US 4,492,352 AGrant
Noise-Adaptive, Predictive Proportional Navigation (NAPPN) Guidance Scheme
Issue Date:1985-01-08
•6 Claims
•3 Drawing Sheets
Abstract
A method for augmenting the value of range uplink information in a predictive proportional navigation terminal guidance scheme. A second order configuration, with noise adaptive varying gain parameters calculated as a function of time-to-go, is particularly useful for high altitude, minimally maneuvering targets. This configuration results in acceptable miss distances, better than conventional proportional navigation, with sensitivity to unmodeled errors being substantially less than that in modern guidance systems. A third order configuration makes enhanced use of range uplink information to provide improved terminal guidance against lower altitude maneuvering targets. The sensitivity of the third order configuration is similar to that of conventional proportional navigation but with better miss distance results.
Metadata
Assignee
- General Dynamics, Pomona Division
Inventor
- William R. Yueh
Application Information
Application Number:US 4215370
Filing Date:1982-09-22
Priority Date:1982-09-22
Art Unit:221
Classifications
IPC:
F41G 722
Field of Search:
2443.15;3.16;3.19;3.2;3.21
Patent Drawings (3 sheets)
Description
Field of the Invention
This invention relates generally to terminal guidance for guided missiles and more particularly concerns a method for improving terminal guidance performance with minimum sensitivity in a range-dependent scheme, the system having gains which vary with time-to-go in a second order system and making enhanced usage of available range uplink information for high speed, maneuvering targets, together with such gains, in a third order system.
Background of the Invention
The principles of controlling guided missiles are well known. Basic principles are comprehensibly set forth in Locke, Guidance (1955). Since that book was published a wealth of information has been developed to refine and improve upon early missile control techniques, and to accommodate new and ever changing environments. Proportional navigation is discussed at various levels of detail in U.S. Pat. Nos. 3,189,300, 3,223,357, 3,603,531 and 3,718,293.
Earlier techniques worked well for large targets and for targets which were either geographically fixed or were moving relatively slowly and predictably. Such targets were at relatively low altitudes, usually not higher than 80,000 feet, flying at speeds not in excess of Mach 2. For target aircraft of these types, well established means are available in missile guidance computers to provide signals to the missile autopilot which, during terminal guidance, rely upon directly observed boresight error, computed line of sight rate and several other computed and estimated factors, to achieve intercept. Examples of such navigation computers and methods of estimating values used in solving navigational problems are shown in U.S. Pat. Nos. 4,128,837, 4,148,029 and 4,179,696.
However, with high speed, high altitude, maneuvering, small airborne targets, the problems to be solved in order to achieve intercept were greatly increased. In some previously available systems employing noise adaptive gains, the gains were also dependent upon recursive calculations from on-board and possible uplink information. Because of their sensitivy, these systems, often referred to as Kalman systems, tended to degrade rapidly due to unmodeled errors such as radome aberration, especially for high altitude targets. Thus while these systems had potentially optimal accuracy at intercept, errors which could not be modeled into them tended to degrade the resulting accuracy beyond that of conventional proportional navigation systems.
A third order system, sometimes referred to as Hanson's tracker, has been devised with optimal Kalman structure of combined prediction and correction loops for processing the line of sight data for a terminal guidance computer configuration. This was intended to improve over the range aided filtering technique (RAFT), and employed fixed gains. However, for extended range missiles, at high altitudes (generally in excess of 80,000 feet) it has been found that such a configuration has drawbacks which lead to instability or unacceptable miss distances.
Summary of the Invention
Broadly speaking, this invention pertains to a terminal guidance system for missiles employing techniques of proportional navigation in a predictive, noise-adaptive, range-dependent scheme. The system employs time varying gains in a system operable in either a second or a third order configuration, the gains depending only upon time-to-go to intercept. The third order scheme makes better use of uplinked range information compared with the prior art range aided filtering techniques. The result is a terminal guidance minimum sensitivity system that is more robust than the Kalman gain design, allowing wider system component tolerances without concomitant degradation of system performance.
In the second order system, which is particularly adapted to high altitude targets, gains A and B vary with the varying time constants .tau..sub.d and .tau..sub.c of adaptive guidance .alpha..sub.d and .alpha..sub.c filters in known systems. It has been found that miss distances are within acceptable tolerances, while the system sensitivity to unmodeled errors such as radome aberration is substantially decreased. The result is a substantial improvement over known systems at minimal cost for system modification and with a reduction in the computations required as compared with Kalman gain calculations.
The third order system employs the range uplink information to particular advantage and has a gain C which is variable in such a way as to combine the advantages of both the second and third order schemes. With A and B gains being dependent upon time-to-go, the third order system is effective against lower altitude maneuvering targets. While the sensitivity is somewhat greater than that of the second order scheme, sources of error such as radome and glint noise are of lesser significance so that terminal guidance results are improved over fixed gains systems employing range information.
This terminal guidance scheme is applicable to either medium range or extended range missiles which may be of the surface-to-air or air-to-air type. The range information may be obtained from either the launching mother ship location (surface or air) as where the missile and target are illuminated by the mother ship radar, an on-board active sensor, or from a passive ranging processor.
Brief Description of the Drawing
This invention will be readily understood from the following detailed description when read in conjunction with the accompanying drawing, in which:
FIG. 1 is a simplified block diagram of the control dynamics of a guidance system having range aided filtering with variable gains in accordance with the prior art;
FIG. 2 is a simplified block diagram of the control dynamics of the guidance system in accordance with the present invention;
FIG. 3 is a graph with the noise adaptive gains A, B and time constants .tau..sub.d, .tau..sub.c plotted against time-to-go; and
FIG. 4 is a sensitivity graph with RMS miss distance plotted against radome slope for the present invention and for prior art systems.
Description of the Preferred Embodiments
Before the details of the present invention are discussed, additional background information will be set forth for expository purposes.
While the present invention is specifically intended for surface-to-air or air-to-air, medium range and extended range missiles, it is not to be so limited. However, the target is preferably continually illuminated by the mother ship or launching platform to provide a source of information to the semi-active homing apparatus in the missile. The missile itself is typically either the non-rolling or roll-dithering type, the flight characteristics of both being well known.
The present missile guidance design adapts during flight to the changing receiver noise. The computer estimates the noise level and adjusts the filtering based upon that estimation. The Fast Fourier Transform (FFT) signal processing algorithm produces receiver output with spectral properties that characterize the present environments.
The guidance computer consists mainly of the guidance filtering algorithms. Guidance filtering serves two major purposes. It smooths the noise in calculating the missile-to-target line of sight rate for guidance, and it filters to stabilize any residual missile body motion coupling from the seeker head space stabilization control or from radome error type distortions at the antenna. The noise and body motion requirements on the guidance filtering vary during flight so the guidance computer is programmed to adapt to these changing environments.
The guidance filtering must be heavy enough to smooth the missile's flight. It reduces the general, somewhat unstable, motion due to large random accelerations responding to the noise. These produce excessive angles of attack with their attendant induced drag which dissipates the missile's kinetic energy. Yet the filtering must be light enough to allow quick missile response when needed to correct for heading errors or to chase a maneuvering target. It must filter out enough noise to prevent excessive breakup of the adaptive autopilot's role dither which becomes very small at high altitude, long range flight conditions. The filtering should knock down spikes and transients without overreacting with a heavy filter which remains after the transient and need for smoothing goes away. It should also minimize the effects of quantization and variable data rate.
These demands on the guidance filtering vary during flight. The guidance computer smooths out the tracking error noise and uses information from the autopilot (A/P) and inertial reference unit (IRU) to determine which demand is the most constraining and applies that to the variable filtering.
With reference now to FIG. 1, there is shown a simplified functional block diagram of the geometry loop, or control dynamics, of one channel of a conventional proportional navigation system as it operates in accordance with principles of the prior art range aided filtering technique (RAFT). Normally there would be two such channels to provide inputs to the quadrature control surfaces of the missile. Since they operate in the same manner, it is only necessary to describe one channel herein. At the left side of the diagram is the line of sight (LOS) angle .sigma. input, which is a measurable quantity. In a system such as that under discussion, where the target is illuminated from a reference location and the missile has only semi-active homing means, the LOS angle is computed in the on-board computer system from other available information. The derivation of the LOS angle input to the system will be referred to later.
The operation of the FIG. 1 system will be discussed at this point only to the extent necessary for reference purposes. It is similar in several respects to the system of the invention shown in FIG. 2.
With the feedback loop switch 5 open as shown, the system operates as a conventional proportional navigation system, without range uplink information in the guidance filter, and without an LOS rate prediction term to facilitate homing. The transfer function of the system with the switch open is represented by: ##EQU1## where
N.sub.c is the acceleration command to the autopilot,
.sigma..sub.M is the measured LOS rate with factors relating to radome slope error and noise,
.LAMBDA. is the navigation ratio,
V.sub.c is the closing velocity between missile and target, .tau..sub.d, .tau..sub.c are filter time constants, and
S is a Laplace operator.
When switch 5 is closed, range uplink infomation is utilized to provide an LOS angle acceleration estimate .sigma. to the input of the guidance filter. To compute .sigma., the output of the first filter is multiplied by 2V.sub.c and combined with the missile and target acceleration vectors Y.sub.M and Y.sub.T. Y.sub.M is information as to actual rate changes from the missile inertial reference unit, and Y.sub.T is the target acceleration vector which may be simulated in the guidance system. This latter input is intended to facilitate terminal guidance performance with respect to maneuvering targets. Note that the Y.sub.T input passes through a 1/2-second low pass filter when switch 6 is closed to simulate the uplinked target information. Because this only provides an additional external leg, the overall impulse response remains unchanged except for miss due to various types of target accelerations. Compensation for heading error, .theta.-bias (rate gyro info as to the seeker head), glint, receiver and amplitude noise are already accounted for in the system and remain unchanged. This combined signal is divided by the uplinked range information, resulting in the .sigma. value. This prediction term is multiplied by a varying .tau..sub.d factor and is combined with the .sigma..sub.M as the guidance filter input. Note that the feedback gains 2V.sub.c and 1/R are both variable, and that .sigma. is therefore range dependent.
It should be observed that the transfer function of equation (1) reflects the fact that the known system of FIG. 1 employs two cascaded first order filters, the .alpha..sub.d filter with a time constant of .tau..sub.d, and the .alpha..sub.c filter with a .tau..sub.c time constant. It is within the confines of this structure that the present invention functions. It is an object of this invention that it be compatible with the existing design, while significantly improving missile guidance performance in the terminal phase.
Kalman filters, that is, methods of providing theoretical optimal estimation in accordance with principles generally attributed to Kalman, are employed in what are commonly referred to as modern guidance systems. Such systems are those which provide both correction and prediction terms to facilitate intercept. It is generally agreed that a Kalman filtering scheme imposes more severe requirements on component tolerances such as radome error, band-limited glint and other correlated unmodeled errors, than the conventional proportional navigation system. If component tolerances can be met, the Kalman estimator should render optimal performance, in terms of minimal variance for RMS miss distance. But in actual practice, optimal performance and optimal results are unlikely. As component tolerances or measurement errors degrade, which is usually the case for radome refraction error for example, the missile performance will degrade much faster than for the conventional proportional navigation case.
One of the primary purposes of this invention is to ease the component tolerance requirements of Kalman, while providing a guidance scheme which is significantly improved over conventional proportional navigation. In order to achieve this result, ultimate miss distance accuracy is sacrificed to an acceptable degree in order to arrive at a system with a lower sensitivity to errors which could cause an unacceptable miss distance at the closest point of approach. Specifically, the sensitivity of this system to unmodelable errors such as radome error, is significantly reduced from Kalman systems. These errors are referred to as unmodelable because they depend upon too many factors during homing, which differ with each flight, to be specifically accounted for in the computer algorithm. These factors include look angle, frequency, beam polarization and dome temperature.
Kalman filters employ time varying gains, and therefore time varying time constants. To the extent that the present system also has time varying gains, it could be referred to as a Kalman type system. However, in Kalman the gains vary based upon recursive calculations from onboard and uplinked data. In the present guidance system the gains depend only upon time-to-go (T.sub.go). It is for this reason, simplification of the data input requirements to the guidance computer, that this guidance scheme, as compared with Kalman, requires less fine tuning of gains, less severe component requirements and has a lower likelihood of degradation as intercept approaches, while providing much of the terminal homing accuracy one would expect from a Kalman system.
Thus the choice of gains for this system has several advantages over known systems. These advantages include: (a) gains which are much simplier to calculate than Kalman gains; (b) a noise-adaptation feature, in common with Kalman; (c) relief from the severe degradation in performance for the unmodeled errors set forth above as occurs in Kalman systems; and (d) allowance for wider system component tolerances.
Reference material on methods of providing estimates for navigation problem values have been enumerated above. Other missile control systems with adaptive features are shown in U.S. Pat. Nos. 3,321,761, 3,575,362 and 3,911,345.
In the system of FIG. 1, body motion coupling due to radome error slope r is accounted for. At the input point the radome problem is simulated by the 1+r block applied to LOS angle .sigma. and the radome error slope r is multiplied by the airframe angle .psi. and subtracted at the same input point. The radome compensation loop K.sub.r, required for stabilizing the negative radome error slope, is inserted at the output rather than the input side of the guidance filters since that enlarges the stability region for positive error slopes and for cross plane coupling slopes. Radome compensation is addressed in U.S. Pat. No. 3,316,549.
With reference now to FIG. 2, there is shown a functional block diagram of one channel of the control dynamics of a system as it operates in accordance with this invention. Note that except for a portion of the guidance computer 17, this diagram has substantial similarities to the RAFT system of FIG. 1. At the left side of the diagram is the line of sight (LOS) angle .sigma. input as discussed above. This input is acted upon by block 11 indicated as 1+r, where r is the radome slope error. That thus modified LOS angle value is combined at combining point 12 with representations of the noise in the system together with .psi., the missile body axis angle with respect to a reference plane, multiplied by r, again the radome slope error, in block 13. The output of combining point 12 is .sigma..sub.M, a representation of the line of sight angle when combined with factors representing the radome error and noise in the system. This signal is then applied to combining point 14 in the seeker head track loop of the seeker section. The seeker head track loop has a .theta. input, which is the rate of change, obtained from a rate gyro, of the angle .theta. between the antenna axis and the missile axis. The .theta. signal is integrated in block 15 and the angle .theta. is also applied to point 14. The output of point 14, .epsilon., results from the fact that .sigma..sub.M -.theta.=.epsilon.. The value .epsilon. is the angle between the missle-to-target LOS and the antenna electrical boresight. Block 16 multiplies the angle .epsilon. by the track loop gain K.sub..epsilon. and that quantity is .theta. to be applied to the integrator 15.
As shown in FIG. 2, one of the external inputs to guidance computer 17 is .epsilon., and another is .theta., both shown at the upper left of the computer. Another input is Y.sub.M which is the lateral missile acceleration value from the inertial reference unit, resulting after acceleration commands have been made to the autopilot and the missile has responded. The airframe axis angle rate of change provides yet another computer input.
As a means of arriving at LOS angle .sigma., the lower portion of FIG. 2 includes a geometry section 21. This portion of the block diagram is not part of the actual mechanization of the guidance system and is only facilitated to derive a geometry line of sight angle .sigma. as an input to the system. The value Y.sub.T is a simulated quantity which, combined with the lateral missile acceleration Y.sub.M in combining point 22 and integrated twice as indicated by block 23, yields the value Y which is the lateral (vertical) displacement of the target and the missile with respect to the LOS, that is, the miss distance when evaluated at intercept. Thus the input to block 25 by means of line 24 is Y=Y.sub.T -Y.sub.M. When this quantity is divided by a simulated range value R in block 25, the result is the LOS angle .sigma..
Now with respect to guidance computer 17, there are two inputs to combining point 31, one of them being .theta. and the other being .sigma., the output of which is .sigma.-.theta.. This quantity is integrated in block 32, the output being .sigma.-.theta. which is applied to combining point 33 together with the input .epsilon.. From the relationship set forth previously,
we can see that the output of point 33, pursuant to the inputs just mentioned is
The significance of equation (3) is that the output of point 33 is essentially an LOS residue value, that value being the difference between the measured LOS angle .sigma..sub.M and the estimated LOS angle .sigma.. It is this value to which the gains of this sytem are applied within the guidance computer to provide the desired results of this invention.
The LOS residue value is applied to block 34, which is the gain B, and then to combining point 35. The output of combining point 35 is applied to .sigma. integrator 36 which also has the inputs of the initial conditions (IC). The output .sigma. of integrator 36 is applied to combining point 37, combining point 38 and to block 41 which multiplies by the value 2V.sub.c. V.sub.c is the closing speed between the missile and the target. At combining point 38 is also applied the value of the LOS residue through block 42 which is the gain A. This combination then completes the loop back to combining point 33 through combining point 31 and integrator 32.
The output of integrator 36, when multiplied by block 41, is applied to combining point 43 together with the lateral acceleration of the missile Y.sub.M, the combination then being applied to combining point 44. The LOS residue value is also multiplied by block 45 which is AV.sub.c, the output of which is applied to combining point 46 together with the output of combining point 44, which value is then divided by R in block 47, R being the range or distance between the missile and the target. The output of block 47 is connected through switch 51 back to combining point 35, all of which, together with the A and B gains, comprise the second order configuration of the present terminal guidance system.
The loop which makes this a third order system applies the LOS residue value to block 52 which is gain C, the output of which goes to combining point 53. Also applied to combining point 53, preferably uplinked from the mother ship, is the value of ##EQU2## together with the internal loop ##EQU3## from block 54. The output of combining point 53 is .eta. which is integrated in block 55 to provide an output of .eta.. This output goes to block 54 in a feedback fashion as well as to the range block 56, the output of which is applied to combining point 44.
To set forth the rest of the system as depicted in FIG. 2, the output of combining point 37 is .sigma.', representing a smoothed value of the LOS rate. This quantity is applied to block 61 which multiplies the LOS rate by .LAMBDA.V.sub.c where .LAMBDA. is the navigation ratio, normally having a value of 4. The output of block 61 is the command acceleration signal N.sub.c which is applied to the autopilot represented by block 62. The IRU is included within the representation of A/P and this provides the output Y.sub.M, which is the actual missile lateral acceleration, measured by the IRU and resulting from the missile flight response to the N.sub.c signals applied to the autopilot to operate the missile flight control surfaces.
The lateral missile acceleration Y.sub.M output from the autopilot is also applied to block 26 to divide by the missile velocity V.sub.M, which is the actual velocity of the missile through the air, independent of missile heading. The output from block 26 is .nu. which is the rate of increase of missile heading as it is changing direction to intercept the target. This missile heading rate is then operated upon by aero block 27, the output of which is .psi., previously stated as the rate of change of the missile body axis angle. Since the radome induced error depends not only upon LOS angle but also upon the actual angle of the missile axis, the factor representing the rate of change of body axis angle is operated upon by block 28 which is the radome compensation factor K.sub.r, and that product is then applied to combining point 37 as previously discussed. The body axis angle rate is then integrated in block 29 to provide body axis angle .psi. which is then multiplied by the radome error r in block 13 and applied to combining point 12 as set forth above.
From FIGS. 1 and 2 it can be seen that the following basic relationship exists:
Another basic relationship is useful and that is
From FIG. 2 and from equations (4) and (5) it may be seen that ##EQU4## In the scheme of this invention, there is provided an estimated LOS angle .sigma., an estimated LOS rate .sigma. and an estimated target acceleration Y.sub.T, all within the framework of the guidance system and within the capability of the person skilled in the art (see for example U.S. Pat. No. 4,148,029). If we define .sigma. in terms of the estimated target lateral position Y.sub.T, we have from equation (6)
The LOS rate equation, where the dot indicates time differentiation and T.sub.go =-1 is defined to be
The well-known Hanson's third order tracker can be described by the following equations:
and
Equation 11 assumes a band-limited target acceleration model with bandwidth 2.lambda.. One further relationship useful for defining .eta., the change of LOS rate due only to target acceleration, is ##EQU5##
By using equations (6)-(12) we can write the following equations of the block diagram of FIG. 2:
It should be noted that each of equations (13)-(15) has a prediction term and a correction term. The correction term in each of these equations is that term which includes the residual LOS measurement, .sigma..sub.M -.sigma., and the predictive portion of each equation is the remainder.
As stated previously, it is desirable to provide a significant improvement within the confines of the prior art guidance computer of the type shown in FIG. 1. In analyzing FIG. 2, if switch 51 is open, all of the range uplink dependent terms are disconnected so that the guidance computer can be reduced to ##EQU7## which is a second order filtering with complex pole. In order for the guidance filter to be conformed to the conventional .tau..sub.d and .tau..sub.c cascaded filters with two real roots of the type shown in FIG. 1, the gains are chosen as follows: ##EQU8## where .tau..sub.d and .tau..sub.c are the varying time constants for the known adaptive guidance .alpha..sub.d and .alpha..sub.c filters and are computed and employed within the guidance system of which the present invention is an improvement. Upon differentiating and substituting equations (17) and (18) into equation (16), we have ##EQU9## It will be noted that equation (19) is identical with equation (1) indicating that with switch 51 open, the guidance computer of FIG. 2 has a configuration where the transfer function is the same as that of the previous scheme shown in FIG. 1. However, in this instance, gains A and B have been chosen to vary to reflect the noise adaptive feature of the system, as shown in FIG. 3.
In the early stage of homing, gains A and B are very small so that heavy noise will not get through the filter. As intercept approaches and the signal-to-noise ratio becomes larger, the filters are gradually opened up by allowing almost exponentially increasing gains as a function of time-to-go. Due to seeker head non-linearities, head-aero loop stability and radome loop stability considerations, gains A and B cannot be allowed to go to infinitely large values at intercept. A ceiling is imposed that reflects the characteristics of the floor in the guidance time constants. The floor of the time constants plays a very important role since the nearly linearly decreasing time constant reaches the floor near a few total system time-constants-to-go. This is in the vicinity of the critical region which determines 90% of the missile performance. The error-inducing noises and bias beyond this critical time occur too early to have a significant effect upon performance since the transient response will die out in a few time constants, while the induced disturbances, at times very close to intercept, are too late for the missile to react and can essentially be ignored. Even though the choice of varying filter gains A and B is slightly sub-optimal compared with the choice of Kalman gains, as stated before a Kalman filtering scheme imposes more severe requirements upon component tolerances. Furthermore, this choice for the gains A and B has advantages in that it is simpler to calculate than Kalman gains while it has already exploited a noise-adaptation feature and will not suffer the severe degradation in performance for the unmodeled errors such as the radome problem and glint noise, among others.
In the guidance computer, the .sigma. integrator 36 is initialized for heading error at the start of homing and for other conditions which can be set prior to the terminal guidance phase of the missile. However, it has been found that in the third order tracker, where the gains A and B are varying according to equations (17) and (18), is not critical that the filter be perfectly initialized.
The system of FIG. 2 would normally operate with switch 51 closed. However, there is then a choice as to whether to operate the system as a second order or a third order scheme. If gain C is set at zero, the target acceleration terms are not involved. In that case equation (15) would not apply and equation (14) would be missing a target acceleration LOS rate term .eta.. Thus it can be seen that there is a reduction in the predictive aspects of the second order scheme (only LOS angle .sigma. and LOS rate .sigma. are estimated) but no reduction in the corrective aspects. For this reason, while the second order configuration does not account for a maneuvering target, it is useful at high altitudes where targets are known to either maneuver very little or not at all. Note that gains A and B are still noise-adaptive and vary dependent upon the time constants .tau..sub.d and .tau..sub.c in accordance with equations (17) and (18). There is also range aided filtering with a feed forward loop through block 45 and a feedback loop through combining point 43 being applied through range-dividing block 47 to combining point 35.
The sensitivity of the second order configuration is shown in FIG. 4, in comparison with conventional proportional navigation and a modern guidance system (MGS), usually referred to as a Kalman system. Note that MGS provides a miss distance at zero or slightly positive radome slope which is better than proportional navigation. However, it is evident from FIG. 4 that the MGS has a high sensitivity to radome error and, as stated earlier, such unmodeled errors cause accelerated degradation of such a guidance system. The conventional proportional navigation system is substantially less sensitive and relatively symmetrical about zero radome slope. Some of the advantages of the present system are readily apparent from FIG. 4 in that the second order scheme according to the present invention has substantially less sensitivity to radome error and at the same time it improves miss distance with respect to conventional proportional navigation. While MGS still provides optimal miss distance, the second order scheme of this invention is less accurate to only a slight and acceptable degree, at the same time being substantially more accurate than the conventional proportional navigation system. The fact that this invention has much less sensitivity to the unmodeled radome error indicates that it is substantially more robust than the MGS scheme. The curves of FIG. 4 do not include RMS miss due to target maneuvering.
When gain C is chosen as something other than zero, FIG. 2 becomes a third order scheme with further range aided filtering and further prediction terms that is, in equation (14) .eta. will have a value, and equation (15) will apply. With the addition of the Y.sub.T input to combining point 44, the output of that combining point is .sigma.R, the acceleration prediction term desirable for providing efficient interception of maneuvering targets. As indicated in FIG. 4, the third order configuration of the present invention has a sensitivity (not including target maneuvering) substantially the same as conventional proportional navigation but a better miss distance performance. Conventional proportional navigation does not have prediction terms in its guidance computer and is not very effective against maneuvering targets, but the third order scheme of the present invention has all of the advantages of the conventional proportional navigation with respect to radome error while at the same time providing improvement in performance against maneuvering targets. With the addition of uplink information from the mother ship of target acceleration (the term ##EQU10## at combining point 53) further information is provided to improve system performance. The value of gain C will typically range from 0 to 3 and is made adaptive to the noise level. It is normally determined before a missile is launched what type of target is to be intercepted and its general capabilities. It is also desirable to vary gain C as a function of time-to-go to take advantage of both the second and third order schemes. It is preferable to choose gain C at some value greater than zero at a relatively long time-to-go in order to reduce the residual miss due to target acceleration. However, it is preferred to have a second order tracker near intercept where C=0 and .eta.=0, especially at about the critical time region of a very few seconds to go, typically within two seconds to go. At this point, the second order tracker is desired for several reasons, such as smaller miss due to thermal noise, heading error, larger component tolerances against possible initial condition error in the .sigma. integrator 36 and the radome error.
To summarize, when gain C=0, FIG. 2 is a second order tracker that only estimates line of sight angle .sigma. and line of sight rate .sigma.. This is particularly adapted for high altitude, relatively non-maneuvering targets and could result in some possibly unacceptable steady state miss due to step target acceleration. For the third order scheme, where gain C is other than zero, and with a target acceleration estimation input, there is substantially no steady state miss but a trade-off result in increased sensitivity to thermal noise and radome coupling. Thus it is preferred to have finite gain C values at a large time-to-go to reduce the steady state miss and then gradually reduce gain C to zero near the critical time region where most of the receiver noise induced transient response characteristics of miss distance have been determined. This enables the invention shown in FIG. 2 to take advantage of both the second and third order tracker configurations.
Claims
I claim:
1. A method for terminal guidance of a missile to intercept a target, the missile having missile control surfaces to control missile flight direction, said method employing a predictive proportional navigation system including an autopilot, an antenna, an inertial reference unit and a noise adaptive guidance computer, said system including noise and radome error compensation, said computer having an adaptive guidance filtering system, said proportional navigation system having input values for line of sight (LOS) angle .sigma., antenna electrical boresight angle .epsilon., angle .theta. between the antenna axis and the missile axis, the rate of change .theta. of the angle .theta., and time varying noise-adaptive time constants .tau..sub.d and .tau..sub.c, where .epsilon.=.sigma..sub.M -.theta., deriving a modified value of LOS angle .sigma..sub.M, an estimate of LOS angle .sigma. and a residual LOS angle .sigma..sub.M -.sigma., the output of said filtering system being a smoothed line of sight rate .sigma. which is then multiplied by a product of the system navigation ratio .LAMBDA. and the closing velocity V.sub.c to provide command acceleration information to said autopilot to modify the position of the missile control surfaces, actual missile lateral acceleration being detected by said inertial reference unit, the output of said inertial reference unit being a measure of actual lateral acceleration Y.sub.M, said method comprising the steps of: applying the boresight error angle .epsilon. to a first combining point; multiplying the residual LOS angle from said first combining point by a gain A to provide the value A(.sigma..sub.M -.sigma.); combining the value A(.sigma..sub.M -.sigma.) with the LOS rate .sigma. at a second combining point to provide an estimated LOS rate .sigma.; combining said estimated LOS rate .sigma. with said rate .theta. to provide the value .sigma.-.theta.; integrating the value .sigma.-.theta. to provide .sigma.-.theta.; feeding back the value .sigma.-.theta. to said first combining point; combining said value .sigma.-.theta. with the angle .epsilon. to provide the value .sigma..sub.M -.sigma. at the input to said filtering system; multiplying the residual LOS angle by a gain B to provide the value B(.sigma..sub.M -.sigma.); multiplying the LOS rate .sigma. by 2V.sub.c ; combining said multiplied LOS rate with acceleration term Y.sub.M at a third combining point to provide the term 2V.sub.c .sigma.-Y.sub.M ; multiplying the residual LOS angle by AV.sub.c to provide the value AV.sub.c (.sigma..sub.M -.sigma.); combining the values AV.sub.c (.sigma..sub.M -.sigma.) and 2V.sub.c .sigma.-Y.sub.M at a fourth combining point to provide the value dividing the value 2V.sub.c .sigma.-Y.sub.M +AV.sub.c (.sigma..sub.M -.sigma.) by the range R between said missile and the target to provide the value ##EQU11## combining the immediately preceding value with the value B(.sigma..sub.M -.sigma.) to provide the corrective and predictive term ##EQU12## then integrating the term .sigma. to provide the smoothed LOS rate .sigma.; where ##EQU13## whereby gains A and B are noise-adaptive and dependent upon time-to-go, while the guidance filter output .sigma. is also range dependent and includes both corrective and predictive terms, said method being a second order guidance scheme.
2. The method recited in claim 1 and comprising the further steps of: multiplying the residual LOS angle by a gain C to provide the term C(.sigma..sub.M -.sigma.); combining external input of uplinked target information ##EQU14## together with C(.sigma..sub.M -.sigma.) and a value ##EQU15## at a fifth combining point to provide a value .eta.; integrating the value .eta. to provide the value of .eta., the change of LOS rate due to target acceleration; multiplying .eta. by ##EQU16## in a feedback loop and applying ##EQU17## to said fifth combining point; multiplying .eta. by R to provide an estimated target acceleration value Y.sub.T ; and combining Y.sub.T with the value 2V.sub.c .sigma.-Y.sub.M at a sixth combining point to provide an estimated LOS acceleration term .sigma.R, said term .sigma.R being combined with the value AV.sub.c (.sigma..sub.M -.sigma.) at said fourth combining point to provide the value 2V.sub.c .sigma.-Y.sub.M +Y.sub.T +AV.sub.c (.sigma..sub.M -.sigma.); said method being a third order guidance scheme.
3. The method recited in claim 2 wherein gain C is initially set at a value greater than zero depending on the boresight error noise level and reduces to zero when T.sub.go is less than about two seconds.
4. The method recited in claim 2 wherein gain C is set at a value greater than zero and remains fixed during homing and terminal guidance.
5. A method for terminal guidance of a missile to intercept a target, the missile having missile control surfaces to control missile flight direction, said method employing a predictive proportional navigation system including an autopilot, an antenna, an inertial reference unit and a noise adaptive guidance computer, said system including noise and radome error compensation, said computer having an adaptive guidance filtering system, said proportional navigation system having input values for the range R between missile and target, line of sight (LOS) angle .sigma., antenna electrical boresight angle .epsilon., angle .theta. between the antenna axis and the missile axis, the rate of change .theta. of the angle .theta., and time varying noise-adaptive time constants .tau..sub.d and .tau..sub.c, where .epsilon.=.sigma..sub.M -.theta., deriving a modified value of LOS angle .sigma..sub.M, an estimate of LOS angle .sigma. and a residual LOS angle .sigma..sub.M -.sigma., the output of said filtering system being a smoothed line of sight rate .sigma. which is then multiplied by a product of the system navigation ratio .LAMBDA. and the closing velocity V.sub.c to provide command acceleration information to said autopilot to modify the position of the missile control surfaces, actual missile lateral acceleration being detected by said inertial reference unit, the output of said inertial reference unit being a measure of actual lateral acceleration Y.sub.M, said method comprising the steps of: applying the boresight error angle .epsilon. to a first combining point; multiplying the residual LOS angle from said first combining point by a gain A to provide the value A(.sigma..sub.M -.sigma.); combining the value A(.sigma..sub.M -.sigma.) with the LOS rate .sigma. at a second combining point to provide an estimated LOS rate .sigma.; combining said estimated LOS rate .sigma. with said rate .theta. to provide the value .sigma.-.theta.; integrating the value .sigma.-.sigma. to provide .sigma.-.theta.; feeding back the value .sigma.-.theta. to said first combining point; combining said value .sigma.-.theta. with the angle .epsilon. to provide the value .sigma..sub.M -.sigma. at the input to said filtering system; multiplying the residual LOS angle by a gain B to provide the value B(.sigma..sub.M -.sigma.); multiplying the LOS rate .sigma. by 2V.sub.c ; combining said multiplied LOS rate with acceleration term Y.sub.M at a third combining point to provide the term 2V.sub.c .sigma.-Y.sub.M ; multiplying the residual LOS angle by AV.sub.c to provide the value AV.sub.c (.sigma..sub.M -.sigma.); multiplying the residual LOS angle by a gain C to provide the term C(.sigma..sub.M -.sigma.); combining external input of uplinked target information ##EQU18## together with C(.sigma..sub.M -.sigma.) and a value ##EQU19## at a fourth combining point to provide a value .eta.; integrating the value .eta. to provide the value of .eta., the LOS rate due to target acceleration; multiplying .eta. by ##EQU20## in a feedback loop and applying ##EQU21## to said fourth combining point; multiplying .eta. by R to provide an estimated target acceleration value Y.sub.T ; combining Y.sub.T with the value 2V.sub.c .sigma.-Y.sub.M at a fifth combining point to provide an estimated LOS acceleration term .sigma.R; combining the values AV.sub.c (.sigma..sub.M -.sigma.) and .sigma. at a sixth combining point to provide the value 2V.sub.c .sigma.-Y.sub.M +Y.sub.T +AV.sub.c (.sigma..sub.M -.sigma.); dividing the immediately preceding value by the range R to provide the value ##EQU22## combining the immediately preceding value with the value B(.sigma..sub.M -.sigma.) at a seventh combining point to provide the corrective and predictive term ##EQU23## then integrating the term .sigma. to provide the smoothed LOS rate .sigma.; where ##EQU24## whereby gains A and B are noise-adaptive and dependent upon time-to-go, while the guidance filter output .sigma. is also range dependent and includes both corrective and predictive terms, said method being a third order guidance scheme.
6. The method recited in claim 5 wherein the values of gain C and LOS rate .eta. due to target acceleration are set at zero, so that all of the terms involving gain C and LOS rate .eta. are zero and estimated target acceleration value Y.sub.T does not exist, the values AV.sub.c (.sigma..sub.M -.sigma.) and 2V.sub.c .sigma.-Y.sub.M are combined at said sixth combining point, and the output thereof is divided by the range R to provide the value ##EQU25## which is combined with B(.sigma..sub.M -.sigma.) at said seventh combining point to provide the estimated value .sigma. to be integrated, said method being a second order guidance scheme.
Patent Citations (15)
| Patent | Date | Inventor | Cited By |
|---|---|---|---|
| US3078041 | 1963-02-01 | Bomberger | |
| US3093821 | 1963-06-01 | Alpers et al. | |
| US3189300 | 1965-06-01 | Robert | |
| US3223357 | 1965-12-01 | Brucker-Steinkuhl | |
| US3316549 | 1967-04-01 | Hallendorff | |
| US3321761 | 1967-05-01 | Biagi et al. | |
| US3414215 | 1968-12-01 | Martin et al. | |
| US3575362 | 1971-04-01 | Hammond et al. | |
| US3603531 | 1971-09-01 | Brucker-Steinkuhl | |
| US3718293 | 1973-02-01 | Willems et al. | |
| US3845481 | 1974-10-01 | Danzer et al. | |
| US3911345 | 1975-10-01 | Totten | |
| US4128837 | 1978-12-01 | Page | |
| US4148029 | 1979-04-01 | Quesinberry | |
| US4179696 | 1979-12-01 | Quesinberry et al. |