초록 ▼

A landing-control device is provided having a new structure for carrying out landing control of an aircraft. A detecting unit 10 detects at least a relative altitude from a landing surface to an aircraft. A parameter-generating unit 20 is constructed by a neural network having a feedback loop which

A landing-control device is provided having a new structure for carrying out landing control of an aircraft. A detecting unit 10 detects at least a relative altitude from a landing surface to an aircraft. A parameter-generating unit 20 is constructed by a neural network having a feedback loop which receives a detection value detected by the detecting unit 10 and outputs a landing-control parameter of the aircraft, an output of a first node among plural nodes constituting the neural network being input to a second node different from the first node. A controlling unit 30 controls the aircraft based on the control parameter output from the parameter-generating unit 20.

대표청구항 ▼

What is claimed is: 1. A landing-control device for an aircraft, comprising: a detecting unit detecting at least a relative altitude from a landing surface to the aircraft; a parameter-generating unit, comprising: a neural network that takes a detection value detected by the detecting unit as its i

What is claimed is: 1. A landing-control device for an aircraft, comprising: a detecting unit detecting at least a relative altitude from a landing surface to the aircraft; a parameter-generating unit, comprising: a neural network that takes a detection value detected by the detecting unit as its input and, based on the detection value, outputs a landing-control parameter of the aircraft, the neural network having a feedback loop wherein an output of a first node of the neural network is input to a second node different from the first node; and a controlling unit controlling the aircraft based on the landing-control parameter output from the parameter-generating unit, wherein the output of the first node is temporally delayed and then input to the second node in the feedback loop, wherein a coupling weight coefficient in the neural network is learned by using a genetic algorithm, and wherein a landing condition of the aircraft is estimated by using plural swing models reproducing different swing motions of the landing surface such that an optimum solution fitted to an evaluation condition is set to the coupling weight coefficient. 2. The landing-control device for the aircraft according to claim 1, wherein the plural swing models are changed to different swing models every evolution of a predetermined generation cycle. 3. The landing-control device for the aircraft according to claim 1, wherein the landing condition of the aircraft is estimated by using an artificial swing model wherein a swing condition of the landing surface is intentionally operated by an operator, such that the optimum solution fitted to the evaluation condition is set to the coupling weight coefficient. 4. The landing-control device for the aircraft according to claim 1, wherein the neural network comprises plural layers, including an input layer having the second node, an intermediate layer, and an output layer having the first node, and has the feedback loop such that the output of the first node at the output layer is fed back to an input of the second node at the input layer. 5. A landing-control device for an aircraft, comprising: a detecting unit detecting at least a relative altitude from a landing surface to the aircraft; a parameter-generating unit, comprising: a neural network that takes a detection value detected by the detecting unit as its input and, based on the detection value, outputs a landing-control parameter of the aircraft, the neural network having a feedback loop wherein an output of a first node of the neural network is input to a second node different from the first node; and a controlling unit controlling the aircraft based on the landing-control parameter output from the parameter-generating unit. wherein a coupling weight coefficient in the neural network is learned by using a genetic algorithm, and wherein a landing condition of the aircraft is estimated by using plural swing models reproducing different swing motions of the landing surface such that an optimum solution fitted to an evaluation condition is set to the coupling weight coefficient. 6. The landing-control device for the aircraft according to claim 5, wherein the plural swing models are changed to different swing models every evolution of a predetermined generation cycle. 7. The landing-control device for the aircraft according to claim 5, wherein the landing condition of the aircraft is estimated by using an artificial swing model wherein a swing condition of the landing surface is intentionally operated by an operator, such that the optimum solution fitted to the evaluation condition is set to the coupling weight coefficient. 8. The landing-control device for the aircraft according to claim 5, wherein the neural network comprises plural layers, including an input layer having the second node, an intermediate layer, and an output layer having the first node, and has the feedback loop such that the output of the first node at the output layer is fed back to an input of the second node at the input layer. 9. A landing-control method for an aircraft, comprising: using a neural network to output a landing-control parameter of the aircraft by inputting at least a relative altitude from a landing surface to the aircraft to the neural network, the neural network comprising a feedback loop such that an output of a first node in the neural network is inputted to a second node different from the first node; controlling the aircraft based on the landing-control parameter output; and learning a coupling weight coefficient in the neural network by using a genetic algorithm, wherein the output of the first node is temporally delayed and then input to the second node in the feedback loop, and wherein said learning said coupling weight coefficient comprises: estimating a landing condition of the aircraft by using plural swing models that reproduce different swing motions of the landing surface, such that an optimum solution fitted to an evaluation condition is set as the coupling weight coefficient. 10. The landing-control method for the aircraft according to claim 9, wherein said learning said coupling weight coefficient comprises: changing the plural swing models to different swing models every evolution of a predetermined generation cycle. 11. The landing-control method for the aircraft according to claim 9, wherein said learning said coupling weight coefficient comprises: estimating the landing condition of the aircraft by further using an artificial swing model wherein a swing condition of the landing surface is intentionally operated by an operator to set the optimum solution fitted to the evaluation condition as the coupling weight coefficient. 12. A landing-control method for an aircraft, comprising: using a neural network to output a landing-control parameter of the aircraft by inputting at least a relative altitude from a landing surface to the aircraft to the neural network, the neural network comprising a feedback loop such that an output of a first node in the neural network is inputted to a second node different from the first node; controlling the aircraft based on the landing-control parameter output; and learning a coupling weight coefficient in the neural network by using a genetic algorithm, wherein said learning said coupling weight coefficient comprises: estimating a landing condition of the aircraft by using plural swing models that reproduce different swing motions of the landing surface, such that an optimum solution fitted to an evaluation condition is set as the coupling weight coefficient. 13. The landing-control method for the aircraft according to claim 12, wherein said learning said coupling weight coefficient comprises: changing the plural swing models to different swing models every evolution of a predetermined generation cycle. 14. The landing-control method for the aircraft according to claim 12, wherein said learning said coupling weight coefficient comprises: estimating the landing condition of the aircraft by further using an artificial swing model wherein a swing condition of the landing surface is intentionally operated by an operator to set the optimum solution fitted to the evaluation condition as the coupling weight coefficient.