Study On Profiling And Ranging Of Cutter Head Of Sugarcane Harvester

Sugarcane is mainly planted in the southern subtropical region,it is one of the three major cash crops in China,and has a pivotal position in Chinese sugar industry.For a long time,the mechanization level of sugarcane production in China has been very low,and its development is mainly limited by the mechanized harvesting field.Cutting blade profiling is a prerequisite for the mechanization of sugarcane harvesting,and the accurate acquisition of the cutting blade position height is the key to achieve profiling.To solve this problem,the specific research contents of this thesis are as follows:(1)A field investigation and analysis of sugarcane cultivation in southern China,a cutter disc profiling ranging scheme is proposed,and multiple ultrasonic sensors are used to distribute measurements to obtain the distance height.In order to verify the feasibility of the knife-disk profiling scheme,a set of experimental devices for real-time acquisition of sugarcane topographic data was designed,which included a data acquisition frame and a real-time topographic data acquisition system.The data acquisition frame was designed,machined and assembled.In the data real-time acquisition system,the sensors were selected,the calibration experiment was completed by using the least squares method,and the upper computer program was prepared to realize the serial communication of multiple sensors.The experimental device was used to go to the sugarcane base of Guizhong Farm in Xiushui Village,Luerong Town,Liuzhou City,Guangxi District for data measurement.(2)In order to solve the problem of large errors in sensor measurement of sugarcane terrain height,research on sensor measurement data processing methods is conducted,which includes data pre-processing and data fusion of multiple sensors.The noise analysis of the original measurement data of the sensor in profiling distance measurement was carried out to study the application of common filtering algorithms and the threshold higher-order filtering method based on wavelet transform to filter and denoise the ultrasonic sensor measurement data.The sensor’s raw data is first filtered by median average filtering and then wavelet filtering,which solves the distortion of port data by median average filtering,and the periodic random noise and spike noise in the raw data are basically filtered out,the smoothness is significantly improved,and the accuracy of measurement data is improved.The time complexity of the filtering algorithm is analyzed,and the running time of the algorithm is only 16.7 ms,which also ensures the requirement of real-time performance in profiling distance measurement.(3)In order to solve the problem of limitations of individual sensors in profiling and further obtain the accurate position height of the cutting blade,data fusion is performed on the pre-processed data from multiple sensors.Two typical algorithms in the data fusion algorithms of stochastic inference and artificial intelligence are selected for the study,including Kalman filter algorithm and adaptive weighting algorithm in the stochastic inference category and deep learning neural network algorithm in the artificial intelligence category.The Kalman filter algorithm is theoretically investigated,the principle of the neural network algorithm for deep learning is analyzed,the data set of the LSTM neural network model is preprocessed and hyperparameters are selected,and a six-input,one-output seven-layer LSTM model is established to predict the measurement data.The adaptive weighting algorithm is theoretically analyzed and then improved twice,and the improved algorithm can well achieve the assignment of optimal weighting factors for multiple sensors.Experimental analysis of the improved adaptive weighting algorithm shows that the algorithm can well fuse the measurement data of multiple sensors in profiling ranging,and the average absolute error of the fused data is only 0.3351 cm,and the variance is only 0.1687,which further improves the accuracy and reliability of the sensor measurement data.(4)The feasibility of the cutter disc profiling program was further verified by completing the profiling experiments by measuring the monopoly height in the field.The actual measured distance from the cutting blade to the monopoly furrow and the height of the monopoly were compared to obtain the final fusion value of the cutting blade position and height,and the average absolute error was only 0.4068 cm compared with the actual measured value.The influence of the sensor distribution on the accuracy of the profiling data was further investigated,and the experimental results showed that the profiling data accuracy was the highest when the sensors were vertically distributed.