A rotating machinery keyphasor and vibration monitoring system supporting multiple sensors

Abstract

The application relates to a rotating machinery key phase and vibration monitoring system supporting multiple sensors, and belongs to the technical field of rotating machinery monitoring. The system converts and outputs standard signals by normalizing and conditioning different key phase signals and vibration sensing signals through a sensing module; the data module pre-processes and data frame caches the standard signals, and outputs standardized original data; the algorithm processing module integrates a rotating machinery key phase vibration comprehensive analysis model, performs time domain-frequency domain decomposition, feature extraction, feature calculation, threshold comparison, trend analysis and data correction on the standardized original data, generates key phase state representation data, vibration feature quantization data and abnormal state judgment results; and the interactive module encodes and adapts the above data and judgment results, and generates standard data frames and control instructions. The system can adapt to multiple sensors, improve the accuracy and reliability of monitoring data, and provide strong support for the operation and maintenance of rotating machinery.

Description

Technical Field

[0001] This invention belongs to the field of rotating machinery monitoring technology, specifically relating to a rotating machinery key phase and vibration monitoring system that supports multiple sensors. Background Technology

[0002] Rotating machinery, as core equipment in many key fields such as industrial production, energy and power, aerospace, and rail transportation, encompasses various types including motors, fans, pumps, steam turbines, and compressors. The stability and safety of its operating status directly determine the continuity of production processes, production efficiency, and equipment safety. If abnormalities occur in rotating machinery and are not detected in time, they can easily lead to equipment failure, shutdowns, and even personal injury and significant property damage. Key phase monitoring and vibration monitoring are core technologies for diagnosing the operating status of rotating machinery and predicting equipment failures. Key phase monitoring can accurately acquire key parameters such as the rotation period and phase shift of the rotating machinery, reflecting the rotational stability of the equipment; vibration monitoring can capture parameters such as vibration amplitude and frequency, identifying potential faults such as bearing wear, rotor imbalance, and misalignment. The combination of these two technologies enables a comprehensive and accurate assessment of the operating status of rotating machinery.

[0003] Currently, existing key phase and vibration monitoring systems for rotating machinery still suffer from numerous technical shortcomings, making it difficult to meet the high-precision and high-adaptability requirements of industrial production for equipment monitoring. Firstly, the problem of poor adaptability is particularly prominent. Existing monitoring systems mostly employ dedicated sensor interface designs, only compatible with a single type and single-format sensor. However, rotating machinery from different manufacturers and of different specifications in industrial settings is equipped with varying types of key phase and vibration sensors, and their output signal formats also differ (such as TTL level signals, analog signals, digital signals, etc.). This results in existing monitoring systems being incompatible with multiple sensors. If monitoring rotating machinery of different specifications is required, the entire monitoring system must be replaced or additional adapter modules must be equipped, increasing monitoring costs and reducing monitoring efficiency, making it difficult to achieve centralized monitoring of rotating machinery of multiple specifications and operating conditions. Secondly, the signal processing accuracy is insufficient. The signal processing flow of the existing monitoring system is relatively simple, lacking a complete signal separation, feature calculation, and data correction mechanism. Since the key phase signal and vibration signal are often transmitted together, the existing system is unable to achieve accurate separation of the two, which easily leads to signal interference problems. At the same time, the feature parameter extraction process lacks scientific calculation methods and does not have an effective data correction step, resulting in large errors in the extracted key phase feature parameters and vibration feature parameters. In the process of threshold comparison and trend analysis, it is easily affected by random noise and systematic errors, resulting in large deviations in the judgment of abnormal states. It is impossible to capture subtle operational abnormalities of rotating machinery in a timely and accurate manner, making it difficult to achieve early prediction of faults. Furthermore, data transmission adaptability is poor. Existing systems mostly use fixed communication protocols, which cannot be adapted to the communication protocols (such as TCP / IP, RS485, Modbus, etc.) of different external devices (such as host computers, PLC controllers, remote monitoring terminals). This results in monitoring data not being able to be uploaded to external devices in real time and smoothly, and control commands not being able to be issued accurately, affecting the remote control and centralized management capabilities of the monitoring system. At the same time, the threshold settings of existing systems are mostly fixed values, which cannot be flexibly adjusted according to the type, specifications, and working conditions of rotating machinery, further reducing the adaptability and practicality of the monitoring system.

[0004] With the continuous improvement of industrial intelligence, industrial sites are placing higher demands on the adaptability, monitoring accuracy, data processing efficiency, and practicality of rotating machinery monitoring systems. Existing monitoring systems suffer from poor adaptability, low monitoring accuracy, inadequate data processing and transmission, and inflexible threshold settings, failing to meet actual monitoring needs. Therefore, addressing the numerous shortcomings of existing rotating machinery key phase and vibration monitoring systems, the development of a new system that is adaptable to various sensors, provides accurate monitoring, efficient data processing, strong adaptability, and high practicality is crucial. This system aims to solve the pain points of existing technologies, achieve comprehensive, accurate, and real-time monitoring of the operating status of rotating machinery, and provide reliable support for equipment fault prediction and maintenance. This has become a significant technical problem to be solved in the field of rotating machinery monitoring technology. Summary of the Invention

[0005] To address the aforementioned problems in the prior art, this invention provides a rotating mechanical key phase and vibration monitoring system that supports multiple sensors. The objective of this invention can be achieved through the following technical solutions: A rotating mechanical key phase and vibration monitoring system supporting multiple sensors includes: a sensing module, a data module, an algorithm processing module, and an interaction module; The sensing module normalizes and conditions different types of bond phase signals and vibration sensing signals, and outputs a standard signal. Based on the standard signal, the data module completes the standardization and storage of the original monitoring data through preprocessing and data frame caching mechanisms, and outputs standardized original data. The algorithm processing module integrates a comprehensive analysis model for the vibration of the key phase of rotating machinery. It performs time-domain-frequency domain decomposition on the standardized raw data to separate the key phase signal from the vibration signal and preliminarily extracts features. It then performs feature calculation on the key phase signal and the vibration signal, extracts key phase feature parameters and vibration feature parameters, performs threshold comparison and trend analysis, and generates key phase state characterization data, vibration feature quantification data and abnormal state judgment results after data correction. The interaction module generates standard data frames and control commands based on the bond phase state characterization data, vibration characteristic quantification data, and abnormal state judgment results through a data encoding adaptation process.

[0006] Specifically, the normalization conditioning conversion process includes: Different types of key phase signals and vibration sensing signals are subjected to amplitude calibration, level conversion and impedance matching processing to uniformly condition them into collectable signals with a preset amplitude range and a unified level standard, while filtering out noise interference to generate the standard signal.

[0007] Specifically, the preprocessing process includes: The standard signal is subjected to adaptive filtering, signal de-jittering and outlier removal processing to filter out random noise and sudden interference signals and remove abnormal data points generated during the acquisition process. Then, through the data frame caching mechanism, the preprocessed signal is regularized, encoded, and stored according to the preset data frame format to generate the standardized raw data.

[0008] Specifically, the time-domain-frequency-domain decomposition process includes: The comprehensive analysis model for the key phase vibration of rotating machinery performs time-domain signal smoothing on the standardized raw data, converts the time-domain signal into a frequency-domain signal through Fourier transform, sets a filtering threshold based on the frequency domain characteristic differences between the key phase signal and the vibration signal, performs frequency band filtering on the frequency-domain signal, separates the frequency bands corresponding to the key phase signal and the frequency bands corresponding to the vibration signal, and finally converts the separated frequency-domain signal back into a time-domain signal through inverse Fourier transform, thus separating the key phase signal and the vibration signal.

[0009] Specifically, the preliminary feature extraction process includes: Time-domain waveform analysis was performed on the separated bond phase signal and vibration signal to obtain the peak point, valley point and zero crossing point of the time-domain signal, and the time-domain peak value was calculated. Frequency domain spectral analysis is performed to identify the fundamental frequency peak in the frequency domain spectrum and extract the fundamental frequency in the frequency domain. The extracted two types of basic features, namely the time domain peak and the frequency domain fundamental frequency, are then standardized and normalized.

[0010] Specifically, the process of performing feature extraction on the bond phase signal and vibration signal includes: Based on the preliminary extracted basic features of the key phase signal, the time-domain waveform of the key phase signal is periodically identified, the time interval between two adjacent key phase pulse trigger points is calculated to obtain the key phase period; the difference between the trigger phase of the key phase signal and the preset reference phase is obtained to obtain the key phase offset; the rising edge and falling edge of the key phase pulse are identified, the time span between the rising edge and the falling edge is calculated to obtain the key phase pulse width; the time interval difference between consecutive key phase pulse trigger points is statistically analyzed to obtain the key phase trigger interval. The bond phase period, bond phase offset, bond phase pulse width, and bond phase trigger interval are extracted as bond phase characteristic parameters.

[0011] Specifically, the process of feature decomposition of the vibration signal includes: Based on the preliminary extracted basic characteristics of the vibration signal, the time-domain waveform of the vibration signal is scanned to capture the maximum and minimum vibration amplitudes, and the difference is calculated to obtain the vibration amplitude. The fundamental frequency and harmonic frequencies of the vibration signal are identified through frequency domain spectrum analysis, and the fundamental frequency with the highest proportion is selected as the vibration frequency. The amplitude of each harmonic component in the frequency domain spectrum is quantized, and the amplitude proportion of each harmonic is extracted as the vibration harmonic component. The ratio of the peak value to the effective value of the vibration signal is calculated to obtain the vibration peak factor. Vibration amplitude, vibration frequency, vibration harmonic components, and vibration peak factor are extracted as vibration characteristic parameters.

[0012] Specifically, the comprehensive analysis model for the vibration of the rotating mechanical key phase has a built-in preset threshold database, which includes: The preset threshold database is classified and stored according to the type, specifications and working conditions of rotating machinery. Each category stores the threshold range of key phase characteristic parameters and the threshold range of vibration characteristic parameters under normal working conditions. When performing threshold comparison, the rotating machinery key phase vibration comprehensive analysis model matches the corresponding threshold range from the preset threshold database based on the currently monitored rotating machinery type, specifications and operating conditions, and then compares the extracted key phase feature parameters and vibration feature parameters with the matched threshold range one by one.

[0013] Specifically, the process of trend analysis includes: The length and step size of the preset sliding window are used to arrange the continuously collected bond phase characteristic parameters and vibration characteristic parameters in chronological order. The mean, variance and linear fitting of the characteristic parameters within the window are calculated in units of the sliding window to obtain the changing trend of the characteristic parameters within the window. The changing trend of the current window is compared with the preset law. The window is slid sequentially and the calculation and comparison process is repeated.

[0014] Specifically, the process for generating the abnormal state judgment result includes: The key phase characteristic parameters and vibration characteristic parameters are compared one by one with the threshold ranges matched in the preset threshold database. Based on the trend analysis results, when parameter abnormalities or trend abnormalities occur, the rotating machinery is determined to be in an abnormal state. The abnormality type is marked according to the abnormal characteristic parameter type. Based on the magnitude of the parameters exceeding the threshold and the degree to which the trend deviates from the preset pattern, the degree of abnormality is quantified and marked, and the abnormal state judgment result is generated.

[0015] Specifically, the data correction process includes: Various error models for the feature solving process are pre-stored, including errors caused by signal noise and system errors caused by algorithm calculation; After the feature calculation is completed, error detection is performed on the extracted bond phase feature parameters and vibration feature parameters to identify error components; Based on the error model, the detected error is compensated in reverse, and the feature parameters are corrected.

[0016] Specifically, the data encoding adaptation process includes: After receiving the key phase state characterization data, vibration characteristic quantification data, and abnormal state judgment results, the system performs encoding processing using a preset encoding protocol to convert them into binary data frames, which are the standard data frames. Based on the communication protocol type, the system performs protocol matching between the standard data frames and control commands.

[0017] The beneficial effects of this invention are as follows: This system normalizes and conditions key phase signals and vibration sensing signals of different formats through a sensing module, enabling the access and adaptation of various types of sensors. It can meet the key phase and vibration monitoring needs of rotating machinery of different specifications and operating conditions without replacing the monitoring system, thus improving the system's versatility and applicability.

[0018] The integrated analysis model of key phase vibration of rotating machinery, which is integrated into the algorithm processing module, achieves accurate separation of key phase signal and vibration signal through time-frequency domain decomposition. Combined with a complete feature calculation and data correction mechanism, it effectively eliminates the impact of signal interference and error on monitoring results, improves the accuracy of key phase feature parameter and vibration feature parameter extraction, ensures the accuracy of abnormal state judgment, and can promptly capture subtle operational abnormalities of rotating machinery.

[0019] The system constructs a complete processing flow of "signal conditioning - data preprocessing - deep analysis - encoding adaptation". The data module ensures the quality of the original data through preprocessing methods such as adaptive filtering and outlier removal. The algorithm processing module achieves deep analysis of the data through multi-step collaborative processing. The interaction module achieves smooth transmission of data and instructions through encoding adaptation. The collaborative operation of each module improves data processing efficiency and transmission reliability. Attached Figure Description

[0020] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 This is a system architecture diagram of a rotating mechanical key phase and vibration monitoring system supporting multiple sensors according to the present invention; Figure 2 This is a data flow diagram of the algorithm processing module in this invention; Figure 3 This is a graph showing the relationship between feature parameters and results in this invention. Detailed Implementation

[0022] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0023] Please see Figures 1-3 A rotating mechanical key phase and vibration monitoring system supporting multiple sensors includes: a sensing module, a data module, an algorithm processing module, and an interaction module; The sensing module normalizes and conditions different types of bond phase signals and vibration sensing signals, and outputs a standard signal. Based on the standard signal, the data module completes the standardization and storage of the original monitoring data through preprocessing and data frame caching mechanisms, and outputs standardized original data. The algorithm processing module integrates a comprehensive analysis model for the vibration of the key phase of rotating machinery. It performs time-domain-frequency domain decomposition on the standardized raw data to separate the key phase signal from the vibration signal and preliminarily extracts features. It then performs feature calculation on the key phase signal and the vibration signal, extracts key phase feature parameters and vibration feature parameters, performs threshold comparison and trend analysis, and generates key phase state characterization data, vibration feature quantification data and abnormal state judgment results after data correction. The interaction module generates standard data frames and control commands based on the bond phase state characterization data, vibration characteristic quantification data, and abnormal state judgment results through a data encoding adaptation process.

[0024] Specifically, the normalization conditioning conversion process includes: Amplitude calibration is performed on key phase signals and vibration sensing signals of different standards. A preset calibration coefficient is used to correct the signal amplitude, eliminating errors caused by differences in the output amplitude of different sensors and ensuring that the amplitudes of various signals are within the same order of magnitude. Subsequently, level conversion is performed to uniformly convert signals of different level standards (such as TTL level and analog level) to a preset standard level, adapting to the data module's acquisition interface requirements. Next, impedance matching is performed, adjusting the output impedance of the sensing module and the input impedance of the data module to make them consistent, reducing reflection loss during signal transmission and improving signal transmission efficiency. During the above processing, a low-pass filter circuit simultaneously filters high-frequency noise interference in the signal, eliminating interference signals unrelated to key phase and vibration monitoring. Finally, a standard signal with stable amplitude, uniform level, and no obvious noise is generated, ensuring the accuracy of subsequent data acquisition.

[0025] Specifically, the preprocessing process includes: The standard signal undergoes adaptive filtering, with the filtering coefficients adjusted in real time according to the signal noise intensity. This automatically identifies and filters random noise while retaining the effective components of the bond phase and vibration signals, preventing over-filtering that could lead to signal loss. Subsequently, signal de-jittering is performed, with a reasonable de-jittering time threshold set to suppress high-frequency jitter and instantaneous fluctuations, ensuring a stable signal waveform. An outlier removal algorithm then identifies and removes abnormal data points generated during acquisition due to poor sensor contact, sudden interference, or other factors, preventing abnormal data from affecting subsequent processing results. After the above preprocessing, the preprocessed signal is organized according to a preset data frame format using the data frame caching mechanism. The signal data is grouped, encoded, and stored in the data module's cache unit. The cache unit uses a circular storage method to ensure the continuity and integrity of data storage, ultimately generating the standardized raw data that can be directly used for algorithm processing.

[0026] Specifically, the time-domain-frequency-domain decomposition process includes: The comprehensive analysis model for the bond phase vibration of rotating machinery performs time-domain signal smoothing on the standardized raw data. A moving average algorithm is used to scan the time-domain signal, eliminating high-frequency glitches and random fluctuations, resulting in a smoother waveform for easier feature identification. Subsequently, a fast Fourier transform algorithm is used to convert the smoothed time-domain signal into a frequency-domain signal, obtaining the frequency spectrum, which clearly shows the frequency distribution characteristics of the signal. Based on the frequency domain differences between the bond phase signal and the vibration signal (bond phase signals are mostly concentrated in the low-frequency band, while vibration signals have a relatively wider frequency distribution), a reasonable filtering threshold is preset to segment the frequency spectrum, accurately separating the frequency bands corresponding to the bond phase signal and the vibration signal, avoiding mutual interference between the two types of signals. Finally, an inverse Fourier transform is used to convert the separated frequency-domain signals back to the time-domain signal, obtaining pure bond phase and vibration signals, providing a high-quality signal foundation for subsequent feature extraction and feature calculation.

[0027] Specifically, the preliminary feature extraction process includes: The time-domain waveform analysis was performed on the bond phase signal and vibration signal obtained after time-domain-frequency domain decomposition. The peak points, valley points, and zero-crossing points of the time-domain signals were captured by the waveform recognition algorithm, and the time coordinates and amplitude information of each feature point were recorded. Based on the amplitude data of the peak points and valley points, the time-domain peak values ​​of the two types of signals were calculated, which intuitively reflects the intensity characteristics of the signals. Subsequently, the frequency domain spectrum analysis was performed on the separated bond phase signal and vibration signal. The fundamental frequency peak in the frequency domain spectrum was identified by the spectral peak recognition algorithm, and the corresponding fundamental frequency was extracted to identify the main frequency components of the signal. Finally, the extracted time-domain peak and frequency-domain fundamental frequency features were standardized and normalized. The feature parameters were converted to a preset range by the normalization algorithm to eliminate the influence of parameters of different magnitudes. The normalized basic features were stored in a temporary feature cache unit to provide a reliable basis for subsequent feature calculation.

[0028] Specifically, the process of performing feature extraction on the bond phase signal and vibration signal includes: Based on the initially extracted basic features of the key phase signal, the time-domain waveform of the key phase signal is periodically identified. A periodic detection algorithm is used to scan the time-domain waveform, identify the trigger points of two adjacent key phase pulses (i.e., the trigger positions where the signal amplitude changes abruptly), and calculate the time interval between the two trigger points to obtain the key phase period, reflecting the rotation period of the rotating machinery. Subsequently, a phase detection algorithm is used to capture the trigger phase of the key phase signal, compare it with a preset reference phase, and calculate the difference between the two to obtain the key phase phase offset, reflecting the phase deviation of the rotating machinery. Next, an edge recognition algorithm is used to identify the rising and falling edges of the key phase pulse, record the time nodes of the rising and falling edges, and calculate the time span between the two time nodes to obtain the key phase pulse width, reflecting the duration of the key phase pulse. Finally, the time intervals between multiple consecutive key phase pulse trigger points are statistically analyzed, and the difference between adjacent time intervals is calculated to obtain the key phase trigger interval, reflecting the stability of the key phase pulse triggering. Combining the above calculation results, the key phase period, key phase phase offset, key phase pulse width, and key phase trigger interval are extracted as core key phase feature parameters to complete the feature decomposition of the key phase signal.

[0029] Specifically, the process of feature decomposition of the vibration signal includes: Based on the preliminary extracted fundamental features of the vibration signal, a peak detection algorithm is used to comprehensively scan the time-domain waveform of the vibration signal, capturing the maximum and minimum vibration amplitudes in the time-domain waveform. The difference between the two is calculated to obtain the vibration amplitude, reflecting the vibration intensity of the rotating machinery. Subsequently, a frequency domain spectrum analysis algorithm is used to analyze the frequency domain spectrum of the vibration signal, identifying the fundamental frequency and each harmonic frequency in the frequency domain spectrum. The amplitude proportion of each frequency component is statistically analyzed, and the fundamental frequency with the highest proportion is selected as the vibration frequency, clarifying the main frequency source of the vibration. Then, the amplitude proportions of each harmonic frequency component in the frequency domain spectrum are analyzed. The harmonic components are quantized for amplitude, and the ratio of the amplitude of each harmonic to the fundamental amplitude is calculated. The amplitude proportion of each harmonic is extracted as the vibration harmonic component, reflecting the harmonic distribution characteristics of the vibration signal. Finally, the effective value of the vibration signal is calculated using an effective value calculation algorithm. Combined with the previously extracted vibration time-domain peak value, the ratio of the peak value to the effective value is calculated to obtain the vibration peak factor, which reflects the impact characteristics of the vibration signal. Based on the above calculation results, the vibration amplitude, vibration frequency, vibration harmonic components, and vibration peak factor are extracted as core vibration characteristic parameters to complete the characteristic solution of the vibration signal.

[0030] Specifically, the comprehensive analysis model for the vibration of the rotating mechanical key phase has a built-in preset threshold database, which includes: The preset threshold database adopts a categorized storage structure, storing data hierarchically according to the type of rotating machinery (e.g., motors, fans, pumps), specifications (e.g., rated speed, power), and operating conditions (e.g., no-load, full-load, low-speed, high-speed). Each category stores the threshold ranges for key phase characteristic parameters and vibration characteristic parameters of the rotating machinery under normal operating conditions. These threshold ranges are calibrated using extensive experimental and field measurement data to ensure their rationality and accuracy. During threshold comparison, the rotating machinery key phase vibration comprehensive analysis model first obtains the currently monitored rotating machinery type, specifications, and operating conditions through an interactive module. Based on this information, it performs precise matching in the preset threshold database, retrieving the corresponding threshold ranges for key phase characteristic parameters and vibration characteristic parameters. Subsequently, the key phase characteristic parameters and vibration characteristic parameters extracted during feature calculation are compared one by one with the matched threshold ranges, recording the deviation of each parameter from the threshold range, providing accurate data support for subsequent abnormal state judgment.

[0031] Specifically, the process of trend analysis includes: Based on the monitoring accuracy requirements and data acquisition frequency of rotating machinery, the length and sliding step size of the sliding window are preset. The window length is set to cover multiple data acquisition cycles, and the sliding step size is set to be consistent with the data acquisition cycle to ensure the continuity and accuracy of trend analysis. Multiple sets of continuously acquired key phase characteristic parameters and vibration characteristic parameters are sorted in chronological order to form a time-series characteristic parameter sequence. Using the set sliding window as a unit, the mean of the characteristic parameters within the window is calculated to obtain the average level of the parameters within the window. Variance is calculated to reflect the dispersion of the parameters within the window and to determine the stability of the parameters. A linear fitting algorithm is used to fit the parameters within the window to obtain a linear trend line of parameter changes, clarifying the direction of parameter change. The current window's trend (average level, dispersion, linear trend) is compared with the preset pattern (parameter change pattern under normal operating conditions) to determine if there are any abnormal trends. Then, the window is moved according to the set sliding step size, and the above mean calculation, variance calculation, linear fitting, and comparison process is repeated to achieve real-time, dynamic monitoring of the continuous characteristic parameter change trends and timely capture of abnormal parameter change trends.

[0032] Specifically, the process for generating the abnormal state judgment result includes: The key phase feature parameters and vibration feature parameters extracted by feature calculation are compared one by one with the threshold ranges matched in the preset threshold database. If any parameter exceeds the corresponding threshold range (greater than the upper threshold or less than the lower threshold), it is marked as a parameter anomaly. At the same time, combined with the trend analysis results, if the trend of the feature parameter changes deviates from the preset law, and this deviation occurs in multiple consecutive sliding windows, excluding the influence of random factors, it is marked as a trend anomaly. When either parameter anomaly or trend anomaly occurs, the rotating machinery is determined to be in an abnormal state. Subsequently, according to the type of abnormal feature parameter, the anomaly type is clearly marked. If the key phase feature parameter is abnormal, it is marked as a key phase anomaly; if the vibration feature parameter is abnormal, it is marked as a vibration anomaly; if both are abnormal, it is marked as a mixed anomaly. Based on the magnitude of the parameter exceeding the threshold (the specific value exceeding the upper or lower threshold) and the degree of trend deviation from the preset law (the deviation from the linear trend line), a quantitative scoring method is used to classify the degree of anomaly (e.g., mild anomaly, moderate anomaly, severe anomaly). Finally, the anomaly state judgment result containing anomaly label, anomaly type, anomaly degree, and anomaly parameter information is generated, providing clear guidance for subsequent equipment maintenance.

[0033] Specifically, the data correction process includes: In the comprehensive analysis model of key phase vibration in rotating machinery, various possible error models that may arise during feature calculation are pre-stored, including random error models caused by signal noise and systematic error models generated during algorithm calculation. These error models are calibrated through a large amount of experimental data and can accurately describe the distribution patterns and impact levels of various errors. After feature calculation is completed, a data correction process is initiated to detect errors in the extracted key phase feature parameters and vibration feature parameters. An error identification algorithm, combined with the pre-set error model, identifies the error components (random or systematic errors) contained in the parameters and calculates the specific values ​​of the errors. Based on the identified error type and error value, the corresponding error compensation algorithm is invoked to perform reverse compensation for the detected errors, correct the feature parameters, and eliminate the impact of errors on parameter accuracy. After correction, the corrected feature parameters are validated to determine whether they are within a reasonable range. If the validation passes, the corrected parameters are used for subsequent threshold comparison and trend analysis. If the validation fails, error detection and compensation are repeated until the parameters meet the validation requirements, ensuring the accuracy of the final generated key phase state characterization data and vibration feature quantification data.

[0034] Specifically, the data encoding adaptation process includes: The system receives key phase state characterization data, vibration characteristic quantification data, and abnormal state judgment results output by the receiving algorithm processing module. It then classifies and organizes these data types, grouping them by data type to ensure data order. Subsequently, it uses a preset industrial standard encoding protocol (such as Modbus encoding protocol) to encode the classified data, converting decimal data into binary data frames that meet transmission requirements. Frame headers, trailers, and checksums are added to the data frames to form the standard data frames, ensuring the integrity and reliability of data transmission. Simultaneously, a protocol identification module identifies the communication protocol type (such as TCP / IP protocol, RS485 protocol) of external devices (host computer, control terminal). Based on the identification results, it performs protocol matching on the standard data frames and control commands, adjusting the data frame format and transmission rate to fully adapt to the communication protocols of external devices. This ensures that the generated standard data frames and control commands can accurately interface with external host devices and control terminals, guaranteeing smooth data interaction and command transmission, and enabling real-time uploading of monitoring data and real-time distribution of control commands.

[0035] This embodiment takes rotating machinery as the monitoring object, and is based on the rotating machinery key phase and vibration monitoring system supporting multiple sensors described in this invention, including the module architecture, sensor adaptation type, communication protocol and error requirements of the EST-8500D series monitoring and protection device.

[0036] I. System Overall Architecture The rotating machinery key phase and vibration monitoring system supporting multiple sensors described in this embodiment adopts a modular architecture design, consistent with the chassis slot layout and module connection method of the EST-8500D series monitoring and protection device. It includes a sensing module, a data module, an algorithm processing module, and an interaction module. Each module achieves signal connection through the I / O slots on the chassis back panel, eliminating the need for additional internal wiring. Connection to external sensors and upper-level devices is achieved solely through the terminal blocks on the module's rear panel, adapting to industrial field installation and maintenance needs. The core system workflow is as follows: the sensing module performs normalization conditioning and conversion of signals of different formats; the data module performs data preprocessing and caching; the algorithm processing module performs data calculation and analysis through its built-in model; and the interaction module performs data encoding adaptation and transmission.

[0037] II. Core Working Process of the Module 2.1 Working process of the sensing module The sensing module is compatible with various types of sensors. The sensor adaptation function of modules such as 8500D-WY / ZZD, 8500D-ZS / JX-1, and 8500D-ZD-1 in the EST-8500D series can receive key phase signals and vibration sensing signals of different formats, including eddy current sensors, magnetoresistive speed sensors, and vibration velocity sensors. It performs normalization, conditioning, and conversion on various signals, as follows: Let the amplitude of the input key phase signal be A1 and the amplitude of the vibration sensing signal be A2. A1 and A2 are calibrated using preset calibration coefficients k1 and k2 respectively. After calibration, the amplitudes are A1' = A1 × k1 and A2' = A2 × k2, ensuring that the amplitudes of the two types of signals are on the same order of magnitude. Then, level conversion is performed to uniformly convert the calibrated signals to a standard level U0. Simultaneously, the impedance is adjusted using the impedance matching formula Z1 = Z2 (Z1 is the output impedance of the sensing module, and Z2 is the input impedance of the data module) to filter high-frequency noise interference, outputting a unified standard signal for transmission to the data module. Meanwhile, the sensing module has four built-in -24VDC power outputs with overcurrent and short-circuit protection functions, providing operating power for external active sensors, consistent with the power output characteristics of the EST-8500D series functional modules.

[0038] 2.2 Data Module Working Process The data module receives the standard signal output from the sensor module, completes preprocessing and data frame buffering, and the data acquisition and preprocessing logic based on the EST-8500D series device is as follows: Let the standard signal be S(t). An adaptive filtering algorithm is used to filter S(t). The filtering coefficient α is adjusted in real time according to the signal noise intensity. The filtered signal is S1(t) = S(t) × α + (1-α) × S(t-1) (S(t-1) is the signal at the previous moment). Then, signal de-jitter processing is performed. A de-jitter time threshold T0 is set (the default alarm delay of the EST-8500D series module can be adjusted within the range of 0~255 seconds). When the signal fluctuation time is less than T0, it is judged as a jitter signal and removed, resulting in a stable signal S2(t). An outlier removal algorithm is used to set an outlier threshold range [U1, U2]. When S2(t) exceeds this range, it is judged as abnormal data and removed. After preprocessing, it is standardized according to the preset data frame format (the communication data frame format of the EST-8500D series device includes frame header, frame tail, data identifier, signal data and check code) to obtain standardized raw data X(t), which is transmitted to the algorithm processing module.

[0039] 2.3 Algorithm Processing Module Working Process The algorithm processing module integrates a comprehensive analysis model for the vibration of key phases in rotating machinery. The core process revolves around time-frequency domain decomposition, feature extraction, data correction, threshold comparison, and trend analysis, as detailed below: (1) Time-frequency domain decomposition: Let the standardized original data be X(t). First, the moving average algorithm is used to smooth X(t) in the time domain. The sliding window length is N. The smoothed signal is X̄(t)=[X(t-N+1)+X(t-N+2)+...+X(t)] / N. Then, the time domain signal X̄(t) is converted into the frequency domain signal X(ω) by the Fast Fourier Transform (FFT), where ω is the angular frequency, X(ω)=∫X̄(t)e^(-jωt)dt (j is the imaginary unit). Based on the difference in frequency domain characteristics between the bond phase signal and the vibration signal, a preset filtering threshold ω0 is set. When ω<ω0, it is determined to be the bond phase frequency domain signal X. k (ω), when ω≥ω0, is determined to be a vibration frequency domain signal Xᵥ(ω); X is then transformed by inverse fast Fourier transform (IFFT). k Xᵥ(ω) and Xᵥ(ω) are converted back to time domain signals to obtain the bond phase signal K(t) and vibration signal V(t), thus achieving the separation of the two types of signals.

[0040] (2) Preliminary feature extraction: Time-domain waveform analysis is performed on the key phase signal K(t) to capture the peak point K. max Valley point K m ᵢ n Calculate the peak value K in the time domain p =K max -K m ᵢ n; Perform frequency domain spectral analysis on K(t) to identify the angular frequency ω corresponding to the peak value of the fundamental frequency in the frequency domain spectrum. k Extract the fundamental frequency f k =ω k / (2π). Using the same method, the time-domain peak value V is calculated for the vibration signal V(t). p =V max -V m ᵢ n Extract the fundamental frequency fᵥ=ωᵥ / (2π); and then use K p f k V p The normalization formula for fᵥ is K. p '=(K p -K pm ᵢ n ) / (K pmax -K pm ᵢ n V p '=(V p -V pm ᵢ n ) / (V pmax -V pm ᵢ n )(K pmax K pm ᵢ n V pmax V pm ᵢ n (For the preset normalization range extreme values), the normalized basic characteristic parameters are obtained.

[0041] (3) Characteristic calculation: ① Calculation of key phase characteristic parameters: Let the pulse trigger points of the key phase signal K(t) be t1, t2, t3...t n The time interval between adjacent trigger points is Δtᵢ=tᵢ +1 -tᵢ(i=1,2,...,n-1), bond phase period T k =∑Δtᵢ / (n-1); Let the preset reference phase be φ0, the key phase trigger phase be φᵢ, and the key phase offset Δφ = φᵢ - φ0; Identify the key phase pulse rise time tᵣ and fall time t f The bond phase pulse width τ = t f -tᵣ;Continuous trigger interval difference Δτᵢ=Δtᵢ +1 -Δtᵢ, we obtain the bond phase triggering interval Δτ=∑Δτᵢ / (n-2), and finally extract T. k Δφ, τ, and Δτ are used as key phase characteristic parameters. ② Calculation of vibration characteristic parameters: Let the maximum amplitude of the vibration signal V(t) be V. max The minimum amplitude is V m ᵢ nThe vibration amplitude Aᵥ=V max -V m ᵢ n Identify the fundamental frequency fᵥ and harmonic frequencies fᵥ of the vibration signal. n (n=2,3,...), the amplitude of each harmonic is Aᵥ n Vibration harmonic component η n =Aᵥ n / Aᵥ;Calculate the effective value Vᵣ of the vibration signal ms =√[∫V(t) 2 [dt / T] (T is the signal period), vibration peak factor C p =V max / Vᵣ ms Finally, Aᵥ, fᵥ, and η are extracted. n C p As a vibration characteristic parameter.

[0042] (4) Data correction: Refer to the output error requirements of the EST-8500D series modules (error less than ±1.0% of full scale), preset the error model, and set the random error ε1~N(0,σ) 2 (σ is the standard deviation of the error), the systematic error ε2 = k × X(t) + b (k and b are error coefficients); let the parameter obtained by feature calculation be P, the detection parameter error ε = P - P0 (P0 is the standard parameter value), separate the random error ε1 and the systematic error ε2; correct it through the error compensation algorithm, and the corrected parameter P' = P - ε1 - ε2; set the error allowable threshold ε0 (ε0 ≤ ±1.0% of full scale), when |P' - P0| ≤ ε0, the verification is passed and the threshold comparison stage is entered; otherwise, the error detection and compensation are repeated to ensure that the parameter accuracy meets the requirements of industrial monitoring.

[0043] (5) Threshold comparison and trend analysis: A preset threshold database is used, referencing the parameter setting logic of the EST-8500D series device, to classify and store thresholds according to rotating machinery type, specifications, and working conditions. The threshold range for key phase characteristic parameters is [T]. km ᵢ n ,T kmax ]、[Δφ m ᵢ n ,Δφ max The threshold range for vibration characteristic parameters is [Aᵥ]. m ᵢ n ,Aᵥ max ]、[fᵥ m ᵢ n ,fᵥ max The threshold values ​​can be adjusted online via a host computer or the interactive module's touchscreen, consistent with the parameter configuration functions of the EST-8500D series modules. The corrected parameter P' is compared with the corresponding threshold range; if P'∉[P...], then...m ᵢ n ,P max [The following is a list of parameters, not part of the original text: ], marked as parameter anomalies. Trend analysis uses the sliding window method, with a window length of M and a sliding step size of L. Let the continuously collected parameter sequence be P1, P2, ..., P... m The mean of the parameters within the window is P̄ = ∑Pᵢ / M (i = 1, 2, ..., M), and the variance is σᵢ. 2 =∑(Pᵢ-P̄) 2 / M, the linear fitting equation is P=at+b (a is the slope, b is the intercept); compare a with the preset trend slope threshold a0, if |a|≥a0 and the condition is met for 3 consecutive windows, it is marked as an abnormal trend, which can promptly capture potential faults such as rotor imbalance, misalignment, and bearing wear in rotating machinery.

[0044] (6) Judgment of abnormal state: Let the abnormal parameter be marked as S1 (S1=1 is abnormal, S1=0 is normal), and the abnormal trend be marked as S2 (S2=1 is abnormal, S2=0 is normal). When S1=1 or S2=1, it is judged as an abnormal state; let the abnormal parameter type be T (T=1 is bond phase abnormal, T=2 is vibration abnormal, T=3 is mixed abnormal), and the parameter exceeds the threshold amplitude ΔP=|P'-P m ᵢd|(P m ᵢd is the median threshold), the degree of trend deviation Δa=|a-a0|, which is quantified by the formula S=k p ×ΔP+k a ×Δa(k p k a The degree of anomaly is calculated using a weighted coefficient. Based on the range of values ​​for S, it is categorized as mild (S∈[S1,S2]), moderate (S∈[S2,S3]), or severe (S≥S3) anomaly, corresponding to the alarm and hazard classification logic of the EST-8500D series device. Finally, key phase state characterization data, vibration characteristic quantification data, and anomaly state judgment results are generated and transmitted to the interactive module, simultaneously triggering a relay alarm output.

[0045] 2.4 Working process of the interaction module The interaction module, based on the functionality of the EST-8500D-TXB communication control module, receives various data and judgment results output by the algorithm processing module. The process is as follows: The input data are D1 (key phase status data), D2 (vibration characteristic data), and D3 (anomaly judgment result). The Modbus encoding protocol (the standard communication protocol for the EST-8500D series devices) is used to encode D1, D2, and D3, with the encoding formula D' = c × D + d (c and d are encoding coefficients), converting them into binary data frames. The external device communication protocol parameters are B (baud rate, which can be flexibly set) and P (parity bit). The data frame format is adjusted according to B and P to obtain the standard data frame F. Simultaneously, the control command C is protocol-matched, adapting to various communication interfaces such as RS485, Ethernet (Modbus-TCP), and CAN, enabling data interaction and command transmission with host computers, DCS, PLCs, and other monitoring and protection devices. Furthermore, the interactive module supports setting communication parameters and updating thresholds via a touchscreen, and displays various data and anomaly statuses in real time. The data display colors follow the EST-8500D series specifications (blue for normal, orange for alarm, and red for shutdown).

[0046] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A rotating mechanical key phase and vibration monitoring system supporting multiple sensors, characterized in that, include: Sensing module, data module, algorithm processing module, and interaction module; The sensing module normalizes and conditions different types of bond phase signals and vibration sensing signals, and outputs a standard signal. Based on the standard signal, the data module completes the standardization and storage of the original monitoring data through preprocessing and data frame caching mechanisms, and outputs standardized original data. The algorithm processing module integrates a comprehensive analysis model of the key phase vibration of rotating machinery, performs time-domain-frequency domain decomposition on the standardized raw data, separates the key phase signal from the vibration signal, and initially extracts features. The bond phase signal and vibration signal are subjected to feature calculation. After extracting the bond phase feature parameters and vibration feature parameters, threshold comparison and trend analysis are performed. After data correction, bond phase state characterization data, vibration feature quantification data and abnormal state judgment results are generated. The interaction module generates standard data frames and control commands based on the bond phase state characterization data, vibration characteristic quantification data, and abnormal state judgment results through a data encoding adaptation process.

2. The system according to claim 1, characterized in that, The normalization conditioning and conversion process specifically includes: Different types of key phase signals and vibration sensing signals are subjected to amplitude calibration, level conversion and impedance matching processing to uniformly condition them into collectable signals with a preset amplitude range and a unified level standard, while filtering out noise interference to generate the standard signal.

3. The system according to claim 1, characterized in that, The preprocessing process specifically includes: The standard signal is subjected to adaptive filtering, signal de-jittering and outlier removal processing to filter out random noise and sudden interference signals and remove abnormal data points generated during the acquisition process. Then, through the data frame caching mechanism, the preprocessed signal is regularized, encoded, and stored according to the preset data frame format to generate the standardized raw data.

4. The system according to claim 1, characterized in that, The specific process of the time-domain-frequency-domain decomposition includes: The comprehensive analysis model for the key phase vibration of rotating machinery performs time-domain signal smoothing on the standardized raw data, converts the time-domain signal into a frequency-domain signal through Fourier transform, sets a filtering threshold based on the frequency domain characteristic differences between the key phase signal and the vibration signal, performs frequency band filtering on the frequency-domain signal, separates the frequency bands corresponding to the key phase signal and the frequency bands corresponding to the vibration signal, and finally converts the separated frequency-domain signal back into a time-domain signal through inverse Fourier transform, thus separating the key phase signal and the vibration signal.

5. The system according to claim 1, characterized in that, The specific process of preliminary feature extraction includes: Time-domain waveform analysis was performed on the separated bond phase signal and vibration signal to obtain the peak point, valley point and zero crossing point of the time-domain signal, and the time-domain peak value was calculated. Frequency domain spectral analysis is performed to identify the fundamental frequency peak in the frequency domain spectrum and extract the fundamental frequency in the frequency domain. The extracted two types of basic features, namely the time domain peak and the frequency domain fundamental frequency, are then standardized and normalized.

6. The system according to claim 1, characterized in that, The specific process of performing feature extraction on the bond phase signal and vibration signal includes: Based on the preliminary extracted basic features of the key phase signal, the time-domain waveform of the key phase signal is periodically identified, the time interval between two adjacent key phase pulse trigger points is calculated to obtain the key phase period; the difference between the trigger phase of the key phase signal and the preset reference phase is obtained to obtain the key phase offset; the rising edge and falling edge of the key phase pulse are identified, the time span between the rising edge and the falling edge is calculated to obtain the key phase pulse width; the time interval difference between consecutive key phase pulse trigger points is statistically analyzed to obtain the key phase trigger interval. The bond phase period, bond phase offset, bond phase pulse width, and bond phase trigger interval are extracted as bond phase characteristic parameters.

7. The system according to claim 6, characterized in that, The specific process of feature extraction of the vibration signal includes: Based on the preliminary extracted basic characteristics of the vibration signal, the time-domain waveform of the vibration signal is scanned to capture the maximum and minimum vibration amplitudes, and the difference is calculated to obtain the vibration amplitude. The fundamental frequency and harmonic frequencies of the vibration signal are identified through frequency domain spectrum analysis, and the fundamental frequency with the highest proportion is selected as the vibration frequency. The amplitude of each harmonic component in the frequency domain spectrum is quantized, and the amplitude proportion of each harmonic is extracted as the vibration harmonic component. The ratio of the peak value to the effective value of the vibration signal is calculated to obtain the vibration peak factor. Vibration amplitude, vibration frequency, vibration harmonic components, and vibration peak factor are extracted as vibration characteristic parameters.

8. The system according to claim 1, characterized in that, The comprehensive analysis model for the vibration of rotating mechanical key phases has a built-in preset threshold database, which specifically includes: The preset threshold database is classified and stored according to the type, specifications and working conditions of rotating machinery. Each category stores the threshold range of key phase characteristic parameters and the threshold range of vibration characteristic parameters under normal working conditions. When performing threshold comparison, the rotating machinery key phase vibration comprehensive analysis model matches the corresponding threshold range from the preset threshold database based on the currently monitored rotating machinery type, specifications and operating conditions, and then compares the extracted key phase feature parameters and vibration feature parameters with the matched threshold range one by one.

9. The system according to claim 1, characterized in that, The specific process of trend analysis includes: The length and step size of the preset sliding window are used to arrange the continuously collected bond phase characteristic parameters and vibration characteristic parameters in chronological order. The mean, variance and linear fitting of the characteristic parameters within the window are calculated in units of the sliding window to obtain the changing trend of the characteristic parameters within the window. The changing trend of the current window is compared with the preset law. The window is slid sequentially and the calculation and comparison process is repeated.

10. The system according to claim 1, characterized in that, The specific process for generating the abnormal state judgment result includes: The key phase characteristic parameters and vibration characteristic parameters are compared one by one with the threshold ranges matched in the preset threshold database. Based on the trend analysis results, when parameter abnormalities or trend abnormalities occur, the rotating machinery is determined to be in an abnormal state. The abnormality type is marked according to the abnormal characteristic parameter type. Based on the magnitude of the parameters exceeding the threshold and the degree to which the trend deviates from the preset pattern, the degree of abnormality is quantified and marked, and the abnormal state judgment result is generated.

11. The system according to claim 1, characterized in that, The specific process of data correction includes: Various error models for the feature solving process are pre-stored, including errors caused by signal noise and system errors caused by algorithm calculation; After the feature calculation is completed, error detection is performed on the extracted bond phase feature parameters and vibration feature parameters to identify error components; Based on the error model, the detected error is compensated in reverse, and the feature parameters are corrected.

12. The system according to claim 1, characterized in that, The data encoding and adaptation process specifically includes: After receiving the key phase state characterization data, vibration characteristic quantification data, and abnormal state judgment results, the system performs encoding processing using a preset encoding protocol to convert them into binary data frames, which are the standard data frames. Based on the communication protocol type, the system performs protocol matching between the standard data frames and control commands.

Images