Remote maintenance guidance method and system for electrical equipment

Abstract

The invention provides an electrical equipment remote maintenance guidance method and system. The method comprises the following steps: capturing an aperiodic torque waveform and a magnetic field gradient abnormal signal by deploying a torque fluctuation sensor and a magnetoresistive sensor; performing multi-scale frequency band division on the non-periodic torque waveform, screening out a transient high-frequency component in a start-stop stage of equipment, and separating out a magnetic field polarity reversal characteristic from a magnetic field gradient abnormal signal; performing cross-domain matching on the occurrence time of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature to generate a mixed feature mark; calculating a frequency band overlapping degree based on the time-frequency energy distribution marked by the mixed features and generating a coherence map; and matching a fault historical case library according to the coherence map, and outputting a maintenance strategy set aiming at the rotating shaft dynamic unbalance and electromagnetic interference superposition fault. According to the invention, through fusion of torque fluctuation and magnetic field distortion collaborative analysis, the precision of mechanical and electromagnetic composite fault diagnosis of the rotating shaft is improved.

Description

Technical Field

[0001] The present application relates to the field of remote maintenance technology, and in particular to a remote maintenance guidance method and system for electrical equipment. Background Art

[0002] Bearings, as critical components in rotating equipment like industrial motors and fans, endure high loads and variable speeds for extended periods. Their health directly impacts equipment safety. With the advancement of intelligent equipment, remote monitoring is required to assess potential bearing failures, such as early wear and localized damage, in real time. This monitoring provides ongoing maintenance guidance to reduce the risk of unplanned downtime and operational costs.

[0003] The current mainstream solution uses an online monitoring system based on vibration signal analysis. By deploying acceleration sensors on the bearing seat to collect vibration spectrum data, combined with a pre-trained fault feature library, a machine learning model is used to identify abnormal frequency components in the vibration signal, generate a bearing health status assessment report and trigger an early warning signal.

[0004] This solution relies on a single vibration signal, which makes it difficult to distinguish between abnormal spectral characteristics caused by mechanical damage and electromagnetic interference. Especially in the variable frequency drive scenario, the motor electromagnetic noise and the bearing fault vibration frequency band overlap, resulting in an increased false alarm rate; in addition, the vibration signal is not sensitive enough to early minor damage and cannot capture subtle changes in the stress distribution inside the bearing, resulting in a delayed fault warning. Summary of the Invention

[0005] The present application provides a remote maintenance guidance method and system for electrical equipment, which is used to solve the problem of insufficient accuracy in the diagnosis of rotating shaft mechanical and electromagnetic composite faults in the prior art.

[0006] In a first aspect, the present application provides a method for remote maintenance guidance of electrical equipment, comprising:

[0007] A torque fluctuation sensor and a magnetoresistive sensor are synchronously deployed on the surface of the rotating shaft of the electrical equipment. The torque fluctuation sensor continuously captures the non-periodic torque waveform caused by the sudden change of the mechanical load on the shaft. At the same time, the magnetoresistive sensor is used to detect the abnormal magnetic field gradient signal generated when the rotating shaft deviates in an orthogonal direction.

[0008] Performing multi-scale frequency band division on the non-periodic torque waveform to screen out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separating the magnetic field polarity reversal characteristics corresponding to the axis offset direction from the magnetic field gradient anomaly signal;

[0009] Cross-domain matching is performed between the appearance time of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature, and a hybrid feature marker including mechanical and electromagnetic coupling relationships is generated based on the matching result;

[0010] Wavelet coherence analysis is used to calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature marker, and a coherence spectrum representing the interaction strength between the torque mutation and the magnetic field distortion is generated according to the frequency band overlap;

[0011] According to the energy concentration area of ​​the coherence spectrum within a specific frequency band, the fault history case library of the electrical equipment is matched, and a maintenance strategy set for the superposition fault of dynamic imbalance of the rotating shaft and electromagnetic interference is output.

[0012] Optionally, the wavelet coherence analysis is used to calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature marker, and a coherence spectrum characterizing the interaction strength between the torque mutation and the magnetic field distortion is generated according to the frequency band overlap, including:

[0013] Dividing the time-frequency energy distribution in the mixed feature mark into analysis windows with adjacent windows partially overlapping at equal intervals along the time axis, and dividing the time-frequency energy distribution into time-frequency units with gradually varying widths in ascending order of frequency within the analysis windows;

[0014] Calculating the product of the torque fluctuation energy ratio and the magnetic field distortion energy ratio in the time-frequency unit, marking the area where the product value exceeds the dual-energy baseline value as a valid overlapping area, and calculating the ratio of the valid overlapping area to the total area of ​​the time-frequency unit as the frequency band overlap coefficient;

[0015] Sliding accumulation is performed on the frequency band overlap coefficients along the frequency axis direction, and during the accumulation process, the accumulation step size is dynamically adjusted according to the overlap change rate of adjacent frequency bands, and a two-dimensional distribution spectrum is formed according to the accumulation results;

[0016] The two-dimensional distribution spectrum is segmented according to a preset interaction strength classification threshold. During the segmentation process, continuous regions exceeding the interaction strength classification threshold and with a frequency span consistent with the mechanical resonance range of the electrical equipment are extracted to generate a coherent spectrum.

[0017] Optionally, cross-domain matching of the appearance time of the transient high-frequency component with the spatial orientation of the magnetic field polarity reversal feature, and generating a hybrid feature marker containing mechanical and electromagnetic coupling relationships according to the matching result, includes:

[0018] Performing mutation detection on the starting point of the sudden increase of the waveform of the transient high-frequency component to generate a mechanical signal time mark sequence including a start timestamp and a duration of the waveform;

[0019] Decomposing the spatial orientation of the magnetic field polarity reversal feature into magnetic field intensity change gradients in three orthogonal detection directions, and generating an electromagnetic signal spatial marker sequence based on the number of positive and negative reversals and amplitude differences in the gradient direction;

[0020] In each phase interval of the equipment operation cycle, the timestamp of the mechanical signal time mark sequence and the azimuth angle change trajectory of the electromagnetic signal space mark sequence are synchronously scanned. When the scanning result shows that there is a continuous azimuth angle offset and the offset direction is consistent with the radial force direction of the shaft, a cross-domain matching event is triggered;

[0021] The transient high-frequency component amplitude peak and the magnetic field polarity reversal amplitude peak in the cross-domain matching event are subjected to geometric scaling processing, and a hybrid feature marker is generated for the processed peak according to a preset mechanical and electromagnetic coupling weight coefficient.

[0022] Optionally, the step of matching the fault history case library of the electrical equipment according to the energy concentration area of ​​the coherence spectrum within a specific frequency band and outputting a maintenance strategy set for the superposition fault of the dynamic imbalance of the rotating shaft and the electromagnetic interference includes:

[0023] Locating an energy concentration region in the coherence map, and extracting a mapping relationship between an energy distribution shape in the energy concentration region and a physical motion pattern of a dynamic imbalance of a rotating shaft;

[0024] Performing similarity search in the fault history case library of the electrical equipment according to the mapping relationship to retrieve a fault case group containing the same electromagnetic interference spectrum characteristics;

[0025] Performing a correlation analysis between the shaft geometry deviation parameters and the electromagnetic shielding failure parameters on the fault case group to generate a candidate parameter set including an allowable range of calibration error and an adjustment value of the shielding layer thickness;

[0026] The candidate parameter sets are prioritized according to the equipment operating environment parameters, and a maintenance strategy set including axis calibration operation steps and electromagnetic shielding optimization paths is formed according to the sorting results.

[0027] Optionally, the multi-scale frequency band division of the non-periodic torque waveform, screening out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separating the magnetic field polarity reversal feature corresponding to the axis center offset direction from the magnetic field gradient anomaly signal includes:

[0028] Continuously intercepting the non-periodic torque waveform according to windows of different time lengths, extracting the fluctuation period of the waveform amplitude in the intercepted results, and screening out window segments with the fluctuation period less than a preset reference value as candidate high-frequency intervals;

[0029] Locking the overlapping window segments associated with the start-up and shutdown phases of the equipment within the candidate high-frequency interval, extracting the short-term fluctuation features of the amplitude mutation direction that continuously reverses within the overlapping window segments and marking them as transient high-frequency components;

[0030] synchronously calculating the gradient difference of adjacent sampling points in a single orthogonal direction for the magnetic field gradient anomaly signal, and determining that the direction is the main direction of the axis deviation when the gradient difference exceeds the sum of the differences in adjacent directions;

[0031] An alternating change region of the gradient difference sign is captured in the main direction of the axis center offset, and a magnetic field polarity reversal feature corresponding to the main direction of the axis center offset is extracted in the alternating change region.

[0032] Optionally, within each phase interval of the device operation cycle, synchronously scanning the timestamp of the mechanical signal time mark sequence and the azimuth angle change trajectory of the electromagnetic signal space mark sequence, and triggering a cross-domain matching event when the scanning result shows that there is a continuous azimuth angle offset and the offset direction is consistent with the radial force direction of the shaft, includes:

[0033] The equipment operation cycle is divided into multiple phase intervals according to the rotation angle of the rotating shaft, where each phase interval corresponds to a preset radial force direction range of the shaft;

[0034] Mapping the timestamp of the mechanical signal time mark sequence to the corresponding phase interval, and extracting the azimuth angle change trajectory of the electromagnetic signal space mark sequence in the same phase interval;

[0035] Detecting a continuous azimuth shift event in the azimuth change trajectory, and determining that the azimuth shift event is a valid continuous shift when the shift direction in the azimuth shift event coincides with the radial force direction of the shaft corresponding to the current phase interval;

[0036] A cross-domain matching event is triggered based on the effective continuous offset, and the starting position of the mechanical signal timestamp is bound to the ending position of the electromagnetic signal azimuth offset through the cross-domain matching event.

[0037] Optionally, the product of the torque fluctuation energy ratio and the magnetic field distortion energy ratio is calculated in the time-frequency unit, an area where the product value exceeds the dual-energy baseline value is marked as a valid overlapping area, and the ratio of the valid overlapping area to the total area of ​​the time-frequency unit is calculated as a frequency band overlap coefficient, including:

[0038] Delineating a maximum envelope of the torque fluctuation energy distribution within the time-frequency unit, and calculating a ratio of the energy value at each time point on the maximum envelope to the total energy of the unit as the torque fluctuation energy ratio;

[0039] Synchronously calculating the local peak density of the magnetic field distortion energy distribution within the same time-frequency unit, and taking the ratio of the local peak density to the total energy of the unit as the magnetic field distortion energy ratio;

[0040] Multiplying the torque fluctuation energy ratio and the magnetic field distortion energy ratio at each time point, screening out continuous time segments where the product value exceeds both the torque reference fluctuation energy and the magnetic field reference distortion energy and marking them as valid overlapping areas;

[0041] The percentage of the time span of the effective overlapping area to the total time length of the time-frequency unit is calculated, and the percentage is combined with the average product value in the effective overlapping area to generate a frequency band overlap coefficient.

[0042] In a second aspect, the present application provides a remote maintenance guidance system for electrical equipment, comprising:

[0043] A deployment module is configured to synchronously deploy a torque fluctuation sensor and a magnetoresistive sensor on the surface of a rotating shaft of an electrical device, wherein the torque fluctuation sensor continuously captures a non-periodic torque waveform caused by a sudden change in the mechanical load of the shaft, and the magnetoresistive sensor simultaneously detects an abnormal magnetic field gradient signal generated when the rotating shaft deviates in an orthogonal direction.

[0044] a screening module for performing multi-scale frequency band division on the non-periodic torque waveform, screening out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separating the magnetic field polarity reversal characteristics corresponding to the axis offset direction from the magnetic field gradient anomaly signal;

[0045] a matching module for performing cross-domain matching between the appearance moment of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature, and generating a hybrid feature signature containing a mechanical and electromagnetic coupling relationship based on the matching result;

[0046] a calculation module, configured to calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature marker using wavelet coherence analysis, and generate a coherence spectrum representing the interaction strength between the torque mutation and the magnetic field distortion according to the frequency band overlap;

[0047] The output module is used to match the fault history case library of the electrical equipment according to the energy concentration area of ​​the coherence spectrum within a specific frequency band, and output a maintenance strategy set for the dynamic imbalance of the rotating shaft and the electromagnetic interference superposition fault.

[0048] In a third aspect, an embodiment of the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a remote maintenance guidance method for electrical equipment as described in the first aspect above.

[0049] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, it implements a remote maintenance guidance method for electrical equipment as described in the first aspect.

[0050] In an embodiment of the present application, a torque fluctuation sensor and a magnetoresistive sensor are synchronously deployed on the surface of a rotating shaft of an electrical device. The torque fluctuation sensor continuously captures a non-periodic torque waveform caused by a sudden change in the mechanical load of the shaft, while the magnetoresistive sensor is used to detect an abnormal magnetic field gradient signal generated when the rotating shaft is offset in an orthogonal direction. The non-periodic torque waveform is divided into multi-scale frequency bands to screen out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and the magnetic field polarity reversal feature corresponding to the axis offset direction is simultaneously separated from the abnormal magnetic field gradient signal. The occurrence time of the transient high-frequency component is cross-domain matched with the spatial orientation of the magnetic field polarity reversal feature, and a hybrid feature signature containing mechanical and electromagnetic coupling relationships is generated based on the matching results. Wavelet coherence analysis is used to calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature signature, and a coherence map representing the interaction strength between the torque sudden change and the magnetic field distortion is generated based on the frequency band overlap. The energy concentration area of ​​the coherence map within a specific frequency band is matched to the fault history case library of the electrical device, and a maintenance strategy set for the superposition of dynamic imbalance and electromagnetic interference faults of the rotating shaft is output.

[0051] The technical solution of this application has the following beneficial effects:

[0052] Through the synchronous acquisition of torque and magnetic field dual-modal signals, the collaborative capture of mechanical load mutations and shaft offset dynamic characteristics is achieved, and the spatiotemporal resolution of composite fault feature extraction is improved; based on multi-scale frequency band division, the transient high-frequency mechanical impact during the start-up and shutdown phases of the equipment is accurately located, and combined with the separation of magnetic field polarity reversal characteristics, the sensitivity to the direction of shaft center offset is enhanced; through cross-domain spatiotemporal matching of mechanical and electromagnetic signals, a coupling relationship model between torque fluctuations and magnetic field distortion is constructed to eliminate the misjudgment interference of single signal analysis; using the time-frequency energy overlap calculation and dynamic accumulation mechanism, the interaction intensity between torque and magnetic field is quantified, and the fault coupling frequency band within the mechanical resonance frequency band is accurately locked; based on the mapping and matching of energy concentration areas and historical fault cases, a targeted maintenance strategy for dynamic imbalance and electromagnetic interference superposition faults is generated to improve the targeted operation and maintenance decision-making.

[0053] Furthermore, the time-frequency energy of the mixed feature markers is divided into adjacent overlapping analysis windows at equal intervals along the time axis, and within each window, it is divided into time-frequency units of gradually varying widths according to increasing frequency. The product of the torque fluctuation and magnetic field distortion energy ratios within each unit is calculated, and the regions exceeding the dual energy baseline are screened and marked as valid overlapping regions, and the ratios are statistically calculated to generate the frequency band overlap coefficient. The step size sliding accumulation coefficient is dynamically adjusted along the frequency axis to form a two-dimensional spectrum, and the continuous regions exceeding the intensity threshold and covering the mechanical resonance range are extracted to generate a coherent spectrum. Through the time-frequency energy overlap analysis and dynamic accumulation mechanism, the misjudgment problem caused by the aliasing of mechanical vibration and electromagnetic noise frequency bands in traditional methods is solved, the interactive frequency bands of torque mutation and magnetic field distortion are accurately identified, and the recognizability of composite fault features within the resonance frequency band is significantly improved, providing a highly reliable basis for maintenance strategy generation.

[0054] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 A flowchart of a remote maintenance guidance method for electrical equipment provided by the present application is shown;

[0057] Figure 2 A schematic diagram of the structure of a remote maintenance guidance system for electrical equipment provided by the present application is shown;

[0058] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown.

[0059] Figure 4 A scenario diagram of a remote maintenance guidance method for electrical equipment provided in this application is shown. DETAILED DESCRIPTION

[0060] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0061] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0062] Among existing rotating equipment bearing health monitoring technologies, online diagnostic systems based on vibration spectrum analysis face two core bottlenecks: First, a single vibration signal cannot effectively distinguish between mechanical bearing damage and abnormal characteristics in the same frequency band caused by electromagnetic interference from the motor. This is especially true under variable frequency drive conditions, where the spectral aliasing of electromagnetic noise and the bearing fault vibration frequency band significantly increases the misjudgment rate. Second, the vibration signal's sluggish response to progressive bearing damage, such as early microcracks and stress concentration, makes it difficult to capture the subtle energy release characteristics of the damage initiation stage, causing fault warning timing to lag behind the actual damage evolution process. These shortcomings stem from traditional methods' neglect of mechanical-electromagnetic coupling effects and the inherent limitations of single-dimensional signal sensitivity.

[0063] In response to the above problems, this application proposes a remote maintenance guidance method for electrical equipment, which constructs a mechanical-electromagnetic coupling feature analysis model by synchronously collecting the torque fluctuation of the rotating shaft and the three-dimensional magnetic field gradient signal. Specifically, the torque fluctuation sensor is used to capture the non-periodic mechanical shock waveform caused by the sudden change of the load, and the magnetic field distortion component caused by the shaft offset is detected orthogonally by the magnetoresistive sensor; the coupling action mode of mechanical damage and electromagnetic interference is separated by the spatiotemporal matching of the transient high-frequency mechanical shock and the magnetic field polarity reversal characteristics; further based on the time-frequency energy interaction intensity analysis, the fault characteristic frequency band within the resonant frequency band is located, and the historical fault cases are associated to generate a targeted maintenance strategy. This method breaks through the dimensional limitations of a single vibration signal, and through the collaborative analysis of mechanical-electromagnetic dual-modal data, it effectively distinguishes the interference in the same frequency band from the real damage characteristics, while enhancing the ability to capture early weak faults, thereby significantly reducing the false alarm rate and realizing the pre-emptive fault warning.

[0064] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0065] Figure 1 A flowchart of a method for remote maintenance guidance of electrical equipment is provided in an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0066] 101. Synchronously deploy a torque fluctuation sensor and a magnetoresistive sensor on the surface of a rotating shaft of an electrical device, wherein the torque fluctuation sensor continuously captures the non-periodic torque waveform caused by a sudden change in the mechanical load of the shaft, and the magnetoresistive sensor simultaneously detects the abnormal magnetic field gradient signal generated when the rotating shaft deviates in an orthogonal direction;

[0067] In the above scheme, the torque fluctuation sensor refers to the measurement of the instantaneous change in torque caused by the sudden change of mechanical load on the surface of the rotating shaft through a strain gauge or a piezoelectric ceramic array. The non-periodic torque waveform it outputs contains the transient impact characteristics of the shaft under abnormal working conditions such as starting, stopping, and jamming. The magnetoresistive sensor refers to a magnetic field detection unit arranged in orthogonal directions based on the principle of magnetoresistive effect, which is used to capture the abnormal magnetic field intensity gradient signal caused by the change of the axis position when the rotating shaft is offset. The non-periodic torque waveform refers to a sudden torque fluctuation signal that has no integer multiple relationship with the steady-state operating frequency of the equipment, and its amplitude mutation is directly related to the equipment load disturbance. The magnetic field gradient abnormality signal refers to an abnormal state in which the difference in the rate of change of the magnetic field intensity in the orthogonal direction at the same detection point exceeds a preset threshold, and is used to characterize the vector characteristics of the axis offset.

[0068] In the embodiment of the present application, first, a torque fluctuation sensor is installed at a circumferentially symmetrical position on the surface of the rotating shaft. The sensor has a built-in high-sampling rate strain signal acquisition module with a sampling rate of >10kHz. When the shaft body is torsionally deformed due to a sudden load change, the sensor generates a torque waveform sequence with a time resolution of 0.1ms by measuring the accumulated surface micro-strain and marks the mechanical impact timestamp corresponding to each peak. At the same time, a three-axis magnetoresistive sensor group is radially installed on the flange at the end of the shaft, with its X / Y axis parallel to the radial plane of the rotating shaft and the Z axis coincident with the axial direction. When the axis is offset, the sensor group calculates the magnetic field distortion rate of each axial direction in real time by comparing the magnetic field intensity change gradient (ΔB / Δt) in the orthogonal direction. When the distortion rate in any direction exceeds 3 times the standard deviation of the historical mean, the magnetic field gradient anomaly signal capture is triggered and its spatial pointing angle is recorded.

[0069] Secondly, a hardware synchronization trigger circuit locks the sampling clocks of the torque fluctuation sensor and the magnetoresistive sensor to the same time base, ensuring that the timestamps of the two signals are strictly aligned with the shaft's rotational angular displacement. The non-periodic torque waveform captured by the torque fluctuation sensor is filtered through a 20-2000Hz bandpass filter, and waveform segments with amplitude fluctuations exceeding the rated torque are extracted. The magnetic field gradient anomaly signal output by the magnetoresistive sensor is synthesized using space vectors, projecting the three-axis magnetic field distortion rate onto the radial plane of the rotating axis to generate a two-dimensional magnetic field gradient vector diagram with the axis as the origin.

[0070] Finally, the filtered torque waveform segments are dynamically matched with the two-dimensional magnetic field gradient vector diagram according to the timestamp. When the sudden increase in the amplitude of the torque waveform and the sudden change in the direction of the magnetic field gradient vector coincide within the ±5ms time window at a certain moment, it is determined to be a valid mechanical and electromagnetic coupling event. The torque waveform segments and magnetic field gradient vector diagram within the time window are packaged as original feature data blocks for subsequent cross-domain analysis.

[0071] 102. Perform multi-scale frequency band division on the non-periodic torque waveform to filter out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separate the magnetic field polarity reversal feature corresponding to the axis offset direction from the magnetic field gradient anomaly signal;

[0072] Optionally, step 102 may specifically include the following steps:

[0073] 1021. Continuously intercept the non-periodic torque waveform according to windows of different time lengths, extract the fluctuation period of the waveform amplitude in the intercepted results, and select window segments with the fluctuation period less than a preset reference value as candidate high-frequency intervals;

[0074] 1022. Lock the overlapping window segments associated with the start-up and shutdown phases of the equipment within the candidate high-frequency interval, extract the short-term fluctuation features of the amplitude mutation direction that continuously reverses within the overlapping window segments, and mark them as transient high-frequency components;

[0075] 1023. Synchronously calculate the gradient difference of adjacent sampling points in a single orthogonal direction for the magnetic field gradient anomaly signal, and determine that the direction is the main direction of the axis deviation when the gradient difference exceeds the sum of the differences in adjacent directions;

[0076] 1024. Capture an alternating change region of the gradient difference sign in the main direction of the axis center offset, and extract a magnetic field polarity reversal feature corresponding to the main direction of the axis center offset in the alternating change region.

[0077] In the above scheme, multi-scale frequency band division refers to intercepting non-periodic torque waveforms through windows of different time lengths to capture high-frequency fluctuation characteristics at different time scales. The candidate high-frequency interval refers to a window segment with a fluctuation period shorter than a preset reference value, which is used to characterize transient mechanical shocks during the start-up and shutdown stages of the equipment. The overlapping window segment refers to the interval where the time ranges intersect in the interception results of multiple time windows, reflecting abnormal fluctuations synchronized with the start-up and shutdown timing of the equipment. The main direction of the axis deviation refers to the orthogonal direction with the largest gradient difference in the magnetic field gradient anomaly signal, which characterizes the vector principal component of the axis deviation. The magnetic field polarity reversal feature refers to the area where the sign of the magnetic field gradient difference alternates in the main direction of the axis deviation, reflecting the magnetic field distortion feature of the axis deviation trajectory.

[0078] In an embodiment of the present application, first, through step 1021, a sliding window mechanism is used to perform multi-scale continuous interception for the non-periodic torque waveform: multiple fixed-length windows are set, and waveform segments are intercepted along the time axis at an overlap rate of 0.5. The mean time interval between adjacent peaks of the waveform segments in each fixed-length window is calculated as the fluctuation period, and the window segments with a fluctuation period less than a preset reference value (for example, 20ms) are screened and marked as candidate high-frequency intervals. This process captures transient fluctuations through short windows and suppresses random noise through long windows. For example, in the motor startup phase, a 100ms sub-interval with a dense fluctuation period is screened out from the waveform intercepted by the 500ms window, and the transient high-frequency impact caused by bearing ball jamming is identified.

[0079] Subsequently, step 1022 is used to superimpose the start and stop timestamps of the equipment operation log records in the candidate high-frequency interval, extract the window segments overlapping with the start and stop phase time, and detect the sudden change direction of the waveform amplitude in these window segments: if the amplitude first increases in a positive direction and then decreases in a reverse direction between two consecutive peaks, for example, it jumps from +15N·m to +30N·m and then falls back to +10N·m, and the sudden change interval is less than the preset threshold (for example, 5ms), it is determined that the amplitude direction is continuously reversed, and this short-term fluctuation feature is marked as a transient high-frequency component - for example, the high-frequency torque oscillation caused by the loosening of the bearing seat when the fan is shut down, its amplitude undergoes two reverse sudden changes within 10ms and is accurately captured.

[0080] When processing the magnetic field gradient anomaly signal in step 1023, the magnetic field intensity difference between adjacent sampling points in the three orthogonal directions X / Y / Z is calculated, for example, ΔB between point t+1 and point t. The absolute value of the magnetic field intensity difference in each direction, |ΔB|, is summed. If the sum of the absolute values ​​of the differences in one direction significantly exceeds the sum of the other two directions, for example, the sum of the differences in the X axis is 1.5 times the sum of the differences in the Y and Z axes, then that direction is determined to be the main direction of axis deviation. For example, when the motor bearing wear causes the axis to deviate along the X axis, the magnetic field gradient difference in the X axis is continuously higher than that in the Y / Z axes.

[0081] Finally, the sign changes of the gradient difference in the main direction of the axis deviation are scanned: if the difference signs of three consecutive sampling points are "+, -, +" or "-, +, -" in sequence, they are marked as areas of alternating sign changes, and the number of sign reversals in these areas is counted. The continuous segments where the number of reversals exceeds the threshold are extracted as the characteristics of the magnetic field polarity reversal. For example, when the axis oscillates periodically, the X-axis magnetic field gradient difference has four sign reversals within 20ms, exceeding the set threshold of three times, which characterizes the reciprocating characteristics of the axis deviation trajectory.

[0082] In practical applications, in the health monitoring scenario of industrial motor bearings driven by variable frequency drives, step 1021 sets three windows of 100ms, 200ms, and 500ms to perform sliding interception of the non-periodic torque waveform. During the motor acceleration phase, a candidate high-frequency interval with a fluctuation period of 18ms caused by cracks in the bearing outer ring is captured. Step 1022 combines the start and stop command timestamps of the motor control system, locks the 200ms window segment overlapping with this phase, and extracts the amplitude that suddenly increases from +20N·m within 4ms. The bidirectional mutation waveform that suddenly drops to +35 N·m and then drops to +5 N·m is marked as a transient high-frequency component; step 1023 calculates the magnetic field gradient signal and finds that the sum of the Y-axis gradient differences reaches twice the sum of the X / Z axes, and determines that the main direction of the axis offset is the Y-axis; step 1024 identifies the continuous sign reversal of "+0.2T, -0.3T, +0.4T, -0.1T" in the Y-axis gradient sequence, extracts this segment as the magnetic field polarity reversal feature, and accurately associates it with the periodic axis swing caused by the crack in the bearing outer ring.

[0083] The overall solution in step 102 above effectively separates transient high-frequency mechanical shock signals during the equipment's startup and shutdown phases through multi-scale window extraction and dynamic filtering of fluctuation cycles, overcoming the traditional method's drawback of high-frequency features being overwhelmed by steady-state noise. Combined with the determination of the main direction of magnetic field gradient differences and the extraction of polarity reversal features, this method accurately analyzes the spatial directional characteristics of the axis offset trajectory. This method transcends the limitations of a single signal dimension. Through dual verification of mechanical shock timing and magnetic field distortion orientation, it significantly distinguishes between bearing mechanical damage and anomalies in the same frequency band caused by electromagnetic interference, improving the accuracy and early warning capabilities of complex fault diagnosis, and providing highly reliable feature evidence for remote maintenance.

[0084] 103. Perform cross-domain matching on the appearance time of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature, and generate a hybrid feature marker including mechanical and electromagnetic coupling relationships according to the matching result;

[0085] Optionally, step 103 may specifically include the following steps:

[0086] 1031. Perform mutation detection on the starting point of the sudden increase of the waveform of the transient high-frequency component to generate a mechanical signal time mark sequence including a start timestamp and a duration of the waveform;

[0087] 1032. Decompose the spatial orientation of the magnetic field polarity reversal feature into magnetic field intensity change gradients in three orthogonal detection directions, and generate an electromagnetic signal spatial marker sequence based on the number of positive and negative reversals and amplitude differences in the gradient directions;

[0088] 1033. In each phase interval of the equipment operation cycle, synchronously scan the timestamp of the mechanical signal time mark sequence and the azimuth angle change trajectory of the electromagnetic signal space mark sequence. When the scanning result shows that there is a continuous azimuth angle offset and the offset direction is consistent with the radial force direction of the shaft, trigger a cross-domain matching event.

[0089] Among them, step 1033 may specifically include the following processes: dividing the equipment operation cycle into multiple phase intervals according to the rotation angle of the rotating shaft, wherein each phase interval corresponds to a preset radial force direction range of the shaft; mapping the timestamp of the mechanical signal time mark sequence to the corresponding phase interval, and extracting the azimuth change trajectory of the electromagnetic signal space mark sequence in the same phase interval; detecting continuous azimuth offset events in the azimuth change trajectory, and determining it as a valid continuous offset when the offset direction in the azimuth offset event coincides with the radial force direction of the shaft corresponding to the current phase interval; triggering a cross-domain matching event based on the valid continuous offset, and binding the starting position of the mechanical signal timestamp and the ending position of the electromagnetic signal azimuth offset through the cross-domain matching event.

[0090] 1034. Perform geometric scaling processing on the transient high-frequency component amplitude peak and the magnetic field polarity reversal amplitude peak in the cross-domain matching event, and generate a hybrid feature marker for the processed peak according to a preset mechanical and electromagnetic coupling weight coefficient.

[0091] In the above scheme, the mechanical signal time mark sequence refers to the sequence data containing timestamps and durations generated by detecting the starting point of the sudden increase in the waveform of the transient high-frequency component, which is used to characterize the timing characteristics of the mechanical impact event. The electromagnetic signal space mark sequence refers to the sequence data that converts the gradient change of the magnetic field polarity reversal feature in the orthogonal direction into spatial azimuth angle and reversal amplitude, reflecting the vector trajectory of the axis offset. The phase interval refers to the fixed interval angle range divided according to the rotation angle of the rotating shaft, which is used to synchronize the spatiotemporal characteristics of the mechanical and electromagnetic signals. The cross-domain matching event refers to the association event in which the mechanical signal timestamp and the electromagnetic signal azimuth angle trajectory meet the continuous offset condition within the phase interval. The hybrid feature label refers to the composite feature label generated by fusing the peak parameters of the mechanical and electromagnetic signals, which includes the coupling strength and spatial correlation of the two types of signals.

[0092] In an embodiment of the present application, first, through step 1031, a differential mutation detection algorithm is used to locate the starting point of the sudden increase for the transient high-frequency component waveform: the absolute value of the amplitude difference between adjacent sampling points is calculated. When three consecutive difference values ​​exceed a preset multiple of the baseline amplitude, for example, when the baseline is 5 N·m and the difference value exceeds 15 N·m, the first exceeding limit point is marked as the starting timestamp, and the time span from this point to the amplitude falling back to the baseline is recorded as the waveform duration, and a mechanical signal time mark sequence is generated - for example, in a motor bearing ball peeling fault, the amplitude is detected to suddenly increase from +10 N·m to +35 N·m within 0.2 ms, lasting for 3 ms, and a timestamp T = 15.3s is generated.

[0093] Secondly, step 1032 decomposes the spatial orientation of the magnetic field polarity reversal feature into orthogonal detection directions: the number of positive and negative sign reversals of the magnetic field gradient difference sequence in the three directions of X / Y / Z is counted respectively, and the sum of the absolute values ​​of the amplitude differences between adjacent reversal points is calculated. For example, the X-axis sequence "+0.3T, -0.2T, +0.4T" is reversed twice, and the sum of the differences is 0.3+0.2+0.4=0.9T. The direction with the most reversals and the largest sum of differences is selected as the main direction, and an electromagnetic signal spatial marker sequence containing azimuth angle (e.g., 30° on the X-axis) and reversal amplitude parameters is generated.

[0094] Subsequently, step 1033 divides the equipment's operating cycle into phase intervals: Each 30° rotation of the rotating shaft is considered a phase interval. The timestamps of the mechanical signal time-stamp sequence are mapped to the corresponding phase interval, and the azimuth change trajectory of the electromagnetic signal space-stamp sequence within this interval is extracted. If, within a phase interval, the azimuth corresponding to the mechanical timestamp shifts by more than 10° within three consecutive sampling points, the azimuth change trajectory is scanned, for example, from 40° to 50° to 55°, and the shift direction aligns with the radial force direction of the shaft within that interval (for example, the 30°-60° interval corresponds to the positive Y-axis direction), triggering a cross-domain matching event. For example, when a wind turbine bearing wears, the electromagnetic azimuth corresponding to the mechanical impact timestamp T = 20.1s shifts continuously from 45° to 60°, matching the force direction of the current interval and triggering an event.

[0095] Finally, step 1034 normalizes the peak mechanical amplitude (e.g., +28 N·m) and the peak magnetic field reversal amplitude (e.g., -0.5 T) in the cross-domain matching event: the mechanical amplitude is scaled to the 0-1 range by a ratio of 1:0.6, and the magnetic field amplitude is scaled by a ratio of 1:0.4. These are then weighted and summed according to a preset weight coefficient, and a hybrid signature is obtained based on the sum and the parameter signs. For example, if the sum of the scaled parameters is 0.828, the hybrid signature is labeled "ME-0.828," representing the strength of the mechanical and electromagnetic coupling.

[0096] In practical applications, in the bearing health assessment of industrial electrical equipment, step 1031 detects the transient high-frequency component caused by the bearing retainer fracture, with a starting timestamp of T = 45.2s and a duration of 5ms. Step 1032 decomposes the Z-axis magnetic field gradient into four reversals with a total difference of 1.5T, generating an electromagnetic marker with an azimuth angle of 120° on the Z axis. Step 1033 scans the phase interval 120°-150° and finds that the electromagnetic azimuth angle corresponding to the timestamp of T = 45.2s continuously shifts from 125° to 140°, which is consistent with the radial force direction of the current interval (negative direction of the X axis), triggering a cross-domain matching event. Step 1034 fuses the mechanical peak value of +40N·m and the magnetic field peak value of -0.8T by weight to generate the marker "ME-0.89", accurately associating the retainer fracture with the shaft eccentricity fault.

[0097] The overall solution of step 103 above, through cross-domain matching of mechanical impact timing and electromagnetic orientation trajectory, breaks through the spatiotemporal limitations of single signal analysis and accurately captures the coupling characteristics of bearing damage and axial offset; based on phase interval force direction consistency verification and dual-modal peak weight fusion, it effectively distinguishes mechanical faults from co-frequency anomalies caused by electromagnetic interference, significantly improving the accuracy of complex fault diagnosis and the credibility of maintenance decisions, and providing a highly robust collaborative analysis paradigm for rotating equipment health management.

[0098] 104. Calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature marker using wavelet coherence analysis, and generate a coherence spectrum representing the interaction intensity between the torque mutation and the magnetic field distortion according to the frequency band overlap;

[0099] Optionally, step 104 may specifically include the following steps:

[0100] 1041. Divide the time-frequency energy distribution in the mixed feature mark into analysis windows with overlapping adjacent windows at equal intervals along the time axis, and divide the time-frequency energy distribution into time-frequency units with gradually varying widths in ascending order of frequency within the analysis windows;

[0101] 1042. Calculate the product of the torque fluctuation energy ratio and the magnetic field distortion energy ratio in the time-frequency unit, mark the area where the product value exceeds the dual-energy baseline value as a valid overlapping area, and calculate the ratio of the valid overlapping area to the total area of ​​the time-frequency unit as the frequency band overlap coefficient;

[0102] Among them, step 1042 may specifically include the following processes: defining the maximum envelope of the torque fluctuation energy distribution within the time-frequency unit, and calculating the ratio of the energy value at each time point on the maximum envelope to the total energy of the unit as the torque fluctuation energy ratio; synchronously calculating the local peak density of the magnetic field distortion energy distribution within the same time-frequency unit, and taking the ratio of the local peak density to the total energy of the unit as the magnetic field distortion energy ratio; multiplying the torque fluctuation energy ratio and the magnetic field distortion energy ratio at each time point, screening continuous time segments whose product values ​​exceed both the torque reference fluctuation energy and the magnetic field reference distortion energy and marking them as valid overlapping areas; counting the percentage of the time span of the valid overlapping area to the total time length of the time-frequency unit, and combining the percentage with the average product value in the valid overlapping area to generate a frequency band overlap coefficient.

[0103] 1043. Slidingly accumulate the frequency band overlap coefficients along the frequency axis, dynamically adjusting the accumulation step size according to the rate of change of overlap of adjacent frequency bands during the accumulation process, and forming a two-dimensional distribution map based on the accumulation results;

[0104] 1044. Perform regional segmentation on the two-dimensional distribution spectrum according to a preset interaction intensity grading threshold. During the segmentation process, continuous regions exceeding the interaction intensity grading threshold and having a frequency span that meets the mechanical resonance range of the electrical equipment are extracted to generate a coherent spectrum.

[0105] In the above scheme, the time-frequency energy distribution refers to the distribution state of mechanical and electromagnetic energy in the mixed feature marker on the time-frequency two-dimensional plane. The analysis window refers to a continuous time period with partial overlap divided along the time axis, which is used for local energy feature extraction. The time-frequency unit refers to a rectangular or trapezoidal area divided by increasing frequency within the analysis window, and its width gradually narrows as the frequency increases. The frequency band overlap coefficient refers to a quantitative indicator of the interaction intensity between torque and magnetic field energy within the time-frequency unit. The dual energy baseline value is the historical statistical mean of torque fluctuation and magnetic field distortion energy, which is used to screen effective interaction areas. The two-dimensional distribution spectrum refers to the frequency and interaction intensity relationship spectrum generated by accumulating the frequency band overlap coefficient. The mechanical resonance range of electrical equipment refers to the natural vibration frequency range of the equipment bearing, which is used to constrain the effective frequency band of fault characteristics.

[0106] In an embodiment of the present application, first, the time-frequency energy distribution in the mixed feature marker is windowed and segmented through step 1041: the analysis windows are equally spaced along the time axis at fixed intervals (for example, 100ms), and some overlapping areas are retained between adjacent windows to ensure time continuity, for example, 30% of the end of the window overlaps with the starting end of the next window; within each analysis window, the time-frequency units are divided in ascending order of frequency, where a wide band is used in the low-frequency band and a narrow band is used in the high-frequency band to form a time-frequency unit grid with a gradually varying width - for example, in motor bearing fault monitoring, the time-frequency energy distribution is divided into 100ms windows, and is divided into 50Hz-wide units below 500Hz and 20Hz-wide units above 1500Hz to enhance the resolution accuracy of high-frequency transient features.

[0107] Secondly, the energy interaction intensity is calculated in each time-frequency unit through step 1042: the proportion of the integral value of the amplitude of the torque fluctuation energy in the unit to the total energy is extracted, and the peak density proportion of the magnetic field distortion energy is simultaneously calculated, and the two are multiplied to obtain the product value; the dual energy baseline value is set to the average product value under historical normal working conditions (for example, 0.2), and the area where the product value exceeds the baseline is screened, and the proportion of the time-frequency unit area covered by it is counted as the frequency band overlap coefficient - for example, in the fan bearing damage scenario, the torque accounts for 65% and the magnetic field accounts for 45% in a certain time-frequency unit, and the product value of 0.29 exceeds the baseline of 0.2, and the marked effective area accounts for 75%.

[0108] Subsequently, step 1043 is used to dynamically accumulate the overlap coefficients along the frequency axis: starting from the starting point of the low-frequency band, the overlap coefficients of the adjacent frequency bands are accumulated by sliding with an initial step size (e.g., 50 Hz). If the rate of change of the adjacent frequency band coefficients exceeds a threshold, the step size is reduced (e.g., from 50 Hz to 25 Hz) to improve the resolution. After accumulation, a two-dimensional distribution map is drawn with the frequency as the horizontal axis and the accumulated value as the vertical axis. For example, in the 800-1200 Hz frequency band, the accumulated value increases from 0.5 to 0.9. Because the rate of change exceeds the threshold, the step size is adjusted to 25 Hz to accurately capture the energy concentration trend of the resonant frequency band.

[0109] Finally, step 1044 is used to extract a coherence spectrum based on the preset intensity threshold and resonance interval constraints: multi-level intensity thresholds are set in the two-dimensional distribution spectrum (for example, a high intensity threshold of 0.8 and a medium intensity threshold of 0.6), and a connected region detection algorithm is used to mark continuous frequency bands whose cumulative values ​​exceed the thresholds; the marked continuous frequency bands are compared with the equipment bearing natural vibration frequency database, and only the frequency bands in the continuous frequency bands whose frequency spans completely fall within the resonance interval are retained; when the coherence spectrum is generated, the frequency bands that meet the dual conditions of intensity and frequency are color-coded according to the cumulative values, with red indicating high-intensity interaction and blue indicating low intensity, forming a coherence spectrum with frequency and intensity as coordinate axes.

[0110] In practical applications, taking the ball wear monitoring of electrical equipment bearings as an example, step 1041 divides the analysis window into 80ms intervals and splits the time-frequency units; step 1042 calculates that the torque accounts for 68% and the magnetic field accounts for 42% in the 600Hz unit, the product value is 0.286 (baseline 0.2), and the effective area accounts for 78%; step 1043 accumulates along the frequency axis to 800Hz, and the coefficient reaches 0.88. Due to the excessive rate of change, the step size is adjusted from 50Hz to 30Hz; step 1044 extracts the area with the accumulated value exceeding 0.85 in the 600-900Hz frequency band (the resonance range of electrical equipment bearings) to generate a coherent spectrum, accurately correlating the coupled faults of ball wear and axial offset.

[0111] The overall solution of step 104 above enhances the local feature analysis capability of high-frequency mechanical shock and magnetic field distortion through time-frequency window segmentation and gradient unit division; quantifies the interaction strength of fault-sensitive frequency bands and suppresses environmental noise interference based on dual energy product screening and dynamic step accumulation mechanism; and extracts coherent spectra by combining resonance interval constraints, significantly improving the recognizability of bearing composite fault characteristics in the time-frequency domain, providing an accurate frequency domain decision-making basis for remote maintenance, and effectively solving the problems of high-frequency feature submersion and resonant frequency band misjudgment in traditional methods.

[0112] 105. According to the energy concentration area of ​​the coherent spectrum within a specific frequency band, match the fault history case library of the electrical equipment and output a maintenance strategy set for the dynamic imbalance of the rotating shaft and the electromagnetic interference superposition fault.

[0113] Optionally, the step of matching the fault history case library of the electrical equipment according to the energy concentration area of ​​the coherence spectrum within a specific frequency band and outputting a maintenance strategy set for the superposition fault of the dynamic imbalance of the rotating shaft and the electromagnetic interference includes:

[0114] 1051. Locate an energy concentration region in the coherence spectrum, and extract a mapping relationship between the energy distribution shape in the energy concentration region and the physical motion pattern of the dynamic imbalance of the rotating shaft;

[0115] 1052. Perform similarity search in the fault history case library of the electrical equipment according to the mapping relationship to retrieve a group of fault cases containing the same electromagnetic interference spectrum characteristics;

[0116] 1053. Perform correlation analysis between the shaft geometry deviation parameters and the electromagnetic shielding failure parameters on the fault case group to generate a candidate parameter set including an allowable range of calibration error and a shielding layer thickness adjustment value;

[0117] 1054. Prioritize the candidate parameter sets according to the equipment operating environment parameters, and form a maintenance strategy set including axis calibration operation steps and electromagnetic shielding optimization paths according to the ranking results.

[0118] In the above scheme, the energy concentration area refers to the continuous frequency band in the coherence spectrum where the interaction intensity value exceeds the preset threshold, and its energy distribution shape includes characteristics such as peak distribution shape, bandwidth range and intensity gradient. The mapping relationship refers to the correspondence rule between the energy distribution shape and the physical mode of the dynamic imbalance of the shaft. The fault case group refers to the set of fault records with similar energy distribution shapes and electromagnetic spectrum characteristics in the historical database. The candidate parameter set refers to the combination of shaft calibration and electromagnetic shielding optimization parameters obtained through correlation analysis. The maintenance strategy set refers to the maintenance operation sequence and parameter adjustment plan adapted according to the environmental parameters.

[0119] In an embodiment of the present application, first, when locating the energy concentration area in step 1051, a multi-level threshold segmentation algorithm is adopted: high, medium and low interaction intensity thresholds are set in the coherence spectrum, and the cumulative value curve is scanned along the frequency axis, and the frequency band that continuously exceeds the high threshold is marked as the core energy area, and the frequency band between the medium and high thresholds is marked as the edge energy area; the peak distribution shape, bandwidth range and intensity gradient of the core energy area are extracted, and shape matching is performed with the dynamic imbalance mode database of the rotating shaft. The database contains energy distribution templates of modes such as eccentricity, bending, and resonance, and finally a mapping relationship between the energy distribution shape and the physical motion mode is established - for example, in the coherence spectrum of a motor bearing, the 1200-1400Hz frequency band presents a double peak shape and a bandwidth of 200Hz, which is matched to a composite mode of shaft eccentricity and superimposed resonance.

[0120] Secondly, similarity retrieval is performed based on the mapping relationship through step 1052: the energy distribution shape characteristics (such as bimodal shape and 200 Hz bandwidth) are compared with the electromagnetic interference spectrum characteristics in the fault history case library in a multi-dimensional manner, the peak position deviation and the bandwidth difference are calculated as the shape similarity, and the harmonic component overlap is calculated as the spectrum similarity, and the fault case groups that meet both the shape similarity and spectrum similarity conditions are screened out - for example, 8 groups of historical cases are retrieved, of which 3 groups have the same second harmonic characteristics in the X-axis magnetic field spectrum.

[0121] Subsequently, when performing parameter correlation analysis on the fault case group through step 1053, a multivariate regression model is used to analyze the statistical correlation between the shaft geometric deviation parameters and the electromagnetic shielding failure parameters: extract the combined data of the shaft deviation value and the shielding parameter in the historical case, calculate the covariance matrix of the two when the fault occurs, screen out the parameter pairs whose absolute value of the covariance exceeds the threshold, and determine the compensation adjustment amount of the shielding parameter when the deviation parameter increases by one unit. For example, the shielding layer needs to be thickened by 0.3mm for every 0.1mm increase in eccentricity. When generating the candidate parameter set based on this, the allowable range of the calibration error is set to the mean ± standard deviation of the deviation parameter when the fault is resolved in the historical case, and the shielding layer thickness adjustment value is an integer multiple of the compensation amount in the correlation model.

[0122] Finally, step 1054 optimizes the strategy priority according to the equipment operating environment parameters, and the priority ranking is achieved through weighted scoring of the environmental parameters: a scoring matrix including load rate, humidity, and temperature environmental factors is constructed, and the environmental adaptability score is calculated for each parameter combination in the candidate parameter set, and finally ranked by the total score. For example, when the load rate is greater than 70%, the calibration weight is set to 0.6 and the shielding weight is 0.4. When the humidity is greater than 50%, the shielding weight is increased to 0.7. When the maintenance strategy set is generated, the parameter combination with the highest score is decomposed into operation steps, and the constraints of the operation sequence are marked. For example, the highest-scoring "calibration 0.25mm + shielding 3.5mm" is decomposed into "the first step is to laser calibrate the shaft to within 0.25mm error, and the second step is to install a 3.5mm silicon steel shielding layer on the inside of the motor end cover."

[0123] In practical applications, taking the monitoring of shaft center offset and electromagnetic interference superposition faults caused by cracks in the outer ring of the bearing of a motor rotating equipment as an example, step 1051 locates the double-peak energy concentration area in the 1200-1500Hz frequency band in the coherence spectrum, mapping it to the shaft eccentricity and electromagnetic harmonic resonance coupling mode caused by the outer ring crack; step 1052 matches 8 groups of fault records with the same X-axis magnetic field second harmonic characteristics in the historical case library; step 1053 analyzes and obtains the associated parameters that a shaft eccentricity of 0.2-0.4mm requires a shielding layer thickness of 3.2-3.8mm; step 1054 generates a maintenance strategy set of "calibrating the eccentricity to 0.25mm and thickening the shielding layer to 3.5mm" based on the motor load rate of 90% and the ambient temperature of 65°C, and prioritizes the axial calibration operation to suppress dynamic imbalance.

[0124] The overall solution in step 105 above quantifies the compensation relationship between mechanical and electromagnetic parameters through a parameter correlation model, breaking through traditional empirical threshold setting methods and improving the scientific nature and adaptability of candidate parameter sets. Combined with an environmental dynamic weighted scoring mechanism, this approach enables adaptive optimization of maintenance strategies based on operating conditions, resolving decision conflicts in complex scenarios such as high loads and high humidity. Through data-driven modeling and multi-objective optimization, this approach significantly improves the accuracy and execution efficiency of maintenance strategies for complex rotating shaft faults, providing a quantifiable and reusable decision-making paradigm for remote operation and maintenance of electrical equipment.

[0125] The following is a complete embodiment of steps 101 to 105:

[0126] In a fault monitoring scenario involving dynamic axis imbalance and electromagnetic interference superimposed on a crack in the outer ring of a bearing of a certain motor rotating equipment, step 101 installs torque fluctuation sensors symmetrically around the surface of the motor bearing's rotating shaft and radially deploys a three-axis magnetoresistive sensor group on the shaft end flange. When the crack in the bearing's outer ring propagates, causing a sudden change in the shaft load, the torque fluctuation sensor captures a non-periodic torque waveform with an amplitude suddenly increasing to +35 N·m. Simultaneously, the magnetoresistive sensor detects that the X-axis magnetic field gradient difference exceeds three standard deviations of the historical mean, ΔB / Δt = 0.4 T / ms, triggering abnormal signal acquisition.

[0127] In step 102, the non-periodic torque waveform is subjected to multi-scale capture in windows of 100ms, 200ms, and 500ms, and a candidate high-frequency interval with a fluctuation period of 18ms is selected. The overlapping window segments are locked in combination with the equipment start-stop log, and the bidirectional mutation waveform whose amplitude suddenly increases from +20N·m to +35N·m and then drops sharply to +5N·m within 4ms is extracted and marked as the transient high-frequency component. Simultaneously, a continuous segment in which the gradient difference sign "+, -, +" alternates three times is separated from the X-axis magnetic field gradient anomaly signal as the magnetic field polarity reversal feature of the main direction of the axis deviation.

[0128] Step 103 performs cross-domain matching on the starting timestamp T = 32.5s of the transient high-frequency component and the azimuth angle of 45° of the X-axis magnetic field polarity reversal feature. The azimuth angle continuously shifts from 40° to 55° within the phase interval of 30°-60°, triggering a matching event. The peak amplitude of +28N·m and the reversal amplitude of -0.5T are fused according to the mechanical weight of 0.7 and the electromagnetic weight of 0.3 to generate a hybrid feature label "ME-0.85".

[0129] In step 104, the time-frequency energy of the mixed feature marker is divided into gradual time-frequency units according to a 100ms window and a 30% overlap rate. The product value of 65% of the torque and 45% of the magnetic field is calculated in the 800Hz unit, which is 0.29. This product value exceeds the baseline by 0.2, and the statistically effective area accounts for 75%. After dynamically accumulating to generate a two-dimensional distribution map, the area with the cumulative value exceeding 0.8 in the 1200-1400Hz frequency band of the bearing resonance range is extracted to generate a coherent map with a bimodal morphology.

[0130] Step 105 is based on the double-peak morphology matching of 5 groups of X-axis magnetic field second harmonic fault cases in the historical case library, and the associated parameters of the eccentricity of 0.3mm requiring a shielding layer of 3.5mm are analyzed; based on the current load rate of 90% and the humidity of 60%, a maintenance strategy set of "calibrating the eccentricity to 0.25mm±0.05mm and thickening the shielding layer to 3.5mm" is generated, and the axial calibration operation is performed first.

[0131] This application achieves accurate capture of the dynamic characteristics of mechanical load mutations and axis center offsets through the coordinated acquisition of dual-modal signals of torque fluctuations and magnetic field distortion; based on multi-scale frequency band division and time-space cross-domain matching, it enhances the modeling of the coupling relationship between transient high-frequency impacts and magnetic field polarity reversal characteristics; and combines the quantification of time-frequency energy interaction intensity with dynamic retrieval of historical cases to break through the misjudgment limitations of traditional single-dimensional analysis. This method significantly improves the time-frequency domain diagnosis accuracy of complex faults such as early cracks in bearings and ball wear. Through the optimization of mechanical-electromagnetic parameter correlation and the generation of working condition adaptive strategies, it provides highly reliable and highly operational remote maintenance decision support for the health management of rotating equipment, effectively reducing the risk of sudden downtime and operation and maintenance costs.

[0132] Figure 2 The present invention provides a schematic diagram of a remote maintenance guidance system for electrical equipment. Figure 2 As shown, the system includes:

[0133] Deployment module 21, configured to simultaneously deploy a torque fluctuation sensor and a magnetoresistive sensor on the surface of a rotating shaft of an electrical device, wherein the torque fluctuation sensor continuously captures a non-periodic torque waveform caused by a sudden change in the mechanical load of the shaft, and the magnetoresistive sensor simultaneously detects an abnormal magnetic field gradient signal generated when the rotating shaft deviates in an orthogonal direction;

[0134] a screening module 22 for performing multi-scale frequency band division on the non-periodic torque waveform, screening out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separating the magnetic field polarity reversal characteristics corresponding to the axis deviation direction from the magnetic field gradient anomaly signal;

[0135] a matching module 23 for performing cross-domain matching between the occurrence moment of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature, and generating a hybrid feature signature including mechanical and electromagnetic coupling relationships based on the matching result;

[0136] a calculation module 24 for calculating the frequency band overlap of the time-frequency energy distribution in the hybrid signature using wavelet coherence analysis, and generating a coherence spectrum representing the interaction strength between the torque mutation and the magnetic field distortion according to the frequency band overlap;

[0137] The output module 25 is used to match the fault history case library of the electrical equipment according to the energy concentration area of ​​the coherence spectrum within a specific frequency band, and output a maintenance strategy set for the superposition fault of the dynamic imbalance of the rotating shaft and the electromagnetic interference.

[0138] Figure 2 The electrical equipment remote maintenance guidance system can be executed Figure 1The implementation principle and technical effects of the remote maintenance guidance method for electrical equipment described in the illustrated embodiment will not be elaborated on here. The specific manner in which each module and unit performs operations in the remote maintenance guidance system for electrical equipment in the above embodiment has been described in detail in the embodiment of the method and will not be elaborated on here.

[0139] In one possible design, Figure 2 The electrical equipment remote maintenance guidance system of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0140] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0141] The processing component 32 is used for the above Figure 1 The embodiment provides a remote maintenance guidance method for electrical equipment.

[0142] The processing component 32 may include one or more processors to execute computer instructions to perform all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0143] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0144] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0145] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0146] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0147] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0148] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The embodiment shown is a method for guiding remote maintenance of electrical equipment.

[0149] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0150] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0151] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned 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 application.

Claims

1. A remote maintenance guidance method for electrical equipment, characterized in that: include: A torque fluctuation sensor and a magnetoresistive sensor are synchronously deployed on the surface of the rotating shaft of the electrical equipment. The torque fluctuation sensor continuously captures the non-periodic torque waveform caused by the sudden change of the mechanical load on the shaft. At the same time, the magnetoresistive sensor is used to detect the abnormal magnetic field gradient signal generated when the rotating shaft deviates in an orthogonal direction. Performing multi-scale frequency band division on the non-periodic torque waveform to screen out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separating the magnetic field polarity reversal characteristics corresponding to the axis offset direction from the magnetic field gradient anomaly signal; Cross-domain matching is performed between the appearance time of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature, and a hybrid feature marker including mechanical and electromagnetic coupling relationships is generated based on the matching result; Wavelet coherence analysis is used to calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature marker, and a coherence spectrum representing the interaction strength between the torque mutation and the magnetic field distortion is generated according to the frequency band overlap; According to the energy concentration area of ​​the coherence spectrum within a specific frequency band, the fault history case library of the electrical equipment is matched, and a maintenance strategy set for the superposition fault of dynamic imbalance of the rotating shaft and electromagnetic interference is output.

2. The method according to claim 1, characterized in that The wavelet coherence analysis is used to calculate the frequency band overlap of the time-frequency energy distribution in the mixed feature mark, and a coherence spectrum representing the interaction intensity between the torque mutation and the magnetic field distortion is generated according to the frequency band overlap, including: Dividing the time-frequency energy distribution in the mixed feature mark into analysis windows with adjacent windows partially overlapping at equal intervals along the time axis, and dividing the time-frequency energy distribution into time-frequency units with gradually varying widths in ascending order of frequency within the analysis windows; Calculating the product of the torque fluctuation energy ratio and the magnetic field distortion energy ratio in the time-frequency unit, marking the area where the product value exceeds the dual-energy baseline value as a valid overlapping area, and calculating the ratio of the valid overlapping area to the total area of ​​the time-frequency unit as the frequency band overlap coefficient; Sliding accumulation is performed on the frequency band overlap coefficients along the frequency axis direction, and during the accumulation process, the accumulation step size is dynamically adjusted according to the overlap change rate of adjacent frequency bands, and a two-dimensional distribution spectrum is formed according to the accumulation results; The two-dimensional distribution spectrum is segmented according to a preset interaction strength classification threshold. During the segmentation process, continuous regions exceeding the interaction strength classification threshold and with a frequency span consistent with the mechanical resonance range of the electrical equipment are extracted to generate a coherent spectrum.

3. The method according to claim 1, characterized in that Cross-domain matching of the appearance time of the transient high-frequency component with the spatial orientation of the magnetic field polarity reversal feature, and generating a hybrid feature marker containing mechanical and electromagnetic coupling relationships according to the matching result, includes: Performing mutation detection on the starting point of the sudden increase of the waveform of the transient high-frequency component to generate a mechanical signal time mark sequence including a start timestamp and a duration of the waveform; Decomposing the spatial orientation of the magnetic field polarity reversal feature into magnetic field intensity change gradients in three orthogonal detection directions, and generating an electromagnetic signal spatial marker sequence based on the number of positive and negative reversals and amplitude differences in the gradient direction; In each phase interval of the equipment operation cycle, the timestamp of the mechanical signal time mark sequence and the azimuth angle change trajectory of the electromagnetic signal space mark sequence are synchronously scanned. When the scanning result shows that there is a continuous azimuth angle offset and the offset direction is consistent with the radial force direction of the shaft, a cross-domain matching event is triggered; The transient high-frequency component amplitude peak and the magnetic field polarity reversal amplitude peak in the cross-domain matching event are subjected to geometric scaling processing, and a hybrid feature marker is generated for the processed peak according to a preset mechanical and electromagnetic coupling weight coefficient.

4. The method according to claim 1, wherein The method of matching the fault history case library of the electrical equipment according to the energy concentration area of ​​the coherence spectrum within a specific frequency band and outputting a maintenance strategy set for the superposition fault of the dynamic imbalance of the rotating shaft and the electromagnetic interference includes: Locating an energy concentration region in the coherence map, and extracting a mapping relationship between an energy distribution shape in the energy concentration region and a physical motion pattern of a dynamic imbalance of a rotating shaft; Performing similarity search in the fault history case library of the electrical equipment according to the mapping relationship to retrieve a fault case group containing the same electromagnetic interference spectrum characteristics; Performing a correlation analysis between the shaft geometry deviation parameters and the electromagnetic shielding failure parameters on the fault case group to generate a candidate parameter set including an allowable range of calibration error and an adjustment value of the shielding layer thickness; The candidate parameter sets are prioritized according to the equipment operating environment parameters, and a maintenance strategy set including axis calibration operation steps and electromagnetic shielding optimization paths is formed according to the sorting results.

5. The method according to claim 1, wherein The multi-scale frequency band division of the non-periodic torque waveform is performed to screen out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and the magnetic field polarity reversal characteristics corresponding to the axis deviation direction are simultaneously separated from the magnetic field gradient anomaly signal, including: Continuously intercepting the non-periodic torque waveform according to windows of different time lengths, extracting the fluctuation period of the waveform amplitude in the intercepted results, and screening out window segments with the fluctuation period less than a preset reference value as candidate high-frequency intervals; Locking the overlapping window segments associated with the start-up and shutdown phases of the equipment within the candidate high-frequency interval, extracting the short-term fluctuation features of the amplitude mutation direction that continuously reverses within the overlapping window segments and marking them as transient high-frequency components; synchronously calculating the gradient difference of adjacent sampling points in a single orthogonal direction for the magnetic field gradient anomaly signal, and determining that the direction is the main direction of the axis deviation when the gradient difference exceeds the sum of the differences in adjacent directions; An alternating change region of the gradient difference sign is captured in the main direction of the axis center offset, and a magnetic field polarity reversal feature corresponding to the main direction of the axis center offset is extracted in the alternating change region.

6. The method according to claim 3, characterized in that In each phase interval of the equipment operation cycle, the timestamp of the mechanical signal time mark sequence and the azimuth angle change trajectory of the electromagnetic signal space mark sequence are synchronously scanned. When the scanning result shows that there is a continuous azimuth angle offset and the offset direction is consistent with the radial force direction of the shaft, a cross-domain matching event is triggered, including: The equipment operation cycle is divided into multiple phase intervals according to the rotation angle of the rotating shaft, where each phase interval corresponds to a preset radial force direction range of the shaft; Mapping the timestamp of the mechanical signal time mark sequence to the corresponding phase interval, and extracting the azimuth angle change trajectory of the electromagnetic signal space mark sequence in the same phase interval; Detecting a continuous azimuth shift event in the azimuth change trajectory, and determining that the azimuth shift event is a valid continuous shift when the shift direction in the azimuth shift event coincides with the radial force direction of the shaft corresponding to the current phase interval; A cross-domain matching event is triggered based on the effective continuous offset, and the starting position of the mechanical signal timestamp is bound to the ending position of the electromagnetic signal azimuth offset through the cross-domain matching event.

7. The method according to claim 2, characterized in that The method comprises: calculating the product of the torque fluctuation energy ratio and the magnetic field distortion energy ratio in the time-frequency unit, marking the area where the product value exceeds the dual energy baseline value as the effective overlapping area, and calculating the ratio of the effective overlapping area to the total area of ​​the time-frequency unit as the frequency band overlap coefficient. Delineating a maximum envelope of the torque fluctuation energy distribution within the time-frequency unit, and calculating a ratio of the energy value at each time point on the maximum envelope to the total energy of the unit as the torque fluctuation energy ratio; Synchronously calculating the local peak density of the magnetic field distortion energy distribution within the same time-frequency unit, and taking the ratio of the local peak density to the total energy of the unit as the magnetic field distortion energy ratio; Multiplying the torque fluctuation energy ratio and the magnetic field distortion energy ratio at each time point, screening out continuous time segments where the product value exceeds both the torque reference fluctuation energy and the magnetic field reference distortion energy and marking them as valid overlapping areas; The percentage of the time span of the effective overlapping area to the total time length of the time-frequency unit is calculated, and the percentage is combined with the average product value in the effective overlapping area to generate a frequency band overlap coefficient.

8. A remote maintenance guidance system for electrical equipment, characterized in that: include: A deployment module is configured to synchronously deploy a torque fluctuation sensor and a magnetoresistive sensor on the surface of a rotating shaft of an electrical device, wherein the torque fluctuation sensor continuously captures a non-periodic torque waveform caused by a sudden change in the mechanical load of the shaft, and the magnetoresistive sensor simultaneously detects an abnormal magnetic field gradient signal generated when the rotating shaft deviates in an orthogonal direction. a screening module for performing multi-scale frequency band division on the non-periodic torque waveform, screening out transient high-frequency components associated with the start-up and shutdown phases of the equipment, and simultaneously separating the magnetic field polarity reversal characteristics corresponding to the axis offset direction from the magnetic field gradient anomaly signal; a matching module for performing cross-domain matching between the appearance moment of the transient high-frequency component and the spatial orientation of the magnetic field polarity reversal feature, and generating a hybrid feature signature containing a mechanical and electromagnetic coupling relationship based on the matching result; a calculation module, configured to calculate the frequency band overlap of the time-frequency energy distribution in the hybrid feature marker using wavelet coherence analysis, and generate a coherence spectrum representing the interaction strength between the torque mutation and the magnetic field distortion according to the frequency band overlap; The output module is used to match the fault history case library of the electrical equipment according to the energy concentration area of ​​the coherence spectrum within a specific frequency band, and output a maintenance strategy set for the dynamic imbalance of the rotating shaft and the electromagnetic interference superposition fault.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an electrical equipment remote maintenance guidance method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the method for remote maintenance guidance of electrical equipment according to any one of claims 1 to 7 is implemented.

Images