Hiromitsu Kumamoto

Brief paper: Random sampling approach to state estimation in switching environments

This paper deals with the state estimation for the systems under measurement noise whose mean and covariance change with Markov transition probabilities. The minimum variance estimate for the state involves consideration of a prohibitively large number of sequences, so that the usual computation method becomes impractical. In the algorithm proposed here, the estimate is calculated with a relatively small number of sequences sampled at random from the set of a large number of sequences. The average risk of the algorithm is shown to converge to the optimal average risk as the number of sampled sequences increases. An ideal sampling probability yielding a very fast convergence is found. The probability is approximated in a minimum mean squared sense by a probability according to which sequences can be sampled sequentially and with great ease. This policy of determination of sampling probability makes it possible to design practical and efficient algorithms. Digital simulation results show a good performance of the proposed algorithm.

Efficient Evaluation of System Reliability by Monte Carlo Method

This paper presents a new Monte Carlo method to estimate the reliability of a large complex system represented by a reliability block diagram or by a fault tree. Two binary functions are introduced; one dominates the system structure function and the other is dominated by the structure function. These functions can be constructed easily by using part of path sets and cut sets of the system. Through the use of these binary functions, two variance-reducing techniques (control variate and importance sampling) are applied to the Monte Carlo evaluation of the system reliability. We prove that the new Monte Carlo method gives a reliability estimate with a smaller variance than that of the crude Monte Carlo method.

Automatic path-tracking controller of a four-wheel steering vehicle

The present paper proposes an automatic path-tracking controller of a four-wheel steering (4WS) vehicle based on the sliding mode control theory. The controller has an advantage in that the front- and rear-wheel steering can be decoupled at the front and rear control points, which are defined as centres of percussion with respect to the rear and front wheels, respectively. Numerical simulations using a 27-degree-of-freedom vehicle model demonstrated the following characteristics: (1) the automatic 4WS controller has a more stable and more precise path-tracking capability than the 2WS controller, and (2) the automatic 4WS controller has robust stability against system uncertainties such as cornering power perturbation, path radius fluctuation, and cross-wind disturbance.

State-Transition Monte Carlo for Evaluating Large, Repairable Systems

This paper presents a new Monte Carlo method to estimate unreliabilities of large, repairable systems which can be modeled by a stationary Markov transition diagram. Sequences of state transitions ending at absorbing states are generated, using random numbers. Times to transitions related to the state-sequences are not generated. Next, the probability of system failure occurring in a mission time along each state-sequence is calculated. Finally, the arithmetic mean of these probabilities estimates the system unreliability. This state transition Monte Carlo method yields better estimates in fewer trials than direct Monte Carlo methods. A cold-standby problem with non-identical units is also solved as a by-product of this paper.

Vehicle stability control strategy for steer by wire system

In this paper, the author focuses on steer by wire control. Its performance is compared with front-wheel angle control and yaw rate control using computer simulation, driving simulator testing and actual vehicle testing.

Top-down Algorithm for Obtaining Prime Implicant Sets of Non-Coherent Fault Trees

An algorithm for finding all the prime implicant sets is given for non-coherent fault trees involving gates other than simple AND and OR, e.g., EOR and NOT. The sets are a generalization of min cut sets and can be used in quantitative and/or qualitative system reliability analysis. The algorithm is a top-down analysis and avoids sum of product expressions of top event, which usually involve a large number of terms. Each step of the algorithm is clearly defined and it is proven that all prime implicant sets can be obtained. The algorithm is efficient, and rather complicated trees can be handled manually.

Application of expert system techniques to fault diagnosis

Abstract An expert system, as viewed in the context of this paper, consists of inference algorithms which manipulate a knowledge database created by subject experts. In a fault diagnosis application, if-then production rules can be used to organize the expert knowledge regarding fault isolation. The process of successive classification of system states can be formulated by the following form of the rule: if [system state i] and (observable fact) then [system state j] For example, if [pump flow is low] and (low pressure exists at the suction valve) then [suction valve is closed] The example provided, which involves an application to an engine cooling system, has 22 if-then rules for cause isolation. The inference algorithm used in the interactive conversation with the operator informs him of (1) the if-then rule currently being applied, (2) the proven states and facts in the prerequisite of the rule and (3) a conclusion when the observable fact about which the computer asks turns out to be true.

Optimal Structure of Sensor Systems with Two Failure Modes

An optimal sensor structure is developed for a sensor system which consists of several channels. Each channel monitors a particular plant state, e.g., temperature or pressure. When some states become abnormal, an event occurs. The channels which monitor these abnormal states then initiate appropriate safety systems. Sensors are either good, or failed-dangerous, or failed-safe. More than one sensor is available for each channel. The problem is to obtain the optimal s-coherent sensor structures for the channels. A theorem is proven and a nonlinear integer programming (NLIP) problem is derived to minimize s-expected total damage. The NLIP problem can be solved by the extended Lawler & Bell algorithm. For a 1-channel sensor system, the optimal structure can be obtained analytically by a simple formula.

Cooperative steering system based on vehicle sideslip angle estimation from side acceleration data at percussion centers

This paper considers how a manual steering scheme can be combined with automatic steering for a cooperative steering system which yields better stability, together with path following capability. A new version of the combination is proposed based on a course error differential equation at a percussion center with respect to rear-wheels. The automatic steering includes the feedback of vehicle sideslip angle which is difficult to measure by an ordinary device. An adaptive estimation scheme is proposed to feedback the sideslip angle. Driving simulator experiments demonstrate good estimates and good performance of the cooperative steering system.

Signal-Flow-Based Graphs for Failure-Mode Analysis of Systems with Control Loops

Control loops make failure-mode analysis via fault trees extremely difficult. This paper proposes a new approach based on signal flow graphs to model systems with control loops. Masons Rule is applied to assess the effect of the loops. The top event of the system is defined by an inequality on a node-variable of the signal flow graph. Basic failures are modeled by source variables. Cut-off failures of the control loops are also considered. The method is useful for uncovering failure modes leading to the top event in complicated systems with control loops. General steps to apply the method to a system are: 1. Draw a SFG for the system. 2. Model basic failures by source variables. 3. Select a node-variable to define a top event. 4. Represent the top event in terms of the source variables, using Masons Rule. 5. Discretize the source variables. 6. Classify loop states. 7. For each loop state, obtain system failure modes, using a search tree like Fig. 5. 8. Review the failure modes by more accurate simulation models. Any model is an approximation of an actual system. Thus, the resulting failure modes like those in Table 3 should be examined again, using past experience, more accurate simulation models, etc. The method should be viewed primarily as useful tool for uncovering failure modes in which complicated systems with the control loops fail.