Models for predicting traffic congestion over time on a road network have been formulated as non-linear optimization problems. In this paper, we construct a new dynamic flow model represented as an optimization problem on a time-space network. In this model, there is a one-to-one correspondence between any directed path on the time-space network and a travel actually taken by a user of the road network. This is an advantage over conventional dynamic flow models in which a travel by a user may correspond to multiple paths in the underlying time-space network. The proposed network model involves arc capacity constraints that depend on the amount of flows of relevant arcs. Since those capacity constraints prevent this model from being treated as network flow problem, we adopt a penalty function techinique to transfer them into the objective function, thereby obtaining a standard non-linear minimum cost flow problem. We report some numerical results to show the validity of the proposed model.
In conventional bill money recognition machine, we develop the recognition algorithm according to the transaction speed and difference of various specifications. However, development of the algorithm for the recognition has been based on the trial and error method. Many researchers have reported that neural networks are suitable for pattern recognition because of the ability of selforganization, parallel processing, and generalization. In this paper, we present, a new bill money recognition method with neural network and show the effectiveness of the present algorithm compared with the pattern matching. Furthemore, we transform bill money data by FFT into frequency domain. Data representation by FFT is more preferable to bill money recognition with neural network and we show that the recognition ability can be improved still more by introducing a new measure of reliability.
This paper proposes a design method of MRACS (Model Reference Adaptive Control System) for unknown plants in the presence of deterministic disturbance or periodic disturbance, so that the output error will converge to zero as time tends to infinity. In the traditional adaptive control systems dealing with disturbance, the disturbance was usually restricted to be deterministic. To remove the effect of disterbance, the general way was to use extended system including a disturbance model, or to use integral compensators. Here, a new disturbance compensating method is proposed, in which disturbance compensating signals are introduced into MRACS, and are combined adaptively to cancel the effect of disturbance. In this proposed method, because the filter signals of input and output are composed in the same way as the disturbance free cases, the MRACS has a simple constructure. Even for the periodic disturbance with unknown shape, if the period is known, by using Fourier series, the proposed method is also possible to remove the effect of this sort of disturbance. Furthermore, being used, disturbance compensating signals in MRACS, it is possible to take the initial responses of filters into account, and the characteristics of parameter convergence can be improved. The effectiveness of this proposed method is illustrated by some computer simulations.
For autonomous assembly operations by a robot, it is required to build a planning function that can generate a series of operations to reach a goal state. This paper describes an automatic planning algorithm that synthesizes the assembly operation strategies. By making an criterion function based on the difficulty of state transitions be minimum, the method can find an optimal path based on the Contact State Network. The Degree of Constraint is proposed by regarding the assembly task as a process to change the constraint, state of the moving object. Then several policies for determining State Transition Difficulty are discussed where the variation of the degree of constraint and the shape information from the geometric model of objects are considered. A criterion function for state transitions is defined based on the discussed policies. Lastly, an algorithm which plans an optimal state transition path from an initial state to a goal state is proposed. Some examples are given to show the validity of this algorithm
In our previous paper, we proposed an approximation model of pitch pattern for the recognition of four tones of Chinese speech. In this paper, we describe how to use the approximation model to recognize the tones of polysyllabic Chinese words. In order to segment a phrase to the syllabic unit, we propose a new segmentation scheme based on the analysis of the sound linking characteristic between syllables. The phrases are classified into non-linked syllabic phrases and linked syllabic phrases. Applying pitch information directly to the former, power parameter and AMDF (Average Magnitude Differential Function) to the latter, phrases are segmented into two syllabic units. Using the approximation model and separated syllabic units, tone recognition of polysyllabic words are carried out. From the results of the recognition experiment which realized the recognition rate of 96. 6% in an average, it is evident that this method is effective to recognize the tone of polysyllabic Chinese words.