| Combines are important agricultural machinery products.In the Fourteenth FiveYear Plan of Agricultural Mechanization Major Challenges and Strategic Tasks and the Plan of Action for Agricultural Machinery Equipment Development(2016-2025),the Ministry of Agriculture and Rural Area proposed to strengthen the construction of quality and reliability of agricultural machinery equipment and improve the reliability of agricultural equipment,especially harvesting machinery.Studying the problems of combine assembly quality inspection,analyzing the inspection methods of combine assembly quality problems and designing the combine assembly quality inspection system are important components to improve the reliability of combine and are of great significance to realize the agricultural machinery reliability index in the national industrial layout planning of the 14 th Five-Year Plan.Therefore,this paper focuses on the problems faced by the detection of assembly quality problems of combine harvesters,and conducts in-depth research on vibration signal characteristics analysis,signal noise reduction,feature fusion extraction and classification model construction.The main research contents include:Analyzing the assembly relationship of each system of combine harvester,researching the assembly quality problem of each system and vibration signal of typical working parts are the basis of the assembly quality inspection of combine harvester.Taking the axial-flow drum combine as the research object and based on the driving structure characteristics of the working parts of the combine,the vibration characteristics of typical working parts of the combine such as cutting table,vibrating screen,separating drum,engine and the common rotor and bearing problems mechanism of each working part were analyzed,and the characteristics of excitation signal and problem response signal contained in the vibration signals of each part of the combine were studied.The frequency modulation,amplitude modulation and multi-frequency simple harmonic combination simulation model of assembly quality signal of combine was built.Vibration signal acquisition test of assembly quality problem of combine provided theoretical and data basis for the establishment of assembly quality detection system.Aiming at the problems of complex mechanical structure of combine and difficulty in identifying and extracting characteristics caused by strong coupling of vibration signals,a noise reduction method for vibration and strong noise signals of combine assembly quality problem was put forward.Vibration signal decomposition performance of complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)was improved by improving interpolation fitting and iteration screening process,and was verified by superposition simulation signal and simulation signal of combine assembly quality problem.The singular value decomposition(SVD)method is analyzed and an improved threshold method is proposed.The improved threshold filtering SVD is used to reduce the noise of the intrinsic modal function(IMF).The noise reduction effect of this method is verified by the simulation signal of the combine assembly quality problem.The feature extraction method for vibration signal of combine assembly quality was studied.The time sequence selection process of multi-scale reverse discrete entropy(MRDE)was improved by sliding coarse-grained method,and an improved feature extraction method of time-frequency domain special Entropy was put forward,which improved the feature characterization ability.The feature evaluation method based on intra-class,inter-class divergence,Pearson correlation coefficient and classification accuracy was established.By extracting multi-class time-domain,frequency-domain and time-frequency-domain features,a multi-category feature set was constructed,and dimension reduction fusion was carried out by multi-objective optimization of nuclear principal component analysis.The performance of feature dimension reduction and fusion is analyzed on the vibration data set of combine assembly quality problem.The results show that the proposed method can achieve90.7% accuracy in the process of combine assembly quality diagnosis and achieve good results.A method for combine assembly quality detection based on improved whale optimization algorithm(WOA)and least square support vector machine(LSSVM)was proposed.In order to solve the problems of WOA such as local optimum and imbalance of search ability,a non-linear control factor and adaptive weight are introduced to improve the algorithm,and the universal applicability of the improved algorithm is verified by 8 benchmark test functions.The improved WOA was used to optimize the constructed multi-objective LSSVM classification diagnosis model,and the effectiveness of this method was verified by the test data of combine assembly quality detection.The results show that the classification accuracy of the diagnostic model for combine assembly quality based on improved WOA and LSSVM reaches95.5% and the standard deviation of accuracy is only 0.77%.The model can diagnose the assembly quality of combine parts stably and accurately,providing a feasible method for the inspection of combine assembly quality.A combine assembly quality inspection system is designed and developed,which consists of signal noise reduction analysis subsystem,problem feature extraction and fusion subsystem and assembly quality problem fusion decision subsystem.Based on the analysis of functional requirements and development environment of the system,the overall structure and each functional subsystem of the combine assembly quality inspection system are designed.The application of function subsystem in combine assembly quality inspection system was verified by combining practical application cases,and the assembly quality inspection of combine in example was completed.The application results of an example verify the validity of the proposed method and the practicability of the combined harvester assembly quality detection system based on vibration signal. |
PDF Full Text Request