A method for joint diagnosis and generation of vibration and noise in motor operation

Abstract

This invention relates to the field of equipment testing technology and discloses a method for generating joint vibration and noise diagnostics of motor operating conditions. The method includes: extracting dual-modal data from operating condition data during motor operation to obtain vibration and noise signals; performing order tracking analysis on the vibration signals to obtain electrical vibration characteristic order components; calculating the time-domain energy envelope and performing synchronous time-domain segmentation on the noise signals to obtain noise signal slices; filtering out non-correlated components from the noise signal slices that are not related to the vibration characteristic order components, and extracting the vibration synchronization noise components that are synchronized with the current rotation cycle of the motor based on the filtered results; constructing an acoustic-vibration fusion spectrum; analyzing the distribution law of acoustic intensity in the acoustic-vibration fusion spectrum, identifying correlation abrupt change points, and diagnosing specific operating condition faults based on the distribution pattern of correlation abrupt change points. This invention can improve the efficiency of generating joint vibration and noise diagnostics of motor operating conditions.

Description

Technical Field

[0001] This invention relates to the field of equipment testing technology, and in particular to a method for generating joint diagnostic data on vibration and noise during motor operation. Background Technology

[0002] Existing technologies have significant shortcomings in the signal processing stage of joint diagnosis of motor vibration and noise. They fail to perform dual-modal synchronous extraction and collaborative processing of motor operating status data, only collecting vibration or noise signals individually, or analyzing the two types of signals in isolation, ignoring the correlation and synergy between the two and the motor operating status, and failing to comprehensively capture fault-related features. Furthermore, they do not conduct in-depth order tracking analysis of vibration signals, only simply extracting time-domain or frequency-domain statistical features, making it difficult to accurately locate the vibration feature order components related to the motor rotation cycle. At the same time, they do not perform synchronous time-domain segmentation and filtering of irrelevant components in noise signals based on vibration features, resulting in interference with effective information in noise signals and the inability to extract noise components synchronized with vibration, thus affecting the accuracy of diagnostic data.

[0003] Existing technologies have significant shortcomings in fault diagnosis and feature fusion. They fail to construct a fusion spectrum to correlate the core characteristics of vibration and noise, relying solely on the characteristics of a single signal for fault diagnosis, making it difficult to intuitively reflect the distribution patterns and correlations of sound and vibration intensity. Furthermore, they fail to identify correlational abrupt changes through spectral distribution analysis, relying only on fixed thresholds or experience to judge faults, failing to accurately capture the feature abrupt changes caused by the fault, leading to ambiguous fault location. Finally, they fail to match specific fault types based on the distribution patterns of abrupt change points, only making a general determination of fault presence or absence, unable to distinguish specific fault types under different operating conditions, resulting in a lack of specificity in diagnostic results and failing to meet the actual needs of accurate motor fault diagnosis and efficient maintenance. Summary of the Invention

[0004] This invention provides a method for generating joint diagnosis of vibration and noise in motor operation to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for joint diagnosis and generation of vibration and noise in motor operating conditions, comprising:

[0006] S1. When the motor is running, perform dual-modal data extraction on the motor's operating status data to obtain the motor's vibration signal and noise signal;

[0007] S2. Perform order tracking analysis on the vibration signal to obtain the vibration characteristic order components of the motor;

[0008] S3. Calculate the time-domain energy envelope of the vibration characteristic order components, and use the time-domain energy envelope as a reference to synchronously segment the noise signal in the time domain to obtain the noise signal slice of the motor;

[0009] S4. Based on the frequency of the vibration characteristic order component, filter out the irrelevant components in the noise signal slice that are not related to the vibration characteristic order component, and extract the vibration synchronization noise component that is synchronized with the current rotation cycle of the motor according to the filtering result.

[0010] S5. Using the vibration characteristic order component as the abscissa and the vibration synchronization noise component as the ordinate, the acoustic-vibration fusion spectrum of the motor is constructed.

[0011] S6. Analyze the distribution pattern of acoustic intensity in the acoustic-vibration fusion spectrum, identify the correlation mutation points of the motor, and diagnose specific operating state faults of the motor based on the distribution pattern of the correlation mutation points.

[0012] In a preferred embodiment, the step of performing dual-modal data extraction on the operating state data of the motor during operation to obtain the vibration signal and noise signal of the motor includes:

[0013] During motor operation, the physical excitation generated by the moving parts of the motor is synchronously collected to obtain the original vibration waveform and original noise waveform of the motor.

[0014] Based on the rated operating frequency of the motor, the original vibration waveform is bandpass filtered to obtain the initial vibration signal of the motor.

[0015] Based on the rated operating frequency of the motor, the original noise waveform is subjected to high-pass filtering to obtain the initial noise signal of the motor;

[0016] The initial vibration signal and the initial noise signal are time-axis calibrated to obtain the vibration signal and noise signal of the motor.

[0017] In a preferred embodiment, the step of performing order tracking analysis on the vibration signal to obtain the vibration characteristic order components of the motor includes:

[0018] The transient impact waveform with abrupt amplitude changes and the steady-state vibration waveform reflecting a stable operating state are separated from the vibration signal.

[0019] The occurrence time interval pattern of the transient impact waveform is analyzed, and the periodically changing impact waveforms in the transient impact waveform are used as the rotational synchronization reference pulses of the motor.

[0020] Based on the instantaneous time interval of the rotation synchronization reference pulse, establish the angle-time correspondence curve of the instantaneous rotation phase in the motor;

[0021] Based on the angle-time correspondence curve, the steady-state vibration waveform is resampled to obtain the angular domain steady-state waveform of the steady-state vibration waveform;

[0022] Within the complete rotation cycle of the angular domain steady-state waveform, locate the specific mechanical angle position corresponding to the peak value of the angular domain steady-state waveform;

[0023] The steady-state vibration amplitude at the specific mechanical angle position is taken as the vibration characteristic order component of the motor.

[0024] In a preferred embodiment, calculating the time-domain energy envelope of the vibration characteristic order components includes:

[0025] The vibration characteristic order components are serialized and arranged to obtain the time-series amplitude sequence of the motor;

[0026] The analysis window for the motor is formed by extracting continuous data points from the time-series amplitude sequence through a sliding interval of fixed length.

[0027] The analysis window is slid along the time-series amplitude sequence to calculate the energy characterization value of the analysis window, wherein the energy characterization value is calculated using the following formula:

[0028] ;

[0029] In the formula, The energy characterization value, The average amplitude of the data points within the analysis window. The total number of data points within the analysis window. The preset morphology adjustment factor, For the analysis window, the first The difference between each data point and the average amplitude;

[0030] By connecting the energy characterization values ​​in chronological order, the time-domain energy envelope of the vibration characteristic order components is obtained.

[0031] In a preferred embodiment, the step of synchronously time-domain segmenting the noise signal using the time-domain energy envelope as a reference to obtain a noise signal slice of the motor includes:

[0032] On the time-domain energy envelope, the interval from which the energy points in the time-domain energy envelope rise from a local minimum until they reach the next local minimum is defined as the energy rising segment of the time-domain energy envelope.

[0033] Within the energy rise segment, locate the inflection point where the amplitude growth rate in the time-domain energy envelope changes from fast to slow.

[0034] Using the time corresponding to the inflection point as the center, extend the time offset to both sides before and after the inflection point to obtain the associated feature window of the inflection point;

[0035] In the noise signal, a noise waveform segment that completely corresponds to the associated feature window is extracted as a noise signal slice of the motor.

[0036] In a preferred embodiment, filtering out components in the noise signal slice that are irrelevant to the vibration characteristic order component based on the frequency of the vibration characteristic order component includes:

[0037] In the frequency domain, the narrowband spectrum with significant energy concentration in the vibration characteristic order component is identified to obtain the dominant oscillation mode of the vibration characteristic order component.

[0038] Based on the center frequency and bandwidth of the dominant oscillation mode, determine the frequency selection template for the vibration characteristic order component;

[0039] The frequency selection template is applied to the spectral representation of the noise signal slice, so that the components in the noise signal slice spectrum that overlap with the frequency selection template are retained, thus obtaining the preliminary filtered noise signal slice of the motor.

[0040] Suppress the portion of the preliminary filtered noise signal slice that is outside the frequency selection template to obtain the standard filtered noise signal slice of the motor.

[0041] In a preferred embodiment, the step of extracting the vibration characteristic order component that is synchronized with the current rotation cycle of the motor based on the filtered result includes:

[0042] Acquire the time-domain waveform between the standard noise-filtered signal slice and the vibration characteristic order component;

[0043] Based on the time-domain waveform, determine the reference time anchor point for the start of the waveform rotation period in the order component of the vibration characteristics;

[0044] Based on the reference time anchor point, the waveform of the standard noise-filtered signal slice is periodically and synchronously segmented to obtain the noise period sequence of the standard noise-filtered signal slice;

[0045] According to the reference time anchor point, the waveform of the noise periodic sequence is aligned to obtain the aligned periodic sequence of the standard filtered noise signal slice;

[0046] Within the same time period, the waveform features that appear stably in the aligned periodic sequence are strengthened, while the waveform features that change randomly in the aligned periodic sequence are weakened, so as to obtain the average periodic waveform of the standard noise-filtered signal slice.

[0047] According to the original time sequence of the reference time anchor point, the average periodic waveform is reconstructed in time sequence to obtain the complete time domain signal of the average periodic waveform;

[0048] The complete time-domain signal is used as the vibration synchronization noise component of the vibration characteristic order component.

[0049] In a preferred embodiment, constructing the acoustic-vibration fusion spectrum of the motor by using the vibration characteristic order component as the abscissa and the vibration synchronization noise component as the ordinate includes:

[0050] On the time axis, establish the correspondence between the amplitude sequence of the vibration characteristic order component and the amplitude sequence of the vibration synchronization noise component;

[0051] Based on the correspondence, the amplitude of the vibration characteristic order component is mapped to the horizontal axis coordinate value of the motor's vibration intensity;

[0052] Based on the aforementioned correspondence, the amplitude of the vibration synchronization noise component is mapped to the acoustic intensity vertical axis coordinate value of the motor;

[0053] The vibration intensity horizontal axis coordinate value and the acoustic intensity vertical axis coordinate value are standardized and paired to obtain the two-dimensional coordinate points of the motor;

[0054] The two-dimensional coordinate points are reconstructed to obtain the acoustic-vibration fusion spectrum of the motor.

[0055] In a preferred embodiment, analyzing the distribution pattern of acoustic intensity in the acoustic-vibration fusion spectrum and identifying the correlation abrupt change points of the motor includes:

[0056] Based on the trajectory curve of the acoustic-vibration fusion spectrum, the inflection point on the trajectory curve where the slope changes significantly is used as the division rule;

[0057] Based on the aforementioned division rules, the trajectory curve is divided into trajectory segments representing the acoustic-vibration coupling relationship of the motor.

[0058] Based on the overall extension direction of the acoustic-vibration coupling relationship trajectory segment and the basic trend of acoustic intensity changing with vibration intensity within the acoustic-vibration coupling relationship trajectory segment, the distribution law of the acoustic-vibration fusion spectrum is determined.

[0059] Based on the distribution pattern, the changing trends between the trajectories in the acoustic-vibration coupling relationship trajectory segment are compared one by one to obtain the correlation mutation point of the motor.

[0060] In a preferred embodiment, diagnosing specific operating state faults of the motor based on the distribution pattern of the correlation mutation points includes:

[0061] The vibration intensity range of the acoustic-vibration fusion spectrum is divided into a normal background range and an abnormal potential abnormal range.

[0062] Observe the frequency of occurrence and clustering density of the relevant mutation points in the normal background interval and the abnormal potential anomaly interval respectively;

[0063] The frequency of occurrence and the clustering density are used to determine significant anomalies, thereby obtaining the target anomaly range of the motor.

[0064] Based on the position of the target abnormal region on the vibration intensity axis, locate the vibration characteristic components of the target abnormal region;

[0065] The vibration characteristic components are mapped to a preset fault-order table, and the fault type that matches the vibration characteristic components is determined in order to identify the specific operating state fault of the motor.

[0066] The specific operating state fault is output to the terminal of the motor to obtain a fault report of the motor.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] 1. This invention provides high-quality data support for motor operation status diagnosis through precise processing and in-depth feature extraction of dual-modal signals. It synchronously acquires the original waveforms of motor vibration and noise, and obtains highly synchronized dual-modal signals after filtering and time-axis calibration. The vibration signal undergoes order tracking analysis, separating transient and steady-state waveforms. An angle-time correspondence curve is established using a rotational synchronization reference pulse. After resampling, the vibration characteristic order component at a specific mechanical angle position is located. The noise signal is segmented based on the time-domain energy envelope of this component, filtering out irrelevant components and extracting the vibration synchronization noise component synchronized with the rotation cycle, thus comprehensively capturing the core characteristics of motor operation.

[0069] 2. This invention significantly improves the accuracy and relevance of motor fault diagnosis by leveraging acoustic-vibration fusion analysis and precise fault identification. It constructs an acoustic-vibration fusion spectrum by using the vibration characteristic order component and the vibration synchronization noise component as the horizontal and vertical axes, respectively, clearly presenting the distribution law of acoustic and vibration intensity. By analyzing the spectrum trajectory curve, it identifies correlation abrupt change points, divides normal and abnormal intervals, and locates the target abnormal interval based on the frequency and cluster density of the abrupt change points, extracting the corresponding vibration characteristic components. These characteristic components are mapped to a preset fault-order table to accurately match fault types and generate fault reports, achieving efficient diagnosis of faults in specific motor operating states and providing a reliable basis for equipment maintenance. Attached Figure Description

[0070] Figure 1 This is a flowchart illustrating a method for generating joint diagnosis of vibration and noise in motor operating state according to an embodiment of the present invention.

[0071] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0072] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0073] This application provides a method for generating a joint diagnosis of vibration and noise in motor operating conditions. The execution subject of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for generating a joint diagnosis of vibration and noise in motor operating conditions can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0074] Reference Figure 1 The diagram shown is a flowchart illustrating a method for generating joint vibration and noise diagnostics of a motor operating state according to an embodiment of the present invention. In this embodiment, the method includes:

[0075] S1. When the motor is running, perform dual-modal data extraction on the motor's operating status data to obtain the motor's vibration signal and noise signal;

[0076] In this embodiment of the invention, the step of performing dual-modal data extraction on the operating state data of the motor during operation to obtain the vibration signal and noise signal of the motor includes:

[0077] During motor operation, the physical excitation generated by the moving parts of the motor is synchronously collected to obtain the original vibration waveform and original noise waveform of the motor.

[0078] Based on the rated operating frequency of the motor, the original vibration waveform is bandpass filtered to obtain the initial vibration signal of the motor.

[0079] Based on the rated operating frequency of the motor, the original noise waveform is subjected to high-pass filtering to obtain the initial noise signal of the motor;

[0080] The initial vibration signal and the initial noise signal are time-axis calibrated to obtain the vibration signal and noise signal of the motor.

[0081] After the motor starts and enters a stable operating state, appropriate vibration and noise sensors are installed at key locations corresponding to the motor's moving parts. The vibration sensors are tightly fitted to the outer surface of the moving parts' housings to ensure accurate capture of the mechanical vibrations generated by the parts' operation. The noise sensors are fixed in unobstructed locations around the motor and at a suitable distance from the moving parts to avoid excessive interference from unrelated environmental noise. By synchronously collecting the physical excitations generated by the moving parts during rotation, the vibration sensors convert the mechanical vibrations into continuous electrical signal waveforms, forming the motor's original vibration waveform. The noise sensors convert the acoustic vibrations into continuous electrical signal waveforms, forming the motor's original noise waveform, ensuring complete synchronization of the two original waveforms in terms of acquisition time.

[0082] Determine the rated operating frequency of the motor, and use this as a benchmark to set the frequency range of the bandpass filter. This range, centered on the rated operating frequency, covers the effective frequency range of the vibration signal during normal motor operation. Input the original vibration waveform into the bandpass filter. The filter will block low-frequency interference signals and high-frequency noise signals outside this frequency range, allowing only the effective vibration signal near the rated operating frequency to pass through. After filtering, an initial vibration signal that accurately reflects the normal vibration state of the motor's operating components is obtained.

[0083] The cutoff frequency of the high-pass filter is set according to the rated operating frequency of the motor. This cutoff frequency is slightly lower than the rated operating frequency to ensure that low-frequency environmental noise and low-frequency signals generated by vibrations unrelated to the equipment below this frequency are filtered out. The original noise waveform is fed into the high-pass filter, which blocks all low-frequency interference signals below the cutoff frequency, retaining only the effective noise signals at and above the rated operating frequency. These effective noise signals are directly related to the operating state of the motor's moving parts. After filtering, the initial noise signal of the motor is obtained.

[0084] The acquisition time records of the initial vibration signal and initial noise signal are retrieved, and a standard time axis is established with the motor start-up time as the unified time origin. The acquisition time corresponding to each data point in the initial vibration signal and initial noise signal is checked one by one. Signal data points that deviate from the standard time axis are adjusted according to a uniform time interval to ensure that each data point of the two signals corresponds one-to-one on the time axis, ensuring that the vibration state and noise state at the same moment can be accurately matched. After time axis calibration, the vibration signal and noise signal of the motor are obtained.

[0085] The beneficial effects are that the physical excitation generated by the motor's operating components is collected synchronously to obtain the original vibration waveform and the original noise waveform, ensuring that the two signals are completely synchronized in the time dimension, fully preserving the synergistic relationship between vibration and noise during motor operation, avoiding the loss of acoustic-vibration correlation information in subsequent analysis due to asynchronous acquisition, and laying the foundation for the collaborative processing of dual-modal data.

[0086] Bandpass filtering is performed on the original vibration waveform based on the rated operating frequency of the motor to accurately select the effective vibration signal near the rated operating frequency, effectively intercept low-frequency interference and high-frequency noise, so that the obtained initial vibration signal can truly reflect the core vibration characteristics of the motor during normal operation, and eliminate the interference of irrelevant signals on the vibration state analysis.

[0087] The original noise waveform is high-pass filtered based on the motor's rated operating frequency, effectively filtering out low-frequency environmental noise and equipment-irrelevant vibrations below that frequency, retaining only the mid-to-high frequency noise signals related to the motor's operating state, allowing the initial noise signal to focus on the effective noise information generated by the motor's operation, thus improving the usability of the noise signal.

[0088] The initial vibration signal and initial noise signal are calibrated on the time axis. Using a unified time origin as a reference, the timing of the data points of the two types of signals is adjusted to ensure that the vibration state and noise state at the same moment correspond accurately. This completely eliminates the timing deviation that may occur during the acquisition process, and makes the vibration signal and noise signal form a strict time correlation. This provides accurate timing assurance for subsequent steps such as segmenting the noise signal based on vibration characteristics and extracting synchronous noise components, thereby improving the data reliability and result accuracy of the entire diagnostic process.

[0089] S2. Perform order tracking analysis on the vibration signal to obtain the vibration characteristic order components of the motor;

[0090] In this embodiment of the invention, the step of performing order tracking analysis on the vibration signal to obtain the vibration characteristic order components of the motor includes:

[0091] The transient impact waveform with abrupt amplitude changes and the steady-state vibration waveform reflecting a stable operating state are separated from the vibration signal.

[0092] The occurrence time interval pattern of the transient impact waveform is analyzed, and the periodically changing impact waveforms in the transient impact waveform are used as the rotational synchronization reference pulses of the motor.

[0093] Based on the instantaneous time interval of the rotation synchronization reference pulse, establish the angle-time correspondence curve of the instantaneous rotation phase in the motor;

[0094] Based on the angle-time correspondence curve, the steady-state vibration waveform is resampled to obtain the angular domain steady-state waveform of the steady-state vibration waveform;

[0095] Within the complete rotation cycle of the angular domain steady-state waveform, locate the specific mechanical angle position corresponding to the peak value of the angular domain steady-state waveform;

[0096] The steady-state vibration amplitude at the specific mechanical angle position is taken as the vibration characteristic order component of the motor.

[0097] The vibration signal is monitored throughout its entire duration, and the changes in signal amplitude are analyzed segment by segment. Waveform segments where the amplitude rises sharply and then falls back rapidly within a short period are identified. These waveform segments are transient impact waveforms, corresponding to signals generated by sudden physical actions such as instantaneous collisions and friction between the motor's moving parts. Simultaneously, waveform segments whose signal amplitude fluctuates smoothly within a certain range without significant abrupt changes are selected. These waveform segments are unaffected by transient impacts and reflect the continuous and stable operation of the motor; these are considered steady-state vibration waveforms.

[0098] The occurrence time of transient impact waveforms is continuously tracked, and the time point of each impact waveform is recorded. The time interval between two adjacent impact waveforms is calculated. All time intervals are analyzed to identify impact waveform groups whose interval values ​​repeat according to a fixed pattern. The occurrence frequency of these impact waveforms is synchronized with the rotation frequency of the motor's rotating parts, and is a direct reflection of the motor's rotational motion. These impact waveform groups are determined as the motor's rotational synchronization reference pulses.

[0099] Using the rotational synchronization reference pulse as the time base, the specific time point corresponding to each reference pulse is recorded. Combined with the rotational characteristics of the motor's rotating components, a correspondence between time and instantaneous rotational phase is established. Based on the time interval between adjacent reference pulses, the change amplitude of the motor's rotational phase per unit time is calculated. Following the time progression, the instantaneous rotational phase corresponding to each time point is determined sequentially, ultimately forming an angle-time correspondence curve that clearly shows the relationship between time and instantaneous rotational phase.

[0100] Based on the angle-time correspondence curve, the steady-state vibration waveform in the time dimension is converted into waveform data in the angle dimension. The steady-state vibration waveform is resampled according to the uniform variation interval of the rotation phase, with one sampling point corresponding to each phase interval. This ensures that the sampling points are uniformly distributed along the rotation angle, eliminating waveform distortion caused by minor fluctuations in motor speed. Through this resampling process, the original time-axis-based steady-state vibration waveform is converted into an angular-domain steady-state waveform based on the rotation angle.

[0101] The entire rotation cycle of the angular domain steady-state waveform is traversed, and the amplitude data corresponding to each angular position on the waveform is recorded one by one. The magnitudes of all amplitude data are compared, and the angular position corresponding to the largest amplitude value is found. This angular position is the position where the vibration intensity of the motor's rotating parts is the greatest during rotation, and it is directly related to the structural characteristics and stress state of the parts. This angular position is determined as the specific mechanical angular position corresponding to the peak value of the angular domain steady-state waveform.

[0102] The steady-state vibration amplitude at a specific mechanical angle position is extracted. This amplitude reflects the vibration intensity of the motor at key vibration locations and is a core characteristic parameter of the motor's operating state. This amplitude is explicitly defined as the vibration characteristic order component of the motor. This component can accurately reflect the correlation between the vibration characteristics of the motor's moving parts and the rotational order, providing crucial characteristic basis for subsequent motor operating state assessment and fault diagnosis.

[0103] The beneficial effects are that by separating the transient impact waveform and the steady-state vibration waveform from the vibration signal, the different types of vibration signals can be accurately separated, eliminating the interference of sudden factors such as instantaneous collisions and friction on the analysis of stable operation, while retaining the core waveform that reflects the continuous operation characteristics of the motor, providing a clean and targeted data foundation for subsequent order analysis.

[0104] By analyzing the time interval patterns of transient impact waveforms and selecting periodic impact waveforms as rotational synchronization reference pulses, we can accurately capture characteristic signals that are synchronized with the motor rotation cycle. These pulses are directly related to the rotational patterns of the motor's operating components, providing a reliable benchmark for establishing the correspondence between time and rotational phase, and ensuring that subsequent analysis remains synchronized with the actual rotational state of the motor.

[0105] An angle-time correspondence curve is established based on the instantaneous time interval of the rotation synchronization reference pulse, which precisely binds discrete time information with the motor rotation phase, clearly showing the instantaneous rotation attitude of the motor at different times. This breaks the limitation of analysis based solely on the time axis and creates conditions for analyzing vibration characteristics from the angle dimension.

[0106] By resampling the steady-state vibration waveform based on the angle-time correspondence curve, the angular domain steady-state waveform can be obtained. This can eliminate waveform distortion caused by small fluctuations in motor speed, make the waveform uniformly distributed across the rotation angle, and more realistically reflect the vibration of the motor's operating components at different mechanical angle positions, thereby improving the accuracy of vibration characteristic analysis.

[0107] Locating a specific mechanical angle position within the complete rotation cycle of the angular domain steady-state waveform, which corresponds to the peak point of the maximum vibration intensity, centrally reflects the vibration characteristics of the motor's operating components in the key phase. This is the core position reflecting the equipment's operating status and provides a precise target point for extracting targeted features.

[0108] Using the steady-state vibration amplitude at a specific mechanical angle position as the vibration characteristic order component, this component is related to both the rotation order of the motor and focuses on the key vibration position. It can accurately characterize the core vibration characteristics of the motor operation, providing a high-value core characteristic basis for subsequent synchronous segmentation of noise signals, construction of acoustic-vibration fusion spectrum and fault diagnosis, and significantly improving the accuracy and pertinence of diagnosis.

[0109] S3. Calculate the time-domain energy envelope of the vibration characteristic order components, and use the time-domain energy envelope as a reference to synchronously segment the noise signal in the time domain to obtain the noise signal slice of the motor;

[0110] In this embodiment of the invention, calculating the time-domain energy envelope of the vibration characteristic order component includes:

[0111] The vibration characteristic order components are serialized and arranged to obtain the time-series amplitude sequence of the motor;

[0112] The analysis window for the motor is formed by extracting continuous data points from the time-series amplitude sequence through a sliding interval of fixed length.

[0113] The analysis window is slid along the time-series amplitude sequence to calculate the energy characterization value of the analysis window, wherein the energy characterization value is calculated using the following formula:

[0114] ;

[0115] In the formula, The energy characterization value, The average amplitude of the data points within the analysis window. The total number of data points within the analysis window. The preset morphology adjustment factor, For the analysis window, the first The difference between each data point and the average amplitude;

[0116] By connecting the energy characterization values ​​in chronological order, the time-domain energy envelope of the vibration characteristic order components is obtained.

[0117] The step of synchronously segmenting the noise signal in the time domain, using the time-domain energy envelope as a reference, to obtain a noise signal slice of the motor, includes:

[0118] On the time-domain energy envelope, the interval from which the energy points in the time-domain energy envelope rise from a local minimum until they reach the next local minimum is defined as the energy rising segment of the time-domain energy envelope.

[0119] Within the energy rise segment, locate the inflection point where the amplitude growth rate in the time-domain energy envelope changes from fast to slow.

[0120] Using the time corresponding to the inflection point as the center, extend the time offset to both sides before and after the inflection point to obtain the associated feature window of the inflection point;

[0121] In the noise signal, a noise waveform segment that completely corresponds to the associated feature window is extracted as a noise signal slice of the motor.

[0122] The vibration characteristic order components are arranged sequentially according to their acquisition time, ensuring that each component value corresponds to a unique acquisition time, forming a continuous and ordered sequence of data. The arrangement process strictly follows the acquisition sequence, without altering the original magnitude or order of any component value, ultimately yielding a time-series amplitude sequence that fully reflects the time-varying characteristics of the motor's vibration characteristic order components.

[0123] Based on the operating characteristics of the motor and the frequency of vibration signal changes, a fixed-length sliding interval is set. The length of this interval must cover a complete fluctuation cycle of the vibration characteristic order component to ensure that the captured data points comprehensively reflect the vibration energy situation within that time period. Using the set sliding interval as a standard, continuously distributed data points are captured starting from the beginning of the time-series amplitude sequence. These data points constitute an independent analysis unit, namely the motor analysis window.

[0124] The constructed analysis window is smoothly slid from the beginning to the end of the time-series amplitude sequence, with each slide distance equal to the interval of one data point. This ensures that no data point is missed during the slide and that adjacent analysis windows remain continuous. At each sliding position, the energy of the amplitude of all data points within the analysis window is calculated. The total energy within the window is represented by summing the squares of the amplitudes of each data point, yielding the energy representation value corresponding to each sliding position.

[0125] Collect the energy characterization values ​​calculated from all sliding positions, and connect them sequentially according to the time order of the analysis window sliding to form a continuous curve. This curve can intuitively show the energy change trend of the vibration characteristic order component over time, clearly presenting the peak, trough and fluctuation patterns of energy, and finally obtaining the time-domain energy envelope of the vibration characteristic order component.

[0126] By comprehensively traversing the entire curve of the time-domain energy envelope, analyzing the amplitude change trend of energy points on the curve point by point, and accurately identifying the local minimum point with the lowest amplitude on the curve, the changes of subsequent energy points are tracked starting from each local minimum point until the amplitude of the energy point drops to the next local minimum point. The curve interval between these two adjacent local minimum points is clearly defined as the energy rising segment of the time-domain energy envelope. This interval completely covers the complete process of energy gradually rising from the lowest level and then falling back to the lowest level.

[0127] An amplitude growth rate analysis is performed on all energy points within the energy rise segment. The amplitude difference between two adjacent energy points is calculated, and the magnitude of the difference determines the rate of increase. Starting from the beginning of the energy rise segment, adjacent differences are compared sequentially along the energy rise direction. When the amplitude difference changes from a continuously increasing state to a continuously decreasing state, the energy point corresponding to this change is the inflection point where the amplitude growth rate slows down. This inflection point is the core dividing point of rate change during the energy rise process.

[0128] Using the specific moment corresponding to the inflection point as a fixed center, a fixed time offset is set based on the correlation characteristics between motor vibration and noise signals. This offset ensures that the complete noise signal segment related to vibration energy changes before and after the inflection point is covered. The set time offset is synchronously extended to both sides of the inflection point moment to form a symmetrical time interval. This interval completely includes the inflection point moment and the surrounding time range related to energy changes, which is the correlation feature window of the inflection point.

[0129] The noise signal, after being calibrated along the time axis, is retrieved. Based on the time range corresponding to the associated feature window, the waveform segment corresponding to that time range is precisely located within the noise signal. The noise waveform within that time range is then completely extracted using a truncation tool, ensuring that the extracted waveform segment is perfectly aligned with the associated feature window in time, with no time deviation. This extracted noise waveform segment is the noise signal slice of the motor, accurately corresponding to the noise characteristics of key energy change stages in the time-domain energy envelope.

[0130] The energy characterization values ​​are derived from all data points within the analysis window. These data points are taken from the time-series amplitude sequence, which is obtained by serializing and arranging the vibration characteristic order components in chronological order of acquisition time.

[0131] The average amplitude is derived from the amplitudes of all data points within the analysis window. The average amplitude of the data points within the analysis window is calculated by summing the amplitudes of each data point within the analysis window and then dividing the sum by the total number of data points within the analysis window.

[0132] The total number of data points in the analysis window comes from a fixed-length sliding interval. This sliding interval is used to extract continuous data points from the time series amplitude sequence to form the analysis window. The fixed length of the sliding interval directly determines the number of data points contained in each analysis window.

[0133] The morphological adjustment factor is pre-set based on the operating characteristics of the motor and the variation law of the vibration signal. During the setting process, the fluctuation range and amplitude variation characteristics of the vibration characteristic order component during normal operation of the motor are taken into account to ensure that the morphological adjustment factor can adapt to the characteristics of the vibration signal and accurately adjust the degree of influence of the difference on the energy characterization value.

[0134] The first analysis window The difference between the nth data point and the average amplitude comes from the calculated amplitude of each data point within the analysis window and the average amplitude. This is achieved by subtracting the average amplitude from the amplitude of each data point within the analysis window, obtaining the difference between the corresponding data point and the average amplitude, and then taking the absolute value of this difference. The absolute difference between each data point and the average amplitude.

[0135] The formula means that by comprehensively analyzing the average amplitude of the data points within the analysis window and the absolute difference between each data point and the average amplitude, the energy characterization value corresponding to the analysis window can be accurately calculated.

[0136] During the calculation process, the average amplitude is used as the basis to ensure that the energy characterization value can reflect the overall amplitude level of the data points within the analysis window. Then, the average of the absolute difference between each data point and the average amplitude is calculated, and the influence weight of the average on the overall result is adjusted by combining the morphological adjustment factor. This ensures that the energy characterization value can not only reflect the overall amplitude of the data points, but also reflect the degree of dispersion of the data points from the average level.

[0137] This calculation method can comprehensively capture the energy characteristics of the vibration characteristic order components within the analysis window, including both the overall amplitude contribution and the local fluctuation differences. The final energy characterization value can accurately reflect the actual energy situation of the vibration characteristic order components within the corresponding analysis window, providing reliable numerical support for the subsequent generation of time-domain energy envelope.

[0138] The beneficial effects are that the vibration characteristic order components are serialized and arranged to obtain a time-series amplitude sequence, which strictly follows the time sequence of data acquisition, ensuring that each vibration characteristic amplitude corresponds to a unique moment, and completely preserving the trajectory of vibration characteristic changes over time, providing an orderly and coherent data foundation for subsequent energy analysis.

[0139] The analysis window is formed by extracting continuous data points from a sliding interval of fixed length. The length of this window is adapted to the fluctuation period of the vibration characteristics, which can comprehensively cover the vibration energy information within a certain period of time. This avoids the omission of energy characteristics due to the data being too short, while ensuring that the data analysis dimensions of each analysis window are consistent, thus ensuring the uniformity of energy calculation.

[0140] The analysis window slides along the temporal amplitude sequence and calculates the energy characterization value. During the sliding process, no data is missed and adjacent windows are continuously connected. The average amplitude of the data within the analysis window and the degree of dispersion of each data point from the average amplitude are comprehensively considered in combination with the formula. The influence weight of the dispersion on the energy value is precisely adjusted by the morphological adjustment factor, so that the energy characterization value can reflect the overall vibration intensity and highlight the local fluctuation differences, accurately depicting the vibration energy state of each time period.

[0141] By connecting the energy characterization values ​​in chronological order, a time-domain energy envelope is obtained, which intuitively presents the fluctuation and change law of the energy of the vibration characteristic order components over time. The peak, valley and change trend of energy are clearly marked, providing an accurate and reliable reference for subsequent synchronous time-domain segmentation of noise signals based on this envelope, ensuring that the noise signal slices can accurately correspond to the key change stages of vibration energy.

[0142] By defining energy rise segments on the time-domain energy envelope, the complete cycle of energy gradually rising from the lowest level and then falling back is precisely locked, ensuring that each rise segment corresponds to a complete energy fluctuation of the vibration characteristic order component, providing a time interval reference that is highly consistent with the vibration energy change for noise signal segmentation.

[0143] Within the energy rise phase, locate the inflection point where the amplitude growth rate slows down. This inflection point is the key dividing point of vibration energy change, which fully reflects the transition characteristic of vibration intensity from rapid increase to gradual change. Using this as the core reference point, subsequent noise signal segmentation can accurately focus on the key stage of vibration energy change.

[0144] The associated feature window is obtained by expanding forward and backward from the moment corresponding to the inflection point. This not only fully covers the time range related to the change in vibration energy before and after the inflection point, but also ensures that the window size is uniform by using a fixed time offset, so as to avoid the omission of effective noise information or the mixing of irrelevant noise due to improper window range.

[0145] By extracting noise waveform segments that perfectly correspond to the associated feature windows from the noise signal as noise signal slices, precise synchronization between the noise signal and the temporal energy changes of the vibration feature order components is achieved. This allows each noise signal slice to accurately correspond to the key stages of vibration energy fluctuations, effectively preserving the effective noise information related to vibration. This lays a high-quality noise data foundation for subsequent filtering of irrelevant components and extraction of vibration synchronization noise components.

[0146] S4. Based on the frequency of the vibration characteristic order component, filter out the irrelevant components in the noise signal slice that are not related to the vibration characteristic order component, and extract the vibration synchronization noise component that is synchronized with the current rotation cycle of the motor according to the filtering result.

[0147] In this embodiment of the invention, filtering out components in the noise signal slice that are irrelevant to the vibration characteristic order component based on the frequency of the vibration characteristic order component includes:

[0148] In the frequency domain, the narrowband spectrum with significant energy concentration in the vibration characteristic order component is identified to obtain the dominant oscillation mode of the vibration characteristic order component.

[0149] Based on the center frequency and bandwidth of the dominant oscillation mode, determine the frequency selection template for the vibration characteristic order component;

[0150] The frequency selection template is applied to the spectral representation of the noise signal slice, so that the components in the noise signal slice spectrum that overlap with the frequency selection template are retained, thus obtaining the preliminary filtered noise signal slice of the motor.

[0151] Suppress the portion of the preliminary filtered noise signal slice that is outside the frequency selection template to obtain the standard filtered noise signal slice of the motor.

[0152] The step of extracting the vibration synchronization noise component that is synchronized with the current rotation cycle of the motor based on the filtered results includes:

[0153] Acquire the time-domain waveform between the standard noise-filtered signal slice and the vibration characteristic order component;

[0154] Based on the time-domain waveform, determine the reference time anchor point for the start of the waveform rotation period in the order component of the vibration characteristics;

[0155] Based on the reference time anchor point, the waveform of the standard noise-filtered signal slice is periodically and synchronously segmented to obtain the noise period sequence of the standard noise-filtered signal slice;

[0156] According to the reference time anchor point, the waveform of the noise periodic sequence is aligned to obtain the aligned periodic sequence of the standard filtered noise signal slice;

[0157] Within the same time period, the waveform features that appear stably in the aligned periodic sequence are strengthened, while the waveform features that change randomly in the aligned periodic sequence are weakened, so as to obtain the average periodic waveform of the standard noise-filtered signal slice.

[0158] According to the original time sequence of the reference time anchor point, the average periodic waveform is reconstructed in time sequence to obtain the complete time domain signal of the average periodic waveform;

[0159] The complete time-domain signal is used as the vibration synchronization noise component of the vibration characteristic order component.

[0160] The vibration characteristic order components are transformed to the frequency domain, and their spectra are comprehensively analyzed, with the energy distribution in each frequency range of the spectrum being examined segment by segment. A key focus is on identifying narrow bands in the spectrum where the energy density is significantly higher than other regions and the frequency range is relatively concentrated. This portion of the spectrum carries the core energy of the vibration characteristic order components and represents the main frequency expression of the vibration signal. These narrow bands are thus identified as the dominant oscillation mode of the vibration characteristic order components.

[0161] The spectrum of the dominant oscillation mode is accurately characterized, and the frequency corresponding to the highest energy in the narrow band spectrum is located and taken as the center frequency of the dominant oscillation mode. The frequency span of the narrow band spectrum from the beginning to the end is measured to determine the bandwidth of the dominant oscillation mode. Using the center frequency as the core and the bandwidth as the boundary, a frequency range model that perfectly matches the spectrum of the dominant oscillation mode is constructed. This model serves as the frequency selection template for the vibration characteristic order components.

[0162] The noise signal slice is converted into a spectral representation, making the frequency distribution characteristics of the noise signal intuitively apparent. The constructed frequency selection template is precisely aligned with the spectral representation of the noise signal slice. Through a spectral matching mechanism, the portion of the noise signal slice's spectrum that completely overlaps with the frequency selection template's frequency range is selected. The spectral components of the overlapping portion are not modified in any way, preserving their frequency and energy information completely. This selection and preservation process yields the initial filtered noise signal slice for the motor.

[0163] Based on the spectrum of the initially filtered noise signal slice, frequency ranges outside the frequency selection template are identified as irrelevant interference components. Spectral suppression techniques are employed to attenuate the spectral components within these irrelevant frequency ranges, reducing their energy to a level that does not affect the core signal, thus completely eliminating interference from unrelated components. After suppressing irrelevant components, a standard filtered noise signal slice of the motor is obtained, retaining only frequency components relevant to the vibration characteristic order components.

[0164] The time-domain data of the standard noise-filtered signal slice and the vibration characteristic order component are acquired simultaneously. The complete waveform changes of the two types of signals on the same time axis are recorded through signal acquisition equipment. It is ensured that the time accuracy of the two types of signals remains consistent during the acquisition process, without any timing deviation, and the amplitude fluctuations of the standard noise-filtered signal slice and the waveform patterns of the vibration characteristic order component are fully preserved, ultimately yielding the corresponding time-domain waveforms of both.

[0165] Observe the time-domain waveform of the vibration characteristic order components, identify the recurring periodic structures in the waveform, and find the key features at the start of each cycle. These features are usually manifested as the starting point of the waveform rising from a trough or the inflection point of the waveform corresponding to a specific phase. Determine the time point corresponding to this key feature as the reference time anchor point for the start of the waveform rotation cycle, ensuring that this anchor point can accurately locate the start time of each rotation cycle.

[0166] Using a reference time anchor point as the segmentation benchmark, starting from the beginning of the time-domain waveform of the standard noise-filtered signal slice, the waveform of the standard noise-filtered signal slice is continuously segmented according to the rotation period length of the vibration characteristic order component. The time span of each segment is completely consistent with the rotation period of the vibration characteristic order component. The segmentation process strictly follows the temporal sequence of the reference time anchor point, without omitting any periodic segment, and finally obtains a noise periodic sequence of the standard noise-filtered signal slice composed of multiple continuous periodic segments.

[0167] Using the waveform position corresponding to the reference time anchor point as the alignment benchmark, the position of each periodic segment in the noise periodic sequence is calibrated. The timing position of each periodic segment is adjusted so that all periodic segments completely coincide with the characteristic waveform features corresponding to the reference time anchor point, ensuring that each periodic segment maintains a high degree of consistency in phase and timing, eliminating minute timing deviations between periods, and obtaining the aligned periodic sequence of the standard filtered noise signal slice.

[0168] Waveform comparison analysis is performed on all periodic segments in the aligned periodic sequence, and the amplitude data of each periodic segment at each time point is statistically analyzed. For amplitude data with stable values ​​and minimal variation across all periodic segments, their proportion in the overall waveform is enhanced by accumulation; for amplitude data with random fluctuations and no fixed pattern, their influence weight is reduced by averaging, thus weakening their interference with the overall waveform. After this enhancement and weakening process, the average periodic waveform of a standard noise-filtered signal slice that reflects the stable waveform characteristics within the period is obtained.

[0169] Following the original time sequence of the reference time anchor points, the average periodic waveforms are sequentially connected and arranged. The start of each average periodic waveform is precisely aligned with the end of the previous waveform based on the reference time anchor points, ensuring that there are no waveform breaks or overlaps at the connection points. This fully restores the temporal variation pattern of the standard noise-filtered signal slice, ultimately forming a complete time-domain signal of continuous and complete average periodic waveforms.

[0170] The correlation between the complete time-domain signal and the vibration characteristic order components was verified, confirming that the phase of the signal period is completely synchronized with the rotation period of the vibration characteristic order components, and that only noise components related to the vibration characteristics are retained in the signal. After verification, this complete time-domain signal is clearly identified as the vibration synchronization noise component of the vibration characteristic order components, providing accurate noise characteristic basis for subsequent motor operating status assessment.

[0171] The beneficial effects are that it can identify the narrowband spectrum with significant energy concentration in the order components of vibration characteristics in the frequency domain to obtain the dominant oscillation mode, accurately lock the core frequency components directly related to the motor's operating state, clarify the main manifestations of vibration characteristics, and provide accurate frequency reference for subsequent noise signal filtering.

[0172] The frequency selection template is determined based on the center frequency and bandwidth of the dominant oscillation mode, so that the template can perfectly match the frequency range of the vibration characteristics. This ensures that the filtering process targets only the frequency components related to the vibration, avoiding the loss of effective signals or the retention of interference signals due to the ambiguity of the frequency range definition.

[0173] By applying a frequency selection template to the spectral representation of the noise signal slice and retaining overlapping components, a precise match between the noise signal and vibration characteristics in the frequency dimension is achieved. Only noise components related to the order components of vibration characteristics are retained, and irrelevant components such as environmental interference and unrelated equipment noise are initially eliminated, resulting in a more targeted preliminary filtered noise signal slice.

[0174] Suppressing the portion of the noise signal slice outside the frequency selection template in the initial filtered noise signal slice further eliminates the interference of irrelevant frequency components, so that the final standard filtered noise signal slice focuses only on the core noise information related to vibration characteristics, significantly improving the purity and effectiveness of the noise signal. This lays a high-quality noise data foundation for the subsequent extraction of vibration synchronization noise components and the construction of acoustic-vibration fusion spectrum, ensuring the accuracy of subsequent diagnostic procedures.

[0175] The time-domain waveforms of standard noise-filtered signal slices and vibration characteristic order components are acquired to ensure that the two types of signals are fully presented in the same time dimension, providing an intuitive waveform reference for subsequent synchronization analysis and ensuring that operations such as periodic synchronization and waveform alignment can be carried out based on real and complete signal data.

[0176] Using the time-domain waveform as a reference, a reference time anchor point is determined to accurately lock the starting position of the rotation cycle of the vibration characteristic order component. This anchor point becomes the core reference for subsequent noise signal segmentation and alignment, ensuring that all operations are strictly synchronized with the motor rotation cycle, and avoiding timing deviations from affecting the extraction accuracy of the synchronous noise component.

[0177] The standard filtered noise signal slices are periodically synchronized by using reference time anchors, so that the time span of each noise cycle sequence is completely consistent with the motor rotation cycle. This achieves precise matching of noise signals and vibration characteristics in the cycle dimension, laying a structural foundation for subsequent extraction of noise components synchronized with the rotation cycle.

[0178] The waveforms of the noise periodic sequence are aligned according to the reference time anchor point to eliminate the micro-timeline deviation between different periods, so that the key waveform features of all periodic segments coincide at the same time node, ensuring that subsequent enhancement and weakening processing can be based on a unified timeline reference and improving the accuracy of the average periodic waveform.

[0179] By strengthening the stable waveform characteristics in the aligned periodic sequence and weakening the random variation characteristics, the random interference components in the noise signal can be effectively filtered out, highlighting the stable noise characteristics related to the motor's operating state. This allows the obtained average periodic waveform to focus on the core effective information and better reflect the noise pattern synchronized with the vibration.

[0180] The average periodic waveform is reconstructed in time sequence according to the original order of the reference time anchor points to fully restore the trajectory of the noise signal change over time, forming a continuous and coherent complete time domain signal. This ensures that the signal can fully cover the noise characteristics of the motor during multi-cycle operation and avoids information breakage or omission.

[0181] Using the complete time-domain signal as the vibration synchronization noise component, this component is not only related to the frequency of the vibration characteristic order component, but also strictly synchronized with the current rotation cycle of the motor. This achieves deep synergy between the core characteristics of vibration and noise, providing high-value noise characteristic basis for subsequent construction of acoustic-vibration fusion spectrum and accurate diagnosis of motor faults, significantly improving the pertinence and reliability of diagnosis.

[0182] S5. Using the vibration characteristic order component as the abscissa and the vibration synchronization noise component as the ordinate, the acoustic-vibration fusion spectrum of the motor is constructed.

[0183] In this embodiment of the invention, the step of constructing the acoustic-vibration fusion spectrum of the motor by using the vibration characteristic order component as the abscissa and the vibration synchronization noise component as the ordinate includes:

[0184] On the time axis, establish the correspondence between the amplitude sequence of the vibration characteristic order component and the amplitude sequence of the vibration synchronization noise component;

[0185] Based on the correspondence, the amplitude of the vibration characteristic order component is mapped to the horizontal axis coordinate value of the motor's vibration intensity;

[0186] Based on the aforementioned correspondence, the amplitude of the vibration synchronization noise component is mapped to the acoustic intensity vertical axis coordinate value of the motor;

[0187] The vibration intensity horizontal axis coordinate value and the acoustic intensity vertical axis coordinate value are standardized and paired to obtain the two-dimensional coordinate points of the motor;

[0188] The two-dimensional coordinate points are reconstructed to obtain the acoustic-vibration fusion spectrum of the motor.

[0189] Using the time axis as a unified benchmark, the acquisition time corresponding to each data point in the amplitude sequence of the vibration characteristic order component and the amplitude sequence of the vibration synchronization noise component is checked one by one. This ensures that the amplitude data acquired at the same time in the two sequences form a one-to-one correspondence, so that the amplitude of each vibration characteristic order component can find a unique corresponding amplitude of the vibration synchronization noise component, and vice versa. This completely eliminates the correspondence misalignment problem caused by timing deviation and establishes a precise time correspondence.

[0190] Based on the established correspondence, all amplitude data in the amplitude sequence of the vibration characteristic order components are extracted. According to the preset coordinate mapping rules, each amplitude data is converted into a horizontal axis coordinate value suitable for the display of the spectrum. This conversion process strictly preserves the relative magnitude relationship between the amplitude data, ensuring that the intensity change of the vibration characteristic can be intuitively reflected through the difference in the horizontal axis coordinate value. Finally, these coordinate values ​​are determined as the horizontal axis coordinate values ​​of the motor's vibration intensity.

[0191] Based on the same correspondence, all amplitude data in the amplitude sequence of the vibration synchronization noise component are extracted. Using a standardization rule consistent with the horizontal axis coordinate mapping, each amplitude data is converted into a coordinate value corresponding to the vertical axis of the spectrum. During the conversion process, the original trend of the amplitude data is maintained, so that the level of acoustic intensity can be clearly presented through the fluctuation of the vertical axis coordinate value, thus obtaining the vertical axis coordinate value of the motor's acoustic intensity.

[0192] The vibration intensity horizontal axis values ​​and acoustic intensity vertical axis values ​​are paired according to the previously established time correspondence. Each vibration intensity horizontal axis value is combined with the corresponding acoustic intensity vertical axis value at the same time, forming a set of coordinate data containing both vibration and acoustic information. All paired coordinate data are then standardized to ensure that each set of data is in a consistent format and corresponds correctly, ultimately yielding the two-dimensional coordinate points of the motor.

[0193] All standardized and paired two-dimensional coordinate points are collected, and the position of each two-dimensional coordinate point is marked one by one in the preset spectral coordinate system according to the distribution pattern of the coordinate values. Adjacent coordinate points are connected in an orderly manner by continuous lines, and densely distributed areas of coordinate points are filled to highlight the feature aggregation, thus fully restoring the correlation and change law between vibration intensity and acoustic intensity. After spectral reconstruction processing, the acoustic-vibration fusion spectrum of the motor that can fully demonstrate the acoustic-vibration synergy characteristics of the motor is obtained.

[0194] The beneficial effect is that it establishes a correspondence between the vibration characteristic order components and the amplitude sequence of vibration synchronization noise components on the time axis, ensuring that the two types of core features are accurately matched in the time dimension, and that each vibration characteristic amplitude can correspond to the noise characteristic amplitude at the same moment, eliminating the correlation deviation caused by time sequence misalignment, and laying a synchronous foundation for acoustic-vibration fusion.

[0195] Based on the correspondence, the amplitude of the vibration characteristic order component is mapped to the horizontal axis coordinate value of vibration intensity, preserving the intensity variation law of vibration characteristics, so that the horizontal axis coordinate can intuitively reflect the core differences of the motor vibration state, making the distribution and variation trend of vibration intensity clearly identifiable, and providing a clear vibration dimension reference for subsequent spectrum analysis.

[0196] Based on the same correspondence, the amplitude of the vibration synchronization noise component is mapped to the vertical axis coordinate value of the acoustic intensity. The standardized mapping rule is consistent with that of the horizontal axis, ensuring that the changes in acoustic intensity can be accurately represented by the vertical axis coordinate. At the same time, it forms a unified quantitative scale with the vibration intensity, which facilitates the correlation analysis between the two.

[0197] By standardizing and pairing the vibration intensity horizontal axis coordinate value with the acoustic intensity vertical axis coordinate value, two-dimensional coordinate points are obtained. Each coordinate point integrates the dual characteristic information of vibration and acoustic at the same moment, realizing the organic fusion of the two types of core data, avoiding the limitations of single feature analysis, and providing data support for comprehensively reflecting the motor's operating status.

[0198] By reconstructing the spectrum of two-dimensional coordinate points, the acoustic-vibration fusion spectrum is obtained. The discrete coordinate points are transformed into a continuous and intuitive spectrum, which clearly shows the distribution law and correlation pattern of acoustic-vibration intensity. This makes the characteristics of motor operation status more visual and provides an intuitive and comprehensive analysis carrier for subsequent identification of correlation mutation points and accurate fault diagnosis.

[0199] S6. Analyze the distribution pattern of acoustic intensity in the acoustic-vibration fusion spectrum, identify the correlation mutation points of the motor, and diagnose specific operating state faults of the motor based on the distribution pattern of the correlation mutation points.

[0200] In this embodiment of the invention, the step of analyzing the distribution pattern of acoustic intensity in the acoustic-vibration fusion spectrum and identifying the correlation abrupt change points of the motor includes:

[0201] Based on the trajectory curve of the acoustic-vibration fusion spectrum, the inflection point on the trajectory curve where the slope changes significantly is used as the division rule;

[0202] Based on the aforementioned division rules, the trajectory curve is divided into trajectory segments representing the acoustic-vibration coupling relationship of the motor.

[0203] Based on the overall extension direction of the acoustic-vibration coupling relationship trajectory segment and the basic trend of acoustic intensity changing with vibration intensity within the acoustic-vibration coupling relationship trajectory segment, the distribution law of the acoustic-vibration fusion spectrum is determined.

[0204] Based on the distribution pattern, the changing trends between the trajectories in the acoustic-vibration coupling relationship trajectory segment are compared one by one to obtain the correlation mutation point of the motor.

[0205] The method of diagnosing specific operating state faults of the motor based on the distribution pattern of the correlation mutation points includes:

[0206] The vibration intensity range of the acoustic-vibration fusion spectrum is divided into a normal background range and an abnormal potential abnormal range.

[0207] Observe the frequency of occurrence and clustering density of the relevant mutation points in the normal background interval and the abnormal potential anomaly interval respectively;

[0208] The frequency of occurrence and the clustering density are used to determine significant anomalies, thereby obtaining the target anomaly range of the motor.

[0209] Based on the position of the target abnormal region on the vibration intensity axis, locate the vibration characteristic components of the target abnormal region;

[0210] The vibration characteristic components are mapped to a preset fault-order table, and the fault type that matches the vibration characteristic components is determined in order to identify the specific operating state fault of the motor.

[0211] The specific operating state fault is output to the terminal of the motor to obtain a fault report of the motor.

[0212] The trajectory curves of the acoustic-vibration fusion spectrum are comprehensively traversed, and the tilt state of the curves is analyzed segment by segment. By comparing the tilt degree of adjacent curve segments, points where the tilt angle changes significantly are accurately identified. These points will cause significant changes in the extension direction of the trajectory curves, and they are clearly used as the core basis for the division rules to ensure that the inflection points can clearly define curve segments with different tilt characteristics.

[0213] Using the identified inflection points where the slope changes significantly as boundaries, starting from the beginning of the trajectory curve, the curve portion between two adjacent inflection points is sequentially divided into independent segments. The trajectory curve within each segment maintains a consistent inclination trend, without obvious abrupt changes in slope, and fully embodies a stable acoustic-vibration correlation characteristic. These independent segments are the acoustic-vibration coupling relationship trajectory segments of the motor.

[0214] A comprehensive feature analysis was conducted on each acoustic-vibration coupling trajectory segment, observing its overall extension direction in the spectral coordinate system to determine whether it exhibits an upward, downward, or horizontal extension trend. Simultaneously, the specific changes in acoustic intensity with vibration intensity within each trajectory segment were tracked point-by-point, summarizing basic trends such as the uniformity and rate of increase of intensity changes. By combining the overall extension direction and intensity change trends, the correlation pattern of acoustic-vibration intensity in the entire acoustic-vibration fusion spectrum was extracted, clarifying the distribution law of the acoustic-vibration fusion spectrum.

[0215] Based on the established distribution patterns, all acoustic-vibration coupling relationship trajectory segments are arranged sequentially, and the changing trends between adjacent trajectory segments are compared one by one. Special attention is paid to the difference between the extension direction and intensity change rate of the subsequent trajectory segment and the previous trajectory segment. When a trajectory segment's changing trend is found to deviate significantly from the previous trajectory segment and the overall distribution pattern, and this deviation exceeds the normal fluctuation range, the starting position corresponding to that trajectory segment is the motor's correlation mutation point.

[0216] Referring to the historical fluctuation range of vibration intensity in the acoustic-vibration fusion spectrum during normal motor operation, a fixed vibration intensity threshold is determined. This threshold is based on the maximum vibration intensity obtained from long-term normal motor operation data. The region in the acoustic-vibration fusion spectrum with vibration intensity below this threshold is defined as the normal background interval, which corresponds to the vibration intensity range when the motor is fault-free. The region with vibration intensity above this threshold is defined as the abnormal potential interval, which covers the vibration intensity range when the motor may experience a fault.

[0217] A range-based approach was used to comprehensively scan both the normal background range and the potential abnormal range, recording the specific location of each relevant abrupt change point within both ranges. The total number of relevant abrupt changes in the normal background range was counted to obtain the frequency of occurrence within that range; the number of abrupt changes per unit vibration intensity was calculated to obtain the cluster density. Using the same statistical method, the frequency of occurrence and cluster density of relevant abrupt changes in the potential abnormal range were obtained, ensuring complete consistency in the statistical standards for both ranges.

[0218] The frequency and density of occurrence in the normal background range are used as benchmark reference values, and compared with the corresponding values ​​in the potential abnormal range. When the frequency of occurrence in the potential abnormal range exceeds a fixed multiple of the benchmark reference value of the normal background range, and the density also exceeds a fixed multiple of the benchmark reference value of the normal background range, the potential abnormal range is determined to have a significant anomaly. The range determined to have a significant anomaly is clearly marked as the target abnormal range of the motor, ensuring that the target abnormal range can accurately correspond to the vibration intensity range where a fault may occur.

[0219] Extract the start and end vibration intensity values ​​of the target anomaly interval on the vibration intensity axis to determine the vibration intensity range covered by this interval. In the acoustic-vibration fusion spectrum, locate all waveform segments whose vibration intensities fall within this range, and extract the corresponding vibration characteristic parameters from these waveform segments. These parameters include core features such as the amplitude variation law and fluctuation period of the vibration. Integrate these extracted parameters into the vibration characteristic components of the target anomaly interval.

[0220] The preset fault-order table is established in advance based on the correlation between various known faults of the motor and their corresponding vibration characteristic components. The table clearly records the specific manifestations of the vibration characteristic components matched for each fault type. The vibration characteristic components of the extracted target abnormal interval are compared one by one with the records in the preset fault-order table to find the table entry that perfectly matches the vibration characteristic component. The fault type corresponding to this table entry is the specific operating state fault that the motor may have.

[0221] The system organizes information related to specific operational state faults, including fault type, corresponding target abnormal range, and details of vibration characteristic components. This organized information is then transmitted to the motor's terminal device via a data transmission module. Upon receiving the information, the terminal device automatically generates a fault report containing key fault information according to a preset report format, thus completing the output of the fault information.

[0222] The beneficial effect is that by using the inflection point where the slope of the acoustic-vibration fusion spectrum trajectory curve changes significantly as the division rule, the key turning point of the trajectory curve can be accurately captured, providing a clear and objective basis for subsequent trajectory segment division, avoiding subjective bias caused by manual division, and ensuring the uniformity and accuracy of the division standard.

[0223] Based on this division rule, the trajectory curve is divided into acoustic-vibration coupling relationship trajectory segments. The slope of each trajectory segment remains stable, fully carrying a specific acoustic-vibration correlation feature. This decomposes the originally continuous trajectory curve into multiple analysis units with clear characteristics, which facilitates targeted analysis of the acoustic-vibration coupling relationship at each stage.

[0224] By analyzing the overall extension direction of the acoustic-vibration coupling trajectory segment and the trend of acoustic intensity changing with vibration intensity, we can fully grasp the overall distribution pattern of the acoustic-vibration fusion spectrum, clearly define the correlation law of acoustic-vibration intensity under normal operating conditions, and provide a reliable benchmark for subsequent identification of abnormal changes.

[0225] Based on established distribution patterns, by comparing the changing trends of each acoustic-vibration coupling trajectory segment, abnormal trajectory segments that deviate from the overall pattern and the trends of adjacent trajectory segments can be accurately identified. These deviation points are the correlation mutation points. This trend comparison-based identification method can effectively capture the abrupt changes in acoustic-vibration correlation characteristics caused by faults, laying a core foundation for subsequent accurate location of fault intervals and diagnosis of specific fault types, and significantly improving the sensitivity and accuracy of fault identification.

[0226] The vibration intensity range of the acoustic-vibration fusion spectrum is divided into a normal background range and an abnormal potential range. The boundary is clearly defined based on the historical vibration intensity data of the motor during normal operation, so that the normal and potentially faulty vibration intensity ranges can be clearly distinguished. This lays the foundation for subsequent targeted analysis of the distribution of abrupt change points and avoids fault characteristics being masked by normal signals.

[0227] By observing the frequency and clustering density of related mutation points in the two intervals respectively, and by statistically analyzing the total number of mutation points and their distribution within a unit range, the abnormal characteristics of different intervals are quantified, ensuring that the analysis of mutation point distribution patterns is objective and quantifiable, and avoiding misjudgment or omission of faults due to subjective judgment alone.

[0228] Significant anomalies in frequency and cluster density are identified. Based on statistical data of the normal background range, the deviation of potential abnormal ranges is clarified by comparison. Areas with significantly abnormal distribution of mutation points are accurately selected as target abnormal ranges, ensuring that fault location focuses on the vibration intensity range where the real problem exists, thus improving the accuracy of fault diagnosis.

[0229] Based on the position of the target abnormal interval on the vibration intensity axis, the corresponding vibration characteristic components are located, and the abnormal interval is directly associated with the specific vibration characteristics. The core characteristic parameters of the vibration within the interval are extracted, so that the fault analysis has clear characteristic basis and avoids general judgments that are divorced from the actual vibration characteristics.

[0230] Vibration characteristic components are mapped to a preset fault-order table and matched with fault types. The preset table establishes the correlation between various faults and corresponding vibration characteristics in advance. Through precise matching, the specific operating state fault of the motor can be directly determined, which changes the traditional general mode of judging whether a fault exists or not. This makes the diagnostic results more targeted and provides a clear direction for precise maintenance.

[0231] The system outputs faults in specific operating states to the motor terminal and generates fault reports. It promptly presents the diagnostic results to staff in a standardized report format, clearly including key information such as fault type, corresponding abnormal range, and characteristic components. This allows staff to quickly grasp the fault situation and take targeted measures, improving equipment maintenance efficiency and ensuring the safe and stable operation of the motor.

[0232] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0233] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0234] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for generating a vibration-noise combined diagnosis of an operating state of an electric machine, characterized in that, The method comprises: S1. When the motor is running, the running state data of the motor is subjected to bimodal data extraction to obtain vibration signals and noise signals of the motor; S2. Order tracking analysis is performed on the vibration signals to obtain vibration characteristic order components of the motor, including: stripping out transient impact waveforms with amplitude mutations in the vibration signals and steady-state vibration waveforms reflecting smooth running states; analyzing the time interval pattern of the transient impact waveforms, and taking the impact waveforms with periodic changes in the time interval in the transient impact waveforms as a rotating synchronous reference pulse of the motor; establishing an angle-time corresponding curve of the instantaneous rotating phase in the motor according to the instantaneous time interval of the rotating synchronous reference pulse; based on the angle-time corresponding curve, resampling the steady-state vibration waveform to obtain an angular domain steady-state waveform of the steady-state vibration waveform; locating a specific mechanical angle position corresponding to a waveform peak value of the angular domain steady-state waveform within a complete rotating period of the angular domain steady-state waveform; taking the steady-state vibration amplitude at the specific mechanical angle position as the vibration characteristic order component of the motor; S3. Time domain energy envelope lines of the vibration characteristic order components are calculated, and the noise signals are subjected to synchronous time domain segmentation with the time domain energy envelope lines as a reference benchmark to obtain noise signal slices of the motor; S4. Based on the frequency of the vibration characteristic order component, non-related components in the noise signal slices are filtered out, and according to the results after filtering, a vibration synchronous noise component synchronized with the vibration characteristic order component and the current rotating period of the motor is extracted, including: in the frequency domain, identifying narrow-band spectra with significant energy concentration in the vibration characteristic order component to obtain a dominant oscillation mode of the vibration characteristic order component; determining a frequency selection template of the vibration characteristic order component according to the center frequency and bandwidth of the dominant oscillation mode; applying the frequency selection template to the spectral representation of the noise signal slices so that the components in the noise signal slice spectrum that coincide with the frequency selection template are retained to obtain a preliminary filtered noise signal slice of the motor; suppressing the parts of the preliminary filtered noise signal slice that are outside the frequency selection template to obtain a standard filtered noise signal slice of the motor; S5. The vibration characteristic order component is taken as the abscissa, and the vibration synchronous noise component is taken as the ordinate to construct a sound-vibration fusion spectrum of the motor; S6. The distribution law of sound-vibration intensity in the sound-vibration fusion spectrum is analyzed, the correlation mutation points of the motor are identified, and based on the distribution mode of the correlation mutation points, specific running state faults of the motor are diagnosed, including: According to the trajectory curve of the sound-vibration fusion spectrum, taking the inflection point on the trajectory curve where the slope changes significantly as a division rule; based on the division rule, the trajectory curve is divided into sound-vibration coupling relationship trajectory segments of the motor; According to the overall extension direction of the acoustic-vibration coupling relationship trajectory segment and the basic trend of acoustic intensity changing with vibration intensity in the acoustic-vibration coupling relationship trajectory segment, the distribution rule of the acoustic-vibration fusion frequency spectrum is determined; Based on the distribution rule, the change trend between trajectories in the acoustic-vibration coupling relationship trajectory segment is compared one by one to obtain the correlation mutation point of the motor; The vibration intensity range of the acoustic-vibration fusion frequency spectrum is divided into a normal background interval and an abnormal potential abnormal interval; The occurrence frequency and aggregation density of the correlation mutation point in the normal background interval and the abnormal potential abnormal interval are observed respectively; The occurrence frequency and the aggregation density are subjected to significant abnormality determination to obtain a target abnormal interval of the motor; According to the position of the target abnormal interval on the vibration intensity axis, a vibration characteristic component of the target abnormal interval is located; The vibration characteristic component is mapped to a preset fault-order table, and a fault type matching the vibration characteristic component is determined to determine a specific operating state fault of the motor; The specific operating state fault is output to a terminal of the motor to obtain a fault report of the motor.

2. The vibration and noise combined diagnosis generation method for motor operation state according to claim 1, characterized in that, The double-mode data extraction of the operating state data of the motor during motor operation obtains the vibration signal and the noise signal of the motor, including: During motor operation, the physical excitation generated by the operating components in the motor is synchronously collected to obtain the original vibration waveform and the original noise waveform of the motor; Based on the rated working frequency of the motor, the original vibration waveform is subjected to band-pass filtering to obtain the initial vibration signal of the motor; Based on the rated working frequency of the motor, the original noise waveform is subjected to high-pass filtering to obtain the initial noise signal of the motor; The initial vibration signal and the initial noise signal are subjected to time axis calibration to obtain the vibration signal and the noise signal of the motor.

3. The vibration and noise combined diagnosis generation method for motor operation state according to claim 1, characterized in that, The time-domain energy envelope of the vibration characteristic order component is calculated, including: The vibration characteristic order component is serialized to obtain a time sequence amplitude sequence of the motor; The continuous data points in the time sequence amplitude sequence are intercepted through a fixed-length sliding interval to form an analysis window of the motor; The analysis window is slid along the time sequence amplitude sequence to calculate an energy representation value of the analysis window, wherein the calculation formula of the energy representation value is as follows: ； wherein, is the energy characterization value, is the average amplitude of the data points within the analysis window, is the total number of data points within the analysis window, is a preset morphology adjustment factor, is the difference between the amplitude of the i-th data point within the analysis window and the average amplitude, is the difference between the amplitude of the i-th data point within the analysis window and the average amplitude. The energy representation values are connected in time sequence to obtain the time-domain energy envelope of the vibration characteristic order component.

4. The vibration and noise combined diagnosis generation method of a motor operating state according to claim 3, characterized in that, The noise signal is synchronously divided in time domain with the time-domain energy envelope as a reference benchmark to obtain a noise signal slice of the motor, including: On the time-domain energy envelope, the interval in which the energy point in the time-domain energy envelope rises from a local minimum value until reaching the next local minimum value is defined as an energy rising segment of the time-domain energy envelope; In the energy rising segment, an inflection point where the amplitude growth rate of the time-domain energy envelope changes from fast to slow is located; Centering on the time corresponding to the inflection point, the time offset is extended to the front and back of the inflection point to obtain an associated feature window of the inflection point; In the noise signal, a noise waveform segment corresponding to the associated feature window is intercepted as a noise signal slice of the motor.

5. The vibration and noise combined diagnosis generation method for motor operation state according to claim 1, characterized in that, The vibration synchronous noise component synchronized with the vibration characteristic order component and the current rotation period of the motor is extracted according to the filtered result, including: Collecting a time domain waveform between the standard filtered noise signal slice and the vibration characteristic order component; Taking the time domain waveform as a reference, a reference time anchor point of waveform rotation period start in the vibration characteristic order component is determined; According to the reference time anchor point, the waveform of the standard filtered noise signal slice is periodically synchronized and divided to obtain a noise period sequence of the standard filtered noise signal slice; According to the reference time anchor point, the noise period sequence is waveform-aligned to obtain an aligned period sequence of the standard filtered noise signal slice; In the same time, the waveform features stably appearing in the aligned period sequence are emphasized, and the waveform features randomly changing in the aligned period sequence are weakened to obtain an average period waveform of the standard filtered noise signal slice; According to the original time sequence of the reference time anchor point, the average period waveform is time-sequenced to obtain a complete time domain signal of the average period waveform; The complete time domain signal is taken as the vibration synchronous noise component of the vibration characteristic order component.

6. The method of claim 1, wherein, The vibration characteristic order component is taken as the abscissa, and the vibration synchronous noise component is taken as the ordinate to construct the sound-vibration fusion spectrum of the motor, including: On the time axis, a corresponding relationship between the amplitude sequence of the vibration characteristic order component and the amplitude sequence of the vibration synchronous noise component is established; According to the corresponding relationship, the amplitude of the vibration characteristic order component is mapped to the vibration intensity abscissa coordinate value of the motor; According to the corresponding relationship, the amplitude of the vibration synchronous noise component is mapped to the acoustic intensity ordinate coordinate value of the motor; The vibration intensity abscissa coordinate value and the acoustic intensity ordinate coordinate value are standardized and paired to obtain a two-dimensional coordinate point of the motor; The two-dimensional coordinate point is graphically reconstructed to obtain the sound-vibration fusion spectrum of the motor.

Images