosis and pulse width analysis, frequency domain energy distribution analysis and time-frequency domain wavelet entropy calculation based on the multi-dimensional features of GIS partial discharge, and generating an optimized feature subset; and the type identification module is used for identifying the partial discharge type by using the integrated learning model to obtain a diagnosis result. According to the invention, the problem of unstable recognition accuracy caused by insufficient signal preprocessing, single feature representation and single classification algorithm in the prior art is solved.

The invention discloses a deep well rock burst early warning system and method based on multi-dimensional monitoring, and belongs to the technical field of deep well rock burst early warning. According to the method, multi-dimensional data such as stress, strain and microseism are collected through a monitoring network, and an aligned multi-source data set is obtained through space-time registration; after dynamic noise suppression processing matched with physical characteristics is adopted, strong correlation characteristics are screened through mutual information entropy; frequency domain, time domain and time-frequency domain features are extracted through principal component extraction and phase-space reconstruction, and a multi-dimensional state space data set is formed; and inputting the prediction model to obtain a danger level and trigger a corresponding early warning signal, and finally dynamically adjusting the monitoring network layout and prevention and control measures based on the early warning signal. According to the method, the early warning accuracy and real-time performance are improved, and effective technical support is provided for deep well rock burst prevention and control.

The invention discloses a bearing fault diagnosis method based on fusion of an improved capsule network and zero sample learning, and relates to the technical field of state monitoring and fault diagnosis of electromechanical equipment, and the method comprises the following steps: collecting a multi-mode signal during the operation of a bearing, employing an improved wavelet threshold denoising algorithm for the multi-mode signal to eliminate environmental noise, and then employing a zero sample learning algorithm for the multi-mode signal; according to the method, the improved wavelet threshold de-noising algorithm and the WPD and VMD fusion decomposition algorithm are adopted to extract the time-frequency domain mixed features as sample data, and the GAN is combined to expand the bearing sample data, so that the data dependence of traditional deep learning is broken through, the time-frequency domain mixed features are extracted through the improved wavelet threshold de-noising algorithm and the WPD and VMD fusion decomposition algorithm, and the time-frequency domain mixed features are extracted through the improved wavelet threshold de-noising algorithm and the WPD and VMD fusion decomposition algorithm. Small sample data learning is realized, and by training a pyramid capsule network and optimizing a cross entropy loss function and combining cross-modal joint optimization and a zero sample inference engine, the diagnosis accuracy of known faults is greatly improved, and unknown fault types can be effectively inferred.

The invention discloses a photoelectric event identification optimization method and system for a single photoelectron detector, and the method comprises the steps: collecting environment noise data, carrying out the modeling of environment noise, and generating a noise processing model; real-time operation monitoring data of the target single photoelectron detector are acquired, noise processing is carried out, and time-frequency domain waveform decoupling is carried out after processing is completed; performing pulse accumulation detection through a time-frequency domain decoupling result, marking an accumulation event, and separating accumulated pulses by using a pre-trained pulse separation network to obtain pulse separation information; generating a floating judgment boundary parameter according to the pulse separation information, and carrying out photoelectric event judgment to obtain a photoelectric event judgment list; and obtaining a historical photoelectric event discrimination list of the target single photoelectron detector, extracting an abnormal discrimination event, analyzing whether maintenance is needed, generating a maintenance self-inspection report, and pushing the maintenance self-inspection report. The detector keeps high robustness and anti-interference capability in the whole life cycle, and the identification error rate improvement caused by aging change is avoided.

The invention belongs to the technical field of intelligent equipment fault diagnosis, and particularly relates to a motor fault real-time diagnosis method and system based on LSTM and random forest. The method comprises the steps that a vibration signal, a current signal and a temperature signal of a motor are collected and preprocessed; performing feature engineering, extracting time domain, frequency domain, time-frequency domain and cross-modal correlation features, and determining an optimal static feature subset through a hybrid screening strategy; constructing a hybrid fault prediction model comprising a random forest model and an LSTM sequential network model; fusing prediction results of the two models by adopting a dynamic credibility weighted fusion mechanism; and performing real-time decision and hierarchical feedback control based on a fusion result. According to the method, advantages of multi-source information and the model are fused, the classification precision of the lag type is remarkably improved, real-time fault diagnosis and active protection are realized, the equipment maintenance cost is reduced, and the method is suitable for motor health management of intelligent agricultural equipment such as mowers.

The invention relates to the technical field of circuit breakers, in particular to a circuit breaker fault diagnosis method and system for switching an alternating current filter in an extra-high voltage converter station, and the method comprises the following steps: S1, synchronously collecting multiple parameters: synchronously collecting vibration acceleration signals, opening and closing coil current waveforms, arc light intensity and environment temperature and humidity parameters; s2, time sequence feature extraction: performing time-frequency domain decomposition on the vibration acceleration signal to obtain a mechanical action time sequence spectrum; s3, operation mode classification: determining the operation mode of the circuit breaker; s4, three-dimensional fault matrix construction: generating a three-dimensional fault matrix including spatial distribution characteristics; s5, dynamic threshold analysis: generating a diagnosis threshold interval dynamically adjusted along with the environmental parameters; and S6, comprehensive diagnosis output: outputting a mechanical jam grade, a contact ablation degree and an insulation deterioration early warning signal. According to the invention, the mechanical, electromagnetic and insulation characteristics of the circuit breaker can be reflected more comprehensively, and reliable guarantee is provided for safe and stable operation of a high-voltage power grid.

The invention discloses a wind turbine generator data analysis and fault diagnosis method and system based on big data and artificial intelligence. According to the method, a blade image, a vibration signal, audio data and operation parameters are synchronously acquired through an unmanned aerial vehicle multi-mode sensor and a ground monitoring system, and a multi-source heterogeneous data set is constructed; after the data is classified and preprocessed, image features, vibration time-frequency domain features and operation parameter key value pairs are extracted respectively; dimensionality reduction is carried out by using an auto-encoder, feature-level space-time alignment is realized through an improved DTW algorithm, and a multi-dimensional fault feature matrix is generated; a hierarchical diagnosis model including a GRU auto-encoder, an MLP network and an attention mechanism CNN is constructed, and training is carried out by taking minimization of sub-model deviation as an optimization target; and finally, fusing multi-source features to realize fault classification, and generating a visual diagnosis report. According to the method, efficient fusion and accurate diagnosis of multi-source heterogeneous data are realized, and the accuracy and the real-time performance of fault detection of the wind turbine generator are remarkably improved.

The invention discloses a method for identifying a multi-model fusion signal modulation mode based on a time-frequency diagram, and belongs to the field of signal processing. Preprocessing the original signal to obtain a signal sample; constructing a data set by using the normalized signal samples; constructing a fusion model; training a modulation identification module in the fusion model; performing modulation mode identification on an input unknown signal by using the fusion model to obtain a preliminary identification result; and performing comprehensive judgment on the preliminary recognition result of the fusion model by using a comprehensive judgment device to obtain a final recognition result. According to the invention, by combining a plurality of time-frequency analysis methods, the time-frequency domain feature information of the signal is fully extracted; the reliability and robustness of a final decision are improved by adopting a multi-model fusion framework, and the modulation recognition performance of the system under the condition of a low signal-to-noise ratio is improved.

The invention discloses a life prediction method based on health index construction and neural network fusion, and belongs to the technical field of equipment state monitoring and predictive maintenance. According to the method, through multi-source degradation feature extraction, common dynamic principal component analysis (CDPCA) dimensionality reduction, health index construction and normalization, deep learning multi-model modeling, integrated learning fusion and Bayesian optimization hyper-parameter optimization, online health assessment and residual life prediction of the equipment part degradation process are realized. Specifically, the method comprises the following steps: firstly, extracting time domain, frequency domain and time-frequency domain features from a sensor acquisition signal, and performing dimension reduction through CDPCA to obtain effective degradation characterization; then, weighting the main features to construct a health index (HI) curve, optimizing the weight through a genetic algorithm, and then performing normalization; a plurality of neural network models such as CNN, Bi-GRU, Bi-RNN, Bi-LSTM and SRNN are constructed based on the normalized HI sequence, and degradation trend modeling is realized; inputting the output results of the neural networks into an integrated learning module for fusion optimization; and finally, carrying out automatic optimization on the key hyper-parameters of the model by utilizing Bayesian optimization. In the equipment operation process, a normalized HI curve can be calculated in real time and input into the fusion model, and the residual life estimation value of the part is dynamically output. According to the method, high-precision, high-robustness and online life prediction can be provided under complex working conditions, the safety and reliability of equipment operation and maintenance are improved, and the method has wide engineering application value.

The invention relates to the technical field of voice processing, can be applied to service system platforms of medical health, financial science and technology, communication and the like, and discloses an audio compression and reconstruction method, device, equipment and medium. Residual vector quantization and inverse quantization are carried out to generate basic spectrum features; and compensating a high-frequency component by adopting a spectrum expansion network, predicting a time domain error in combination with a long-short-term memory network, generating a time domain residual compensation signal, and superposing the time domain residual compensation signal with the reconstructed time domain audio signal to obtain a target audio signal. Through residual vector quantization and neural network modeling, the audio reconstruction quality under low-bit-rate compression is improved, spectrum detail reservation and time domain error compensation are optimized, meanwhile, the calculation complexity is reduced, and the method is suitable for high-sampling-rate and resource-limited scenes.

The invention relates to the field of equipment diagnosis, and particularly discloses an equipment predictive maintenance method based on deep learning, and the method comprises the steps: carrying out the time domain amplitude normalization and frequency domain weighted normalization of training data, and carrying out the dual-channel feature fusion, so as to obtain a fusion feature vector; performing feature extraction on the fused feature vector by using multiple groups of self-adaptive wavelet kernels to obtain a self-adaptive time-frequency feature vector; constructing a weight population through statistical characteristics, and screening the optimal initial weight of the initial diagnosis model; carrying out multi-scale depth feature calculation, time-frequency domain attention feature fusion, fault prototype comparative learning and a pre-constructed total loss function on the adaptive time-frequency feature vector, and carrying out iterative updating on the initial diagnosis model to obtain a diagnosis model; and the equipment is diagnosed through the diagnosis model. Multi-scale feature fusion and a double-path attention mechanism can cooperatively capture short-time impact and a long-period mode, and the limitation of a traditional method in diversified fault scenes is overcome.

The invention discloses a GNSS deception signal detection method based on a time-frequency domain feature fusion deep learning model, and relates to the technical field of intelligent detection. The method comprises the following steps: acquiring original GNSS signal data, preprocessing the data, and dividing the data into a training set and a test set; in the model training stage, a pre-training lightweight deep learning model is utilized to extract time domain convolution features and multi-resolution frequency domain features, and weighted correction and fusion among the features are realized through a bidirectional attention mechanism; and carrying out iterative training on the model based on the fusion feature vector, and finally carrying out deception signal detection on a test set by utilizing the trained model. According to the invention, through a time-frequency domain feature deep interaction and model training mechanism, the deception signal is detected through the model, and the detection precision of the GNSS deception signal in a complex scene and the anti-interference capability of the model are significantly improved.

The invention belongs to the field of machining, and relates to a machining process quality online monitoring and feedback adjusting system and method. The system comprises a multi-modal data acquisition module used for acquiring a mechanical vibration signal and a high-frequency micro-damage signal; the feature preprocessing and cross-modal fusion module is used for carrying out time domain, frequency domain and time-frequency domain feature extraction on the collected mechanical vibration signals and high-frequency micro-damage signals, carrying out edge, texture and shape feature extraction on workpiece machining surface image signals, and introducing a cross-modal attention mechanism to achieve weighted fusion of heterogeneous signals. And finally generating a fusion feature vector with high discrimination. The machine tool / cutter state and workpiece quality intelligent analysis module is used for constructing an intelligent analysis model for the equipment operation state and the workpiece machining quality; inputting the fusion feature vector into an intelligent analysis model to obtain a quality evaluation result; and the quality judgment and feedback adjustment module is used for adopting a dynamic threshold adjustment mechanism to realize adaptive comparison with a process standard according to a quality evaluation result.

The invention discloses a lithium ion battery internal short circuit fault detection method based on time-frequency fusion, and the method comprises the steps: firstly constructing a fractional order equivalent circuit model fused with an electrochemical aging mechanism, and simulating the paths of conductivity reduction, active material loss and lithium inventory reduction in combination with an aging empirical formula; joint modeling of the aging process and the random internal short circuit is achieved, and multi-cycle voltage and current and electrochemical impedance spectroscopy data are obtained; secondly, extracting time domain features by using a time domain attention enhanced long-short-term memory network, and analyzing frequency domain information by using a multi-scale frequency sensing convolutional neural network; and then weighting and screening two types of modal features through a dynamic gating fusion module, introducing a cross attention mechanism to establish dependency mapping between time-frequency domain features, and finally outputting an internal short circuit fault detection result. The method can solve the problems that in the lithium ion battery aging process, due to lithium dendrite growth, the internal short circuit early fault is high in concealment, and single time-frequency characteristics are difficult to detect, and early high-precision recognition of the internal short circuit fault can be achieved.

The invention belongs to the field of power system engineering, and discloses a power distribution automation terminal diagnosis method and system based on multi-source wave recording feature fusion, and the method comprises the steps: obtaining the electric quantity data and equipment operation state data collected by a power distribution automation terminal; performing adaptive decomposition on the electrical quantity data by using a variational mode decomposition algorithm to obtain an intrinsic mode function; constructing a deep residual network model, carrying out fusion analysis on the time-frequency domain features of the intrinsic mode function, and generating a fault feature vector; establishing a multi-dimensional evaluation matrix based on the fault feature vectors, and integrating a plurality of indexes to output fault types and credibility scores; according to the fault type and the credibility score, generating a fault isolation strategy based on a Petri network model; and executing a dynamically adjusted self-adaptive self-healing control algorithm. According to the method, complex and changeable fault modes can be effectively identified, a complete collaborative verification mechanism is formed, seamless connection from fault diagnosis to self-healing control is realized, and the operation reliability of the power distribution network is remarkably improved.

The invention discloses a multi-mode pet health monitoring and motion artifact elimination method and system based on a millimeter wave radar, and relates to the technical field of pet health monitoring, and the method comprises the following steps: S001, building a unified time baseline and an energy fingerprint auditing surface, constructing an energy distribution model from a chest to a tail, and taking the model as a reference for artifact evolution, identifying a signal spectrum coupling trend in a time-frequency domain; and S002, based on the energy distribution model, performing causal playback on continuous time sequence signals acquired by the millimeter-wave radar, extracting a pseudo-motion energy nucleus caused by tail or limb movement, and calibrating a phase anchor point and a space observation area of a respiratory signal. According to the method, multi-source physiological data are fused, pure respiratory signals are extracted through energy distribution modeling, causal playback, phase anchor point calibration and three-dimensional resampling, risk assessment is achieved based on physiological credibility tensor, and the accuracy and anti-interference capacity of health monitoring are improved by combining phase conjugate traction and a space-time regulation strategy.

The invention relates to the technical field of cable partial discharge detection and diagnosis, provides a high-voltage cable multi-parameter intelligent diagnosis method, system and equipment and a storage medium, and aims to solve the problems of low accuracy and poor reliability of high-voltage cable multi-parameter diagnosis caused by multiple factors. The method comprises the following steps: decomposing a partial discharge signal after interference filtering, extracting a time domain amplitude change rate and an energy distribution parameter, and generating a multi-dimensional time-frequency domain feature vector; calibrating a multi-dimensional time-frequency domain feature vector, carrying out nonlinear manifold mapping, generating a low-dimensional feature vector matched with dielectric spectrum characteristics, matching a partial discharge type with the highest similarity with an identification result, judging an insulation degradation level in combination with a final loss threshold interval, and generating a comprehensive diagnosis report. According to the invention, the ultrahigh-frequency sensor array captures the partial discharge signal matched with the dielectric spectrum, precise identification of the discharge type and dynamic evaluation of the insulation degradation level are realized through multi-scale analysis and deep learning fusion, and the reliability of high-voltage cable diagnosis is improved.

The invention relates to an AI server cooling water pump vibration quantity self-adaptive calibration test method and system, an intelligent test platform integrates a six-dimensional force sensor and a temperature compensation vibration exciter, the platform rigidity is automatically calibrated, a water pump-pipeline system transfer function is obtained, and a rotating speed-lift-modal frequency three-dimensional mathematical model is established; a three-axis MEMS accelerometer array is arranged at sensitive parts such as a water pump bearing seat and a motor shell, and vibration, current and pressure signals are synchronously collected; carrying out time-frequency domain signal processing by adopting variational mode decomposition in combination with self-adaptive S transformation, and extracting a 128-dimensional full-frequency domain feature vector containing a modulation side frequency band; effective values of vibration acceleration, speed and displacement are calculated through a frequency domain integration algorithm, and current harmonic interference is corrected through an electromagnetic vibration compensation model; and finally, a vibration health degree evaluation model containing 18 characteristic parameters is established based on a support vector machine, the test system comprises an intelligent test platform unit, a multi-source sensing unit, an edge calculation unit and a data management unit, and microsecond-level synchronous acquisition and GB-level data throughput are realized through a time sensitive network. And online incremental learning and automatic generation of a test report are supported. The method has the effect of improving the water pump vibration quantity test precision.

The invention relates to a signal detection and accurate positioning method for a lift-off platform unmanned aerial vehicle, and the method comprises the steps: deploying an unmanned aerial vehicle-mounted platform, capturing a global electromagnetic signal in real time, outputting multichannel original signal data, carrying out the time-frequency domain joint analysis, extracting a frequency hopping period, a modulation mode and pseudo-random code features, and recognizing a target unmanned aerial vehicle signal. A control instruction link signal and an image transmission link signal are separated; the arrival time difference and the arrival phase difference of the target unmanned aerial vehicle signal are converted into a dynamic space coordinate system, a three-dimensional positioning equation set is constructed, and real-time space position coordinates of the target unmanned aerial vehicle are obtained through calculation; and extracting transmitting end characteristic parameters of control instruction link signals, constructing a control signal propagation path model in combination with the real-time position of the target unmanned aerial vehicle, and positioning geographic coordinates of an unmanned aerial vehicle control source by reversely tracking a signal transmitting direction and an attenuation law. The terrain shielding influence is effectively avoided, the target identification accuracy and positioning precision are improved, and the unmanned aerial vehicle control source traceability is realized.

The invention discloses a vibration signal anomaly detection method based on unsupervised learning, which relates to the technical field of vibration anomaly detection, and comprises the following steps: collecting vibration signals of electromechanical equipment during normal operation and synchronously recording working condition information; the collected vibration signals are preprocessed; setting a plurality of window segmentation lengths, and enabling each segment of vibration signal to generate a multi-stage sub-sequence; extracting time-frequency domain features of the vibration signals in the subsequences; the time-frequency domain features form feature vectors in a feature matrix splicing mode, feature standardization processing is carried out on the feature vectors, the feature vectors are fused with real-time working condition feature vectors obtained through working condition information, and a fused feature matrix is generated; and constructing an OCSVM model, and carrying out vibration anomaly detection on the electromechanical equipment by utilizing fusion feature matrix training. The method has the advantages that robustness and abnormal interpretability of single-class data are enhanced, the misjudgment rate is reduced through the dynamic confidence interval algorithm and probability distribution modeling, and the defect of insufficient model generalization is overcome through cross-modal feature fusion and working condition correlation modeling.

The invention relates to an intelligent power grid communication optimization method based on HPLC and HRF dual-mode communication, and the method is characterized in that the method comprises the following steps: S1, obtaining power grid communication data, and carrying out the preprocessing of the data; s2, according to the preprocessed intelligent power grid communication data, a feature extractor based on a deep neural network automatically extracts two-channel time-frequency domain features, feature importance of different modes is dynamically weighted through an attention mechanism, and a unified channel state matrix CSM after fusion is obtained; s3, constructing a routing strategy model based on reinforcement learning, and obtaining an optimal path decision according to the CSM matrix, the power grid network topology and the service priority label; s4, according to the path obtained by the optimal path decision, adaptive modulation coding is adopted to obtain an optimal modulation coding scheme and a dynamic adaptive channel condition; and S5, performing communication according to the optimal modulation coding scheme and the dynamic adaptive channel condition. According to the method, the communication state sensing capability in a complex scene is remarkably improved.

The invention discloses a bearing degradation trend prediction method and system based on multi-domain feature dynamic fusion and dimensionality reduction, and the method comprises the steps: collecting full-life vibration signals of a bearing, synchronously marking three stages of health, degradation and fault, constructing multi-dimensional features such as a time domain, a multi-scale frequency domain, a time-frequency domain, and the like; evaluating the cross-stage difference of the features by using double criteria of mahalanobis distance and information entropy, and adaptively adjusting the weight to complete optimization; threshold cutting, linear proportion, Softmax or hierarchical weighting strategy empowerment are automatically selected according to data distribution, and energy is reserved through PCA for dimension reduction. And a TCN-GRU deep network model is constructed. Real-time data are input into the model to predict the degradation state, if errors exceed the limit, feature reconstruction and model retraining are triggered, and full-life-cycle high-precision high-robustness multi-stage continuous online monitoring is achieved. The method aims at solving the problems that the diagnosis precision is limited and the working condition adaptability is insufficient due to the fact that single time domain or frequency domain features are excessively depended and the features of each stage of fault evolution are difficult to comprehensively characterize.

The present invention provides a dual-level thermal runaway warning method of a lithium battery based on a sound signal, comprising: obtaining a battery sound signal sequence; performing outlier identification on the battery sound signal sequence, and providing a level-1 thermal runaway warning when an abnormal data point exists; extracting a time-frequency domain feature of the abnormal data point, and identifying a presence of a thermal runaway expansion sound through a sparrow search algorithm-optimized eXtreme Gradient Boosting (SSA-XGBoost) algorithm, to providing a level-2 thermal runaway warning. In the SSA-XGBoost algorithm, optimal parameter adjustment is performed on a number of iterations, a learning rate, and a decision tree depth of the XGBoost algorithm through the SSA. A dual-level thermal runaway warning strategy is adopted to perform grading identification on general anomalies or deep anomalies, thereby effectively improving identification accuracy of a weak abnormal sound signal in an early stage.

The invention provides a baseband radio frequency signal data feedback processing method based on vector modulation, and the method comprises the steps: firstly receiving a baseband signal data set transmitted by a target radio frequency link in a preset frequency band, and the baseband signal data set comprises a plurality of signal frame original modulation parameters and corresponding radio frequency feedback signal waveforms; calculating a feedback error index composed of a phase offset, an amplitude distortion factor and a spectrum leakage factor according to a radio frequency feedback signal waveform, quantifying a modulation error, carrying out time-frequency domain conjoint analysis on the feedback signal waveform, extracting an error feature vector, inputting the error feature vector into a pre-training model for cross-frame correlation learning, and carrying out the correlation learning; according to the method, a dynamic adjustment parameter sequence containing a time sequence dependency relationship is generated, and finally a closed-loop feedback control link is constructed based on the dynamic adjustment parameter sequence and original modulation parameters, so that precise processing and optimal control of baseband radio frequency signals are realized, and the signal transmission performance is effectively improved.

The invention discloses a multi-person posture recognition method based on CSI (Channel State Information) and an attention mechanism, and the method comprises the steps: constructing a multi-person posture recognition system MultiFormer based on the CSI and the attention mechanism, designing a double-TokenTransformer architecture of a time-frequency double-domain Token TFDDT, converting an original CSI signal into a time domain Token and a frequency domain Token, maintaining the local feature continuity of the time domain and the frequency domain, and carrying out the recognition of the posture of a plurality of persons. And a multi-stage feature fusion network MSFN is developed to optimize a part thermodynamic diagram PCM and a part associated field PAF, and the fusion of CSI features and an intermediate attitude thermograph is realized through an adaptive channel-space attention mechanism, so that the anatomical consistency is ensured, and a multi-person scene is supported. According to the method, a MultiFormer system is provided and is different from existing CSI imaging processing, an interpretable TFDDT preprocessing method is adopted, the system adopts a multi-stage heat map estimation method to realize global attitude perception, the estimation precision is improved through iterative optimization, a multi-person scene is supported, attitude topology conforming to an anatomical structure is generated, and the method has the advantages of being simple in structure, convenient to operate and high in practicability. The method has the characteristics of privacy protection, low cost and no influence of illumination.

The invention discloses a resting electroencephalogram quality evaluation method and system based on double-branch comparative learning, and the method comprises the steps: collecting an original EEG signal X, carrying out the data preprocessing and data enhancement, generating two different enhanced views, transmitting the two different enhanced views to a double-branch encoder in parallel, respectively extracting a time domain waveform and a time-frequency domain rhythm feature, and carrying out the deep fusion, the fused features X1 and X2 are sent to a projection head, and through a self-supervised contrast learning mechanism, the network weight is optimized by using contrast loss; a small amount of labeled fine-tuning data sets including X and corresponding labels y are adopted, after flowing through a pre-trained double-branch encoder, the fine-tuning data sets are directly sent to a classification head connected with the back of the double-branch encoder so as to output a prediction result of an input data segment, and a mixed loss function including classification loss and comparison loss is adopted for training in the optimization process. According to the invention, a complete online real-time quality control system is established, and end-to-end real-time closed loop from data acquisition to quality evaluation is realized.