Improved rank selection algorithm

Problem statement. A rank algorithm for selecting radio emission modes for operation in conditions of heterogeneity of sources and complex interference conditions, including the possible presence of mutual interference, is synthesized.Objective. The synthesis purpose is to ensure the independence of mode recognition from particular features of radio emission observation. Algorithm input is the primary signal processing result that includes such estimations as pulses durability, frequency and amplitude dynamics, and absolute variations. Primary decision statistics are formed using these values: observable signal base and relation variations of frequency and amplitude. Secondary statistics are formed based on primary ones using median and recursive or maximum and recursive smoothing. Each of the decision statistics in the multi-threshold procedure is transformed into a row of ranks, the size of which corresponds to the number of recognized modes. In aggregate, these lines form a ranking table (matrix) with colons representing recognized modes’ discrete descriptions. Fluent observation processing includes rank formation for used decision statistics. Mode recognition is performed either following a ranking table or using an additional voting procedure 2/3. An alternative approach consists of constructing the Manhattan mismatch metric of the current and reference ranks and making a decision on the criterion of the minimum mismatch metric.Results. Mode recognition performed on results of this comparison using unbalance metrics minimum criterion. Thresholds in frames of the ranking procedure are formed heuristically at ranking table formation. They are then used at fluent rank formation for observable modes. The performed numerical experiment shows that maximal and recursive filtration provides an errorless selection of all observable modes. This filtration represents the composition of maximum selection in sliding window and subsequent recursive first-order filtration. An additional advantage of this filtration is a simpler maximum selection in comparison with the median one. In perspective, it can provide increased operating speed.Practical implications. Performed consideration shows that rank selection is worthwhile at the observation of heterogeneous irradiation sources. Algorithm strength is decision simplicity in a complex situation. Additional algorithm advantage is the possibility of extending alternative irradiation modes and, hence, for more representative data sets.

A NEW STUDY ON ASYMPTOTIC OPTIMALITY OF LEAST SQUARES MODEL AVERAGING

In this article, we present a comprehensive study of asymptotic optimality of least squares model averaging methods. The concept of asymptotic optimality is that in a large-sample sense, the method results in the model averaging estimator with the smallest possible prediction loss among all such estimators. In the literature, asymptotic optimality is usually proved under specific weights restriction or using hardly interpretable assumptions. This article provides a new approach to proving asymptotic optimality, in which a general weight set is adopted, and some easily interpretable assumptions are imposed. In particular, we do not impose any assumptions on the maximum selection risk and allow a larger number of regressors than that of existing studies.

Object-oriented backtracking

Several versions of the backtracking are known. In this paper, those versions are in focus which solve the problems whose problem space can be described with a special directed tree. The traversal strategies of this tree will be analyzed and they will be implemented in object-oriented style. In this way, the traversal is made by an enumerator object which iterates over all the paths (partial solutions) of the tree. Two different “acktracking enumerators” are going to be presented and the backtracking algorithm will be a linear search over one of these enumerators. Since these algorithms consist of independent objects (the enumerator, the linear search and the task which must be solved), it is very easy to exchange one component in order to solve another problem. Even the linear search could be substituted with another algorithm pattern, for example, with a counting or a maximum selection if the task had to be solved with a backtracking counting or a backtracking maximum selection.

GENETIC DIVERSITY INAUS RICE LANDRACES OF BANGLADESH USING SSR MARKERS

Assessment of genetic diversity is essential for germplasm characterization, utilization and conservation. Genetic diversity of 31 Aus rice landraces of Bangladesh was assessed using 36 SSR (simple sequence repeats) markers. A total of 141 alleles were detectedand the number of alleles per locus ranged from two (RM1216, RM145, RM282, RM293, RM567and RM496) to 10 alleles (RM304), with an average of 3.92. The gene diversity varied from 0.06 (RM145) to 0.80 (RM304) with an average of 0.54 and the PIC values ranged from 0.06 (RM145) to 0.78 (RM304), with an average of 0.48.PIC value revealed that RM304 was the best marker for characterizing the studied Aus rice genotypes. The dendrogram from unweighted pair-group method with arithmetic average clustering of markers classified the genotypes into five major groups with a coefficient of 0.49. Two and three-dimensional graphical views of Principal Coordinate Analysis (PCA) revealed that the genotypes Hashikalmi, Chaina, Puitraaijang, Saithsail, Kuchmuch, Kalodhan, Ausdhan and Itcriewere found far away from the centroid of the cluster and can be selected as parents for further breeding programs.The results provided some useful implications for establishment of sovereignty of Bangladeshi rice gene pool. This information will provide maximum selection of diverse parents, background selection during backcross breeding programs and assist in broadening germplasm-based rice breeding programs in future.

Fuzzy Logic Based Approach for Power System Fault Section Analysis

In this chapter, a fuzzy expert system is developed to assist the operators in fault detection. It requires much less memory to store the database (power system topology and the post fault status of circuit breakers and protective relays). The fuzzy expert system identifies two basic network section sets, Shealthy for the healthy sub network and Sisland for the fault islands, using the post fault status of circuit breakers and relays. It then calculates membership function for each possible fault section. The objective of this calculation is to determine the likelihood of each candidate fault section as the actual fault section. Moreover membership functions provide a convenient means of ranking among possible (or candidate) fault sections, and are the most important factors in decision making. During decision making, the most possible fault section is determined by maximum selection method. In this method most possible fault section is the one which is having highest membership grade. MATLAB code for the proposed scheme is developed and the results obtained in four cases for a power- system network.