A vertical take-off and landing aircraft duct drive lubrication detection method

Abstract

The present application relates to a kind of vertical take-off and landing aircraft duct drive lubrication detection method, belong to vertical take-off and landing aircraft power system state monitoring field.It includes the following steps: setting temperature sensor on the outer side of upper and lower annular guide rail of duct motor, mirror image orthogonal distribution is synchronously collected temperature data and storage;Extract 7 temperature basic characteristics, calculate lubrication state comprehensive quantitative index and pulse stability index, construct normalized seven-dimensional feature vector, match 5 typical fault modes by weighted Euclidean distance method;Fusion mirror image symmetry and theoretical model health index realizes sensor self-checking and fault tolerance, temperature compensation is carried out to failed sensor using first-order fourier series model;Based on four-stage cumulative loss model, the residual life of lubricating grease is quantitatively evaluated, and finally a comprehensive diagnostic report is generated.The present application has low hardware cost, easy installation, early warning, precise fault classification ability, which can effectively ensure the safe and stable operation of duct drive.

Description

Technical Field

[0001] This invention belongs to the field of power system condition monitoring technology for vertical takeoff and landing aircraft, specifically relating to a lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft. Background Technology

[0002] Vertical takeoff and landing (VTOL) aircraft utilize a hollow annular ducted motor as their core lift system. The rotor integrates fan blades at its center, which, during rotation, expel air downwards, generating thrust. The rotor achieves axial and radial positioning via upper and lower annular guide rails. Each guide rail is equipped with multiple ceramic sliders that slide with the rotor. Taking a typical motor with a guide rail diameter of 1370mm, 18 sliders on each side, and a maximum speed of 3000rpm as an example, the maximum linear velocity of the sliders can reach 215m / s. A single motor generates a maximum lift of approximately 300kg, corresponding to an axial thrust of 2940N. This force compresses the upper guide rail upwards through the rotor, causing the upper guide rail to bear an axial load 3 to 5 times that of the lower guide rail. This load asymmetry results in higher contact stress and more frictional heat generation on the upper guide rail sliders, significantly accelerating grease degradation. Therefore, during normal operation, the temperature baseline of the upper guide rail is higher than that of the lower guide rail. The stable positive temperature difference between the two and its variation characteristics become the core physical basis for determining lubrication status and fault modes.

[0003] Currently, the five mainstream technologies for lubrication monitoring of rotating machinery all have significant limitations: traditional single-temperature threshold alarms suffer from severe lag, cannot distinguish fault types, and are insensitive to pulse-type temperature rises; vibration monitoring is not directly sensitive to lubrication status, the multi-slider structure leads to signal aliasing, and it is also susceptible to aerodynamic noise interference; infrared thermal imaging cannot achieve continuous online monitoring and is greatly affected by the installation field of view and environmental convection; oil analysis is not suitable for grease-lubricated systems, is a post-event detection method, and has a cumbersome process; multi-source sensor fusion technology has a complex architecture, high hardware costs, is limited by the compact structure of ducted motors, and single-point failures can easily lead to system failure. In summary, existing technologies lack a lubrication monitoring solution specifically designed for ducted motors, operating independently with only a very small number of temperature sensors, capable of utilizing the asymmetrical force characteristics of the upper and lower guide rails, and possessing sensor self-calibration and remaining life prediction capabilities. Summary of the Invention

[0004] To achieve the above objectives, the present invention employs the following technical solution: This invention provides a method for detecting lubrication of ducted drive systems in vertical takeoff and landing aircraft, comprising the following steps: An application to a ducted motor having a hollow annular guide rail and multiple sliders, wherein the hollow annular guide rail includes an upper guide rail and a lower guide rail, includes the following steps: S1. Set on the outer surface of the annular guide rail Each temperature sensor synchronously collects temperature data and stores it. A circular buffer queue; among which... The temperature sensors are mounted on the upper guide rail and are orthogonally distributed circumferentially. A temperature sensor is installed on the lower guide rail and forms a mirror image of the temperature sensor on the upper guide rail. S2. Extract seven basic features from the temperature data: global maximum temperature, temperature pulse amplitude, temperature rise rate, temperature rise acceleration, global average temperature, deviation, and mirror temperature difference. S3. Calculate the comprehensive quantitative index of lubrication status based on the global maximum temperature, temperature pulse amplitude, and temperature rise rate. and pulse stability index The lubrication status comprehensive quantitative index, temperature pulse amplitude, deviation, temperature rise rate, temperature rise acceleration, mirror temperature difference, and pulse stability index are assembled into a normalized seven-dimensional feature vector. S4. Perform fault mode matching between the seven-dimensional feature vector and the preset fault mode feature library to obtain the current fault mode; S5. Calculate the mirror-symmetric health index and the theoretical model health index, and weightedly fuse them to obtain the comprehensive health index of the sensor; perform self-calibration and fault tolerance of the temperature sensor based on the comprehensive health index of the sensor. S6. Calculate the total loss using a four-level cumulative loss model to assess the remaining life of the grease. S7. Generate a comprehensive diagnostic report based on the remaining life assessment of the grease.

[0005] Furthermore, in step S1, the number of temperature sensors is greater than or equal to 4 and is an even number, and the sampling frequency is greater than or equal to 200Hz.

[0006] Furthermore, step S2 specifically includes: Extract the maximum value of the current sampling time readings from N temperature sensors to obtain the global highest temperature. ; For the highest global temperature For the corresponding temperature sensor, the temperature sampling sequence within the most recent complete rotation cycle is extracted from its circular buffer queue. The difference between the maximum and minimum values ​​in the sequence is calculated to obtain the temperature pulse amplitude. The formula is expressed as follows: , , in, Represents a temperature sampling sequence; This represents the number of sampling points within a single rotation cycle. , Sampling frequency, The motor's rotational speed per second; This indicates the operation of taking the minimum value; For the highest global temperature The temperature sequence of the corresponding temperature sensor within a preset sliding time window is subjected to least-squares linear fitting, and the absolute value of the slope of the fitted line is taken as the heating rate. ; Continuously save the most recent C calculation cycles The values ​​form a time series. A second-order difference operation is performed on the time series, and the average of all second-order difference values ​​is taken as the temperature rise acceleration. ; Pick The global average temperature is obtained by taking the arithmetic mean of the current readings of each temperature sensor. ; Calculate the global maximum temperature With global average temperature The difference yields the deviation. ; For each pair of upper and lower mirror images, calculate the current temperature difference between the upper and lower guide rail temperature sensors, and take the maximum value of the temperature difference for all mirror pairs to obtain the mirror temperature difference. .

[0007] Furthermore, step S3 specifically includes: Obtain comprehensive quantitative indicators of lubrication status The calculation process is as follows: , in, Indicates the temperature acceleration coefficient; Represents the natural constant; Indicates the reference temperature; This indicates the empirical coefficient for the grease type; Continuous acquisition Calculate the temperature pulse amplitude for one complete rotation cycle. The reciprocal of the coefficient of variation over a complete rotation cycle yields the pulse stability index. .

[0008] Furthermore, in step S3, the comprehensive quantitative index of lubrication state, temperature pulse amplitude, deviation, temperature rise rate, temperature rise acceleration, mirror temperature difference, and pulse stability index are normalized to construct a seven-dimensional feature vector. : , in, A comprehensive quantitative index representing the normalized lubrication condition; This represents the normalized temperature pulse amplitude; Indicates the deviation from normalization; This represents the normalized rate of temperature rise. This represents the normalized acceleration due to temperature rise; This represents the normalized mirror temperature difference; This represents the normalized pulse stability index.

[0009] Furthermore, in step S4, during pattern matching, a weighted Euclidean distance method is used for matching and discrimination, and the current feature vector is defined. With the center point of the kth reference mode Calculate the weighted distance between the current feature vector and the center points of all five reference modes, and take the minimum distance d_min; when d_min≤0.35, take the corresponding mode as the current diagnostic conclusion; when d_min>0.35, the current feature vector does not match all known modes sufficiently, continue to collect temperature data, and repeat steps S2-S4.

[0010] Furthermore, in step S5, the specific process for obtaining the comprehensive health indicators of the sensors is as follows: For each pair of mirror sensors, calculate the deviation between the measured temperature difference and the reference temperature difference. The maximum deviation between the measured temperature difference and the reference temperature difference in all mirror pairs is taken to obtain the mirror-symmetric health index. The formula is expressed as follows: , , in, , These represent the temperatures collected by the temperature sensors on the upper and lower guide rails, respectively, from the g-th mirror sensor. This represents the reference temperature difference of the g-th mirror sensor under healthy conditions during the break-in period; This indicates the absolute value operation; Indicates the number of mirror sensor pairs; This indicates the operation of retrieving the maximum value; Using historical normal data from all N temperature sensors, an empirical temperature field prediction model is established, and the theoretical predicted temperature value is calculated. The formula is expressed as follows: , in, , , , , The least squares fit coefficients represent five different health data points; Indicates the angular position of the temperature sensor; Indicates the current rotational speed; Indicates ambient temperature; For each temperature sensor, the maximum deviation between the measured value and the theoretically predicted temperature value is calculated to obtain the theoretical model health index. The formula is expressed as follows: , in, This represents the measured value of the i-th temperature sensor; This represents the predicted value of the i-th temperature sensor; Calculate the overall health index of sensors The calculation method is as follows: , in, Indicates the weight of mirror-symmetric health indicators; This represents the weight of the health indicators in the theoretical model.

[0011] Furthermore, in step S5, the temperature sensor undergoes self-calibration and fault tolerance based on the sensor's comprehensive health indicators: Sensor health grading: Determine sensor health. Determine if the sensor is in a sub-healthy state; : If the sensor fails, the fault tolerance mechanism is triggered; The fault tolerance mechanism is as follows: The circumferential temperature field of the guide rail is fitted using a first-order Fourier series model with least squares. , in, Indicates angle The theoretical temperature at that location; , , This represents three different coefficients to be estimated; The angular coordinates along the circumference of the guide rail are expressed in radians; compensation is performed on the temperature values ​​of the failed sensor. , in, This represents the temperature after the model has been fitted. Indicates the angular position of the failed sensor; If the faulty sensor is located on the upper guide rail. If the faulty sensor is located on the lower guide rail, ;in, This represents the estimated temperature after compensation for the failed sensor. This represents the reference temperature difference of the i-th pair of mirror sensors under healthy conditions during the break-in period.

[0012] Furthermore, step S6 specifically includes: Calculate total accumulated loss The formula is expressed as follows: , in, This indicates the cumulative load of historical early warning events; Indicates the base loss coefficient; Indicates the cumulative operating hours; Indicates the impact loss coefficient; Indicates the cumulative number of starts and stops; This represents the acceleration weighting coefficient; This represents the difference between the K value of the excess portion and 0.3; Indicates the duration of the event; Through total accumulated losses Calculate the percentage of remaining lifespan The formula is expressed as follows: , in, This indicates the critical load value calibrated during accelerated life testing.

[0013] The advantages of this invention are: This invention adopts a minimalist temperature sensor layout and is installed via surface mounting. It eliminates the need for drilling into the guide rail body, thus avoiding damage to the equipment structure, reducing hardware costs and installation difficulty, and perfectly adapting to the compact hollow structure of ducted motors. The system operates completely independently, without relying on any external monitoring devices such as vibration, air gap, infrared thermal imaging, or oil analysis. It can complete all monitoring tasks solely through multi-dimensional feature analysis of the temperature signal itself. It transforms the inherent force asymmetry of the upper and lower guide rails of the ducted motor into a diagnostic advantage, turning the physical deviations that originally needed compensation into core clues for fault root cause identification, obtaining rich diagnostic information without additional hardware investment. It achieves accurate differentiation of multiple typical fault modes within the pure temperature domain, overcoming the inherent defect of traditional single temperature threshold schemes that can only alarm but not classify, while also possessing excellent early warning capabilities, able to identify the inflection point of lubrication deterioration trends in advance. It adopts a sensor self-calibration and fault-tolerance mechanism, maintaining basic diagnostic functions even when some sensors fail, avoiding overall monitoring paralysis due to single-point failures. By constructing a multi-dimensional loss accumulation model, it comprehensively considers the impact of various operating conditions on grease life, enabling quantitative assessment of remaining life, providing a scientific and reliable decision-making basis for condition-based maintenance of equipment. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0015] Figure 1This is a flowchart of the steps of the method of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 In this embodiment, as Figure 1 As shown, this invention provides a lubrication detection method for ducted drive systems in vertical takeoff and landing aircraft, the specific steps of which include: An application to a ducted motor having a hollow annular guide rail and multiple sliders, wherein the hollow annular guide rail includes an upper guide rail and a lower guide rail, includes the following steps: S1. Set on the outer surface of the annular guide rail Each temperature sensor synchronously collects temperature data and stores it. A circular buffer queue; among which... The temperature sensors are mounted on the upper guide rail and are orthogonally distributed circumferentially. A temperature sensor is installed on the lower guide rail and forms a mirror image of the temperature sensor on the upper guide rail. Specifically, the number of temperature sensors is greater than or equal to 4 and is an even number, and the sampling frequency is greater than or equal to 200Hz.

[0018] In one embodiment, N=4, sensor 1 is installed at the 0° position of the upper guide rail, sensor 2 is installed at the 90° position of the upper guide rail, sensor 3 is installed at the 0° position of the lower guide rail, and sensor 4 is installed at the 90° position of the lower guide rail, forming a double orthogonal mirror layout.

[0019] All sensor data are collected synchronously at a sampling frequency of 500Hz, at the maximum speed of the ducted motor of 3000rpm (corresponding to the rotational frequency). Under these conditions, 10 sampling points are generated within each rotation cycle (20ms), which is sufficient to capture the instantaneous temperature pulse waveform generated when a single slider passes through the temperature measuring point. The collected raw data is stored in a circular buffer queue after being digitally low-pass filtered at a cutoff frequency of 100Hz.

[0020] S2. Extract seven basic features from the temperature data: global maximum temperature, temperature pulse amplitude, temperature rise rate, temperature rise acceleration, global average temperature, deviation, and mirror temperature difference. Specifically, the maximum value of the readings at the current sampling time of N temperature sensors is extracted to obtain the global highest temperature. This point, which reflects the highest temperature on the entire circumference of the guide rail, is a direct indicator of the degree of deterioration in lubrication. , in, This represents the current temperature reading of the i-th sensor, in degrees Celsius (°C). This indicates the operation of retrieving the maximum value.

[0021] In one embodiment, since the lift direction of the ducted motor is upward, the upper guide rail bears the main axial load (approximately 3 to 5 times that of the lower guide rail), under normal lubrication conditions. This is typically found in the two sensors (S1 or S2) on the upper guide rail. Under full load conditions with a maximum lift of 300 kg, and under normal lubrication... The typical value range is 55℃ to 70℃. If If the temperature exceeds 85°C and continues to rise, it is determined that the grease has entered a stage of accelerated deterioration.

[0022] As a floating-point number record, its corresponding sensor number is recorded synchronously. and angle position This is used for subsequent fault location analysis.

[0023] For the highest global temperature For the corresponding temperature sensor, the temperature sampling sequence within the most recent complete rotation cycle is extracted from its circular buffer queue. The difference between the maximum and minimum values ​​in the sequence is calculated to obtain the temperature pulse amplitude. The formula is expressed as follows: , , in, Represents a temperature sampling sequence; This represents the number of sampling points within a single rotation cycle. , Sampling frequency, The motor's rotational speed per second; This indicates the operation of taking the minimum value; ΔT is one of the most critical local fault characteristic quantities in this invention. When a slider is poorly lubricated, the slider will experience a local instantaneous temperature rise due to increased friction every time it passes the location of the temperature sensor, which manifests as a periodic pulse in the temperature time series. ΔT captures the peak amplitude of this pulse within a single rotation cycle, directly reflecting the instantaneous friction intensity between the most dangerous slider and the guide rail when it passes the temperature measuring point.

[0024] Under normal lubrication conditions, ΔT typically does not exceed 0.5°C. The typical ΔT at the initial stage of lubrication failure in a single slider is approximately 1.5 to 3°C. In cases of severe slider jamming or foreign object embedding, ΔT can increase dramatically to over 5°C.

[0025] Since the upper guide rail is the main load-bearing guide rail, if ΔT increases significantly and T_max is located at the upper guide rail sensor at the same time, it indicates that the lubrication condition of a certain slider on the upper guide rail is deteriorating rapidly, and the slider at the corresponding angle position on the upper guide rail should be checked first.

[0026] For the highest global temperature The temperature sequence of the corresponding temperature sensor within a preset sliding time window is subjected to least-squares linear fitting, and the absolute value of the slope of the fitted line is taken as the heating rate. .

[0027] In one embodiment, let the width of the sliding window be... Seconds, including 1 sampling point. W = 5 seconds, M = 2500 sampling points. Data points within the window are... ,in, , , These represent the start and end temperatures within the sliding window, respectively. Establish a univariate linear regression equation for temperature and time: , in, The slope to be estimated is... For the intercept to be estimated, The function is a univariate linear regression function of temperature and time; the least squares method is used to solve it. : Temperature rise rate Pick absolute value: , in, This indicates an absolute value operation, with the temperature rise rate measured in degrees Celsius per minute.

[0028] v_T reflects the persistent trend of temperature change. During normal operation, v_T is close to zero (≤0.1 degrees Celsius per minute), and the guide rail is in thermal equilibrium. During mild degradation, v_T is between 0.1 and 0.3 degrees Celsius per minute. During moderate degradation, v_T is between 0.3 and 0.5 degrees Celsius per minute. During rapid deterioration, v_T exceeds 0.5 degrees Celsius per minute, indicating that the grease may reach a critical state within hours.

[0029] Under a lift of 300 kg, the upper guide rail, due to concentrated load, typically deteriorates at a rate 2 to 3 times faster than the corresponding sensor v_T on the lower guide rail. This is an important clue to distinguish between heavy-load degradation of the upper guide rail and a global failure.

[0030] Continuously save the most recent C calculation cycles The values ​​form a time series. A second-order difference operation is performed on the time series, and the average of all second-order difference values ​​is taken as the temperature rise acceleration. .

[0031] In one embodiment, let the most recent C computation cycles be... The values ​​constitute the time series as follows: Preferably, This corresponds to 10 consecutive calculation cycles; for every three consecutive values ​​in the time series, the second difference is calculated, and the arithmetic mean of all the second difference values ​​is taken to obtain the temperature rise acceleration. The unit is degrees Celsius per square minute; This is the core characteristic quantity for achieving early warning in this invention, used to predict the inflection point of the temperature rise trend. Compared with a single temperature threshold alarm, it can issue an early warning signal 8 to 15 minutes earlier. Its change pattern has a clear diagnostic meaning: A continuously positive and increasing value indicates that the rate of temperature rise is accelerating, the fault is deteriorating faster, and immediate attention is required.

[0032] A change from a positive value to a negative value indicates that the temperature rise curve has reached an inflection point, and the system may be about to enter a temperature plateau or decline period. The alarm level can be appropriately reduced.

[0033] The system continues to fluctuate slightly around zero: the system is in a relatively stable state and there is no trend of accelerated deterioration.

[0034] Pick The global average temperature is obtained by taking the arithmetic mean of the current readings of each temperature sensor. The unit is Celsius; This reflects the overall thermal balance level of the guide rail. Due to the inherent force asymmetry between the upper and lower guide rails, It is primarily used to track the long-term drift trend of overall thermal balance, rather than for short-term fault detection. Short-term fault detection mainly relies on deviation. and pulse stability index .when When a sustained rise occurs over a long timescale (e.g., hours to days), it indicates an increase in the overall thermal load of the guide rail system, which usually corresponds to the end of the natural life of the grease or global uniform deterioration.

[0035] Calculate the global maximum temperature With global average temperature The difference yields the deviation. The unit is Celsius; The magnitude of the temperature field directly reflects the degree of non-uniformity—that is, whether there are significant local hot spots. Under good lubrication conditions, heat is evenly distributed among multiple sliders, resulting in a relatively uniform circumferential temperature distribution along the guide rail. Typically within the range of 5 to 15°C. When When the temperature exceeds 25°C, it indicates that there are significant local hot spots in the circumferential direction of the guide rail, which most likely correspond to the lubrication failure of one or more sliders. When the temperature exceeds 40°C, the local hot spots are already very serious and the machine needs to be stopped immediately for inspection.

[0036] For each pair of upper and lower mirror images, calculate the current temperature difference between the upper and lower guide rail temperature sensors, and take the maximum value of the temperature difference for all mirror pairs to obtain the mirror temperature difference. The unit is Celsius.

[0037] In one embodiment, P pairs of mirror sensors are provided, where P = N / 2. The location number of each pair of mirror sensors is... , , Let be the position numbers of the temperature sensors on the upper and lower guide rails of the j-th mirror sensor pair, respectively. .but: , Preferably, The two pairs of mirror images correspond to the 0° position (sensor 1, sensor 3) and the 90° position (sensor 2, sensor 4). , The temperature collected by the temperature sensors on the upper and lower guide rails of the j-th mirror sensor is denoted as .

[0038] M_up_down is the core feature quantity with the most innovativeness and diagnostic value of this invention. It directly uses the physical characteristic of the asymmetrical force on the upper and lower guide rails caused by the lift direction of the ducted motor to identify the fault mode.

[0039] Normal baseline: Due to the upward lift, the upper guide rail bears the main axial load, resulting in greater frictional heat generation. Under normal operating conditions, the temperature of the upper guide rail is higher than that of the lower guide rail. Under a full load of 300kg and good lubrication, M_up_down remains within a stable range of 8 to 15℃, with fluctuations not exceeding ±3℃ per hour.

[0040] The quantitative criteria for "unidirectional increase" are as follows: if M_up_down increases monotonically over 6 consecutive calculation cycles (each cycle is 1 minute, i.e., a cumulative 6 minutes), and the cumulative increase exceeds 30% of the baseline value or the absolute value increase exceeds 5°, it is judged as "unidirectional increase".

[0041] Diagnostic Mode 1 – Accelerated Deterioration of Upper Guide Rail Under Heavy Load: M_up_down increases continuously in one direction, such as rising from 12℃ to above 25℃ within 48 hours, accompanied by a synchronous increase in v_T. This indicates that the thermal load on the upper guide rail (heavy-load guide rail) is intensifying relative to the lower guide rail, and it is highly likely that the upper guide rail grease has entered the accelerated deterioration stage, requiring maintenance.

[0042] Diagnostic Mode Two – Guide Rail Thermal Deformation: M_up_down increases sharply to above 30°C within a short period (e.g., within 30 minutes), but the pulse stability index ΔT_stability remains at a moderate level (3~10 (thermal deformation / slight unevenness, pulse fluctuations but discernible period)), without significant increase. This distinguishes it from the single slider failure mode – thermal deformation causes a full-circumference or large-scale temperature increase, rather than pulse-type heating of a single slider.

[0043] Diagnostic Mode 3 – Abnormal Heating of the Lower Guide Rail: This mode is further subdivided into two sub-stages. The early stage is characterized by a significant decrease in M_up_down from the baseline (e.g., a drop from 12℃ to 3℃ within 30 minutes), indicating that the lower guide rail is beginning to generate an independent heat source. The severe stage is characterized by M_up_down changing from a positive value to a negative value, meaning the lower guide rail temperature has exceeded the upper guide rail temperature. This is a crucial abnormal fault signal, requiring immediate shutdown and troubleshooting. If the lower guide rail, under normal operating conditions with a light load, experiences an abnormal temperature increase or surpasses that of the upper guide rail, it is necessary to investigate for abnormal off-center loading, lower guide rail slider failure, or foreign object embedding.

[0044] Diagnostic Mode 4 – Global Uniform Deterioration: M_up_down remains stable, but T_avg increases overall. This indicates that the upper and lower guides deteriorate at similar rates, with a uniform increase in heat load, typically corresponding to the end of the grease's natural lifespan.

[0045] S3. Calculate the comprehensive quantitative index of lubrication status based on the global maximum temperature, temperature pulse amplitude, and temperature rise rate. and pulse stability index The lubrication status comprehensive quantitative index, temperature pulse amplitude, deviation, temperature rise rate, temperature rise acceleration, mirror temperature difference, and pulse stability index are assembled into a normalized seven-dimensional feature vector. Comprehensive quantitative index of lubrication condition The calculation process is as follows: , in, Indicates the temperature acceleration coefficient; Represents the natural constant; Indicates the reference temperature; Indicates the empirical coefficient of grease type; a comprehensive quantitative index of lubrication condition. Dimensionless.

[0046] In one embodiment, Diagnostic meaning: When lubrication is good, Approaching zero, the ratio is extremely small; when the grease ages overall, The ratio continues to increase steadily; when individual sliders suddenly fail, A sharp increase in temperature may temporarily decrease the ratio, as the accelerating temperature term gradually becomes dominant. The value will then rise rapidly. This dynamic change pattern of "first a slight decline, then a rapid rise" is a key signal for distinguishing between global aging and local failure.

[0047] Based on the accelerating effect of temperature on the chemical reaction rate in the Arrhenius equation, the rate of oxidative degradation of grease increases exponentially with increasing temperature. The reference temperature is set to 50℃. The temperature acceleration coefficient is taken, considering the basic thermal aging characteristics of the lubricating grease under a lift condition of 300 kg, and is selected as follows: This coefficient is slightly larger than the recommended value for general rotating machinery to reflect the actual operating conditions where the lift load of the ducted motor subjectes the grease to higher thermal stress.

[0048] when When the correction factor is No additional acceleration. When The correction factor is The degradation rate is approximately twice that of the baseline state. When When the correction factor is .

[0049] Different types of grease were calibrated through accelerated life testing to obtain empirical coefficients for grease types. The reference grease... Other models The value is selected within the range of 0.8 to 1.5 based on its high-temperature resistance and additive formulation. Greases with superior high-temperature resistance... The smaller the value, the lower the corresponding K value.

[0050] K-value grading criteria: The lubrication is good and no special attention is required. The lubricating grease has slightly deteriorated; it is recommended to pay attention to this during routine inspections. The grease is moderately deteriorated; it is recommended to have it inspected or replaced within one week. The grease is severely degraded; it is recommended to replace it immediately.

[0051] Continuous acquisition Calculate the temperature pulse amplitude for one complete rotation cycle. The reciprocal of the coefficient of variation over a complete rotation cycle yields the pulse stability index. , dimensionless.

[0052] In one embodiment, continuous calculation The ΔT values ​​for each rotation cycle constitute a temperature pulse amplitude sequence: In a preferred embodiment, P_cycle=50, which corresponds to a cumulative duration of 1 second at 3000rpm.

[0053] Calculate the arithmetic mean of the temperature pulse amplitude sequence. and standard deviation The coefficient of variation is , Pulse stability index .

[0054] Boundary case handling: When At that time, set directly (Upper limit truncation), and judged as "high stability", indicating that the pulse is extremely weak and extremely stable, corresponding to good lubrication conditions. When =0 and When =0 (i.e., ΔT is always zero), the same setting is also used. .

[0055] ΔT_stability is the most critical discriminant feature for distinguishing between local faults and global degradation in the pure temperature domain; High stability (ΔT_stability>10): This means that temperature pulses of almost the same amplitude appear in each revolution, strongly indicating that there is continuous and stable abnormal friction in a certain fixed slider. This is a typical characteristic of lubrication failure or slight jamming of a single slider. Medium stability (3 ≤ ΔT_stability ≤ 10): The pulse amplitude of each loop fluctuates to some extent, but obvious periodic pulse characteristics can still be identified, which may correspond to thermal deformation of the guide rail or mild non-uniform degradation. Low stability (ΔT_stability<3): The amplitude of each pulse fluctuates randomly and lacks a stable periodic pattern, indicating a global random deterioration process, such as large-area failure of grease.

[0056] The comprehensive quantitative indicators of lubrication status, temperature pulse amplitude, deviation, temperature rise rate, temperature rise acceleration, mirror temperature difference, and pulse stability indicators are normalized. The normalization method is to divide each characteristic quantity by its preset calibration upper limit value. The preset calibration upper limits are: 1, 5°C, 40°C, 0.5°C / min, 0.2°C / min; 2, 30.0°C, 10°C.

[0057] The seven-dimensional eigenvector F is represented as follows: , in, A comprehensive quantitative index representing the normalized lubrication condition; This represents the normalized temperature pulse amplitude; Indicates the deviation from normalization; This represents the normalized rate of temperature rise. This represents the normalized acceleration due to temperature rise; This represents the normalized mirror temperature difference; This represents the normalized pulse stability index.

[0058] S4. Perform fault mode matching between the seven-dimensional feature vector and the preset fault mode feature library to obtain the current fault mode; Fault Mode and Feature Library Construction: Based on extensive experimental data and equipment operation history, reference feature vector center points for five typical fault modes were predefined. The normalized feature value distribution of each mode in each feature dimension is shown in the table below: The criteria for determining the five typical failure modes are as follows: Failure mode 1: Slow aging of global grease; K_norm trend: Slowly increasing (K_norm rate of change is in the range of 0.02-0.05). :Low( ≤0.2); :Low( ≤0.15); :Low( ≤0.2); Low to medium ( ≤0.3); Low to medium (in the range of 0.1-0.4); Near zero ( ≤0.05).

[0059] Failure mode 2: Lubrication failure in some sliders; K_norm trend: It first drops slightly, then rises rapidly (K_norm changes are initially negative, in the range of -0.02 to -0.05, and then become positive, K_norm>0.05). High (suddenly increased to, ≥0.5); :high( ≥0.5); :high( ≥0.5); Low to medium ( ≤0.3); Medium to high (within the range of 0.3-0.7); : First positive, then stable; First positive: a_T (temperature rise acceleration) is positive at first, indicating that the temperature rise rate is accelerating and the fault is worsening; Then stable: a_T then returns to near zero (<0.01), with small and stable fluctuations, indicating that the deterioration rate has slowed down and the fault has entered a relatively stable deterioration stage.

[0060] Failure mode 3: Thermal deformation of the guide rail; K_norm trend: moderately increasing (K_norm rate of change 0.05-0.1); Medium (within the range of 0.2-0.5); Medium (within the range of 0.2-0.5); :Low( ≤0.2); High (increasing unidirectionally) ≥0.5), M_up_down continues to increase within 6 consecutive calculation cycles (1 minute / cycle, 6 minutes in total), and the increase is ≥30% of the baseline or the absolute value increment is ≥5°C; Medium (within the range of 0.2-0.5); : remains positive.

[0061] Fault mode 4: Slider jammed or foreign object embedded; K_norm trend: First, it drops slightly, then rises rapidly (K_norm changes are initially negative, between -0.05 and -0.02, then become positive, with a change rate ≥ 0.1). Extremely high ( ≥0.8); :high( ≥0.5); Extremely high ( ≥0.8); Medium (within the range of 0.2-0.5); :high( ≥0.5); First positive, then quickly turn negative.

[0062] Failure mode 5: Extensive failure of lubricating grease; K_norm trend: High across the entire region (K_norm consistently ≥ 0.3); Medium (within the range of 0.2-0.5); :Low( ≤0.15); :Low( ≤0.2); :Low( ≤0.15); Medium (within the range of 0.2-0.5); Near zero ( ≤0.05).

[0063] Detailed explanation of the discriminative role in each mode: yes The variation pattern of the normalized value under different fault modes is the key criterion for distinguishing different fault types in this invention: In the scenario of "accelerated degradation under heavy load on the upper guide rail" (corresponding to "slow aging of global grease" or "large-area failure of grease" in the table above, but specifically referring to the upper guide rail), The core characteristic is unidirectional increase – because the upper guide rail bears a greater axial load, its grease deteriorates faster, and its temperature continues to rise relative to the lower guide rail. It increases accordingly.

[0064] In the "guide rail thermal deformation" mode, Increased size is also a key characteristic, but the rate of increase is faster and the magnitude is greater, and The temperature remained at a moderate level without a significant increase—because thermal deformation causes an overall temperature rise in the guide rail, rather than a pulsed heating of a single slider. This is the key criterion for distinguishing between thermal deformation of the guide rail and failure of a single slider.

[0065] In the "abnormal overheating of the lower guide rail" mode, In the early stages, the temperature decreases significantly from the baseline, and in the severe stages, it turns from positive to negative—this is an extremely unusual and unconventional signal. Under normal operating conditions, the lower guide rail has a light load, and its temperature should not exceed that of the upper guide rail. Once this occurs… If the value decreases or turns negative, it indicates that an independent heat source has appeared on the lower guide rail, which needs to be investigated immediately.

[0066] During pattern matching, a weighted Euclidean distance method is used for matching and discrimination, and the current feature vector is defined. With the center point of the kth reference mode Calculate the weighted distance between the current feature vector and the center points of all five reference modes, and take the minimum distance d_min; when d_min≤0.35, take the corresponding mode as the current diagnostic conclusion; when d_min>0.35, the current feature vector does not match all known modes sufficiently, continue to collect temperature data, and repeat steps S2-S4.

[0067] In one embodiment, the current feature vector With the center point of the kth reference mode Weighted distance between The calculation method is as follows: , in, This represents the weight coefficient of the i-th feature component; This represents the i-th feature component; This represents the normalized reference value for the k-th fault mode and the ith feature dimension. The weighting coefficients are designed based on the differences in discriminative power of each feature: The weight w is 1.5 (the highest, as it is the core criterion for distinguishing between local faults and global degradation). The weight w = 1.2; The weight w = 1.2; The weight w = 1.0; The weight w = 1.0; The weight w = 0.8; The weight is w=0.8.

[0068] The values ​​are fixed templates obtained from offline calibration, experiments, and historical data statistics: For the five typical faults, the following steps were taken: Bench tests (heavy load on upper guide rail, slider jamming, thermal deformation, aging, large-area failure). Record a large number of operating conditions: speed, load, and temperature sequences; For each fault, extract 7 (K_norm, ...) values ​​from all samples. , , , , , Normalization characteristics; For each type of fault, the mean value is taken to obtain the standard center vector of that fault. ,Right now =The mean vector of the features of the k-th type of fault samples.

[0069] S5. Calculate the mirror-symmetric health index and the theoretical model health index, and weightedly fuse them to obtain the comprehensive health index of the sensor; perform self-calibration and fault tolerance of the temperature sensor based on the comprehensive health index of the sensor. For each pair of mirror sensors, calculate the deviation between the measured temperature difference and the reference temperature difference. The maximum deviation between the measured temperature difference and the reference temperature difference in all mirror pairs is taken to obtain the mirror-symmetric health index. The formula is expressed as follows: , , in, , These represent the temperatures collected by the temperature sensors on the upper and lower guide rails, respectively, from the g-th mirror sensor. This represents the reference temperature difference of the g-th pair of mirror sensors under healthy conditions during the break-in period, typically between 8 and 15°C, with the specific value varying depending on the specific device. Using historical normal data from all N temperature sensors, an empirical temperature field prediction model is established, and the theoretical predicted temperature value is calculated. The formula is expressed as follows: , in, , , , , The least squares fit coefficients represent five different health data points; Indicates the angular position of the temperature sensor; Indicates the current rotational speed; Indicates ambient temperature.

[0070] In one embodiment, the least squares fitting coefficients for health data are obtained as follows: During the healthy operation phase of the equipment, which is fresh from the factory or after a grease change, break-in complete, and without any faults, a large number of historical samples are collected. Based on the historical operating data during the healthy period (sensor angle, motor speed, ambient temperature, measured temperature), there are n sets of valid healthy sample data (n... 5) The sample feature vector is The temperature observation vector is .

[0071] Establish a design matrix (n×5): , in, This indicates the position angle of the k-th sensor, where the subscript k indicates which sensor it is, and k = 1, 2, 3, 4; Indicates the transpose operation; It is the actual rotational speed corresponding to the kth sensor; This represents the ambient temperature corresponding to the k-th sensor; This indicates the actual temperatures measured by all sensors; coefficient vector Solve using the equation: .

[0072] For each temperature sensor, the maximum deviation between the measured value and the theoretically predicted temperature value is calculated to obtain the theoretical model health index. The unit is Celsius, and the formula is as follows: , in, This represents the measured value of the i-th temperature sensor; This represents the predicted value of the i-th temperature sensor.

[0073] Obtain comprehensive health indicators from sensors The calculation method is as follows: , in, Indicates the weight of mirror-symmetric health indicators; This represents the weight of the health indicators in the theoretical model; preferably, , ; The weighting is higher because it can more directly reflect the drift or failure of a single sensor during normal device operation.

[0074] Sensor health grading: The sensor is determined to be healthy and participates in all subsequent calculations with a weight of 1.0.

[0075] The sensor is determined to be in a sub-healthy state and is included in the calculation with a weight of 0.85. At the same time, a "Sensor maintenance recommended" prompt is triggered in the diagnostic report.

[0076] If the sensor is determined to be faulty, the fault tolerance mechanism is triggered, and the sensor will no longer directly participate in the calculation of global statistics such as T_max and T_avg.

[0077] The fault tolerance mechanism is as follows: The circumferential temperature field was fitted using health sensor data to estimate the temperature of the failed sensor and capture the first-order temperature distribution caused by the asymmetry of the upper and lower guide rail loads. The circumferential temperature field of the guide rail is fitted using a first-order Fourier series model with least squares. , in, Indicates angle The theoretical temperature at that location; , , Three different coefficients to be estimated are obtained by least squares solution based on effective temperature data provided by health and sub-health sensors. The angular coordinates along the circumference of the guide rail are expressed in radians. The reason for choosing a first-order Fourier series instead of a higher-order one is that the temperature field of the annular guide rail is approximately uniformly distributed under good lubrication conditions. The main non-uniformity comes from the circumferential first-order variation caused by the asymmetry of the loads on the upper and lower guide rails and the possible off-center load torque. The first-order model can effectively capture this non-uniformity. With a sensor layout of N=4, the first-order model needs to estimate 3 parameters, requiring at least 3 valid data points, which just meets the minimum redundancy solution requirement.

[0078] Compensation processing is performed on the temperature values ​​of the failed sensor: , in, This represents the temperature after the model has been fitted. Indicates the angular position of the failed sensor; If the faulty sensor is located on the upper guide rail. If the faulty sensor is located on the lower guide rail, ;in, This represents the estimated temperature after compensation for the failed sensor. This represents the reference temperature difference of the i-th pair of mirror sensors under healthy conditions during the break-in period; Specifically, the compensation treatment for the inherent temperature difference between the upper and lower guide rails: Due to the inherent temperature difference of 8 to 15°C between the upper and lower guide rails, when the failed sensor and the healthy sensor used for fitting are located on different upper and lower guide rails, the temperature field obtained by the first-order Fourier model reflects the comprehensive temperature distribution of the guide rail planes where all effective sensors are located. Directly substituting the angle of the failed sensor will not accurately restore the actual temperature level of its guide rail layer. To solve this problem, the following compensation strategy is adopted: Determine the guide rail layer where the failed sensor is located. If there is a sensor on the same guide rail layer as the failed sensor among the effective sensors used for fitting, then the fitted value is directly used as the substitute value without additional compensation; if there are no effective sensors on the guide rail layer where the failed sensor is located (i.e., all sensors on this layer are failed, and all fitting data comes from another guide rail layer), then the fitted value is used as the substitute value. Based on this, the reference temperature difference T_ref(i) calibrated in the healthy state of the mirror pair to which the failed sensor belongs is superimposed. The specific rule is: if the failed sensor is located on the upper guide rail, If the faulty sensor is located on the lower guide rail, This ensures that the generated replacement values ​​accurately reflect the actual temperature level of the guide rail layer where the failed sensor is located. Complete the failed sensor data to ensure accurate calculations of features such as T_max and K, avoiding diagnostic biases caused by guide rail temperature differences. Substitute these values ​​into all subsequent feature calculations (T_max, ΔT, K, etc.) to ensure diagnostic continuity.

[0079] in, This represents the estimated temperature after compensation for the failed sensor. This represents the reference temperature difference of the i-th pair of mirror sensors under healthy conditions during the break-in period.

[0080] In one embodiment, when the number of health and sub-health sensors is less than 3, the first-order Fourier model cannot be solved stably. In this case, it degenerates into mean estimation—using the arithmetic mean of the remaining effective sensors as the replacement value for the failed sensors, and at the same time issuing a "sensor network degradation alarm" to indicate that sensor hardware repair is required.

[0081] S6. Calculate the total loss using a four-level cumulative loss model to assess the remaining life of the grease. Receive total accumulated loss The formula is expressed as follows: , in, The value range is 0-1, and values ​​exceeding 1 are recorded as 1; This indicates the cumulative load of historical early warning events; The coefficient represents the basic loss factor, which is calibrated through accelerated life testing. Under a full load of 300kg, α = 0.008 / h. This coefficient reflects the rate of natural oxidation aging of the grease at normal operating temperature. Indicates the cumulative operating hours; Represents the impact loss coefficient, taken as... The value is 0.12 per cycle. The physical basis for using the square root function is that the initial few start-stop cycles cause the greatest damage to the grease thickener fiber structure. As the number of start-stop cycles increases, the marginal damage effect of a single start-stop decreases. For example, after 10 start-stop cycles, this factor contributes approximately 0.38, and after 100 start-stop cycles, it contributes approximately 1.2. Indicates the cumulative number of starts and stops; This represents the acceleration weighting coefficient, which is set to 1.5 under a lift of 300 kg, reflecting the stronger cumulative damage effect of overload events under heavy load conditions. This represents the difference between the K value exceeding the limit and 0.3, where the excess part refers to the value when K > 0.3. This indicates the duration of the event, which refers to any event where the K value exceeds the threshold of 0.3.

[0082] In one embodiment, This is the cumulative load sum of all historical warning events experienced by the grease, in dimensionless units. Each time the system triggers a "moderate degradation" or "severe degradation" warning, the warning duration is multiplied by the corresponding severity coefficient and then added to this item. Historical warning events can cause irreversible performance degradation of the grease; even if current operating conditions return to normal, its remaining life has been permanently reduced.

[0083] For example, each time the system issues a warning of moderate degradation (0.3≤K<0.6) or severe degradation (K≥0.6), the duration is multiplied by the severity coefficient to form an irreversible damage integral.

[0084] coefficient: Moderate degradation: coefficient K value equals 0.3; Severe degradation: coefficient K value equals 0.6; Assume that this ducted motor has issued 3 warnings during operation: First warning: Moderate degradation warning, lasting 0.2 hours. Contribution: 0.2 × 0.3 = 0.06; Second: Severe degradation warning, lasting 0.5 hours, contribution: 0.5 × 0.6 = 0.3; Third: Moderate degradation warning, lasting 0.3 hours. Contribution: 0.3 × 0.3 = 0.09; Total historical cumulative load: 0.06 + 0.3 + 0.09 = 0.45.

[0085] Calculating the remaining life percentage through the total cumulative loss is as follows: The formula is expressed as follows: , where represents the critical load value calibrated in the accelerated life test. Preferably, take (dimensionless); When , the remaining life of the grease is sufficient, and it is carried out according to the regular maintenance cycle; When , the grease enters the end stage of its life, and attention should be paid to inspection; When , the grease is approaching the end of its life, and it is recommended to arrange for replacement.

[0086] When , the grease is approaching the end of its life, and it is recommended to arrange for immediate replacement.

[0087] S7. Generating a comprehensive diagnostic report based on the evaluation of the remaining life of the grease.

[0088] The comprehensive diagnostic report includes: Summary of lubrication status: Normal: K < 0.1, RemainingLife > 0.3; Attention: 0.1 ≤ K < 0.3, 0.2 < RemainingLife ≤ 0.3; Warning: 0.3 ≤ K < 0.6, 0.1 < RemainingLife ≤ 0.2; Danger: K ≥ 0.6, RemainingLife ≤ 0.1.

[0089] Fault mode: When the weighted distance minimum d_min ≤ 0.35, output the corresponding mode; when the weighted distance minimum d_min > 0.35, output "to be observed"; Sensor health: H_m ≤ 3 is healthy; 3 < H_m ≤ 6 is sub-healthy; H_m > 6 is failed; Maintenance suggestions: Classify according to RemainingLife (RemainingLife > 0.3: regular; 0.2 - 0.3: check within one week; RemainingLife ≤ 0.2: recommend replacement).

[0090] In one embodiment, when comparing the present invention with five common lubrication monitoring technologies, it has the following significant advantages, as shown in Table 1: Table 1 Comparison table of the present invention and five common lubrication monitoring technologies Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A lubrication detection method for ducted motor drives in vertical takeoff and landing aircraft, applied to a ducted motor having a hollow annular guide rail and multiple sliders, wherein the hollow annular guide rail includes an upper guide rail and a lower guide rail, characterized in that, Includes the following steps: S1. Set on the outer surface of the annular guide rail Each temperature sensor synchronously collects temperature data and stores it. A circular buffer queue; among which... The temperature sensors are mounted on the upper guide rail and are orthogonally distributed circumferentially. A temperature sensor is installed on the lower guide rail and forms a mirror image of the temperature sensor on the upper guide rail. S2. Extract seven basic features from the temperature data: global maximum temperature, temperature pulse amplitude, temperature rise rate, temperature rise acceleration, global average temperature, deviation, and mirror temperature difference. S3. Calculate the comprehensive quantitative index of lubrication status and the pulse stability index based on the global maximum temperature, temperature pulse amplitude, and temperature rise rate. Assemble the comprehensive quantitative index of lubrication status, temperature pulse amplitude, deviation, temperature rise rate, temperature rise acceleration, mirror temperature difference, and pulse stability index into a normalized seven-dimensional feature vector. Obtain comprehensive quantitative indicators of lubrication status The calculation process is as follows: ， in, Indicates the temperature acceleration coefficient; Represents the natural constant; Indicates the reference temperature; This indicates the empirical coefficient for the grease type; Indicates the highest global temperature; Indicates the rate of temperature increase; Indicates the amplitude of the temperature pulse; Continuous acquisition Calculate the temperature pulse amplitude for one complete rotation cycle. The reciprocal of the coefficient of variation over a complete rotation cycle yields the pulse stability index. ; S4. Perform fault mode matching between the seven-dimensional feature vector and the preset fault mode feature library to obtain the current fault mode; S5. Calculate the mirror-symmetric health index and the theoretical model health index, and weightedly fuse them to obtain the comprehensive health index of the sensor; perform self-calibration and fault tolerance of the temperature sensor based on the comprehensive health index of the sensor. S6. Calculate the total loss using a four-level cumulative loss model to assess the remaining life of the grease. Calculate total accumulated loss The formula is expressed as follows: ， in, This indicates the cumulative load of historical early warning events; Indicates the base loss coefficient; Indicates the cumulative operating hours; Indicates the impact loss coefficient; Indicates the cumulative number of starts and stops; This represents the acceleration weighting coefficient; This represents the difference between the K value of the excess portion and 0.3; Indicates the duration of the event; Through total accumulated losses Calculate the percentage of remaining lifespan The formula is expressed as follows: ， in, This indicates the critical load value calibrated during accelerated life testing; S7. Generate a comprehensive diagnostic report based on the remaining life assessment of the grease.

2. The lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft according to claim 1, characterized in that, Step S2 specifically includes: Extract the maximum value of the current sampling time readings from N temperature sensors to obtain the global highest temperature. ; For the highest global temperature For the corresponding temperature sensor, the temperature sampling sequence within the most recent complete rotation cycle is extracted from its circular buffer queue. The difference between the maximum and minimum values ​​in the sequence is calculated to obtain the temperature pulse amplitude. The formula is expressed as follows: ， ， in, Represents a temperature sampling sequence; This represents the number of sampling points within a single rotation cycle. , Sampling frequency, The motor's rotational speed per second; This indicates the operation of taking the minimum value; For the highest global temperature The temperature sequence of the corresponding temperature sensor within a preset sliding time window is subjected to least-squares linear fitting, and the absolute value of the slope of the fitted line is taken as the heating rate. ; Continuously save the most recent C calculation cycles The values ​​form a time series. A second-order difference operation is performed on the time series, and the average of all second-order difference values ​​is taken as the temperature rise acceleration. ; Pick The global average temperature is obtained by taking the arithmetic mean of the current readings of each temperature sensor. ; Calculate the global maximum temperature With global average temperature The difference yields the deviation. ; For each pair of upper and lower mirror images, calculate the current temperature difference between the upper and lower guide rail temperature sensors, and take the maximum value of the temperature difference for all mirror pairs to obtain the mirror temperature difference. .

3. The lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft according to claim 1, characterized in that, In step S3, the comprehensive quantitative indicators of lubrication status, temperature pulse amplitude, deviation, temperature rise rate, temperature rise acceleration, mirror temperature difference, and pulse stability indicators are normalized to construct a seven-dimensional feature vector. : ， in, A comprehensive quantitative index representing the normalized lubrication condition; This represents the normalized temperature pulse amplitude; Indicates the deviation from normalization; This represents the normalized rate of temperature rise. This represents the normalized acceleration due to temperature rise; This represents the normalized mirror temperature difference; This represents the normalized pulse stability index.

4. The lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft according to claim 3, characterized in that, In step S4, during pattern matching, the weighted Euclidean distance method is used for matching and discrimination, and the current feature vector is defined. With the center point of the kth reference mode Calculate the weighted distance between the current feature vector and the center points of all five reference modes, and take the minimum distance d_min; when d_min≤0.35, take the corresponding mode as the current diagnostic conclusion; when d_min>0.35, the current feature vector does not match all known modes sufficiently, continue to collect temperature data, and repeat steps S2-S4.

5. The lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft according to claim 1, characterized in that, In step S5, the specific process for obtaining the comprehensive health index of the sensor is as follows: For each pair of mirror sensors, calculate the deviation between the measured temperature difference and the reference temperature difference. The maximum deviation between the measured temperature difference and the reference temperature difference in all mirror pairs is taken to obtain the mirror-symmetric health index. The formula is expressed as follows: ， ， in, , These represent the temperatures collected by the temperature sensors on the upper and lower guide rails, respectively, from the g-th mirror sensor. This represents the reference temperature difference of the g-th mirror sensor under healthy conditions during the break-in period; This indicates the absolute value operation; Indicates the number of mirror sensor pairs; This indicates the operation of retrieving the maximum value; Using historical normal data from all N temperature sensors, an empirical temperature field prediction model is established, and the theoretical predicted temperature value is calculated. The formula is expressed as follows: ， in, , , , , The least squares fit coefficients represent five different health data points; Indicates the angular position of the temperature sensor; Indicates the current rotational speed; Indicates ambient temperature; For each temperature sensor, the maximum deviation between the measured value and the theoretically predicted temperature value is calculated to obtain the theoretical model health index. The formula is expressed as follows: ， in, This represents the measured value of the i-th temperature sensor; This represents the predicted value of the i-th temperature sensor; Calculate the overall health index of sensors The calculation method is as follows: ， in, Indicates the weight of mirror-symmetric health indicators; This represents the weight of the health indicators in the theoretical model.

6. The lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft according to claim 5, characterized in that, In step S5, the temperature sensor undergoes self-calibration and fault tolerance based on the sensor's comprehensive health indicators: Sensor health grading: Determine sensor health. Determine if the sensor is in a sub-healthy state; : If the sensor fails, the fault tolerance mechanism is triggered; The fault tolerance mechanism is as follows: The circumferential temperature field of the guide rail is fitted using a first-order Fourier series model with least squares. ， in, Indicates angle The theoretical temperature at that location; , , This represents three different coefficients to be estimated; The angular coordinates along the circumference of the guide rail are expressed in radians; compensation is performed on the temperature values ​​of the failed sensor. ， in, This represents the temperature after the model has been fitted. Indicates the angular position of the failed sensor; If the faulty sensor is located on the upper guide rail. If the faulty sensor is located on the lower guide rail, ;in, This represents the estimated temperature after compensation for the failed sensor. This represents the reference temperature difference of the i-th pair of mirror sensors under healthy conditions during the break-in period.

7. The lubrication detection method for ducted drive systems of vertical takeoff and landing aircraft according to claim 1, characterized in that, In step S1, the number of temperature sensors is greater than or equal to 4 and is an even number, and the sampling frequency is greater than or equal to 200Hz.

Images