Experiment Modal Parameter Identification For Continuous Miner Speed Reducer Based On Mixture Genetic Algorithm

The continuous miner reducer is the core component in key mining equipment of the fully-mechanized face, its reasonable design and reliable operation is the necessary to ensure the high productivity and efficiency of the modern mine. And the modal analysis is the primary mean to the estimate of the structural dynamic characteristics and the optimization in the new product design. The modal analysis can obtain the several modal characteristics of its structure in a sensitive frequency range, and then prejudges the structure’s actual vibration response under the action of the external and internal oscillator, so provides the basis for the fault diagnosis and the structural optimization design. Genetic Algorithm, in the parameter identification, uses the probabilistic search mechanism to make the objective function towards the reduced direction, not depends on gradient information, has strong adaptability, robustness and global search capability. In this paper, the continuous miner reducer, with more drive series, complex structure and the errors existed in the design, manufacture and installation, results in the complex search space of the algorithm. So the paper introduces Lamarckian learning mechanism in the traditional genetic algorithm. By the simulation of their own learning behavior in the life cycle, the mechanism strengthens the local search capabilities, making the individual obtain a higher fitness level in the life cycle and inherit it to the next generation by the way of "gene", so to improve the convergence efficiency.First, this paper describes the development and status of the modal analysis technique and the application of the genetic algorithm in the modal parameter identification, understands the efficiency of the hybrid genetic algorithm for the parameter identification systematically; Second, the cause and mechanism of the vibration of the reducer is summarized, the modal analysis theory is elaborated, and the reducer is simplified as a viscous proportional damping vibration system with multiple degrees of freedom, and its modal expansion is exported. Third, this paper focuses on the mathematical basis of genetic algorithm, formulates Lamarckian operator based on Powell law and builds the experimental modal parameter identification model based on hybrid genetic algorithm, plan the correct genetic program, introduce the Lamarckian strategy, develop a Lamarckian operator to let the each chromosome produced by the standard genetic algorithm learn the mountain climbing. Fourth, the reducer finite element model is established and pre-analyzed. The modal testing is done in accordance with the detail process, and the test data is acquired. Fifth, the collected data is processed by Singular Value Decomposition (SVD) noise reduction, according to the analysis results determines the sensor arrangement, and then the experimental modal parameter is identified by using standard genetic algorithm, hybrid genetic algorithm, LMS Test. Lab. Be contrast with the standard genetic algorithm, the hybrid genetic algorithm is more stable obviously. Compared with LMS Test. Lab commercial systems, the hybrid genetic algorithm, under the Lamarckian learning mechanism, has the ability of deep local search, and can identify the modal parameters with high accuracy. The hybrid genetic algorithm improved by Lamarckian learning mechanism is applied to the experimental modal parameter identification of the continuous miner reducer, and the good results are achieved, which verify the method is accurate and effective.