US20060155506A1 - Multi-function air data probes employing neural networks for determining local air data parameters - Google Patents
Multi-function air data probes employing neural networks for determining local air data parameters Download PDFInfo
- Publication number
- US20060155506A1 US20060155506A1 US11/319,966 US31996605A US2006155506A1 US 20060155506 A1 US20060155506 A1 US 20060155506A1 US 31996605 A US31996605 A US 31996605A US 2006155506 A1 US2006155506 A1 US 2006155506A1
- Authority
- US
- United States
- Prior art keywords
- air data
- neural network
- local
- inputs
- electrical signals
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 82
- 239000000523 sample Substances 0.000 title claims abstract description 57
- 230000004044 response Effects 0.000 claims abstract description 6
- 230000003068 static effect Effects 0.000 description 14
- 238000012545 processing Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 10
- 238000000034 method Methods 0.000 description 9
- 238000012546 transfer Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000011144 upstream manufacturing Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013529 biological neural network Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000008867 communication pathway Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000013011 mating Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64D—EQUIPMENT FOR FITTING IN OR TO AIRCRAFT; FLIGHT SUITS; PARACHUTES; ARRANGEMENT OR MOUNTING OF POWER PLANTS OR PROPULSION TRANSMISSIONS IN AIRCRAFT
- B64D43/00—Arrangements or adaptations of instruments
- B64D43/02—Arrangements or adaptations of instruments for indicating aircraft speed or stalling conditions
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01P—MEASURING LINEAR OR ANGULAR SPEED, ACCELERATION, DECELERATION, OR SHOCK; INDICATING PRESENCE, ABSENCE, OR DIRECTION, OF MOVEMENT
- G01P13/00—Indicating or recording presence, absence, or direction, of movement
- G01P13/02—Indicating direction only, e.g. by weather vane
- G01P13/025—Indicating direction only, e.g. by weather vane indicating air data, i.e. flight variables of an aircraft, e.g. angle of attack, side slip, shear, yaw
Definitions
- the present invention relates generally to air data sensing systems for use on an air vehicle. More particularly, the present invention relates to multi-function air data probes employed in an air data sensing system.
- Air data systems that calculate the aerodynamic aircraft angle of attack (AOA) and angle of sideslip (AOS) of an air vehicle utilizing independent probes that are not pneumatically coupled, but which have processors for interchanging electrical signals between the probes, are known in the art. These probes are sometimes referred to as multi-function probes (MFPs). MFPs include processing circuitry located at the probe itself as part of its instrument package. With the help of the processing circuitry, MFPs are capable of converting pressure, measured by sensors included in the MFPs, into various useful local (to the probes) parameters, such as local AOA ⁇ 1 , local Mach number M 1 and local static pressure P s1 . These local parameters are utilized for the determination of aircraft AOA, aircraft AOS and other aircraft parameters including determination of altitude from static pressure or other means.
- AOA aerodynamic aircraft angle of attack
- AOS angle of sideslip
- One current technique, used with MFPs, for determining the above-mentioned local parameters includes employing a multiple table lookup and interpolation scheme. Although this technique produces relatively accurate results, it is slow, requires the use of a relatively large amount of memory and is costly to implement. Further, it demonstrates a substantial degradation in performance when there are relatively large variations in the input measured pressure. Other current techniques likewise suffer from several of the above-mentioned disadvantages. Consequently, a faster, less expensive technique for determining local air data parameters, which can produce more robust results from noisy input data, would be a significant improvement in the art.
- An air data sensing probe or MFP includes a barrel having multiple pressure sensing ports for sensing multiple pressures. Instrumentation coupled to the pressure sensing ports provides electrical signals related to the multiple pressures. A neural network, coupled to the instrumentation, receives as inputs the electrical signals related to the multiple pressures, and in response, the neural network provides, as an output, electrical signals indicative of at least one local air data parameter for the air data sensing probe.
- FIG. 1 is a top plan view of an aircraft illustrating a sideslip condition.
- FIGS. 2A and 2B are a schematic cross sectional front representation and a schematic side view, respectively, of a nose portion of an aircraft showing placement of air data sensing probes, such as multi-function probes (MFPs), in one configuration.
- MFPs multi-function probes
- FIG. 3 is a schematic representation of a cross section of a nose portion of an aircraft showing placement of the MFPs.
- FIG. 4 is a block diagram illustrating circuitry of one of the probes, with the circuitry including a neural network in embodiments of the invention.
- FIG. 5-1 is a block diagram showing the neural network of FIG. 4 with two non-dimensional inputs.
- FIG. 5-2 is diagrammatic illustration of a 3 - 3 - 3 cascade neural network used to provide local air data parameters in accordance with an embodiment of the present invention.
- FIG. 5-3 is diagrammatic illustration of a 3 - 3 - 3 feed-forward neural network used to provide local air data parameters in accordance with an embodiment of the present invention.
- FIG. 5-4 is a block diagram showing multiple neural networks employed to obtain different local air data parameters in accordance with an embodiment of the present invention.
- FIG. 6 is a diagrammatic illustration of a node or threshold gate of the neural network.
- an aircraft indicated generally at 10 is shown with a nose portion 12 .
- Mounted air data sensing probes or MFPs are indicated schematically and generally at 14 , 16 and 18 .
- the positioning of aircraft 10 is with respect to a center plane or center line 20 that is perpendicular to the normal plane of the wings 21 .
- Center line 20 is shown in a greatly exaggerated sideslip condition where the path of travel of aircraft 10 is indicated at 22 , and where there is a substantial angle ⁇ between the path of travel line 22 and the line or plane 20 .
- Angle ⁇ is the aircraft AOS, which is defined herein as an aircraft parameter.
- Other aircraft parameters or air data parameters include aircraft static pressure and Mach number, for example.
- the aircraft essentially is yawing to the left in FIG. 1 . As the airflow passes across the aircraft, the probes 14 and 18 will be subjected to different local flow conditions, insofar as the local angle of wind and local static pressure is concerned, than is the probe 16 .
- Another aircraft parameter is the aircraft AOA. It also can be envisioned that if the aircraft changes angle of attack, because of the yaw angle, the amount of change in pressure on one side of the aircraft would be different than that on the other side of the aircraft. When the probes are only electrically connected together, there is no way of simply averaging these changes to obtain a correct static pressure reading. An angle of sideslip value can be used as a correction factor to compensate for this angle.
- the above-mentioned aircraft parameters are determined from local air data parameters such as local AOA ( ⁇ 1 ), local Mach number (M 1 ), local static pressure (P s1 ), and Pitot pressure (P t ) corrected for the effects of airspeed and local AOA which are typically computed for each of the MFPs 14 , 16 and 18 .
- local air data parameters such as local AOA ( ⁇ 1 ), local Mach number (M 1 ), local static pressure (P s1 ), and Pitot pressure (P t ) corrected for the effects of airspeed and local AOA which are typically computed for each of the MFPs 14 , 16 and 18 .
- FIGS. 2A and 2B diagrammatically illustrate an arrangement of the MFPs 14 , 16 and 18 shown in FIG. 1 , from a front view and side view of the aircraft 10 for one example embodiment.
- These probes measure Pitot pressure P t , and angle of attack pressures comprising P ⁇ 1 and P ⁇ 2 , and calculate local parameters such as local AOA ⁇ 1 , local Mach number M 1 , local static pressure P s1 and Pitot pressure corrected for the effects of airspeed and local AOA.
- probes 14 and 16 are symmetrically located, and are on the same “water line” or vertical height on the aircraft, so that they are in identical positions.
- Probe 18 is mounted on the same side of the aircraft as probe 14 .
- This arrangement of MFPs helps provide for redundancy in correcting for sideslip of an aircraft arriving at air data parameters, such as aircraft AOA, static pressure or pressure attitude, and Mach number. Advantages of such a three-probe air data system and details regarding the compensation for aircraft AOS in such a system are set forth in U.S. patent application Ser. No.
- An MFP or an air data probe is defined herein as a probe in which a computer or processing circuitry is integrally attached to the probe, and the sensing portions of the probe itself protrude into the airstream that moves past the aircraft skin.
- the processing circuitry is in an integrated housing immediately inside the skin.
- the outputs from the probes can include digital electrical signals representing pressures derived from sampling the pressure readings from internal pressure sensors.
- the internal pressure sensors of the probes can be either differential sensors or absolute sensors that receive the pressure signals and convert them into electrical signals that are then digitized.
- the processing circuitry of an MFP includes a neural network trained or configured to provide local air data parameters which are utilized for determining aircraft parameters, for example.
- the probes 14 , 16 and 18 may be probes similar to those shown in the U.S. Pat. No. 4,378,696, or other similar patents. As shown schematically in FIG. 3 , the probes have barrels 14 A, 16 A and 18 A, with suitable ports for sensing Pitot pressure (or total pressure) P t at leading ends of the probes indicated at 14 B, 16 B and 18 B. The angle of attack sensing ports are placed on the top and bottom of the probes, and the top ports are illustrated schematically at 14 C, 16 C and 18 C. Mating ports ( 14 E, 16 E and 18 E not shown in FIG.
- the lower angle of attack pressure port on the probes will sense a pressure designated Pal, and the upper angle of attack pressure ports shown at 14 C, 16 C and 18 C sense a pressure designated P ⁇ 2 .
- Each of the probes 14 , 16 and 18 is provided with a separate instrument housing 14 D, 16 D and 18 D, respectively.
- the electronics within instrument housings 14 D, 16 D and 18 D are electrically coupled via electrical connection 24 .
- Electrical connection 24 can be a data bus, electrical cables, or other types of communication pathways for communicating information in the form of electrical signals.
- the neural networks which are trained to provide local air data parameters, are implemented within the processing circuitry contained within the housings 14 D, 16 D and 18 D associated with probes 14 , 16 and 18 .
- an optional remote (air data computer) ADC 38 within which the neural networks used in the present invention can alternatively be implemented to calculate the local parameters for one or more of MFPs 14 , 16 and 18 .
- circuitry 23 is shown schematically coupled to one of probes ( 14 , 16 or 18 ) which are labeled as multi-function probes. Circuitry 23 is positioned within the corresponding housing 14 D, 16 D or 18 D, and in the illustrated embodiment includes a neural network 37 .
- Inputs are illustrated schematically as the ⁇ 1 pressure port 14 E, 16 E or 18 E, the ⁇ 2 pressure port 14 C, 16 C or 18 C, and the total pressure port 14 B, 16 B or 18 B.
- the ⁇ 2 pressure port and total pressure port correspond to those shown in FIG. 3 for probes 14 , 16 and 18 .
- the ⁇ 1 pressure port corresponds to the lower port (not shown in FIG. 3 ) on the probes.
- pressures are transmitted through a manifold 30 to individual pressure sensors 32 , 33 and 34 , respectively.
- the outputs of pressure sensors 32 , 33 and 34 which may be absolute pressure sensors or differential pressure sensors, are provided to analog-to-digital converter 35 for digitization. Electrical signals corresponding to pressures P ⁇ 1 , P ⁇ 2 and P t , are output from analog-to-digital converter 35 .
- quantities related to the output from digitization circuitry 35 are input to neural network 37 , which is configured to accurately map the input quantities into desired output quantities (nondimensional local air data parameters P LOCAL ).
- processing circuitry 36 can include a microcontroller and related memory and communications circuits all forming part of the MFP circuitry 23 .
- Non-dimensionalization of the three pressure values (P ⁇ 1,m , P ⁇ 2,m and P t,m ) is carried out within processing circuitry 36 , to reduce the pressure values to independent parameters I ND1 and I ND2 .
- Non-dimensional quantities I ND1 and I ND2 can then serve as inputs to neural network 37 .
- Neural network 37 outputs nondimensional local air data parameters P LOCAL 38 as a function of quantities I ND1 and I ND2 .
- neural network 37 can output at 38 local air data parameters P LOCAL that include local AOA ⁇ 1 , local Mach number M 1 and
- P sl P sl [ 1 2 ⁇ ( P ⁇ ⁇ ⁇ 1 , m + P ⁇ ⁇ ⁇ 2 , m ) ] Equation ⁇ ⁇ 3
- P s1 can be easily obtained from Equation 3.
- the computation of additional local parameters, such as P s1 is carried out by post processing circuitry 39 .
- pressures values P ⁇ 1,m , P ⁇ 2,m and P t,m are input without non-dimensionalization into the neural network 37 .
- neural network 37 outputs local AOA ⁇ 1 , local Mach number M 1 and/or local static pressure P s1 for the corresponding MFP.
- processing circuitry 36 does not carry out the computation of non-dimensional input quantities I ND1 and I ND2 , and is therefore optional.
- neural network 37 can be implemented as a computer program that is stored and executed in processing circuitry 36 .
- the MFP circuitry can also include other circuits and devices such as electromagnetic filters for reducing EMI interference, heater control circuits, and power supplies.
- Circuitry 23 also includes, and/or is coupled to, data bus 24 for receiving information from, or transmitting information to, other probes or devices in the system.
- Neural network 37 can be any of a number of known neural network architectures.
- neural network 37 can be a feed-forward architecture, a cascade architecture, a three-layer architecture, or other types of neural networks.
- the neural network architecture is first chosen with the required number of inputs and outputs (for example two inputs ⁇ 2 and ⁇ 3 and three outputs ⁇ 1 , M 1 and ⁇ 4 ).
- Neural network 37 is next trained using a data set of inputs and ideal target outputs spanning the expected operating range of values. This data set can be, for example, experimental data correlating various inputs to their corresponding outputs. The experimental data can be obtained using wind tunnel testing, flight testing, or by other methods.
- neural network 37 is a deterministic algorithm consisting of weighted sums and transfer functions interconnected in series and/or parallel. It can then be used to map the probe inputs to the desired outputs.
- FIG. 5-1 is a block diagram showing neural network 37 receiving two non-dimensional inputs I ND1 and I ND2 as described above, and providing local air data parameters 38 as outputs. Also included in FIG. 5-1 , is an optional feedback path 40 . In embodiments of the present invention that employ static neural networks, feedback path 40 is excluded. However, embodiments that employ adaptive neural networks include feedback path 40 . Also, although shown externally to neural network 37 , feedback path 40 can be considered to be internal to the neural network since neural networks are frequently implemented in a processor environment. Details of different embodiments of neural network 37 are provided further below in connection with FIGS. 5-2 through 5 - 4 .
- separate neural networks can be developed based on training data for a number of appropriate local air data parameters including, but not limited to local AOA ⁇ 1 , local Mach number M 1 and a nondimensional parameter related to local static pressure P s1 .
- Artificial neural networks are computer programs that simulate biological neural networks. Artificial neural networks such as neural network 37 are unlike artificial intelligence programs in that, instead of using deductive reasoning to apply known rules to situations to produce outputs, artificial neural networks automatically construct associations based upon the results of known situations (i.e., combinations of inputs and outputs). In training, neural networks such as neural network 37 automatically adjust themselves and eventually generalize associations.
- neural network 37 Once neural network 37 is constructed and trained (by being presented with historical cause and effect situations) in a wind tunnel simulation, using other simulated data, or flight test data, the network 37 then shapes itself to build an internal representation of the governing rules. Later, after artificial neural network 37 is trained and is in use in one of the air data probes, it can be presented with various combinations of inputs to produce a real event result. In other words, neural network 37 will be trained to produce particular local air data parameter values at output(s) 38 in response to specific combinations of Pitot pressure P t , and angle of attack pressures comprising P ⁇ 1 and P ⁇ 2 and/or dimensional and non-dimensional form. Shown in FIG.
- 5-2 is a 3 - 3 - 3 neural network (i.e., a three-layered neural network with three nodes in the first, second, and third layers) used to provide local air data parameters 38 - 1 , 38 - 2 and 38 - 3 .
- the illustration is not intended to limit the invention to any particular type of neural network, nor to the number of inputs or layers of the neural network.
- inputs I ND1 and I ND2 to neural network 37 , are provided to input nodes 51 and 52 , respectively.
- the outputs of each of the input nodes is then provided to each of threshold gates or nodes 71 , 72 , and 73 in first network layer 70 of neural network 37 .
- each of these threshold gates are provided to each of threshold gates 81 , 82 and 83 in second network layer 80 of network 37 . Further, outputs from each of the threshold gates of second network layer 80 are provided to each of gates 91 , 92 and 93 of third network layer 90 , which in turn provides nondimensional local air data parameters 38 - 1 , 38 - 2 and 38 - 3 as outputs.
- a bias (BIAS 1 ) is provided to each of the threshold gates of first network layer 70 from bias node 61 .
- biases BIAS 2 and BIAS 3 are provided to each of the threshold gates of the second and third network layers 80 and 90 from bias nodes 62 and 63 , respectively.
- Different types of transfer functions can be employed at each threshold gate to provide appropriate gate outputs. For example, a hyperbolic tangent (tanh) transfer function, which is a “squashing” function that takes inputs ranging from ⁇ to + ⁇ and outputs a corresponding value between ⁇ 1 and +1, can be applied to the inputs at each threshold gate of layers 70 and 80 , to produce corresponding intermediate gate outputs. Further, a linear transfer function, which simply outputs the input value, can be applied to the inputs at each threshold gate of third layer 90 , to produce local air data parameters 38 - 1 , 38 - 2 and 38 - 3 .
- FIG. 5-3 shows a 3 - 3 - 3 neural network that is substantially similar to the neural network of FIG. 5-2 .
- the same reference numerals are used to represent the same or similar elements in FIGS. 5-2 and 5 - 3 .
- the example neural network shown in FIG. 5-2 is a cascade neural network (i.e., the upstream nodes influence multiple network layers). Therefore, in the embodiment of FIG. 5-2 , upstream nodes 51 and 52 influence network layers 70 , 80 and 90 and nodes 71 , 72 and 73 influence network layers 80 and 90 .
- the neural network shown in FIG. 5-3 is merely a feed forward neural network (i.e., the upstream nodes do not influence multiple network layers). Therefore, in the embodiment of FIG. 5-3 , upstream nodes 51 and 52 only influence network layer 70 , nodes 71 , 72 and 73 only influence network layer 80 , etc.
- the neural networks of FIG. 5-2 and 5 - 3 are substantially similar in structure and operation.
- a single neural network is employed to obtain multiple local air data parameters 38 - 1 , 38 - 2 and 38 - 3 .
- first neural network 102 outputs local air data parameter 38 - 1
- second neural network 104 outputs air data local parameter 38 - 2
- third neural network 106 outputs local air data parameter 38 - 3 .
- FIG. 6 illustrates a plurality of separate weights (W 1 , W 2 , . . . , W N and W B ) applied to a node or threshold gate 108 of a neural network (such as 37 ). Also illustrated is a bias (BIAS) which is associated with gate or node 108 .
- BIAS bias
- each node in the neural network can have its own unique bias as well as its own unique set of weights applied to its inputs.
- the gate output is determined as a function of the inputs (I 1 , I 2 , . . I N ), the weights, the bias, and the transfer function.
- an arbitrary number of inputs (I 1 , I 2 , . . . I N ) and a Bias are each multiplied by corresponding weights (W 1 , W 2 , . . . , W N and W B ) and summed. This sum is then passed through a transfer (in this case a hyperbolic tangent) function, to produce the gate output.
- the inputs can be from the neural network inputs and/or outputs of other gates.
- the present invention includes the use of a neural network in this manner in other types of MFPs. Therefore, the phrases “air data probes” and “MFPs” should also be interpreted to include any type of MFP capable of delivering local Mach number, local angle of attack, and/or local static pressure.
- FIG. 5-1 through 5 - 4 illustrate neural networks with two non-dimensional inputs I 1 and I 2
- the invention includes MFPs utilizing a neural network having an appropriate number of dimensional inputs, for example P ⁇ 1,m , P ⁇ 2,m and P t,m .
- the number and type of dimensional or non-dimensional inputs and outputs can be changed without departing from the spirit and scope of the invention.
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measuring Fluid Pressure (AREA)
- Feedback Control In General (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
An air data sensing probe or MFP includes a barrel having multiple pressure sensing ports for sensing multiple pressures. Instrumentation coupled to the pressure sensing ports provides electrical signals related to the multiple pressures. A neural network, coupled to the instrumentation, receives as inputs the electrical signals related to the multiple pressures, and in response, the neural network provides, as an output, electrical signals indicative of at least one local air data parameter for the air data sensing probe.
Description
- The present application is a Divisional Application of, and claims priority from, U.S. patent application Ser. No. 10/328,487, filed Dec. 23, 2002.
- The present invention relates generally to air data sensing systems for use on an air vehicle. More particularly, the present invention relates to multi-function air data probes employed in an air data sensing system.
- Air data systems that calculate the aerodynamic aircraft angle of attack (AOA) and angle of sideslip (AOS) of an air vehicle utilizing independent probes that are not pneumatically coupled, but which have processors for interchanging electrical signals between the probes, are known in the art. These probes are sometimes referred to as multi-function probes (MFPs). MFPs include processing circuitry located at the probe itself as part of its instrument package. With the help of the processing circuitry, MFPs are capable of converting pressure, measured by sensors included in the MFPs, into various useful local (to the probes) parameters, such as local AOA α1, local Mach number M1 and local static pressure Ps1. These local parameters are utilized for the determination of aircraft AOA, aircraft AOS and other aircraft parameters including determination of altitude from static pressure or other means.
- One current technique, used with MFPs, for determining the above-mentioned local parameters includes employing a multiple table lookup and interpolation scheme. Although this technique produces relatively accurate results, it is slow, requires the use of a relatively large amount of memory and is costly to implement. Further, it demonstrates a substantial degradation in performance when there are relatively large variations in the input measured pressure. Other current techniques likewise suffer from several of the above-mentioned disadvantages. Consequently, a faster, less expensive technique for determining local air data parameters, which can produce more robust results from noisy input data, would be a significant improvement in the art.
- An air data sensing probe or MFP includes a barrel having multiple pressure sensing ports for sensing multiple pressures. Instrumentation coupled to the pressure sensing ports provides electrical signals related to the multiple pressures. A neural network, coupled to the instrumentation, receives as inputs the electrical signals related to the multiple pressures, and in response, the neural network provides, as an output, electrical signals indicative of at least one local air data parameter for the air data sensing probe.
-
FIG. 1 is a top plan view of an aircraft illustrating a sideslip condition. -
FIGS. 2A and 2B are a schematic cross sectional front representation and a schematic side view, respectively, of a nose portion of an aircraft showing placement of air data sensing probes, such as multi-function probes (MFPs), in one configuration. -
FIG. 3 is a schematic representation of a cross section of a nose portion of an aircraft showing placement of the MFPs. -
FIG. 4 is a block diagram illustrating circuitry of one of the probes, with the circuitry including a neural network in embodiments of the invention. -
FIG. 5-1 is a block diagram showing the neural network ofFIG. 4 with two non-dimensional inputs. -
FIG. 5-2 is diagrammatic illustration of a 3-3-3 cascade neural network used to provide local air data parameters in accordance with an embodiment of the present invention. -
FIG. 5-3 is diagrammatic illustration of a 3-3-3 feed-forward neural network used to provide local air data parameters in accordance with an embodiment of the present invention. -
FIG. 5-4 is a block diagram showing multiple neural networks employed to obtain different local air data parameters in accordance with an embodiment of the present invention. -
FIG. 6 is a diagrammatic illustration of a node or threshold gate of the neural network. - Referring to
FIG. 1 , an aircraft indicated generally at 10 is shown with anose portion 12. Mounted air data sensing probes or MFPs are indicated schematically and generally at 14, 16 and 18. The positioning ofaircraft 10 is with respect to a center plane orcenter line 20 that is perpendicular to the normal plane of thewings 21.Center line 20 is shown in a greatly exaggerated sideslip condition where the path of travel ofaircraft 10 is indicated at 22, and where there is a substantial angle β between the path oftravel line 22 and the line orplane 20. Angle β is the aircraft AOS, which is defined herein as an aircraft parameter. Other aircraft parameters or air data parameters include aircraft static pressure and Mach number, for example. The aircraft essentially is yawing to the left inFIG. 1 . As the airflow passes across the aircraft, theprobes probe 16. - Another aircraft parameter is the aircraft AOA. It also can be envisioned that if the aircraft changes angle of attack, because of the yaw angle, the amount of change in pressure on one side of the aircraft would be different than that on the other side of the aircraft. When the probes are only electrically connected together, there is no way of simply averaging these changes to obtain a correct static pressure reading. An angle of sideslip value can be used as a correction factor to compensate for this angle.
- The above-mentioned aircraft parameters are determined from local air data parameters such as local AOA (α1), local Mach number (M1), local static pressure (Ps1), and Pitot pressure (Pt) corrected for the effects of airspeed and local AOA which are typically computed for each of the
MFPs FIG. 1 , circuitry integrally attached to the MFPs and the determination of local air data parameters are described below in connection with FIGS. 2 to 6. It must be noted that the present invention is not limited to three MFP systems, but can instead be used in systems utilizing any number of MFPS. -
FIGS. 2A and 2B diagrammatically illustrate an arrangement of theMFPs FIG. 1 , from a front view and side view of theaircraft 10 for one example embodiment. These probes measure Pitot pressure Pt, and angle of attack pressures comprising Pα1 and Pα2, and calculate local parameters such as local AOA α1, local Mach number M1, local static pressure Ps1 and Pitot pressure corrected for the effects of airspeed and local AOA. - As can be seen in
FIG. 2A ,probes Probe 18 is mounted on the same side of the aircraft asprobe 14. This arrangement of MFPs helps provide for redundancy in correcting for sideslip of an aircraft arriving at air data parameters, such as aircraft AOA, static pressure or pressure attitude, and Mach number. Advantages of such a three-probe air data system and details regarding the compensation for aircraft AOS in such a system are set forth in U.S. patent application Ser. No. 09/851,289, filed May 8, 2001 and entitled “MULTI-FUNCTION AIR DATA PROBES USING NEURAL NETWORK FOR SIDESLIP COMPENSATION,” U.S. patent application Ser. No. 09/851,485, filed May 8, 2001 and entitled “ITERATIVE METHOD OF AIRCRAFT SIDESLIP COMPENSATION FOR MULTI-FUNCTION PROBE AIR DATA SYSTEMS,” U.S. patent application Ser. No. 09/850,863, filed May 8, 2001 and entitled “METHOD TO CALCULATE SIDESLIP ANGLE AND CORRECT STATIC PRESSURE FOR SIDESLIP EFFECTS USING INERTIAL INFORMATION,” and U.S. patent application Ser. No. 09/850,796, filed May 8, 2001 and entitled “SIDESLIP CORRECTION FOR A MULTI-FUNCTION THREE PROBE AIR DATA SYSTEM,” which are incorporated herein by reference in their entirety. However, the present invention is not limited to systems employing three MFPs, but instead can be used in systems employing a greater or lesser number. - An MFP or an air data probe is defined herein as a probe in which a computer or processing circuitry is integrally attached to the probe, and the sensing portions of the probe itself protrude into the airstream that moves past the aircraft skin. The processing circuitry is in an integrated housing immediately inside the skin. The outputs from the probes can include digital electrical signals representing pressures derived from sampling the pressure readings from internal pressure sensors. The internal pressure sensors of the probes can be either differential sensors or absolute sensors that receive the pressure signals and convert them into electrical signals that are then digitized. In accordance with the invention, the processing circuitry of an MFP (or other similar air data sensing probe) includes a neural network trained or configured to provide local air data parameters which are utilized for determining aircraft parameters, for example. The
probes FIG. 3 , the probes havebarrels FIG. 3 ) are on the lower portion, as is well known, for measuring the angle of attack by sensing the differentials in pressure between the upper port and the lower port relative to a change in local AOA. The lower angle of attack pressure port on the probes will sense a pressure designated Pal, and the upper angle of attack pressure ports shown at 14C, 16C and 18C sense a pressure designated Pα2. Each of theprobes separate instrument housing - As shown in
FIG. 3 , the electronics withininstrument housings electrical connection 24.Electrical connection 24 can be a data bus, electrical cables, or other types of communication pathways for communicating information in the form of electrical signals. In embodiments of the present invention, the neural networks, which are trained to provide local air data parameters, are implemented within the processing circuitry contained within thehousings probes FIG. 3 is an optional remote (air data computer)ADC 38 within which the neural networks used in the present invention can alternatively be implemented to calculate the local parameters for one or more ofMFPs - The arrangement of the circuitry in one example embodiment of an air data sensing probe of the invention is shown in block diagram form in
FIG. 4 . Since the circuitry integrally attached to each probe is substantially similar in some embodiments of the present invention, the circuitry shown can apply to any one ofprobes Circuitry 23 is shown schematically coupled to one of probes (14, 16 or 18) which are labeled as multi-function probes.Circuitry 23 is positioned within the correspondinghousing neural network 37. Inputs are illustrated schematically as the α1 pressure port 14E, 16E or 18E, the α2 pressure port 14C, 16C or 18C, and thetotal pressure port FIG. 3 forprobes FIG. 3 ) on the probes. - These pressures are transmitted through a manifold 30 to
individual pressure sensors pressure sensors digital converter 35 for digitization. Electrical signals corresponding to pressures Pα1, Pα2 and Pt, are output from analog-to-digital converter 35. In accordance with the present invention, quantities related to the output fromdigitization circuitry 35, either in dimensional or non-dimensional form, are input toneural network 37, which is configured to accurately map the input quantities into desired output quantities (nondimensional local air data parameters PLOCAL). - In embodiments of the present invention that utilize non-dimensional quantities as inputs to
neural network 37, electrical signals corresponding to pressures Pα1, Pα2 and Pt, which are output fromdigitization circuitry 35, are provided toprocessing circuitry 36 resulting in measured pressures Pα1,m, Pα2,m and Pt,m. Processing circuitry 36 can include a microcontroller and related memory and communications circuits all forming part of theMFP circuitry 23. Non-dimensionalization of the three pressure values (Pα1,m, Pα2,m and Pt,m) is carried out withinprocessing circuitry 36, to reduce the pressure values to independent parameters IND1 and IND2. Non-dimensional quantities IND1 and IND2 can then serve as inputs toneural network 37.Neural network 37 outputs nondimensional local airdata parameters P LOCAL 38 as a function of quantities IND1 and IND2. - In one embodiment of the present invention, quantities IND1 and IND2 are equal to parameters π2 and π3, respectively, which are computed as:
With π2 and π3 as inputs,neural network 37 can output at 38 local air data parameters PLOCAL that include local AOA α1, local Mach number M1 and a non-dimensional quantity π4 in one example embodiment. Output quantity π4 can be related to the local static pressure Ps1 according to:
Thus, the value of Ps1 can be easily obtained fromEquation 3. The computation of additional local parameters, such as Ps1, is carried out bypost processing circuitry 39. - In embodiments of the present invention that utilize dimensional quantities as inputs to
neural network 37, pressures values Pα1,m, Pα2,m and Pt,m, output bydigitization circuitry 35, are input without non-dimensionalization into theneural network 37. In response,neural network 37 outputs local AOA α1, local Mach number M1 and/or local static pressure Ps1 for the corresponding MFP. Thus, in such embodiments, processingcircuitry 36 does not carry out the computation of non-dimensional input quantities IND1 and IND2, and is therefore optional. However, when dimensional input quantities are utilized, a minimum of three inputs (Pα1,m, Pα2,m and Pt,m) toneural network 37 are required, instead of a minimum requirement of two inputs (IND1 and IND2) when non-dimensional quantities are utilized. - Although
neural network 37 is shown as a separate block inFIG. 4 ,neural network 37 can be implemented as a computer program that is stored and executed inprocessing circuitry 36. Further, the MFP circuitry can also include other circuits and devices such as electromagnetic filters for reducing EMI interference, heater control circuits, and power supplies.Circuitry 23 also includes, and/or is coupled to,data bus 24 for receiving information from, or transmitting information to, other probes or devices in the system. -
Neural network 37 can be any of a number of known neural network architectures. For example,neural network 37 can be a feed-forward architecture, a cascade architecture, a three-layer architecture, or other types of neural networks. Typically, the neural network architecture is first chosen with the required number of inputs and outputs (for example two inputs π2 and π3 and three outputs α1, M1 and π4).Neural network 37 is next trained using a data set of inputs and ideal target outputs spanning the expected operating range of values. This data set can be, for example, experimental data correlating various inputs to their corresponding outputs. The experimental data can be obtained using wind tunnel testing, flight testing, or by other methods. Once trained,neural network 37 is a deterministic algorithm consisting of weighted sums and transfer functions interconnected in series and/or parallel. It can then be used to map the probe inputs to the desired outputs. -
FIG. 5-1 is a block diagram showingneural network 37 receiving two non-dimensional inputs IND1 and IND2 as described above, and providing localair data parameters 38 as outputs. Also included inFIG. 5-1 , is anoptional feedback path 40. In embodiments of the present invention that employ static neural networks,feedback path 40 is excluded. However, embodiments that employ adaptive neural networks includefeedback path 40. Also, although shown externally toneural network 37,feedback path 40 can be considered to be internal to the neural network since neural networks are frequently implemented in a processor environment. Details of different embodiments ofneural network 37 are provided further below in connection withFIGS. 5-2 through 5-4. - In general, separate neural networks can be developed based on training data for a number of appropriate local air data parameters including, but not limited to local AOA α1, local Mach number M1 and a nondimensional parameter related to local static pressure Ps1. Artificial neural networks are computer programs that simulate biological neural networks. Artificial neural networks such as
neural network 37 are unlike artificial intelligence programs in that, instead of using deductive reasoning to apply known rules to situations to produce outputs, artificial neural networks automatically construct associations based upon the results of known situations (i.e., combinations of inputs and outputs). In training, neural networks such asneural network 37 automatically adjust themselves and eventually generalize associations. - Once
neural network 37 is constructed and trained (by being presented with historical cause and effect situations) in a wind tunnel simulation, using other simulated data, or flight test data, thenetwork 37 then shapes itself to build an internal representation of the governing rules. Later, after artificialneural network 37 is trained and is in use in one of the air data probes, it can be presented with various combinations of inputs to produce a real event result. In other words,neural network 37 will be trained to produce particular local air data parameter values at output(s) 38 in response to specific combinations of Pitot pressure Pt, and angle of attack pressures comprising Pα1 and Pα2 and/or dimensional and non-dimensional form. Shown inFIG. 5-2 is a 3-3-3 neural network (i.e., a three-layered neural network with three nodes in the first, second, and third layers) used to provide local air data parameters 38-1, 38-2 and 38-3. The illustration is not intended to limit the invention to any particular type of neural network, nor to the number of inputs or layers of the neural network. As illustrated inFIG. 5-2 , inputs IND1 and IND2, toneural network 37, are provided to inputnodes nodes neural network 37. In turn, the outputs of each of these threshold gates are provided to each ofthreshold gates second network layer 80 ofnetwork 37. Further, outputs from each of the threshold gates ofsecond network layer 80 are provided to each ofgates third network layer 90, which in turn provides nondimensional local air data parameters 38-1, 38-2 and 38-3 as outputs. - A bias (BIAS1) is provided to each of the threshold gates of first network layer 70 from
bias node 61. Similarly, biases BIAS2 and BIAS3 are provided to each of the threshold gates of the second and third network layers 80 and 90 frombias nodes layers 70 and 80, to produce corresponding intermediate gate outputs. Further, a linear transfer function, which simply outputs the input value, can be applied to the inputs at each threshold gate ofthird layer 90, to produce local air data parameters 38-1, 38-2 and 38-3. -
FIG. 5-3 shows a 3-3-3 neural network that is substantially similar to the neural network ofFIG. 5-2 . The same reference numerals are used to represent the same or similar elements inFIGS. 5-2 and 5-3. The example neural network shown inFIG. 5-2 , and described above, is a cascade neural network (i.e., the upstream nodes influence multiple network layers). Therefore, in the embodiment ofFIG. 5-2 ,upstream nodes nodes FIG. 5-3 is merely a feed forward neural network (i.e., the upstream nodes do not influence multiple network layers). Therefore, in the embodiment ofFIG. 5-3 ,upstream nodes nodes influence network layer 80, etc. In other respects, the neural networks ofFIG. 5-2 and 5-3 are substantially similar in structure and operation. - In the example embodiments of the present invention described above in connection with
FIGS. 5-1 and 5-2, a single neural network is employed to obtain multiple local air data parameters 38-1, 38-2 and 38-3. - In the embodiment shown in
FIG. 5-4 , multiple neural networks are employed, with each neural network of the multiple neural networks outputting a different local air data parameter as a function of inputs IND1 and IND2. Thus, firstneural network 102 outputs local air data parameter 38-1, secondneural network 104 outputs air data local parameter 38-2 and thirdneural network 106 outputs local air data parameter 38-3. - There are a wide variety of training schemes, which are aimed at minimizing the neural network output error. One such technique suitable for use with the networks of
FIGS. 5-1 through 5-4 is the Levenberg-Marquardt back-propagation scheme. In general, this training technique involves recursively nudging the network weight values toward a solution that minimizes the mean absolute error between the network output and the target output. -
FIG. 6 illustrates a plurality of separate weights (W1, W2, . . . , WN and WB) applied to a node orthreshold gate 108 of a neural network (such as 37). Also illustrated is a bias (BIAS) which is associated with gate ornode 108. In general, each node in the neural network can have its own unique bias as well as its own unique set of weights applied to its inputs. The gate output is determined as a function of the inputs (I1, I2, . . IN), the weights, the bias, and the transfer function. - In one embodiment of the present invention, the gate output is determined by the relation:
Gate Output=tanih(Bias×W B +Σ[I K ×W K]) Equation 4
Here, an arbitrary number of inputs (I1, I2, . . . IN) and a Bias are each multiplied by corresponding weights (W1, W2, . . . , WN and WB) and summed. This sum is then passed through a transfer (in this case a hyperbolic tangent) function, to produce the gate output. The inputs can be from the neural network inputs and/or outputs of other gates. In one embodiment employing Equation 4 to determine the gate output, a single Bias with a constant value of 1 can input to each gate, such that Bias×WB=WB and represents an arbitrary constant to each gate. - Although the present invention has been described with reference to preferred embodiments, workers skilled in the art will recognize that changes may be made in form and detail without departing from the spirit and scope of the invention. For example, the present invention includes the use of a neural network in this manner in other types of MFPs. Therefore, the phrases “air data probes” and “MFPs” should also be interpreted to include any type of MFP capable of delivering local Mach number, local angle of attack, and/or local static pressure. Further, while
FIG. 5-1 through 5-4 illustrate neural networks with two non-dimensional inputs I1 and I2, the invention includes MFPs utilizing a neural network having an appropriate number of dimensional inputs, for example Pα1,m, Pα2,m and Pt,m. The number and type of dimensional or non-dimensional inputs and outputs can be changed without departing from the spirit and scope of the invention.
Claims (2)
1. An air data sensing probe comprising:
a plurality of pressure sensing ports sensing a plurality of pressures;
instrumentation coupled to the plurality of pressure sensing ports providing electrical signals related to the plurality of pressures; and
a neural network coupled to the instrumentation, the neural network configured to receive as inputs the electrical signals related to the plurality of pressures, and in response, the neural network configured to provide as an output electrical signals indicative of at least one local air data parameter for the air data sensing probe.
2. An air data sensing probe comprising:
pressure sensing means for sensing a plurality of pressures;
electrical signal generating means for providing electrical signals related to the plurality of pressures; and
neural network means coupled to the electrical signal generating means for receiving as inputs the electrical signals related to the plurality of pressures, and for providing in response electrical signals indicative of at least one local air data parameter for the air data sensing probe.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/319,966 US20060155506A1 (en) | 2002-12-23 | 2005-12-28 | Multi-function air data probes employing neural networks for determining local air data parameters |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/328,487 US7379839B2 (en) | 2002-12-23 | 2002-12-23 | Multi-function air data probes employing neural networks for determining local air data parameters |
US11/319,966 US20060155506A1 (en) | 2002-12-23 | 2005-12-28 | Multi-function air data probes employing neural networks for determining local air data parameters |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/328,487 Division US7379839B2 (en) | 2002-12-23 | 2002-12-23 | Multi-function air data probes employing neural networks for determining local air data parameters |
Publications (1)
Publication Number | Publication Date |
---|---|
US20060155506A1 true US20060155506A1 (en) | 2006-07-13 |
Family
ID=32507342
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/328,487 Expired - Lifetime US7379839B2 (en) | 2002-12-23 | 2002-12-23 | Multi-function air data probes employing neural networks for determining local air data parameters |
US11/319,966 Abandoned US20060155506A1 (en) | 2002-12-23 | 2005-12-28 | Multi-function air data probes employing neural networks for determining local air data parameters |
Family Applications Before (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/328,487 Expired - Lifetime US7379839B2 (en) | 2002-12-23 | 2002-12-23 | Multi-function air data probes employing neural networks for determining local air data parameters |
Country Status (2)
Country | Link |
---|---|
US (2) | US7379839B2 (en) |
EP (1) | EP1437599B1 (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040122615A1 (en) * | 2002-12-23 | 2004-06-24 | Cronin Dennis J. | Multi-function air data probes employing neural networks for determining local air data parameters |
US20060212181A1 (en) * | 2005-03-16 | 2006-09-21 | Rosemount Aerospace Inc. | Method and apparatus for extending useful range of air data parameter calculation in flush air data systems |
US20060225496A1 (en) * | 2005-04-06 | 2006-10-12 | Rosemount Aerospace Inc. | Method and apparatus for obtaining improved accuracy and range for air data parameters inferred from independent measurements of interdependent pressures |
US20070107510A1 (en) * | 2005-10-24 | 2007-05-17 | Mark Agami | Multi-function air data sensor |
US20070130096A1 (en) * | 2005-12-01 | 2007-06-07 | Rosemount Aerospace, Inc. | Fault detection in artificial intelligence based air data systems |
US20070150122A1 (en) * | 2004-10-29 | 2007-06-28 | Rosemount Aerospace Inc. | Fault isolation method and apparatus in artificial intelligence based air data systems |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8620495B2 (en) * | 2006-12-19 | 2013-12-31 | Rosemount Aerospace Inc. | Air data stall protection system |
US8479581B2 (en) | 2011-05-03 | 2013-07-09 | General Electric Company | Device and method for measuring pressure on wind turbine components |
FR2983965B1 (en) * | 2011-12-12 | 2014-07-04 | Thales Sa | LOCAL IMPACT MEASUREMENT PROBE AND METHOD USING THE SENSOR |
US8949090B2 (en) * | 2013-01-28 | 2015-02-03 | The Boeing Company | Formation flight control |
US9297714B2 (en) * | 2013-03-05 | 2016-03-29 | Rosemount Aerospace Inc. | Air data probes |
US9341533B2 (en) | 2013-03-05 | 2016-05-17 | Rosemount Aerospace Inc. | Air data probes |
FR3008073B1 (en) * | 2013-07-04 | 2015-08-07 | Thales Sa | AIRCRAFT COMPRISING A MEASURING PROBE AND METHOD FOR DETERMINING FLIGHT PARAMETERS OF SUCH AN AIRCRAFT |
US9821903B2 (en) | 2014-07-14 | 2017-11-21 | The Boeing Company | Closed loop control of aircraft control surfaces |
CN104568295A (en) * | 2014-12-08 | 2015-04-29 | 太原航空仪表有限公司 | Monitoring and processing method for pressure faults of embedded air data system |
US10913545B2 (en) | 2018-06-15 | 2021-02-09 | Rosemount Aerospace Inc. | Architecture for providing enhanced altitude functionality to aircraft air data system |
US11015955B2 (en) | 2018-06-15 | 2021-05-25 | Rosemount Aerospace Inc. | Dual channel air data system with inertially compensated backup channel |
US10852316B2 (en) | 2018-06-15 | 2020-12-01 | Rosemount Aerospace Inc. | Advanced air data system architecture with air data computer incorporating enhanced compensation functionality |
US20210174198A1 (en) * | 2019-12-10 | 2021-06-10 | GM Global Technology Operations LLC | Compound neural network architecture for stress distribution prediction |
CN112163271B (en) * | 2020-09-04 | 2023-11-03 | 北京空天技术研究所 | Atmospheric parameter calculation method of atmospheric data sensing system |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3318146A (en) * | 1966-02-14 | 1967-05-09 | Rosemount Eng Co Ltd | Pressure sensing instrument for aircraft |
US4096744A (en) * | 1975-09-05 | 1978-06-27 | Rosemount Inc. | Pressure sensor for determining airspeed, altitude and angle of attack |
US4378697A (en) * | 1981-07-06 | 1983-04-05 | Rosemount Inc. | Strut mounted multiple static tube |
US4379696A (en) * | 1981-02-12 | 1983-04-12 | Color Communications, Inc. | Latex mylar chip |
US5205169A (en) * | 1991-08-08 | 1993-04-27 | Rosemount Inc. | Angle of attack sensor using inverted ratio of pressure differentials |
US5319970A (en) * | 1991-03-22 | 1994-06-14 | Rosemount, Inc. | Continuously curved strut mounted sensor |
US5485412A (en) * | 1993-10-29 | 1996-01-16 | Rautaruukki Oy | Method and apparatus for successive horizontal and vertical data filtering |
Family Cites Families (32)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4303978A (en) | 1980-04-18 | 1981-12-01 | The Boeing Company | Integrated-strapdown-air-data sensor system |
US4378696A (en) | 1981-02-23 | 1983-04-05 | Rosemount Inc. | Pressure sensor for determining airspeed altitude and angle of attack |
WO1984000071A1 (en) | 1982-06-16 | 1984-01-05 | Boeing Co | Autopilot flight director system |
US4615213A (en) | 1983-12-22 | 1986-10-07 | Rosemount Inc. | Pressure sensing instrument for aircraft |
JPH0765661B2 (en) * | 1986-09-08 | 1995-07-19 | 日産自動車株式会社 | Shift control device for continuously variable transmission |
DE3727906A1 (en) | 1987-08-21 | 1989-03-02 | Messerschmitt Boelkow Blohm | Device for the redundant measurement of the incidence angle and slip angle of flying vehicles, in particular aircraft |
US4836019A (en) | 1987-08-27 | 1989-06-06 | Rosemount Inc. | Compact air data sensor |
RU2041136C1 (en) | 1988-04-25 | 1995-08-09 | Ульяновское конструкторское бюро приборостроения | Device for determination of angle of attack of aircraft |
DE3816057C1 (en) | 1988-05-11 | 1989-04-13 | Daimler-Benz Aktiengesellschaft, 7000 Stuttgart, De | |
JP2913005B2 (en) | 1992-04-06 | 1999-06-28 | 科学技術庁航空宇宙技術研究所長 | Flight velocity vector detection system using a truncated polygonal pitot tube probe and a truncated polygonal pitot tube probe |
US5369993A (en) | 1993-06-30 | 1994-12-06 | The B. F. Goodrich Company | Three axis air data system for air vehicles |
DE4410709A1 (en) | 1994-03-28 | 1995-10-05 | Bodenseewerk Geraetetech | Monitoring device for monitoring the flight safety of aircraft |
JP2952397B2 (en) * | 1994-08-23 | 1999-09-27 | 科学技術庁航空宇宙技術研究所長 | Active air control aircraft using air speed vector measurement device |
US5610845A (en) | 1994-08-30 | 1997-03-11 | United Technologies Corporation | Multi-parameter air data sensing technique |
RU2103718C1 (en) | 1994-12-07 | 1998-01-27 | Санкт-Петербургская государственная академия аэрокосмического приборостроения | Gear testing pickups of automatic control system of aircraft |
US5616861A (en) | 1995-06-07 | 1997-04-01 | Rosemount Aerospace Inc. | Three pressure pseudo -Δ-P sensor for use with three pressure air data probe |
US5615119A (en) | 1995-06-07 | 1997-03-25 | Aurora Flight Sciences Corporation | Fault tolerant automatic control system utilizing analytic redundancy |
US5901272A (en) * | 1996-10-24 | 1999-05-04 | The United States Of America As Represented By The Secretary Of The Navy | Neural network based helicopter low airspeed indicator |
FR2760268B1 (en) | 1997-03-03 | 1999-04-30 | Aerospatiale | METHOD AND DEVICE FOR VERIFYING THE CONSISTENCY OF THE MEASUREMENTS OF AN INCIDENCE SENSOR |
AU2306599A (en) | 1997-12-22 | 1999-07-12 | Accurate Automation Corporation | Air data sensor apparatus and method |
US6253166B1 (en) * | 1998-10-05 | 2001-06-26 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Stable algorithm for estimating airdata from flush surface pressure measurements |
US6305218B1 (en) | 1999-02-22 | 2001-10-23 | Rosemount Aerospace Inc. | Method of and apparatus for using an alternate pressure to measure mach number at high probe angles of attack |
US6466888B1 (en) * | 1999-08-26 | 2002-10-15 | The United States Of America As Represented By The Secretary Of The Navy | Neural network system for estimation of aircraft flight data |
US6594559B2 (en) | 2001-05-08 | 2003-07-15 | Rosemount Aerospace Inc. | Iterative method of aircraft sideslip compensation for multi-function probe air data systems |
US6609421B2 (en) * | 2001-05-08 | 2003-08-26 | Rosemount Aerospace Inc. | Sideslip correction for a multi-function three probe air data system |
US6604029B2 (en) * | 2001-05-08 | 2003-08-05 | Rosemount Aerospace Inc. | Multi-function air data probes using neural network for sideslip compensation |
US6561020B2 (en) * | 2001-05-08 | 2003-05-13 | Rosemount Aerospace Inc. | Method to calculate sideslip angle and correct static pressure for sideslip effects using inertial information |
US6761057B2 (en) | 2001-09-13 | 2004-07-13 | Rosemount Aerospace Inc. | Error detection and fault isolation for multi-function air data probes and systems |
US6668640B1 (en) | 2002-08-12 | 2003-12-30 | Rosemount Aerospace Inc. | Dual-channel electronic multi-function probes and methods for realizing dissimilar and independent air data outputs |
US7379839B2 (en) | 2002-12-23 | 2008-05-27 | Rosemount Aerospace, Inc. | Multi-function air data probes employing neural networks for determining local air data parameters |
US20060212181A1 (en) | 2005-03-16 | 2006-09-21 | Rosemount Aerospace Inc. | Method and apparatus for extending useful range of air data parameter calculation in flush air data systems |
US7213454B2 (en) | 2005-04-06 | 2007-05-08 | Rosemount Aerospace Inc. | Method and apparatus for obtaining improved accuracy and range for air data parameters inferred from independent measurements of interdependent pressures |
-
2002
- 2002-12-23 US US10/328,487 patent/US7379839B2/en not_active Expired - Lifetime
-
2003
- 2003-11-28 EP EP20030257533 patent/EP1437599B1/en not_active Expired - Lifetime
-
2005
- 2005-12-28 US US11/319,966 patent/US20060155506A1/en not_active Abandoned
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3318146A (en) * | 1966-02-14 | 1967-05-09 | Rosemount Eng Co Ltd | Pressure sensing instrument for aircraft |
US4096744A (en) * | 1975-09-05 | 1978-06-27 | Rosemount Inc. | Pressure sensor for determining airspeed, altitude and angle of attack |
US4379696A (en) * | 1981-02-12 | 1983-04-12 | Color Communications, Inc. | Latex mylar chip |
US4379696B1 (en) * | 1981-02-12 | 1991-01-15 | Color Communications | |
US4378697A (en) * | 1981-07-06 | 1983-04-05 | Rosemount Inc. | Strut mounted multiple static tube |
US5319970A (en) * | 1991-03-22 | 1994-06-14 | Rosemount, Inc. | Continuously curved strut mounted sensor |
US5205169A (en) * | 1991-08-08 | 1993-04-27 | Rosemount Inc. | Angle of attack sensor using inverted ratio of pressure differentials |
US5485412A (en) * | 1993-10-29 | 1996-01-16 | Rautaruukki Oy | Method and apparatus for successive horizontal and vertical data filtering |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040122615A1 (en) * | 2002-12-23 | 2004-06-24 | Cronin Dennis J. | Multi-function air data probes employing neural networks for determining local air data parameters |
US7379839B2 (en) | 2002-12-23 | 2008-05-27 | Rosemount Aerospace, Inc. | Multi-function air data probes employing neural networks for determining local air data parameters |
US20070150122A1 (en) * | 2004-10-29 | 2007-06-28 | Rosemount Aerospace Inc. | Fault isolation method and apparatus in artificial intelligence based air data systems |
US7257470B2 (en) * | 2004-10-29 | 2007-08-14 | Rosemount Aerospace Inc. | Fault isolation method and apparatus in artificial intelligence based air data systems |
US20060212181A1 (en) * | 2005-03-16 | 2006-09-21 | Rosemount Aerospace Inc. | Method and apparatus for extending useful range of air data parameter calculation in flush air data systems |
US20060225496A1 (en) * | 2005-04-06 | 2006-10-12 | Rosemount Aerospace Inc. | Method and apparatus for obtaining improved accuracy and range for air data parameters inferred from independent measurements of interdependent pressures |
US7213454B2 (en) | 2005-04-06 | 2007-05-08 | Rosemount Aerospace Inc. | Method and apparatus for obtaining improved accuracy and range for air data parameters inferred from independent measurements of interdependent pressures |
US20070107510A1 (en) * | 2005-10-24 | 2007-05-17 | Mark Agami | Multi-function air data sensor |
US7490510B2 (en) * | 2005-10-24 | 2009-02-17 | Ametek, Inc. | Multi-function air data sensor |
US20070130096A1 (en) * | 2005-12-01 | 2007-06-07 | Rosemount Aerospace, Inc. | Fault detection in artificial intelligence based air data systems |
Also Published As
Publication number | Publication date |
---|---|
EP1437599A1 (en) | 2004-07-14 |
EP1437599B1 (en) | 2015-05-20 |
US20040122615A1 (en) | 2004-06-24 |
US7379839B2 (en) | 2008-05-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20060155506A1 (en) | Multi-function air data probes employing neural networks for determining local air data parameters | |
US6604029B2 (en) | Multi-function air data probes using neural network for sideslip compensation | |
CN107021237B (en) | Synthetic air data output generation | |
US6609421B2 (en) | Sideslip correction for a multi-function three probe air data system | |
US6490510B1 (en) | Fixed multifunction probe for aircraft | |
US7213454B2 (en) | Method and apparatus for obtaining improved accuracy and range for air data parameters inferred from independent measurements of interdependent pressures | |
US6561020B2 (en) | Method to calculate sideslip angle and correct static pressure for sideslip effects using inertial information | |
EP0236587A2 (en) | Time-responsive flight optimization system | |
US7257470B2 (en) | Fault isolation method and apparatus in artificial intelligence based air data systems | |
Rohloff et al. | Air data sensing from surface pressure measurements using a neural network method | |
EP1293784A2 (en) | Error detection and fault isolation for multi-function air data probes and systems | |
US20030010109A1 (en) | Method of reducing total temperature errors and multi-function probe implementing same | |
Lerro et al. | Survey on a neural network for non linear estimation of aerodynamic angles | |
US20020169526A1 (en) | Iterative method of aircraft sideslip compensation for multi-function probe air data systems | |
EP0629166A1 (en) | Aerodynamic pressure sensor systems | |
Prabhu et al. | An innovative analytic redundancy approach to air data sensor fault detection | |
Samy et al. | Fault detection and flight data measurement: Demonstrated on unmanned air vehicles using neural networks | |
Hagen et al. | Deutsche airbus flight test of rosemount smart probe for distributed air data systems | |
Majeed et al. | Multi sensor data fusion based approach for the calibration of airdata systems | |
BOBBITT | The pros and cons of code validation | |
Rohloff | Development and evaluation of neural network flush air data sensing systems | |
Rohloff et al. | Fault tolerance and extrapolation stability of a neural network air-data estimator | |
EP4328547A1 (en) | Training data for air data estimation in aircraft | |
CN108088469A (en) | A kind of long endurance airplane inertial navigation error compensation method | |
Türkmen et al. | A NEW ALTERNATIVE AIR DATA COMPUTATION METHOD BASED ON ARTIFICIAL NEURAL NETWORKS |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION |