FR2826445A1 - Aircraft/missile central navigation system having inertial detectors generating angle/speed increments and pre processing with sculling/coning compensation applied providing second lower frequency increments deriving angle/speed increments - Google Patents

Abstract

The navigation information method produces angle and speed increments with inertial detectors. Pre processing is applied to the parameters producing second increments at a lower frequency. The pre processing is carried out with compensation for the phenomenon of coning and sculling and filtering carried out modifying the pass band. The angle and speed information are then derived from the second increments.

Description

<Desc / Clms Page number 1>

METHOD AND DEVICE FOR GENERATING INFORMATION
NAVIGATION AND INERTIAL NAVIGATION CENTER
INCORPORATING SUCH A DEVICE
The present invention relates to a method and a device for generating navigation information, comprising attitude information and speed information, as well as an inertial navigation unit incorporating such a device.

 It relates to the field of inertial navigation, and finds applications in on-board computers for a carrier such as an airplane, a missile, etc.

 The principle of inertial navigation consists in integrating acceleration values in a local geographic benchmark, linked to the earth, obtained from acceleration values supplied, in a measurement benchmark linked to the carrier, by accelerometers integral with the carrier. .

 In the right part of the diagram in Figure 1, we note [g] the local geographic coordinate system defined by a trihedron comprising three orthogonal axes Xg, Yg and Zg linked to the earth. The coordinate system [g] is also called TGL, for "Local Geographic Trihedron". In the left part of the diagram, we note [m] the measurement frame defined by three orthogonal axes Xm, Ym, and Zm linked to the carrier 1. Inertial sensors (not shown in Figure 1) are integral with the carrier 1. They include at least three accelerometers which supply acceleration values of the carrier along the three aforementioned axes of the reference [m]. They also include gyros which give the onboard computer a reference to the TGL.

mesure [m] au vecteur accélération ; ? g exprimé dans le TGL, est donné par la relation suivante :

où Tg/m désigne une matrice 3x3 appelée"matrice de passage", cette relation définissant un produit de matrices. The passage of the acceleration vector Ym expressed in the reference frame

measure [m] at the acceleration vector; ? g expressed in the TGL, is given by the following relation:

where Tg / m denotes a 3x3 matrix called "passage matrix", this relation defining a product of matrices.

 In practice, the generation of attitude information (heading, roll, pitch) is carried out by an algorithm based on the integration of a quaternion called attitude quaternion. Such an algorithm is indeed simpler to implement

<Desc / Clms Page number 2>

 work that an algorithm based on the angles of Euler. It is recalled that a quaternion is a mathematical tool which is in the form of a complex vector with four components, which is equivalent to the aforementioned passage matrix.

 As illustrated in the diagram in FIG. 2, the attitude quaternion is integrated periodically, at integration instants denoted tq, tq + AT, tq + 2. AT, etc. separated by a determined time interval AT. In other words, the quaternion is integrated at a frequency 1 / AT.

où w désigne le vecteur vitesse de rotation, et où a désigne l'intégrale de w à l'instant courant sur l'intervalle de temps AT. The variation Aq of the attitude quaternion between two successive integration instants is constructed using the rotation vector Qq, resulting from the integration over a time interval AT of the following differential equation:

where w denotes the speed of rotation vector, and where a denotes the integral of w at the current instant over the time interval AT.

mesure donnée par les gyromètres cumulée sur l'intervalle de temps AT), on

commet une erreur valant-j Am. dt. Cette erreur ne dépend que des 2 mouvements coniques du porteur (phénomène dit de"coning"en anglais). Using only the integral a of the speed of rotation w over the time interval AT (which is convenient since a is nothing other than the

measurement given by the gyrometers accumulated over the time interval AT), we

commits an error worth-Am. dt. This error depends only on the 2 conical movements of the carrier (phenomenon known as "coning" in English).

où, toujours, a désigne l'intégrale de la vitesse de rotation w à l'instant courant sur l'intervalle de temps AT. In practice, too, the generation of speed information is carried out periodically, at the frequency 1 / AT of integration of the attitude quaternion. The variation Av of the speed vector between two successive instants of computation is constructed using the specific force Fs and the attitude matrix, by integration of the following differential equation:

where, always, a designates the integral of the speed of rotation w at the current instant over the time interval AT.

 Using only the integral of the specific force Fs over the time interval AT (which is convenient since this is nothing other than the measurement given by the accelerometers accumulated over the time interval

<Desc / Clms Page number 3>

 AT), an error equivalent to jo is made. FAs. dt. This error depends only on the carrier's scull movements (phenomenon known as "sculling" in English).

 In a manner known per se, compensation algorithms are used, known as coning and culling compensation algorithms in the language of a person skilled in the art. These compensation algorithms are implemented in preprocessing means making it possible to produce angle increments and speed increments at low frequency (the integration frequency of the attitude quaternion) from the angle increments. and high frequency speed increments (the frequency at the output of accelerometers and gyrometers). More precisely, the time interval AT is divided into M smaller time intervals 6T (where M is an integer strictly greater than unity), on each of which we have the integral, respectively, of the speed of rotation w or the specific force Fs given by, respectively, the gyrometers or the accelerometers. These algorithms make it possible to cumulate the measurements given by the accelerometers and by the gyrometers over the time interval AT, by adding, respectively, a term of compensation of coning or a term of compensation of sculling. They make it possible to reduce the number of operations to be performed in real time, because the integration of the attitude quaternion and the speed are carried out at low frequency and not at high frequency, while compensating for coning and sculling.

 In certain applications, it is sought to introduce filtering of the navigation information, in order to define the bandwidth of the device. However, filtering the final attitude and speed information has several drawbacks.

 On the one hand, filtering at the level of attitude information is not equivalent to direct filtering at the level of angle increments. In the first case, the attitude angle can be indeterminate for certain attitudes of the carrier (the course is for example indeterminate when the pitch is equal to if / 2) which makes filtering impossible. In the second case, on the contrary, the filtering of the angle increments is similar to a kind of "virtual suspension", the characteristics of which do not depend on the attitude of the wearer.

<Desc / Clms Page number 4>

 On the other hand, the filtering at the level of the attitude information or of the speed information is unsuitable for certain system architectures, where it may be required to provide to a third-party computer angle increments and / or speed increments at low frequency compensated for coning and culling phenomena, with an imposed transfer function.

 Finally, the filtering of information sampled at low frequency induces aliasing of the spectrum, so that part of the energy of the signal can be found within the bandwidth which it is desired to eliminate, which makes totally ineffective filtering.

 The invention aims to propose a solution to this filtering problem in order to define the bandwidth of a device for generating navigation information comprising attitude information and speed information.

 This object is achieved by a method and a device for generating navigation information according to the invention.

 The method includes the steps of: - producing first angle increments and first speed increments, at a first determined frequency, using inertial sensors; applying a pre-processing of the first angle increments and of the first speed increment audits to produce second angle increments and second speed increments, at a second determined frequency lower than said first determined frequency, this pretreatment comprising compensation coning and culling phenomena, and, in addition, filtering to modify the bandwidth; - processing said second angle increments and said second speed increments to produce attitude information and speed information.

 Thus the digital calculation of the filtering, which makes it possible to modify the bandwidth of the device, is integrated into the coning and culling compensation algorithms.

 Preferably, the preprocessing includes a correction of the filtering in order to correct the effect of the filtering on the algorithms implemented for the

<Desc / Clms Page number 5>

 coning and culling compensation. In this way, one can use the coning and sculling compensation algorithms of the prior art, without modifying them or degrading their performance.

 The device comprises: - inertial sensors producing first angle increments and first speed increments at a first determined frequency; - pre-processing means for producing, at a second determined frequency lower than said first determined frequency, second angle increments and second speed increments from said first angle increments and said first speed increments, comprising means for compensating for the coning and culling phenomena, and, moreover, filtering means for modifying the bandwidth; processing means producing attitude information and speed information from said second angle increments and from said second speed increments.

 The device is therefore suitable for implementing the method.

The invention also relates to an inertial navigation unit incorporating such a device.

 Other characteristics and advantages of the invention will become apparent on reading the description which follows. This is purely illustrative and should be read with reference to the appended drawings, in which there is shown: - in Figure 1, already analyzed: a schematic representation of a measurement coordinate system linked to a carrier and a geographic coordinate system linked to the earth; in FIG. 2, also already analyzed: a diagram illustrating the division at a division level of a time interval between two successive instants of integration of the quaternion of attitude and of speed; - in Figure 3: a block diagram of a navigation information generation device; - in Figure 4: a block diagram illustrating the pre-processing means of a device according to the prior art; and, - in Figure 5: a block diagram illustrating the pre-processing means of a device according to the invention.

<Desc / Clms Page number 6>

 In Figure 3, there is shown schematically a device for generating navigation information.

 The device includes inertial sensors, in particular gyroscopes 31 and accelerometers 32. More particularly, it includes a gyroscope and an accelerometer for each of the axes Xm, Ym and Zm of the measurement mark [m]. The gyroscopes 31 produce increments of angle o, regularly spaced in time by a time interval 6T. The accelerometers 32 produce increments of speed V, also regularly spaced over time in the time interval oT. In other words, the angle increments oe and OV are produced at a determined frequency 1 / oT (hereinafter called high frequency).

 The device further comprises pre-processing means 33. These means have the function of producing angle increments AO and speed increments AV, from angle increments oe and speed increments OV at high frequency . The angle increments AO and the speed increments at are regularly spaced in time by a time interval AT.

 As illustrated by the diagram in FIG. 2, the time interval AT is divided into M time intervals OV (where M is an integer strictly greater than unity) for which there is a value high frequency angle increment ôO and a high frequency speed increment value 5V. The time interval AT is therefore M times greater than the time interval oT. In other words, the increments A6 and AV are produced at a second determined frequency 1 / AT (hereinafter low frequency) lower than the high frequency 1 / oT mentioned above.

 In an example, the high frequency is equal to 2000 Hz (hertz), so that the time interval oT is equal to 0.5 ms (millisecond), and the low frequency is equal to 100 Hz, so that l AT time interval is 10 ms.

 The device finally comprises processing means 34 producing navigation information comprising attitude information and speed information from the angle increments AO and speed increments AV at low frequency. These means 34 carry out the integration of the

<Desc / Clms Page number 7>

 attitude and speed quaternion, according to an algorithm known per se. Advantageously, the integration of the attitude quaternion and the integration of the speed are thus carried out at low frequency, which limits the computational load.

trois grandeurs scalaires 08x et oVx, 08y et oVy, et 08z et ôVz fournies à haute fréquence, respectivement par les gyroscopes 31 et par les accéléromètres 32. Dans la suite de la description et aux figures, le pré-traitement par les moyens 33 est explicité par référence aux grandeurs vectorielles, ce qui est plus commode. En réalité, il est bien évident que les calculs mis en oeuvre le sont à partir des grandeurs scalaires correspondantes, qui sont les seules valeurs disponibles. The invention relates to the pre-processing means 33. As indicated in FIG. 3, and as will be considered in the following description, the high frequency increments 08 and OV and the increments at low frequency AO and àV are vector quantities defined from

three scalar quantities 08x and oVx, 08y and oVy, and 08z and ôVz provided at high frequency, respectively by the gyroscopes 31 and by the accelerometers 32. In the following description and in the figures, the preprocessing by the means 33 is explained with reference to vector quantities, which is more convenient. In reality, it is quite obvious that the calculations implemented are based on the corresponding scalar quantities, which are the only values available.

 Figure 4 gives a schematic representation of the pretreatment means 33 of a device according to the prior art. In this figure, the suffix "c" and the suffix "s" are conventionally used to distinguish the references which designate the dual components of the means of the device participating, respectively, in the coning compensation algorithm and in the culling compensation.

The pre-processing means 33 comprise above all means 41 c for accumulating first angle increments 08 at high frequency and means 41s for accumulating first UV speed increments at high frequency, over a determined integration period AT corresponding to the inverse of the low frequency 1 / AT. The means 41c and 41s respectively produce an accumulated angle value aM and an accumulated speed value aM
The preprocessing means 33 also include means 43c for generating, from the first angle increments o, a term CC1 of first order coning compensation. Likewise, they include means 43s for generating, from the first angle increments oe and further from the first speed increments OV, a term CS1 of first order sculling compensation.

<Desc / Clms Page number 8>

 The pre-processing means 33 finally comprise means 42c for adding the term CC1 of first order coning compensation to the accumulated angle value aM, in order to generate the angle increments AO at low frequency. In a dual way, the preprocessing means 33 comprise means 42s for adding the term CS1 of first order culling compensation to the accumulated speed value aM, in order to generate the speed increments AV at low frequency.

 The preprocessing means 33 moreover generally comprise means 45c for generating, from the high frequency angle increments 08, a term CC2 of coning compensation at an order greater than unity, and means 45s for generating, from the first angle increments 08 at high frequency and further from the first speed increments oV at high frequency, a term CS2 of compensation for sculling at an order greater than unity. By order greater than unity, we mean here an order N where N is an integer strictly greater than unity.

 In this case, the preprocessing means 33 also comprise means 44c for additionally adding the term CC2 of coning compensation to an order greater than unity to the accumulated angle value aM, in order to generate the increments d 'AO angle at low frequency. In a dual way, the preprocessing means 33 also comprise means 44s for additionally adding the term CS2 of sculling compensation to an order greater than one to the value of accumulated speed aM, in order to generate the speed increments AV at low frequency.

où les coefficients aij sont choisis de façon connue en soi, de manière à annuler l'erreur à l'ordre N en présence d'un mouvement conique. In summary, the pre-processing means 33 make it possible to generate angle increments at low frequency AO according to the following relation:

where the coefficients aij are chosen in a manner known per se, so as to cancel the error at order N in the presence of a conical movement.

<Desc / Clms Page number 9>

où les coefficients bij sont choisis de façon connue en soi, de manière à annuler l'erreur à l'ordre N en présence d'un mouvement de godille. In addition, they generate speed increments at low AV frequency according to the following relationship:

where the bij coefficients are chosen in a manner known per se, so as to cancel the error in order N in the presence of a scull movement.

 The above expressions of the terms which are added in relations (4) and (5) are known per se, and implemented in all the coning and culling compensation algorithms known to date.

 FIG. 5 gives a schematic representation of the pretreatment means 33 of a device according to the invention. As in FIG. 4, the suffix "c" and the suffix "s" are conventionally used to distinguish the references which designate the dual components of the means of the device participating, respectively, in the coning compensation algorithm and to the culling compensation algorithm.

 These preprocessing means 33 comprise a first filter 50c, 50s having a determined transfer function, for filtering the angle increments 56 and the speed increments oV at high frequency. The function of this filter is to produce filtered angle increments H (08) and filtered speed increments H (oV). It allows to modify the bandwidth of the device. The filter transfer function, denoted H (p) in the following (where p denotes the Laplace variable), is determined according to the needs of the application. Typically this will be a low pass filtering function.

 The pre-processing means 33 also include means 51c for accumulating first filtered angle increments H (08) and means 51s for accumulating first filtered speed increments H (oV), over a determined integration period AT corresponding to the inverse of the bass

<Desc / Clms Page number 10>

 frequency 1 / AT. The means 51c and 51s respectively produce an accumulated angle value aMf and an accumulated speed value off.

 The preprocessing means 33 further comprise means 53c for generating, from the first filtered angle increments H (58), a term CC1f of first-order coning compensation. Similarly, they comprise means 53s for generating, from the first filtered angle increments H (58) and also from the first filtered speed increments H (5V), a term CS1f of first order sculling compensation .

il (P) II 2 + Il H, (P) 112 = 1. Ce second filtre 58c, 58s a pour fonction de filtrer les premiers incréments d'angle non filtrés 58 et les premiers incréments de vitesse non filtrés 5V (i. e., incréments non filtrés par le premier filtre 50c, 50s). Furthermore, the pre-treatment means 33 also include a second filter 58c, 58s, having a transfer function, denoted H ′ (p), such that

il (P) II 2 + Il H, (P) 112 = 1. This second filter 58c, 58s has the function of filtering the first unfiltered angle increments 58 and the first unfiltered speed increments 5V (ie, increments not filtered by the first filter 50c, 50s).

This second filter produces second filtered angle increments H '(60) and second filtered speed increments H' (5V).

 In addition, the preprocessing means 33 comprise means 57c for generating, from the second filtered angle increments H '(58), a first filter correction term CFc. Likewise, they comprise means 57s for generating, from the second filtered angle increments H '(50) and also from the second filtered speed increments H' (oV), a second filter correction term CFs .

 The preprocessing means 33 finally comprise means 52c and 56, for adding the term CC1f of coning compensation to the first order and the first term CFc of filtering correction to the accumulated angle value aMf, in order to generate the increments angle AO at low frequency. In a dual way, the preprocessing means 33 comprise means 52s and 56s, for adding the term CS1f of sculling compensation to the first order and the second term CFs of filtering correction to the accumulated speed value crmf, in order to generate the low frequency AV speed increments.

 In a preferred but nonlimiting embodiment, the pre-processing means 33 also comprise means 55c for generating, from the first unfiltered angle increments 50, a term CC2f of

<Desc / Clms Page number 11>

 coning compensation at an order greater than unity, and means 55s for generating, from the first unfiltered angle increments woe) and also from the first unfiltered speed increments ÓV, a compensation term CS2f of culling to an order greater than unity. By order greater than unity, we mean here an order N, where N is an integer strictly greater than unity.

accumulée aMf, afin de générer les incréments d'angle AO à basse fréquence. In this embodiment, the preprocessing means 33 also include means 54c for additionally adding the term CC2f of coning compensation to an order greater than one to the angle value

accumulated aMf, in order to generate the angle increments AO at low frequency.

In a dual way, the preprocessing means 33 also comprise means 54s for additionally adding the term CS2f of culling compensation to an order greater than one to the accumulated speed value friend in order to generate the speed increments AV at low frequency.

où to est l'instant de mise en route de l'algorithme. The pre-processing means 33 according to the invention thus make it possible to generate angle increments at low frequency AG according to the following relationship:

where to is the instant when the algorithm is started.

In addition, the pre-processing means 33 according to the invention make it possible to generate speed increments at low frequency AV according to the following relationship:

<Desc / Clms Page number 12>

où to est l'instant de mise en route de l'algorithme.

where to is the instant when the algorithm is started.

 It will be noted that the terms of coning and sculling compensation to an order greater than unity (CC2 and CS2) are not affected by the method according to the invention. In addition, more generally, the coning and sculling compensation algorithms are not modified by the method according to the invention. The coning and sculling compensations at the first are just implemented from the filtered high frequency angle increments H (ó8) and the filtered high frequency speed increments H (#V), that is to say say from the measurements provided by the inertial sensors not raw, but after filtering by the filter 50c, 50s. In other words, they are implemented from the filtered high frequency angle increments H (zoo) and the filtered high frequency speed increments H (#V), without being otherwise modified. On the other hand, coning and sculling compensation of higher order to unity, are always implemented, if necessary, from unfiltered high frequency angle increments ## and speed increments at high unfiltered frequency #V, that is to say from the raw measurements provided by the inertial sensors.

 It will also be noted that, in the above expression of the value ssm involved in the first and second filtering correction terms CFc and CFs respectively, the second angle increments fi! very H '(59i) and the second filtered speed increments H' (V1) are accumulated since the instant to which is the instant of starting of the algorithm, and not the instant of the start of the interval of AT time considered.

<Desc / Clms Page number 13>

 The invention has been described above in a preferred but nonlimiting embodiment. Variants are possible without departing from the scope of the invention. In particular, the method is entirely applicable in the case of the implementation of coning and culling compensation algorithms with two levels of division of the time interval AT. Such algorithms make it possible to reduce the number of operations to be performed, and are known in themselves by those skilled in the art.

 The pre-processing means 33, as moreover the processing means 34 of the device are advantageously produced in the form of a program carrying out digital calculations. This program is executed, for example, in an on-board computer of an inertial unit.

Claims (9)

1. Method for generating navigation information, comprising the steps consisting in: - producing first angle increments (58) and first speed increments (5V), at a first determined frequency (1 / 5T), at l using inertial sensors; - apply a pre-processing of the first angle increments (58) and first speed increment audits (5V) to produce second angle increments (ÀûO) and second speed increments (AV), at a

 second determined frequency (1 / AT) lower than said first determined frequency (1 / 5T), this pre-processing comprising compensation for the phenomena of coning and sculling, and, moreover, filtering to modify the bandwidth; - processing said second angle increments (b. 8) and said second speed increments (AV) to produce attitude information and speed information.  2. Method according to claim 1, according to which the preprocessing comprises a correction of the filtering in order to correct the effect of the filtering on the algorithms implemented for the compensation of the phenomena of coning and sculling.  3. Method according to claim 2, according to which the preprocessing comprises the steps consisting in: - filtering the first angle increments (58) and the first speed increments (5V) by a first filter (50c, 50s) having a determined transfer function, denoted H (p), for producing first filtered angle increments (H (58)) and first filtered speed increments (H (5V)); - accumulating first filtered angle increments (H (58)) and accumulating first filtered speed increments (H (5V)), over a determined integration period (AT) corresponding to the inverse of said second frequency <Desc / Clms Page number 15>  determined (1 / AT), to respectively produce an accumulated angle value (aMf) and an accumulated speed value (off); - generate, from the first filtered angle increments (H (08)), a term (CC1f) of first order coning compensation, and generate, from the first filtered angle increments (H (08)) and furthermore from the first filtered speed increments (H (oV)), a term (CS1f) of first order culling compensation; - filter the first angle increments (zoo) and the first speed increments (OV) by a second filter (58c, 58s) having a transfer function, denoted H '(p), such as #H (p) # 2 + # H '(p) # 2 = 1, to produce second filtered angle increments (H' (08)) and second filtered speed increments (H '(V)); - generate, from the second filtered angle increments (H '(08)), a first filter correction term (CFc), and generate, from the second filtered angle increments (H' (08)) and further from the second filtered speed increments (H '(oV)), a second filter correction term (CFs); - add the term (CC1f) of first order coning compensation and the first term (CFc) of filtering correction to the accumulated angle value (aMf) to generate said second angle increments (##), and add the term (CS1f) of first order sculling compensation and the second term (CFs) of filtering correction to the accumulated speed value (aMf) to generate said second speed increments (AV).  4. The method as claimed in claim 4, further comprising the steps consisting in: generating, from the first angle increments (8), a term (CC2f) of coning compensation at an order greater than unity, and generating, from the first angle increments (08) and further from the first speed increments (#V), a term (CS2f) of sculling compensation at an order greater than unity; - additionally add the term (CC2f) of coning compensation to an order greater than one to the accumulated angle value (aMf) to generate said <Desc / Clms Page number 16>  second angle increments and further adding the term (CS2f) of sculling compensation to an order greater than one to the accumulated speed value (off) to generate said second speed increments (AV).  5. Device for generating navigation information, comprising: - inertial sensors (31, 32) producing first angle increments (o) and first speed increments (OV) at a first determined frequency (1 / sot) ; - pre-processing means (33) for producing, at a second determined frequency (1 / AT) lower than said first determined frequency (1 / # T), second angle increments (AG) and second speed increments (AV) from said first angle increments (##) and said first speed increments (OV), comprising means of compensation (53c, 55c, 53s, 55s) of the coning and sculling phenomena, and, in further, filtering means (50c, 50s) for modifying the bandwidth; - processing means (34) producing attitude information and speed information from said second angle increments (A9) and said second speed increments (AV).  6. Device according to claim 5, in which the preprocessing means (33) comprise means (56c-58c, 56s-58c) for correcting the filtering in order to correct the effect of the filtering on the algorithms used for the compensation. coning and sculling phenomena.  7. Device according to claim 2, in which the preprocessing means comprise: - a first filter (50c, 50s) having a determined transfer function, denoted H (p), for filtering the first angle increments (##) and the first speed increments (OV), which produces first filtered angle increments (H (oS)) and first filtered speed increments (H (oV)); - means (51c) for accumulating first filtered angle increments (H (bone)) and means (51s) for accumulating first filtered speed increments (H (oV)), over a determined integration period (AT) <Desc / Clms Page number 17>  - a second filter (58c, 58s), having a transfer function, denoted H '(p), such that jH (p)) + H' (p) = 1, for filtering the first angle increments (zoo) and the first speed increments (OV), which produces second filtered angle increments (H '(08)) and second filtered speed increments (H' (oV)); - means (57c) for generating, from the second filtered angle increments (H '(08f)), a first term (CFc) for filtering correction, and means (57s) for generating, from the second filtered angle increments (H '(08)) and further from second filtered speed increments (H' (oV)), a second filter correction term (CFs); - Means (52c, 56c) for adding the term (CC1f) of first order coning compensation and the first term (CFc) of filtering correction to the accumulated angle value (aMf), in order to generate said second increments angle (AO), and means (52s, 56s) for adding the term (CS1f) of first order sculling compensation and the second term (CFs) of filtering correction to the accumulated speed value (Oj \ / Mf), in order to generate said second speed increments (AV).

 corresponding to the inverse of said second determined frequency (1 / AT), respectively producing an accumulated angle value (aMf) and an accumulated speed value (aMf); - means (53c) for generating, from the first filtered angle increments (H (08)), a term (CC1f) of first-order coning compensation, and means (53s) for generating, from the first filtered angle increments (H (08)) and further from the first filtered speed increments (H (oV)), a first order sculling compensation term (CS1f);  8. Device according to claim 7, further comprising: - means (55c) for generating, from the first angle increments (08), a term (CC2f) of coning compensation at an order greater than unity , and means (55s) for generating, from the first angle increments (08) and also from the first speed increments (OV), a term (CS2f) of culling compensation at an order greater than l 'unit; <Desc / Clms Page number 18>  - means (54c) for additionally adding the term (CC2f) of coning compensation to an order greater than one to the accumulated angle value (aMf) in order to generate said second angle increments (they), and means (54s) for further adding the term (CS2f) of culling compensation to an order greater than one to the accumulated speed value (#Mf) in order to generate said second speed increments (AV).  9. Inertial navigation unit comprising a device according to any one of claims 5 to 8.