420 results about "Sensitivity coefficient" patented technology

The invention relates to the technical field of electrical fault monitoring, in particular to an intelligent mobile substation electrical fault monitoring method, which comprises the following steps of: synchronously acquiring an electrical monitoring signal, a multi-dimensional environment noise signal and a dynamic working condition parameter of a mobile substation through a multi-source sensor; the method comprises the following steps: constructing an environmental noise floor vector group by adopting multi-scale spectrum mode decomposition, filtering an environmental noise interference component from an electrical monitoring signal through orthogonal projection filtering, and outputting a baseline correction signal; a working condition disturbance response field matrix is constructed based on time domain and frequency domain correlation analysis of dynamic working condition parameters, gradient sensitivity coefficients are calculated, and components strongly related to working condition disturbance and residual components weakly related to equipment faults are separated out; reconstructing the residual component into a pure fault feature vector; and finally, fault type diagnosis and risk early warning are carried out on the basis. The method effectively solves the problem of fault feature annihilation caused by noise pollution in a complex environment and the problem of false alarm and missing alarm caused by confusion of working condition disturbance and real fault signals.

The invention provides a tunnel deformation monitoring system and method based on a distributed optical fiber sensing technology, and relates to the technical field of tunnel monitoring. A spiral winding type optical fiber is arranged in the longitudinal axis direction of the surface of a tunnel lining, a transverse optical fiber is arranged on a key section, and the data acquisition frequency is set to be 1Hz. Dividing the optical fiber into a plurality of data acquisition points at equal intervals, obtaining the real-time temperature and Brillouin frequency shift change of each point, measuring the temperature and strain sensitivity coefficient, and calculating the total strain; calculating long-term temperature strain compensation based on the long-term temperature component, generating an instantaneous temperature difference through the real-time temperature and the long-term temperature component, further calculating instantaneous temperature strain compensation, subtracting the two kinds of strain compensation from the total strain to obtain stress strain, and evaluating the deformation level of the tunnel based on the stress strain. And the accuracy of evaluation is judged through clustering analysis.

The invention discloses a large language model weight and activation combined quantification method and system, and belongs to the technical field of model quantification. The method comprises the following steps: collecting and preprocessing a calibration set, inputting a large language model to execute forward propagation, and recording an activation matrix; for each embedding dimension, counting maximum activation absolute values of all lexical elements on the dimension; generating a global threshold value by combining a quantile statistical method with the global sensitivity coefficient, and determining that the dimension with the maximum activation absolute value exceeding the global threshold value in the dimensions is an outlier dimension; respectively designing scaling factors for the normal dimension and the outlier dimension, and generating a reconstruction weight matrix; using Bayesian-gradient joint optimization to reconstruct a truncation threshold value of the weight; calculating a scaling factor of the reconstructed weight matrix to obtain a reconstructed quantized weight matrix; quantizing the activation matrix of the current layer according to the embedding dimension by applying a scaling factor; and performing multiplication calculation with the reconstructed quantization weight matrix to obtain a multiplication output result in an integer domain, and then performing unified inverse quantization recovery.

The invention belongs to the technical field of eddy current sensor calibration, and discloses an eddy current sensor for a high-temperature environment and a calibration method thereof. The method comprises the following steps: synchronously acquiring working current, voltage, surface position, temperature, standard displacement signals and original output voltage of an eddy current sensor, constructing an oxide layer equivalent circuit model, and optimizing to obtain thickness distribution; building a thickness increase model, and building a correlation model to calculate time-varying impedance; calculating a sensitivity coefficient function, and designing a piecewise calibration function to obtain a calibration parameter set; and finally, dynamically compensating the original output voltage in combination with the calibration parameter set to obtain the calibrated displacement. A temperature field, a standard displacement signal and an original voltage signal are collected, a temperature-oxide layer-impedance three-dimensional dynamic model is established, and the nonlinear influence of oxide layer growth and temperature gradient is compensated in real time.

The invention provides a decoupling and Kalman filtering temperature strain demodulation method and system for double fiber gratings, and relates to the technical field of optical fiber sensing. The method comprises the following steps: firstly, designing an FBG sensing module for measuring environment temperature change and strain information; the FBG sensing module comprises a two-channel sensing system composed of an FBG1 sensor and an FBG2 sensor, and is used for measuring environment temperature change and strain information at the same time. Then, aiming at optical signals which are acquired by the two FBG sensors and contain physical quantity information, a wavelength demodulation module is used for acquiring the central wavelength drift distance of reflection spectrums of the two FBG sensors; temperature sensitivity coefficients and strain sensitivity coefficients of the two FBG sensors are obtained, and a linear equation set is constructed for independent variation of temperature and strain to perform matrix decoupling so as to distinguish the influence of the temperature and the strain on the change of the central wavelength; and performing optimization estimation on the decoupled temperature and strain by using the prior information of the temperature and the strain and combining a real-time measurement value by adopting a Kalman filtering method.

The invention relates to the technical field of enterprise system construction, in particular to an enterprise standard system construction system based on dynamic classification and strategic adaptation, which comprises a dynamic classification module, a strategic adaptation verification module and a feedback optimization channel module, wherein strategic document semantic analysis and industry benchmark data are fused in real time through a dynamic classification module to generate a strategic weight coefficient, an environment sensitivity coefficient is calculated by combining compliance dynamics and a market risk model, and the fusion weight ratio of the strategic document semantic analysis and the industry benchmark data is dynamically adjusted based on threshold judgment, so that standard label weight distribution and hierarchical division are realized; the strategic adaptation verification module tracks the dynamic contribution degree of a standard node by using a node influence attenuation model of the association map, and detects strategic deviation by executing the deviation between an index and an expected value in real time; and after the feedback optimization channel module analyzes the deviation signal, the fusion weight is adjusted according to the deviation direction and amplitude, and standard level elastic lifting is triggered through sliding window monitoring, so that strategic suitability real-time verification and quantitative response to external environment change are realized.

The invention discloses a focusing control method and related equipment, and the method comprises the steps: synchronously collecting the data of a multi-modal sensor, extracting features and parameters, inputting a multi-modal fusion model subjected to joint optimization to generate a motor control instruction, and achieving the focusing through the hierarchical control of rough adjustment, fine adjustment and dynamic tracking. The dynamic weight fusion network deploys the weight of each modal data according to the environmental sensitivity coefficient, and combines the historical memory and trajectory pre-judgment of the alignment trajectory prediction network to reduce invalid calculation and repeated search and greatly improve the focusing speed; the multi-mode sensor works cooperatively to replace a special hardware function, so that the cost and the power consumption are reduced; the dynamic weight fusion network can adjust the modal confidence according to the environment and enhance the anti-interference capability in the complex light environment; the alignment trajectory prediction network and the fuzzy self-adaptive PID controller are in hierarchical cooperation, the problem of erroneous judgment of a complex scene is solved, and the micro-distance and dynamic tracking precision is improved; the model is subjected to pruning, distillation and quantitative optimization, light weight is achieved, and low-cost equipment is adapted.

The invention relates to a power capacity evaluation method based on a dynamic confidence interval and double-layer optimization, which solves the problems of inappropriate single-side modeling, static evaluation, source-load coupling modeling data processing mode and the like in the prior art, and adopts the technical scheme that a prediction error of wind power output is obtained based on historical data and a prediction value of the wind power output; based on the prediction error, determining an influence parameter generating the error; creating a power supply side model based on the prediction error and the influence parameter; obtaining a demand response sensitivity coefficient and a time-varying adjustment parameter based on the historical load data, the demand response execution record and the external influence factor; creating a load side model based on the demand response sensitivity coefficient and the time-varying adjustment parameter; creating a source-load joint scene model based on the power supply side model and the load side model; optimizing the source-load joint scene model through a double-layer robust optimization model; an assessment risk probability index and a capacity vacancy expectation are determined based on the optimized source-load joint scenario. The method has the advantages that a source-load joint scene model is created, and the coverage range and precision of an evaluation result in time and space dimensions can be remarkably improved; meanwhile, by means of a dynamic confidence interval adjustment mechanism and a double-layer optimization strategy, effective balance between the electric power market economy and the operation safety is achieved.

The invention relates to the technical field of industrial equipment monitoring, and discloses an industrial equipment state monitoring and fault early warning method and system based on big data, and the system comprises a data collection and fusion module, an intelligent diagnosis module and an early warning execution module. Through innovative design of a spatio-temporal feature fusion engine, the problem of spatio-temporal mismatch of vibration, temperature, current and other signals is solved, dynamic characteristics of millisecond-level vibration signals are captured through a long-short-term memory network, a spatial topological relation of equipment is modeled through a graph convolutional network, cross-scale alignment of multi-physical-quantity data is achieved, the limitation of physical significance isolation in a traditional method is broken through, and the method has a good application prospect. The composite fault recognition capability is improved, the diagnostic value of unstructured data such as sound emission is released, a full-life-cycle self-adaptive mechanism is constructed through a dynamic threshold algorithm, the working condition fluctuation tolerance is intelligently adjusted through a sensitivity coefficient, an early warning model continuously optimized along with the operation time is formed, the stiffness defect of a static threshold is changed, and the early warning effect is improved. And an accurate intelligent early warning system is established.

The invention provides a myopia occurrence risk assessment method based on multi-source data, and relates to the technical field of myopia prevention and control, and the method comprises the steps: collecting user physiological parameters and dynamic behavior data through multiple sensors, and carrying out the preprocessing and labeling; detecting a behavior short-term trend by applying a sliding window algorithm, generating a myopia risk index change process, identifying a false risk reduction interval and a delay window period, and marking a myopia risk index rebound event and an accumulated rebound risk peak value; when the behavior continuously reaches the standard, a physiological parameter verification process is started, a closed-loop intervention suggestion is generated based on the rebound early warning point, and intervention feedback is collected; comparing the score change with the actually measured physiological data direction, suspending the process when a conflict event is triggered, and generating a fusing log; and extracting prevention and control compliance abnormal data to update the causal atlas, non-linearly adjusting the model weight and the sensitivity coefficient according to the comprehensive feedback index, iteratively optimizing the model until early warning is removed, and outputting a report.

The invention relates to the technical field of data processing, and provides a resistance line-associated underwater bench blasting numerical simulation error correction method, which is characterized in that land and underwater measured data are matched through the same resistance line and similar material parameters, and the resistance line and the water depth are used as basic input of an error prediction model; a reference basis is provided for error correction of subsequent numerical simulation; calculating a target result in combination with a prediction error after initial simulation, comparing a simulation result with the target result during iterative simulation to generate an iterative error, and analyzing and determining sensitivity coefficients of different error classifications in a resistance line and the iterative error so as to adjust material parameters to reduce an invalid iterative process; simulation results are processed according to stress areas in a zoning mode through an equidistant slicing method, adjustment precision is verified in combination with the root rate and the damage area, and the reliability of the simulation results is improved; and meanwhile, parameter data passing verification are input into a back propagation neural network, a correction parameter mapping library is constructed, and the design efficiency of an underwater blasting scheme is improved.

The invention discloses an intelligent evaluation system for advertisement putting effect monitoring, and relates to the technical field of advertisement putting, and the system comprises the steps: loading a bidirectional pre-training language model and a multi-modal fusion model, and synchronizing advertisement metadata to generate a structured feature matrix; collecting user behaviors, biological characteristics and environment data in real time, generating a four-channel environment map, splicing the four-channel environment map with the structured characteristic matrix, and inputting the spliced four-channel environment map and the structured characteristic matrix into a Transform encoder to form a joint characteristic vector; matching a high-density high-emotion region according to the candidate budget allocation scheme, and generating a cognitive load index in combination with the facial micro-expression of the user; user interaction data cognition load scenes are analyzed, core advertisement group weights are evaluated and optimized through a regret value mechanism and a Shapley value, and a final dynamic putting scheme is generated; the response difference of the advertisement group to the cognitive load is quantified through the sensitivity coefficient, and the marginal contribution of the advertisement group is evaluated in combination with the Shapley value, so that the accurate identification and budget inclination of the core advertisement group are realized.

The invention provides an intelligent optical cable performance prediction method and system for a data center, and relates to the technical field of data center performance analys.The intelligent optical cable performance prediction method comprises the steps that firstly, an optical cable microstructure disturbance quantization model is constructed, and an optical cable is divided into a plurality of virtual quantization units in the longitudinal direction; each unit packages a real-time stress-strain state parameter and a response sensitivity coefficient of the basic structure unit, inputting a Brillouin scattering spectrum frequency shift data flow collected in real time into the model for disturbance source analysis, generating a physical excitation decomposition sequence, driving a virtual quantization unit to perform state evolution iterative operation according to the physical excitation decomposition sequence, and obtaining a Brillouin scattering spectrum frequency shift data flow model; a stress-strain state evolution track set is obtained, a time-space correlation network of optical cable link disturbance propagation is constructed according to the stress-strain state evolution track set, network flow characteristic analysis is conducted on the time-space correlation network, an early warning node set is extracted, the future evolution trend of the early warning node set is calculated, and serialized optical cable performance abnormity early warning information is generated. The optical cable performance abnormity can be predicted in advance, and stable operation of the data center optical cable is guaranteed.

The invention discloses a face gear worm grinding machine error compensation method based on sensitivity analysis and a multi-source error mapping model, which comprises the following steps of: 1, constructing a multi-source error-tooth surface error mapping model based on a vector decomposition and truncation function method, the truncation function method is used for quantitatively describing propagation mechanisms of geometric errors and thermal errors in a machining chain and influences of the propagation mechanisms on tooth surface deviation, and high-order error terms are eliminated to improve calculation efficiency; 2, responding to the output of the mapping model, applying an improved Morris sensitivity analysis method, distributing a weighting coefficient higher than a geometric error component for a thermal error component, and calculating a weighting sensitivity coefficient so as to identify a key error component which has the greatest influence on the tooth surface precision; and 3, based on the key error component, a motion axis compensation value is generated through an error compensation model, and the compensation value is input into a machine tool control system to adjust the relative position of the worm grinding wheel and the workpiece, so that the tooth surface error is reduced.

The invention relates to the technical field of earthquake monitoring, and discloses a comprehensive earthquake prediction method, which comprises the following steps of: acquiring a continuous micro-motion data sequence to reconstruct an empirical green function sequence, and dividing the empirical green function sequence into a compression phase subset and a stretching phase subset by utilizing a synchronously acquired theoretical earth tide stress phase; respectively extracting a causal branch waveform corresponding to the positive time axis and a non-causal branch waveform corresponding to the negative time axis, and determining a weight index for inhibiting the non-structural interference by calculating a reciprocity residual error between a causal wave velocity offset and a non-causal wave velocity offset; correcting a wave velocity offset mean value by using the weight index to obtain a stress sensitivity coefficient reflecting the stress state of the earth crust medium; according to the method, a physical criterion is established by using an elastic wave propagation reciprocity principle, decoupling of medium attribute change and noise source spatial drift is realized, and visual wave velocity offset induced by environmental interference is eliminated.

The invention relates to an active power regulation and control method and system based on boundary constraint, and the method comprises the following steps: constructing a boundary constraint condition model, and achieving the dynamic updating of the boundary constraint condition model through a visual maintenance tool; the load rate of the power grid equipment is monitored in real time, whether the power grid equipment is out of limit is judged according to the scene definition table, working condition classification is performed, and generation of a control strategy is triggered; a control strategy is generated based on sensitivity analysis and resource priority, and a comprehensive index calculation model is adopted to screen and adjust resources according to a strategy constraint definition table; sensitivity analysis obtains a sensitivity coefficient based on real-time topology of power grid equipment in combination with load flow calculation, and the resource priority is determined according to a resource rule definition table; and when a single type of resources cannot meet the regulation and control requirements, generating a comprehensive control strategy containing multi-type resource regulation variables, and executing closed-loop effect verification and strategy iteration through a strategy index definition table.

The invention discloses a button switch service life detection method, a detection device and a medium, button use parameters and physical response signals are collected in real time, instantaneous pressing force and accumulated fatigue damage amount are separated by adopting a signal decoupling technology, and a composite damage index is calculated based on a nonlinear rule. And the life prediction module dynamically selects an exponential decay model or a power law decay model in combination with the total damage amount, and outputs a residual life evaluation result. A calibration unit is arranged in the device, standard pressure is periodically applied, response signals are collected, the model sensitivity coefficient and fatigue parameters are dynamically adjusted through a recursive optimization algorithm, and long-term stability of prediction precision is guaranteed. According to the method, full-process closed-loop control from data acquisition, damage feature extraction to intelligent prediction is realized, the accuracy and real-time performance of button service life evaluation are remarkably improved, and the equipment shutdown risk and maintenance cost caused by button failure are effectively reduced.

The invention discloses an HPC curtain wall safety intelligent monitoring method and system, and the method comprises the steps: obtaining curtain wall multi-domain monitoring data and a cable climbing robot scanning record, recognizing a vibration abnormal component, generating a defect sensitivity coefficient, and constructing a damage response field domain; scanning and generating a modal capacity contour surface, setting a resonance early warning trigger node, extracting frequency response distribution to determine a resonance dangerous area, and generating a high-sensitivity monitoring area and a sensor control strategy; an environment load sequence is collected and projected to a damage response field domain to form an effective load bandwidth, and a safety control range is determined by combining structure bearing requirements; performing path search by applying a layered monitoring and adjusting scheme, identifying a stress concentration position through keel deformation modulation, and generating a structure control anchor; analyzing a structure control anchor to generate a fatigue accumulation rate, forming a damage development speed in combination with a scanning record, and extracting collaborative deformation parameters; and comparing the cooperative deformation parameter with a preset value to form a control deviation signal, and generating a self-adaptive control sequence to realize intelligent monitoring and active adjustment of structure safety.

The invention discloses a side plate type heat exchanger performance testing system and method, and relates to the technical field of heat exchanger testing, and the system comprises a computer which is used for constructing a digital twinborn model, generating multiple groups of dynamic working condition data, inputting the multiple groups of dynamic working condition data into the digital twinborn model to obtain dynamic performance data, screening according to the sensitivity coefficient of each piece of dynamic performance data to obtain sensitive working condition data; the test equipment executes the sensitive working condition data and collects actual measurement state data of the side plate type heat exchanger; and the computer judges whether the performance of the side plate heat exchanger meets the design requirement according to the actually measured state data. Physical modeling is carried out on the heat exchanger in a digital twinning mode, heat exchange simulation is carried out on the heat exchanger, invalid test working conditions are deleted from a parameter set, only sensitive working conditions are reserved, and time, manpower and energy cost needed by testing is greatly saved.

The invention provides a medical insurance fund health monitoring method and system based on a neural network, and a medium. Comprising the steps that multi-source heterogeneous medical insurance data are collected and input into a two-channel neural network model in parallel after space-time normalization processing, a time sequence prediction channel predicts a fund sustainability index through a long and short-term memory network in combination with an attention mechanism, an anomaly detection channel calculates a medical fund abuse risk probability and a regional circulation balance degree through a graph neural network, and the medical fund abuse risk probability and the regional circulation balance degree are calculated. And finally, fusing the indexes to generate a health index, comparing the health index with a dynamically adjusted threshold value to realize three-level early warning, constructing a medical risk propagation network, generating a sensitivity coefficient based on a clustering coefficient, betweenness centrality and historical risk intensity, dynamically optimizing an early warning threshold value, and enabling the system to support anti-factual causal analysis and generate policy intervention suggestions. According to the method, the problems of insufficient medical insurance fund space-time heterogeneity modeling, neglect of a risk conduction mechanism and poor static threshold adaptability of a traditional method are solved, and accurate monitoring and active prevention and control of fund health are realized.

The invention relates to the technical field of power system signal measurement, in particular to a mutual inductor digital sampling method and system. The method comprises the following steps: acquiring a signal output by a secondary side of a mutual inductor for a power system in real time; determining the load fluctuation degree at the current moment; determining a dynamic sensitivity coefficient at the current moment; obtaining a dynamic reference mean value, a dynamic relaxation parameter and a dynamic decision threshold value used in the CUSUM algorithm at the current moment; determining the cumulative sum between the signal and the dynamic reference value in the CUSUM algorithm at the current moment; and determining the sampling frequency of the mutual inductor according to the cumulative sum between the signal and the dynamic reference value in the CUSUM algorithm at the current moment and the size of the dynamic decision threshold so as to realize digital sampling of the mutual inductor. According to the invention, through adaptive adjustment of the reference mean value, the relaxation parameter and the decision threshold value, the rate of missing report and false report is reduced, and the identification capability of initial faults such as mutual inductor turn-to-turn short circuit and the like is improved.

The invention relates to a proportion optimization method for a cement stabilized macadam base of recycled aggregate. Taking an interface failure sensitivity coefficient IFSI as a quantitative index; the method comprises the following steps: performing microstructure reconstruction on recycled aggregate particles through SEM scanning, resistivity testing and CT scanning, extracting two indexes of defect density and surface adsorption strength, and constructing a degradation prediction model ITZ; the ITZ result is used as a basic parameter for adjusting the water cement ratio and the fine aggregate content, so that a mechanism from aggregate defects to interface failure is connected; establishing unit volume interface energy consumption minimization as an optimization target based on an IFSI value output by the interface failure model; the influence of the water cement ratio, the grading difference and the fine particle wrapping degree on the interface bonding energy consumption is analyzed, and a matching point under the specific recycled aggregate condition is found out through a multivariable function coupling model.

The invention discloses an intelligent decision-making system for dynamic stress sensitive area identification and efficiency compensation of a hydroelectric generating set, and belongs to the field of intelligent monitoring and optimal control of hydroelectric equipment. Running state data, environment data and task instruction data of key parts of the hydroelectric generating set are acquired in real time through a data acquisition module; the sensitive area identification and analysis module identifies a dynamic stress sensitive area by utilizing time-frequency domain analysis and sensitive coefficient calculation; the risk division module judges operation risk intervals under different working condition parameter combinations based on clustering analysis and outlier detection; the execution decision module screens an optimal working condition parameter group in combination with the working condition risk value and the power generation efficiency; the efficiency compensation module generates a compensation priority sequence according to the sensitivity coefficient, and monitors and triggers a compensation mechanism in real time; according to the invention, the intelligent operation management of the hydroelectric generating set is realized, the safety, the power generation efficiency and the economic benefit of the equipment are obviously improved, and a new thought and method are provided for the technical progress and sustainable development of the hydroelectric industry.

The present invention relates to the technical field of electric fault monitoring, and specifically to an intelligent mobile substation electric fault monitoring method. The method synchronously collects electrical monitoring signals, multi-dimensional environmental noise signals and dynamic operating parameters of the mobile substation through multi-source sensors; adopts multi-scale spectral modal decomposition to construct an environmental noise basis vector group, filters out environmental noise interference components from the electrical monitoring signal through orthogonal projection filtering, and outputs a baseline correction signal; constructs an operating condition disturbance response field matrix based on time domain and frequency domain correlation analysis of dynamic operating condition parameters, calculates gradient sensitivity coefficients, separates components strongly correlated with operating condition disturbances and residual components weakly correlated with equipment faults; reconstructs the residual components into pure fault feature vectors; and finally performs fault type diagnosis and risk warning based on the pure fault feature vectors. The method effectively solves the problem of fault feature annihilation caused by noise pollution in complex environments, and the problem of false alarms and missed alarms caused by confusion between operating condition disturbances and real fault signals.

The invention discloses a machine learning-based kernel data verification method, which comprises the following steps of: firstly, obtaining simulated effective proliferation factors and actually measured effective proliferation factors of a plurality of kernel experiment references, and calculating to obtain a difference value; 2, calculating nuclear data sensitivity characteristics of each experimental reference, wherein the nuclear data sensitivity characteristics comprise a sensitivity coefficient for a nuclear reaction cross section and experimental metadata; thirdly, constructing a machine learning model based on a random forest, and performing training by taking the kernel data sensitivity characteristics as input and the effective multiplication factor difference value as output; and finally, carrying out model feature importance analysis, and identifying kernel data sensitivity features which have significant influence on effective proliferation factor differences. The nuclear data verification method based on machine learning is established for the first time, and is used for identifying potential defects in nuclear reaction data or deviation of experimental references.

The invention relates to the technical field of artificial intelligence, in particular to a pavement quality detection method and device based on artificial intelligence, and the method comprises the following steps: obtaining pavement images and load data, extracting crack boundary coordinates through edge detection, calculating a displacement difference to obtain crack distribution, calculating an extension rate in combination with time, and carrying out the weighted analysis of a direction difference. The method comprises the following steps: extracting crack boundary displacement difference, identifying defect evolution parameters, calculating stability and environmental sensitivity coefficients through cross coupling, judging degradation types in combination with load fluctuation, calculating defect grades through classification comparison threshold values of a support vector machine, and outputting a pavement overall quality result. Angle difference variation recognizes evolution direction difference, coupling extension rate and direction consistency depicts stability, environmental sensitivity and load fluctuation linkage calculation is introduced to present external action association, and grading is completed based on parameter and threshold difference to realize dynamic grading recognition of defect development rules.

The invention is suitable for the technical field of model prediction, and provides a method and system for constructing a pulmonary infection risk prediction model based on machine learning, and the method comprises the following steps: collecting data information, and carrying out the standardization processing of the data information, the data information comprising structured data, image data, time series data and environment data; performing multi-modal feature extraction based on the data information to obtain an image feature F1, a physiological time sequence feature F2 and an environment feature F3; performing multi-modal feature fusion, determining weights of the image features, the physiological time sequence features and the environment features, and adaptively adjusting the weights of the environment features based on seasonal factors; processing class imbalance through focus loss, determining a time decay weight according to infection latency characteristics, and determining a total loss function; and carrying out model training by adopting a multi-layer perceptron to obtain a pulmonary infection risk prediction model. According to the invention, by introducing the seasonal factor and the seasonal sensitivity coefficient, adaptive adjustment of the environmental feature weight is realized.

The invention discloses an intelligent standard substance storage management cabinet, and belongs to the technical field of intelligent storage. The intelligent standard substance storage management cabinet comprises a shell, an access is formed in the middle of the front end of the shell, an operation screen is arranged on one side of the front end of the shell, and a first area, a second area and a third area are transversely distributed in the shell. The problems that an existing three-dimensional lifting intelligent storage cabinet cannot accurately control the temperature, and the efficiency is low when a certain object is frequently taken out are solved, the second baffle transversely moves along with the fixing frame and is attached to the fixing frame, the inner structure of the fixing frame can be completely sealed and limited by attaching the fixing frame, and the storage cabinet is convenient to use. Calculation is carried out through a built-in formula of a position distribution calculation module in the intelligent storage system, the taking-out use frequency of the stored articles can be solved through specific data matching frequency weight coefficients, time attenuation coefficients and interval sensitivity coefficients, and calculation of article placement is provided.

The invention relates to the technical field of digital twinning, in particular to a smart factory data management system based on digital twinning. The device comprises an abnormal lagging unit, a window changing unit and a window updating unit. According to the method, a window module is defined to obtain a change speed sensitive coefficient through historical data, and then a data change speed value and a digital twin model update frequency are combined with the change speed sensitive coefficient to calculate an update frequency weight coefficient; the size of the sliding window of the digital twin model is defined by using the data change speed value, the update frequency of the digital twin model, the change speed sensitivity coefficient and the update frequency weight coefficient, and the size of the sliding window is adjusted, so that newest data can be quickly included, model parameters can be updated in time, update lag caused by interference of old data is avoided, and the update efficiency is improved. According to the method, the data utilization efficiency is improved, and when the data change speed is high or low, fine changes can be captured in time while the sliding window is properly enlarged, so that model parameters are updated more accurately.

The invention relates to the field of data analysis, in particular to an information collecting and processing method for heat stroke pathology data integration based on deep learning, which comprises the following steps: determining the data state of user analysis data according to the data richness of the user analysis data and the image conversion completion degree; when the keyword selection mode is associated phrase selection generation, determining the number of the selected keywords according to the phrase influence coefficient, and selecting the keywords according to the sensitivity coefficient from large to small; when the keyword selection mode is image conversion supplementary generation, determining that a keyword generation mode is generated according to a median point or an outlier according to the feature representation degree of the divided paragraph; under the condition that the initial search is completed, optimization analysis is conducted, the search optimization mode is determined to be combined search or sequential search according to the repeated keyword analysis state, and the pathological data integration search efficiency is improved.