| Piling machine is a kind of indispensable important equipment of construction machinery, and machine frame as one of the most important part of the piling machine, have the effect of supporting the whole pressure machine, piling machine performance depends largely on the quality of the frame and structure form.Taking YZY800D static hydraulic piling machine frame as the research object, studied the frame lightweight methods, and discusses the stress, strain and acceleration with load frame frequency and phase angle change rule.The main contents are as follows:Piling machine working status and loading is complex. There are two different installation ways on piling machine based on the rack, to simplify the frame stress for 6 kinds of working condition, and calculate the frame stress under the working condition of each size, direction and functional point, analyze the working condition of the six kinds of stress and strain contours of each condition and to determine the frame stress the worst working condition, calculate the frame of the allowable stress and allowable deformation for strength and rigidity of frame.Frame parameterized model is set up in the DM module in ANSYS Workbench, and eight different evaluation methods validate meshing quality of grid, using GDO module to complete the optimization of the frame.Using central composite design (CCD) of DOE to complete the response surface, it is concluded that the correlation between input and output parameters, using a multi-objective genetic algorithm (MOGA) to complete the optimization of the frame, analysis shows that the maximum stress appears in the frame on the inside of the plate, the frame quality decreased by 13.26% after optimization. Verify the optimized model under 6 kinds of working conditions, and results show that the optimized frame model meets the design requirements.Analysis of defects of BP neural network algorithm, through the improved learning algorithm to improve the network convergence speed and stability, and comparing the advantages and disadvantages of three methods for the training sample, using orthogonal test to select the sample data, select 50 groups of sample data, and use the Sigmoid transfer function to normalize the sample data. Comparing the advantages and disadvantages of 7 kinds of BP neural network algorithm, and choose LM algorithm as the BP neural network algorithm, call load trained four implicit function of BP neural network, using the penalty function method dealing with constraints, selection of fitness function to evaluate the individual fitness tests by using the genetic algorithm for global, and get the optimal solution, the frame weight is reduced by 13.53%. Provide practical engineering problems for the nonlinear reference.Compare two methods in the harmonic response analysis. The modal superposition method for harmonic response analysis is carried out on the rack. Through the modal analysis, the frequency range in the harmonic response analysis can be determined for 4～21 Hz. Divide the analysis steps, determine the analysis object, set the analysis phase and coordinates. According to the harmonic response analysis curve, the phase Angle a little influence the result of the harmonic response analysis, and the frequency range affects obviously the result of the harmonic response analysis. Within 4～21 Hz selected 10 points, calculate stress and strain of each point. According to the stress and strain amplitude frequency curve, ensure the safety of the frame frequency range from 4 to 11.25 Hz. |
