Kupershtein L. Methods and Facilities of the Data Neural-alike Processing for a Control Systems

A research object is processes of the threshold parallel adding up and threshold algebraic parallel adding up of vector data array. The research subject is structural organization of pipeline processors as digital neurochips for the threshold parallel processing of numerical information. Research methods are based on the basis of mathematical logic theory, mathematical analysis and statistics, theory of computer systems analysis and synthesis, theory of digital signals and images processing, theory of neural computing and neural structures. The purpose of dissertation work is an improvement, namely, increase of speed of operation and extending compatibility option of methods and facilities of neural-like processing of data arrays for the tasks of identification and control dynamic objects using the neural networks technologies. The theoretical meaning of the got results consists of that the method of the parallel adding up of vector data array on a different cuts base is improved as basic in neural-like processing of vector arrays, improved and researched the modified model of neuron (perceptron) with the lateral braking, the structures of pipeline processors are offered for the neural-like processing of vector data, the method of efficiency estimation of pipeline processors is modified for the neural-like data processing. The practical meaning of the got results consists of that the realization ways of the neural-like data processing are offered on a different cuts base, which allows increasing a speed of operation and extending compatibility option of neural-like for processing of large arrays of information. Degree of impression lie under bounds of field. A field of impression is companies of telecommunication equipment.