Research On Evolutionary Algorithm For Optimization Three-dimensional Integrated Sensors

This research is from extension of the Harbin Municipal Science and Technology re-search project "Giant Magneto-resistive precision electronic compass ". Optimization for theVector field measurement system is often needed by many measuring devices. Optimizingwith algorithm is the most effective way now. More foreign researchers using recursive leastsquares, Kalman filtering,maximum likelihood estimation and other methods to solve thisproblem. In this paper, the method is first put forward that using computational intelli-gence techniques solving the orthogonal correction and alignment problem for vectormeasurements.This paper is mainly about the evolutionary algorithms, especially the com-mon geneticalgorithm and particle swarm optimization and the application in the calibration and align-ment of a vector field measurement system, including:1．A concept of self-evolutionary genetic algorithm is proposed. Since the genetic algo-rithm was proposed, its major evolutionary mechanism is to imitate the biological evolutionprocess, and the most sensitive evolutionary parameters is given by the artificial, this makesome uncertainties for the algorithm results, particularly it unable to escapethe dependence on human intervention. Now that these disadvantage of the classical geneticalgorithm, the “self-evolution genetic algorithm” expands the genetic and evolutionary me-chanisms to the selecting of the parameters for the algorithm in the optimization process, andthe parameters of the algorithm in some sense deemed to be variables that need to be opti-mized, so the algorithm get rid of the human experience and artificial participation, throughgenetic and evolutionary, the optimizal parameters for the algorithm and the "optimum solu-tion" will be generated automatically. This is an algorithm that fully reflect the mechanism ofgenetic and evolution, compared to the classcal genetic algorithm, the new algorithm can bet-ter embody the genetic and evolutionary characteristics, and the objectivity of the result ismore stronger, the operation is simple, at the same time it does not significantly increase thecomputational complexity.2. An improved adaptive particle swarm optimization algorithm is proposed. The algo-rithm introduces self-learning factor to particle swarm optimization in the process by dynam-ically adjusting the diversity of population and the optimization direction.3. On the basis of the principle of mapping the measurement system to the ideal ortho-gonal system, and according to the equalation of each vector modulus after coordinate systemtransformation, the measurement system error correction model is established for the ortho-gonal correction of three axis measuring system. 4. In addition, in order to solve the alignment problem for the measuring systems, thispaper put forward two kinds of new methods: one is used for the same vector measurementsystem, another is used for the different vector measurement system. Applicating theself-evolution genetic algorithm and the improved adaptive particle swarm optimization algo-rithm on the identification for the same vector measurement system transformation matrixparameter, experimental results prove the method was correct and feasible.These efforts is valuable for verifying the stability of the algorithms, and provides afeasible scheme for the use of genetic algorithms and particle swarm algorithms to correctand to registrate the measurement system in online or quasi-online mode.