Search

Bridge structures deteriorate due to various factors, and their maintenance can be made more efficient by utilizing relevant information. This paper describes an algorithm called Self-Organizing Map-based Cluster Merging (SOMCM) using a multi-dimensional matrix composite neural network, and a surface image identification system. The algorithm is designed to investigate the relevance between the main components of bridges and their types of deterioration. 6140 records on bridge maintenance were collected for all bridges in the Taoyuan region of Taiwan in 2021. The SOMCM algorithm involves finding the winner neuron through merging processes. 61 clusters were merged into 8 clusters after a predetermined number of iterations. Clustering analysis of the final 8 clusters revealed 9 major bridge maintenance association rules. Follow-up studies can apply the algorithm to integrate more technologies including GIS coordinates, material availability, real-time traffic conditions, and weather information to be of benefit to engineering practitioners. • The study is to develop a new algorithm, SOM-based Cluster Merging (SOMCM). • The paper is to investigate the relevance for bridge main components and their deterioration types. • The initial result involves 61 clusters being merged to 8 clusters with the predetermined iterations. • The clustering analysis for the final 8 clusters shows 9 major bridge maintenance association rules. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, a multi-layer perceptron based on 6 important features is proposed for predicting the time needed by our aircraft to approach enemy aircraft when they are in a combat situation. Then the prediction module is incorporated into a SOMO-based decision aid for solving multi-aircraft assignment problems. The problem of assigning our aircraft to meet head-on enemy aircraft in a combat situation is formulated as a constrained resource allocation problem. The objective function of the multi-aircraft assignment problem is a measure for computing the expected overall predominance value of aircraft over enemy aircraft based on four predominance factors. The SOMO algorithm which is an extension version of the SOM algorithm for optimization is executed to find an assignment such that the expected overall predominance value of aircraft can be maximized. Four hundred combat situations are simulated to test the effectiveness of the proposed SOMO-based decision aid. The simulation results demonstrate that the SOMO-based decision aid could provide appealing solutions within 0.5 s even for combat situations with a large number of aircraft. [Copyright &y& Elsevier]

Abstract: This paper presents a new approach to incrementally constructing a neural network that is capable of learning new information without forgetting old knowledge. The proposed neural network, called hyper-spherical ARTMAP network (HS-ARTMAP network), is a synthesis of an RBF-network-like module and an ART-like module. The HS-ARTMAP network is trained via a training algorithm similar to the training algorithm for the fuzzy ARTMAP system. To demonstrate the performance of the proposed HS-ARTMAP network, several pattern recognition and function approximation problems were tested. [Copyright &y& Elsevier]

Abstract: A classifier system is a machine learning system that learns syntactically simple string rules (called classifiers) through a genetic algorithm to guide its performance in an arbitrary environment. In a classifier system, the bucket brigade algorithm is used to solve the problem of credit assignment, which is a critical problem in the field of reinforcement learning. In this paper, we propose a new approach to fuzzy classifier systems and a neuro-fuzzy system referred to as ACSNFIS to implement the proposed fuzzy classifier system. The proposed system is tested by the balancing problem of a cart pole and the back-driving problem of a truck to demonstrate its performance. [Copyright &y& Elsevier]

Abstract: In this paper we propose a clustering algorithm to cluster data with arbitrary shapes without knowing the number of clusters in advance. The proposed algorithm is a two-stage algorithm. In the first stage, a neural network incorporated with an ART-like training algorithm is used to cluster data into a set of multi-dimensional hyperellipsoids. At the second stage, a dendrogram is built to complement the neural network. We then use dendrograms and so-called tables of relative frequency counts to help analysts to pick some trustable clustering results from a lot of different clustering results. Several data sets were tested to demonstrate the performance of the proposed algorithm. [Copyright &y& Elsevier]