Microwave coaxial resonance-based rotational speed measurement system and method

Abstract

The application discloses a rotational speed measurement system and method based on microwave coaxial resonance, and belongs to the technical field of microwave detection and rotational speed measurement. The system comprises a coaxial resonance microwave sensor, a rotational speed measurement analyzer and a signal processing module. The sensor sensitive end is provided with a ceramic protective layer and is installed in front of the measured rotating blade. The method comprises the following steps: calibrating a fixed working frequency point and obtaining the number of blades; continuously sampling the amplitude and phase of the reflected signal at the fixed frequency point, performing baseline deduction and normalization; extracting the blade event based on the amplitude, phase, edge feature and template similarity and outputting the time stamp; calculating the rotational speed according to the event time stamp and the number of blades, outputting the signal quality index, and automatically triggering the resonance tracking correction of the working frequency point when the quality decreases. The application takes into account the low complexity of fixed frequency point speed measurement and the adaptability of resonance tracking, and can still stably extract the blade event under harsh conditions such as oil pollution and variable gap, thereby improving the reliability of rotational speed measurement.

Description

Technical Field

[0001] This invention relates to the field of microwave detection and rotational speed measurement technology, and in particular to a rotational speed measurement system and method based on microwave coaxial resonance. Background Technology

[0002] Aero engines, turbomachinery, and other high-speed rotating equipment all require reliable speed information during startup, acceleration, steady-state operation, and transient conditions. Speed ​​is not only used for operational control and overspeed protection, but also affects blade tip timing, blade vibration detection, and equipment health status assessment. Such equipment typically has limited installation space and is subject to high temperatures, strong vibrations, oil mist, particle deposition, and electromagnetic interference, thus requiring speed measurement devices to be non-contact, miniaturized, contamination-resistant, and capable of real-time output.

[0003] Existing non-contact velocimetry technologies mainly include optical velocimetry, eddy current velocimetry, electrostatic velocimetry, magnetoelectric velocimetry, and blade tip timing-related methods. Optical velocimetry offers high accuracy but relies on a stable optical path and a clean window, making long-term operation in oil-contaminated, obstructed, or enclosed structures difficult. Eddy current and magnetoelectric methods are compact in structure but are sensitive to target materials, installation gaps, and electromagnetic boundary conditions. Blade tip timing technology can obtain the arrival time of the blade, but its system design often revolves around blade vibration and phase relationships. If used directly as a dedicated speed measurement link, issues such as stable signal extraction, installation adaptability, and engineering integration still need to be addressed.

[0004] Microwave near-field measurement features non-contact operation, insensitivity to oil contamination, ease of miniaturization and integration, and suitability for sensing rotating metal targets. For rotating components with blades, the blades alter the equivalent electromagnetic boundary at the sensor tip when passing through the sensor's sensitive area, causing detectable amplitude and phase disturbances in the reflected signal. If the blade passage event corresponding to this disturbance can be stably extracted, the rotational speed can be inverted based on the number of blades and the event interval.

[0005] However, existing microwave velocimetry solutions still have the following shortcomings: First, sensor installation gaps, oil deposits, and environmental drift can change the resonance state, and a single fixed frequency point may deviate from the optimal operating point during long-term operation; second, simple amplitude threshold or zero-crossing detection is easily affected by baseline drift, blade surface differences, and occasional noise, leading to false triggering or missed triggering; third, although continuous broadband frequency sweeping or complex multi-subband processing can obtain more frequency domain information, its real-time performance, hardware complexity, and cost are not conducive to engineering applications in confined spaces.

[0006] Therefore, a microwave coaxial resonant velocimetry system and method for high-speed rotating blades is needed. This system should not only take advantage of the miniaturization and high sensitivity of the coaxial resonant sensor, but also avoid continuous frequency sweeping during online velocimetry. Furthermore, it should automatically perform short-time resonance tracking and operating frequency correction when signal quality deteriorates or the resonant state drifts, thereby improving the stability of blade event extraction and the accuracy of speed measurement. Summary of the Invention

[0007] To address the challenge of reliably capturing blade passage events in high-speed rotating equipment under conditions of limited installation space, oil contamination, varying installation gaps, and high dynamic operating conditions, this invention proposes a microwave coaxial resonance-based speed measurement system and method. While retaining the advantages of low complexity in fixed-frequency online speed measurement, it adds mechanisms for resonant state self-checking, operating frequency correction, event quality evaluation, and abnormal event rejection. This improves the reliability of microwave speed measurement from both system structure and signal processing perspectives, thereby resolving the aforementioned problems.

[0008] This application discloses a speed measurement system based on microwave coaxial resonance, including a microwave sensor, radio frequency cable, speed measurement and analysis instrument and signal processing module; The microwave sensor is connected to the speed measurement and analysis instrument via an RF cable, and the speed measurement and analysis instrument is connected to the signal processing module. The sensitive end of the microwave sensor is oriented toward the area swept by the rotating blade under test, and is used to generate amplitude and phase perturbations of the reflected signal when the blade passes by. The rotational speed measurement and analysis instrument is used to output microwave signals to the microwave sensor and receive reflected signals, and to detect the amplitude and phase of the reflected signals. The signal processing module is used to identify blade passing events and calculate rotational speed based on the amplitude and phase changes of the reflected signal.

[0009] Preferably, the microwave sensor is a single-port coaxial resonant structure, including a non-metallic ceramic protective layer, an outer conductor, a central conductor, a resonant cavity, and a coaxial connection end; The outer conductor and the central conductor are arranged in a coaxial nested configuration, with the outer conductor annularly covering the radially outer side of the central conductor. The resonant cavity is located at the axial top end of the outer conductor, and the center conductor passes through the resonant cavity; The non-metallic ceramic protective layer is disposed on the outside of the sensitive end of the microwave sensor, covering the end face and outer peripheral area of ​​the resonant cavity; The lower end of the outer conductor and the lower end of the center conductor together form the coaxial connection end.

[0010] This application also discloses a rotational speed measurement method based on microwave coaxial resonance, implemented using the aforementioned rotational speed measurement system, comprising the following steps: S1. Installation, Calibration, and Fixed Operating Frequency Selection: Measure the sensor's reflection response under no-load and blade-facing conditions to select a fixed operating frequency. Record the resonant bias , Set the resonant frequency and save the calibration parameters; S2, Obtain the number of blades and calibration parameters; S3, at a fixed operating frequency The amplitude and phase of the continuously sampled reflected signals are preprocessed by baseline subtraction and normalization. S4. Based on amplitude, phase, edge features and template similarity, perform joint extraction of spatiotemporal features of blade events, confirm valid blade events and output event timestamps; S5. Based on the valid blade event timestamps and the number of blades Calculate the rotational speed and output the signal quality index. When the signal quality index is lower than the threshold, trigger the resonance tracking mode to correct the operating frequency.

[0011] Preferably, the resonant bias described in S1 It remains unchanged after initial calibration, until the current resonant frequency is detected. When drift occurs, the operating frequency is corrected to .

[0012] Preferably, the fixed operating frequency point mentioned in S1 Evaluation index based on comprehensive frequency points Sure:

[0013] in, , The reflection amplitude and phase under no-load conditions. , The reflection amplitude and phase are for the blade facing the same direction. , For noise estimation, This indicates phase foldback processing. This is the local slope stability coefficient of the resonance curve. As a penalty item, , , , As weight, To prevent small quantities with a denominator of zero.

[0014] Preferably, the number of blades in S2 Obtain it through the following methods: For rotating components with known structures, the information can be obtained from design drawings or manual input. For a blade number unknown but with an external reference rotational speed The rotating component, in time Internal statistical count of valid leaf events Calculate using the following formula

[0015]

[0016] If there is no external reference speed, the blade event frequency is output first, and then converted into speed after the user inputs the number of blades; The calibration parameters include the installation clearance. Operating frequency Resonant bias Event templates and threshold parameters.

[0017] Preferably, the baseline subtraction in S3 is: Estimating the amplitude baseline using moving median or moving average and phase baseline Calculate the amplitude disturbance. and phase perturbation ,in and The direction of the response to the event that the blade passes through during calibration is determined; The normalization preprocessing is as follows: Calculate the normalized amplitude and normalized phase ,in and This represents the absolute median deviation within the sliding window.

[0018] Preferably, the candidate events in S4 are determined by the comprehensive response function. produce:

[0019] in, For amplitude perturbation weights, For amplitude, For phase perturbation weights, For phase-assisted response, For edge feature weights, This refers to the edge sharpness index. For template-related weights, The template correlation coefficient; Candidate events must simultaneously meet the following conditions: Comprehensive response function The value is a maximum within the local window and exceeds the adaptive threshold. ; The time interval between adjacent candidate peaks is not less than , This refers to the maximum permissible speed. The direction of the phase change is consistent with the calibration event; Template correlation coefficient Not lower than the threshold ; Event interval consistency Not exceeding the maximum permissible angular acceleration A defined permissible range.

[0020] Preferably, the rotational speed calculation method described in S5 is as follows: Let the confirmation event timestamp be If the instantaneous rotational speed is calculated using adjacent blade events, then:

[0021] in, Instantaneous rotational speed; If in order to reduce random error, cross If the number of events is calculated, then:

[0022] If a fixed update time window is used Calculate the average rotational speed, and the first and last valid event times within the window are respectively: and The number of valid event intervals is ,but:

[0023] in, This represents the window average rotational speed calculated within an update time window or a statistical period.

[0024] Preferably, the quality indicators mentioned in S5 are as follows:

[0025] in, This is a normalized index for the signal-to-noise ratio of blade events, used to characterize the significance of blade events relative to background noise. The proportion of abnormal events. For the resonant bandwidth, , , , As weight; when When the speed is below the preset quality threshold, the output speed value and low quality flag are displayed, and the resonance tracking mode is triggered for short-term resonance tracking.

[0026] The beneficial effects of this invention are: (1) This invention combines a coaxial resonant sensor, resonant state monitoring, working mode switching, and fixed frequency event extraction into a closed-loop measurement system. During normal speed measurement, continuous sampling at a fixed frequency is maintained to reduce hardware complexity and online computation. When the signal quality deteriorates, the working frequency is corrected through short-time resonant tracking to avoid a single fixed frequency deviating from the optimal sensitive position for a long time.

[0027] (2) The event extraction of the present invention does not rely on a single amplitude threshold or zero crossover, but integrates amplitude peak height, event width, rising / falling edge, phase swing, amplitude and phase delay, template correlation coefficient, and consistency of adjacent event intervals. This method can reduce the risk of false triggering or duplicate counting caused by sporadic noise, baseline drift, and multi-peak structure of the same event.

[0028] (3) The present invention sets a non-metallic ceramic protective layer at the sensitive end of the sensor and incorporates the protective layer into the initial calibration and resonance tracking process. This structure can reduce the direct contamination of the metal resonant end face by oil, particles and external media, while avoiding misinterpreting the static response changes caused by the ceramic layer as blade events.

[0029] (4) The present invention outputs the number of valid events, the number of abnormal events and the signal quality index (SQI) at the same time as the output speed, which facilitates the determination of whether the current speed result is reliable in engineering applications and provides a basis for recalibration or correction of the working frequency. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the speed measurement system based on microwave coaxial resonance according to an embodiment of the present invention.

[0031] Figure 2 This is a schematic cross-sectional view of the coaxial resonant microwave sensor according to an embodiment of the present invention.

[0032] Figure 3 This is a diagram showing the electric field distribution of the microwave sensor in the resonant state according to an embodiment of the present invention.

[0033] Figure 4 This is a comparison chart of the S-parameters of the microwave sensor in this embodiment of the invention, showing the conditions of the sensor not being directly facing the blade and being directly facing the blade.

[0034] Figure 5 This is a schematic diagram of the amplitude / phase time-domain event at a fixed frequency point when the blade passes through the sensor's sensitive area, according to an embodiment of the present invention.

[0035] Figure 6 This is a schematic diagram of adaptive resonant tracking and fixed-frequency dual-mode collaborative velocity measurement according to an embodiment of the present invention.

[0036] Figure 7 This is a schematic flowchart of a speed measurement method based on microwave coaxial resonance according to an embodiment of the present invention.

[0037] Figure 8 This is a schematic diagram of the joint extraction of spatiotemporal features of blade events according to an embodiment of the present invention.

[0038] The attached figures are labeled as follows: 1-Microwave sensor, 2-RF cable, 3-Speed ​​measurement and analysis instrument, 4-Signal processing module, 5-Non-metallic ceramic protective layer, 6-Blade, 7-Outer conductor, 8-Center conductor, 9-Resonant cavity, 10-Coaxial connection end. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided with reference to the accompanying drawings and embodiments.

[0040] One embodiment of this application discloses a rotational speed measurement system based on microwave coaxial resonance, the structure of which is as follows: Figure 1 As shown, it includes a microwave sensor 1, an RF cable 2, a speed measurement and analysis instrument 3, a signal processing module 4, and a blade 6.

[0041] In a specific embodiment, taking engine speed measurement as an example, microwave sensor 1 is fixed to the engine housing by a mounting bracket. The sensitive end of microwave sensor 1 faces the area swept by the rotating blade 6 being measured. The transmitting port of the microwave signal and the receiving port of the reflected signal can be the same port, i.e., a single-port reflective measurement method is adopted. Microwave sensor 1 is connected to speed measurement analyzer 3 via radio frequency cable 2, and speed measurement analyzer 3 is connected to signal processing module 4. Signal processing module 4 is used to analyze radio frequency signals and convert the signals into actual speed values ​​for output.

[0042] In a specific embodiment, such as Figure 2 As shown, the microwave sensor 1 is a single-port coaxial resonant structure, comprising at least a non-metallic ceramic protective layer 5, an outer conductor 7, a central conductor 8, a resonant cavity 9, and a coaxial connection end 10. The outer conductor 7 and the central conductor 8 are arranged in a nested, coaxial configuration, with the outer conductor 8 annularly covering the radially outer side of the central conductor 7. They are radially spaced apart and do not contact each other. The lower end of the outer conductor 7 and the lower end of the central conductor 8 together form the coaxial connection end 10. The outer conductor 7 is electrically connected to the outer conductor of the RF connector (integrated with the RF cable), and the central conductor 8 is electrically connected to the inner conductor of the RF connector. The central conductor 8 and the outer conductor 7 together constitute a coaxial transmission and resonant structure, capable of forming a locally enhanced electric field distribution in the sensitive area at the top of the sensor, providing the basic electric field conditions for the sensing detection of the blade 6.

[0043] The resonant cavity 9 is located at the axial tip of the outer conductor, and the center conductor 8 passes through the resonant cavity 9. The resonant cavity 9 is the core sensitive detection area of ​​the sensor, corresponding to the end of the sensor near the blade 6, and is used to enhance the near-field coupling sensitivity of the sensor end face to the external metal blade. When the blade 6 enters the sensitive area, the external boundary conditions and equivalent capacitance of the sensor end change, causing amplitude and phase perturbations in the reflection coefficient near the resonant frequency band. Compared with ordinary non-resonant probes, the coaxial resonant structure can obtain a more significant perturbation response in a smaller volume, facilitating fixed-frequency time-domain sampling.

[0044] A non-metallic ceramic protective layer 5 is fixedly mounted on the outside of the sensor's sensitive end, completely covering the end face and outer periphery of the resonant cavity. It can be a sheet-like, cap-like, or end-face covering structure. Specifically, the non-metallic ceramic protective layer 5 is coaxially sleeved outside the resonant cavity 9, coaxially aligned with the inner resonant cavity 9, outer conductor 7, and center conductor 8. As the outer protective structure of the sensor's sensitive end, it effectively isolates the metal resonant end face from direct erosion and contamination by external media such as oil, solid particles, and high-temperature airflow, ensuring the long-term operational stability of the sensor. The thickness of the non-metallic ceramic protective layer 5... Dielectric constant and loss tangent It will affect the initial resonance state of the sensor. Therefore, the non-metallic ceramic protective layer 5 should be included in the calibration as part of the overall sensor structure, and should not be replaced arbitrarily after calibration.

[0045] Mounting gap between microwave sensor 1 and blade 6 Defined as the shortest distance between the outer surface of the ceramic protective layer and the trajectory of the blade. Based on the current prototype, the sensing distance can cover a range of 0-10 mm; before formal submission, the optimal gap range, allowable error, and corresponding signal-to-noise ratio should be determined based on actual measurement results. If subsequent prototype verification indicates that the installation distance can be increased, it should be supplemented with actual test data.

[0046] In one specific embodiment, the speed measurement and analysis instrument 3 includes at least a microwave source, a reflected signal separation unit, an amplitude / phase detection unit, an analog-to-digital conversion unit, a resonance state monitoring unit, a working mode switching unit, a calibration parameter storage unit, a blade event spatiotemporal feature extraction unit, and a speed calculation and quality evaluation unit.

[0047] The microwave source can output swept or narrowband frequency step signals during calibration or resonant tracking, and can also output continuous wave signals at fixed frequencies during online speed measurement. The reflection signal separation unit can employ a directional coupler, circulator, or other equivalent reflection signal separation structure. The amplitude / phase detection unit can use a detector with a phase detection circuit, or it can use I / Q demodulation to obtain the complex reflection signal. The resonant state monitoring unit is used to estimate the current resonant frequency of the sensor during installation calibration, timed self-test, or signal quality degradation. Resonance depth, local slope, and quality factor The operating mode switching unit switches between adaptive resonance tracking mode M1 and fixed-frequency high-speed speed measurement mode M2 ​​based on the resonance state, event quality indicators, and user-defined update cycle. The calibration parameter storage unit stores the operating frequency. Resonant frequency bias Amplitude / phase baseline, event template, threshold parameters, number of blades and installation gap Parameters such as these.

[0048] The microwave source is used to output a swept or discrete frequency signal near the resonant frequency band in calibration mode, and to output a fixed operating frequency in speed measurement mode. The microwave source can be a continuous wave signal. A phase-locked loop frequency synthesizer or a microwave frequency source can be used; for example, the ADF4351 can generate output frequencies ranging from 35MHz to 4.4GHz. If the system operates at a higher frequency, a higher frequency source can also be used.

[0049] The reflection signal separation unit is used to separate the incident signal output from the microwave source to the sensor from the reflected signal returned by the sensor, enabling the system to use single-port reflective measurements. This unit can be implemented using a directional coupler, circulator, or reflective bridge; for example, the Mini-Circuits ZUDC6-0283-S+ is a directional coupler in the 2 GHz to 8 GHz range.

[0050] The amplitude / phase detection unit is used to detect the amplitude and phase changes of the reflected signal at the operating frequency, providing input for subsequent blade event identification. This unit can be implemented using an I / Q demodulator to calculate the amplitude and phase from the I and Q signals; alternatively, a detector can be used when only amplitude detection is required. For example, the ADL5380, a 400MHz to 6GHz broadband quadrature I / Q demodulator, is suitable for joint amplitude and phase detection.

[0051] The analog-to-digital converter (ADC) unit converts the analog signal output from the amplitude / phase detection unit into a digital sequence. If I / Q demodulation is used, the ADC unit synchronously samples the I and Q signals. If amplitude / phase voltage output is used, the amplitude and phase channels are synchronously sampled. This unit can be implemented using a synchronous sampling ADC, with the specific sampling rate determined based on the maximum rotational speed, number of blades, and event width.

[0052] The resonant state monitoring unit is used to determine whether the sensor's resonant state has shifted based on calibration or narrowband retest results. This unit extracts the current resonant frequency fr(t), the response intensity at a fixed frequency point, baseline drift, and signal-to-noise ratio. When the resonant frequency shifts or the signal quality deteriorates, it triggers operating frequency correction or recalibration. This unit is typically implemented in software / firmware using a processor, DSP, or FPGA. For example, the TM4C1294NCPDT microcontroller can serve as an exemplary processor in low-speed or medium-speed systems, featuring a 120MHz Arm Cortex-M4F core, 1MB Flash, and 256KB RAM.

[0053] The operating mode switching unit controls the system to switch between calibration mode, resonant tracking mode, fixed frequency speed measurement mode, and anomaly verification mode. It enters calibration mode upon initial installation or sensor replacement. During normal operation, it enters fixed frequency speed measurement mode; when the resonant state monitoring unit determines a deterioration in signal quality, it enters resonant tracking or anomaly verification mode. This unit is primarily implemented by the processor's internal state machine, and can be used with an RF switch to switch between different RF channels when necessary.

[0054] The calibration parameter storage unit is used to store the number of blades B and the initial resonant frequency. Fixed operating frequency Frequency bias Parameters include amplitude / phase weights, threshold parameters, event templates, and quality evaluation thresholds. This unit can be implemented using non-volatile memory, such as EEPROM, Flash, or F-RAM, to ensure that existing calibration results can be retrieved after a power outage and restart.

[0055] The blade event spatiotemporal feature extraction unit is used to identify blade passing events from continuously acquired amplitude and phase sequences at a fixed frequency. This unit performs background subtraction, normalization, adaptive thresholding, phase consistency assessment, template similarity assessment, and event interval constraints on the signal to confirm valid blade events and output event timestamps. Compared to single amplitude peak detection, this unit utilizes amplitude, phase, and event temporal shape for joint judgment, reducing the probability of false triggers and missed triggers.

[0056] The speed calculation and quality evaluation unit is used to calculate the speed based on the effective blade event timestamps and the number of blades, and to determine the reliability of the speed measurement results. Let the timestamps of adjacent effective events be... and The number of leaves is Then the rotation speed can be calculated according to... The calculation can also calculate the average speed across multiple events or a fixed time window. This unit simultaneously outputs the number of valid events, the number of abnormal events, event interval stability, signal-to-noise ratio, and quality flags to indicate the reliability of the current speed measurement results.

[0057] During the laboratory verification phase, the speed measurement analyzer 3 can be implemented using the single-port frequency sweep and fixed-frequency sampling functions of a vector network analyzer. In the engineering application phase, it can be integrated into a single analyzer consisting of a fixed frequency source, a directional coupler, an I / Q demodulator, a data acquisition card, and an embedded processor. The common feature of both implementations is that, during normal speed measurement, continuous broadband frequency sweep is not performed; instead, high-speed time-domain sampling is conducted at a pre-calibrated or adaptively corrected fixed operating frequency.

[0058] The microwave coaxial resonance-based speed measurement system of this application has two operating modes: adaptive resonance tracking mode M1 and fixed-frequency high-speed speed measurement mode M2. M1 is used for installation calibration, resonance state self-check, and operating frequency correction. M2 is used for real-time speed measurement. The core of this dual-mode mechanism is that it does not continuously sweep the frequency throughout the entire speed measurement process, but performs short-time, narrow-band, low-duty-cycle resonance state monitoring when necessary, so that the fixed operating frequency is always kept at a frequency position sensitive to blade events.

[0059] During initial calibration, the system enters M1 to perform frequency sweeps on microwave sensor 1 under no-load and blade-facing conditions to determine the initial resonant frequency. and initial operating frequency And record the resonant bias between the two. The system then enters M2, where... The amplitude and phase are continuously sampled. During operation, if the resonance state monitoring unit detects the current resonant frequency... If drift occurs, the operating mode switching unit can correct the operating frequency to... This allows the system to continue operating at the same sensitive frequency point as the slope of the resonance curve.

[0060] The conditions that trigger M1 can include any of the following: First, the signal quality index (SQI) output by the speed calculation and quality evaluation unit is lower than the set threshold. .

[0061] Second, the number of abnormal events occurring within K consecutive update time windows exceeds a set proportion. Abnormal events include false triggering events, missed detection compensation events, repeated triggering events, events where the phase change direction is inconsistent with the calibration template, events with abnormal event width or edge slope, and events where the template correlation coefficient is lower than a threshold. In one implementation, K is 3 to 5. If the proportion of abnormal events to the total number of candidate events exceeds 20%, M1 is triggered; under strong noise conditions, this proportion can be relaxed to 30%.

[0062] Third, the time since the last resonance state monitoring exceeds the set period. .

[0063] Fourth, the drift of the amplitude baseline or phase baseline relative to the calibration value exceeds a threshold determined by the calibration noise. This threshold is not a subjectively fixed value, but is adaptively determined by the noise statistics during the calibration phase and the allowable speed range of the equipment. Specifically, the system acquires amplitude and phase sequences at a fixed no-load frequency and calculates the amplitude baseline. Phase baseline and the corresponding noise standard deviation , And according to the maximum permissible speed and number of leaves Calculate the minimum blade event interval This is used to determine the real-time baseline estimation window and avoid peak regions of blade events. During real-time operation, if the current amplitude baseline... or phase baseline satisfy or If the baseline drift exceeds the threshold, then it is determined that the baseline drift exceeds the threshold. and The coefficient is determined by the calibrated noise level and the number of effective sampling points, and is preferably between 3 and 5.

[0064] Through this mechanism, the system maintains low complexity and high refresh rate of fixed frequency sampling during normal speed measurement. When environmental changes cause the sensitivity of the fixed frequency to decrease, the system can automatically correct the operating point through short-time resonance tracking, thereby avoiding the risk of failure when a single fixed frequency is used for a long time.

[0065] One embodiment of this application also discloses a rotational speed measurement method based on microwave coaxial resonance. This method is implemented based on the aforementioned rotational speed measurement system based on microwave coaxial resonance, and its process is as follows: Figure 7 As shown, it includes the following S1-S5.

[0066] S1. Installation calibration and selection of fixed operating frequency.

[0067] After installation, the sensor's reflection response was first measured under both unloaded and blade-facing conditions. The unloaded state refers to the reflection amplitude and phase when the blade is not within the sensor's sensitive area, denoted as follows: and The blade facing forward means that the blade is located directly in front of the sensor end face, with an installation gap of [missing information]. The reflection amplitude and phase at time are denoted as . and .

[0068] To avoid relying solely on "maximum response change" leading to frequency points falling into areas of high noise or abrupt phase changes, this embodiment constructs a comprehensive frequency point evaluation index:

[0069] in, and For amplitude and phase noise estimation obtained from no-load repeated measurements or short-time static measurements, This indicates phase foldback processing. This is the local slope stability coefficient of the resonance curve. This is a penalty term for phase jumps, amplitude saturation, or frequency instability. , , , As weight, To prevent small quantities with a denominator of zero.

[0070] Within the candidate frequency bands, select The fixed operating frequency is a frequency with a relatively large local slope, stable phase, no abnormal phase jumps, and low static noise. If only the amplitude channel is reliable, then... =0; if the phase channel is stable, the phase is used as an auxiliary criterion. Unlike simply selecting the resonant peak or the frequency point of the maximum amplitude difference, the above indicators simultaneously consider blade disturbance sensitivity, noise level, and frequency stability, making them more suitable for long-term fixed-frequency time-domain sampling.

[0071] After completing the frequency selection, record... , , The parameters include: no-load baseline, blade alignment response, event template, and threshold parameters. When a subsequent resonance state drifts, a complete broadband frequency sweep is not re-executed; instead, the current frequency sweep is used instead. Narrowband short-term tracking is performed nearby, and based on... Correct the operating frequency.

[0072] S2, number of blades and calibration parameters.

[0073] Number of leaves It is a necessary parameter for speed inversion. For engines, turbines, or test rotors with known structures, Preferably, the data is obtained from equipment design drawings, impeller model, maintenance log, or manual input, and stored in the calibration parameter storage unit.

[0074] When the number of blades cannot be obtained directly, a low-speed calibration method can be used to determine it: using an external reference speed. Given or having a reference signal per revolution, at time Internal statistical count of valid leaf events and calculate If there is no external reference speed, the blade event frequency can be output first. The number of blades is then converted to r / min after the user inputs the number of blades.

[0075] Calibration parameters excluding the number of blades In addition, it also includes installation gaps. Ceramic protective layer parameters, operating frequency Resonant bias Amplitude / phase noise statistics, event templates Threshold coefficient Maximum permissible speed Maximum permissible acceleration and update time window These parameters should be tied to the specific prototype and the object under test to avoid direct application under different blade structures or different installation clearances.

[0076] S3. Fixed frequency amplitude-phase joint sampling and preprocessing.

[0077] In M2 mode, at a fixed operating frequency The reflected signal is continuously acquired at the location. If I / Q demodulation is used, the sampled signal is obtained... and It can calculate amplitude and phase:

[0078]

[0079] in, express The amplitude of the reflected signal at any given time. express The phase of the reflected signal at any given moment. express The in-phase component of the reflected signal at any given moment. express The quadrature components of the reflected signal at any given time.

[0080] If a detector and phase detection circuit are used, the result can be obtained directly. and .

[0081] Since temperature, cable condition, installation vibration, and instrument drift can cause slowly varying baselines, this embodiment first uses the moving median, moving average, or low-order trend term to estimate the background. , And calculate the disturbance:

[0082]

[0083] in, Indicates the amplitude disturbance amount, This represents the phase perturbation. This represents the direction coefficient of amplitude disturbance. This represents the phase perturbation direction coefficient. express The amplitude baseline at time [time]. express Phase baseline at time.

[0084] and The direction of the response to the blade passing event is determined by the calibration time, so that the effective blade event will show a positive response after processing.

[0085] To reduce the difference between amplitude and phase units, and Perform robust normalization:

[0086]

[0087] in, and This is the median absolute deviation of a sliding window or a calibrated static segment. This processing allows the threshold to be determined based on noise statistics, rather than relying on fixed amplitude units.

[0088] S4. Joint extraction of spatiotemporal features of blade events.

[0089] like Figure 8 As shown, this embodiment upgrades leaf event extraction from single peak detection to multi-dimensional feature joint matching. For each candidate peak, a feature vector is extracted within its neighborhood:

[0090] in, For the peak amplitude, For the event width, For the event area, The slope of the rising edge, The slope of the falling edge, For phase swing, The time deviation between the amplitude peak and the phase characteristic point. To match the calibrated event template The correlation coefficient, This is an indicator of consistency between adjacent event intervals.

[0091] Candidate events are first determined by the comprehensive response function. produce:

[0092] in, For amplitude perturbation weights, For amplitude, For phase perturbation weights, For phase-assisted response, For edge feature weights, This refers to the edge sharpness index. For template-related weights, This is the template correlation coefficient. The amplitude term is the primary criterion, while the phase and template terms are used to eliminate non-blade disturbances, occasional spikes, and baseline fluctuations.

[0093] Candidate events must simultaneously meet the following conditions: First, The value is a maximum within the local window and exceeds the adaptive threshold. , Desirable , Determined by calibration data, the optimal value is 3-6.

[0094] Second, the time interval between two adjacent candidate peaks is not less than the minimum time interval. This is used to prevent the same leaf event from being counted repeatedly.

[0095] Third, the direction of phase change is consistent with the calibration event.

[0096] Fourth, template correlation coefficient Not less than , The template correlation coefficient is the ratio between candidate leaf events and the calibration template, used to characterize the similarity between the waveform of the candidate event and the waveform of the actual leaf passage event. The threshold for the template correlation coefficient can be determined based on the mean and standard deviation of the correlation coefficients of the calibrated events. and standard deviation according to Confirmed, among which The threshold coefficient is preferably 2 to 3.

[0097] Fifth, consistency of event intervals If it does not exceed the threshold determined by the maximum permissible rate of change of rotational speed, it can be calculated as follows: Calculation, where The time interval between current adjacent events. It is the median of the previous stable event interval or the event interval within the sliding window.

[0098] To adapt to acceleration and deceleration conditions, It doesn't require all events to be at equal intervals, but rather based on the most recent... A number of valid events form a local prediction interval, which is combined with the device's allowable angular acceleration. The allowable range of variation is given. Specifically, the system obtains the reference interval based on the interval of the most recent N=5~10 valid events. And spaced by blade angles Calculate the reference angular velocity Then, based on the maximum permissible angular acceleration... Calculate the allowable range of the next event interval. When the candidate event interval satisfy or When, it is marked as a low-quality event, where , The standard deviation of the event interval. The sampling rate is defined as follows: Within a consecutive update time window of K = 3-5, if the proportion of low-quality events to the total number of candidate events exceeds 20%, the reliability of the current fixed frequency speed measurement is determined to have decreased, triggering M1 resonance tracking or recalibration. Under strong noise conditions, this proportion can be relaxed to 30%. If the proportion of low-quality events exceeds 40% within a single window, the abnormality verification mode can be directly entered.

[0099] Event confirmation is implemented using a state machine, including a waiting state, a candidate state, a confirmed state, and a dead-time state. In the waiting state, a comprehensive response exceeding a threshold is searched. Upon entering the candidate state, phase consistency and template similarity are checked. After confirmation, the event timestamp is output, and the system enters the dead-time state. The dead-time length is determined by... Confirmed; after the dead zone ends, it re-enters the waiting state. This state machine can suppress multi-peak repeated counting of the same blade event.

[0100] S5, timestamp correction, rotation speed calculation, and quality evaluation.

[0101] To improve timestamp accuracy, subsampling correction is applied to the discrete sampling peaks of confirmed events. One approach is the local centroid method.

[0102] in, For the first The timestamps of individual leaf events corrected using the local centroid method. For the local window of the candidate event Inner The time corresponding to each sampling point This is the comprehensive response value of the blade event at this sampling point.

[0103] The summation range is the sampling points near the candidate peak where G(t) is greater than a certain proportion of the peak value.

[0104] Another method is the three-point parabolic interpolation method, which fits a parabola to the peak sampling point and the sampling points before and after it to obtain the subsampling peak position.

[0105] If the corrected timestamp is obtained by the local centroid method Corrected timestamps obtained by the three-point parabolic interpolation method satisfy If so, the candidate event is marked as a low-quality event. The sampling interval is... The optimal decision coefficient is 0.5. In high-precision speed measurement scenarios, A value of 0.25 can be chosen, especially in scenarios with high noise levels or wide event peaks. It can be relaxed to 1.

[0106] Let the confirmation event timestamp be If the instantaneous rotational speed is calculated using adjacent blade events, then:

[0107] in, This represents the instantaneous rotational speed, expressed in r / min.

[0108] If in order to reduce random error, cross If the number of events is calculated, then:

[0109] If a fixed update time window is used Calculate the average rotational speed, and the first and last valid event times within the window are respectively: and The number of valid event intervals is ,but:

[0110] in, This represents the window average rotational speed calculated within an update time window or a statistical period.

[0111] The speed calculation and quality evaluation unit simultaneously outputs signal quality indicators. . It can be composed of event signal-to-noise ratio, template correlation coefficient, proportion of anomalous events, resonance drift, and baseline stability, for example:

[0112] in, This is a normalized index for the signal-to-noise ratio of blade events, used to characterize the significance of blade events relative to background noise. The proportion of abnormal events. For the resonant bandwidth, , , , As weight. When the speed is below a threshold, the system does not directly output the rotational speed without a quality flag. Instead, it outputs the rotational speed value and a low-quality flag, triggering M1 to perform short-term resonance tracking. For example, within a certain update time window, the system detects 30 candidate blade events, of which 23 are valid events and 7 are abnormal events, with an abnormal event ratio of 23.3%. Simultaneously, the signal-to-noise ratio of blade events decreases within this window, the template correlation coefficient decreases, and the current resonant frequency deviates somewhat from the initial calibration resonant frequency. The system calculates the quality evaluation formula... If the preset quality threshold is Then at this time, it is satisfied. This indicates a deterioration in the quality of the current speed measurement signal. In this case, the system does not directly output the speed result as a completely reliable one, but instead outputs the currently calculated speed value, for example... =18500r / min, and simultaneously attach a "low quality flag" or "requires verification flag", and trigger the short-time resonance tracking mode of M1.

[0113] When the rotational speed change of adjacent windows exceeds the maximum allowable acceleration of the equipment When deriving the scope, priority is given to checking for false triggers or missed triggers. For false triggers, the system uses a combined approach of event dead time, template similarity, and event shape consistency to eliminate them. The event dead time can be determined based on the maximum allowable speed and the number of blades, for example... ,in Preferably select values ​​between 0.3 and 0.6. If multiple candidate events fall within the same dead zone, only the event with the highest combined response value and template correlation coefficient is retained. For events with amplitudes reaching the threshold but template correlation coefficients below a certain threshold... Candidate peaks with inconsistent phase change directions or abnormal event widths and edge slopes are marked as false triggering events by the system. For missed triggers, the system obtains a reference event interval based on several recent valid events and determines the allowable range of the event interval by combining it with the equipment's allowable angular acceleration. If the interval between adjacent valid events is significantly larger than the allowable range, it is marked as a suspected missed trigger interval. The system does not generate new blade event timestamps out of thin air; instead, it uses real valid events to estimate the rotational speed across events and outputs a low-quality flag. If the anomaly cannot be corrected using the current data, a low-confidence result is output and a recalibration prompt is given.

[0114] In one specific embodiment, laboratory dual-mode verification was performed. A coaxial resonant microwave sensor was mounted on a bracket, with its sensitive end face directly facing the blade tip region of the rotating blade test specimen. A non-metallic ceramic protective layer 5 was installed, and the mounting gap was fixed. A vector network analyzer was used to connect the sensor via a single port. Frequency sweeps were performed first under no-load and blade-facing conditions to calculate... And select The network analyzer was then set to fixed-frequency sampling mode to continuously record amplitude and phase. Blade event timestamps were obtained using a spatiotemporal feature joint extraction method, and rotational speed was calculated based on the number of blades.

[0115] In one specific embodiment, integrated engineering measurement is performed. The vector network analyzer is replaced with an integrated speed measurement analyzer, which internally includes a swept / fixed frequency multiplexed microwave source, a reflection signal separation unit, an I / Q demodulation unit, a data acquisition unit, a resonance state monitoring unit, a working mode switching unit, and an embedded processor. After system power-on, pre-stored... , Threshold parameters and number of blades Output rotational speed within 100ms or other set window. When During descent, it automatically enters short-time resonance tracking and corrects. .

[0116] In one specific embodiment, switching is performed between amplitude-only velocity measurement and amplitude-phase combined velocity measurement. When the phase signal is significantly affected by cable bending or temperature drift, the system can... Setting it to 0 uses only amplitude and template shape for event extraction. When the phase signal is stable, the system enables a phase-assisted threshold to confirm the direction of blade events and eliminate non-blade noise peaks. This switch does not change the sensor structure or the rotational speed calculation formula.

[0117] In one specific embodiment, blade count self-learning calibration is performed. This is for cases where the number of blades is unknown but an external reference rotational speed is available. The rotating components, during the low-speed stable operation phase, are statistically analyzed. Valid number of events within ,calculate If the calculation results are consistent across multiple calibrations, then B is written to the parameter storage unit; if they are inconsistent, the user is prompted to check the installation clearance, reference speed, or blade event extraction quality.

[0118] This application proposes a coaxial resonant microwave sensor, ceramic protective layer involved in calibration, adaptive resonance tracking and fixed-frequency high-speed velocities dual-mode collaboration, and based on... Operating frequency selection, joint extraction of spatiotemporal features of blade events based on amplitude-phase-template-interval consistency, and based on The application addresses the switching of operating modes and correction of operating frequencies. Compared with microwave velocimetry schemes based on multi-subband energy comparison, tracking filters, or zero-crossing pulse generation, this application does not rely on continuous multi-subband processing during normal velocimetry, nor does it use simple zero-crossing as the sole triggering condition. Instead, it uses the resonance state of the coaxial resonant sensor as the closed-loop monitoring object and confirms events based on the multidimensional spatiotemporal characteristics of blade events.

[0119] Compared with coaxial resonant cavity sensors used for measuring medium concentration, the sensor proposed in this application, although also a coaxial resonant structure, has different working objects, signal meanings, and processing procedures: this application focuses on transient boundary disturbances and event timestamps caused by the passing of metal blades, rather than inverting the medium concentration based on the mapping relationship between resonant frequency and dielectric constant.

[0120] Compared with speed measurement methods based on Doppler frequency shift or difference frequency analysis of transmitted / reflected signals, this application does not use Doppler frequency shift as the main speed measurement quantity. Instead, it uses the fixed frequency reflection disturbance envelope caused by the periodic entry of the blade into the coaxial resonance sensitive region to generate an event sequence, and calculates the rotational speed by the number of blades and the event timestamp.

[0121] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A rotational speed measurement system based on microwave coaxial resonance, characterized in that, It includes a microwave sensor (1), an RF cable (2), a speed measurement and analysis instrument (3), and a signal processing module (4); The microwave sensor (1) is connected to the speed measurement and analysis instrument (3) via an RF cable (2), and the speed measurement and analysis instrument (3) is connected to the signal processing module (4); The sensitive end of the microwave sensor (1) is directed toward the area swept by the rotating blade (6) to generate amplitude and phase disturbances of the reflected signal when the blade passes by. The rotation speed measurement and analysis instrument (3) is used to output microwave signals to the microwave sensor (1) and receive reflected signals, and to detect the amplitude and phase of the reflected signals; The signal processing module (4) is used to identify the passing events of the blades and calculate the rotational speed based on the amplitude and phase changes of the reflected signal.

2. The speed measurement system based on microwave coaxial resonance according to claim 1, wherein the microwave sensor (1) is a single-port coaxial resonance structure, including a non-metallic ceramic protective layer (5), an outer conductor (7), a central conductor (8), a resonant cavity (9), and a coaxial connection end (10). The outer conductor (7) and the center conductor (8) are arranged in a coaxial nested configuration, with the outer conductor (7) circumferentially covering the radial outer side of the center conductor (8). The resonant cavity (9) is disposed at the axial top end of the outer conductor (7), and the center conductor (8) is disposed through the resonant cavity (9); The non-metallic ceramic protective layer (5) is disposed on the outside of the sensitive end of the microwave sensor (1), covering the end face and outer peripheral area of ​​the resonant cavity (9); The lower end of the outer conductor (7) and the lower end of the center conductor (8) together form the coaxial connection end (10).

3. A method for measuring rotational speed based on microwave coaxial resonance, characterized in that, The speed measurement system according to claim 1 or 2 is used, characterized by comprising the following steps: S1. Installation, Calibration, and Fixed Operating Frequency Selection: Measure the sensor's reflection response under no-load and blade-facing conditions to select a fixed operating frequency. Record the resonant bias , Set the resonant frequency and save the calibration parameters; S2, Obtain the number of blades and calibration parameters; S3, at a fixed operating frequency The amplitude and phase of the continuously sampled reflected signals are preprocessed by baseline subtraction and normalization. S4. Based on amplitude, phase, edge features and template similarity, perform joint extraction of spatiotemporal features of blade events, confirm valid blade events and output event timestamps; S5. Based on the valid blade event timestamps and the number of blades Calculate the rotational speed and output the signal quality index. When the signal quality index is lower than the threshold, trigger the resonance tracking mode to correct the operating frequency.

4. The rotational speed measurement method based on microwave coaxial resonance according to claim 3, characterized in that, The resonant bias described in S1 It remains unchanged after initial calibration, until the current resonant frequency is detected. When drift occurs, the operating frequency is corrected to .

5. The rotational speed measurement method based on microwave coaxial resonance according to claim 4, characterized in that, The fixed operating frequency point mentioned in S1 Evaluation index based on comprehensive frequency points Sure: in, , The reflection amplitude and phase under no-load conditions. , The reflection amplitude and phase are for the blade facing the same direction. , For noise estimation, This indicates phase foldback processing. This is the local slope stability coefficient of the resonance curve. As a penalty item, , , , As weight, To prevent small quantities with a denominator of zero.

6. The rotational speed measurement method based on microwave coaxial resonance according to claim 5, characterized in that, The number of blades mentioned in S2 Obtain it through the following methods: For rotating components with known structures, the information can be obtained from design drawings or manual input. For a blade number unknown but with an external reference rotational speed The rotating component, in time Internal statistical count of valid leaf events Calculate using the following formula If there is no external reference speed, the blade event frequency is output first, and then converted into speed after the user inputs the number of blades; The calibration parameters include the installation clearance. Operating frequency Resonant bias Event templates and threshold parameters.

7. The rotational speed measurement method based on microwave coaxial resonance according to claim 6, characterized in that, The baseline subtraction described in S3 is: Estimating the amplitude baseline using moving median or moving average and phase baseline Calculate the amplitude disturbance. and phase perturbation ,in and The direction of the response to the event that the blade passes through during calibration is determined; The normalization preprocessing is as follows: Calculate the normalized amplitude and normalized phase ,in and This represents the absolute median deviation within the sliding window.

8. The rotational speed measurement method based on microwave coaxial resonance according to claim 7, characterized in that, In S4, candidate events are determined by the comprehensive response function. produce: in, For amplitude perturbation weights, For amplitude, For phase perturbation weights, For phase-assisted response, For edge feature weights, This refers to the edge sharpness index. For template-related weights, The template correlation coefficient; Candidate events must simultaneously meet the following conditions: Comprehensive response function The value is a maximum within the local window and exceeds the adaptive threshold. ; The time interval between adjacent candidate peaks is not less than , This refers to the maximum permissible speed. The direction of the phase change is consistent with the calibration event; Template correlation coefficient Not lower than the threshold ; Event interval consistency Not exceeding the maximum permissible angular acceleration A defined permissible range.

9. The rotational speed measurement method based on microwave coaxial resonance according to claim 8, characterized in that, The rotational speed calculation method described in S5 is as follows: Let the confirmation event timestamp be If the instantaneous rotational speed is calculated using adjacent blade events, then: in, Instantaneous rotational speed; If in order to reduce random error, cross If the number of events is calculated, then: If a fixed update time window is used Calculate the average rotational speed, and the first and last valid event times within the window are respectively: and The number of valid event intervals is ,but: in, This represents the window average rotational speed calculated within an update time window or a statistical period.

10. The rotational speed measurement method based on microwave coaxial resonance according to claim 9, characterized in that, The quality indicators described in S5 are as follows: in, This is a normalized index for the signal-to-noise ratio of blade events, used to characterize the significance of blade events relative to background noise. The proportion of abnormal events. For the resonant bandwidth, , , , As weight; when When the speed is below the preset quality threshold, the output speed value and low quality flag are displayed, and the resonance tracking mode is triggered for short-term resonance tracking.

Images