Equipment self-adaptive control method and device, equipment and medium

Abstract

The invention relates to the technical field of intelligent decision making, can be applied to business system platforms of financial science and technology, medical health and the like, and discloses an equipment adaptive control method and device, equipment and a medium, and the method comprises the following steps: obtaining a multi-modal signal of an equipment detection target, and constructing a dynamic reference system according to the multi-modal signal; performing space-time alignment optimization on the multi-modal signal according to the dynamic reference system to obtain a space-time alignment signal, and performing dynamic calibration on the space-time alignment signal to obtain a calibration signal; performing transient waveform distortion feature extraction on the calibration signal to obtain a signal distortion feature; performing feature space mapping on the signal distortion features to obtain a feature mapping result; and generating an equipment early warning grade parameter according to the feature mapping result and a preset safety boundary distance, and dynamically adjusting a preset equipment control parameter according to the equipment early warning grade parameter. By analyzing the multi-mode signal, the dynamic adjustment of the equipment control parameters is realized, and the equipment detection precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent decision-making technology, and in particular to a device adaptive control method, device, equipment and medium. Background Art

[0002] Equipment refers to devices, instruments, mechanical or electronic systems designed and manufactured to achieve specific functions. For example, equipment in the medical and health field includes traditional medical imaging equipment, home health monitoring equipment, or small devices such as portable ultrasound detectors. These devices can conduct systematic examinations on healthy or sub-healthy people and detect potential disease risks at an early stage; equipment in the financial technology field includes ATMs, POS machines, smart teller machines, etc., which can avoid transaction interruptions or security vulnerabilities by detecting hardware abnormalities in real time.

[0003] At present, the traditional device control method is to directly use the preset parameters of the device itself to detect the target to be detected and obtain the corresponding detection information. However, the targets have diverse shapes, and using unified detection equipment parameters for detection may lead to insufficient detection accuracy. Summary of the Invention

[0004] The present invention provides a device adaptive control method, apparatus, equipment and medium, which realize dynamic adjustment of device control parameters and improve device detection accuracy by analyzing multimodal signals.

[0005] In a first aspect, a device adaptive control method is provided, comprising:

[0006] Acquire multimodal signals of a target detected by the device, and construct a dynamic reference system based on the multimodal signals;

[0007] performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal alignment signal, and dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal;

[0008] Extracting transient waveform distortion features of the calibration signal to obtain signal distortion features;

[0009] Performing feature space mapping on the signal distortion feature to obtain a feature mapping result;

[0010] An equipment warning level parameter is generated according to the feature mapping result and a preset safety boundary distance, and a preset equipment control parameter is dynamically adjusted according to the equipment warning level parameter to obtain an adjusted equipment control parameter.

[0011] In a second aspect, a device adaptive control apparatus is provided, comprising:

[0012] An acquisition and construction module is used to acquire multimodal signals of a target detected by the device and to construct a dynamic reference system based on the multimodal signals;

[0013] an optimization module, configured to perform spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal aligned signal;

[0014] a calibration module, configured to dynamically calibrate the spatiotemporal alignment signal to obtain a calibration signal;

[0015] An extraction module, configured to extract transient waveform distortion features from the calibration signal to obtain signal distortion features;

[0016] A mapping module, configured to perform feature space mapping on the signal distortion feature to obtain a feature mapping result;

[0017] A generation and adjustment module is used to generate a device warning level parameter according to the feature mapping result and a preset safety boundary distance, and dynamically adjust the preset device control parameter according to the device warning level parameter to obtain the adjusted device control parameter.

[0018] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned device adaptive control method when executing the computer program.

[0019] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned device adaptive control method are implemented.

[0020] In the solution implemented by the above-mentioned device adaptive control method, device, computer equipment and storage medium, a dynamic reference system is constructed based on multimodal signals, and a coordinate system is constructed based on real-time signals to adapt to the variable working conditions of the equipment, solve the error problem of the traditional static reference system in dynamic scenarios, and establish a unified cross-modal measurement standard; spatiotemporal alignment achieves nanosecond synchronization through timestamp calibration and spatial coordinate transformation, and at the same time, dynamic calibration corrects deviations caused by temperature drift, sensor aging, and mechanical deformation by comparing the reference system with the measured signal in real time; transient waveform distortion feature extraction captures early fault micro-features, and distinguishes normal fluctuations from abnormal mutations, maps high-dimensional original features to low-dimensional feature space, retains key fault features, and reduces computational complexity; generates warning levels based on the distance between the feature mapping results and the safety boundary, avoids the "black or white" alarm mode, and dynamically optimizes the equipment control parameters according to the health status of the equipment, thereby improving detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0022] Figure 1 This is a schematic diagram of an application environment of a device adaptive control method according to an embodiment of the present invention;

[0023] Figure 2 This is a flow chart of a device adaptive control method according to an embodiment of the present invention;

[0024] Figure 3 It is a structural diagram of a device adaptive control apparatus according to an embodiment of the present invention;

[0025] Figure 4 is a structural diagram of a computer device in one embodiment of the present invention;

[0026] Figure 5 It is another structural schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0028] The embodiment of the present invention provides a device adaptive control method, which can be applied in the following situations: Figure 1In an application environment, a client communicates with a server via a network. The server can obtain multimodal signals of a device detecting a target and construct a dynamic reference system based on the multimodal signals; perform spatiotemporal alignment optimization on the multimodal signals based on the dynamic reference system to obtain a spatiotemporal aligned signal; dynamically calibrate the spatiotemporal aligned signal to obtain a calibration signal; extract transient waveform distortion features from the calibration signal to obtain signal distortion features; perform feature space mapping on the signal distortion features to obtain a feature mapping result; generate a device warning level parameter based on the feature mapping result and a preset safety boundary distance; dynamically adjust preset device control parameters based on the device warning level parameter to obtain adjusted device control parameters, and feed the adjusted device control parameters back to the client. The present invention provides a device adaptive control device that analyzes multimodal signals for adjusted device control parameter services, thereby dynamically adjusting device control parameters and improving device detection accuracy. The client can include, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. The server can be implemented as a standalone server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0029] See also Figure 2 As shown, Figure 2 A flow chart of a device adaptive control method provided by an embodiment of the present invention includes the following steps:

[0030] S1. Acquire multimodal signals of a detection target from a device, and construct a dynamic reference system based on the multimodal signals.

[0031] In an embodiment of the present invention, in an embodiment of the present invention, the acquisition refers to collecting signal data of the target object in different dimensions and of different types through a variety of detection equipment or technologies, and the construction refers to establishing a benchmark model or reference framework that can be dynamically adjusted with time, environment, and target status based on the historical data and real-time changes of multimodal signals.

[0032] Specifically, by collecting multimodal signals through multi-dimensional data, a comprehensive understanding of the target is constructed, and then based on the dynamic characteristics of the data, such as temporal changes and cross-modal correlations, an adaptively updateable "standard template" (dynamic reference system) is established for comparison and judgment of whether the current state is abnormal.

[0033] In specific medical and health scenarios, for chronic diseases such as hypertension and diabetes, wearable devices are used to continuously obtain multimodal signals such as blood pressure, blood sugar, and exercise data to build a dynamic reference system for personal health (such as the normal fluctuation range of blood sugar 2 hours after a meal). When the data deviates from the reference system, intervention reminders are triggered.

[0034] In the fintech scenario, the dynamic reference system can integrate multimodal signals such as macroeconomic indicators, market transaction data, and public opinion sentiment indexes to learn the patterns of historical market cycles, which can be used to predict the degree of deviation from the current market status and assist in investment decisions.

[0035] In an embodiment of the present invention, constructing a dynamic reference system based on the multimodal signal includes:

[0036] Extracting timestamps from the multimodal signal and eliminating time deviations from the extracted timestamps to obtain a time synchronization signal;

[0037] Performing signal space position mapping on the time synchronization signal to obtain a signal space position result;

[0038] Extracting the signal period of the time synchronization signal according to the signal spatial position result, and calculating the time correlation of the time synchronization signal;

[0039] Constructing a spatiotemporal weight matrix based on the signal spatial position results, signal period, and time correlation;

[0040] The time correlation is fused with the spatiotemporal weight matrix, and a dynamic reference system is constructed according to the fusion result.

[0041] In an embodiment of the present invention, the timestamp extraction refers to extracting information used to mark the data acquisition time from a multimodal signal, the time deviation elimination refers to correcting the time error of the multimodal signal caused by the clock asynchrony of the acquisition device, transmission delay, etc., to achieve time alignment, the signal space position mapping refers to associating the multimodal signal with the position in the physical space or logical space, and giving the signal space attributes, the extraction refers to identifying the periodic law of data change from the time synchronization signal, the calculation refers to measuring the degree of temporal correlation between the time synchronization signal at different time points or different modal signals, the construction refers to assigning weights to different signals in the space-time dimension based on the spatial position and periodic characteristics of the signal, reflecting the difference in their importance to the dynamic reference system, and the fusion refers to integrating the time correlation characteristics of the signal with the space-time weight to form a dynamically adjustable benchmark model.

[0042] Specifically, for different modal signals, their inherent time stamp fields are identified, and the system clocks of each acquisition device are calibrated using a unified time base to reduce initial deviations. The transmission path of the signal from acquisition to storage is analyzed, and the delay value is calculated through historical testing or real-time monitoring to perform offset correction on the timestamp. The data is resampled to a unified time axis through a time interpolation algorithm to ensure that multimodal signals correspond one-to-one in the time dimension.

[0043] In detail, signal space position mapping includes physical space mapping and logical space mapping. For example, in the field of medical health, physical space mapping determines the spatial coordinates of medical images in the human anatomical structure through equipment parameters (such as the position of the scanning bed and the angle of the probe). The specific operation of logical space mapping is to classify different modal signals into abstract space dimensions according to business logic (such as the "heart function" dimension in medicine to integrate electrocardiogram and cardiac ultrasound data, or the "risk factor" dimension in finance to integrate market data and user behavior data).

[0044] Furthermore, the extraction of signal period is divided into time domain analysis, frequency domain conversion, and period parameter calibration. Time domain analysis detects whether the signal has a repetitive fluctuation pattern through waveform observation or statistical methods (such as autocorrelation function). Frequency domain conversion converts the signal from time domain to frequency domain, such as Fourier transform, and determines the main periodic components through the spectrum peak. Period parameter calibration specifically calculates the characteristic parameters such as the time length of the period (such as 24 hours), amplitude (fluctuation range), and phase (starting point of the period).

[0045] Furthermore, calculating time correlation includes autocorrelation calculation, cross-correlation calculation, and correlation matrix construction. Autocorrelation calculation is to analyze the correlation between the same modal signal at different time points. Cross-correlation calculation is to calculate the time lag relationship between different modal signals. The correlation coefficients under different time offsets are calculated through a sliding window to determine the optimal time lag parameter. Finally, the time correlation degree of the multimodal signal is quantified into a matrix.

[0046] Furthermore, when constructing a spatiotemporal weight matrix based on the signal spatial position results, signal period, and time correlation, the steps of spatial weight allocation, period weight allocation, and matrix fusion are required. Spatial weight allocation is based on the importance of the spatial position of the signal. For example, the imaging data weight of core organs in medical treatment is higher than that of marginal tissues. Period weight allocation is to allocate period weights to signals with significant periodicity based on their period stability (the more stable the fluctuation pattern, the higher the weight). The spatial weights are combined with period characteristic parameters (such as period amplitude and stability) to construct a two-dimensional matrix.

[0047] Furthermore, the time correlation matrix is ​​weighted by the spatiotemporal weight matrix to highlight the time correlation characteristics of signals at important spatiotemporal positions; when a new time synchronization signal is introduced, the weight matrix and correlation parameters are updated through the incremental learning algorithm, and the final dynamic reference system is expressed as a joint model of "spatiotemporal weight-time correlation".

[0048] In an embodiment of the present invention, timestamp extraction and deviation elimination are used to solve the fundamental problem of multimodal signal asynchrony, and a unified time axis is established for subsequent analysis. The spatiotemporal weight matrix quantifies the importance of the signal, and a dynamic benchmark is constructed in combination with time correlation, so that the reference system can reflect historical laws and adapt to new data.

[0049] In the embodiment of the present invention, dynamic data learning is used to adapt to changes in the target, so that abnormality judgment is more in line with the actual scenario and misjudgment and missed judgment are reduced.

[0050] S2. Performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal alignment signal, and dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal.

[0051] In an embodiment of the present invention, the spatiotemporal alignment optimization refers to the joint alignment of multimodal signals in the time and space dimensions based on a dynamic reference system, eliminating the spatiotemporal deviations between modalities, and making the signals comparable in a unified spatiotemporal coordinate system. The dynamic calibration refers to the deviation correction of the spatiotemporally aligned signals according to the historical laws and real-time update mechanism of the dynamic reference system, so that they conform to the expected characteristics of the benchmark model while retaining the effective abnormal information in the signal.

[0052] Specifically, using the dynamic reference system as the standard template, the difference between the real-time signal and the template is compared, and the deviation is calibrated through weighted adjustment, noise filtering, etc. The reference system will be continuously updated as new data is input, and the calibration rules will also be adjusted accordingly. For example, in medicine, the physiological benchmark of the human body changes with age, and in finance, the normal market fluctuation range changes with policy adjustments.

[0053] In medical and health scenarios, CT images, pathology reports, and real-time physiological monitoring data are unified into the patient's spatiotemporal model to calibrate measurement deviations of different devices and avoid misdiagnosis due to equipment differences.

[0054] In financial risk control, stock trading data, macroeconomic news, and social media public opinion are unified into the "market time and space grid", and the timestamps and price units of different trading markets are calibrated to avoid risk misjudgment due to time and space deviations.

[0055] In an embodiment of the present invention, performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal aligned signal includes:

[0056] Performing spatiotemporal reference parameter analysis on the dynamic reference system to obtain spatiotemporal alignment reference parameters;

[0057] Performing hardware clock calibration on the multimodal signal according to the reference parameters, and performing time delay compensation on the multimodal signal according to the calibration result to obtain a compensated signal;

[0058] Performing regional fusion on the compensation signal using the spatiotemporal weight matrix to obtain a fused signal;

[0059] The fusion signal is subjected to spatiotemporal consistency verification and optimization according to the dynamic reference system to obtain a spatiotemporal alignment signal.

[0060] In an embodiment of the present invention, the spatiotemporal reference parameter analysis refers to extracting key reference parameters for spatiotemporal alignment from a dynamic reference system, including a time reference and a space reference. The hardware clock calibration refers to correcting the deviation of the physical clock of the hardware device that collects multimodal signals based on the spatiotemporal reference parameters to ensure the accuracy of the timestamp. The delay compensation refers to calculating and compensating for the delay generated during the transmission and processing of the signal based on the spatiotemporal reference parameters. The regional fusion refers to weighted aggregation of the compensated signal in the spatial region based on the spatiotemporal weight matrix. The spatiotemporal consistency verification and optimization refers to the process of verifying the consistency of the fused signal in the time and space dimensions through the benchmark rules of the dynamic reference system and optimizing the deviation.

[0061] Specifically, time synchronization parameters are collected from the historical data of the dynamic reference system, and the spatial benchmark rules are analyzed to form a quantifiable benchmark parameter set; each hardware clock is compared with the time benchmark of the dynamic reference system, the clock offset is calculated, and the clock frequency is adjusted through the hardware driver or software interface, or the offset compensation is directly performed on the collected timestamp; delay compensation is achieved by subtracting the corresponding delay amount from the collected signal timestamp through signal loop testing or historical data statistics to ensure time synchronization with other modal signals.

[0062] Furthermore, regional fusion is divided into spatial region division and weighted fusion. According to the definition of the spatiotemporal weight matrix, the compensation signal is mapped to the corresponding region. The multimodal signals in the same region are linearly combined according to the weight values ​​assigned by the weight matrix to generate a regional fusion signal. Finally, the spatiotemporal consistency verification and optimization include time consistency verification, spatial consistency verification, and optimization iteration. The specific operations are to check whether the time series of the fusion signal conforms to the time law of the dynamic reference system, smooth or correct the abnormal time series; verify whether the spatial characteristics of the fusion signal conform to the benchmark model, redistribute the weights or mark the spatial conflicts as abnormalities, adjust the spatiotemporal weight matrix or benchmark parameters according to the verification results, and reversely optimize the regional fusion rules.

[0063] In an embodiment of the present invention, dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal includes:

[0064] Performing a limiting process on the spatiotemporal alignment signal according to a preset condition to obtain a limited signal;

[0065] performing phase deviation compensation on the amplitude limiting signal to obtain a deviation compensation signal;

[0066] Interference filtering is performed on the deviation compensation signal to obtain a calibration signal.

[0067] In the embodiments of the present invention, the clipping process refers to truncating or attenuating signals beyond a certain range by setting the upper and lower thresholds of the signal amplitude. The phase deviation compensation refers to correcting the phase inconsistency of multi-modal signals caused by transmission delay, sampling error, etc. The interference filtering refers to removing the noise or non-target interference components doped in the signal through a filtering algorithm.

[0068] Specifically, set the threshold, preset the upper limit value (Max) and the lower limit value (Min) according to the normal fluctuation range of the signal, detect the signal amplitude point by point. If the amplitude > A_max, set it to A_max; if the amplitude < A_min, set it to A_min, and keep the intermediate values unchanged. Analyze the phase offset between different signals through the cross-correlation function or Fourier transform, for example, calculate the phase difference between the electrocardiogram signal and the respiration signal. Convert the phase difference into a time delay. Where f is the signal frequency, perform interpolation delay (such as linear interpolation) on the lagging signal, or truncate the leading signal, and add a phase compensation factor e ^ (-j2πfΔt) and then perform inverse transformation. Select a suitable filter to remove high-frequency noise, retain a specific frequency band, and optimize the signal smoothness through methods such as wavelet transform after filtering to obtain a calibrated signal.

[0069] In the embodiments of the present invention, the spatio-temporal reference parameter parsing is a rule extractor for a dynamic reference system, providing a standard for subsequent calibration. The hardware clock calibration and time delay compensation solve the time heterogeneity of multi-modal signals, ensuring the comparability of time sequences. The regional fusion integrates spatial features through a weight matrix to achieve cross-modal spatial unity.

[0070] In the embodiments of the present invention, based on the dynamic reference system, while filtering noise, real anomalies are retained. The spatio-temporal alignment optimization and dynamic calibration jointly ensure the analysis effectiveness of multi-modal data in a dynamic scenario.

[0071] S3. Extract the transient waveform distortion features of the calibrated signal to obtain signal distortion features. [[ID=]19]

[0072] In the embodiments of the present invention, the transient waveform distortion feature extraction refers to capturing the abnormal waveform changes occurring within a short time from the calibrated signal and extracting the quantitative features that can characterize these anomalies.

[0073] Specifically, through time-domain, frequency-domain or time-frequency domain analysis, locate the instantaneous mutations deviating from the normal mode in the waveform, such as spikes, sudden drops, oscillations, etc., and convert them into quantifiable characteristic parameters, such as distortion amplitude, duration, spectral energy mutation rate, etc.

[0074] In the specific scenario of medical health, in arrhythmia monitoring, transient distortions of the ECG waveform, such as abnormal wave group morphology of ventricular premature beats, are extracted to identify abnormal cardiac electrical activity in advance, providing a basis for sudden death warning.

[0075] In financial scenarios, the system can extract transient distortions of minute-level stock K-lines, such as a 5% price drop within one minute and a sudden increase in trading volume, to identify risk events such as flash crashes and market manipulation, triggering the trading system to automatically close positions.

[0076] In an embodiment of the present invention, extracting transient waveform distortion features from the calibration signal to obtain signal distortion features includes:

[0077] Dividing the calibration signal into a plurality of segmented signals according to a preset time window, and performing abnormal point detection on the segmented signals according to a preset distortion characteristic condition to obtain abnormal segmented signals;

[0078] Performing a distortion frequency domain feature analysis on the abnormal segmented signal to obtain a distortion frequency domain feature;

[0079] Performing time-frequency domain joint feature fusion on the abnormal segmented signal and the distorted frequency domain feature to obtain a time-frequency joint feature vector;

[0080] Distortion features are screened for the time-frequency joint feature vector to obtain signal distortion features.

[0081] In an embodiment of the present invention, the outlier detection refers to identifying instantaneous outliers that deviate from the normal pattern in the segmented signal; the distortion frequency domain feature analysis refers to performing frequency domain transformation on the abnormal segmented signal, extracting frequency features related to the distortion, and revealing the abnormal energy distribution pattern; the time-frequency domain joint feature fusion refers to cross-domain correlation between the time domain abnormal segmented signal and the frequency domain features; and the distortion feature screening refers to screening out the most discriminative features for the distortion from the time-frequency joint features, and eliminating redundant or noise features.

[0082] Specifically, according to a preset time window, such as 500ms for medical signals and 100ms for financial signals, the calibration signal is divided into overlapping or non-overlapping segments, each segment is demeaned and normalized to eliminate baseline drift, and the mean μ and standard deviation σ of the segmented signal are calculated. Points exceeding μ±3σ are judged as abnormal. If the proportion of abnormal points in the segmented signal exceeds a preset threshold, where the preset threshold is μ±3σ, the segment is marked as an abnormal segmented signal.

[0083] Furthermore, the distortion frequency domain feature analysis is divided into three steps: frequency domain transformation, feature parameter extraction, and frequency domain feature standardization. First, the time domain anomaly segments are converted into frequency domain spectra, which is suitable for stationary signals. Multi-resolution analysis is performed on non-stationary distortion to obtain time-frequency localization features. When extracting feature parameters, energy distribution features are performed, including the energy proportion of abnormal frequency bands. For example, when the ECG is abnormal, the energy of the 1-2Hz frequency band corresponding to ST segment elevation is increased by 20%. Finally, the frequency domain features, such as the energy of each frequency band, are scaled to the [0,1] interval.

[0084] Furthermore, the time domain segmented signal and frequency domain features are timestamp aligned to ensure that the features correspond to the same physical moment. The time domain waveform sampling points and frequency domain features are directly spliced ​​into an N+M dimensional vector, and weights are assigned according to the importance of the features. For example, in medical treatment, the weight of time domain morphological features is 0.6, and the weight of frequency domain energy features is 0.4. The time-frequency matrix is ​​generated through short-time Fourier transform and directly used as the time-frequency joint feature vector.

[0085] Furthermore, the specific operation of distortion feature screening is to calculate the Pearson correlation coefficient r between each feature in the time-frequency joint feature vector and the distortion label, retain the features with |r|>0.5, and use genetic algorithm or particle swarm optimization to search for the optimal feature combination. The goal is to maximize the classification accuracy. The screened features are combined into a signal distortion feature vector, such as [time domain mutation slope, abnormal frequency band energy ratio, time-frequency entropy], for subsequent classification or prediction.

[0086] In an embodiment of the present invention, the abnormal segmented signal and the distorted frequency domain features are subjected to a joint time-frequency domain feature fusion. While fully capturing the multi-dimensional features of the signal, the feature representation capability in complex scenarios can also be improved. The joint time-frequency feature can compress the multi-dimensional information into the feature vector to form a joint representation of "time, space-frequency".

[0087] In the embodiment of the present invention, by quantifying transient distortion characteristics, abnormal phenomena are converted into calculable risk indicators, providing data support for real-time warning and decision-making.

[0088] S4. Perform feature space mapping on the signal distortion feature to obtain a feature mapping result.

[0089] In the embodiment of the present invention, the feature space mapping refers to the process of converting the original signal distortion features from the feature space where they are located (such as the space composed of time-frequency joint feature vectors) to another feature space through mathematical transformation or mapping function.

[0090] Specifically, through dimensionality conversion, dimensionality reduction, dimensionality increase or spatial transformation, the features are made more suitable for subsequent analysis, classification or modeling tasks.

[0091] In the embodiment of the present invention, performing feature space mapping on the signal distortion feature to obtain a feature mapping result includes:

[0092] Performing dimension alignment on the signal distortion features to obtain aligned signal features;

[0093] Calculating the spatial coordinates of the alignment signal features one by one to obtain feature coordinate points;

[0094] Performing topological marking on the characteristic coordinate points using preset historical distortion spatial distribution data to obtain topological characteristic coordinate points;

[0095] The mapping result of the topological feature coordinate points is verified to obtain a feature mapping result.

[0096] In an embodiment of the present invention, the dimensional alignment refers to adjusting the signal distortion characteristics of different dimensions, different scales or different feature spaces to a unified dimensional space, the spatial coordinate calculation refers to mapping the aligned feature vectors into coordinate points in a high-dimensional space, the topological labeling refers to labeling the topological relationship of the current feature coordinate points based on the spatial distribution law of historical distortion data, and the mapping result verification refers to verifying the accuracy of the feature space mapping by comparing the topological labeling results with the actual signal characteristics.

[0097] Specifically, if the feature vector dimensions are different (e.g., a segmented signal contains 10 time domain features, and another segment contains 8 frequency domain features), the dimensions can be unified by zero padding, interpolation, or feature fusion. Assume that the aligned feature vector is F = [f1, f2, ..., f n ], then its coordinates in n-dimensional space are (f1,f2,...,f n ), if the n-dimensionality is too high (e.g., n>3), reduce the dimension to 2 / 3 dimensions through methods such as PCA and t-SNE to generate visual coordinates; then construct a spatial distribution database of historical distortion features, divide the topological areas of different distortion types through clustering (such as K-means), calculate the distance between the current feature coordinate point and the historical topological area, mark its topological category or outlier, and finally use cross-validation to train the topological model with part of the historical data and verify the classification accuracy of the current mapping result with the other part. If the verification finds a mapping deviation, such as mistakenly marking a normal signal as an abnormal topology, adjust the mapping function parameters and re-map the feature space.

[0098] In the embodiment of the present invention, dimensional alignment solves the problem of feature heterogeneity and lays a unified foundation for subsequent analysis. Spatial coordinate calculation converts abstract features into points in geometric space to facilitate topological analysis. Mapping result verification ensures the accuracy of the entire feature processing flow and avoids misjudgment.

[0099] In the embodiment of the present invention, the spatial structure of the features is reshaped through mathematical transformation to make it more suitable for subsequent analysis tasks.

[0100] S5. Generate a device warning level parameter according to the feature mapping result and a preset safety boundary distance, and dynamically adjust the preset device control parameter according to the device warning level parameter to obtain an adjusted device control parameter.

[0101] In an embodiment of the present invention, the generation refers to quantifying the risk level of the equipment operation status based on the distance between the feature mapping result and the preset safety boundary, and generating parameters that can directly guide the early warning. The dynamic adjustment refers to modifying the control strategy of the equipment in real time according to the early warning level parameters, and adjusting the equipment operation status by adjusting the control parameters.

[0102] Specifically, by comparing signal characteristics with safety thresholds, early warning of equipment operating status and adaptive adjustment of control parameters are achieved.

[0103] In an embodiment of the present invention, generating a device warning level parameter based on the feature mapping result and a preset safety boundary distance includes:

[0104] Calculate the shortest geometric distance between the feature space coordinate point in each feature mapping result and the preset safety boundary;

[0105] Defining warning level conditions, and determining initial equipment warning level parameters based on the warning level conditions and the shortest geometric distance;

[0106] The initial equipment warning level parameter is modified according to preset historical warning data to obtain the equipment warning level parameter.

[0107] In an embodiment of the present invention, the calculation refers to calculating the minimum distance from a feature space coordinate point to a preset safety boundary through a geometric measurement method; the definition refers to establishing a distance-risk level mapping rule to convert the geometric distance into an understandable warning level; the determination refers to preliminarily determining the current risk level of the equipment based on the shortest distance and warning level conditions to generate uncorrected warning parameters; the correction refers to optimizing the initial parameters using historical warning data to compensate for the limitations of a single distance calculation.

[0108] Specifically, the shortest geometric distance between the feature space coordinate point in each feature mapping result and the preset safety boundary is calculated by the distance measurement method. Specifically, an appropriate distance formula is selected according to the shape of the safety boundary, such as the point-to-hyperplane distance. The distance from the feature point P(x1, y1, z1) to the plane ax+by+cz+d=0 is as follows:

[0109]

[0110] Furthermore, for example, the distance from a point to the spherical boundary, the distance from the feature point P to the center of the sphere is L, then the shortest distance to the boundary is |Lr| (when L>r, it is an external point, and when L<r, it is an internal point).

[0111] Furthermore, according to the business scenario requirements, the distance range is divided into multiple intervals, each interval corresponds to a warning level,

[0112] Safe range: D ≥ 3 → Level 1 (normal);

[0113] Warning range: 1≤D<3→Level 2 (Warning);

[0114] Emergency zone: D<1→Level 3 (Emergency).

[0115] Furthermore, the calculated shortest distance D is substituted into the warning level condition and directly mapped to the initial level. For example, if D = 0.8, corresponding to the warning level condition "D < 1 → emergency", the initial level is level 3. Further collect real abnormal events that have occurred in the past and their corresponding characteristic distances and actual risk consequences, construct a labeled data set, and summarize the correction rules based on historical cases (such as "when the distance ∈ [0.8, 1.0] and 80% of similar historical cases are minor abnormalities, the initial level is -1"), or use models such as logistic regression and random forest, with "historical distance + initial level" as input and "actual risk level" as output, train the correction model, and thus correct the initial equipment warning level parameters.

[0116] In an embodiment of the present invention, the shortest geometric distance between the feature space coordinate point in each feature mapping result and the preset safety boundary is calculated, and the safety of the feature point is quantified using mathematical methods; at the same time, the initial value is determined to complete the preliminary mapping from data to decision.

[0117] In an embodiment of the present invention, by converting physical signals into coordinates in a feature space and then comparing them with safety boundaries, the abstract risk level is converted into a specific control action, thereby realizing an automated process from state perception to intelligent response.

[0118] It can be seen that in the above scheme, by acquiring the multimodal signal of the device detection target and constructing a dynamic reference system based on the multimodal signal; performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal alignment signal, and dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal; extracting transient waveform distortion features from the calibration signal to obtain signal distortion features; performing feature space mapping on the signal distortion features to obtain feature mapping results; generating device warning level parameters based on the feature mapping results and the preset safety boundary distance, and dynamically adjusting the preset device control parameters based on the device warning level parameters to obtain adjusted device control parameters. By analyzing the multimodal signal, dynamic adjustment of the device control parameters is achieved, thereby improving the device detection accuracy.

[0119] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0120] In one embodiment, a device adaptive control apparatus is provided, which corresponds one-to-one to a device adaptive control method in the above embodiment. Figure 3 As shown, the device adaptive control apparatus includes an acquisition and construction module 101, an optimization module 102, a calibration module 103, an extraction module 104, a mapping module 105, and a generation and adjustment module 106. The functional modules are described in detail as follows:

[0121] An acquisition and construction module 101 is used to acquire multimodal signals of a target detected by a device and to construct a dynamic reference system based on the multimodal signals;

[0122] an optimization module 102, configured to perform spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal aligned signal;

[0123] A calibration module 103 is configured to dynamically calibrate the spatiotemporal alignment signal to obtain a calibration signal;

[0124] An extraction module 104 is configured to extract transient waveform distortion features from the calibration signal to obtain signal distortion features;

[0125] A mapping module 105 is configured to perform feature space mapping on the signal distortion feature to obtain a feature mapping result;

[0126] The generation and adjustment module 106 is configured to generate a device warning level parameter according to the feature mapping result and a preset safety boundary distance, and dynamically adjust the preset device control parameter according to the device warning level parameter to obtain the adjusted device control parameter.

[0127] In one embodiment, the acquisition and construction module 101, when constructing a dynamic reference system according to the multimodal signal, is configured to:

[0128] Extracting timestamps from the multimodal signal and eliminating time deviations from the extracted timestamps to obtain a time synchronization signal;

[0129] Performing signal space position mapping on the time synchronization signal to obtain a signal space position result;

[0130] Extracting the signal period of the time synchronization signal according to the signal spatial position result, and calculating the time correlation of the time synchronization signal;

[0131] Constructing a spatiotemporal weight matrix based on the signal spatial position results, signal period, and time correlation;

[0132] The time correlation is fused with the spatiotemporal weight matrix, and a dynamic reference system is constructed according to the fusion result.

[0133] In one embodiment, when performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal aligned signal, the optimization module 102 is configured to:

[0134] Performing spatiotemporal reference parameter analysis on the dynamic reference system to obtain spatiotemporal alignment reference parameters;

[0135] Performing hardware clock calibration on the multimodal signal according to the reference parameters, and performing time delay compensation on the multimodal signal according to the calibration result to obtain a compensated signal;

[0136] Performing regional fusion on the compensation signal using the spatiotemporal weight matrix to obtain a fused signal;

[0137] The fusion signal is subjected to spatiotemporal consistency verification and optimization according to the dynamic reference system to obtain a spatiotemporal alignment signal.

[0138] In one embodiment, when the calibration module 103 dynamically calibrates the spatiotemporal alignment signal to obtain the calibration signal, it is configured to:

[0139] Performing a limiting process on the spatiotemporal alignment signal according to a preset condition to obtain a limited signal;

[0140] performing phase deviation compensation on the amplitude limiting signal to obtain a deviation compensation signal;

[0141] Interference filtering is performed on the deviation compensation signal to obtain a calibration signal.

[0142] In one embodiment, when extracting transient waveform distortion features from the calibration signal to obtain signal distortion features, the extraction module 104 is configured to:

[0143] Dividing the calibration signal into a plurality of segmented signals according to a preset time window, and performing abnormal point detection on the segmented signals according to a preset distortion characteristic condition to obtain abnormal segmented signals;

[0144] Performing a distortion frequency domain feature analysis on the abnormal segmented signal to obtain a distortion frequency domain feature;

[0145] Performing time-frequency domain joint feature fusion on the abnormal segmented signal and the distorted frequency domain feature to obtain a time-frequency joint feature vector;

[0146] Distortion features are screened for the time-frequency joint feature vector to obtain signal distortion features.

[0147] In one embodiment, when the mapping module 105 performs feature space mapping on the signal distortion feature to obtain a feature mapping result, it is configured to:

[0148] Performing dimension alignment on the signal distortion features to obtain aligned signal features;

[0149] Calculating the spatial coordinates of the alignment signal features one by one to obtain feature coordinate points;

[0150] Performing topological marking on the characteristic coordinate points using preset historical distortion spatial distribution data to obtain topological characteristic coordinate points;

[0151] The mapping result of the topological feature coordinate points is verified to obtain a feature mapping result.

[0152] In one embodiment, when generating the device warning level parameter based on the feature mapping result and the preset safety boundary distance, the generation and adjustment module 106 is configured to:

[0153] Calculate the shortest geometric distance between the feature space coordinate point in each feature mapping result and the preset safety boundary;

[0154] Defining warning level conditions, and determining initial equipment warning level parameters based on the warning level conditions and the shortest geometric distance;

[0155] The initial equipment warning level parameter is modified according to preset historical warning data to obtain the equipment warning level parameter.

[0156] The present invention provides an adaptive control device for equipment. The device acquires multimodal signals of a target to be detected by the equipment and constructs a dynamic reference system based on the multimodal signals. The device then performs spatiotemporal alignment optimization on the multimodal signals according to the dynamic reference system to obtain a spatiotemporal alignment signal. The device then dynamically calibrates the spatiotemporal alignment signal to obtain a calibration signal. The device then extracts transient waveform distortion features from the calibration signal to obtain signal distortion features. The device then performs feature space mapping on the signal distortion features to obtain a feature mapping result. The device generates a warning level parameter based on the feature mapping result and a preset safety boundary distance. The device then dynamically adjusts the preset device control parameters based on the warning level parameters to obtain the adjusted device control parameters. By analyzing the multimodal signals, the device control parameters can be dynamically adjusted, thereby improving the detection accuracy of the device.

[0157] For the specific definition of a device adaptive control device, please refer to the definition of a device adaptive control method above and will not be repeated here. Each module in the above-mentioned device adaptive control device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0158] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the server side of a device adaptive control method.

[0159] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, memory, network interface, display screen, and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of the client side of a device adaptive control method.

[0160] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0161] Acquire multimodal signals of a target detected by the device, and construct a dynamic reference system based on the multimodal signals;

[0162] performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal alignment signal, and dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal;

[0163] Extracting transient waveform distortion features of the calibration signal to obtain signal distortion features;

[0164] Performing feature space mapping on the signal distortion feature to obtain a feature mapping result;

[0165] An equipment warning level parameter is generated according to the feature mapping result and a preset safety boundary distance, and a preset equipment control parameter is dynamically adjusted according to the equipment warning level parameter to obtain an adjusted equipment control parameter.

[0166] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0167] Acquire multimodal signals of a target detected by the device, and construct a dynamic reference system based on the multimodal signals;

[0168] performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal alignment signal, and dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal;

[0169] Extracting transient waveform distortion features of the calibration signal to obtain signal distortion features;

[0170] Performing feature space mapping on the signal distortion feature to obtain a feature mapping result;

[0171] An equipment warning level parameter is generated according to the feature mapping result and a preset safety boundary distance, and a preset equipment control parameter is dynamically adjusted according to the equipment warning level parameter to obtain an adjusted equipment control parameter.

[0172] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0173] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0174] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0175] The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. If software tools or components other than those of the company appear in the application embodiments, they are merely used for illustration and do not represent actual use. Although the present invention has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above-mentioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A device adaptive control method, characterized in that: include: Acquire multimodal signals of a target detected by the device, and construct a dynamic reference system based on the multimodal signals; performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal alignment signal, and dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal; Extracting transient waveform distortion features of the calibration signal to obtain signal distortion features; Performing feature space mapping on the signal distortion feature to obtain a feature mapping result; An equipment warning level parameter is generated according to the feature mapping result and a preset safety boundary distance, and a preset equipment control parameter is dynamically adjusted according to the equipment warning level parameter to obtain an adjusted equipment control parameter.

2. The device adaptive control method according to claim 1, characterized in that: The constructing of a dynamic reference system according to the multimodal signal includes: Extracting timestamps from the multimodal signal and eliminating time deviations from the extracted timestamps to obtain a time synchronization signal; Performing signal space position mapping on the time synchronization signal to obtain a signal space position result; Extracting the signal period of the time synchronization signal according to the signal spatial position result, and calculating the time correlation of the time synchronization signal; Constructing a spatiotemporal weight matrix based on the signal spatial position results, signal period, and time correlation; The time correlation is fused with the spatiotemporal weight matrix, and a dynamic reference system is constructed according to the fusion result.

3. The device adaptive control method according to claim 2, wherein: The performing spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal aligned signal includes: Analyzing the dynamic reference system to obtain reference parameters for time-space alignment; Performing hardware clock calibration on the multimodal signal according to the reference parameters, and performing time delay compensation on the multimodal signal according to the calibration result to obtain a compensated signal; Performing regional fusion on the compensation signal using the spatiotemporal weight matrix to obtain a fused signal; The fusion signal is subjected to spatiotemporal consistency verification and optimization according to the dynamic reference system to obtain a spatiotemporal alignment signal.

4. The device adaptive control method according to claim 1, wherein: The dynamically calibrating the spatiotemporal alignment signal to obtain a calibration signal includes: Performing a limiting process on the spatiotemporal alignment signal according to a preset condition to obtain a limited signal; performing phase deviation compensation on the amplitude limiting signal to obtain a deviation compensation signal; Interference filtering is performed on the deviation compensation signal to obtain a calibration signal.

5. The device adaptive control method according to claim 1, wherein: The extracting transient waveform distortion features of the calibration signal to obtain signal distortion features includes: Dividing the calibration signal into a plurality of segmented signals according to a preset time window, and performing abnormal point detection on the segmented signals according to a preset distortion characteristic condition to obtain abnormal segmented signals; Performing a distortion frequency domain feature analysis on the abnormal segmented signal to obtain a distortion frequency domain feature; Performing time-frequency domain joint feature fusion on the abnormal segmented signal and the distorted frequency domain feature to obtain a time-frequency joint feature vector; Distortion features are screened for the time-frequency joint feature vector to obtain signal distortion features.

6. The device adaptive control method according to claim 1, wherein: The performing feature space mapping on the signal distortion feature to obtain a feature mapping result includes: Performing dimension alignment on the signal distortion features to obtain aligned signal features; Calculating the spatial coordinates of the alignment signal features one by one to obtain feature coordinate points; Performing topological marking on the characteristic coordinate points using preset historical distortion spatial distribution data to obtain topological characteristic coordinate points; The mapping result of the topological feature coordinate points is verified to obtain a feature mapping result.

7. The device adaptive control method according to claim 1, wherein: Generating a device warning level parameter according to the feature mapping result and a preset safety boundary distance includes: Calculate the shortest geometric distance between the feature space coordinate point in each feature mapping result and the preset safety boundary; Defining warning level conditions, and determining initial equipment warning level parameters based on the warning level conditions and the shortest geometric distance; The initial equipment warning level parameter is modified according to preset historical warning data to obtain the equipment warning level parameter.

8. A device adaptive control device, characterized in that: include: An acquisition and construction module is used to acquire multimodal signals of a target detected by the device and to construct a dynamic reference system based on the multimodal signals; an optimization module, configured to perform spatiotemporal alignment optimization on the multimodal signal according to the dynamic reference system to obtain a spatiotemporal aligned signal; a calibration module, configured to dynamically calibrate the spatiotemporal alignment signal to obtain a calibration signal; An extraction module, configured to extract transient waveform distortion features from the calibration signal to obtain signal distortion features; A mapping module, configured to perform feature space mapping on the signal distortion feature to obtain a feature mapping result; A generation and adjustment module is used to generate a device warning level parameter according to the feature mapping result and a preset safety boundary distance, and dynamically adjust the preset device control parameter according to the device warning level parameter to obtain the adjusted device control parameter.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the device adaptive control method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the device adaptive control method according to any one of claims 1 to 7 is implemented.

Images