SIMPLIFIED IDENTIFICATION ALGORITHM FOR CLASSICAL LINEAR REGRESSION CONTAINING POWER FUNCTIONS OF UNKNOWN PARAMETER
Annotation
The classical linear regression equation is considered, containing the measured signal in the left part and the sum of terms consisting of the product of unknown parameters and known functions (regressors) in the right part. A distinctive feature of the considered equation from the classical one is the assumption that the unknown parameters are non-linear combinations of one. Namely, each of the unknown parameters is obtained by raising one unknown parameter to a power. The article proposes a new simplified procedure for searching for the unknown parameter, which, unlike the widely used gradient descent method, allows, on the one hand, to significantly simplify the identification algorithm, and, on the other hand, to expand the assumptions for regressors.
Keywords
Постоянный URL
Articles in current issue
- NEURAL NETWORK MODEL OF ERRORS OF AN AUTONOMOUS AIRCRAFT STRAPDOWN INERTIAL NAVIGATION SYSTEM
- AUTHENTICATION OF A CRITICAL OBJECT WORKSTATION OPERATOR BASED ON COMPUTER HANDWRITING
- METHODOLOGICAL ERRORS OF AN ELECTRONIC SENSOR OF AIRCRAFT AIRSPEED VECTOR PARAMETERS
- THREE-LENS SIX-COLOR SUPERACHROMAT OF A WIDE SPECTRAL RANGE
- ROTARY COMPLEX FOR DYNAMIC VIBRATION TESTING OF NANOSATELLITES
- MODELING THE REINFORCEMENT CORROSION PROCESS IN REINFORCED CONCRETE STRUCTURE AT THE OBJECT OF TRANSPORT INFRASTRUCTURE
- MILLIMETER-RANGE ELLIPSOMETRY IN PROBLEMS OF COMPOSITE MATERIALS DIAGNOSTICS
- APPLICATION OF THE SPECTROPHOTOMETRIC METHOD FOR STUDYING ETHYL ALCOHOL IN LIQUID PHASE MEDIA
- FORMALIZING NATURAL PARALLELISM OF MANAGEMENT AND INFORMATION PROCESSES
- FEATURES OF THE PRACTICE OF DEVELOPING STANDARDS TO IMPROVE THEIR QUALITY IN METROLOGICAL SUPPORT TASKS
- FEATURES OF ROUGHNESS EVALUATION OF OPTICAL SURFACES