Adaptive learning controller for synthetic aperture radar

• US 4,692,765 A
• Filed: 12/05/1985
• Issued: 09/08/1987
• Est. Priority Date: 12/05/1985
• Status: Expired due to Fees

• Current Assignee: Environmental Research Institute Incorporated
• Original Assignee: Environmental Research Institute of Michigan
• Inventors: Politis, Demetrios T., Licata, William H.
• Primary Examiner(s): Tubbesing, T. H.
• Assistant Examiner(s): Barron, Jr., Gilberto
• Application Number:
  US06/805,186
• Time in Patent Office: 642 Days
• Field of Search: 343/5 CM, 343/5 DP, 343/5 ST, 343/8, 343/5 PL, 343/5 MM, 364/453, 364/456, 342/25, 342/106
• US Class Current: 342/106
• CPC Class Codes: G01C 21/188 for accumulated errors, e.g. by coupling inertial systems with absolute positioning systems
• CPC Class Codes: G01S 13/9019 Auto-focussing of the SAR signals

Abstract

An adaptive learning controller (ALC) for use with an inertial navigation system (INS). To correct quadratic position errors which result from acceleration bias errors, the ALC produces a position correction signal. The position correction signal is generated by twice integrating an acceleration correction signal produced by the ALC. The ALC receives signals which indicate a current system state based on the corrected position signal and the velocity bias signal. The ALC also receives a failure signal determined by comparing the corrected position signal to predetermined failure criteria, these criteria relating to excursions of the corrected position signals beyond acceptable error limits.

Claims

1. For use with an inertial navigation system adapted to produce a position measurement signal by integrating both an acceleration measurement signal and the resulting velocity estimate signal, an apparatus for generating velocity bias estimate and position correction signals in order to correct bias errors in the acceleration measurement signal, comprising:

means for adding the position correction signal to the position measurement signal, thereby generating a corrected position signal;

means for comparing the corrected position signal to first predetermined error criteria and generating a first failure signal indicative of a failure of the corrected position signal to meet the first predetermined error criteria;

means, adapted to receive the corrected position signal and velocity bias estimate signal, for generating a system state signal;

adaptive learning controller means, adapted to receive the first failure signal, a second failure signal and the system state signal, for generating an acceleration correction signal in response thereto;

means for integrating the acceleration correction signal to generate the velocity bias estimate signal;

means for integrating the velocity bias estimate signal to create the position correction signal;

means for averaging the acceleration correction signal over a period of time; and

means for comparing the time-averaged acceleration correction signal to second predetermined error criteria and generating a second failure signal indicative of a failure of the time-averaged acceleration correction signal to meet the second predetermined error criteria.

2. A method for processing data produced by a mobile radar system, the data comprising radar, elapsed time, and measured position signals, comprising the steps of:

(a) sampling the measured position, elapsed time, and radar signal data at discrete points in time;

(b) integrating a rate of change of velocity bias signal, thereby producing a velocity bias signal;

(c) integrating the velocity bias signal, thereby producing a position correction signal;

(d) subtracting the position correction signal from the measured position signal, thereby producing a corrected position signal;

(e) comparing the corrected position signal to a predetermined upper signal threshold and generating a positive count for each sample time when the corrected position signal exceeds the upper threshold;

(f) comparing the corrected position signal to a predetermined lower signal threshold and generating a negative count for each sample time when the corrected position signal falls below the lower threshold;

(g) accumulating the positive and negative counts produced in steps (e) and (f), to produce a first failure criterion signal;

(h) dividing the value of the rate of change of velocity bias by the value of the elapsed time and integrating the resulting signal to produce a second failure criterion signal;

(i) producing a state variable signal as a function of the values of the velocity bias, the position correction, and the first and second failure criteria signals;

(j) processing the state variable signal by an adaptive learning controller, responsive to failures signified by the first and second failure criteria independently exceeding predetermined first and second upper and lower failure bounds, for producing the two-valued rate of change of velocity bias signal; and

(k) recording the radar, elapsed time, and corrected position signal data,

whereby the radar signals and elapsed time signals are placed in correspondence with the corrected position signal data.

3. The method of claim 2, wherein the state variable signal produced in step (i) is a binary-valued vector signal, with exactly one component having a value of one and the remainder of the components having a value of zero.

4. The method of claim 2, wherein the two-valued rate of change of velocity bias signal takes on the values of +a and -a, where a is a predetermined positive number.

5. The method of claim 2, wherein step (j) further comprises the steps of:

(j1) receiving the state variable signal and the first and second failure criterion signals and producing an internal reinforcement signal r* in response thereto, the internal reinforcement signal conveying information concerning the value of the state variable and the existence of at least one failure; and

(j2) receiving the state variable signal, the internal reinforcement signal r*, and the next last value of the two-valued rate of change of velocity bias signal produced by this step, for constructing associations between the state variable and the rate of change of velocity bias signals.

6. The method of claim 5, wherein:

step (g) further comprises, when the value of the first failure criterion signal exceeds the predetermined first upper and lower failure bonds, the step of resetting the value of the first failure criterion signal to a predetermined first value; and

step (h) further comprises, when the value of the second failure criterion signal exceeds the predetermined second upper and lower failure bounds, the steps of resetting the value of the elapsed time to zero and resulting the value of the second failure criterion signal to a predetermined second value.

7. A method for producing recorded data produced by a mobile radar system, the data comprising radar, elapsed time, and measured position signals Y, comprising the steps of:

(a) sampling the measured position, elapsed time, and radar signal data at uniformly separated discrete points in time;

(b) integrating a rate of change of velocity bias signal V thereby producing a velocity bias signal Y_c ;

(c) integrating the velocity bias signal Y_c, thereby producing a position correction signal Y_c ;

(d) subtracting the position correction signal Y_c from the measured position signal Y, thereby producing a corrected position signal Y-Y_c ;

(e) comparing the corrected position signal to a predtermined upper signal threshold Y_b and generating a positive count of 1 for each sample time when the corrected position signal exceeds the upper threshold;

(f) comparing the correction position signal to a predetermined lower signal threshold -Y_b and generating a negative count of -1 for each sample time when the corrected position signal falls below the lower threshold;

(g) accumulating the positive and negative counts produced when the corrected position signal exceeds the positive threshold Y_b or falls below the negative threshold -Y_b, respectively, to produce a first failure criterion signal E, and, when the value of the first failure criterion signal E exceeds predetermined first upper and lower failure bounds E_b and -E_b, resetting the value of the first failure criterion signal E to a value of zero;

(h) dividing the value of the rate of change of velocity bias by the value of the elapsed time and integrating the resulting signal to produce a second failure criterion signal F, and when the value of the second failure criterion signal F exceeds the predetermined second upper and lower failure bounds Fb and -F_b, resetting the value of the elapsed time to zero and resetting the value of the second failure criterion signal to zero;

(i) producing a binary state variable vector X, the state variable having N states which are functions of the values of the velocity bias Y_c, the position correction Y_c, and the first and second failure criteria, E and F, such that X has N binary components at the time of the kth sample, x_i (k), for i=1 to N, exactly one of the components having the value of one, the remainder of the components having a value of zero;

(j) receiving the binary state variable vector X, the first failure criterion signal E, and the second failure criterion F, and producing an internal reinforcement signal r* in response thereto in accordance with the following formulas, applicable at the kth time sample;

##EQU1## γ

is a non-negative constant less than 1, β and

δ

are positive constants;

(k) receiving the binary state variable vector X, the internal reinforcement signal r*, and the next last value of the two-valued rate of change of velocity bias signal produced by this step, in accordance with the following formulas;

##EQU2## α

is a positive constant, and 0<

β

<

1; and

(l) recording the radar, elapsed time, and corrected position signal (Y-Y_c) data,

whereby the radar signals and elapsed time signals are placed in correspondence with the corrected position signal data.