Airplane real-time track prediction and route correction method based on Beidou positioning

Abstract

The application discloses a method for real-time flight path prediction and air route correction of an airplane based on Beidou positioning, relates to the technical field of aviation navigation control, collects space sensing signal parameters and electromechanical characteristic parameters, and generates an abnormal jump interception mark; according to the abnormal jump interception mark, an observation covariance is reconstructed, an asymmetric hysteresis constant related to electromechanical hysteresis is introduced, and a limited correction state sequence is obtained; based on the limited correction state sequence, carrier transient airspeed and a structural safety extreme value envelope, an absolute maximum critical curvature is obtained, a continuous feedforward manifold is derived, and when a wideband data link is rejected, topological stationary points and derivative transition information are extracted from the continuous feedforward manifold and encoded into narrowband short message output. The method can distinguish link degradation and body shielding false anomalies, avoid digital rebound and mechanical hysteresis coupling oscillation, and maintain the continuity of flight path correction under the condition of limited bandwidth.

Description

Technical Field

[0001] This invention relates to the field of aviation navigation and control technology, specifically to a method for real-time aircraft trajectory prediction and route correction based on BeiDou positioning. Background Technology

[0002] The main applications of satellite navigation, airborne flight control, and trajectory calculation are manned and unmanned aircraft approach, go-around, ferry operations, and maintaining flight paths and navigating complex airspace (except when there is no suitable navigation equipment or poor communication conditions on some land-based systems), and in missions with significant weather disturbances (or large changes in flight attitude). Currently, airborne navigation systems typically consist of a satellite positioning system, a navigation computer, a display system, and a flight control system. Position, heading, ground speed, and deviation are calculated using navigation and then input into flight guidance or control systems. The navigation computer also performs basic technical processes such as segment information calculation, flight plan management, map display, and deviation alerts.

[0003] Chinese patent CN110488864A proposes a method and system for correcting the heading beacon signal in an aircraft flight control system. Based on the heading beacon signal received during a precision approach, the system continuously collects the heading beacon signal and flight parameters, including ground speed and heading. It uses the current heading beacon signal symbol and heading to determine whether its absolute value exhibits a true trend of change and calculates the true rate of change. Based on the change in the absolute value of the heading beacon signal, it determines a reference trend of change and a reference rate of change. After the correction activation condition is met, the heading beacon signal is corrected according to the current true rate of change, and the corrected signal is output to the flight control system. The principle is to use flight parameters to verify and correct navigation deviation signals to reduce the impact of abnormal signals on approach control.

[0004] While the above technologies can correct the localizer signal in approach scenarios, they still start from the abnormal judgment and correction output of a single navigation deviation signal. Their application is for the judgment of localizer deviation during runway approach.

[0005] When an aircraft is under complex conditions such as satellite positioning link quality issues, large-scale airframe motion, actuator delays, and limited air-to-ground data link bandwidth, relying solely on the relationship between a single type of navigation quantity and flight parameters for anomaly detection cannot distinguish the boundaries between external link degradation and airframe obstruction or changes in maneuvering attitude. Directly inputting correction results into the flight control system without further constraints on actuator response timing and structural safety boundaries can easily introduce new control instabilities when digital updates outpace mechanical responses. Closed-loop control of the aircraft focuses primarily on its own internal closed-loop control, which cannot guarantee consistent expression of flight path information between the air and ground under limited bandwidth conditions. It also cannot transmit or reconstruct identical technology chains, and there is information incompatibility between the airborne control chain and the ground surveillance chain. The core problem is how to stably predict and correct aircraft flight paths under complex disturbances. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method for real-time aircraft trajectory prediction and route correction based on BeiDou positioning. It reconstructs the observation covariance based on anomaly jump interception markers, introduces an asymmetric hysteresis constant related to electromechanical hysteresis, and obtains a constrained correction state sequence. Based on the constrained correction state sequence, the carrier's transient airspeed, and the structural safety extreme value envelope, it obtains the absolute maximum critical curvature, deriving a continuous feedforward manifold. When a broadband data link is denied, it extracts topological stationary points and derivative transition information from the continuous feedforward manifold and encodes them into a narrowband short message for output. This method can distinguish between link degradation and false anomalies caused by airframe obstruction, avoids digital bounce and mechanical hysteresis coupled oscillations, and solves the technical problems described in the background art.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: The method for real-time flight path prediction and route correction of aircraft based on BeiDou positioning includes the following steps: collecting spatial sensing signal parameters and electromechanical characterization parameters, and generating an abnormal jump interception flag based on the deterioration indication of spatial sensing signal parameters and the occlusion interpretation relationship of electromechanical characterization parameters. In response to the abnormal jump interception flag, the observation covariance of the corresponding abnormal segment is reconstructed, and an asymmetric hysteresis constant related to electromechanical hysteresis is introduced when the abnormal jump interception flag is released, so as to obtain the constrained correction state sequence. The absolute maximum critical curvature is determined based on the constrained correction state sequence, the current transient air velocity of the carrier, and the structural safety extreme envelope, and a continuous feedforward manifold constrained by it is generated. When the broadband data link is refused, the topological stationary point and derivative transition information are extracted from the continuous feedforward manifold, encoded into a narrowband short message and output.

[0008] Furthermore, the dual-frequency carrier noise variation in the spatial sensing signal parameters is de-strained and differentially analyzed by adjacent epochs to form a degradation indicator; the electromechanical characterization parameters are written into the circular buffer with a unified timestamp, and an abnormal jump interception flag is generated when the degradation indicator meets the electromagnetic trigger threshold and the electromechanical characterization parameters do not meet the occlusion rejection threshold.

[0009] Furthermore, the timestamp, degradation indication, electromechanical characterization parameters, and abnormal jump interception flag of each navigation epoch are combined into a synchronization judgment segment and written into the time alignment buffer; in response to the abnormal jump interception flag, the observation covariance reconstruction is performed on the entry segment, hold segment, and release segment of the corresponding abnormal segment, and the boundary of the corresponding abnormal segment is written into the time alignment buffer.

[0010] Furthermore, when the abnormal jump interception flag changes from set to cleared, the initial servo pulse width offset is latched and released. The servo return-to-zero progress is determined based on the current servo pulse width offset, and a release factor is generated by combining the asymmetric hysteresis constant. The restricted correction state sequence is updated after scaling the current measurement residual with the release factor, and the updated restricted correction state sequence is used as the input for subsequent curvature constraint calculation.

[0011] Furthermore, when the broadband data link is rejected, a complete topology semantic message is output. The complete topology semantic message includes, in order, a message type field, an event count field, topology stationary parameters arranged in ascending order of parameter variables, a curvature quantization code corresponding to each topology stationary parameter, a derivative transition symbol corresponding to each topology stationary parameter, a time base stamp, a mode code, and a check field.

[0012] Furthermore, a complete topology semantic message is sent when the current sending window can accommodate the complete topology semantic message; when the current sending window is insufficient, a minimum status message is sent. The minimum status message includes a message type field, a timestamp, a mode code, and a check field, and the minimum status message and the complete topology semantic message share the same time synchronization base.

[0013] Furthermore, the topology stationary parameters, curvature quantization code, and derivative transition symbol are serially written into the narrowband transmission message according to a one-to-one correspondence, and a mode code and a check field are appended after the timestamp; the complete topology semantic message is output in a fixed field order, and the field order of adjacent transmission cycles is kept consistent.

[0014] Furthermore, when the computing power occupancy rate of online manifold solving reaches a preset threshold, online manifold solving is stopped, and a static lookup table is invoked instead. The static lookup table outputs a conservative control segment with at least the current vacuum velocity and the upper bound of the allowable curvature as indexes, and uses this conservative control segment as a downgraded replacement object for the continuous feedforward manifold.

[0015] (III) Beneficial Effects This invention provides a method for real-time aircraft trajectory prediction and route correction based on BeiDou positioning, which has the following beneficial effects: Based on the constrained correction state sequence, the current transient airspeed of the carrier, and the structural safety extreme envelope, the absolute maximum critical curvature is obtained, and a continuous feedforward manifold constrained by it is further formed to control the route correction within the physical boundary and avoid sending trajectories that do not meet the airframe carrying conditions into the subsequent control chain.

[0016] By coupling abnormal jump interception markers, observation covariance reconstruction, asymmetric hysteresis constants, and continuous feedforward manifolds, and by implementing a link from perception to deduction and control to prevent the superposition and amplification of link anomalies, algorithm divergence, and mechanical hysteresis, the controllability of the trajectory prediction and route correction process is improved. When broadband data links are rejected, topological stationary points and derivative transition information are extracted from the continuous feedforward manifold and encoded into narrowband short messages, a large number of discrete trajectory points are transformed into reconstructable geometric semantics. This allows the ground side to have a state of continuous trajectory constraints even in the absence of broadband lines, thereby improving the overall capability of the solution in BeiDou positioning aircraft trajectory prediction and route correction scenarios. Attached Figure Description

[0017] Figure 1 A schematic diagram of the overall process of the BeiDou-based real-time flight path prediction and route correction method for aircraft provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the joint anti-counterfeiting logic for abnormal jumps provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of observation covariance reconstruction and asymmetric hysteresis release control provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of feedforward manifold generation and degradation control under dynamic redline constraints provided in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating topological semantic short message generation, narrowband transmission, and ground reconstruction provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figures 1-5This invention provides a method for real-time flight path prediction and route correction of aircraft based on BeiDou positioning, including the following steps: The method sequentially processes spatial sensing signal parameters, electromechanical characterization parameters, constrained correction state sequences, continuous feedforward manifolds, and topological semantic short messages, and uses the abnormal jump interception identifier and time sequence segment output in step one as the input for the subsequent covariance reconstruction stage.

[0020] Step 1: When a quality drop occurs in the spatial positioning link, the electromechanical actions of the aircraft itself are used to reject the false anomaly of physical obstruction, and only the abnormal jump that is sufficient to cause the trajectory simulation drift is sent to the subsequent stages.

[0021] Numerical discrimination on the pure navigation side often only observes noise fluctuations, pseudorange jumps, or epoch lock-offs. When the aircraft performs maneuvers such as steep turns, rolls to break free, or low-altitude obstacle avoidance, a brief obstruction occurs between the dorsal antenna and the wings, vertical tail, and payload bay doors. On the surface, the spatial positioning link appears to experience a sudden drop, but this drop is not caused by distortion from the external electromagnetic environment; rather, it is a projection of the aircraft's geometry. If this type of sudden drop induced by aircraft movements is confused with genuine link degradation, step two can easily treat physically interpretable transients as dirty data and over-inflate the weights.

[0022] Based on this, the first step is to extract the electromagnetic attenuation trend from the spatial perception signal parameters, and then press the attitude, control surface and visual motion traces from the electromechanical characterization parameters into the same time scale. Only when the two do not support each other can the abnormal jump be identified.

[0023] The process unfolds in the order of evidence gathering, rejection, and release. First, the carrier noise levels at the first and second frequencies are acquired, and within the same navigation epoch, the roll rate, pitch rate, and servo pulse width offset are obtained. If the high-precision angular velocity path is downgraded, the equivalent roll rate and pitch rate are derived from the servo pulse width offset using a calibration mapping table, thus keeping the subsequent decision-making formula unchanged. Then, the aforementioned quantities are uniformly written into the same circular buffer, forming continuous segments with timestamp alignment. Based on whether the electromagnetic attenuation gradient exceeds the electromagnetic trigger threshold and whether the airframe obstruction judgment value exceeds the obstruction rejection threshold, an abnormal jump interception flag is output. When the abnormal jump interception flag is set, it means that electromagnetic link degradation exists and airframe obstruction cannot be explained; when the abnormal jump interception flag is not set, it means that the fall has been fully explained by physical action.

[0024] This method first calculates the carrier noise at the first frequency point at each navigation epoch. With the second frequency carrier noise We use the median value from the previous epoch to remove glitches, and then use the difference between adjacent epochs to form a more sensitive attenuation characterization of the drop edge. The reason for not directly using single-frequency threshold comparison is that single-frequency sudden drops would mix receiver automatic gain adjustment, slight contact jitter of antenna feed lines, and actual link blockage, while dual-frequency synchronous changes can preserve the trajectory of the overall link degradation.

[0025] For example, when the aircraft flies level along the sidewall of a valley and enters the local reflection zone, a ground observer sees that the antenna attitude remains almost unchanged. This method records that both carrier noise paths simultaneously contract downwards and their differences remain convergent. This method retains this segment as a candidate for electromagnetic attenuation to be verified, rather than immediately classifying it as aircraft obstruction. Wherein: ; Where: electromagnetic attenuation gradient : indicates the first The normalized attenuation intensity of each navigation epoch, taking values ​​in the range of non-negative real numbers; a larger value indicates a steeper dual-frequency synchronization drop; carrier noise at the first frequency point. : Represents the carrier noise measurement value at the first frequency point, with a range of positive real numbers obtainable by the positioning measurement link, used to characterize the electromagnetic quality at the first frequency point; carrier noise at the second frequency point. : Represents the noise carrier measurement value at the second frequency point. The value range is the positive real number that can be obtained by the positioning measurement link, and it is used to characterize the electromagnetic quality of the second frequency point; Navigation epoch sequence number : Represents the time discrete index, with a value range of positive integers, used to unify the time position of various parameters; electromagnetic attenuation gradient First, we need to gather evidence to determine if a sudden descent occurred, providing a target for subsequent rejection of any attempts to veto the obstruction of the aircraft.

[0026] After obtaining the electromagnetic attenuation gradient Afterwards, this method does not immediately output an anomaly, but instead traces the aircraft's actions back along the same timestamp. The electromechanical representation parameters used here are not abstract flight status words, but rather physical action traces directly related to antenna obstruction: roll rate. The lateral sweep rate and pitch rate of the wing and vertical tail relative to the antenna determine the lateral sweep rate and vertical tail relative to the antenna. The shielding rotation determines the direction of the fuselage dome and nose, and the servo pulse width offset. This reflects the tendency of the aircraft attitude to develop due to rapid deflection of the control surfaces. All three factors are compressed into a single aircraft occlusion judgment value. This is used to answer whether the noise drop at this moment can be explained by the machine itself.

[0027] For example, when an aircraft performs a high-angle roll-off maneuver, an observer can see the ailerons and rudder deflect, and the antenna relative to the satellite direction is swept past by the wing root. Even if the noise load drops briefly at this time, this method will classify this segment as part of the physical blocking chain, not the anomalous jump chain. Wherein: ; Where: Body occlusion judgment value : indicates the first The comprehensive interpretability of the aircraft's actions on antenna obstruction within a navigation epoch, with a value range of [value range missing]. The closer the value is to 1, the more sufficient the body movement is to explain the noise-carrying drop; roll rate : Represents the angular velocity of the aircraft about its longitudinal axis, with values ​​ranging from signed real numbers allowed by the flight control system; pitch angular velocity : Represents the angular velocity of the aircraft about the horizontal axis, with a value range of signed real numbers allowed by the flight control system; servo pulse width offset. : Indicates the offset of the current servo control pulse width relative to the level flight reference pulse width. The value range is the signed real number allowed by the servo control chain and is used to compensate for the situation of angular velocity path reduction or short-term distortion. Rolling normalization upper limit : Represents the normalized scale of roll angular velocity, with values ​​ranging from positive real numbers, used to suppress the occlusion judgment value of a single channel on the aircraft. Exclusive dominance; upper limit of pitch normalization : Represents the normalized scale of pitch angular velocity, with values ​​ranging from positive real numbers; Upper limit of servo normalization. : Represents the normalized scale of the servo pulse width offset, with a range of positive real numbers, used to enable servo surface actions to be included in the same decision formula as angular velocity actions; In scenarios with strong light reflection or drastic changes in near-ground texture, the abrupt change rate of optical flow divergence can also be obtained. This value is only incorporated into the rejection side of the machine occlusion determination value, not into the trigger side of the electromagnetic attenuation gradient. The visual motion traces prove that the machine is performing actions sufficient to explain the occlusion, but cannot directly prove that the mechanical link has deteriorated. Therefore, the inverse constraint on whether the machine occlusion determination value can be explained by the machine gives step one the characteristic of cross-domain cross-spoofing detection rather than single-domain scoring.

[0028] When visual motion is enabled, the optical flow field is first calculated from adjacent image frames, and then the absolute value of the difference in optical flow divergence between two consecutive navigation epochs is used as the visual motion. This visual motion only participates in the weighting of the rejection side of the aircraft occlusion determination value and is not used as the trigger source for the electromagnetic attenuation gradient. The visual motion, roll rate, pitch rate, and servo pulse width offset are written to the circular buffer using the same timestamp; if the direction of change of the visual motion is consistent with the current aircraft attitude establishment direction, the weight of the aircraft occlusion interpretation is increased; otherwise, it is not included in the rejection.

[0029] When the aforementioned electromagnetic attenuation gradient With the body occlusion judgment value This method activates the joint anti-counterfeiting logic valve only after the construction is completed at the same time scale. This logic valve adopts a trigger-then-reject approach, that is, it first checks whether there is an electromagnetic drop that is of concern, and then checks whether the drop has been interpreted by the body's actions.

[0030] If the action judgment is performed first, this method will treat all large maneuvers as potential abnormal premises, causing the valve to be in a state of alert for a long time during normal tactical actions; while performing electromagnetic evidence first and then electromechanical rejection can lock the judgment range to the time slice in which the noise abnormality actually occurs.

[0031] For example, during the glide and turn phase at an airport, ground maintenance personnel observe the aircraft first increasing its bank angle and then leveling off; this method simultaneously records the electromagnetic attenuation gradient. Increase; if the machine body obstructs the judgment value The value also increases, and the joint counterfeit detection logic valve remains locked; if the machine body obstructs the judgment value... Low flat electromagnetic attenuation gradient The joint anti-counterfeiting logic valve immediately outputs an abnormal jump interception flag. This segment is then frozen for later use in refactoring. Specifically: ; Where: Abnormal jump interception identifier : indicates the first Whether a navigation epoch should include the current segment in the subsequent covariance reconstruction chain, with a value range of [value range missing]. A value of 1 indicates that the abnormal jump has occurred, and a value of 0 indicates that the current fall has been rejected by the machine's action; gate function : Represents the step gate function, which takes a value of 1 when the input is not less than 0 and a value of 0 when the input is less than 0. It is used to convert a continuous quantity into a discrete decision; electromagnetic trigger threshold. : Represents the electromagnetic attenuation gradient The release threshold, which takes the value of a positive real number, is used to limit the entry into the next judgment layer only when the sudden drop reaches the edge of attention. Obstruction veto threshold : Indicates the occlusion judgment value of the machine body The veto threshold has a range of values. The positive real numbers within are used to intercept the physical actions sufficient to explain the occlusion within step one; In engineering implementation, the 2×2 cross-decision relationship corresponds to four states: when electromagnetically triggered and not obstructed, an abnormal jump interception flag is output. Locking occurs when electromagnetic triggering is confirmed and the action is blocked; locking occurs when there is no electromagnetic triggering but the machine's movement is obvious; and remaining silent occurs when neither condition is met.

[0032] Preferably, the RF acquisition thread, data parsing thread, and logic valve thread run in independent tasks within the same defined period kernel, with the circular buffer length being eight to sixteen consecutive navigation epochs, so as to preserve the complete edge established by the occlusion without dragging in old data.

[0033] Abnormal jump interception flag The corresponding segments are fed into step two as the trigger for observing covariance expansion and hysteresis damping injection. Then, electromagnetic link degradation and body occlusion projection are separated from the same type of noise drop phenomenon to prevent subsequent steps from absorbing interpretable actions as external anomalies.

[0034] Since the aircraft occlusion determination value incorporates physical motion traces of roll rate, pitch rate, and servo pulse width offset, this method gains direct interpretability for steep-slope turns, roll evasion, and low-altitude obstacle avoidance scenarios. Therefore, step two only expands the observation covariance for the segments that truly require isolation. Step two writes the output anomalous jump interception flag, electromagnetic attenuation gradient, and aircraft occlusion determination value into the time alignment buffer.

[0035] Step 2: Within the intervals before and after the appearance and resolution of the anomalous jump segment, reconstruct the observation covariance matrix using the asymmetric hysteresis law of the electromechanical servo. And correct the absorption rhythm, blocking the secondary excitation of the body by the instantaneous rebound of digital weights.

[0036] Step 1: Output abnormal jump interception flag Subsequently, if the inference chain continues to accept new observations at the original pace, dirty data will be directly injected into the track state along the measurement residuals; if full-weight absorption is immediately restored after the abnormal jump is resolved, the digital inference will stabilize before the aileron servo, elevator servo, and hydraulic circuit, and then give a large correction again before the physical mechanism leaves the dead zone, forming a phase misalignment where the algorithm rushes ahead and the mechanics follow. Step two therefore adopts a continuous action of isolation, slowing down, and release: first, the confidence of the observation corresponding to the abnormal segment is reduced, then an asymmetric hysteresis constant is constructed based on the servo dead zone and oil circuit temperature, and finally, the absorption rate of the digital domain for new observations is released again at a restricted pace. In this scheme, electromechanical hysteresis is not noise, but a structural boundary. Only by writing this boundary directly back into the inference formula can the digital domain be prevented from misjudging short-term recovery as complete recovery.

[0037] First, the abnormal jump interception flag given in step one is read. The leading edge, holding segment, and release segment of the abnormal segment are extracted from the time alignment buffer, and the total oil temperature, fuel consumption ratio, and servo motor return-to-zero progress are obtained. Then, an expansion in the same direction as the electromagnetic attenuation gradient is applied to the observation covariance matrix so that new observations found in the abnormal segment only serve as boundary references and do not directly dominate the state write-back. Next, an asymmetric hysteresis constant is generated based on the servo motor step calibration curve, the ground hydraulic test bench calibration curve, and the total oil temperature. Different time constants are used for entering and exiting the abnormal segment. Finally, the original weights are not restored, but the residual absorption of each epoch is limited to the allowable range of the current servo motor return-to-zero progress.

[0038] For example, when the BeiDou noise level rises again after the aircraft leaves the cloud wall, but the ailerons are still on their way back to center, this method first maintains a low absorption state. After the servo feedback shows that the dead zone has been crossed, the absorption rhythm is gradually relaxed, so as to avoid a second sudden pull when the signal has improved but the aircraft has not yet caught up.

[0039] First, handle the entry of abnormal segments. This method intercepts abnormal transitions upon detection of the abnormal transition flag. After setting the value, the observations are not deleted, nor is the state frozen. Instead, the baseline observation covariance matrix is ​​set. Transform into reconstructing the observation covariance matrix .

[0040] Completely discarding observations would cut off the recovery reference, while a fixed threshold attenuation would fail to reflect the electromagnetic attenuation gradient. With the body occlusion judgment value The joint source.

[0041] Preferably, the covariance reconstruction thread expands the segment header sequentially towards the epoch according to the navigation epoch order. Observations entering anomalous segments still retain a small amount of directional information, thus providing time anchoring for segment observations. ; Where: Reconstructed observation covariance matrix : No. The observation covariance matrix used for state correction in each navigation epoch, with values ​​ranging from a set of positive definite matrices, is used to reduce the dominance of new observations within anomalous segments; the baseline observation covariance matrix... The observation covariance matrix obtained from the normal flight segment calibration has a value range that is a positive definite matrix set; abnormal jump interception identifier. Step 1 inputs the discrete trigger value from Step 2, with a value range of [range missing]. This is used to determine whether the current epoch enters the covariance reconstruction chain; Electromagnetic attenuation gradient The spatial positioning link is in the 1st The attenuation intensity of each navigation epoch, taking values ​​in the range of non-negative real numbers, is used to characterize the steepness of the edge of the abnormal segment; electromagnetic trigger threshold. The trigger threshold from step one is used, with values ​​ranging from positive real numbers to prevent minor fluctuations from being amplified into reconstruction actions; [This refers to] the occlusion judgment value. : No. The ability of the aircraft's movements within a navigation epoch to explain noise drop is within a range of values. Navigation epoch sequence number : Time discrete index, with values ​​ranging from positive integers, used for unified reconstruction of the observation covariance matrix. The temporal position relative to the preceding quantity; Reconstruct the observation covariance matrix This transforms observations within anomalous segments from a directly driven state to a restricted participation state, thereby reducing dirty data absorption, preserving recovery references, and maintaining the temporal continuity of segment boundaries.

[0042] After the aforementioned reconstructed observation covariance matrix is ​​established, during the abnormal segment removal phase in step two, the most dangerous event is not the observation interruption, but the instant the abnormal jump interception flag changes from set to zero. The digital domain tends to view link recovery as immediate full-scale availability, but hydraulic servos remain in the zero-return hysteresis phase due to factors such as oil viscosity, seal rebound, and aerodynamic load on the control surfaces. Therefore, this paper synthesizes the servo dead time measured on the ground test bench, the total oil circuit temperature, and the fuel consumption ratio into an asymmetric hysteresis constant. The fuel consumption ratio does not directly correct attitude but reflects the change in zero-return load on the servo due to forward and backward shifts in the center of gravity.

[0043] Preferably, a ground test bench applies advance and return step movements under rated supply pressure, recording the servo motor current inflection point, the moment the stick end displacement returns to zero, and the return oil temperature to isolate the hysteresis sources entering and exiting the abnormal segment. Total oil circuit temperature. The lower the value, the greater the asymmetric hysteresis constant. The larger the value, the greater the abnormal jump interception flag. When a transition from 1 to 0 occurs, an additional hysteresis is released. Wherein: ; Where: asymmetric hysteresis constant : No. The hysteresis time constant injected into the correction release thread during each navigation epoch, with a value range of positive real numbers, is used to slow down the digital bounce beat after the exception is resolved; the baseline hysteresis constant... The base time constant obtained from servo dead-time calibration takes values ​​in the range of positive real numbers; fuel influence coefficient. Fuel consumption ratio The proportional gain for hysteresis growth, taking values ​​in the range of non-negative real numbers, is used to incorporate the change in rudder surface return-to-zero load caused by center of gravity shift into the asymmetric hysteresis constant. ; Fuel consumption ratio The remaining fuel quantity represents the degree of fuel consumption relative to the takeoff fuel quantity, with a value range of [range missing]. Release additional hysteresis constant Abnormal jump interception indicator Remove the additional time constant for edge detection, with a value range of positive real numbers; gate function. The step gate function takes a value of 1 when the input is not less than 0 and a value of 0 when the input is less than 0. It is used to release the additional hysteresis constant only when triggered at the release edge. ; Temperature gain constant The magnitude constant of the contribution of total oil circuit temperature to hysteresis growth, taking values ​​in the range of positive real numbers, is used to map the low-temperature viscosity effect as an additional hysteresis; reference temperature. The reference temperature used for ground calibration, with values ​​ranging from positive real numbers, is used to provide a comparison benchmark for the total oil circuit temperature; total oil circuit temperature. : Oil circuit temperature used to characterize the temperature of the return oil pipe, with a value range of real numbers within the allowable temperature range of the hydraulic system, used to reflect changes in oil viscosity; temperature scale : The temperature scale of the exponential mapping, with a range of positive real numbers, is used to limit the growth slope of the low-temperature hysteresis; Furthermore, the clearing of abnormal segments is no longer performed through hard switching, but rather by releasing digital updates based on the time it takes for the electromechanical servo to respond. Asymmetric hysteresis constant. After obtaining the result, this method starts the correction and release thread. Preferably, the correction and release thread runs in a fixed epoch using forward Euler discretization and saturation arithmetic, and the parallel implementation paths read the same asymmetric hysteresis constant when discretized using bilinear transformation. .

[0044] In practice, the covariance reconstruction thread first writes the current measurement residual into the residual buffer, and then the correction release thread determines the residual based on the asymmetric hysteresis constant. Generate a monotonically increasing release sequence and multiply it by the servo return-to-zero progress. Only the result is allowed to enter the state modification process.

[0045] For example, after a slope turn, the aileron returns from a large deflection angle to the neutral position, the servo current is high at first and then low, and the hydraulic return oil temperature is still low. During this period, this method only releases the first few measurement residuals. As the servo returns to zero, the subsequent measurement residuals are released. Therefore, the correction amount returns slowly in a single direction and there will be no pull in opposite directions between adjacent epochs.

[0046] Servo homing progress is recorded as When obtaining the servo pulse width offset At this time, this method latches and releases the initial offset at the point of abnormal release. And calculate the servo motor's return-to-zero progress. : ; In the formula, This is the absolute servo pulse width offset at the point where the anomaly is resolved. To prevent extremely small positive numbers with a denominator of zero, the servo motor's homing progress is monitored. The value range is [0,1], where the closer the value is to 1, the closer the servo is to completing the return to zero.

[0047] The correction and release process is based on the servo motor's return-to-zero progress. and asymmetric hysteresis constant Generate release factor Its form is ; In the formula, The interval between adjacent navigation epochs. It is an asymmetric hysteresis constant, and its dimensions are the same as those of the asymmetric hysteresis constant. same; ∈[0,1] represents the servo motor's homing progress. Therefore, the release factor... The value range is [0,1). Measurement residuals for the current epoch. via releasing factor After scaling, the release residuals that participated in the state correction are obtained. Its form is: .

[0048] This method only addresses the release of residuals. By inputting existing state correction equations, the rate at which measurement residuals are incorporated into the state estimation process after anomaly resolution is limited, thus matching the state correction process with the servo motor's return-to-zero progress.

[0049] In the parallel implementation, when the high-precision angular velocity path is missing, the servo pulse width offset used in step one is still retained in the buffer. The correction and release thread uses the servo pulse width offset return process as the alternative input for the servo return to zero progress, so that the downgraded model still executes the same delayed release logic.

[0050] By using the abnormal jump interception flag output in step one Electromagnetic attenuation gradient And the value of body occlusion judgment Connected to the reconstructed observation covariance matrix In the generative formula, step two first reduces the direct absorption of track status by abnormal segment internal data. This is achieved by reducing the total temperature of the fuel system. Fuel consumption ratio The asymmetric hysteresis constant is written together with the release edge. Step two further injects the physical boundaries of hydraulic viscosity, center of gravity shift, and servo dead zone into the digital simulation in reverse. Then, by releasing the correction and absorption cycle according to the servo motor's return-to-zero progress, step two ensures that the recovery speed of the digital domain is always constrained by the physical mechanism's ability to follow the cycle, thus preserving a non-oscillating and continuously inheritable starting point for step three, which generates a continuous feedforward manifold according to the dynamic redline.

[0051] Step 3 receives the reconstructed observation covariance matrix, asymmetric hysteresis constant, and corrected state sequence output from Step 2, and simultaneously acquires the parameters in the vacuum velocity, angle of attack, normal overload, load occupancy rate, and structural safety extreme value envelope.

[0052] Step 3: Transform the corrected state sequence released in Step 2 into a continuous feedforward manifold constrained by the structural safety extreme value envelope, thereby completing the route correction while the aircraft is still in a recoverable state, without amplifying the digital correction into an overload maneuver.

[0053] Although step two already uses an asymmetric hysteresis constant While the digital rebound speed is reduced, the corrected state sequence output in this step is still only a time series showing how much measurement residual can be absorbed, and does not inherently equate to the trajectory the aircraft can follow. If conventional proportional-integral-derivative closed-loop commands are reintegrated on this basis, the digital domain will follow the instantaneous deviation, and the aircraft is prone to a chain reaction of problems such as excessive command curvature, excessively rapid roll build-up, and pitch coupling exceeding limits under conditions of high airspeed, high overload, or center of gravity shift.

[0054] Therefore, the current approach employs a single-chain process: first, a suspended closed loop is established; then, the redline is calculated; and finally, the manifold is generated. The suspended closed loop refers to temporarily halting the real-time reverse calculation of the rudder input based on the deviation when entering the trajectory correction phase. Instead, this method calculates the absolute maximum critical curvature based on the current airspeed, normal overload, roll allowable angle, and structural safety envelope. This curvature is then used as the rigid upper bound of the continuous feedforward manifold. The starting point for this approach is not smoothness or aesthetics, but rather to strictly confine digital correction within the boundaries that the airframe structure and aerodynamic load can follow.

[0055] This method first uses the corrected state sequence at the end of step two as the starting state, and then reconstructs the observation covariance matrix. The directional information that has not yet been fully released is transcribed into the target displacement trend.

[0056] Subsequently, the proportional-integral-derivative closed-loop output is frozen at the previous effective rudder position to prevent new high-frequency feedback from being inserted during the generation of the feedforward manifold; vacuum velocity, normal overload, roll allowable angle, fuel consumption ratio, and load occupancy rate are acquired simultaneously to form the structural safety extreme value envelope at the current moment; and the absolute maximum critical curvature is obtained within this envelope, and a third-order tangential feedforward manifold is constructed with the initial tangent, target tangent, and upper curvature bound as common constraints.

[0057] For example, when an aircraft needs to slowly backtrack from the side deviating from its flight path after crossing the edge of a cumulonimbus cloud, a ground observer sees the wing stabilize first, and then the nose returns to its flight path along a continuous arc. This is because this method does not directly pursue the shortest distance during manifold generation, but instead first completes the physical redline clamping. If the computational load of the online manifold solution rises to a set threshold, the manifold differential solution is cut off, and a static lookup table is used instead to keep the curvature boundary first.

[0058] First, the abstract concept of avoiding overload is concretized into a single, callable upper bound of curvature. Vacuum speed, roll allowable angle, normal overload allowable value, and fuel consumption ratio are obtained at the navigation epoch. Then, based on the structural safety extreme value envelope, the corresponding center of gravity contraction coefficient and wing root load contraction coefficient are retrieved. Here, the fuel consumption ratio no longer affects hysteresis but is used to correct for the reduction in the allowable envelope caused by forward and backward shifts of the center of gravity.

[0059] Preferably, the structural safety extreme value envelope consists of a four-dimensional index array of aircraft model number, center of gravity segment, airspeed segment, and roll segment, with each storage unit corresponding to a set of calibrated contraction coefficients. Therefore, step three does not require temporary calculation of complex elastic wing equations and avoids mixing load boundaries from different aircraft models, wherein: ; Where: absolute maximum critical curvature : No. Each navigation epoch allows the writing of the upper bound of the curvature of the feedforward manifold, taking values ​​in the range of positive real numbers, used to limit the minimum turning radius of the aircraft's steering geometry; gravitational acceleration. : Constant term, taking values ​​of positive real numbers, used to convert the roll allowance angle into equivalent curvature capability; Roll allowance angle The upper bound of the roll angle allowed by the current structural safety extreme value envelope, the value range of which is a non-negative real number and less than a right angle; vacuum speed The vacuum speed, obtained from the airspeed correction process, is a range of positive real numbers and is used to reflect the squared amplification effect of speed on the turning radius; the center of gravity contraction coefficient. : A dimensionless coefficient obtained by looking up the fuel consumption ratio and the current center of gravity segment, with values ​​ranging from 0 to 1; wing loading contraction coefficient : A dimensionless coefficient determined by load occupancy rate data and structural safety envelope, with values ​​ranging from 0 to 1. Overload coupling coefficient : The proportionality coefficient of the normal overload allowable value to the curvature reduction, taking values ​​in the range of non-negative real numbers, used to incorporate the vertical load margin into the upper limit of the lateral curvature; normal overload allowable value The upper limit of the normal overload allowed by the current structural safety extreme value envelope, with a range of positive real numbers, is used to characterize the vertical load margin that the structure can still withstand; navigation epoch sequence number. : Time discrete index, with values ​​ranging from positive integers, used to unify the critical curvature of absolute maximum. The time position relative to other input quantities; Among them, the absolute maximum critical curvature The airspeed, roll, and structural loads are directly compressed into an executable geometric boundary, so that no subsequent manifold can exceed this upper bound.

[0060] When the aforementioned absolute maximum critical curvature Once determined, the manifold generation thread no longer uses polylines to connect the target points, nor does it employ low-order splines that only guarantee positional continuity. Instead, it uses the initial position, initial tangent, target position, target tangent, and upper curvature bound to jointly determine the third-order tangent feedforward manifold. .

[0061] Specifically, the starting position is recorded as The finish line location is recorded as The initial unit tangent vector is denoted as The target unit tangent vector is denoted as The airborne trajectory is described using cubic Bézier curves, where... The four control vertices of the cubic Bézier curve are initialized as follows: ; in, and These are the initial tangential length parameters and the target tangential length parameters, respectively, with the initial value being the Euclidean distance between the starting and ending points. A scale factor is defined to determine curvature constraints. The total length of the control polyline formed by the current four control vertices, i.e.: ; in, Describing the Euclidean norm, Used to match the rate of change of angle over the normalized parameter domain with the actual track length scale.

[0062] This method is applied at preset discrete parameter points. The generated curve is checked to see if it satisfies the curvature constraint. If any discrete parameter point exists... Make: ; Then according to the reduction factor synchronous decrease and And regenerate the control vertices until the curvature constraint is satisfied for the first time. It is a real number between 0 and 1.

[0063] The polygonal trajectory will experience abrupt curvature changes at the turning points. Even if the low-order splines are continuously positioned, the servo motor will still exhibit sudden jumps in angular acceleration, thus disrupting the hysteresis yielding relationship established in step two. Preferably, the numerical solver serves only as a tool for realizing continuous manifolds. The invention lies in defining the absolute maximum critical curvature. As a priori constraint controlling vertex generation and verification, this constraint ensures that any solver used outputs a feedforward manifold within the structural safety extremum envelope. For example, a B-spline solver or a piecewise cubic Hermite interpolator can be implemented under this prior constraint.

[0064] For example, when making a turnback at the valley exit, the aircraft's trajectory does not turn directly toward the centerline of the route. Instead, it first extends along the existing tangential direction for a while, then gradually increases the turning curvature, and then gradually decreases the curvature after approaching the centerline of the route. What the ground sees is a continuous arc without any turning points.

[0065] ; In the formula, For the feedforward trajectory curve, its domain is the parameter interval [0,1], and its range is the position vector space in the horizontal plane; Indicates parameters Corresponding waypoints; For the first One Bézier control point; These are cubic Bernstein basis functions; Here are the manifold parameters, and their values ​​range from 1 to 2. The feedforward trajectory curve satisfies the condition of the maximum permissible curvature. Limited curvature constraints: ; In the formula, This is the feedforward trajectory curve. For parameters The corresponding horizontal track point.

[0066] Parameter variables Manifold parameter, taking values ​​in a closed interval. Used to uniformly represent the progression position of a manifold from its starting point to its ending point; controls the vertices. The control vertices of the third-order manifold, whose values ​​are in the range of three-dimensional coordinate vectors, are used to determine the feedforward manifold. Shape and tangency; index variables : Controls the numbering of vertices and basis functions, with values ​​ranging from the set of integers. basis functions Third-order basis functions, whose range is the set of non-negative functions and whose sum is 1, are used to smoothly stitch control vertices into a continuous curve; tangential angle. Feedforward manifold The trajectory direction angle in the horizontal plane is used to characterize the change in the lateral trajectory direction as the manifold advances.

[0067] scale factor : Positive real numbers obtained from the estimation of the arc lengths at the start and end points, used to convert the tangential rate of change in the parameter domain into physical curvature constraints; absolute maximum critical curvature The obtained upper bound of curvature, taking values ​​in the range of positive real numbers, is used to directly constrain the feedforward manifold. The rate of change of direction; This ensures the simultaneous continuity of position, tangential, and curvature boundaries, thus transforming digital correction into a geometric object that the solid control surface can follow. In some cases, such as drastic weather changes or multiple concurrent links, the computational cost of differential solving in the manifold generation process can increase. To prevent the solution process itself from becoming a weak link in step three, we designed computational cost monitoring and degradation switching processing.

[0068] When the computing power utilization rate continuously reaches a predetermined threshold, the propulsion speed of the current parameter variable is frozen, online differentiation is cut off, and the three-dimensional static lookup table of vacuum speed-absolute maximum critical curvature-target deflection angle is read. The corresponding control segments are directly selected and spliced ​​into a conservative feedforward trajectory. Preferably, the lookup table units are organized hierarchically according to aircraft type, configuration, airspeed segment, and center of gravity segment. Each unit pre-stores the endpoints, tangents, and durations of control segments with curvature not exceeding the limit. Parallel implementation paths are transitioned to adjacent units using bilinear interpolation or trilinear interpolation.

[0069] For example, when online manifold solver resources encounter peak occupancy, ground maintenance personnel can see on the playback interface that the nose still cuts back along a conservative arc, only the arc is wider and the cutback is gentler. This indicates that the degradation path does not overturn the aforementioned physical red line, but simply replaces online manifold solver with lookup table splicing. During evaluation, the number of curvature overruns within the same correction segment, the number of tangential jumps at segment connections, and the trajectory continuity during periods of occupancy exceeding limits can be recorded, corresponding to the three characteristics of red line preservation, segment splicing, and smooth degradation.

[0070] First, calculate the absolute maximum critical curvature. Then, a feedforward manifold is generated using this upper bound of curvature. Step three rewrites the original correction intent from the digital domain into a continuous trajectory constrained by the structural safety extreme value envelope, thus preventing the closed-loop command from being amplified into an overload maneuver under high airspeed conditions. By freezing the proportional-integral-derivative closed-loop output and using the correction state sequence from step two only as the initial trend rather than a direct rudder input, step three maintains the single-chain connection between the preceding and following steps, and does not reintroduce high-frequency pursuit.

[0071] Step four receives the feedforward manifold generated in step three and performs topological semantic extraction, short message encapsulation, narrowband transmission, and ground-side reconstruction.

[0072] Step 4: Process the feedforward manifold generated in Step 3. It is converted into a topological semantic short message that can be transmitted in an extremely narrowband channel and synchronously reconstructed by the ground side, while retaining the minimum closed-loop indication capability when the channel deteriorates further.

[0073] Conventional discrete coordinate point matrix messages require the continuous transmission of a large number of position and attitude points. After a short-term link loss, the ground reconstruction step can only obtain an incomplete polyline, which does not retain the critical curvature obtained in step three. The applied upper bound of curvature cannot indicate where the curvature of the track increases, where it decreases, or where it remains tangential.

[0074] Step four therefore does not compress the original sampling points, but only from the feedforward manifold The topological stationary points and derivative transition symbols of the geometric shape are extracted, quantized with a fixed dictionary, and then encapsulated to form a short message. The control value of the continuous manifold is not distributed across the entire sampling point. The reconstruction result is determined by a few curvature extrema, tangential flip direction, and small time points.

[0075] Step four: Traverse the parameter range of the feedforward manifold, extract the boundary positions of the curvature rising segment, curvature falling segment, and tangential preservation segment; write the boundary positions, boundary types, and the curvature code obtained after normalizing the absolute maximum critical curvature into the message dictionary; concatenate the message dictionary with the time base stamp, pattern enumeration code, and check field to form a fixed-length payload and send it to the ground side. After receiving the fixed-length payload, the ground side does not perform point interpolation, but instead reverse-engineers the piecewise constraints of the feedforward manifold using the same dictionary, and then generates a continuous trajectory.

[0076] For example, when an aircraft flies out of the interference zone and the broadband link has not yet been restored, the maintenance station sees not hops, but a continuous route band composed of a few arcs; this is because the aircraft sends control semantics rather than the original point sequence.

[0077] First, the geometric nodes with the highest transmission value are selected. This method reads the feedforward manifold in each communication cycle. The parameterized expression first performs a coarse scan with a fixed step size, and then performs a binary approximation in the local interval where the curvature derivative changes sign, to avoid misjudging small numerical fluctuations in flat segments as topological events.

[0078] Preferably, the coarse sweep step size is taken from the control vertex spacing when generating the manifold in step three, and the binary approximation ensures that the change in the topological stationary point parameters between two adjacent steps does not exceed the parameter tolerance. Termination time; parameter tolerance This method is set during the initialization of the manifold parsing thread to uniformly limit the accuracy of topology stationary point positioning.

[0079] For example, in a traverse path where curvature first increases and then decreases, this method leaves only a topological stationary point near the curvature peak, without retaining the dense sampling points on both sides of the peak. Wherein: ; Where: topological stationary point sequence Feedforward manifold The set of parameter positions used for message encapsulation, with values ​​taking the values ​​of a closed interval. A finite set of points on the manifold is used to determine the geometrically critical locations that should be retained in short messages; manifold curvature function Feedforward manifold With parameter variables The varying curvature function, taking the values ​​of a set of non-negative real functions, is used to characterize the turning intensity of the manifold at different locations; parameter variables. Feedforward manifold The normalized arc parameter takes values ​​on a closed interval. Used to unify the labeling of manifold positions on the aircraft side and the ground side; topological stagnation point parameters : No. The location of each topological stationary point takes values ​​within a closed interval. This is used for subsequent quantization and packaging; Index variables : Topology station number, a positive integer, used to distinguish multiple topology station parameters ; ; In the formula: the derivative transition symbol : No. The sign of the second derivative corresponding to each topological stationary point, taking the value of Where -1 represents the peak of curvature, 1 represents the valley of curvature, and 0 represents a flat stagnation point, used to distinguish the local arc segment morphology that should be established in the ground reconstruction step; Second derivative operator : Curvature function of manifold Relative to parameter variables Operators that perform second-order differentiation are used to determine the local polarity of topological stationary points; sign function. : Symbol discrimination function, output 1 when the input is greater than 0, output -1 when the input is less than 0, and output 0 when the input is equal to 0. It is used to compress local polarity into fixed codewords; Furthermore, topological stationary sequence Tell us where the key points of the ground reconstruction steps are, and the derivative transition sign. The ground reconstruction steps are instructed on how to flip the arcs near the location, and together they replace the lengthy dot matrix.

[0080] After obtaining the topological stationary point sequence and derivative transition sign Subsequently, the semantic encapsulation thread does not directly send floating-point numbers, but instead normalizes the local curvature of each topological stationary point to the absolute maximum critical curvature. Then, it is mapped to hexadecimal codewords. What the ground reconstruction step really needs is how much curvature budget is used relative to the upper limit of the permission, not the aircraft-side floating-point mantissa.

[0081] Preferably, the fixed-length payload consists of an event count field, a time base stamp field, a curvature code field, a derivative transition field, a pattern enumeration field, and a cycle check field, and the order of the fields remains consistent in the narrowband transmission step and the ground reconstruction step.

[0082] For example, when only two topology stations are extracted in a certain communication cycle, the narrowband transmission step will first write the event count field, and then write the two topology station parameters in sequence. Quantization code, corresponding curvature code, and derivative transition symbol Finally, complete the remaining fields to keep the payload length unchanged. Among them: ; Where: curvature quantization code : No. The hexadecimal curvature code at each topological stationary point takes the value of a set of integers. Used to represent the percentage of local curvature occupancy in half a byte; floor function. : Floor down operation, used to compress the normalized curvature into discrete codewords, ensuring that the aircraft side and the ground side use consistent integer representation; Constant 15: Upper limit of half-byte quantization, with a fixed positive integer value, used to map the normalized curvature to hexadecimal units; manifold curvature function : No. The local curvature values ​​at each topological stationary point are non-negative real numbers, used to provide the bending strength corresponding to that stationary point; the absolute maximum critical curvature. The upper bound of curvature given in step three is a positive real number, used to provide a unified normalization benchmark for local curvature; Furthermore, this prevents fixed-length payloads from carrying a large number of unreproducible floating-point mantissas, instead carrying discrete curvature semantics directly related to the physical redline, thereby maintaining the same feedforward manifold on both the aircraft and ground sides. A consistent understanding.

[0083] Once the fixed-length payload is completed, the narrowband transmission step sends short messages according to the currently available channel type.

[0084] Preferably, when the BeiDou short message link is available, the BeiDou short message channel is used for transmission; when the broadband link is unavailable, the service message is transmitted through the narrowband digital link; each link only changes the outer physical bearer and does not change the service meaning of the topology semantic field.

[0085] Regardless of the narrowband link used, the service field semantics of the complete message and the minimum state message remain consistent, while the outer link encapsulation can be adjusted according to the specific channel. The ground reconstruction steps are also always performed in reverse order using the same dictionary; therefore, what changes is the transmission medium, not the topology semantics.

[0086] The topological stationary sequence is rewritten as an ordered sequence arranged in ascending order of the parameter variables. In the formula, This represents the number of topology resident points extracted during the current communication cycle. The ground reconstruction step first extracts the parameters of each topology resident point in field order. Corresponding curvature quantization code and derivative transition sign And press: ; Restore the curvature values ​​at each topological stationary point. Subsequently, the ground reconstruction step constructs the derivative transition symbols between adjacent topological stationary points. Piecewise monotonic curvature functions with consistent polarity And press: ;as well as: ; Restore lateral track. Complete messages and minimum state messages are two independent types of service messages: complete messages carry topology stationary parameters. Curvature quantization code Derivative transition symbol and timestamp; the minimum status message only carries the timestamp. With downgrade mode code It is used to keep the state of both the air and ground sides synchronized when the current sending window is insufficient.

[0087] Furthermore, when a single frame transmission window is insufficient to accommodate a fixed-length payload, the truncated thread discards all topology stationary content, retaining only the time reference stamp and the degradation mode code, ensuring that the ground reconstruction step knows at least the previous feedforward manifold. Should we continue with the current mode or have we switched to conservative mode?

[0088] For example, during a narrowband link congestion, several complete payload frames appear sequentially in the ground seat interface, and then only the timestamp and degradation mode code are received. Here, no random time jump is assigned to each frame; the previous restricted arc is maintained, and the frame is displayed in a degraded state until the next complete payload frame arrives. Wherein: ; In the formula: net load sent : No. Each transmission cycle consists of the bit sequence actually emitted by the narrowband transmission step, and its value is a set of binary strings of finite length, used to unify the two paths of complete transmission and truncated transmission. Complete Clean Lotus : No. A complete topological semantic short message for each transmission cycle, consisting of a set of fixed-length binary strings, is used to carry the topological stationary point sequence. Derivative transition symbol Curvature quantization code and related fields; net load length : No. Complete payload for each transmission cycle The length of is a positive integer, used for comparison with the current channel's maximum capacity. Channel capacity length : The maximum length that the current channel is allowed to transmit within a transmission window, a positive integer value, used to determine whether to trigger truncation and degradation; Timestamp : No. The time identifier field, retained in each transmission cycle, is a fixed-length binary string used by the ground reconstruction step to identify the previous feedforward manifold. Timeliness; Degradation mode code : No. A single-byte status enumeration field sent in each transmission cycle, with values ​​from a finite enumeration set, is used to indicate whether the ground reconstruction step should maintain the current manifold, switch to a conservative manifold, or pause updates. splicing operator The operation of concatenating binary fields in a predetermined order is used to form the smallest executable message in the truncated transmission path; transmission period number. : The discrete periodic sequence number of the short message transmission, which takes the value of a positive integer, is used to keep the aircraft and the ground synchronized at the same communication time; In the narrowband transmission step, preferably two service messages are transmitted: one is a complete topology semantic message, which carries topology stationary point parameters, curvature quantization code, derivative transition symbol, and time base stamp; the other is a minimum state message, which only carries the time base stamp and degradation mode code. If the current transmission window can accommodate all topology semantic messages, the complete topology semantic message is transmitted; otherwise, the complete topology semantic message is transmitted.

[0089] Furthermore, the aircraft side maintains a single terminology system in all three states: normal channel, restricted channel, and extremely restricted channel. The complete payload prioritizes the recovery of all manifold semantics, while the truncated payload at least retains the closed-loop state indication. The ground reconstruction steps then execute the same set of reverse reconstruction and conservative preservation logic accordingly.

[0090] First, extract the topological stationary point sequence. and derivative transition sign Step four: transform the feedforward manifold The few events that truly determine the geometry are separated from a large number of intermediate sampling points, allowing the ultra-narrowband channel to carry control semantics rather than data noise. This is achieved by using an absolutely maximum critical curvature. Generate curvature quantization code for normalized reference Step four involves transmitting the physical redline established in step three to the communication side to prevent the ground reconstruction step from recovering a false trajectory that exceeds the upper limit of curvature.

[0091] Each step of the method described in this application can be executed by the computational process in the flight mission according to a predetermined sequence, or it can be implemented by software instructions; regardless of how the specific execution platform changes, the order of the above-mentioned anomaly detection, covariance reconstruction, hysteresis release, curvature constraint and topological semantic encapsulation remains unchanged.

[0092] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for real-time aircraft trajectory prediction and route correction based on BeiDou positioning, characterized in that, The steps include the following: The system collects spatial sensing signal parameters and electromechanical characterization parameters. Based on the degradation indication of the spatial sensing signal parameters and the occlusion interpretation relationship of the electromechanical characterization parameters, an abnormal transition interception flag is generated. The dual-frequency carrier noise variation in the spatial sensing signal parameters is de-squeezed in the same window and differentially analyzed between adjacent epochs to form a degradation indication. The electromechanical characterization parameters are written into a circular buffer with a uniform timestamp. An abnormal transition interception flag is generated when the degradation indication meets the electromagnetic trigger threshold and the electromechanical characterization parameters do not meet the occlusion rejection threshold. In response to the anomalous jump interception flag, the observation covariance of the corresponding anomalous segment is reconstructed, and an asymmetric hysteresis constant related to electromechanical hysteresis is introduced when the anomalous jump interception flag is released to obtain a constrained correction state sequence; wherein, a release factor is generated by combining the asymmetric hysteresis constant; wherein, the constrained correction state sequence is updated after scaling the current measurement residual with the release factor, and the updated constrained correction state sequence is used as the input for subsequent curvature constraint calculation; The absolute maximum critical curvature is determined based on the constrained correction state sequence, the current transient airspeed of the carrier, and the structural safety extreme envelope, and a continuous feedforward manifold constrained by it is generated. When the broadband data link is rejected, the topological stationary point and derivative transition information are extracted from the continuous feedforward manifold, encoded into a narrowband short message and output. The structural safety extreme envelope consists of a four-dimensional index array of aircraft number, center of gravity segment, airspeed segment, and roll segment, and the storage unit corresponds to a set of calibrated contraction coefficients.

2. The method for real-time aircraft trajectory prediction and route correction according to claim 1, characterized in that: The timestamp, degradation indication, electromechanical characterization parameters, and abnormal jump interception flag of each navigation epoch are combined into a synchronization decision segment and written into the time alignment buffer. In response to the abnormal jump interception flag, the observation covariance reconstruction is performed on the entry segment, hold segment, and release segment of the corresponding abnormal segment, and the boundary of the corresponding abnormal segment is written into the time alignment buffer.

3. The method for real-time aircraft trajectory prediction and route correction according to claim 2, characterized in that: When the abnormal transition interception flag changes from set to cleared, the initial servo pulse width offset is latched and released, and the servo return-to-zero progress is determined based on the current servo pulse width offset.

4. The method for real-time aircraft trajectory prediction and route correction according to claim 1, characterized in that: When the broadband data link is refused, a complete topology semantic message is output. The complete topology semantic message includes, in order, a message type field, an event count field, topology stationary parameters arranged in ascending order of parameter variables, curvature quantization codes corresponding to the topology stationary parameters, derivative transition symbols corresponding to the topology stationary parameters, a time base stamp, a mode code, and a check field.

5. The method for real-time aircraft trajectory prediction and route correction according to claim 4, characterized in that: When the current sending window can accommodate a complete topology semantic message, a complete topology semantic message is sent; when the current sending window is insufficient, a minimum status message is sent. The minimum status message includes a message type field, a timestamp, a mode code, and a check field, and the minimum status message and the complete topology semantic message share the same time synchronization base.

6. The method for real-time aircraft trajectory prediction and route correction according to claim 4, characterized in that: The topology stationary point parameters, curvature quantization code, and derivative transition symbol are serially written into the narrowband transmission message according to a one-to-one correspondence, and the mode code and check field are appended after the timestamp; the complete topology semantic message is output in a fixed field order, and the field order of adjacent transmission cycles is kept consistent.

7. The method for real-time aircraft trajectory prediction and route correction according to claim 1, characterized in that: When the computing power occupancy rate of online manifold solving reaches a preset threshold, online manifold solving is stopped, and a static lookup table is invoked instead. The static lookup table outputs a conservative control segment with at least the current vacuum velocity and the upper bound of the allowable curvature as indexes, and uses this conservative control segment as a degradation replacement object for the continuous feedforward manifold.

Images