WO2001036913A1

WO2001036913A1 - Inertial measurement system

Info

Publication number: WO2001036913A1
Application number: PCT/US2000/031088
Authority: WO
Inventors: Kenneth S. Morgan
Original assignee: Honeywell Inc.
Priority date: 1999-11-18
Filing date: 2000-11-13
Publication date: 2001-05-25
Also published as: IL149729A0; EP1248942A1; JP2003515117A

Abstract

An apparatus for accurately measuring inertial measurements for use in an inertial measurement system includes a high range sensor, a low range sensor, and an apparatus for combining data from the high range sensor and the low range sensor to provide optimal accurate inertial measurements.

Description

INERTIAL MEASUREMENT SYSTEM

BACKGROUND OF THE INVENTION

Inertial measurement systems are used in aircraft to determine multiple characteristics of the aircraft such as velocity, position, and many other characteristics of the aircraft. Current inertial measurement systems require both good measurement sensitivity and a high operational range. Current systems use expensive inertial measurement sensors to provide accuracy in a wide range. Typically, these systems have one sensor per axis of measurement. A typical system would have three accelerometers and three gyros. The problem was that inexpensive high rate sensors could measure at high rates, but were not highly sensitive and inexpensive low rate sensors couldn't measure at high rates accurately, but were more accurate at low rates. Fig. 1 shows a chart of a low rate sensor compared to a high rate sensor. As can be seen in the chart, low rate sensor are extremely accurate at low rates. However, as the rates increase, they become very inaccurate very quickly. High rate sensors, on the other hand, are not as accurate but it's accuracy is consistent at all rates. As seen in Fig. 1 , the high rate sensor becomes considerably more accurate than low rate sensors at higher rates.

Fig. 1 shows the benefit of using a low rate sensor at low rates and a high rate sensor at high rates. However, if a single inertial sensor were to be designed to be highly sensitive at low rates as well as having a wide dynamic range, combining both these features in a single device results in an expensive as well as complex inertial device. It would be beneficial if a low cost inertial measurement system existed which could accurately measure at high and low rates as well as measure with a high accuracy. It would also be beneficial if such a device was not complex or expensive.

SUMMARY OF THE INVENTION An apparatus for accurately measuring inertial measurements for use in an inertial measurement system combines high range sensor data with low range sensor data to attain the optimal accurate inertial measurements. An optimal sensor filter combines data from a high range sensor and a low range sensor. wJEF DESCRIPTION OF THE DRAWINGS

Fig. 1 shows a chart displaying low rate and high rate sensor accuracy.

Fig. 2 shows an upper level block diagram of the present invention.

Fig. 3 shows the analog to digital conversion of the low rate and high rate sensor data.

Fig. 4 shows biasing and compensation of the sensor data after it has been converted to digital form.

Fig. 5 shows the combination of the high rate and low rate sensor data.

Fig. 6 shows a block diagram of the Kalman filter processing.

Fig. 7 shows the noise matrix used in the present invention.

DETAILED DESCRIPTION OF THE PRESENT INVENTION The present invention is an optimal processing apparatus and technique which allows for the efficient processing of two sets of low and high range data in an inertial measurement system 1. Fig. 2 shows a block diagram of the present invention. As can be seen, two sets of data, one set from the high range sensor 2 and one set from the low range sensor 4 come into the present invention 6. A priori sensor 8 information is also sent to the present invention 6. The a priori sensor 8 information assists in combining the low and high sensor data and optimizing the sensor data. All this data is sent through an optimal sensor filter 10 which provides optimal sensor data which is a combination of the low and high range data. Some of the sensor data is sent directly to the Kalman filter 12 which will be discussed in further detail later in the description. The sensor data is processed by navigation solution apparatus 14 to determine position, velocity, attitude and other well known outputs which are output out of the present invention 6. This sensor data is also sent to the Kalman filter 12 for further optimizing.

External aiding data 16 is sent to the Kalman filter 12 for processing in which the aiding data provided to the Kalman filter 12 includes inputs from an odometer or GPS. Aiding data 16 is used to improve the inertial navigation solution. The aiding data 16 is not limited to the aforementioned inputs which are used for example purposes only. Other resources and data could be used depending on the requirements of the respective system 1. Further, the aiding data 16 can be used as calibration for accuracy or compensation for possible errors. Also, at times no aiding data is required and thus, used. The present invention 6 is not limited to use of aiding data, but could perform without aiding data as well depending on the quality of the sensors and the system requirements. After the Kalman filter 12 finishes processing, the Kalman filter 12 outputs data which is sent back to the optimal sensor filter 10 for more processing to determine the gain of the optimal sensor data. This ^ cess continues until the optimal sensor data > tained and then is output creating an output better than the previously used inertial devices. The description of Fig. 2 is an upper level description of the present invention 6 and does not show many of the compensation means involved in refining the data to produce the optimal sensor data. A more specific description of the present invention 6 will be discussed in the rest of the detailed description.

Fig. 3 shows the data going into the present invention 6. The high range sensor 2 will provide data both of a high rate gyro 20 and a high g accelerometer 22. The low rate sensor 4 will provide data both of a low rate gyro 24 and a low g accelerometer 26. Temperature sensors 28 provide data important in processing. The temperature data is considered as a priori sensor 8 information and is used to compensate for errors by calibrating to account for the temperature sensitive errors of the gyros 20, 24 and accelerometers 22, 26. All the data is converted from analog to digital data by an analog to digital converter 29. The data from the sensors 2, 4 is taken in a voltage form and this conversion changes the data from voltage to frequency data. A low rate angle (ΔΘ^G1 _LOW), high rate angle (ΔΘ^G2 _Hlgh)_. low rate velocity (ΔV^A1 ,_w) and high rate velocity (ΔV^High) data are created wherein the "A" represents the reference frame for accelerometers 22, 26 and "G" represents the reference frame for gyros 20, 24. The "High", "Low" for the Δθ is for high and low rate gyros 20, 24 and the "High", "Low" for ΔV is for high and low g accelerometers 22, 26. The temperature data are filtered and used as thermal scale factors and thermal bias data.

Fig. 4 shows more processing performed by the present invention 6 with the angle and velocity data with more a priori sensor 8 information. The angle and velocity data is first compensated by the temperature scale factors, in a scale factor compensation means 30, for the respective sensors. The data output after the scale factor compensation means 30 is biased by the thermal bias data for the respective sensors and biased by Kalman filter corrections. This bias compensation is performed by a bias compensation means 32. The Kalman filter corrections are sent from the Kalman filter 12 to the Optimal Sensor Filter 10 as shown in Fig. 2 and are used to optimize the optimal sensor data output. The angle and velocity data is then orthogonally aligned by an orthogonal compensation means 34 to create nominally orthogonal outputs. Orthogonal data is required for Kalman filter processing which will be performed later in the Kalman filter 12. The orthogonal frame is referred to as the GR frame and the data is now represented as ΔΘ^GR _.w, ΔΘ^GR _Hlgh, ΔV^GR _Low and ΔV°^R _Hιgh with the GR frame of reference.

Fig. 5 shows the present invention 6 continuing to process the orthogonal outputs and with some high and low rate sensor compensation and filtering by a low rate compensation and filtering means 36 and a high rate compensation and filtering means respectively, i nis data is then sent to the optimal sensor filter 10. The compensation and filtering is to achieve the data from 1200 Hz to 200 Hz which is a much more manageable frequency for computations. An angle and velocity vector is created for each high and low range sensor data and this data is sent to the optimal sensor data filter 12. Also, this data in the vehicle body frame of reference is input into the optimal data sensor filter 10 as well so that the vehicle body is an additional reference frame for accuracy. This is represented by the data with the "B" in the superscript as opposed to the "GR" which is the orthogonal data of the sensors. The optimal sensor filter 10 then takes the high and low rate data and combines it to provide the outputs ΔΘ^GR, ΔV°^R which are the delta angle vector and the delta velocity vector used to determine characteristics of the vehicle. Again, the combination of the high and low rate sensor data provides an accurate output of the vehicle characteristics.

Th following equations are used in determining the delta angle vector and the delta velocity vector:

Δθ GR lLo ΔΘ^G lH^Righ

^■=[!]

σ² represents the variance in the data based on a priori information 8 which is well known and for example can include sensor specifications, sensor calibration data, known operating ranges, and known frequency characteristics. These are just examples but other sensor information can be used in to obtain accurate estimates of the magnitude of the sensor errors. The above equations show the delta angle vectc The delta velocity vector is determined by S Λituting V in place of θ in the respective portions of the equations.

Other outputs are provided as well such as angular rate, angular acceleration, and linear acceleration. Also, the weighting of how much of the high and low rate sensor data used in determining the delta angle vector and the delta velocity vector are determined as well. These are represented by K_LOWΔΘ^GRLOW, K_H,ghΔΘ^GR _Hlgj₁, LOW ΔV^GR _o„ and K_H._gh ΔV^GR _Hlgh where K _w and K_Hlgh are the weighting factors representing the low and high rate sensor data respectively. Since the present invention 6 combines the high and low rate sensor data to achieve optimal accuracy, differing percentages of each respective data is used. Looking at Fig. 1 again, the ranges where low rate sensor data is more accurate and where the high range sensor data is more accurate can be seen. If the sensor is operating in a range where the low rate sensor is more accurate, the low rate data is weighted more and if the sensor is operating in a range where the high rate data is more accurate, the high rate data is weighted more. Again, these values indicate the amount the combined data is represented by the high and low rate sensors.

A navigation solution means 6 combines the angle data and the velocity data. Navigation solution means 6 uses navigation equations which are well known in this area of technology. Obtaining the combination of low rate and high rate data is the key element of the present invention. After that data is obtained by the optimal sensor filter 10. Known equations can be used to determine position, velocity, attitude and be output for use. The present invention 6 is not limited to these outputs, but are used for example purposes only. Other output can be determined with known mathematics and methods.

The Kalman filter 12 is used to aid in the navigation system 1 in providing the optimal information. The Kalman filter 12 produces data to correct and optimize the optimal sensor filter 10 data taking into account of all the error state variables that could possibly exist. Fig. 6 shows a block diagram of the determination for what is required in producing updated error states which are sent back to the optimal sensor filter 10 to provide optimal processing. Firstly, an error state dynamics matrix Φ is formed. This matrix is formed with the aid of output from the navigation solution means 6 as well as aiding data from other external aiding sources 16 which provide error states (X). Aiding sources are well known in this area of technology and will not be discussed in any further detail in this description. The aiding data and the error state matrix (X) used in the present invention 6 are:

Aiding data is not limited to the above recited data, but could include other data which is well known in this area of technology. All the data comes in the form of the F matrix which is made up of the data shown by:

Fl2j ψ^L,δ

F₁₂ δvL,6ASFGR ) =

The error state dynamics matrix Φ is determined based on the F matrix, an Identity matrix I (well known in this area of technology) and the current time, t. Once the error state dynamics matrix Φ is determined, the error state dynamics matrix Φ and the error states matrix (X) from the current time period are multiplied to determine the error states matrix X of the next time period. The Covariance matrix P is also determined to optimize the optimal sensor filter data. The current error state matrix X and the current covariance matrix P are:

where P is a matrix wuh a size equal to the number of Kalman filer stales squared The diagonals of P contain the initial variance of each Kalman filter state.

An example of this is the following and is repeated for all the error states. ) is the initial variance of the integrated velocity error states.

is the initial variance of the position error states. δp^l

is the initial variance of the aiding states δ>id

Determining P for the next time period uses the error state dynamics matrix Φ and a noise matrix (n) shown in Fig. 7 provided as aiding data from external aiding sources 10. The aiding sources 10 provide yet more aiding data in the form of a measurement and measurement sensitivity matrix (Z, H respectively). These matrices are used to form a Kalman gain matrix K where R is observation noise again from aiding data sources 10 and P is the current covariance matrix determined earlier. The Kalman Gain matrix K is then used to update both the covariance matrix P and the error state matrix X and this data is sent back to the optimal sensor filter and processed so that the output of the optimal sensor filter is optimized and accurate.

The invention has been described herein in detail in order to comply with the Patent Statutes and to provide those skilled in the art with the information needed to apply the novel principles and to construct and use such specialized materials and components as are required. However, it is to be understood that the invention can be carried out by specifically different materials and components, and that various modifications, both as to the processing details and operating procedures, can be accomplished without departing from the scope of the invention itself.

Claims

The embodiments of the invention in which an exclusive property or right is claimed are defined as follows:

1. An apparatus for accurately measuring inertial measurements for use in an inertial measurement system, comprising: high range sensing apparatus; low range sensing apparatus; and apparatus for combining data from the high range sensing apparatus and the low range sensing apparatus to provide optimal accurate inertial measurements.

2. The apparatus of claim 1 wherein the combining apparatus is an optimal sensor filter.

3. The apparatus of claim 1 , further comprising a Kalman filter refining the combined data to assist in providing the optimal accurate inertial measurements.

4. An apparatus for accurately measuring inertial measurements for use in an inertial measurement system, comprising: high range sensing apparatus; low range sensing apparatus; apparatus for providing historical data; apparatus for providing additional external sensing data; apparatus for combining data from the high range sensing apparatus and the low range sensing apparatus; apparatus for determining inertial measurements for an vehicle; apparatus for refining the inertial measurements with the historical data and the external data to provide refined data which is sent to the determining apparatus which determines optimal accurate inertial measurements with the refined data.

5. The apparatus of claim 4 wherein the combining apparatus is an optimal sensor filter.

6. The apparatus of claim 4 wherein the refining apparatus is a Kalman filter.

7. An apparatus for accurately measuring inertial measurements for use in an inertial measurement system, comprising: a plurality of sensing means to sense an inertial measurement producing a plurality of inertial signals; and a control means connected to the plurality of sensing means receiving the plurality of inertial signals and combining the inertial signals.

8. The apparatus of claim 7 wherein the plurality of sensing means is a low range sensing means and a high range sensing means.

9. A method of optimizing data in an apparatus for providing optimal inertial measurements, comprising the steps of: forming current error states; propagating current error states to a future time; forming measurement and measurement sensitivity data; updating the error states based on propagated error states, measurement and measurement sensitivity data to provide optimal inertial measurements.