Communication device running state detection method based on multi-source sensing data fusion

Abstract

This invention relates to the field of data processing and discloses a method for detecting the operational status of communication equipment based on multi-source sensor data fusion. The method includes: acquiring multi-source sensor data collected by multiple sensors deployed on the communication equipment; performing spatiotemporal alignment on the multi-source sensor data to obtain time-synchronized and spatially correlated aligned sensor data; extracting time-domain and frequency-domain features corresponding to each sensor based on the aligned sensor data to obtain a multi-source feature matrix; mapping the multi-source feature matrix to a preset state basis space to obtain a fused state basis vector; and determining the operational status detection result of the communication equipment based on the dynamic offset between the projection vector of the real-time acquired sensor data in the state basis space and the fused state basis vector. This invention improves the sensitivity of detecting early faults and performance degradation in communication equipment.

Description

Technical Field

[0001] This invention belongs to the field of data processing, and in particular relates to a method, device and storage medium for detecting the operating status of communication equipment based on multi-source sensor data fusion. Background Technology

[0002] When communication equipment is in operation, multiple sensors are typically deployed for status monitoring. Because the sampling frequencies and installation locations of various sensors differ, the multi-source sensor data collected are asynchronous and misaligned in time and space, making direct fusion analysis difficult.

[0003] Traditional methods often rely on independent judgment using data from a single sensor or simple data aggregation followed by threshold comparison. These methods are easily affected by environmental noise and fluctuations in operating conditions, resulting in insufficient sensitivity for detecting early performance degradation and minor faults in equipment. A new technical approach is needed to address these issues. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method, device and storage medium for detecting the operating status of communication equipment based on multi-source sensor data fusion, which can solve the problem that the threshold setting in related technologies is not objective and is easily decoupled from the actual performance of the pipeline.

[0005] The first aspect of this invention provides a method for detecting the operating status of communication equipment based on multi-source sensor data fusion, comprising: Acquire multi-source sensing data collected by multiple sensors deployed on communication equipment, including vibration sensing data, temperature sensing data, current sensing data, and radio frequency sensing data; The multi-source sensor data is spatiotemporally aligned to obtain time-synchronized and spatially correlated aligned sensor data. Based on the alignment sensing data, extract the time-domain features and frequency-domain features corresponding to each sensor to obtain a multi-source feature matrix; The multi-source feature matrix is ​​mapped to a preset state basis space to obtain a fused state basis vector; The operating status detection result of the communication device is determined based on the dynamic offset between the projection vector of the real-time collected sensor data in the state basis space and the fused state basis vector.

[0006] Optionally, in a first implementation of the first aspect of the present invention, mapping the multi-source feature matrix to a preset state basis space to obtain a fused state basis vector includes: Obtain the historical multi-source feature matrix of the communication device under normal operating conditions for multiple historical periods; The intrinsic dimension decomposition is performed on the historical multi-source feature matrix to obtain the intrinsic modal features; Generate a constraint basis orthogonal to the intrinsic mode features, wherein the constraint basis is used to suppress the interference of environmental noise on the feature space; The intrinsic modal features are orthogonally projected onto the constraint basis to obtain the fused state basis vector.

[0007] Optionally, in a second implementation of the first aspect of the present invention, the step of performing intrinsic dimension decomposition on the historical multi-source feature matrix to obtain intrinsic modal features includes: The historical multi-source feature matrix is ​​decomposed into multiple intrinsic mode components, and the structural mode frequency, thermal mode frequency, and electromagnetic mode frequency of the communication device are determined. Calculate the correlation coefficients between each of the intrinsic modal components and the structural modal frequencies, the thermal modal frequencies, and the electromagnetic modal frequencies; Target modal components associated with the physical characteristics of the communication device are selected based on the correlation coefficient. The target modal components are sorted by frequency from low to high and then combined to obtain the intrinsic modal features.

[0008] Optionally, in a third implementation of the first aspect of the present invention, calculating the correlation coefficients between each of the intrinsic modal components and the structural modal frequencies, the thermal modal frequencies, and the electromagnetic modal frequencies includes: The intrinsic modal components and the structural modal frequencies are subjected to coherence function analysis to obtain the structural coherence coefficient; The thermal coupling coefficient is obtained by performing transfer function analysis on the intrinsic modal components and the thermal modal frequencies. The intrinsic mode components are matched with the electromagnetic mode frequencies using spectral energy matching to obtain electromagnetic matching coefficients; The correlation coefficient is the weighted sum of the structural coherence coefficient, the thermal coupling coefficient, and the electromagnetic matching coefficient.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, determining the structural mode frequency, thermal mode frequency, and electromagnetic mode frequency of the communication device includes: Based on the geometric model and material parameters of the communication device, the structural modal frequencies are calculated using a preset finite element analysis. The thermal mode frequency is calculated using a preset thermodynamic model based on the heat conduction path and heat source distribution of the communication device. The electromagnetic mode frequency is obtained by electromagnetic simulation calculation based on the radio frequency circuit topology and component parameters of the communication device.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, determining the operating status detection result of the communication device based on the dynamic offset between the projection vector of the real-time acquired sensing data in the state basis space and the fused state basis vector includes: Calculate the Euclidean or Mahalanobis distance between the projection vector and the fused state basis vector to obtain the dynamic offset; Calculate the cumulative sum and rate of change of the dynamic offset within the sliding time window; If the cumulative sum exceeds the first threshold and the rate of change exceeds the second threshold, the communication device is determined to be in an abnormal state; otherwise, the degree of degradation of the communication device is determined based on the magnitude of the cumulative sum, so as to obtain the operating status detection result of the communication device.

[0011] Optionally, in a sixth implementation of the first aspect of the present invention, the step of extracting the time-domain features and frequency-domain features corresponding to each of the sensors based on the aligned sensing data to obtain a multi-source feature matrix includes: For the aligned sensing data of each sensor, time-domain statistical features and frequency-domain energy spectrum features are extracted to obtain the single-source feature set corresponding to each sensor. The various single-source feature sets are combined to obtain the multi-source feature matrix.

[0012] Optionally, in a sixth implementation of the first aspect of the present invention, the step of extracting the time-domain features and frequency-domain features corresponding to each of the sensors based on the aligned sensing data to obtain a multi-source feature matrix includes: For the aligned sensing data of each sensor, time-domain statistical features and frequency-domain energy spectrum features are extracted to obtain the single-source feature set corresponding to each sensor. The various single-source feature sets are combined to obtain the multi-source feature matrix.

[0013] Optionally, in a seventh implementation of the first aspect of the present invention, the step of performing spatiotemporal alignment on the multi-source sensing data to obtain time-synchronized and spatially correlated aligned sensing data includes: The multi-source sensor data is time-interpolated and aligned to obtain time-synchronized sensor data; Based on the physical topology of the communication device, spatial coordinate registration is performed on the time-synchronized sensor data to obtain the spatially correlated aligned sensor data.

[0014] Secondly, embodiments of the present invention provide an electronic device, 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, it implements the steps of the above-described method for detecting the operating status of a communication device based on multi-source sensor data fusion.

[0015] Thirdly, embodiments of the present invention provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for detecting the operating status of a communication device based on multi-source sensor data fusion.

[0016] Fourthly, embodiments of the present invention provide a computer program product that, when run on an electronic device, causes the electronic device to execute the aforementioned communication device operation status detection method based on multi-source sensor data fusion.

[0017] The beneficial effects of this invention compared to the prior art are as follows: by performing spatiotemporal alignment on multi-source sensor data, the data inconsistency caused by time asynchrony and spatial misalignment is eliminated, and multi-physics field sensor data are fused under a unified spatiotemporal reference; on this basis, time-domain features and frequency-domain features are extracted to generate a multi-source feature matrix, which is then mapped to a preset state basis space to obtain a fused state basis vector reflecting the normal operating state of the device; by using the dynamic offset between the projection vector of the sensor data collected in real time in this basis space and the fused state basis vector, the operating state of the device can be accurately determined, improving the detection sensitivity of early faults and performance degradation of communication equipment. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of an embodiment of the communication device operation status detection method based on multi-source sensor data fusion in this invention. Figure 2 This is a schematic diagram of Embodiment 2 of the communication device operation status detection method based on multi-source sensor data fusion in this invention. Figure 3 This is a schematic diagram of Embodiment 3 of the communication device operation status detection method based on multi-source sensor data fusion in this invention. Figure 4 This is a schematic diagram of Embodiment 4 of the communication device operation status detection method based on multi-source sensor data fusion in this invention. Figure 5 This is a schematic diagram of one embodiment of the electronic device in this invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are protected by this invention.

[0021] It should be noted that the terms "comprising," "including," and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this invention, are intended to cover non-exclusive inclusion. For example, a process, method, terminal, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. In the claims, specification, and accompanying drawings of this invention, relational terms such as "first" and "second" are used merely to distinguish one entity / operation / object from another entity / operation / object, and do not necessarily require or imply any such immediate relationship or order between these entities / operations / objects.

[0022] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] When communication equipment is in operation, multiple sensors are typically deployed for status monitoring. Because the sampling frequencies and installation locations of various sensors differ, the multi-source sensor data collected are asynchronous and misaligned in time and space, making direct fusion analysis difficult.

[0024] Traditional methods often rely on independent judgment using data from a single sensor or simple data aggregation followed by threshold comparison. These methods are easily affected by environmental noise and fluctuations in operating conditions, resulting in insufficient sensitivity for detecting early performance degradation and minor faults in equipment. A new technical approach is needed to address these issues.

[0025] In view of this, embodiments of the present invention provide a method, device, and storage medium for detecting the operating status of communication equipment based on multi-source sensor data fusion. By performing spatiotemporal alignment on multi-source sensor data, data inconsistencies caused by time asynchrony and spatial misalignment are eliminated, and multi-physics sensor data are fused under a unified spatiotemporal reference. On this basis, time-domain features and frequency-domain features are extracted to generate a multi-source feature matrix, which is then mapped to a preset state basis space to obtain a fused state basis vector reflecting the normal operating status of the equipment. By using the dynamic offset between the projection vector of the sensor data collected in real time in this basis space and the fused state basis vector, the operating status of the equipment can be accurately determined, improving the detection sensitivity of early faults and performance degradation of communication equipment.

[0026] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0027] Example 1: Figure 1 This illustration shows a flowchart of a communication device operation status detection method based on multi-source sensor data fusion, provided by an embodiment of the present invention. This method can be applied to electronic devices, such as servers, service clusters, mobile phones, tablets, laptops, ultra-mobile personal computers (UMPCs), netbooks, etc.

[0028] Specifically, the above-mentioned communication equipment operation status detection method based on multi-source sensor data fusion may include the following steps S101 to S104.

[0029] Step S101: Acquire multi-source sensing data collected by multiple sensors deployed on the communication device. The multi-source sensing data includes vibration sensing data, temperature sensing data, current sensing data, and radio frequency sensing data.

[0030] In embodiments of the present invention, the electronic device acquires multi-source sensing data collected in real time from multiple sensors deployed on the communication device via a data acquisition interface. The multi-source sensing data includes vibration sensing data, temperature sensing data, current sensing data, and radio frequency sensing data.

[0031] Optionally, each sensor is polled sequentially according to a preset sampling period, or data frames from all sensors are received in parallel via a data bus; vibration sensing data is collected by an accelerometer mounted on the casing or key board of the communication equipment to reflect the mechanical vibration state of the equipment; temperature sensing data is collected by a temperature sensor attached to a heating element or heat dissipation channel to reflect the thermal state of the equipment; current sensing data is collected by a current transformer connected in series at the power input terminal to reflect the power consumption and electrical characteristics of the equipment; and radio frequency sensing data is collected by a directional coupler or power detection circuit coupled to the radio frequency output terminal to reflect the radio frequency transmission performance of the equipment.

[0032] Optionally, after acquiring the raw sensing data, a data validity check is performed. The data validity check includes checking whether the data is within a reasonable range, whether there are any dropped frames or out-of-order data, and adaptive processing of invalid data.

[0033] Step S102: Perform spatiotemporal alignment on the multi-source sensor data to obtain time-synchronized and spatially correlated aligned sensor data.

[0034] In an embodiment of the present invention, a spatiotemporal alignment operation is performed on the acquired multi-source sensor data to eliminate time asynchrony and spatial misalignment caused by differences in sampling frequency, asynchronous start-up time, and dispersed installation positions between different sensors, thereby generating time-synchronized and spatially correlated aligned sensor data.

[0035] Optionally, the spatiotemporal alignment operation includes two sub-steps: alignment in the time dimension and alignment in the spatial dimension.

[0036] In the time alignment sub-step, a unified reference time axis is determined. This reference time axis can be selected as the timestamp of the sensor data with the highest sampling frequency, or a new time axis generated by equal-interval interpolation can be used. Then, for sensor data with a lower sampling frequency, linear interpolation, cubic spline interpolation, or a model-based resampling method are used to map its data values ​​to each time point on the reference time axis, so that all sensor data have the same sampling time and sampling interval in time.

[0037] In the spatial alignment sub-step, the physical topology information of the communication device is acquired. This information includes the installation coordinates of each sensor in the device coordinate system and the device's geometric dimensions. Based on the physical topology, a unified spatial reference system is generated, for example, a three-dimensional spatial coordinate system is established with the geometric center of the communication device or a fixed reference point as the origin. Then, each time-synchronized sensor data is associated with its corresponding sensor spatial coordinates to obtain spatiotemporal data points with spatial location labels. For data that needs to be compared or fused at the same spatial location, spatial interpolation methods are used to extrapolate the data from discrete sensor locations to a unified grid point or region of interest, thereby achieving spatial coordinate registration.

[0038] Optionally, if the physical topology of the communication device is complex, a finite element mesh model can be invoked to map the sensor data onto the model nodes, so that data from different physical fields can be expressed at the same spatial discrete points.

[0039] Step S103: Based on the alignment sensing data, extract the time-domain features and frequency-domain features corresponding to each sensor to obtain a multi-source feature matrix.

[0040] In an embodiment of the present invention, using aligned sensing data as input, time-domain features and frequency-domain features that can characterize the operating state are extracted for each sensor's data sequence, and then the features extracted from all sensors are organized into a multi-source feature matrix according to a preset arrangement order.

[0041] Optionally, for the aligned sensing data of each sensor, data preprocessing is performed, including detrending, filtering and noise reduction, and outlier removal.

[0042] For time-domain feature extraction, at least one of the following is calculated: mean, root mean square (RMS), peak value, kurtosis, skewness, peak factor, margin factor, and waveform factor. The mean reflects the central tendency of the data; the RMS reflects the intensity of data fluctuation; the peak value reflects the maximum instantaneous deviation of the data; the kurtosis reflects the sharpness of the data distribution and is sensitive to impact faults; the skewness reflects the asymmetry of the data distribution; the peak factor, which is the ratio of the peak value to the RMS, is used to detect impulsive faults; and the margin factor and waveform factor are dimensionless indicators.

[0043] For frequency domain feature extraction, Fourier transform or power spectral density estimation is performed on the time-domain data sequence to obtain the frequency domain energy spectrum. The entire frequency band is divided into several sub-bands, such as low-frequency, mid-frequency, and high-frequency bands, and the proportion of energy in each sub-band to the total energy is calculated; alternatively, frequency domain statistics such as centroid frequency, root mean square frequency, and frequency variance are extracted. Optionally, for vibration sensing data, the focus is on the frequency band related to the mechanical resonance frequency of the equipment; for radio frequency sensing data, the focus is on the spectral purity and spurious components near the operating frequency.

[0044] After obtaining the time-domain and frequency-domain features of each sensor, these feature values ​​are arranged according to a two-dimensional structure based on the sensor and feature type. Specifically, multiple features of each sensor are grouped into a row, and the row vectors of all sensors are stacked vertically to obtain a matrix of size equal to the number of sensors multiplied by the total number of features. This matrix is ​​the multi-source feature matrix. Optionally, the multi-source feature matrix can be normalized to make features with different dimensions and amplitude ranges comparable.

[0045] Step S104: Map the multi-source feature matrix to a preset state basis space to obtain the fused state basis vector.

[0046] In an embodiment of the present invention, the multi-source feature matrix is ​​mapped to a low-dimensional state basis space through a preset transformation model, and a fused state basis vector is output in the space. This vector is used to characterize the overall operating state of the communication device at the current moment.

[0047] Optionally, the state basis space and the corresponding mapping transformation model are pre-trained in the offline phase. Specifically, when the communication equipment is in a known normal operating state, a large number of historical multi-source feature matrix samples are collected. Then, through dimensionality reduction algorithms such as principal component analysis, linear discriminant analysis, autoencoders, or manifold learning, a low-dimensional subspace is learned from these samples. This subspace is called the state basis space. The mapping transformation model includes a set of basis vectors or an encoder network, used to project any input multi-source feature matrix onto this subspace to obtain the corresponding coordinate vector.

[0048] During runtime, the current multi-source feature matrix is ​​input into the mapping transformation model. If a linear mapping method is used, the inner product of the current feature matrix and each basis vector is calculated to obtain a set of projection coefficients. These coefficients are then arranged in order to form the fusion state basis vectors. If a nonlinear mapping method is used, the feature matrix is ​​propagated forward layer by layer, and low-dimensional vectors are extracted from the bottleneck layer as the fusion state basis vectors.

[0049] Optionally, the dimension of the fused state basis vector is much smaller than the dimension of the original multi-source feature matrix, preferably 2 to 10 dimensions, to facilitate subsequent dynamic offset calculation and state visualization. Each dimension of this vector represents the response intensity of the device in a certain potential state dimension. For example, the first dimension may correspond to the overall health, the second dimension may correspond to the thermal load level, and the third dimension may correspond to the degree of RF performance degradation, etc.

[0050] Step S105: Determine the operating status detection result of the communication device based on the dynamic offset between the projection vector of the real-time collected sensor data in the state basis space and the fused state basis vector.

[0051] In this embodiment of the invention, the real-time sensor data collected at the current moment is processed according to the same procedure as described above to obtain the current multi-source feature matrix. Then, using the same mapping transformation model, the current feature matrix is ​​projected onto the state basis space to obtain a real-time projection vector. The dynamic offset between the real-time projection vector and the fused state basis vector is calculated, and the magnitude and trend of the dynamic offset are used to determine whether the communication device is currently in a normal, abnormal, or degraded state, and the operating status detection result is output.

[0052] Optionally, when calculating the dynamic offset, Euclidean distance can be used, which is the square root of the sum of squares of the differences between corresponding components of two vectors; or Mahalanobis distance can be used, which standardizes the distance by considering the variance and correlation of each dimension in the state basis space. The dynamic offset is a scalar value that reflects the degree to which the current state deviates from the normal operating baseline.

[0053] Optionally, instead of directly using instantaneous offsets for judgment, a sliding time window is generated, accumulating historical offset values ​​within the window. The cumulative sum and rate of change of offsets within the window are calculated. The cumulative sum is compared with a preset first threshold, and the rate of change is compared with a preset second threshold. If the cumulative sum exceeds the first threshold and the rate of change exceeds the second threshold, it indicates that the device not only has a significant deviation but the deviation is also accelerating, and the communication device is determined to be in an abnormal state. If the cumulative sum exceeds the first threshold but the rate of change does not exceed the second threshold, it indicates that the device has experienced performance degradation but has not yet deteriorated sharply, and the device is determined to be in a degraded state, with the degree of degradation quantified based on the magnitude of the cumulative sum. If the cumulative sum does not exceed the first threshold, the device is determined to be in a normal state.

[0054] Optionally, a time-series curve of the dynamic offset can also be output to monitor the evolution trend of the device status; when an abnormal state is detected, an alarm signal is generated, the time point of the abnormality and the peak value of the offset are recorded, and the operation status detection results are output in the form of a display interface, network interface or log file.

[0055] The beneficial effects of this invention compared to the prior art are as follows: by performing spatiotemporal alignment on multi-source sensor data, the data inconsistency caused by time asynchrony and spatial misalignment is eliminated; multi-physics field sensor data are fused under a unified spatiotemporal reference; based on this, time-domain features and frequency-domain features are extracted to generate a multi-source feature matrix, which is then mapped to a preset state basis space to obtain a fused state basis vector reflecting the normal operating state of the device; by using the dynamic offset between the projection vector of the sensor data collected in real time in this basis space and the fused state basis vector, the operating state of the device can be accurately determined, improving the detection sensitivity of early faults and performance degradation of communication equipment.

[0056] Example 2: In an optional specific embodiment of the present invention, considering that traditional methods often employ unsupervised dimensionality reduction techniques such as principal component analysis or linear discriminant analysis to directly project the feature matrix, and given that sensor data is mixed with environmental noise and electromagnetic interference, these noise components are coupled with the intrinsic state information of the device itself in the feature space. Direct dimensionality reduction cannot effectively distinguish between the true state of the device and external interference, resulting in the established baseline vector containing noise components, which reduces the accuracy of subsequent state detection. Therefore, Embodiment Two is proposed to solve the above-mentioned technical problems. (Refer to...) Figure 2 , Figure 2 This is a schematic diagram of Embodiment 2 of the communication device operation status detection method based on multi-source sensor data fusion in the present invention. The multi-source feature matrix is ​​mapped to a preset state basis space to obtain a fused state basis vector, which may specifically include steps S201 to S202.

[0057] Step S201: Obtain the historical multi-source feature matrix of the communication device under normal operating conditions for multiple historical periods.

[0058] In an embodiment of the present invention, multiple sets of multi-source feature matrices collected by the communication device during a known normal operating state are read from the storage unit. These are historical multi-source feature matrices, with each set of feature matrices corresponding to a historical sampling period.

[0059] Optionally, based on the operation logs of the communication equipment or manual annotations, data from periods when the equipment is stable and fault-free can be filtered out, and feature matrices corresponding to non-steady-state conditions such as start-up and shutdown, and load changes can be removed.

[0060] Optionally, the historical multi-source feature matrix is ​​arranged in chronological order to obtain a historical feature sequence, and then preprocessed by detrending and normalization to eliminate slow drift between different operating cycles.

[0061] Step S202: Perform intrinsic dimension decomposition on the historical multi-source feature matrix to obtain intrinsic modal features.

[0062] In an embodiment of the present invention, multiple historical multi-source feature matrices are regarded as a high-dimensional dataset. An intrinsic dimension decomposition operation is performed on the dataset to extract low-dimensional features that represent the inherent modal characteristics of the device, which are called intrinsic modal features.

[0063] Optionally, algorithms such as variational mode decomposition, empirical mode decomposition, or singular spectrum analysis are used to decompose the historical feature matrix sequence into several intrinsic mode components, each corresponding to a specific time scale or frequency component. Further, by analyzing the correlation between each component and the equipment's physical field, components strongly correlated with the equipment's intrinsic dynamic behavior are selected, while components related to random noise or external disturbances are discarded. The retained components are then combined into eigenmode eigenvectors or feature matrices according to preset rules.

[0064] Step S203: Generate a constraint basis orthogonal to the intrinsic modal features, wherein the constraint basis is used to suppress the interference of environmental noise on the feature space.

[0065] In an embodiment of the present invention, a set of basis vectors orthogonal to the intrinsic modal features in the feature space are generated based on the intrinsic modal features. The basis vectors are used to suppress the interference of environmental noise on the feature space during subsequent projection.

[0066] Optionally, the intrinsic modal features are used as a signal subspace, and the orthogonal complement space of this subspace is calculated. Specifically, if the intrinsic modal features constitute a column vector or matrix, a set of basis vectors orthogonal to the subspace spanned by the intrinsic modal features is obtained by using Schmitt orthogonalization or solving for the null space, and these orthogonal basis vectors are combined into a constraint basis matrix.

[0067] Optionally, prior physical knowledge of the communication device can be introduced, such as the theoretical modal orientation provided by the finite element model. After the correlation between the theoretical modal orientation and the intrinsic modal features is matched, only the basis that is orthogonal to the intrinsic modal features and consistent with the orientation of the main noise source is retained, thereby enhancing the noise suppression capability of the constrained basis.

[0068] Step S204: Orthogonally project the intrinsic modal features onto the constrained basis to obtain the fused state basis vector.

[0069] In an embodiment of the present invention, the intrinsic modal features are projected onto the subspace spanned by the constraint basis. Specifically, the component coefficients of the intrinsic modal features in each orthogonal direction of the constraint basis are calculated, and these coefficients are arranged in the basis order to form a vector, which is the fusion state basis vector.

[0070] Optionally, the constraint basis matrix is ​​orthogonally normalized, and then the projection coefficients are calculated as the product of the transpose of the constraint basis matrix and the intrinsic modal eigenvectors. This projection process is equivalent to extracting the components of the intrinsic modal features that are orthogonal to the constraint basis, while filtering out the components that are parallel to the intrinsic modal features, thereby obtaining a purer device state representation.

[0071] In this embodiment of the invention, the historical multi-source feature matrix of the communication device under normal operating conditions is obtained, and its intrinsic dimension is decomposed to extract intrinsic modal features reflecting the inherent physical characteristics of the device. Then, a constraint basis orthogonal to these intrinsic modal features is generated. This constraint basis is used to suppress the interference of environmental noise on the feature space. Finally, the intrinsic modal features are orthogonally projected onto the constraint basis to obtain the fused state basis vector. This method can actively filter out feature directions related to environmental noise when generating the state reference space, making the final fused state basis vector more purely represent the health state of the device, thereby improving the accuracy of subsequent dynamic offset detection for early faults.

[0072] Example 3: In an optional specific embodiment of the present invention, considering that traditional methods typically only perform purely mathematical decomposition on historical multi-source feature matrices, the decomposed components lack clear physical meaning. Since vibration, temperature, current, and radio frequency sensing data correspond to different physical fields such as the mechanical structure, thermodynamics, and electromagnetics of the device, the modal components derived solely through data-driven decomposition often contain coupling information from multiple physical fields, making it difficult to establish a correspondence with the actual structural, thermal, and electromagnetic modes of the device. This results in a lack of quantitative basis for subsequent screening of modal components related to the device's physical characteristics, leading to poor physical interpretability of the extracted intrinsic modal features and an inability to accurately reflect the health status of each physical subsystem of the device. Therefore, Embodiment Three is proposed to solve the above-mentioned technical problems. (Refer to...) Figure 3 , Figure 3 This is a schematic diagram of Embodiment 3 of the communication device operation status detection method based on multi-source sensor data fusion in the present invention, which may specifically include steps S301 to S304.

[0073] Step S301: Decompose the historical multi-source feature matrix into multiple intrinsic mode components, and determine the structural mode frequency, thermal mode frequency and electromagnetic mode frequency of the communication device.

[0074] In an embodiment of the present invention, historical multi-source feature matrices of multiple historical periods are obtained, these matrices are treated as a high-dimensional multivariate time series, and a signal decomposition algorithm is used to decompose them into several intrinsic mode components, each component representing an independent oscillation mode or trend.

[0075] Optionally, a variational mode decomposition algorithm can be used, with a preset number of decomposition layers, to iteratively decompose the historical multi-source feature matrix sequence into a specified number of intrinsic mode components; or an empirical mode decomposition algorithm can be used, which adaptively decomposes layer by layer according to the time-scale characteristics of the data itself to obtain a set of intrinsic mode function components. Each intrinsic mode component has the same dimensional structure as the original feature matrix, that is, it contains information from all sensors and all feature types.

[0076] The inherent modal frequencies of the communication device in the three physical fields of structure, heat and electromagnetic are obtained or calculated, and used as a reference for subsequent correlation coefficient calculation.

[0077] Optionally, the structural modal frequency values, thermal modal frequency values ​​(and electromagnetic modal frequency values) obtained through offline simulation can be read from a preset configuration file. If not pre-saved, the finite element analysis module, thermodynamic model, and electromagnetic simulation module can be invoked to perform calculations based on the device's geometric model, material parameters, heat conduction path, and RF circuit topology, and the calculation results can be stored for later use.

[0078] Step S302: Calculate the correlation coefficients between each intrinsic modal component and the structural modal frequency, thermal modal frequency, and electromagnetic modal frequency.

[0079] In an embodiment of the present invention, for each intrinsic modal component, the correlation metric between it and the structural modal frequency, thermal modal frequency and electromagnetic modal frequency is calculated to obtain three sets of correlation coefficients.

[0080] Optionally, each intrinsic modal component is expanded in the time and frequency domains to extract its power spectral density distribution; simultaneously, the structural modal frequency, thermal modal frequency, and electromagnetic modal frequency are treated as discrete reference frequency points or frequency bands. The ratio of the peak energy of the intrinsic modal component's spectrum near the structural modal frequency to the total energy is calculated as the structural correlation coefficient; similarly, the energy ratio near the thermal modal frequency is calculated as the thermal correlation coefficient; and the energy ratio near the electromagnetic modal frequency is calculated as the electromagnetic correlation coefficient. Alternatively, the electronic device uses a frequency domain coherence function to calculate the average coherence coefficient between the intrinsic modal component and each reference frequency.

[0081] For each intrinsic modal component, the correlation metric between it and the structural modal frequency, thermal modal frequency, and electromagnetic modal frequency is calculated, resulting in three sets of correlation coefficients.

[0082] Optionally, each intrinsic modal component is expanded in the time and frequency domains to extract its power spectral density distribution; simultaneously, the structural modal frequency, thermal modal frequency, and electromagnetic modal frequency are treated as discrete reference frequency points or frequency bands. The ratio of the peak energy of the intrinsic modal component's spectrum near the structural modal frequency to the total energy is calculated as the structural correlation coefficient; similarly, the energy ratio near the thermal modal frequency is calculated as the thermal correlation coefficient; and the energy ratio near the electromagnetic modal frequency is calculated as the electromagnetic correlation coefficient. Alternatively, the electronic device uses a frequency domain coherence function to calculate the average coherence coefficient between the intrinsic modal component and each reference frequency.

[0083] Step S303: Filter target modal components that are associated with the physical characteristics of the communication equipment based on the correlation coefficient.

[0084] In an embodiment of the present invention, the three correlation coefficients of each intrinsic modal component are compared with a preset correlation coefficient threshold, and the components that have significant correlation in at least one physical mode are selected and marked as target modal components.

[0085] Optionally, different correlation coefficient thresholds can be set for the structural, thermal, and electromagnetic physical fields. If the structural correlation coefficient of an intrinsic mode component exceeds the structural threshold, it is classified as a structurally correlated target mode component; if the thermal correlation coefficient exceeds the thermal threshold, it is classified as a thermally correlated target mode component; and if the electromagnetic correlation coefficient exceeds the electromagnetic threshold, it is classified as an electromagnetically correlated target mode component. An intrinsic mode component may be significantly correlated with multiple physical fields simultaneously; such a component is retained and marked as a multi-physics coupled component. Electronic devices will treat intrinsic mode components that do not meet any threshold conditions as noise or redundant components.

[0086] Step S304: Sort the target modal components by frequency from low to high and combine them to obtain intrinsic modal features.

[0087] In an embodiment of the present invention, all the selected target modal components are sorted in ascending order of their characteristic frequencies, and then the sorted components are combined into a whole by vector concatenation or matrix stacking, which is the intrinsic modal feature.

[0088] Optionally, the characteristic frequencies of each target modal component are calculated. All target modal components are sorted in ascending order of representative frequencies. For components with similar frequencies, a secondary sort is performed based on their physical properties; for example, structurally related components are sorted first, followed by thermally related components, and then electromagnetically related components. After sorting, each component is flattened into a one-dimensional vector and joined end-to-end in sequence to form a long vector, or the components are arranged as column vectors side-by-side to form a matrix. This vector or matrix is ​​the intrinsic modal feature, used to replace the original historical multi-source feature matrix for subsequent steps.

[0089] In this embodiment of the invention, the historical multi-source feature matrix is ​​decomposed into multiple intrinsic mode components, and the structural mode frequency, thermal mode frequency, and electromagnetic mode frequency of the communication device are predetermined. Then, the correlation coefficient between each intrinsic mode component and the above three physical mode frequencies is calculated. Based on the correlation coefficient, target mode components that are strongly correlated with the physical characteristics of the device are selected. Finally, the intrinsic mode features are combined in ascending order of frequency. This ensures that each selected target mode component has a clear physical affiliation, thereby accurately reflecting the health status of each physical subsystem of the device.

[0090] Example 4: In an optional specific embodiment of the present invention, considering that traditional methods typically use a single Pearson correlation coefficient or cross-correlation function to perform overall matching of the entire time-domain sequence or frequency-domain spectrum, structural mode frequencies usually exhibit discrete resonance peaks, thermal mode frequencies often manifest as low-frequency slow variations or periodic thermal fluctuations, while electromagnetic mode frequencies may present as high-frequency narrowband interference or harmonic components. These three physical modes have drastically different frequency-domain distribution characteristics and coupling mechanisms, making it difficult to accurately characterize the true correlation strength between intrinsic mode components and each physical field using a unified correlation coefficient calculation method, leading to insufficient accuracy in subsequent mode selection. Therefore, Embodiment Four is proposed to solve the above-mentioned technical problems. (Reference) Figure 4 , Figure 4 This is a schematic diagram of Embodiment 4 of the communication equipment operation status detection method based on multi-source sensor data fusion in the present invention. The calculation of the correlation coefficient between each intrinsic mode component and the structural mode frequency, thermal mode frequency and electromagnetic mode frequency may specifically include steps S401 to S404.

[0091] Step S401: Perform coherence function analysis on the intrinsic modal components and structural modal frequencies to obtain the structural coherence coefficient.

[0092] In an embodiment of the present invention, for each intrinsic mode component, frequency domain coherence function analysis is performed between it and the structural mode frequency of the communication device to calculate a value that quantifies the degree of linear correlation between the two in the frequency domain, which is used as the structural coherence coefficient.

[0093] Optionally, a Fourier transform is performed on the time series of the intrinsic mode components to obtain their power spectral density. A narrow-band frequency window is defined with the structural mode frequency as the center frequency. The ratio of the average energy to the total energy of the power spectrum of the intrinsic mode components within this frequency window is calculated, and this ratio is used as the output of the coherence function analysis. Alternatively, the standard coherence function calculation method is used to calculate the coherence coefficient between the intrinsic mode components and an ideal sinusoidal signal, obtaining a coherence value between 0 and 1.

[0094] Step S402: Perform transfer function analysis on the intrinsic modal components and thermal modal frequencies to obtain the thermal coupling coefficient.

[0095] In the embodiments of the present invention, the intrinsic modal components are regarded as the output response of the system, and the thermal excitation corresponding to the thermal modal frequency is regarded as the system input. The response intensity of the intrinsic modal components to the thermal excitation is evaluated by transfer function analysis to obtain the thermal coupling coefficient.

[0096] Optionally, a reference signal simulating thermal excitation is generated. The frequency of this signal is equal to the thermal mode frequency, and the waveform can be a sine wave or a square wave. The time series of the intrinsic mode components is used as the output response. Least squares estimation is used to estimate the amplitude gain of the transfer function from thermal excitation to the intrinsic mode components at the thermal mode frequency. This gain value is normalized and used as the thermal coupling coefficient. If the thermal mode frequency is within a frequency band, the integral average of the transfer function amplitude within that band is calculated.

[0097] Step S403: Perform spectral energy matching between the intrinsic modal components and the electromagnetic modal frequencies to obtain the electromagnetic matching coefficients.

[0098] In an embodiment of the present invention, the spectrum of the intrinsic modal component is subjected to energy matching analysis with the frequency of the electromagnetic mode, and the energy concentration of the intrinsic modal component near the frequency of the electromagnetic mode is calculated as the electromagnetic matching coefficient.

[0099] Optionally, a Fast Fourier Transform (FFT) is performed on the intrinsic mode components to obtain their discrete spectra. A frequency window is defined with the electromagnetic mode frequency as the center. The sum of the energies of all frequency points within this window is calculated, and then divided by the total spectral energy of the intrinsic mode components. The resulting ratio is used as the electromagnetic matching coefficient. If there are multiple electromagnetic mode frequencies, the energy proportion within each frequency window is calculated separately, and then the maximum value or a weighted sum is taken as the final electromagnetic matching coefficient.

[0100] Step S404: The weighted sum of the structural coherence coefficient, thermal coupling coefficient, and electromagnetic matching coefficient is used as the correlation coefficient.

[0101] In an embodiment of the present invention, the three types of coefficients are weighted and summed according to preset weights to obtain the comprehensive correlation coefficient of each intrinsic mode component.

[0102] Optionally, a first weight, a second weight, and a third weight are assigned to the structural coherence coefficient, thermal coupling coefficient, and electromagnetic matching coefficient, respectively, with the sum of the three weights equal to 1. The weight values ​​can be preset according to the actual operating conditions of the communication equipment. For example, for equipment where mechanical vibration is the primary failure mode, the weight of the structural coherence coefficient is set to a higher value, while the weights for thermal and electromagnetic coupling coefficients are each set to a lower value. For equipment with significant heat dissipation problems, the weight of the thermal coupling coefficient is increased. For example, the weighted sum for electronic equipment is calculated as follows: Correlation coefficient = (Weight 1 × Structural coherence coefficient) + (Weight 2 × Thermal coupling coefficient) + (Weight 3 × Electromagnetic matching coefficient).

[0103] In this embodiment of the invention, differentiated correlation measurement methods are used for three physical modes to measure the energy concentration of intrinsic mode components within a specific electromagnetic frequency band. Finally, the three coefficients are weighted and summed to obtain a comprehensive correlation coefficient. This achieves correlation measurement under multi-physics coupling, improving the accuracy of determining the association between intrinsic mode components and the actual physical modes of the device.

[0104] Example 5: In an optional embodiment of the present invention, determining the structural mode frequency, thermal mode frequency, and electromagnetic mode frequency of the communication device may specifically include steps S501 to S502.

[0105] Step S501: Based on the geometric model and material parameters of the communication equipment, the structural modal frequencies are calculated using a preset finite element analysis.

[0106] In an embodiment of the present invention, a three-dimensional geometric model and material parameter data of a communication device are obtained, a preset finite element analysis module is called to perform modal analysis calculations on the device, and the natural frequencies of the device in multiple degrees of freedom are output as structural modal frequencies.

[0107] Optionally, the 3D geometric model is imported into finite element analysis software, mesh generation parameters are set, and material parameters for each component are set. Boundary conditions are set, such as fixing the installation points or support locations of the equipment. Modal analysis is performed to calculate the first few natural frequencies of the equipment; these frequencies are the structural modal frequencies. Optionally, only modal frequencies below a preset upper limit are retained to match the measurement range of the vibration sensor.

[0108] Step S502: Based on the heat conduction path and heat source distribution of the communication equipment, the thermal mode frequency is calculated using a preset thermodynamic model.

[0109] In an embodiment of the present invention, the heat conduction path description and heat source distribution data of the communication device are obtained, a preset thermodynamic model is called, transient thermal analysis or periodic steady-state thermal analysis is performed, and the dominant response frequency of the device under periodic thermal excitation is calculated as the thermal mode frequency.

[0110] Optionally, the heat conduction path and heat source distribution are input into thermodynamic simulation software to generate thermal resistance and thermal capacity network models. The heat source is set to generate heat in the form of a periodic square wave or sine wave, and the temperature response at key temperature measurement points is simulated and calculated. The temperature response curve is extracted, and the excitation frequency with the largest response amplitude is identified through Fourier transform; this frequency is the thermal mode frequency. Optionally, the electronic device can also calculate multiple thermal mode frequencies, corresponding to thermal diffusion modes at different locations and time scales.

[0111] Step S503: Based on the radio frequency circuit topology and component parameters of the communication equipment, the electromagnetic mode frequency is obtained through electromagnetic simulation calculation.

[0112] In this embodiment of the invention, the radio frequency circuit topology and component parameters of the communication device are obtained, and a preset electromagnetic simulation module is invoked to perform circuit simulation or electromagnetic field simulation, calculating the self-oscillation frequency or resonant frequency of the radio frequency circuit as the electromagnetic mode frequency. The circuit topology includes the connection relationships of modules such as power amplifiers, filters, and antennas, and the component parameters include inductor, capacitor, and resistor values, as well as transistor models.

[0113] Optionally, the RF circuit topology is converted into a netlist file, imported into an electromagnetic simulation tool, and power and bias conditions are set. Small-signal AC analysis or harmonic balance analysis is performed, the frequency range is scanned, and the circuit's transfer function or input impedance is calculated. The frequency point corresponding to the peak gain in the transfer function, or the frequency point where the imaginary part of the impedance is zero, is identified; this frequency is the electromagnetic mode frequency.

[0114] Example 6: In an optional embodiment of the present invention, considering that traditional methods typically directly compare the instantaneous offset at the current moment with a fixed threshold, and determine an anomaly if the threshold is exceeded, but due to sensor noise, environmental disturbances, or normal fluctuations in operating conditions, the instantaneous offset is prone to random spikes, leading to false alarms, Embodiment Six is ​​proposed. Based on the dynamic offset between the projection vector of the real-time acquired sensor data in the state basis space and the fused state basis vector, the operating status detection result of the communication device is determined, specifically including steps S601 to S603.

[0115] Step S601: Calculate the Euclidean distance or Mahalanobis distance between the projection vector and the fused state basis vector to obtain the dynamic offset.

[0116] Step S602: Calculate the cumulative sum and rate of change of the dynamic offset within the sliding time window.

[0117] In an embodiment of the invention, a fixed-length sliding time window is maintained, which stores the dynamic offsets of the most recent sampling times in chronological order. At each new sampling time, the current dynamic offset is added to the end of the window, and the oldest offset at the beginning of the window is removed. Then, the cumulative sum and rate of change are calculated based on the offset sequence within the window.

[0118] Step S603: If the cumulative sum exceeds the first threshold and the rate of change exceeds the second threshold, the communication device is determined to be in an abnormal state; otherwise, the degree of degradation of the communication device is determined based on the magnitude of the cumulative sum to obtain the operating status detection result of the communication device.

[0119] In an embodiment of the present invention, the cumulative sum is compared with a preset first threshold, and the rate of change is compared with a preset second threshold. Based on the comparison results, the operating status of the communication device is determined.

[0120] If the cumulative sum exceeds the first threshold and the rate of change exceeds the second threshold, it indicates that the device not only has a large cumulative deviation but the deviation is also accelerating, thus the communication device is determined to be in an abnormal state. If the cumulative sum exceeds the first threshold but the rate of change does not exceed the second threshold, it indicates that the device has experienced performance degradation but has not yet deteriorated rapidly, and the degree of degradation is quantified based on the magnitude of the cumulative sum. If the cumulative sum does not exceed the first threshold, the device is determined to be in a normal state. The final determination result is output as the operating status detection result.

[0121] In this embodiment of the invention, a dynamic offset is obtained by calculating the Euclidean or Mahalanobis distance between the projected vector and the fused state basis vector. Then, the cumulative sum and rate of change of this offset are calculated within a sliding time window. When the cumulative sum exceeds a first threshold and the rate of change exceeds a second threshold, the device is determined to be in an abnormal state; otherwise, the degree of device degradation is determined based on the magnitude of the cumulative sum. This achieves unified detection of sudden anomalies and gradual degradation, improving the robustness of state judgment and early warning capabilities.

[0122] Example 7: In an optional specific embodiment of the present invention, considering that traditional communication equipment status monitoring methods usually rely on data from a single type of sensor for analysis, and their feature extraction dimensions are singular, it is difficult to comprehensively and deeply capture the comprehensive status information of the equipment under complex operating conditions formed by the coupling effects of multiple physical fields such as mechanical, thermal, and electrical fields. Therefore, this embodiment seven is proposed. Based on the aligned sensing data, the time-domain features and frequency-domain features corresponding to each sensor are extracted to obtain a multi-source feature matrix, which may specifically include steps S701 to S702.

[0123] Step S701: For the aligned sensing data of each sensor, extract the time-domain statistical features and frequency-domain energy spectrum features to obtain the single-source feature set corresponding to each sensor.

[0124] In embodiments of the present invention, statistical features in the time domain of single sensor data are extracted, including but not limited to mean, variance, peak value, RMS value, waveform factor, and impulse factor.

[0125] For single-sensor data, frequency domain transformation, such as Fourier transform, is performed to obtain its spectral distribution, and frequency domain energy spectrum features are extracted from it, including but not limited to dominant frequency, spectral centroid, spectral variance, and frequency band energy proportion.

[0126] Step S702: Combine the individual single-source feature sets to obtain a multi-source feature matrix.

[0127] In an embodiment of the present invention, the time-domain statistical feature set and the frequency-domain energy spectrum feature set corresponding to the sensor are combined to obtain a single-source feature set representing the sensor. The processor summarizes and arranges the single-source feature sets obtained from each sensor, and can combine them according to sensor type or spatial location order to obtain a structured multi-source feature matrix to comprehensively characterize the multi-dimensional state of the communication device at a specific moment.

[0128] In this embodiment of the invention, by synchronously extracting time-domain statistical features and frequency-domain energy spectrum features from the aligned data of each sensor, the statistical regularity changes of the device state in the time dimension and the energy distribution transfer in the frequency dimension can be captured respectively. This allows for a more comprehensive characterization of the device's operational behavior from two complementary time and frequency dimensions. Combining these features from multiple sensors into a unified matrix effectively integrates heterogeneous information from multiple sources such as vibration, temperature, current, and radio frequency, generating a high-dimensional, comprehensive state representation and improving the comprehensiveness and accuracy of state detection.

[0129] Example 8: In an optional embodiment of the present invention, considering that the original data streams from different locations and types of sensors on the communication device may have different sampling frequencies and start times, and that there is no clear spatial physical correlation between the data points, if fusion analysis is performed directly, feature extraction and state interpretation will be based on spatiotemporally misaligned data, introducing additional errors. Therefore, this embodiment eight is proposed. Spatiotemporal alignment of multi-source sensor data is performed to obtain time-synchronized and spatially correlated aligned sensor data, which may specifically include steps S801 to S802.

[0130] Step S801: Time interpolation and alignment are performed on the multi-source sensor data to obtain time-synchronized sensor data.

[0131] In an embodiment of the present invention, data from different sampling times are unified to a series of common, equally spaced time points through an interpolation algorithm, thereby obtaining a time-synchronized set of sensor data.

[0132] Step S802: Based on the physical topology of the communication device, spatial coordinate registration is performed on the time-synchronized sensor data to obtain spatially correlated aligned sensor data.

[0133] In an embodiment of the present invention, the specific installation location and spatial relationship of each sensor on the device are determined based on the physical topology information of the communication device pre-stored in the memory. Based on this topology, the time-synchronized sensor data are mapped and associated with spatial coordinates, assigning each data point its corresponding physical spatial coordinate information, and finally outputting aligned sensor data that is aligned in both time and spatial dimensions.

[0134] In this embodiment of the invention, by synchronously extracting time-domain statistical features and frequency-domain energy spectrum features from the aligned data of each sensor, the statistical regularity changes of the device state in the time dimension and the energy distribution transfer in the frequency dimension can be captured respectively. This allows for a more comprehensive characterization of the device's operational behavior from two complementary time and frequency dimensions. Combining these features from multiple sensors into a unified matrix effectively integrates heterogeneous information from multiple sources such as vibration, temperature, current, and radio frequency, generating a high-dimensional, comprehensive state representation and improving the comprehensiveness and accuracy of state detection.

[0135] like Figure 5 The diagram illustrates an electronic device according to an embodiment of the present invention. The electronic device 800 may include a processor 801, a memory 802, and a computer program 803 stored in the memory 802 and executable on the processor 801, such as a communication device operation status detection program based on multi-source sensor data fusion. When the processor 801 executes the computer program 803, it implements the steps described in the various embodiments of communication device operation status detection based on multi-source sensor data fusion.

[0136] A computer program can be divided into one or more modules / units. One or more modules / units are stored in memory 802 and executed by processor 801 to complete the present invention. One or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in an electronic device.

[0137] The electronic device may include, but is not limited to, a processor 801 and a memory 802. Those skilled in the art will understand that... Figure 5 This is merely an example of an electronic device and does not constitute a limitation on the electronic device. It may include more or fewer components than shown, or combine certain components, or different components. For example, an electronic device may also include input / output devices, network access devices, buses, etc.

[0138] The processor 801 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0139] The memory 802 can be an internal storage unit of an electronic device, such as a hard drive or RAM. The memory 802 can also be an external storage device of the electronic device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 802 can include both internal and external storage units. The memory 802 is used to store computer programs and other programs and data required by the electronic device. The memory 802 can also be used to temporarily store data that has been output or will be output.

[0140] It should be noted that, for the sake of convenience and brevity, the structure of the above-mentioned electronic device can also be referred to the specific description of the structure in the method embodiment, which will not be repeated here.

[0141] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement the steps in the above-described method for detecting the operating status of communication devices based on multi-source sensor data fusion.

[0142] This invention provides a computer program product that, when run on a mobile terminal, enables the mobile terminal to implement the steps in the above-mentioned communication device operation status detection method based on multi-source sensor data fusion.

[0143] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0144] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for various specific applications, but such implementations should not be considered beyond the scope of this invention.

[0145] In the embodiments provided by this invention, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the electronic device embodiments described above are merely illustrative. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0146] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0147] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0148] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0149] The embodiments described above are merely illustrative of the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for detecting the operating status of communication equipment based on multi-source sensor data fusion, characterized in that, include: Acquire multi-source sensing data collected by multiple sensors deployed on communication equipment, including vibration sensing data, temperature sensing data, current sensing data, and radio frequency sensing data; The multi-source sensor data is spatiotemporally aligned to obtain time-synchronized and spatially correlated aligned sensor data. Based on the alignment sensing data, extract the time-domain features and frequency-domain features corresponding to each sensor to obtain a multi-source feature matrix; The multi-source feature matrix is ​​mapped to a preset state basis space to obtain a fused state basis vector; The operating status detection result of the communication device is determined based on the dynamic offset between the projection vector of the real-time collected sensor data in the state basis space and the fused state basis vector.

2. The communication equipment operation status detection method based on multi-source sensor data fusion as described in claim 1, characterized in that, The step of mapping the multi-source feature matrix to a preset state basis space to obtain a fused state basis vector includes: Obtain the historical multi-source feature matrix of the communication device under normal operating conditions for multiple historical periods; The intrinsic dimension decomposition is performed on the historical multi-source feature matrix to obtain the intrinsic modal features; Generate a constraint basis orthogonal to the intrinsic mode features, wherein the constraint basis is used to suppress the interference of environmental noise on the feature space; The intrinsic modal features are orthogonally projected onto the constraint basis to obtain the fused state basis vector.

3. The communication equipment operation status detection method based on multi-source sensor data fusion as described in claim 2, characterized in that, The intrinsic dimension decomposition of the historical multi-source feature matrix to obtain intrinsic modal features includes: The historical multi-source feature matrix is ​​decomposed into multiple intrinsic mode components, and the structural mode frequency, thermal mode frequency, and electromagnetic mode frequency of the communication device are determined. Calculate the correlation coefficients between each of the intrinsic modal components and the structural modal frequencies, the thermal modal frequencies, and the electromagnetic modal frequencies; Target modal components associated with the physical characteristics of the communication device are selected based on the correlation coefficient. The target modal components are sorted by frequency from low to high and then combined to obtain the intrinsic modal features.

4. The communication equipment operation status detection method based on multi-source sensor data fusion as described in claim 3, characterized in that, The calculation of the correlation coefficients between each of the intrinsic modal components and the structural modal frequencies, the thermal modal frequencies, and the electromagnetic modal frequencies includes: The intrinsic modal components and the structural modal frequencies are subjected to coherence function analysis to obtain the structural coherence coefficient; The thermal coupling coefficient is obtained by performing transfer function analysis on the intrinsic modal components and the thermal modal frequencies. The intrinsic mode components are matched with the electromagnetic mode frequencies using spectral energy matching to obtain electromagnetic matching coefficients; The correlation coefficient is the weighted sum of the structural coherence coefficient, the thermal coupling coefficient, and the electromagnetic matching coefficient.

5. The communication equipment operation status detection method based on multi-source sensor data fusion as described in claim 3, characterized in that, Determining the structural mode frequency, thermal mode frequency, and electromagnetic mode frequency of the communication device includes: Based on the geometric model and material parameters of the communication device, the structural modal frequencies are calculated using a preset finite element analysis. The thermal mode frequency is calculated using a preset thermodynamic model based on the heat conduction path and heat source distribution of the communication device. The electromagnetic mode frequency is obtained by electromagnetic simulation calculation based on the radio frequency circuit topology and component parameters of the communication device.

6. The method for detecting the running state of the communication device based on multi-source sensing data fusion according to claim 1, characterized in that, The step of determining the operating status detection result of the communication device based on the dynamic offset between the projection vector of the real-time acquired sensor data in the state basis space and the fused state basis vector includes: Calculate the Euclidean or Mahalanobis distance between the projection vector and the fused state basis vector to obtain the dynamic offset; Calculate the cumulative sum and rate of change of the dynamic offset within the sliding time window; If the cumulative sum exceeds the first threshold and the rate of change exceeds the second threshold, the communication device is determined to be in an abnormal state; otherwise, the degree of degradation of the communication device is determined based on the magnitude of the cumulative sum, so as to obtain the operating status detection result of the communication device.

7. The communication equipment operation status detection method based on multi-source sensor data fusion as described in claim 1, characterized in that, The step of extracting the time-domain and frequency-domain features corresponding to each sensor based on the alignment sensing data to obtain a multi-source feature matrix includes: For the aligned sensing data of each sensor, time-domain statistical features and frequency-domain energy spectrum features are extracted to obtain the single-source feature set corresponding to each sensor. The various single-source feature sets are combined to obtain the multi-source feature matrix.

8. The method for detecting the running state of the communication device based on multi-source sensing data fusion according to claim 1, characterized in that, The step of performing spatiotemporal alignment on the multi-source sensor data to obtain time-synchronized and spatially correlated aligned sensor data includes: The multi-source sensor data is time-interpolated and aligned to obtain time-synchronized sensor data; Based on the physical topology of the communication device, spatial coordinate registration is performed on the time-synchronized sensor data to obtain the spatially correlated aligned sensor data.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the communication device operation status detection method based on multi-source sensor data fusion as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the communication device operation status detection method based on multi-source sensor data fusion as described in any one of claims 1 to 8.

Images