Turboshaft engine complete machine test run structure damage monitoring method and system

Abstract

The invention relates to the technical field of aero-engine detection, in particular to a turboshaft engine complete machine test run structure damage monitoring method and system, and the method comprises the steps: S1, installing a multi-mode sensor, an airflow detection assembly and a multispectral imaging assembly on an engine; s2, constructing a digital twinborn model; s3, driving the whole machine to perform test run, collecting real-time data, performing data fusion, and synchronously mapping the data to the digital twinborn model; s4, analyzing the real-time data through a deep learning algorithm based on the digital twinborn model, and generating a health state evaluation report; and S5, based on the health state evaluation report, dynamically adjusting an early warning threshold Td and a sampling frequency fs. The dynamic damage generated in the test run process can be monitored in real time without shutdown, the detection time is greatly shortened, and therefore the detection efficiency and the detection reliability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of aero-engine detection, and more particularly to a method and system for monitoring structural damage during a complete run-in of a turboshaft engine. Background Art

[0002] A turboshaft engine is a compact, lightweight, high-power aircraft engine commonly used in helicopters, drones, and other aircraft. Its primary operating principle is to drive a compressor through a turbine to produce continuous and stable power output, making it suitable for scenarios requiring high power density and flexible maneuverability. During the turboshaft engine production process, ground-based test runs are required to simulate the operating conditions the engine may encounter in actual use, thereby assessing its health status and failure trends. Test runs not only identify potential structural defects and design issues in advance but also provide a scientific basis for optimizing design and improving operational stability. They are a critical step in ensuring engine safety and performance. During the engine-based test runs, engineers often use endoscopic inspection technology to detect internal cracks in turboshaft engines. However, endoscopic inspections must be performed with the engine shut down, making it impossible to monitor the ground-based test status in real time. This results in the inability to detect dynamic damage during operation in a timely manner. Summary of the Invention

[0003] The purpose of the present invention is to overcome the shortcomings of the existing technology that it must be detected in a shutdown state, cannot be monitored in real time, and cannot detect dynamic damage in an operating state. A method and system for monitoring structural damage in a whole-machine test of a turboshaft engine is provided, which realizes online monitoring, can detect dynamic damage and reduce detection time.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] A method for monitoring structural damage of a turboshaft engine during a complete run is provided, comprising the following steps:

[0006] S1. With a turboshaft engine in a stationary state, a multimodal sensor is mounted on the turboshaft engine, an airflow detection assembly is mounted at the air inlet and exhaust regions of the turboshaft engine, and a multispectral imaging assembly is mounted at the exhaust region, wherein the multimodal sensor is used to collect temperature, strain, vibration, and pressure information of the turboshaft engine, the airflow detection assembly is used to collect flow velocity and flow rate information of the intake and exhaust gases, and the multispectral imaging assembly is used to monitor the temperature field distribution of the exhaust gases in real time;

[0007] S2. Constructing a digital twin model with the same structure as the turboshaft engine, integrating the installation positions of the multimodal sensor, the airflow detection component, and the multispectral imaging component into the digital twin model, so that the installation positions of the multimodal sensor, the airflow detection component, and the multispectral imaging component in the digital twin model correspond one-to-one to the actual turboshaft engine positions;

[0008] S3. Driving the turboshaft engine to perform a full-machine test run to collect real-time data, where the real-time data includes multimodal sensor data, airflow detection component data, and multispectral imaging component data; performing data fusion processing on the real-time data and synchronously mapping the data to the digital twin model in real time;

[0009] S4. Based on the digital twin model, analyze the real-time data using a deep learning algorithm to obtain actual operation information, actual fault information, and fault prediction information of the component, and generate a health status assessment report;

[0010] S5. Based on the health status assessment report, dynamically adjust the warning thresholds T of the multimodal sensor, airflow detection component, and multispectral imaging component. d and sampling frequency f s ;

[0011] S6, based on the warning threshold T updated in step S5 d and sampling frequency f s , repeat steps S1-S5 to obtain the actual operation information, actual fault information and fault prediction information of the component.

[0012] The present invention provides a method for monitoring structural damage during a whole-machine test of a turboshaft engine. The method comprises the following steps: installing a multimodal sensor inside the turboshaft engine to collect temperature, strain, vibration and pressure information during the whole-machine test of the turboshaft engine; installing an airflow detection component in the air intake and exhaust port areas to collect flow velocity and flow rate information of the intake and exhaust gases; and a multispectral imaging component installed in the exhaust port area to detect the temperature field distribution of the exhaust gases. During the monitoring process, a digital twin model having the same structure as the engine needs to be constructed, and the installation positions of the multimodal sensor, the airflow detection component and the multispectral imaging component are all mapped in the digital twin model to ensure the actual operation. The engine is then driven to perform a whole-machine test, and real-time data is collected through the multimodal sensor, the airflow detection component and the multispectral imaging component. The real-time data is fused and mapped into the digital twin model to ensure that the actual installation positions correspond one-to-one in the digital twin model. A deep learning algorithm is used to analyze the real-time data to obtain the actual operating status and actual fault information, and a fault prediction is made, which is finally integrated into a health status assessment report. The warning thresholds T of the multimodal sensor, the airflow detection component and the multispectral imaging component are dynamically adjusted according to the health status assessment report. dand sampling frequency f s ; Finally, based on the updated warning threshold T d and sampling frequency f s Repeat steps S1-S5 to obtain actual component operation information, actual fault information, and fault prediction information. By installing multimodal sensors on the turboshaft engine to collect real-time strain and pressure change information during the entire engine test run, combined with the real-time data mapping function of the digital twin model, dynamic damage incurred during the test run can be monitored in real time without stopping the engine, significantly reducing detection time and improving detection efficiency and reliability.

[0013] Preferably, in step S1, the multimodal sensors include a temperature sensor, a fiber Bragg grating (FBG) strain sensor, a triaxial vibration sensor, and a dynamic pressure sensor. The multimodal sensors are deployed at key locations within the turboshaft engine. The temperature sensor is used to collect thermodynamic parameters during engine operation, ensuring accurate reflection of temperature trends under varying operating conditions. The fiber Bragg grating (FBG) strain sensor is installed in areas where stress concentration may occur, monitoring the strain state of key components in real time and helping to determine structural stability. The triaxial vibration sensor is deployed at locations with strong vibration, such as the turbine housing and bearing seat, to capture vibration characteristics and provide a basis for spectral analysis. The dynamic pressure sensor is located in the combustion chamber and intake duct to record pressure changes and analyze airflow conditions. These sensors, through their specific acquisition functions, work together to form a comprehensive, multidimensional operational data monitoring system, providing accurate and reliable input for subsequent data analysis.

[0014] Preferably, in step S3, the operating conditions of the whole machine test run include normal operating conditions, extreme temperature conditions and high load conditions. The test run process under normal operating conditions is used to simulate the operating state of the engine under standard design conditions, serving as a benchmark for health status assessment and providing a reference standard for subsequent abnormal assessment of extreme temperature conditions and high load conditions. Extreme temperature conditions are used to simulate the operation of the engine in extremely high or low temperature environments, evaluate the stability and durability of the engine, identify potential failure risks in extreme environments, such as crack initiation, material aging, etc., and provide durability data of components under extreme temperature conditions, providing a basis for improving design and material selection; high load conditions are used to simulate the operation of the engine under overload conditions, test its operating stability and limit performance, predict the potential failure hazards of the engine under overload conditions, generate a detailed diagnostic report, and determine the operating limits and design margins of the engine, providing guidance for formulating operating specifications and maintenance strategies.

[0015] Preferably, in step S3, the data fusion processing includes time synchronization adjustment, noise filtering optimization, spatial mapping alignment, and real-time consistency processing of the real-time data. Time synchronization adjustment is used to eliminate timing deviations between multiple sets of data caused by acquisition frequency or delay; noise filtering optimization improves the signal-to-noise ratio by removing noise and outliers in sensor data; spatial mapping alignment is to unify multiple sets of data into the same coordinate system in the digital twin model to avoid errors caused by perspective differences; real-time consistency processing is to ensure that the data fusion results maintain spatiotemporal consistency and logical consistency in a dynamic environment, thereby improving confidence.

[0016] Preferably, in step S4, based on the digital twin model, real-time data and historical data are jointly analyzed through a deep learning algorithm to obtain the actual operating information, actual fault information, and fault prediction information of the component. Relying on the digital twin model, real-time data and historical data are jointly analyzed, and a deep learning algorithm is used to analyze the multi-dimensional data provided by the multimodal sensor and the airflow detection component. Feature extraction and pattern recognition are used to comprehensively evaluate the operating status of the engine, analyze existing fault problems, and identify potential fault trends. The use of historical data can improve the accuracy of fault prediction. The generated health status assessment report contains the actual operating status of the component, fault prediction information, and possible fault causes, providing a comprehensive and intuitive basis for maintenance decisions.

[0017] Preferably, in step S5, the dynamic adjustment of the sampling frequency and the detection threshold includes: when the health status assessment report feedback indicates that the operating status is normal, increasing the warning threshold and reducing the sampling frequency; when the health status assessment report feedback indicates that there is an abnormality or potential risk in the detection, reducing the warning threshold and increasing the sampling frequency. When the engine is operating normally, maintain a low sampling frequency and standard threshold to reduce resource consumption; when an abnormal trend or potential risk is detected, automatically increase the sampling frequency and optimize the threshold sensitivity to ensure real-time capture of key data. Through the dynamic adjustment mechanism, it is possible to quickly respond to state changes, improve the efficiency and accuracy of fault monitoring, and at the same time improve the reliability of the detection structure and reduce resource consumption.

[0018] Preferably, in step S5, the warning threshold calculation formula is:

[0019] T d =α·V real-time +(1-α)·T base

[0020] Among them, T d is the warning threshold, V real-time is the real-time sensor data value, T base The threshold is set based on α, which is a weighting coefficient and the range is 0<α<1.d The dynamic adjustment of V based on real-time sensor data value real-time With the basic setting threshold T base The deviation between them is adjusted by using the weighting coefficient α for dynamic correction. The dynamic adjustment logic is as follows: When the real-time sensor data value V real-time Stable and close to the basic setting threshold T base When the weighted coefficient α changes slightly, the warning threshold T d The adjustment range is small to ensure that the normal operation state of slight fluctuations does not trigger false alarms; when the real-time sensor data value V real-time When the change range increases or approaches the abnormal range, the warning threshold T is quickly adjusted by reducing the weighting coefficient α. d , thereby improving detection sensitivity to capture abnormal engine trends.

[0021] Preferably, in step S5, the sampling frequency calculation formula is:

[0022] f s =f base ×(1+k×|ΔV|)

[0023] Among them, f s is the sampling frequency, f base is the basic sampling frequency, k is the adjustment coefficient, and ΔV represents the rate of change of sensor data. Sampling frequency f s The dynamic adjustment is used to achieve a quick response when an abnormal state is detected by increasing the data sampling frequency f s To capture key data and ensure accurate analysis of faults, the adjustment logic is as follows: When the engine is running normally, the sampling frequency f s Maintain the basic sampling frequency f base level, reducing resource consumption. When the sensor data fluctuation intensifies or approaches the warning threshold, the sampling frequency f is dynamically increased according to the change rate ΔV of the sensor data. s , to capture more detailed operating data, record data change trends in real time, provide high-precision input for the digital twin model, and support abnormal state analysis and fault prediction.

[0024] Preferably, the system further includes step S7, triggering a multi-level warning mechanism based on the real-time data and the warning threshold, collecting abnormal data, generating an abnormality analysis report, or stopping the whole machine test run. Through the multi-level warning mechanism, engine failures of varying degrees can be classified and responded to. When the abnormality level is not high, it is only necessary to continuously monitor, collect abnormal data, and generate an abnormality analysis report to facilitate subsequent analysis and processing. When the abnormality level is large, the whole machine test run should be stopped and an alarm should be issued to prompt staff to intervene. The multi-level warning mechanism can not only achieve accurate classification of abnormal conditions, but also flexibly select response measures based on the severity of the abnormality, avoiding unnecessary downtime and enabling rapid response at critical moments, thereby maximizing the safety and reliability of system operation.

[0025] The present invention also provides a turboshaft engine complete machine test structural damage monitoring system, which is used to implement the turboshaft engine complete machine test structural damage monitoring method, comprising:

[0026] Engine test module: used to drive the turboshaft engine for engine test;

[0027] Data acquisition module: including multimodal sensors, airflow detection components and multispectral imaging components, used to obtain real-time data during the operation of the whole machine test module and transmit the real-time data to the data processing module;

[0028] Data processing module: used to perform data fusion processing on the real-time data, map it to the digital twin model, and analyze the real-time data through deep learning algorithms to generate a health status assessment report:

[0029] Control module: used to dynamically adjust the warning threshold T of the multimodal sensor, airflow detection component and multispectral imaging component d and sampling frequency f s And control the working status of the whole machine test module.

[0030] The turboshaft engine whole-machine test structural damage monitoring system of the present invention uses a whole-machine test module to drive the turboshaft engine to conduct a whole-machine test on the ground, simulating the operating conditions that the engine may encounter in actual use, discovering potential dynamic defects during operation, and thus evaluating its health status and failure trends; the data acquisition module obtains real-time data of the whole-machine test module in multiple dimensions during the operation process through multi-modal sensors, airflow detection components and multi-spectral imaging components, including temperature, strain, vibration, pressure, flow rate, flow rate and other information, and transmits the relevant information to the data processing module; the data processing module fuses the data and maps it to the digital twin model for analysis to generate a health status assessment report; the control module adjusts the warning threshold T of the multi-modal sensor, airflow detection component and multi-spectral imaging component according to the health status assessment status.d and sampling frequency f s , the control data processing module records the engine data under abnormal conditions and regulates the working status of the whole machine test module.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. By installing multimodal sensors to collect real-time strain and pressure change information on key components and combining it with the real-time data mapping function of the digital twin model, dynamic damage generated during vehicle testing can be monitored in real time without downtime, significantly reducing detection time and thus improving detection efficiency and reliability.

[0033] 2. Through the digital twin model and deep learning algorithm, real-time data and historical operation data are jointly analyzed to generate fault predictions and actual fault reports, achieving fault location and classification, improving detection accuracy and the safety of the test process;

[0034] 3. The sampling frequency and warning threshold of the sensor are adjusted in real time through the health status assessment report, which reduces the resource consumption of the detection system under normal conditions. At the same time, high-frequency data is captured under abnormal conditions, solving the problem of slow response to abnormal situations and waste of resources caused by fixed sampling strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of the method for monitoring structural damage during the whole-machine test of a turboshaft engine;

[0036] Figure 2 This is a schematic diagram of the structural damage monitoring system for the turboshaft engine complete test run;

[0037] Figure 3 Schematic diagram of the early warning mechanism of the control module. DETAILED DESCRIPTION

[0038] The present invention is further described below with reference to specific embodiments. The accompanying drawings are for illustrative purposes only and are schematic, not actual, representations. They should not be construed as limiting this patent. To better illustrate the embodiments of the present invention, some components in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted from the drawings.

[0039] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right" and the like indicate directions or positional relationships based on the directions or positional relationships shown in the drawings, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting this patent. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0040] Example 1

[0041] This embodiment is a first embodiment of a method for monitoring structural damage during a complete turboshaft engine test run, comprising the following steps:

[0042] S1. When the turboshaft engine is stationary, a multimodal sensor is installed on the turboshaft engine, an airflow detection assembly is installed at the air inlet and exhaust port areas of the turboshaft engine, and a multispectral imaging assembly is installed at the exhaust port area. The multimodal sensor is used to collect temperature, strain, vibration, and pressure information of the turboshaft engine. The airflow detection assembly is used to collect flow velocity and flow rate information of the intake and exhaust gases. The multispectral imaging assembly is used to monitor the temperature field distribution of the exhaust gas in real time.

[0043] S2. Build a digital twin model with the same structure as the turboshaft engine, and integrate the installation locations of the multimodal sensor, airflow detection component, and multispectral imaging component into the digital twin model, so that the installation locations of the multimodal sensor, airflow detection component, and multispectral imaging component in the digital twin model correspond one-to-one to the actual turboshaft engine locations;

[0044] S3: Drive the turboshaft engine to conduct a full-machine test run and collect real-time data, including multimodal sensor data, airflow detection component data, and multispectral imaging component data. The real-time data is fused and processed and mapped to the digital twin model in real time.

[0045] S4. Based on the digital twin model, real-time data is analyzed through deep learning algorithms to obtain the actual operation information, actual fault information, and fault prediction information of the components, and generate a health status assessment report;

[0046] S5. Based on the health status assessment report, dynamically adjust the warning threshold T of the multimodal sensor, airflow detection component and multispectral imaging component. d and sampling frequency f s ;

[0047] S6, based on the warning threshold T updated in step S5 dand sampling frequency f s , repeat steps S1-S5 to obtain the actual operation information, actual fault information and fault prediction information of the component.

[0048] like Figure 1 As shown, the method for monitoring structural damage during the whole-machine test of the turboshaft engine in this embodiment installs a multimodal sensor inside the turboshaft engine to collect temperature, strain, vibration and pressure information during the whole-machine test of the turboshaft engine, installs the airflow detection component in the air inlet and exhaust port areas to collect the flow velocity and flow rate information of the intake and exhaust gases, and the multispectral imaging component installed in the exhaust port area is used to detect the temperature field distribution of the exhaust gas; during the monitoring process, a digital twin model with the same structure as the engine needs to be constructed, and the installation positions of the multimodal sensor, airflow detection component and multispectral imaging component are all mapped in the digital twin model to ensure the actual; then the engine is driven to move for the whole-machine test, and real-time data is collected through the multimodal sensor, airflow detection component and multispectral imaging component, and the real-time data is fused and mapped into the digital twin model to ensure that the actual installation position corresponds one-to-one in the digital twin model; a deep learning algorithm is used to analyze the real-time data, obtain the actual operating status and actual fault information, and make a fault prediction, which is finally integrated into a health status assessment report; the warning threshold T of the multimodal sensor, airflow detection component and multispectral imaging component is dynamically adjusted according to the health status assessment report d and sampling frequency f s ; Finally, based on the updated warning threshold T d and sampling frequency f s Repeat steps S1-S5 to obtain actual component operation information, actual fault information, and fault prediction information. By installing multimodal sensors on the turboshaft engine to collect real-time strain and pressure change information during the entire engine test run, combined with the real-time data mapping function of the digital twin model, dynamic damage incurred during the test run can be monitored in real time without stopping the engine, significantly reducing detection time and improving detection efficiency and reliability.

[0049] In this embodiment, multimodal sensors are deployed at key locations within the turboshaft engine. These sensors include temperature sensors, fiber Bragg grating (FBG) strain sensors, triaxial vibration sensors, and dynamic pressure sensors. The temperature sensors are used to collect thermodynamic parameters during engine operation, ensuring accurate reflection of temperature trends under varying operating conditions. Fiber Bragg grating (FBG) strain sensors are installed in areas where stress concentration may occur, monitoring the strain state of key components in real time and helping to determine structural stability. Triaxial vibration sensors are deployed in locations with high vibration levels, such as the turbine housing and bearing seat, to capture vibration signatures and provide a basis for spectral analysis. Dynamic pressure sensors are located in the combustion chamber and intake duct to record pressure changes and analyze airflow conditions. These sensors, each with its own specific acquisition capabilities, work together to form a comprehensive, multidimensional operational data monitoring system, providing accurate and reliable input for subsequent data analysis.

[0050] In step S3, the operating conditions of the whole-machine test run include normal operating conditions, extreme temperature conditions, and high-load conditions. The test run process under normal operating conditions is used to simulate the operating state of the engine under standard design conditions. It serves as a benchmark for health status assessment and provides a reference standard for subsequent abnormal assessment of extreme temperature conditions and high-load conditions. Extreme temperature conditions are used to simulate the operation of the engine in extremely high or low temperature environments, evaluate the engine's stability and durability, and identify potential failure risks in extreme environments, such as crack initiation and material aging. At the same time, it provides durability data of components under extreme temperature conditions, providing a basis for improving design and material selection. High-load conditions are used to simulate the operation of the engine under overload conditions, test its operating stability and extreme performance, predict the potential failure risks of the engine under overload conditions, generate a detailed diagnostic report, and determine the engine's operating limits and design margins, providing guidance for the formulation of operating specifications and maintenance strategies.

[0051] In this embodiment, under normal operating conditions, fiber Bragg grating (FBG) strain sensors record strain changes in real time to analyze the stress state of engine components under standard loads. Triaxial vibration sensors capture vibration characteristics to determine whether the vibration spectrum is within the normal operating range. Dynamic pressure sensors measure internal pressure fluctuations to verify airflow stability. Under extreme temperature conditions, temperature sensors monitor the thermal response of key components under high or low temperature conditions, analyzing temperature gradients and thermal expansion effects. Fiber Bragg grating (FBG) strain sensors record strain changes to assess stress concentration and fatigue characteristics of components under extreme temperatures. Triaxial vibration sensors detect vibration amplitude and frequency changes under high or low temperature conditions to determine whether there is structural loosening or abnormal vibration. Dynamic pressure sensors monitor pressure fluctuations in the combustion chamber and intake duct in real time to identify airflow instability caused by temperature changes. Under high load conditions, temperature sensors monitor temperature rise under high load to analyze heat concentration locations and cooling effectiveness. Fiber Bragg grating (FBG) strain sensors record strain changes under high load to identify plastic deformation or stress concentration caused by overload. Triaxial vibration sensors capture high-frequency vibration characteristics to determine whether there is dynamic instability caused by excessive load. Dynamic pressure sensors monitor pressure peaks and fluctuations within the combustion chamber to assess combustion efficiency and airflow stability.

[0052] In this embodiment, by real-time collection of intake air velocity and flow, the working stability of the intake system can be monitored and intake blockage or other abnormal fluctuations can be identified; monitoring of exhaust air velocity and flow can timely reflect the internal combustion state of the engine and the exhaust emission characteristics. The multi-spectral imaging component in the exhaust port area is used to monitor the temperature field distribution of the exhaust gas, capturing changes in combustion efficiency and temperature characteristics of abnormal areas in real time. The combined application of these two types of components can fully reflect the airflow and temperature distribution during the test process, providing key data support for engine performance analysis and fault trend prediction.

[0053] In step S3, data fusion processing includes using a multi-layer data fusion module to perform time synchronization adjustment, noise filtering optimization, spatial mapping alignment, and real-time consistency processing on real-time data. Time synchronization adjustment is used to eliminate timing deviations between multiple sets of data caused by acquisition frequency or delay; noise filtering optimization improves the signal-to-noise ratio by eliminating noise and outliers in sensor data; spatial mapping alignment is to unify multiple sets of data into the same coordinate system in the digital twin model to avoid errors caused by perspective differences; real-time consistency processing is to ensure that the data fusion results maintain temporal, spatial, and logical consistency in a dynamic environment, thereby improving confidence. Using data fusion technology, data from different sources are synchronized in time and space to ensure data consistency and accuracy. The fused data is mapped in real time through the digital twin model, enabling the virtual model to synchronously reproduce the actual operating status of the engine, thereby providing accurate dynamic data support for subsequent operation evaluation and fault diagnosis.

[0054] In step S4, based on the digital twin model, a deep learning algorithm is used to jointly analyze real-time and historical data to obtain information about the component's actual operation, actual faults, and fault predictions. Relying on the digital twin model, real-time and historical data are integrated and compared. Deep learning algorithms are used to analyze the multidimensional data provided by multimodal sensors and airflow detection components. Feature extraction and pattern recognition are used to comprehensively assess the engine's operating status and identify potential fault trends. The use of historical data improves the accuracy of fault predictions. The generated health status assessment report contains the component's actual operating status, fault prediction information, and possible fault causes, providing a comprehensive and intuitive basis for maintenance decisions.

[0055] In the digital twin model, the installation positions of multimodal sensors, airflow detection components, and multispectral imaging components are mapped one-to-one to the actual physical system. The data twin model accurately defines the position, detection range, and mode of action of each detection component to ensure that each data point can be accurately mapped in the virtual space. The digital twin model can not only dynamically reflect the data input of sensors and detection components, but also simulate data changes during operation, forming a high-precision operating status model that combines virtual and real, laying a solid foundation for fault analysis and performance optimization.

[0056] The method for monitoring structural damage during a turboshaft engine full-machine test run in this embodiment further includes step S7, triggering a multi-level warning mechanism based on real-time data and warning thresholds to collect abnormal data, generate an abnormality analysis report, or halt the full-machine test run. The warning mechanism includes a first warning level, a second warning level, and a third warning level. At the first warning level, abnormal data is collected; at the second warning level, an abnormality analysis report is generated and an alarm signal is issued; at the third warning level, the full-machine test run is halted, an abnormality analysis report is generated, and an alarm signal is issued. This multi-level warning mechanism enables classification and response to engine failures of varying severity. For low-severity abnormalities, continuous monitoring, abnormal data collection, and abnormality analysis report generation are sufficient for subsequent analysis and processing. For high-severity abnormalities, the full-machine test run is halted and an alarm is issued, prompting personnel to intervene. This multi-level warning mechanism not only accurately classifies abnormal conditions but also flexibly selects response measures based on the severity of the abnormality, avoiding unnecessary downtime while enabling rapid response at critical moments, maximizing the safety and reliability of system operation.

[0057] like Figure 2As shown, the warning mechanism is graded based on real-time data and warning thresholds. When real-time data is greater than 80% and less than 100% of the warning threshold, the control module enters the first warning level; when real-time data is greater than or equal to 100% and less than 120% of the warning threshold, the control module enters the second warning level; and when real-time data is greater than 120% of the warning threshold, the control module enters the third warning level. Warning level classification is based on the magnitude of anomalies in real-time sensor data, and the warning level is determined by comparing the deviation between the real-time data and the warning threshold. When the magnitude of the anomaly in real-time data is small, between 80% and 100% of the warning threshold, the control system enters the first warning level. When the real-time data exceeds the set threshold, the warning level rises to the second warning level; and when the real-time data exceeds 1.2 times the warning level, it is marked as the third warning level. This warning level classification method based on the magnitude of the anomaly allows the system to classify faults of varying severity and respond accordingly, improving overall system security.

[0058] Example 2

[0059] This embodiment is the second example of a method for monitoring structural damage during a turboshaft engine full-machine test. This embodiment is similar to the first example, differing in that, in step S5, the dynamic adjustment of the sampling frequency and detection threshold involves increasing the warning threshold and decreasing the sampling frequency when the health assessment report indicates normal operating status; and decreasing the warning threshold and increasing the sampling frequency when the health assessment report indicates an abnormality or potential risk. When the engine is operating normally, the sampling frequency and standard threshold are maintained at a low level to reduce resource consumption. When an abnormal trend or potential risk is detected, the sampling frequency is automatically increased and the threshold sensitivity is optimized to ensure real-time capture of critical data. This dynamic adjustment mechanism enables rapid response to state changes, improving the efficiency and accuracy of fault monitoring, while also enhancing the reliability of the detection structure and reducing resource consumption. To ensure sensitive fault detection and prediction accuracy, the present invention incorporates a dynamic adjustment mechanism based on fault prediction and actual fault report data. This mechanism dynamically adjusts the warning threshold and sampling frequency by analyzing operational data collected by sensors in real time to meet monitoring requirements under varying operating conditions, ensuring timely response to abnormal conditions while optimizing resource utilization.

[0060] In step S5, the early warning threshold calculation formula is:

[0061] T d =α·V real-time +(1-α)·T base

[0062] Among them, T d is the warning threshold, V real-time is the real-time sensor data value, Tbase The threshold is set based on α, which is a weighting coefficient and the range is 0<α<1. d The dynamic adjustment of V based on real-time sensor data value real-time With the basic setting threshold T base The deviation between them is adjusted by using the weighting coefficient α for dynamic correction. The dynamic adjustment logic is as follows: When the real-time sensor data value V real-time Stable and close to the basic setting threshold T base When the weighted coefficient α changes slightly, the warning threshold T d The adjustment range is small to ensure that the normal operation state of slight fluctuations does not trigger false alarms; when the real-time sensor data value V real-time When the change range increases or approaches the abnormal range, the warning threshold T is quickly adjusted by reducing the weighting coefficient α. d , thereby improving detection sensitivity to capture abnormal engine trends.

[0063] In step S5, the sampling frequency calculation formula is:

[0064] f s =f base ×(1+k×|ΔV|)

[0065] Among them, f s is the sampling frequency, f base is the basic sampling frequency, k is the adjustment coefficient, and ΔV represents the rate of change of sensor data. Sampling frequency f s The dynamic adjustment is used to achieve a quick response when an abnormal state is detected by increasing the data sampling frequency f s To capture key data and ensure accurate analysis of faults, the adjustment logic is as follows: When the engine is running normally, the sampling frequency f s Maintain the basic sampling frequency f base level, reducing resource consumption. When the sensor data fluctuation intensifies or approaches the warning threshold, the sampling frequency f is dynamically increased according to the change rate ΔV of the sensor data. s , to capture more detailed operating data, record data change trends in real time, provide high-precision input for the digital twin model, and support abnormal state analysis and fault prediction.

[0066] Example 3

[0067] This embodiment is the first embodiment of the turboshaft engine whole-machine test structural damage monitoring system, the whole-machine test module is used to drive the turboshaft engine to perform whole-machine test;

[0068] Data acquisition module: including multimodal sensors, airflow detection components and multispectral imaging components, used to obtain real-time data during the operation of the whole machine test module and transmit the real-time data to the data processing module;

[0069] Data processing module: used to perform data fusion processing on real-time data, map it to the digital twin model, and analyze the real-time data through deep learning algorithms to generate health status assessment reports:

[0070] Control module: used to dynamically adjust the warning threshold T of the multimodal sensor, airflow detection component and multispectral imaging component d and sampling frequency f s And control the working status of the whole machine test module.

[0071] like Figure 3 As shown, the turboshaft engine whole-machine test structural damage monitoring system in this embodiment uses a whole-machine test module to drive the turboshaft engine to conduct a whole-machine test on the ground, simulating the operating conditions that the engine may encounter in actual use, discovering potential dynamic defects during operation, and thus evaluating its health status and failure trends; the data acquisition module obtains real-time data of the whole-machine test module during operation through multimodal sensors, airflow detection components, and multispectral imaging components, including temperature, strain, vibration, pressure, flow rate, flow rate and other information, and transmits the relevant information to the data processing module; the data processing module fuses the data and maps it to the digital twin model for analysis to generate a health status assessment report; the control module adjusts the sampling frequency of the data collection module according to the health status assessment status, controls the data processing module to record engine data under abnormal conditions, and regulates the working status of the whole-machine test module.

[0072] In the specific contents of the above-mentioned specific implementation methods, the various technical features can be combined in any non-contradictory manner. In order to make the description concise, not all possible combinations of the above-mentioned technical features are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0073] Obviously, the above embodiments of the present invention are merely examples for the purpose of clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. A method for monitoring structural damage during a complete turboshaft engine test run, characterized in that: The following steps are involved: S1. With a turboshaft engine in a stationary state, a multimodal sensor is mounted on the turboshaft engine, an airflow detection assembly is mounted at the air inlet and exhaust regions of the turboshaft engine, and a multispectral imaging assembly is mounted at the exhaust region, wherein the multimodal sensor is used to collect temperature, strain, vibration, and pressure information of the turboshaft engine, the airflow detection assembly is used to collect flow velocity and flow rate information of the intake and exhaust gases, and the multispectral imaging assembly is used to monitor the temperature field distribution of the exhaust gases in real time; S2. Constructing a digital twin model with the same structure as the turboshaft engine, integrating the installation positions of the multimodal sensor, the airflow detection component, and the multispectral imaging component into the digital twin model, so that the installation positions of the multimodal sensor, the airflow detection component, and the multispectral imaging component in the digital twin model correspond one-to-one to the actual turboshaft engine positions; S3. Driving the turboshaft engine to perform a full-machine test run to collect real-time data, where the real-time data includes multimodal sensor data, airflow detection component data, and multispectral imaging component data; performing data fusion processing on the real-time data and synchronously mapping the data to the digital twin model in real time; S4. Based on the digital twin model, analyze the real-time data using a deep learning algorithm to obtain actual operation information, actual fault information, and fault prediction information of the component, and generate a health status assessment report; S5. Based on the health status assessment report, dynamically adjust the warning thresholds T of the multimodal sensor, airflow detection component, and multispectral imaging component. d and sampling frequency f s ; S6, based on the warning threshold T updated in step S5 d and sampling frequency f s , repeat steps S1-S5 to obtain the actual operation information, actual fault information and fault prediction information of the component.

2. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 1, wherein: In step S1, the multimodal sensor includes a temperature sensor, a fiber Bragg grating strain sensor, a triaxial vibration sensor, and a dynamic pressure sensor.

3. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 1, wherein: In step S3, the operating conditions of the whole machine test run include normal operating conditions, extreme temperature conditions and high load conditions.

4. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 1, wherein: In step S3, the data fusion processing includes time synchronization adjustment, noise filtering optimization, spatial mapping alignment and real-time consistency processing of the real-time data.

5. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 1, wherein: In step S4, based on the digital twin model, real-time data and historical data are jointly analyzed through deep learning algorithms to obtain the actual operation information, actual fault information and fault prediction information of the components.

6. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 1, wherein: In step S5, the dynamic adjustment of the sampling frequency and the detection threshold includes: when the health status assessment report feedback indicates that the operating status is normal, increasing the warning threshold and reducing the sampling frequency; when the health status assessment report feedback indicates that there is an abnormality or potential risk in the detection, reducing the warning threshold and increasing the sampling frequency.

7. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 5, wherein: In step S5, the warning threshold calculation formula is: T d =α·V real-time +(1-a)·T base Among them, T d is the warning threshold, V real-time is the real-time sensor data value, T base The threshold is set based on α, and α is the weighting coefficient, ranging from 0<α<1.

8. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 5, wherein: In step S5, the sampling frequency calculation formula is: f s =f base ×(1+k×|ΔV|) Among them, f s is the sampling frequency, f base is the basic sampling frequency, k is the adjustment coefficient, and ΔV represents the rate of change of sensor data.

9. The method for monitoring structural damage during a complete turboshaft engine test run according to claim 1, wherein: The method further includes step S7, triggering a multi-level warning mechanism based on the real-time data and the warning threshold, collecting abnormal data, generating an abnormal analysis report or stopping the whole machine test run.

10. A turboshaft engine complete machine test structural damage monitoring system, used to implement the turboshaft engine complete machine test structural damage monitoring method according to any one of claims 1 to 9, characterized in that: include: Engine test module: used to drive the turboshaft engine for engine test; Data acquisition module: including multimodal sensors, airflow detection components and multispectral imaging components, used to obtain real-time data during the operation of the whole machine test module and transmit the real-time data to the data processing module; Data processing module: used to perform data fusion processing on the real-time data, map it to the digital twin model, and analyze the real-time data through deep learning algorithms to generate a health status assessment report: Control module: used to dynamically adjust the warning threshold T of the multimodal sensor, airflow detection component and multispectral imaging component d and sampling frequency f s And control the working status of the whole machine test module.

Images