EP1631907A1 - Method and device for processing information output by redundant primary flight equipment - Google Patents

Abstract

The invention relates to redundant architectures of processing lines placed between primary flight equipment that are doubled or tripled for safety purposes and one or more flight control systems. The invention also relates, in particular, to the addition, at the beginning of the processing lines, of anti-noise filters serving to prevent untimely disconnections of an automatic control of a flight control system that are not justified by a suspected failure. These anti-noise filters are digital filters that operate at a sampling rate of the output signals of the primary flight equipment and not at the lower sampling rate of the flight control systems.

Description

 METHOD AND DEVICE FOR PROCESSING INFORMATION FROM REDUNDANT PRIMARY FLIGHT EQUIPMENT

The present invention relates to the processing of signals supplied to a flight control system by primary flight equipment doubled or tripled for safety. It relates more particularly to the processing operations intended to avoid untimely disconnections of an automatic control system of the flight control not justified by a real failure of an equipment delivering or processing the signals used by the automatic control.

A certain number of flight information items, including the attitude of the aircraft, the module and the orientation of the aircraft speed vector, and the altitude of the aircraft are essential for piloting an aircraft. This information is provided by sensors belonging to on-board equipment called "primary flight equipment". Among the primary flight equipment, there are the static and dynamic pressure sensors and their associated computer ("Air Data System" in English) making it possible to know the air speed of the aircraft and the inertial unit or units delivering the accelerations and angular speeds of the aircraft which can be combined within the same equipment designated by the acronym ADIRS (acronym of the Anglo-Saxon expression "Air Data Inertial Reference System").

The essential information for piloting is exploited in a raw form or preprocessed by one or more flight control systems bringing together automatisms which facilitate piloting by ensuring either attitude stabilizations or follow-up of attitude, heading, slope instructions , route, altitude, speed, etc. The best known of these automations is the autopilot and / or flight director. Primary flight equipment, such as flight control systems, must have an extremely low failure rate which is often achieved only by redundancy, the same equipment being doubled or tripled as well as a flight control system, each copy of the same flight control system being linked to the various copies of the primary flight equipment by processing chains whose function is to choose the most credible version, from the different versions available of the same information, and to detect any discrepancy between the different versions available of the same information being able to make suspect an unreported failure of one of the copies of the primary flight equipment at the origin of the information.

Usually, the primary flight equipment as well as the processing lines are in duplicate. Each processing chain operates independently of the others and includes a voting device whose function is to choose at all times the copy of the flight equipment which provides the information taken into account by the flight system or systems. This voting device makes its choice by applying a proximity criterion with respect to a value corresponding to the median value of the available versions of the same item of information offset by an arbitrary threshold. When two versions of the same information have very similar values, which is the case when two copies of the same equipment function correctly, the voting device tends to switch randomly from one copy to another, which introduces a spurious switching noise which can disturb the operation of flight control systems. It is known to limit this random switching by introducing a certain threshold in the operation of the voting device.

The flight information is delivered by the primary flight equipment at rates, for example of the order of 20 milliseconds, compatible with the speed of variation of the flight parameters that they measure, this in an asynchronous manner, two copies of the same equipment operating in completely independent ways with separate clocks. Flight control systems use the flight information delivered by the primary flight equipment at a lower rate adapted to their needs, for example of the order of 50 milliseconds, based on the reaction speed of the aircraft to their instructions. Information from primary flight equipment is preprocessed in flight control systems at the lowest rate, to minimize their computational load. Asynchronisms between the different copies of the same equipment connected to their inputs and between their inputs and their output are taken into account at the start of the chain by the use of buffer memories.

In the case of double processing chains, monitoring the proper functioning of the processing chains and the equipment placed upstream is done by detecting discrepancies between the versions of the same information available at the output of the two processing chains. When this discrepancy becomes too great, one suspects an unreported malfunction of one of the elements of the processing chains or of one of the equipment placed upstream of the processing chains and we prefer to stop the automations depending on the processing chains. treatment. This is obtained using a subtractor circuit placed at the outputs of the chains and followed by a threshold comparator. The threshold comparator generates, in the event of its threshold being exceeded by the difference existing between the versions of the same information available at the output of the two processing chains, an alarm leading to the stopping or disconnection of the automations in operation flight control systems using the channel exit information.

In an architecture with a processing chain doubled in order to reinforce security, hardware and software dissymmetries, the lack of synchronization and the change of timing between inputs and outputs cause noise disturbing the signals of primary flight equipment, such as noise due to parasitic vibrations of the aircraft cell in the case of inertial units, propagate differently along the two chains and can cause, by artifact, sufficient discrepancies to trigger the threshold detector placed at the outputs of the two signaling chains possibility of malfunction. This is even more true in the case of processing chains liable to significant drifts such as those comprising integrating circuits used to extract speed information from acceleration information, position information from speed information, or for develop precision terms intended for steady state control commands.

The object of the present invention is to reduce the frequency of untimely disconnections of automatic systems of a flight control system due to artefacts from processing chains of primary flight information doubled or tripled to reinforce safety. It relates to a method for processing information originating from primary flight equipment mounted on board an aircraft, in a form sampled at a first rate, with a view to being delivered to a flight control system of the aircraft, at a second rate lower than the first rate, this method being remarkable in that it consists in subjecting the information samples to digital noise reduction filtering carried out at the first sampling rate.

Advantageously, the digital noise reduction filtering is an anti-aliasing filtering blocking the undesirable components having frequencies lower than half of the first sampling rate and higher than half of the second sampling rate

Advantageously, digital noise reduction filtering is first order low pass filtering.

Advantageously, the digital noise filtering is a second order low-pass filtering.

Advantageously, the digital noise reduction filtering is a low-pass or notch filtering of the Butterworth type.

Advantageously, when the processed information originating from primary flight equipment is affected by noise exhibiting energy peaks, the noise reduction filtering is a cut-band filtering corresponding to the energy peaks of the noise.

Advantageously, the digital filtering implements a transfer function depending on the flight configuration of the aircraft.

Advantageously, the noise reduction filter is a sliding average filter operating on several samples.

The invention also relates to a device for implementing the above method.

Other advantages and characteristics of the invention will emerge from the description below of an embodiment of the invention given by way of example. This description will be made with reference to the drawing in which:

- a figure 1 represents a redundant architecture used, in the prior art, for a lateral acceleration information processing chain provided by a inertial unit in order to make it available to a yaw stabilizing automation,

FIGS. 2a to 2d are diagrams of curves illustrating the evolution of the differences between the operating drifts of the two processing chains of the redundant architecture of FIG. 1 in the presence of highly noisy sensor signals and the untimely disconnections resulting therefrom for the yaw stabilizing automation,

FIG. 3 represents a redundant architecture according to the invention for a chain for processing lateral acceleration information supplied by an inertial unit with a view to making it available to a yaw stabilizing automation, and

- Figures 4a to 4d are diagrams of curves illustrating the controlled operating drifts of the two processing chains of the redundant architecture of Figure 3 and the very infrequent blockages resulting therefrom for the yaw stabilizing automation.

FIG. 1 shows a known type of redundant architecture used for the generation of integrated lateral acceleration information YD intended for a yaw stabilizing automation forming part of a flight control system and having the function of dampen the yaw oscillations of the aircraft and cancel the skid angle of the aircraft. This redundant architecture takes advantage of lateral acceleration information γl delivered in parallel and independently by two copies 10, 1 1 of INS inertial unit to also independently generate two versions of YD anti-skid control information all two intended for yaw stabilizing automation. Constant monitoring of the difference between the two supplied versions YDa and YDb of the control information is used as a test of the correct functioning of the elements of the redundant architecture of the yaw stabilizing automation. This type of redundant architecture contains two functionally identical parallel chains FGMa and FGMb, installed on separate hardware modules. The two copies 10, 11 of inertial unit INS deliver, in digital form, and at a repetition rate of 50 Hz (periodicity of 20 ms), two versions γl1 and γl2 of the lateral acceleration information. Their data flows are neither completely identical nor synchronized because they are subjected to different vibrational atmospheres due to their installations in different places of the aircraft cell and operate independently of each other with independent clocks .

The two parallel processing chains FGMa, FGMb use the two versions γl1 and γl2 of the lateral acceleration information delivered asynchronously and at a repetition rate of 50 Hz (periodicity of 20 ms) by the two central units 10, 11 inertial INS to generate by integration, with a repetition rate of 20 Hz (periodicity of 50 ms), two versions of the anti-skid control information YDa, YDb.

Each processing chain FG a, FGMb comprises at input a double buffer memory 20, 21, followed by a voting circuit 30, 31, an amplifier 40, 41 and an integrator 50, 51.

The double buffers 20, 21 are loaded at a rate of 20 Hz with pairs of samples of the lateral acceleration information, the two samples of the same pair coming from one of copy 10 and the other from copy 11 of inertial unit INS and corresponding to the lateral acceleration information γl1 and γl2 delivered by the two copies 10, 11 of inertial unit for the same time slot.

The voting circuits 30, 31 select, whenever necessary, that is to say every 20 ms, in the double buffer memory 20, 21 placed upstream, one of the samples of the last pair d 'lateral acceleration information samples entered. For this selection, they implement an arbitrary choice criterion consisting of a voting mechanism such as that described in the preamble. The amplifier circuits 40, 41 followed by the integrator circuits 50, 51 make it possible to extract, by integration, samples retained by the voting circuits 30, 31, two sampled versions YDa and YDb of anti-skid control information. intended for a yaw damping automation of a flight control system.

At the outputs of the two processing chains FGMa and FGMb, there is a subtractor circuit 60, supplying a threshold comparator 61 which receives its threshold from a register 62 and which issues a blocking order for the automations supplied by the two processing chains FGMa and FGMb in the event of detection of an overstepping of the threshold by the difference existing between samples of the same rank coming from the two chains. Indeed, an excessively large difference between samples of the same rank of the anti-skid control information originating from the two processing chains can raise fears of a malfunction of one of the circuits of the two processing chains FGMa and FGMb.

In the absence of noise from sensors, the redundant architecture which has just been described in relation to FIG. 1 makes it possible to safely supply the automatisms of a flight control system. On the other hand, in the presence of highly noisy sensor signals, as is encountered with inertial units disturbed by vibrational phenomena affecting the structure of the aircraft, in particular during the exit of the landing gear, the exit of flaps or airbrakes, opening of cargo door, carrying external load, etc., this architecture leads to the issuance of untimely orders to decoupled automation.

This can be seen by studying the diagrams in FIGS. 2a to 2d relating to the operation over the same period of time of the two processing chains FGMa and FGMb, in the absence of a fault, with strongly input signals noise from inertial units in good working condition but subject to parasitic vibrations from the aircraft cell. The diagram in FIG. 2a represents a highly noisy signal γl1 originating from one of the two examples of inertial unit INS affected by parasitic vibrations of the aircraft cell. The diagram in FIG. 2b represents the highly noisy signal γl2 coming from the other 11 of the two examples of inertial unit INS also subjected to parasitic vibrations from the cell of the aircraft. The general evolutions of the signals γl1 and γl2 are the same but the noises which affect them are different, the inertial units which deliver them do not have the same vibratory environment because they are not mounted exactly at the same location in the airframe. The diagram in FIG. 2c represents the two versions YDa and YDb of the anti-skid control information delivered by the two chains FGMa and FGMb in response to the signals γl1 and γl2. The relative evolution of the difference between the two versions YDa and YDb due to the slow drift of the integrator circuits 50 and 51 is aggravated by the effects, on the hardware and software asymmetries of the two processing chains FGMa and FGMb, of vibrational noises affecting the γl1 and γl2 signals. The diagram in FIG. 2d shows the state resulting from the blocking order B generated by the threshold comparator 61 intended for automations depending on the processing chains. We notice, in the middle of the diagram, untimely automatic blocking orders.

To reduce the frequency of untimely blocking orders of automatic systems dependent on redundant processing chains, it is proposed to place a noise filter upstream of the voting circuits. The redundant architecture used for the generation of YD anti-skid control information intended for a yaw stabilizing automation forming part of a flight control system is then modified in accordance with FIG. 3. In this Figure 3, the elements unchanged from Figure 1, that is to say the two copies of inertial unit, the voting circuits and the circuits arranged downstream of the voting circuits, keep the same indexing while the circuits already present but having changed timing resume the same assignment with a premium.

The redundant architecture of FIG. 3 differs from that of FIG. 1 by the presence in the two processing chains FGM'a and FGM'b, of two double noise filters 70, 71 interposed between the voting circuits 30 , 31 and the double buffers 20 ', 21' and by the fact that the double buffers 20 'and 21' operate at a rate of 50 Hz which is that of the data transmitted by the two copies 10, 11 of inertial unit and not at the 20 Hz rate of the output signals from the processing chains.

Each double noise filter 70 or 71 filters in parallel, the two series of samples delivered by the two copies 10, 11 of the inertial unit INS, at the rate of these suites, without subsampling but under the control of its own clock not synchronized with one or the other of the clocks of copies 10, 11 of inertial unit.

The double buffer memories 20 ′ and 21 ′ passing at the higher rate of the data transmitted by the two copies 10, 11 of INS inertial unit, are no longer used for sub-sampling but only for taking into account the absence of synchronism between the clocks at the same frequency two copies 10, 11 of INS inertial unit and double noise filters 70, 71.

The double noise filters 70, 71 are digital filters operating at the rate of 50 Hz from the data sequences from the two copies 10, 11 of the INS inertial unit, therefore before subsampling, in order to avoid at their level , possible problems posed by the folding of the tape accompanying a sub-sampling. The sub-sampling enabling the output samples of the two processing chains FGM'a and FGM'b to be passed at the rate of 20 Hz is done at the outputs of the double noise filters 70, 71 whose registers also serve as buffer memories .

The transfer functions of the double noise filters 70, 71 are chosen so as to block as best as possible the noises affecting the signals of the two copies 10, 11 of inertial unit INS1, INS2, while minimizing the useful signals. They are chosen after studying the vibrational environment of each exemplary inertial unit as a function of the flight configurations of the aircraft more particularly conducive to the appearance of vibrations on the airframe, such as outgoing landing gear configurations, flaps out, airbrakes out, open cargo door, carrying of external load, etc. They can even be changed depending on the current flight configuration. They are advantageously of the low-pass or band-cut type and of any order depending on the stiffness of the desired cut. (1 ^st , 2 ^nd order, Butterworth, etc.). They can also be obtained by sliding average over any number of samples.

When the annoying noises are broadband noises brought back into the useful band by the phenomenon of aliasing resulting from the sub-sampling making it possible to pass from the sampling rate of 50 Hz of the units 10, 11 of inertial unit to the rate sampling of 20 Hz output from the processing chains FGM'a, FGM'b, the transfer function chosen for the double anti-noise filters is that of an anti-aliasing filter. This anti-aliasing filter can be adjusted to block the components of frequency lower than half of the first sampling frequency of 50 Hz and of frequency higher than half of the second sampling frequency of 20 Hz.

When the annoying noises are due to resonance vibrations of the aircraft cell at precise frequencies in the useful signal band, the transfer function adopted for the double noise filters can be of the notch or plug type , with one or more stop frequencies placed at the frequencies of the resonance vibrations developing in the airframe of the aircraft at the locations of copies 10, 11 of inertial unit.

A double noise filter 70 or 71 as well as the control of the double buffer memory 20 'or 21' which precedes it and carries out the acquisitions of the samples of the lateral acceleration information delivered by the two copies 10, 11 of the inertial unit INS can be the subject in each FGM chain of the same software task executed at the same rate. The diagrams in FIGS. 4a to 4d illustrate the improvement brought by the noise filters. They are plotted over the same period of time and with the same highly noisy input signals as those of FIGS. 2a to 2d, always in the absence of failure of the inertial units and of the measurement chains. The diagram in FIG. 4a represents the signal γi'1 coming after noise reduction filtering from one of the two examples of inertial unit INS affected by parasitic vibratory phenomena of the aircraft cell. The diagram in FIG. 4b represents the signal γl'2 coming, after anti-noise filtering, from the other 1 1 of the two examples of INS inertial unit also affected by parasitic vibrational phenomena of the aircraft cell. The effects of anti-noise filtering make the general evolutions of the signals γl'1 and γl'2 more apparent and their resemblance. The diagram in FIG. 4c represents the two versions YDa and YDb of the anti-skid control information delivered by the two chains FGMa and FGMb in response to the signals γl1 and γl2. The noise filters considerably slow down the evolution of the gap between the two versions YDa and YDb during the slow drift of the integrator circuits 50 and 51, to the point of eliminating the untimely orders for blocking the automatic mechanisms as shown in the diagram in FIG. 4d representing the output state of the comparator at threshold 61. L redundant architecture with double noise filters which has just been described can be used with any flight control system receiving primary flight information, redundant processing chains having, by nature, a significant drift capacity due to terms of precision including integrators, such as those dealing with information coming from an inertial unit 1RS, AHRS or accelerometer block and this in any aircraft, whether it be an airplane, a helicopter, a drone, a missile, etc.

Claims

1. Method for processing information (γl1, γl2) coming from primary flight equipment (10, 11) mounted on board an aircraft, in a form sampled at a first rate in order to be delivered after processing (Yda, Ydb), to a flight control system of the aircraft, in a form sampled at a second rate lower than the first rate, characterized in that it consists in submitting the samples of information coming from primary flight equipment (10, 11) with digital noise reduction filtering (70, 71) carried out at the first sampling rate. 2. Method according to claim 1, characterized in that the digital noise reduction filtering (70, 71) is an anti-aliasing filtering blocking the frequency components greater than half the second sampling rate. 3. Method according to claim 1, characterized in that the digital noise reduction filtering (70, 71) is an anti-aliasing filtering blocking the components of frequencies lower than half of the first sampling rate. 4. Method according to claim 1, characterized in that the digital noise-reduction filtering (70, 71) is an anti-aliasing filtering blocking the components of frequencies greater than half of the second sampling rate and those of frequency less than the half of the first sampling rate. 5. Method according to claim 1, characterized in that the digital noise reduction filtering (70, 71) is a first order low pass filtering. 6. Method according to claim 1, characterized in that the digital noise reduction filtering (70, 71) is a second order low pass filtering. 7. Method according to claim 1, characterized in that the digital noise reduction filtering (70, 71) is a low-pass filtering of Butterworth type. 8. Method according to claim 1, characterized in that the digital noise reduction filtering (70, 71) is a notch filtering of Butterworth type. 9. Method according to claim 1, characterized in that, when the processed information coming from a primary flight equipment is affected by a noise having energy peaks, the digital noise reduction filtering (70, 71) is a filtering with cut bands corresponding to the energy peaks of the noise. 10. Method according to claim 1, characterized in that the digital anti-noise filtering (70, 71) is a sliding average filtering operating on several samples 11. Method according to claim 1, characterized in that the digital anti-noise filtering (70, 71) implements a transfer function depending on the flight configuration of the aircraft. 12. Device with redundant architecture with two parallel chains (FGM'a, FGM'b) for processing signals from primary flight equipment (10, 11) mounted on board an aircraft, said signals being available at a first cadence, in a sampled form and in several copies (γl1, γl2) and intended to be delivered after processing, always in several copies (Yda, Ydb), to a flight control system of the aircraft, in a sampled form at a second rate lower than the first rate, characterized in that it comprises, at the head of each chain (FGM'a, FGM'b), following a multiple buffer memory (20 ', 21'), a digital multiple noise filter (70, 71) which filters in parallel the various available copies of primary flight equipment signals (10, 11) and operates, like the multiple buffer memory (20 ', 21') at the first rate sampling. 13. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is an anti-aliasing filter blocking the frequency components greater than half of the second sampling rate. 14. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is an anti-aliasing filter blocking the components of frequencies lower than half of the first sampling rate. 15. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is an anti-aliasing filter blocking the components of frequencies greater than half of the second sampling rate and those of frequency less than half of the first sampling rate. 16. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is a first order low pass filter. 17. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is a second order low pass filter. 18. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is a low-pass filter of the Butterworth type. 19. Device according to claim 12, characterized in that the digital multiple noise reduction filter (70, 71) is a type notch filter Butterworth. 20. Device according to claim 12, characterized in that, when the processed information originating from primary flight equipment is affected by noise exhibiting energy peaks, the digital filter multiple noise canceling (70, 71) is a cut-band filter corresponding to noise energy peaks. 21. Device according to claim 12, characterized in that the digital multiple anti-noise filter (70, 71) is a sliding average filter operating on several samples 22. Device according to claim 12, characterized in that the digital multiple anti-noise filter (70, 71) has a transfer function depending on the flight configuration of the aircraft.