RU2625603C2 - Method of unmanned aircraft motion parameters minimax filtration with correction from satellite navigation system - Google Patents

Abstract

FIELD: instrument making.

SUBSTANCE: invention relates to aviation and space instrument-making devices and unmanned aerial vehicles (UAV) parameters filtration systems, determining location in space using correlation data from several navigation instruments and can be used for UAV flight parameters filtration coming from on-board navigation system (BNS) to increase accuracy of UAV motion parameters determining. For this purpose, UAV parameters filtration process is performed in discrete time intervals based on processing information on UAV current position, coming from BNS and satellite navigation system (SNS). Filtering UAV motion parameters in current position consists of coming from BNS motion parameters minimax filtration, and periodic BNS correction from SNS. UAV parameters minimax filtration is based on calculation of data areas, taking into account measuring device possible error range and approachability areas (OA). Based on analysis of mutual position information areas and OA UAV motion parameters vector evaluation is determined, based on which as UAV control for transition into new position is determined. At periodic BNS correction from SNS in discrete time intervals measured data areas are abruptly reduced to minimum sizes determined by SNS motion parameters determining accuracy, and then vary in accordance with features of BNS operation till the next correction moment.

EFFECT: increase in accuracy.

1 cl, 3 dwg

Description

The present invention relates to methods for filtering the motion parameters of unmanned aerial vehicles (UAVs) that determine the location in space using data correlation from several navigation devices.

A known method of filtering the parameters of the trajectory of the object and the device for it is a patent RU No. 2307376 C1 (publ. 2007). The essence of this method is that the reduction of the filtering error of the parameters of the trajectory of the tracking radar objects occurs by adapting the smoothing coefficients of the filter to the current conditions of tracking the object. This allows you to accompany objects in conditions when the coordinates of the objects change in the general case according to unknown nonlinear laws (for example, when maneuvering the object), and their measurements are carried out with constant or variable discreteness and various unknown errors. Since this method allows filtering the parameters of the trajectory of the tracked object only from the radar station, this method is not suitable for determining the motion parameters of the UAV when placing the navigation system on board.

Known modified Kalman filter described in patent RU No. 2160496 C2 (publ. 2000), based on the preliminary averaging of measurements by using the unit for calculating the optimal weight coefficients.

Modification of the Kalman filter complicates algorithms for estimating UAV motion parameters, requires a large amount of memory and speed of an onboard computer, and does not guarantee the possibility of using UAVs for arbitrary unknown maneuvers.

Another way of adaptive signal filtering is the method described in patent WO No. 201336154 A1 (publ. 2013), based on signal filtering by an adaptive Kalman filter, the passband of which is adjusted in accordance with a direct estimate of the filtered signal parameter characterizing the dynamics of the filtered signal.

The essence of the method is that the control bandwidth of the Kalman filter is carried out through the values of the weight coefficients of the filter based on a direct estimate of the intensity of the forming process with a known (measured) estimate of the power spectral density of the additive Gaussian white observation noise. In this case, the passband of an adaptable Kalman filter at any moment of filtration is close to optimal in a wide range of operating conditions, which leads to an increase in filtering accuracy. However, the Kalman adaptive filter only works in cases where white noise is a disturbance of the observed flight parameters. Another drawback of the Kalman filter is that the convergence of the estimates depends on how accurately linear equations describe the behavior of a real system.

Known non-linear adaptive automatic control system described in patent RU No. 2267147 C1 (publ. 2005), solving the problem of automatic control (with an error asymptotically tending to zero) of a multidimensional dynamic object with an unknown mathematical description and arbitrary perturbing effects in the presence of non-linear restrictions in the form of compatible equalities and inequalities for controlled variables, control actions and trajectories of transition of the control object to the required state. The invention is made in the form of an adaptive control system with self-tuning PID controllers and generates state variable estimates and control actions using modified Kalman filter algorithms, in which moving average autoregression is used as a priori models of the control object and control actions generator.

Modified Kalman filter algorithms complicate the UAV motion parameter estimation algorithms and work only in cases when white noise is a disturbance of the observed flight parameters.

For the closest analogue of the claimed invention adopted the method of inertial-satellite navigation of aircraft (LA) described in patent RU No. 2536768 C1 (publ. 2013), consisting in the fact that the joint processing of input data on the position of the aircraft, formed independently inertial sensors that generate the angular velocity and acceleration vectors of the aircraft, a barometric altimeter and a satellite receiver of the global navigation satellite system with a well-known satellite almanac and the working composition determined in flight about the constellation of satellites. In the input processing, data on the position of the aircraft is generated in an inertial manner, calculating them based on the readings of inertial sensors and a bar altimeter, and parallel data on the position of the aircraft determined by the satellite method in the satellite receiver is extracted. In the output processing, on the basis of the mentioned estimates of the inertial and satellite methods, the errors of the inertial method are estimated using the extended Kalman filter. Next, the error correction of the inertial method in the input processing is performed and the updated position of the aircraft is determined in the form of the difference between the estimates of the position of the aircraft determined by the inertial method and the mentioned error estimates of the inertial method.

The main disadvantage of this method is the lack of accuracy in determining the motion parameters of a maneuvering UAV, since the Kalman filter algorithms work well only in cases where the disturbance of the observed flight parameters is white noise and when the UAV control program is known in advance.

The claimed invention has the task of increasing the accuracy of determining the motion parameters of a maneuvering UAV, the control of which is determined during the flight in the presence of measurement errors of motion parameters, the statistical properties of which are not known.

The solution of this problem is carried out by filtering the UAV motion parameters coming from the onboard navigation system (BSS), using the minimax filtering method based on the analysis of information areas and the minimax filtering algorithm based on approximating information areas with parallelepipeds. The method and algorithm do not require knowledge of the statistical characteristics of measurement errors at the BNS output and the current UAV control.

Since when using the navigation system on board the UAV, measurement errors increase with increasing measurement time, to increase the accuracy of filtering, periodic correction of the BNS should be carried out using the satellite navigation system (SSS), while at discrete times T _corr , corresponding to the period of correction of the BNS from SNA, the measured information areas are abruptly reduced to the minimum size, determined by the accuracy of determining the parameters of the SNA movement, and then change in accordance with the features of the BNS until the next moment of correction. The moments of correction time of the BNS from the SNA T _corr are selected minimum to ensure the necessary accuracy in determining the motion parameters of the maneuvering UAV.

Achievable technical result from the implementation of the proposed method is to increase the accuracy of determining the motion parameters of a maneuvering UAV, the control of which is determined during the flight in the presence of measurement errors of the motion parameters, the statistical properties of which are not known, due to the use of minimax filtering of the motion parameters at the BPS output, while discrete time moments corresponding to the period of correction of the BNS from the SNA T _corr , the measured information areas are abruptly reduced to the minimum size, determined by the accuracy of determining the motion parameters of the SNA, and then change in accordance with the features of the BPS UAV until the next moment of correction.

In FIG. 1 shows the information area and the reachability area (OD) of the UAV.

In FIG. 2 shows the approximation of information areas and OD of a UAV by parallelepipeds.

In FIG. 3 is a block diagram illustrating a minimax algorithm for filtering UAV motion parameters with correction from a satellite navigation system.

The method is as follows. When solving the minimax filtering problem with correction from the SNA, the UAV motion parameters are determined at discrete time instants t ₀ , t ₁ = t ₀ + Δt, etc. until the end of the movement, where Δt is the sampling step.

When implementing the minimax filtering method, the signal at the output of the UAV UAV UAV is set in the form

χ (t) = z (t) + ξ (t),

where z (t) is the vector of motion parameters of the UAV;

ξ (t) is the vector of measurement errors of the BNS.

The statistical properties of measurement errors are unknown, but they are limited:

| ξ (t) | ≤ξ.

It is assumed that, without correction of the BNS, measurement errors increase over time. The rate of increase of measurement errors depends on the type of BNS used.

The movement of the UAV is determined by the vector differential equation:

where α (t) is the unknown UAV control satisfying the constraint:

It is assumed that the vector χ (t) is obtained as a result of processing primary information using stochastic linear filtering methods.

The minimax filtering method with correction from the SNA consists of the following steps.

вектора параметров движения БПЛА z(t) (см. фиг. 3) 1. Строится информационная область W(t) (см. фиг. 1), совместимая с измеренным сигналом χ(t) (см. фиг. 3) 2.For UAVs in position {t, z (t)}, the estimate

vector of motion parameters of the UAV z (t) (see Fig. 3) 1. The information area W (t) (see Fig. 1) is constructed, compatible with the measured signal χ (t) (see Fig. 3) 2.

An OD is constructed for the system (1) G (t + Δt, W (t)) at time t + Δt from the region W (t) in the presence of constraints (2) (see Fig. 3) 3.

The UAV moves to the position {t + Δt, z (t + Δt)} with control determined by estimating the vector of UAV motion parameters and limited (2) (see Fig. 3) 4.

With minimax filtering of the UAV motion parameters at the BNS output with correction from the SNA at discrete time instants corresponding to the BNS correction period from the SNA T _corr (see Fig. 3) 5, the measured information areas are abruptly reduced to the minimum dimensions determined by the accuracy of determining the SNA motion parameters , and then change in accordance with the features of the operation of the BNS UAV until the next moment of correction (see Fig. 3) 6. The time of correction of the BNS from the SNA T _corr are selected as minimum to ensure the necessary accuracy of determination dividing the motion parameters of a maneuvering UAV.

For time t + Δt, the information area W _y (t + Δt) is constructed, compatible with the measured signal χ (t + Δt) (see Fig. 1), taking into account the possible error range of the measuring device (see Fig. 3) 7.

The information region W (t + Δt) (see Fig. 3) 8 is constructed as the intersection of the regions G (t + Δt, W (t)) and W _y (t + Δt) (see Fig. 1).

как Чебышевский центр области W(t+Δt) (см. фиг. 3) 9 и переходят к следующему шагу фильтрации и т.д. до момента окончания движения БПЛА.The vector estimate is determined

like the Chebyshev center of the region W (t + Δt) (see Fig. 3) 9 and go to the next filtering step, etc. until the end of the UAV movement.

In the general case, the construction of ODs and information areas is possible only on the basis of their approximation. For approximation, information areas and ODs are placed in parallelepipeds _{W W} (t), _{G G} (t), the dimensions of which are equal to the dimensions of the filtered vector. In FIG. 2 presents a two-dimensional case.

In this case, the operation of intersecting the regions P _Wy (t + Δt) and P _G (t + Δt) is quite simple, as a result, the information region W (t + Δt) will be in the parallelepiped P _W (t + Δt).

The minimax filtering algorithm with correction from the SNA based on the approximation of information areas and OD by parallelepipeds consists of the following steps:

- the information area W (t) is placed in an n-dimensional parallelepiped П _W (t) = [z (t): β _i (t) ≤z _i (t) ≤γ _i (t), i = 1, ... n] where n is the dimension of the UAV motion parameters vector;

- the parallelepiped is calculated, which majorizes the OD G (t + Δt, W (t)):

PG (t + Δt) = [z (t + Δt): β _iG (t + Δt) ≤z _i (t + Δt) ≤γ _iG (t + Δt), i = 1, ... n];

- to determine the boundaries of the OD parallelepiped, β _iG (t + Δt) and γ _iG (t + Δt) are calculated (Tolpegin OA, Telyakov RF, Minimax filtering of the motion parameters of a descent aircraft. Collection of “Actual Problems of Protection and Security”, Volume 1 Proceedings of the XVI All-Russian Scientific and Practical Conference - St. Petersburg: NGO "Special Materials", 2013, p. 437-442);

- according to the measurement results χ (t + Δt), the box is determined:

П _Wy (t + Δt) = [z (t + Δt): β _iχ (t + Δt) ≤z _i (t + Δt) ≤γ _iχ (t + Δt), i = 1, ... n],

where γ _iχ (t + Δt), β _iχ (t + Δt) are set a priori, and the vector χ _i (t + Δt) is used as the center;

- if the current time t + Δt corresponds to the moment of correction of the BNS from the SNA T _corr , the measured information areas P _Wy (t) are abruptly reduced to the minimum size, determined by the accuracy of determining the motion parameters of the SNA;

- the box is defined:

П _W (t + Δt) = [z (t + Δt): β _i (t + Δt) ≤z _i (t + Δt) ≤γ _i (t + Δt), i = 1, ... n],

where γ _i (t + Δt) = min [γ _iG (t + Δt), γ _iχ (t + Δt)];

β _i (t + Δt) = max [β _iG (t + Δt), β _iχ (t + Δt)].

находятся составляющие вектора

и переходят к следующему шагу фильтрации.- according to the formulas

vector components are found

and proceed to the next filtering step.

The claimed method works as follows. An estimate of the vector of UAV motion parameters at the current position is determined, an information area is constructed that is compatible with the measured BPS signal at the current position. The OD of UAV motion parameters is built for the next position from the current information area in the presence of control restrictions.

The UAV makes a real movement from the current position to a new position with control determined by evaluating the vector of UAV motion parameters. In the new position, an information area is built that is compatible with the measured signal, taking into account the possible range of errors of the measuring device. If the new moment of time corresponds to the moment of correction of the BNS from the SNA, the measured information areas are abruptly reduced to the minimum size, determined by the accuracy of determining the parameters of the SNA movement.

The information area is constructed as the intersection of the reachability area and the information area, taking into account the possible range of errors of the measuring device.

The estimation of the UAV motion parameters vector in the new position as the Chebyshev center of the information area is determined. Goes to the next filtering step.

Thus, the invention allows to obtain a technical result, namely to increase the accuracy of determining the motion parameters of a maneuvering UAV, the control of which is determined during the flight in the presence of measurement errors of the motion parameters, the statistical properties of which are not known due to the use of minimax filtering of the motion parameters at the BPS output, while at discrete instants of time corresponding to the period of the correction LBP SNA T _corr measured information areas abruptly reduced to minim lnyh sizes determined accurately determine motion parameters SNA and then changed in accordance with the peculiarities of LBP UAV until the next time the correction.

Claims (1)

Method for minimax filtering of motion parameters of an unmanned aerial vehicle (UAV) with correction from a satellite navigation system (SNA), consisting of joint processing of input data on the position of UAVs generated independently by the onboard navigation system (BSS) and SNA, characterized in that for filtering motion parameters UAVs arriving from the BNS use the minimax filtering method and the minimax filtering algorithm, at the same time at discrete time points corresponding to the period of BNS correction from the SSS T _corr . These information areas are abruptly reduced to the minimum size, determined by the accuracy of determining the motion parameters of the SNA, and then change in accordance with the features of the UAV operation of the UAV until the next moment of correction.

Images