An online differential model-based flight control and pilot operation monitoring system optimization method

Abstract

The application discloses an online differential model-based flight control and pilot operation monitoring system optimization method, and belongs to the field of airplane design and airplane system design. Firstly, the flight stages of an airplane in flight are divided according to the flight characteristics of the airplane. Then, initial control models of the flight stages are established by using the operation statistical data of different pilots of the airplane in the flight stages. Finally, the initial control models of the flight stages of the same pilot are updated in the flight process of the airplane, and are changed into the differential model of the pilot, and the differential model of each pilot is taken as a benchmark to solve the pilot abnormal operation monitoring and the pilot-induced oscillation problem. The application can update the differential model online, provide more accurate model input for airplane related systems, and improve the system operation precision and efficiency.

Description

Technical Field

[0001] This invention belongs to the field of aircraft design and aircraft system design, specifically relating to a method for designing and optimizing parameters of a flight control and pilot operation monitoring system based on online differentiated operation modeling. Background Technology

[0002] Modern aircraft incorporate numerous automated systems, such as flight control systems and onboard surveillance systems. These systems monitor and record pilot inputs and, based on the control logic and laws established during the aircraft design phase, translate pilot intentions into control commands for various control surfaces, thereby controlling the aircraft's motion. These control logics and laws must consider the intervention and impact of pilot inputs on the automated systems and constrain and limit inappropriate pilot actions to prevent pilot errors from causing loss of control of the aircraft.

[0003] In the traditional design process of avionics systems and flight control laws, the pilot operation model is abstracted into a "standard digital pilot model" (hereinafter referred to as "standard model"), which is an "average pilot" model obtained by statistical analysis of a large amount of data that can replace the pilot's operation frequency and amplitude, and the control logic and control laws are designed accordingly.

[0004] Because of the use of a "standard model," almost all current control algorithms face the same difficulty: distinguishing between deviations in flight technique and deviations in pilot operation. Since each pilot's flight technique is different, their operation is not entirely equivalent to that of a "standard pilot." This means that various alarm thresholds and control law parameters set based on the characteristics of a "standard pilot" are not suitable for all pilots. For some pilots, the thresholds may be too low; for others, the settings may be too high. While in flight, these minor deviations will not have a serious impact. However, in the following two specific use cases, such deviations can lead to serious accidents:

[0005] Pilot Abnormal Operation Monitoring. This system is primarily used to monitor abnormal pilot maneuvers caused by special reasons (such as emotional fluctuations, fatigue, and incapacitation), thereby providing timely warnings or restraints to prevent flight accidents. In this scenario, if pilot operation data is collected and compared with a standard model, the system can hardly distinguish whether the deviation between the pilot's current operation and the standard model is due to flight technique or abnormal operation, rendering the system unusable in practice.

[0006] Pilot-induced oscillation suppression. Pilot-induced oscillation occurs when there is coupling between the pilot's control commands and the aircraft's control system commands, creating a phenomenon similar to "resonance." This leads to the aircraft gradually oscillating and becoming unstable until it loses control. During aircraft design, notch filters are typically placed near the manual control frequency to suppress coupled oscillations. However, because the design references a standard model, the system cannot completely recognize and suppress actual pilot commands during operation, and coupled oscillations may still occur. Summary of the Invention

[0007] To address the aforementioned issues, this invention proposes a parameter design optimization method for flight control and pilot operation monitoring systems based on online differentiated operation modeling. Building upon a "standard model," the method calculates and updates the actual pilot model in real time during flight, thereby providing more accurate pilot model inputs for systems or functions such as abnormal pilot operation monitoring and pilot-induced oscillation suppression. This enables these systems to perform parameter design more effectively and improve system performance.

[0008] An optimization method for a flight control and pilot operation monitoring system based on an online differentiation model includes the following steps:

[0009] Step one: Statistical analysis of aircraft currently in flight, and division of flight phases based on the flight characteristics of each aircraft.

[0010] The flight characteristics of the aircraft include engine status, aircraft flaps, landing gear status, air-to-ground signal status, and flight altitude.

[0011] The flight phases include takeoff, climb, cruise, descent, approach, landing, and go-around.

[0012] Step 2: For each flight phase, establish an initial control model for each flight phase by statistically analyzing the operational data of different pilots;

[0013] The specific process is as follows:

[0014] The operation data of different pilots in each flight phase are collected, and Fourier transform is performed to convert them into frequency domain data. The peak frequency and amplitude of the frequency domain data are extracted and averaged to obtain the control frequency and control amplitude, thereby obtaining the standard model for each flight phase, which serves as the initial control model for the corresponding flight phase.

[0015] Pilot operation data includes time history data for inputting control stick, control wheel, throttle, and pedal operations.

[0016] Step 3: Update the initial control model of the same pilot in each flight phase, and transform it into a differentiated model for that pilot.

[0017] Specifically:

[0018] First, during flight, determine the current flight phase of the aircraft and select the corresponding initial control model for updating.

[0019] Then, the pilot's control command data at each stage of flight is recorded in real time. Except for the cruise phase, the recorded data is divided into 10-second segments and subjected to Fourier transform; during the cruise phase, the data is divided into 100-second segments and subjected to Fourier transform.

[0020] Finally, the control frequency and control amplitude data corresponding to the peak values ​​are extracted from the frequency domain data of each flight phase, averaged with the initial control model corresponding to that phase, and used as a new model for the next update cycle.

[0021] As more tests are conducted, the actual models for each flight phase will gradually shift from standard models to customized models for each pilot.

[0022] Step four involves using a differentiated model as a benchmark to address issues related to monitoring abnormal pilot maneuvers and pilot-induced oscillations.

[0023] Pilot Abnormal Operation Monitoring: The technical characteristics of the corresponding pilots are obtained from the differentiation model. The deviation level is classified according to the degree to which the pilot deviates from his normal operating level. When a small deviation occurs, the pilot can be prompted by the instrument. When the deviation is large, the pilot can be alarmed and another pilot in the same flight can take over the operation.

[0024] Suppress pilot-induced oscillations: The difference between the control frequency and control amplitude of each pilot's differentiated model and the corresponding parameters of the standard model is used as a fine-tuning input to the flight control system, thereby correcting the relevant logic and control parameters of the flight control system.

[0025] The advantages of this invention are:

[0026] 1. This invention uses a "standard model" as the initial model. During flight, the model is continuously calculated and updated based on the pilot's actual operation. The model will gradually become more accurate and tend to stabilize as the flight time increases.

[0027] 2. The differentiated model obtained by this invention can provide more accurate model input for related systems, thereby improving system operating accuracy and efficiency.

[0028] 3. The principle of this invention is simple and easy to implement in programming; the computational load is small and the model can be updated online. Attached Figure Description

[0029] Figure 1This is a flowchart of a parameter design and optimization method for a flight control and pilot operation monitoring system based on online differentiated operation modeling, according to the present invention.

[0030] Figure 2 Example graph of time-domain curve of a pilot's stick operation;

[0031] Figure 3 This is an example graph showing the frequency domain characteristics of a pilot's stick control. Detailed Implementation

[0032] The present invention will now be described in further detail with reference to the accompanying drawings.

[0033] This invention relates to an optimization method for flight control and pilot operation monitoring systems based on an online differentiation model, such as... Figure 1 As shown, the specific steps are as follows:

[0034] Step 1: For aircraft that are in flight, divide the flight phase according to the flight characteristics of each aircraft.

[0035] The pilot's flying characteristics can vary significantly across different phases of flight. For example, during takeoff and landing, the focus is on safe takeoff / landing, requiring the pilot to frequently adjust the aircraft's flight status. During en-route flight, the focus is on flight quality, requiring the pilot to control the aircraft as gently as possible to provide a better passenger experience. Based on the typical characteristics of civil aircraft flight, the flight phases can be divided into takeoff, climb, cruise, descent, approach, landing, and go-around phases, according to engine status, flaps, landing gear status, air-to-ground signal status, and altitude (these phases can also be further subdivided based on aircraft type).

[0036] 1. Takeoff Phase: An aircraft is considered to be in the takeoff phase when all of the following conditions are met:

[0037] a) Air-to-ground signals last less than 30 seconds on the ground or in the air.

[0038] b) Landing gear down

[0039] c) The flaps are in takeoff configuration.

[0040] d) Engine throttle is set at takeoff thrust

[0041] 2. Climb Phase: An aircraft is considered to be in the climb phase when all of the following conditions are met:

[0042] a) The previous stage was the takeoff stage.

[0043] b) Air-to-ground signals in the air

[0044] c) The engine is located at the takeoff / climb thrust.

[0045] 3. Cruise Phase: An aircraft is considered to be in the cruise phase when all of the following conditions are met:

[0046] a) Air-to-ground signals in the air

[0047] b) The aircraft reached its cruising altitude.

[0048] c) Landing gear is in the retracted position.

[0049] d) The flaps are in the retracted position.

[0050] e) The engine provides cruise thrust.

[0051] 4. Decline Phase

[0052] a) The previous stage was in cruise mode.

[0053] b) The aircraft descends from cruising altitude

[0054] c) Engine revert to slow speed

[0055] d) The airspace is separated into airspace.

[0056] 5. Approach and Landing Phases

[0057] a) Landing gear down

[0058] b) Flaps are in approach / landing configuration

[0059] c) Air-to-ground signals in the air

[0060] 6. Resumption of flights phase

[0061] a) Air-to-ground signals in the air

[0062] b) The engine provides takeoff / go-around thrust.

[0063] c) Flaps are in approach / landing configuration

[0064] 7. Single-shot state

[0065] a) Air-to-ground signals in the air

[0066] b) Single engine failure

[0067] Step 2: For each flight phase, establish an initial control model for each flight phase by statistically analyzing the operational data of different pilots;

[0068] The control model comprises two parameters: control frequency and control amplitude. Fourier transforms are performed on the control data of different pilots at each flight phase to convert them into frequency domain data. The peak frequency and amplitude of the frequency domain data are extracted and averaged to obtain the standard model, which serves as the initial model. Pilot control statistics mainly include the time history data of stick operations, steering wheel operations, throttle operations, and pedal operations. Therefore, the standard model is divided into 7 flight phases, with each phase consisting of 4 control operations.

[0069] Taking the climb phase as an example, the time history data of the pilot's control stick input during a certain climb phase is as follows: Figure 2 As shown, the amplitude-frequency response curve of the data after Fourier transform is as follows: Figure 3 As shown in the figure, the pilot's joystick manipulation frequency during the climb phase is concentrated at around 2Hz, with an amplitude of 2-3 degrees per second. By collecting joystick manipulation data from other pilots during the same flight phase, performing the same processing, and obtaining the corresponding manipulation frequency and amplitude data, and averaging these data, a standard joystick manipulation model for the climb phase can be obtained.

[0070] The standard models for the control wheel, throttle, and pedals during the climb phase, as well as other standard models for other flight phases, are established using the same method.

[0071] Step 3: Update the initial control model of the same pilot in each flight phase and transform it into a differentiated model for that pilot.

[0072] During flight, the current flight phase is first determined based on the status of the aircraft's landing gear, flaps, throttle, and air-to-ground signals, in order to decide which phase of the model needs to be updated.

[0073] Then, the current driver's control commands are recorded in real time. Except during the cruise phase, the recorded data is divided into 10-second segments and subjected to Fourier transform; during the cruise phase, the data is divided into 100-second segments and subjected to Fourier transform.

[0074] Finally, the control frequency and control amplitude data corresponding to the peak values ​​are extracted from the frequency domain data of each flight phase, averaged with the initial control model corresponding to that phase, and used as the new model to await the next update cycle. During each update cycle, when updating the control model for each flight phase, the data is averaged with the data from the control models of the previous cycle to update the control model. Before any updates are made to the control model, it is considered the standard model.

[0075] As more tests are conducted, the actual models for each flight phase will gradually change from the standard model to the pilot's actual (differentiated) model.

[0076] Step 4: Use differentiated models to address issues related to monitoring abnormal pilot behavior and pilot-induced oscillations.

[0077] After multiple updates, the characteristics represented by the model have changed from "the average operating characteristics of a large number of pilots" to "the operating characteristics of current pilots". Based on this, it can more accurately describe the operating characteristics of pilots at each stage, thereby solving the problems brought about by the standard model described above.

[0078] 1) Pilot Abnormal Operation Monitoring. Pilot fatigue, physical incapacitation, or severe emotional fluctuations can all affect their normal operation, causing changes in their control characteristics and deviations from the control frequency and amplitude characteristics of their normal operating state. Since the differentiated model obtained using the method of this invention can accurately describe the pilot's technical characteristics after long-term flight experience, it solves the problem that standard models struggle to distinguish between flight technique deviations and operational deviations caused by pilot abnormal operations. Multiple deviation levels are categorized based on the degree to which the pilot deviates from their normal operating level. Small deviations can be indicated to the pilot via instruments; larger deviations can trigger an alarm, allowing another pilot on the same flight to take over control.

[0079] 2) Suppress pilot-induced oscillations. When designing the flight control system, which involves pilot model-related logic and control parameters, a pilot model fine-tuning interface can be reserved. When the model is not updated (still the standard model), the model fine-tuning amount is 0. When the model undergoes multiple updates and obtains new control frequency and control amplitude data, the difference between these data and the standard model is used as the fine-tuning amount input to the flight control system, thereby correcting the relevant logic and control parameters of the flight control system and optimizing the system's performance.

Claims

1. A method for optimizing a flight control and pilot operation monitoring system based on an online differential model, characterized in that, Includes the following steps: First, we collect statistics on the aircraft currently in flight and divide the flight phases according to the flight characteristics of each aircraft. Then, for each flight phase, an initial control model for each flight phase is established by statistically analyzing the operational data of different pilots; The setup process is as follows: For each flight phase, operational data from different pilots is collected, Fourier transforms are performed on each data to convert it into frequency domain data, and the peak frequency and amplitude of the frequency domain data are extracted and averaged to obtain the control frequency and control amplitude, which serve as the initial control model for the corresponding flight phase. Next, the initial control model of the same pilot in each flight phase is updated and transformed into a differentiated model for that pilot; The transformation process is as follows: For the current pilot, the pilot's control command data in each flight phase is recorded in real time. Except for the cruise phase, the recorded data is divided into 10-second segments and subjected to Fourier transform; during the cruise phase, the data is divided into 100-second segments and subjected to Fourier transform. Then, the control frequency and control amplitude corresponding to the peak values ​​are extracted from the frequency domain data of each flight phase, averaged with the initial control model corresponding to that phase, and used as a new model for the next update cycle. The control data in each update cycle is averaged with the control model data of the previous cycle to obtain the updated model. As the update tests increase, the actual model in each flight phase will gradually change from the standard model to the pilot's differentiated model. Finally, based on the differentiated models of each pilot, the problems of monitoring abnormal pilot operations and pilot-induced oscillations were solved. The solution to abnormal pilot operation monitoring is as follows: The technical characteristics of each pilot are obtained from the differentiated models of each pilot. The deviation level is divided according to the degree to which the pilot deviates from his normal operating level. The pilot is then given a prompt or warning according to the deviation level. The solution to the driver-induced oscillation problem is as follows: The difference between the control frequency and control amplitude of each pilot's differentiated model and the corresponding parameters of the standard model is used as a fine-tuning input to the flight control system, thereby correcting the relevant logic and control parameters of the flight control system.

2. The optimization method for a flight control and pilot operation monitoring system based on an online differentiation model according to claim 1, characterized in that, The flight characteristics of the aircraft include engine status, aircraft flaps, landing gear status, air-to-ground signal status, and flight altitude; the flight phases include takeoff, climb, cruise, descent, approach, landing, and go-around.

3. The optimization method for a flight control and pilot operation monitoring system based on an online differentiation model according to claim 1, characterized in that, The pilot's operational data includes the time history data of inputting stick operation, steering wheel operation, throttle operation, and pedal operation.

Images