Shinji Suzuki

Neural Network Modeling of Lateral Pilot Landing Control

Human pilot landing control has been analyzed by the authors using neural network modeling. Although the previous research considered only longitudinal control, analysis of the lateral control is the focus of this paper. The lateral control can be quite difficult under crosswind conditions, because the combination of two lateral control methodologies (crab method and wing-low method) is needed for a successful landing. To analyze lateral control, this paper makes a first step: a lateral pilot control model is established. The necessary visual cues for lateral control are defined, and neural network models for aileron and rudder controls are constructed. The adequacy of the proposed neural network model is checked by Monte Carlo simulations.

Analysis of Visual Cues During Landing Phase by Using Neural Network Modeling

A neural network modeling approach has been developed to analyze the pilots use of visual cues for landing a transport airplane. Time sequences of the visual cues and pilot control inputs obtained by using a flight simulator can be analyzed to quantitatively estimate the relationship between the visual cues and the pilot control inputs. In this paper, visual cues such as the horizon, runway shape, and runway marker are compared based on their importance. By using a flight simulator, neural network models are obtained in all cases wherein a pilot intentionally alters his or her attentiveness to the visual cues. The contribution ratios analysis reflects the attentiveness to each visual cue. The Monte Carlo landing simulation shows the difference in robustness of each obtained neural network model. It is confirmed that the timely choice of the appropriate visual cue is necessary for a smooth and safe landing.

Analysis of Human Pilot Control Inputs Using Neural Network

A neural-network-modeling approach is applied to analyze human-pilot control inputs during the landing phase in the visual approach. Flight data that contain the aircraft state variables and pilot control inputs are recorded using a flight simulator. The time history of visual cues and control inputs is utilized as teaching data for neural networks that can emulate the movements of a human pilot. A genetic algorithms approach is proposed to improve the generalization ability of the network by determining the network structure and initial values of its parameters. Generalization capabilities are evaluated by analyzing the flight data of a personal-computer-based simulator. The contribution ratios of each visual cue and their sensitivities to the control inputs of a pilot are estimated by analyzing the obtained neural-network models using a training simulator. The obtained results reveal that the proposed method can be used for analyzing the skill of a pilot.

Optimization of Neural Network Modeling for Human Landing Control Analysis

[Abstract] A neural network modeling technique is developed to analyze human pilot control, making a connection between visual cues and pilot control during the final landing phase. When constructing the neural network model, its generalization capability has to be watched closely. Generalization capability means that the neural network model can obtain robustness. In order to improve it, a genetic algorithm has been used in the previous study. However, the genetic algorithm was very time consuming, which is an obstacle for making an analysis tool for pilot training. In this paper, the neural network parameters, such as selection of inputs and noise reduction method, are refined and the generalization capability can be improved without the use of genetic algorithms. Its generalization is confirmed by Monte Carlo landing simulations. Moreover, the potential skill of the pilot can be modeled using the landing data of several flights done by the same pilot.

Modeling of Pilot Landing Approach Control Using Stochastic Switched Linear Regression Model

DOI: 10.2514/1.C000204 Human pilot control during the landing approach is modeled using a stochastic switched model, which is a combination of the conventional linear regression model and a hidden Markov model. Using this model, the time histories of pilot control can be categorized in several states, such as pitch stabilizing mode. First, the selection of modelinputsisdiscussed.Basedondataobtainedin flightsimulatorexperiments,thepilotapproachcontrolmodelis constructed for several cases, with different wind conditions applying an expectation-maximization algorithm. The obtained models are analyzed with respect to the timing of state transitions and the stochastic-switched-model gain parameters. The current findings suggest that the proposed analysis method has a great potential to reveal a pilot’s decision-making process.

Analysis of pilot landing control in crosswind using neural networks

Human pilot control at visual approach has been analyzed using a neural network modeling technique. Neural network models simulate the relationship between control (e.g., elevator, throttle, etc.) and human input (visual information) and can be analyzed mathematically. In previous research, only longitudinal operation was analyzed, because the characteristic flare maneuver is said to be one of the most difficult maneuvers in normal operations. However, lateral control is also difficult especially under crosswind conditions. In a crosswind approach, crab control is applied first, and then winglow sideslip control is applied. The transition process of these two controls is called decrab, which is recognized as quite a difficult maneuver. Moreover, the longitudinal control also depends on the lateral control, and the influence of this coupling deserves some interest, too. In this paper we focus on the lateral control analysis. As in the previous research, neural network plot models are investigated with sensitivity analysis. Several simulator experiments were conducted with different pilots and under various wind conditions, and the analysis results clarify the differences in control strategies.

Analysis of Pilot Landing Control Using Neural Network

A neural network (NN) modeling approach has been developed to analyze human pilots’ control by utilizing the recorded time history of visual cues and pilot control inputs. The contribution ratios of each visual cue are estimated by analyzing the obtained NN models. This paper compares the importance of visual cues such as the horizon, a runway shape, and a runway marker. NN models are obtained in all cases wherein a pilot intentionally changes the attentiveness to visual cues by using a flight simulator. The obtained contribution ratio and flight simulation results that are controlled by the estimated NN model clarify the role of each visual cue. Additionally, an experimental method is developed to record the time histories of visual cues and pilot control inputs by analyzing recorded video data during a real flight. The developed method makes it possible to obtain the NN model of a pilot for a real landing case. I. Introduction HE most difficult maneuver for airline pilots is perhaps a landing approach. However, it has yet not been automated. In particular, an airline pilot cannot afford to read instruments during the final approach. Therefore, they have to estimate airspeed, descent rate, altitude, and pitch angle of the aircraft by using visual cues from the cockpit. It is considered that this estimation skill plays an important role in the smoothness of the landing. 1 Since it is difficult to analyze the estimation skill directly, the author’s team has developed a new analysis tool using neural network (NN) modeling techniques. 2,3,4 This method utilizes the recorded time histories of visual cues such as the horizon, runway shape, and runway marker and the corresponding time history of the pilot control inputs. The contribution ratio and sensitivity of each visual cue to the pilot control can be estimated. Artificial NNs are mathematical models that emulate biological nervous systems and consist of a large number of highly interconnected processing elements such as neurons. 5 The NN model is adjusted by a learning process. Note that the NNs can relate input/output data that have high nonlinearity. The obtained NNs can be applied to automatic recognition systems and automatic control systems of complicated problems including human control. We have applied the NNs to analyze airplane pilots’ maneuvers during the landing phase. The input data for the NNs are visual cues, e.g., runway geometries and the horizon, and control column input. The output data from the NNs are control column and throttle lever deflections. The contribution ratios of each NN input to the pilot control column are computed to analyze the pilot maneuvers. It has been recognized that the most difficult problem in creating the NN models is deciding whether or not the obtained model has generalization capability. Generalization refers to the NN producing reasonable outputs for inputs not encountered during the training. In our analysis, Genetic Algorithms (GAs) 6 are applied to determine some parameters in the NN models, as described in Ref. 4, where this approach can increase the generalization capability. 4

Simple Adaptive Control with PID for MIMO Fault Tolerant Flight Control Design

This paper describes a fault-tolerant flight control method using Simple Adaptive Control augmented with a PID controller (PID-SAC). The aim of this paper is to demonstrate control of an aircraft using PID-SAC control even if the dynamic characteristics change due to damage to the aircraft. We show actual flight test results in the case where the aileron or rudder authority suddenly diminishes mid-flight. Through actual flight tests, we demonstrate the superiority of this new control method in comparison to a conventional control method. One of the biggest advantages of PID-SAC approach is that a PID-SAC controller has high flexibility and extensibility to another control scheme.

Neural Network Analysis of Pilot Landing Control under Real Flight Condition

A neural network modeling approach has been developed to analyze a pilot’s control during a visual approach. While flight simulator operations were analyzed in our previous research, real flight landing cases are analyzed in this paper. An experimental method which uses recorded video data is developed to obtain the necessary data for analysis, such as visual cues and column deflection. Using this method, data from various aircraft and pilots can be obtained and analyzed. A contribution ratios analysis and sensitivity analysis of the NN model reveal differences in control strategy for different aircraft and different pilots. I. Introduction ITH the increase of air traffic, the number of aircraft accidents has increased steadily. A large number of aircraft accidents occurred during landing attempts, and some of them would not have occurred if the pilot’s control would have been appropriate. A pilot’s skill depends much on his/her experience, and it is difficult to make a guideline for control. Even if the pilot’s skill seems to be enough under normal conditions, nobody knows whether the pilot can perform appropriate control in case of an emergency. Therefore, it is important to examine the pilot’s potential control skill quantitatively, and the authors have developed an evaluation method for pilot control. 1

Flight test of fault-tolerant flight control system using simple adaptive control with PID controller

Purpose n n n n nThis paper aims to demonstrate the effectiveness of a fault-tolerant flight control method by using simple adaptive control (SAC) with PID controller. n n n n nDesign/methodology/approach n n n n nNumerical simulations and flight tests are executed for pitch angle and roll angle control of research aircraft MuPAL-α under the following fault cases: sudden reduction in aileron effectiveness, sudden reduction in elevator effectiveness and loss of longitudinal static stability. n n n n nFindings n n n n nThe simulations and flight tests reveal the effectiveness of the proposed SAC with PID controller as a fault-tolerant flight controller. n n n n nPractical implications n n n n nThis research includes implications for the development of vehicles’ robustness. n n n n nOriginality/value n n n n nThis study proposes novel SAC-based flight controller and actually demonstrates the effectiveness by flight test.