Synthesis of quasi-optimal fast filters by the least square criterion

The subject of the article research is special signal processing methods based on optimal discrete filtering theory. goal to increase efficiency model-based for information signals by reducing computational costs and increasing speed algorithms. Applied methods: description dynamic processes in terms state space using elements vector-matrix algebra, weighted least squares method, Kalman's theory filtering, basic concepts O'Reilly–Luenberger functional observers, probability theory, statistical modeling Monte Carlo method. Results: a new method proposed, which uses approximation Kalman filter transfer matrix time dependence given piecewise linear functions according criterion. effectiveness was evaluated example second-order dynamical system. On basis comparative analysis, several acceptable variants considered are proposed. practical significance work lies further development synthesis quasi-optimal high-speed filters. operability proposed modifications confirmed algorithms criterion "accuracy-computational costs". It shown that total savings number multiplication addition operations can reach tens times due insignificant losses accuracy process.