A turboprop integrated control system for a multi-engine aircraft

Abstract

The application belongs to the technical field of aircraft control, and specifically discloses a turbo-propeller integrated control system for a multi-engine aircraft, which comprises a total control panel, a plurality of independent control modules, a redundant bus and a fault-tolerant module, adopts a triple-redundancy control logic, and the priority from high to low is in turn manual backup control, MCP instruction control and automatic control. The automatic control identifies a flight stage through a multi-source sensor and sets a corresponding optimal rotating speed; the MCP instruction control issues a unified instruction through the total control panel to realize the collaborative operation of the multi-engine; and the manual backup control automatically switches the control authority when an engine fails to realize system-level fault tolerance. The application simplifies the operation and significantly improves the collaboration, reliability and safety of the multi-engine control.

Description

Technical Field

[0001] This invention belongs to the field of aircraft control technology, and specifically relates to an integrated turboprop control system for multi-engine aircraft. Background Technology

[0002] Modern multi-engine turboprop aircraft are typically equipped with multiple engines to provide ample power and redundancy for safety. Traditional control systems, such as those for the PW150A engine, usually employ a multi-stick (e.g., independent throttle and pitch sticks) independent or semi-independent operation mode. Pilots need to coordinate multiple control stick positions simultaneously to precisely manage the power output of each engine and the propeller speed. This operating mode is not only cumbersome and greatly increases the pilot's workload, but also prone to risks due to operational errors during critical phases such as takeoff and landing.

[0003] With the development of aviation technology, the need to simplify cockpit human-machine interaction, improve operational efficiency, and enhance system safety is becoming increasingly urgent. For example, the PW150C engine has been designed to integrate with avionics and control systems to reduce pilot workload. Existing technologies have also explored control system architecture. For instance, Chinese patent CN111216904B discloses an "engine and propeller control system," which provides a dual-channel redundant electronic controller. By setting independent control and protection processors within a single controller and employing dual-coil sensors, it improves the reliability and safety of single-engine control, preventing it from entering dangerous operating modes.

[0004] However, this existing technology still has the following limitations: First, its core focus is on ensuring the reliability of a single engine controller, without addressing the issue of coordinated control between multiple engines, and thus failing to achieve the single-lever function that simplifies operation. Second, this solution lacks consideration for system-level fault tolerance across multiple engines. When one engine fails, the solution is to switch channels within the controller, but it does not address how to isolate the failed engine from the normal engines, or how to reconfigure the entire control system to maintain maximum availability. Summary of the Invention

[0005] The purpose of this invention is to overcome the deficiencies in the prior art and provide an integrated turboprop control system for multi-engine aircraft.

[0006] This invention provides an integrated turboprop control system for a multi-engine aircraft, the multi-engine aircraft comprising at least two engines, each engine driving a propeller, the control system comprising:

[0007] The main control panel is used to receive mode selection instructions from the pilot and generate unified control instructions.

[0008] Multiple independent control modules, the number of which corresponds to the number of engines, each of which is configured to communicate with the corresponding engine and propeller for independent control of the engine and propeller;

[0009] A redundant bus is used for communication connection between the main control panel and the independent control modules;

[0010] A fault-tolerant module, connected to the main control panel, is used to determine whether the engine has malfunctioned and to send the fault type to the main control panel.

[0011] The control system adopts triple redundancy control logic, with the priority from high to low being manual backup control, MCP command control, and automatic control.

[0012] The automatic control is configured to identify the flight phase and set the optimal speed in normal operating mode; the MCP command control is configured to send the unified control command to the independent control modules through the redundant bus to enable multiple engines to operate in coordination; the manual backup control is configured to: when the fault-tolerant module detects a failure in any engine, send a fault signal to the main control panel; the main control panel responds to the fault signal and sends a control switching command to all independent control modules through the redundant bus, so that the control authority of the failed engine is assigned to the independent control module corresponding to the failed engine, while other engines maintain their original settings.

[0013] A further embodiment is that the automatic control process includes:

[0014] The system acquires signals from multiple sensors, identifies flight phases based on these signals, sets the optimal propeller speed for each flight phase, continuously monitors the flight phases, and automatically updates the target speed command when the flight phase changes.

[0015] The multi-source sensor signals include at least: wheel load signal, landing gear position signal, flap position signal, autopilot status signal, autothrottle status signal, and altitude interception status signal;

[0016] The mapping relationship between the flight phase and the target rotation speed is as follows:

[0017] When the wheel load signal is valid, it is determined to be in ground condition, and the target speed is set to 1020 RPM;

[0018] When the wheel load signal is invalid, the landing gear is retracted, the flaps are retracted, and the flight time exceeds the climb threshold, it is determined to be in climb mode, and the target speed is set to 900 RPM.

[0019] When the autopilot is engaged, the auto throttle is engaged, and the altitude has been intercepted, it is determined to be in cruise mode, and the target speed is set to 850 RPM;

[0020] When the landing gear is down or the flap angle is greater than or equal to the takeoff set angle, it is determined to be the maximum takeoff state, and the set target speed is 1020 RPM.

[0021] A further embodiment is that the main control panel includes a push-button switch; the push-button switch includes a power button and three mutually exclusive command switches: MCR, MCL, and MTO.

[0022] A further embodiment is that the MCP instruction control includes:

[0023] The push-in switch responds to the unified control command to switch the aircraft between main states and synchronously sends a unified rotation speed command to all the independent control modules.

[0024] The main states include cruise, climb and maximum takeoff, with different propeller speeds corresponding to different main states.

[0025] A further embodiment is that the independent control module is a parking switch, which has multiple physical positions corresponding to different engine control states; the parking switch is equipped with a resistance sensor, and different physical positions correspond to different resistance values, and the main control panel identifies the physical positions based on the resistance values.

[0026] A further embodiment is that the parking switch includes an outer large-ring knob and an inner small-ring knob. The outer large ring is provided with three mechanical positions: CUT / OFF, FTHS, and RUN. The position is selected by controlling the outer large-ring knob.

[0027] The inner small circle has three virtual gear positions: MCR, MCL, and MTO. When the outer large circle knob is in the RUN position, the inner small circle knob is used to control the selection of the virtual gear.

[0028] The parking switch integrates a display to show the parking switch's position.

[0029] A further solution is that the virtual gear position control of the parking switch is determined by the knob offset. and holding time Joint decision:

[0030] when <- and > At that time, the monitor displays the MCR setting;

[0031] When | |≤ At that time, the monitor displays the MCL setting;

[0032] when > and > At that time, the display shows the MTO gear position;

[0033] If the above conditions are not met, the display shows the previous indicator position;

[0034] in, As the trigger threshold, Dead zone threshold, For the time to remove jitter.

[0035] A further solution includes a fault detection module, which is connected to the fault tolerance module and each of the engines, respectively, for acquiring engine operating status parameters, determining the fault type of the engine based on the operating status parameters, and sending the fault type to the fault tolerance module.

[0036] The operating parameters include: engine speed, turbine inlet temperature, lubricating oil pressure, engine vibration amplitude, and fuel flow rate.

[0037] A further approach is to determine the fault type using a fault diagnosis model;

[0038] The process of constructing the fault diagnosis model is as follows:

[0039] Obtain a first parameter group, which includes operating status parameters collected by the fault detection module from the engine, and system health status labels corresponding to the operating status parameters; the system health status labels include normal operation or specific fault types.

[0040] Obtain a second parameter group, which includes high-level features extracted from the running status parameters and system health status labels corresponding to the high-level features; the high-level features include: time-domain statistical features, frequency-domain features, time-frequency-domain features, and correlation features between different parameters;

[0041] The first parameter group and the second parameter group were labeled by human experts, and the first parameter group and the second parameter group were divided into training set and test set according to the ratio;

[0042] The training set of the first parameter group is input into the first machine learning unit for iterative training, so that the first machine learning unit learns the mapping relationship from the running state parameters to the system health state.

[0043] The training set of the second parameter group is input into the second machine learning unit for iterative training, so that the second machine learning unit learns the mapping relationship from high-level features to the system health status.

[0044] The first machine learning unit and the second machine learning unit, after training, are combined to obtain a fault diagnosis model that outputs fault types based on running state parameters and high-level features.

[0045] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0046] This invention generates unified control commands through a central control panel and distributes them to all independent control modules via a redundant bus, achieving coordinated and integrated control of multiple engines. This frees pilots from cumbersome multi-stick independent operations, significantly simplifying the operation process and reducing the risk of human error. A fault-tolerant mechanism is established through triple-redundant control logic: in normal operating modes such as automatic control or MCP command control, the system operates efficiently and collaboratively; once any independent control module detects an engine failure, manual backup control is immediately activated. The central control panel, through the fault-tolerant module, detects the failure and issues a control switch command, precisely isolating the control authority of the failed engine and transferring it to the corresponding module, while other non-failed engines maintain their original efficient collaborative settings. This ensures that even in the event of single-point or multi-point failures, the entire power system maintains maximum availability and flight stability, achieving a leap from traditional single-engine internal redundancy to multi-engine system-level fault tolerance, greatly enhancing the overall safety margin of multi-engine aircraft.

[0047] This invention intelligently identifies flight phases using multi-source sensor signals and sets the optimal propeller speed for each phase. It transforms pilot experience into precise, automated decision-making, ensuring the engine operates at preset optimal conditions in all flight phases: automatically setting a high speed of 1020 RPM for ground and maximum takeoff states to provide sufficient thrust; setting 900 RPM during climb to balance climb efficiency and fuel economy; and setting 850 RPM during cruise to minimize noise, vibration, and fuel consumption. This not only continuously maintains the engine within its high-efficiency range, improving overall flight efficiency and economy, but also effectively avoids underutilization or excessive wear of engine performance caused by pilot subjective judgment delays or errors, achieving refined and intelligent power management throughout the entire flight process.

[0048] This invention achieves a high degree of integration in multi-engine control by setting up a power button and three mutually exclusive command switches (MCR, MCL, and MTO), and associating these switches with main states such as cruise, climb, and maximum takeoff. Pilots no longer need to operate multiple independent throttle and pitch levers separately; they only need to select the appropriate flight mode through an integrated switch, and a unified RPM command can be synchronously issued to all independent control modules via a redundant bus. This significantly reduces cockpit complexity, allowing pilots to focus more on flight path management and situational awareness. The mutually exclusive command switches physically prevent the issuance of conflicting commands, ensuring the clarity and uniqueness of the control logic, thereby guaranteeing a high degree of synchronization and consistency in thrust output of multiple engines during critical flight phases, improving flight stability and safety.

[0049] The parking switch of this invention combines the outer large-circle mechanical positions CUT / OFF, FTHS, and RUN with the inner small-circle virtual positions MCR, MCL, and MTO, and integrates a display and a resistance sensor. The physical positions provide direct and reliable tactile feedback for critical operations, reducing the risk of misoperation. In the RUN position, precise mode selection is achieved by controlling the virtual positions via the inner small-circle knob. The switching of virtual positions is determined by the knob offset and hold time, and trigger thresholds, dead-zone thresholds, and de-jitter time are set, effectively filtering out accidental inputs caused by aircraft vibration or unintentional touches. This ensures that every position switch is the result of the pilot's explicit intention, greatly improving operational accuracy and anti-interference capability.

[0050] This invention continuously collects multi-dimensional operating status parameters, including engine speed, turbine inlet temperature, lubricating oil pressure, vibration amplitude, and fuel flow, through a fault detection module. It then utilizes a machine learning-based fault diagnosis model for in-depth analysis, achieving accurate diagnosis and fault prediction. The fault diagnosis model integrates a first parameter set and a second parameter set, trained separately by a first machine learning unit and a second machine learning unit, ultimately combining them into a powerful diagnostic engine. This enables the system not only to identify obvious parameter exceedances but also to uncover early, potential fault symptoms from complex data features and correlations, thereby achieving accurate fault type identification and early warning. Attached Figure Description

[0051] The following figures are for illustrative purposes only and are not intended to limit the scope of the invention, wherein:

[0052] Figure 1 Overall architecture diagram of the integrated turboprop control system for multi-engine aircraft;

[0053] Figure 2 Automatic control flight phase mapping diagram;

[0054] Figure 3 Schematic diagram of push-button switch and stop switch layout;

[0055] Figure 4 Schematic diagram of the parking switch structure;

[0056] Figure 5 Schematic diagram of the layout of the push-button switch and another stop switch;

[0057] Figure 6 : Another schematic diagram of the parking switch structure;

[0058] In the diagram: 1. Controlled object layer; 2. Actuator layer; 3. Main control panel; 4. Fault-tolerant module; 5. Fault diagnosis model; 6. Fault detection module; 7. Independent control module; 8. Redundant bus; 9. Push-in switch; 10. Automatic control module; 11. Stop switch; 12. Inner small-circle knob; 13. Outer large-circle knob; 14. Display; 15. Indicator panel; 16. Rebound knob. Detailed Implementation

[0059] To make the objectives, technical solutions, design methods, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0060] like Figure 1 As shown, the present invention provides an integrated turboprop control system for a multi-engine aircraft, the multi-engine aircraft comprising a controlled object layer 1 consisting of four engines and a propeller driven by each engine, the control system comprising:

[0061] A central control panel 3 is installed in the cockpit center console to receive pilot commands and generate unified control signals;

[0062] The actuator layer 2 is equipped with four independent control modules 7, which are used to control the controlled object layer 1, corresponding to four engines respectively. Each independent control module 7 is connected to the fuel regulation mechanism and propeller pitch mechanism of the corresponding engine through a dedicated wiring harness, and has the ability to independently control the speed, thrust and propeller pitch of a single engine. The independent control module 7 has a built-in processor that can store the control parameters of a single engine.

[0063] The redundant bus 8 adopts a dual-redundant bus architecture, connecting the main control panel 3 with four independent control modules 7. Bus A is the main channel and bus B is the backup channel. Under normal operating conditions, the two channels transmit commands synchronously. If either channel fails, the other channel automatically takes over to ensure uninterrupted command transmission.

[0064] The fault-tolerant module 4 is connected to the main control panel 3 and communicates with four independent control modules 7 in real time via the CAN bus. It continuously receives engine status data (such as speed, temperature, and vibration) uploaded by each independent control module 7 and has a built-in fault judgment algorithm to determine whether the engine has malfunctioned and send the fault type to the main control panel 3.

[0065] The control system adopts triple redundancy control logic, with the priority from high to low being manual backup control, MCP command control, and automatic control.

[0066] During normal flight, the system defaults to automatic control mode. The automatic control module 10 uses multi-source sensors to acquire wheel load signals, landing gear signals, etc., to identify the flight phase and generate unified optimal speed commands for the four engines. For example, during cruise, the command is to maintain all four engines at 850 RPM, adjusting the fuel supply and propeller pitch of the corresponding engines to ensure that the speed deviation is ≤ ±5 RPM. If the pilot needs to intervene manually, a unified control command is sent through the push-in switch 9 on the main control panel 3 to control the four engines and propellers. The manual backup control is configured as follows: when the fault-tolerant module 4 detects a fault in any engine, it sends a fault signal to the main control panel 3. The main control panel 3 responds to the fault signal by sending a control switching command to all independent control modules 7 through the redundant bus 8, so that the control authority of the faulty engine is assigned to the independent control module 7 corresponding to the faulty engine, while the other engines maintain their original settings. For example, if the fault-tolerant module 4 detects a sudden drop in the speed of an engine (such as from 850 RPM to 400 RPM, exceeding the normal threshold), it immediately sends a fault signal (including the faulty engine number and fault type) to the main control panel 3. After the main control panel 3 responds, it sends a control switching command to the four independent control modules 7 via the redundant bus 8: the control authority of the faulty engine is locked to its corresponding independent control module 7, while the independent control modules 7 of the other engines maintain the original cruise speed.

[0067] In this embodiment, multiple source sensors are deployed on the aircraft to collect signals in real time and transmit them to the fault-tolerant module 4. These multiple source sensors include: wheel load sensors (one for each of the four wheels to detect whether the wheels are under load); landing gear position sensors to detect the retracted or extended state of the landing gear; flap position sensors (one for each of the left and right wings to detect the flap angle); autopilot sensors and autothrottle status sensors (integrated into the flight control system, outputting on / off signals); and altitude acquisition sensors (integrated into the navigation system, outputting a signal indicating whether the target altitude has been acquired or not). The automatic control process includes:

[0068] like Figure 2As shown, multi-source sensor signals are acquired through multiple sensors, flight phases are identified based on these signals, and an optimal propeller speed is set for each flight phase. The system continuously monitors the flight phases and automatically updates the target speed command when the flight phase changes. The multi-source sensor signals include: wheel load signals, landing gear position signals, flap position signals, autopilot status signals, autothrottle status signals, and altitude interception status signals. Different aircraft states and control processes are as follows:

[0069] Ground condition determination and control: When all four wheel load sensors output valid signals, that is, when the wheels are under force, the fault-tolerant module 4 determines that it is a ground condition and generates a command for the four engines to reach a speed of 1020 RPM. Then, it controls the fuel flow of the engines to make the speed reach 1020 RPM. At the same time, the propeller pitch mechanism adjusts the pitch to the ground idle speed matching value to ensure that the engines are not overloaded.

[0070] Climb Status Determination and Control: When all four wheel load sensors output invalid signals, all four landing gear position sensors output "retracted," the flap sensor outputs "retracted," and the flight time exceeds the climb threshold, a climb status is determined, and the target engine speed command is adjusted to 900 RPM. The engine is controlled to reduce fuel supply, causing the engine speed to smoothly decrease from 1020 RPM to 900 RPM, avoiding sudden thrust drops that could cause flight attitude fluctuations.

[0071] Cruise Status Determination and Control: When the autopilot sensor output is on, the autothrottle sensor output is on, and the altitude interception sensor output is intercepted, the system is determined to be in cruise mode, and the target engine speed command is set to 850 RPM. Fuel flow and propeller pitch are further optimized to minimize fuel consumption of all four engines while maintaining stable engine speed.

[0072] Maximum takeoff state determination and control: When any landing gear position sensor outputs "down" or the flap sensor outputs a takeoff angle ≥15°, the maximum takeoff state is determined, and the target engine speed command is restored to 1020 RPM. The engine is controlled to increase fuel supply, raising the speed to 1020 RPM to provide maximum thrust for takeoff, while the engine turbine inlet temperature is monitored in real time to prevent overheating.

[0073] like Figure 3As shown, in this embodiment, a push-button switch 9 is provided on the front of the main control panel 3, including a power button and three mutually exclusive command switches. The power button is a self-resetting button. When pressed, it connects the power supply to the entire control system, and when released, it remains powered. Pressing and holding it for 3 seconds will cut off the power. The mutually exclusive command switches are set with three positions: MCR (cruise), MCL (climb), and MTO (maximum takeoff). When any position is pressed, the other positions are locked and cannot be pressed. The specific functions include: pressing the MCR position: sending a cruise speed command (850 RPM) to the engine; pressing the MCL position: sending a climb speed command (900 RPM); and pressing the MTO position: sending a maximum takeoff speed command (1020 RPM). According to flight requirements, the pilot presses the MTO position of the push-in switch 9 on the main control panel 3. The main control panel 3 generates a unified MTO speed command. After receiving the command, the four engines immediately start their local control logic: adjusting the opening of the engine fuel regulating valve to the corresponding MTO value, and simultaneously controlling the propeller pitch mechanism to adjust the pitch to the maximum thrust matching value. After the command is executed, a feedback signal is sent to the main control panel 3. The display screen of the main control panel 3 updates the speed status of the four engines, completing the MCP command control closed loop.

[0074] Continue to refer to Figure 3 In this embodiment, the independent control module 7 is a parking switch 11, which has three physical positions corresponding to different engine control states. The parking switch 11 has an internal resistance sensor, with different resistance values ​​corresponding to different physical positions. The main control panel 3 identifies the physical positions based on these resistance values. The three physical positions are: CUT / OFF: cuts off engine fuel supply; FTHS: propeller constant speed control; RUN: manual adjustment, where the speed mode can be adjusted via the inner small knob 12. The parking switch 11 has a built-in precision resistance sensor, with each physical position corresponding to a unique resistance value. In this embodiment, the CUT / OFF position corresponds to 100Ω, the FTHS position to 200Ω, and the RUN position to 300Ω. The main control panel 3 collects the resistance values ​​of the four parking switches 11 in real time via the redundant bus 8. For example, when the resistance value of a parking switch 11 is detected to be 300Ω, it is identified as the RUN position and displayed on the screen.

[0075] like Figure 4As shown, the parking switch 11 includes an outer large-circle knob 13 and an inner small-circle knob 12. The outer large-circle knob has three mechanical positions: CUT / OFF, FTHS, and RUN, which are selected by the outer large-circle knob 13. The inner small-circle knob has three virtual positions: MCR, MCL, and MTO. When the outer large-circle knob 13 is in the RUN position, the inner small-circle knob 12 is used to select the virtual position. The parking switch 11 integrates a display 14 to show the position of the parking switch 11. The virtual position control of the parking switch 11 is achieved by the knob offset. and holding time Joint decision: When <- and > When | |≤ When, monitor 14 displays the MCL setting; when > and > When the above conditions are not met, the display 14 shows the MTO gear position; when the above conditions are not met, the display 14 shows the previous indicated gear position; wherein, As the trigger threshold, Dead zone threshold, To reduce jitter. For example: rotate the inner small knob 12 to the left. =-6°<- =-5°, and hold time =0.6 seconds> =0.5 seconds: Display 14 switches to "RUN-MCR", independent control module 7 sends an 850RPM command to the engine; rotation angle =1°,| |≤ =2°: Regardless of the duration, the display 14 maintains the previous position; rotate the inner small ring knob 12 to the right. =7°> =5°, and holding time =0.7 seconds> =0.5 seconds: Display 14 switches to "RUN-MTO", engine speed increases to 1020 RPM; rotate the inner small knob 12 to the right, =6°, but holding time =0.3 seconds < =0.5 seconds: After release, the display 14 returns to the previous gear position to avoid gear changes caused by vibration or accidental touch. In this embodiment, another structure for the parking switch 11 is also provided, such as... Figure 5 and Figure 6As shown, the parking switch 11 is a spring-loaded knob type, specifically including an indicator panel 15 and a spring-loaded knob 16. The indicator panel 15 also integrates a display 14. In its initial state, pressing the spring-loaded knob 16 allows selection of OFF, FTHS, and RUN positions by rotating it left or right. When the RUN position is selected, lifting the spring-loaded knob 16 allows it to be turned left or right, and the internal spring mechanism automatically returns it to the center. The three physical positions of left turn, neutral, and right turn are defined as three virtual positions, corresponding to the three target speeds of cruise (MCR, 850 RPM), climb (MCL, 900 RPM), and maximum takeoff (MTO, 1020 RPM), respectively. The CUT / OFF and FTHS positions employ a pin-type anti-misoperation design.

[0076] In this embodiment, a fault detection module 6 is installed in the fuselage electronics bay of the four-engine aircraft. It is directly connected to the sensors of the four engines through a dedicated detection harness, and communicates with the fault-tolerant module 4 through the CAN bus to realize a closed loop of parameter acquisition, fault diagnosis and signal transmission.

[0077] Operating status parameters include:

[0078] Engine speed: acquired via a speed sensor;

[0079] Turbine inlet temperature: acquired via thermocouple sensor;

[0080] Lubricating oil pressure: acquired via a pressure sensor;

[0081] Engine vibration amplitude: collected by vibration sensors;

[0082] Fuel flow rate: Collected via a flow sensor.

[0083] In the above, the fault type is determined by fault diagnosis model 5;

[0084] The construction process of the fault diagnosis model 5 is as follows:

[0085] Collect historical operating data of the aircraft: including 1,000 sets of normal operating condition data and 500 sets of fault operating condition data. Among them, the fault operating condition data covers 10 types of faults such as turbine overheating, insufficient lubricating oil pressure, and abnormal speed, forming the first parameter group.

[0086] Advanced features are extracted: Signal processing is performed on the original parameters to generate time-domain statistical features such as mean speed and temperature variance, peak frequency features of vibration signals, time-frequency domain features, and parameter correlation features (such as the correlation between temperature and fuel flow), forming the second parameter group; health labels of the two parameter groups are reviewed and corrected by aero-engine maintenance experts. The first and second parameter groups are both divided into training and test sets in a 7:3 ratio, with no data overlap between the training and test sets.

[0087] The training set of the first parameter group is input into the first machine learning unit for iterative training, so that the first machine learning unit learns the mapping relationship from the running state parameters to the system health state.

[0088] The training set of the second parameter group is input into the second machine learning unit for iterative training, so that the second machine learning unit learns the mapping relationship from high-level features to the system health status.

[0089] The first machine learning unit and the second machine learning unit, after training, are combined to obtain a fault diagnosis model 5 based on the operating state parameters and high-level features to output fault types.

[0090] The first machine learning unit is configured to use a neural network model suitable for processing time-series data, specifically a one-dimensional convolutional neural network or a long short-term memory network. Through iterative training, the first machine learning unit learns the end-to-end nonlinear mapping relationship from the original operating state parameters to the system health state, with the advantage of automatically capturing local patterns and time dependencies in the parameters. The second machine learning unit is configured to use a machine learning model with strong processing capabilities for structured features, specifically a gradient boosting decision tree or a support vector machine. Through iterative training, the second machine learning unit learns the mapping relationship from the high-level features to the system health state, providing highly interpretable and accurate judgments on fault mechanisms. The first machine learning unit and the second machine learning unit are combined through a feature fusion layer. Specifically, during model training and inference: the first parameter set is input into the first machine learning unit, and the first machine learning unit outputs a first feature vector at the last fully connected layer; the second parameter set is input into the second machine learning unit, and the second machine learning unit outputs a second feature vector before the decision function; the first feature vector and the second feature vector are concatenated to form a joint feature vector; the joint feature vector is fed into a shared fully connected classification layer, processed by the Softmax activation function, and finally outputs the diagnostic probability distribution corresponding to different fault types.

[0091] The fault detection module 6 synchronously inputs the first and second parameter groups of the four engines into the fault diagnosis model 5. The fault diagnosis model 5 analyzes the parameters of each engine independently. For example, if the original parameters of engine #2 are 700 RPM, 880℃, and 1.8 bar, and the advanced features are increased speed variance and decreased correlation between temperature and flow rate, the fault diagnosis model 5 will output a fault of insufficient lubricating oil pressure for engine #2. The fault diagnosis model 5 sends the fault type to the fault tolerance module 4. After confirming the fault, the fault tolerance module 4 triggers the alarm and manual backup control logic of the main control panel 3.

[0092] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. An integrated turboprop control system for a multi-engine aircraft, the multi-engine aircraft comprising at least two engines, each engine driving a propeller, characterized in that, The control system includes: The main control panel is used to receive mode selection instructions from the pilot and generate unified control instructions. Multiple independent control modules, the number of which corresponds to the number of engines, each of which is configured to communicate with the corresponding engine and propeller for independent control of the engine and propeller; A redundant bus is used for communication connection between the main control panel and the independent control modules; A fault-tolerant module, connected to the main control panel, is used to determine whether the engine has malfunctioned and to send the fault type to the main control panel. The control system adopts triple redundancy control logic, with the priority from high to low being manual backup control, MCP command control, and automatic control. The automatic control is configured to identify flight phases and set optimal speeds using multi-source sensors during normal operation. The MCP command control is configured to send unified control commands to the independent control modules via the redundant bus to enable multiple engines to operate collaboratively. The manual backup control is configured to send a fault signal to the main control panel when the fault-tolerant module detects a fault in any engine. The main control panel responds to the fault signal by sending a control switching command to all independent control modules via the redundant bus, so that control authority for the faulty engine is assigned to the independent control module corresponding to the faulty engine, while other engines maintain their original settings.

2. The integrated turboprop control system for multi-engine aircraft according to claim 1, characterized in that, The automatic control process includes: The system acquires signals from multiple sensors, identifies flight phases based on these signals, sets the optimal propeller speed for each flight phase, continuously monitors the flight phases, and automatically updates the target speed command when the flight phase changes. The multi-source sensor signals include at least: wheel load signal, landing gear position signal, flap position signal, autopilot status signal, autothrottle status signal, and altitude interception status signal; The mapping relationship between the flight phase and the target rotation speed is as follows: When the wheel load signal is valid, it is determined to be in ground condition, and the target speed is set to 1020 RPM; When the wheel load signal is invalid, the landing gear is retracted, the flaps are retracted, and the flight time exceeds the climb threshold, it is determined to be in climb mode, and the target speed is set to 900 RPM. When the autopilot is engaged, the auto throttle is engaged, and the altitude has been intercepted, it is determined to be in cruise mode, and the target speed is set to 850 RPM; When the landing gear is down or the flap angle is greater than or equal to the takeoff set angle, it is determined to be the maximum takeoff state, and the set target speed is 1020 RPM.

3. The integrated turboprop control system for multi-engine aircraft according to claim 1, characterized in that, The main control panel includes a push-button switch; the push-button switch includes a power button and three mutually exclusive command switches: MCR, MCL, and MTO.

4. The integrated turboprop control system for a multi-engine aircraft according to claim 3, characterized in that, The MCP command control includes: The push-in switch responds to the unified control command to switch the aircraft between main states and synchronously sends a unified rotation speed command to all the independent control modules. The main states include cruise, climb and maximum takeoff, with different propeller speeds corresponding to different main states.

5. The integrated turboprop control system for a multi-engine aircraft according to claim 1, characterized in that, The independent control module is a parking switch, which has multiple physical positions corresponding to different engine control states. The parking switch is equipped with a resistance sensor, and different physical positions correspond to different resistance values. The main control panel identifies the physical positions based on the resistance values.

6. The integrated turboprop control system for a multi-engine aircraft according to claim 5, characterized in that, The parking switch includes an outer large-circle knob and an inner small-circle knob. The outer large-circle knob has three mechanical positions: CUT / OFF, FTHS, and RUN. The position is selected by controlling the outer large-circle knob. The inner small circle has three virtual gear positions: MCR, MCL, and MTO. When the outer large circle knob is in the RUN position, the inner small circle knob is used to control the selection of the virtual gear. The parking switch integrates a display to show the parking switch's position.

7. The integrated turboprop control system for a multi-engine aircraft according to claim 6, characterized in that, The virtual gear position control of the parking switch is achieved by the knob offset. and holding time Joint decision: when <- and > At that time, the monitor displays the MCR setting; When | |≤ At that time, the monitor displays the MCL setting; when > and > At that time, the display shows the MTO gear position; If the above conditions are not met, the display shows the previous indicator position; in, As the trigger threshold, Dead zone threshold, For the time to remove jitter.

8. The integrated turboprop control system for a multi-engine aircraft according to claim 1, characterized in that, It also includes a fault detection module, which is connected to the fault tolerance module and each of the engines respectively, for acquiring engine operating status parameters, determining the fault type of the engine based on the operating status parameters, and sending the fault type to the fault tolerance module; The operating parameters include: engine speed, turbine inlet temperature, lubricating oil pressure, engine vibration amplitude, and fuel flow rate.

9. The integrated turboprop control system for a multi-engine aircraft according to claim 8, characterized in that, The fault type is determined by a fault diagnosis model; The process of constructing the fault diagnosis model is as follows: Obtain a first parameter group, which includes operating status parameters collected by the fault detection module from the engine, and system health status labels corresponding to the operating status parameters; The system health status label includes whether it is operating normally or a specific type of fault; Obtain a second parameter group, which includes high-level features extracted from the running status parameters and system health status labels corresponding to the high-level features; the high-level features include: time-domain statistical features, frequency-domain features, time-frequency-domain features, and correlation features between different parameters; The first parameter group and the second parameter group were labeled by human experts, and the first parameter group and the second parameter group were divided into training set and test set according to the ratio; The training set of the first parameter group is input into the first machine learning unit for iterative training, so that the first machine learning unit learns the mapping relationship from the running state parameters to the system health state. The training set of the second parameter group is input into the second machine learning unit for iterative training, so that the second machine learning unit learns the mapping relationship from high-level features to the system health status. The first machine learning unit and the second machine learning unit, after training, are combined to obtain a fault diagnosis model that outputs fault types based on running state parameters and high-level features.

Images