Research On Accurate Detection Methods Of Hardwood Log Quality Through Stress Wave Technology

Hardwood logs are ideal raw materials for home decoration industry and construction industry because of its rich texture and colors and high mechanical properties.The quality of hardwood logs varies widely within species,harvest site,and even the same tree.The location,type,and size of defects on hardwood logs dictate the potential grade and value of the resulting lumber and affect the utilization ratio of logs.Accurate quality inspection of hardwood logs usually includes multi-stage quality grading and accurate defect detection.Therefore,early quality detection and grading of hardwood logs in logging area could achieve a quick diversion conveniently to save transportation cost.On the other hand,detecting internal defect in hardwood logs accurately in mill could help to improve sawing procedures and increase the board recovery and maximize the profits.Stress wave technique is the main means of log quality detection,but it is difficult to obtain its typical characteristics in time domain or frequency domain for a complex response signal acquired,resulting in a finite precision in quality assessment and difficult to popularize and apply.In view of this,this paper focuses on analyzing the stationary and non-stationary signal processing methods on the impact stress wave signal,put forward the application of modern signal processing methods such as time-frequency analysis,wavelet transform,moment analysis,autoregressive model to extract acoustic characteristic parameters including time center,velocity,damping ratio,kurtosis,spectral kurtosis,and carries out a research in the two aspects of log quality classification and internal defects prediction.The main work and research results are as follows:The attenuation properties of stress wave propagation simulated by small damped vibration system was studied to obtain the theoretical basis of log quality evaluating by acoustic parameters is given.Based on the assumption of log equivalent to the uniform straight pole,the wave equation of longitudinal free vibration of log was derived,by which the quantitative relationship between elasticity modulus and acoustic velocity and medium density was obtained.Then,based on the free boundary conditions for both ends of logs,the equation of stress wave propagation velocity,natural frequency and rod length was derived.The influence of internal longitudinal force in log on the vibration system was discussed,and the attenuation of stress wave propagation in the log was simulated by the microelement of log approximately equivalent to a single degree-of-freedom mass-spring damped system.Through discussing the quantitative relationship between damping vibration period,amplitude attenuation,energy dissipation and damping ratio and analyzing the relationship between acoustic characteristic parameters and vibration attenuation of a log,the theoretical basis for evaluating log quality using the acoustic parameters related to signal waveform was given.An accurate log quality grading strategy based on multiple acoustic parameters was established,by which the quality of a log with minor defects couldnot be graded by acoustic velocity was supplemented effectively.The physical significance of the first moment based time centroid and damping ratio was analyzed from the perspective of energy dissipation,and then the significant influence of typical acoustic signals on the time centroid and damping ratio was analyzed.The extraction process of damping ratio based on continuous wavelet transform was derived in detail,and quality grade of yellow poplar log was predicted by using the acoustic parameters such as the time centroid,damping ratio,two mixed parameters and acoustic velocity.The grading results showed that the high grade board rates in the high-quality log group predicted by the first four parameters were 74.2%,74.1%,79.6%and 69.0%respectively,much higher than that of 43.9%predicted by the acoustic velocity.In view of the performance differences of several acoustic parameters on quality assessment and the influence of log geometry defects on assessment parameters,a multi-parameter joint prediction scheme for accurate quality grading of hardwood logs was proposed.The defect signal separation and enhancement algorithm was studied,and classification of log quality based on kurtosis was realized.The possible components of sound source signal were discussed,and then the acoustic signal was modeled as the convolution between the response function of transmission path with the combined signal of periodic components,defect components and background noise.Two signal separation and enhancement algorithms based on AR-MED and AR-SK were proposed and the realization process of the MED algorithm was given in detail,and then the signal separation and enhancement process using AR-MED method was demonstrated with an numerical example.The process of determining the center frequency and bandwidth of the filter according to the short-time Fourier transform based Kurtogram was given and the effective separation of defect signals was realized based on the optimal bandpass filter.Then the fourth moment was calculated for each separated defect signal to extract the kurtosis reflecting defect characteristics,and then the kurtosises were used to predict log quality.The high grade board rates in the high-quality log group predicted by kurtosis of K_MED and K_SK,extracted by based AR-MED and AR-SK,reached 72.5%and 77.2%,respectively.The prediction model of defect ratio based on multi-acoustic parameters was established to realize accurate prediction of log defect ratio.In order to realize the quantitative detection of the log internal defects,the quantitative relationship between the internal defects of four species of mixed logs（black cherry,white oak,red oak and cottonwood）and the acoustic parameters（time centroid,the first-order damping ratio,the second-order damping ratio and acoustic velocity）was studied.Radial stress wave transmission time was used to determine the locations of internal defects of log to construct defect map,and by which the defect ratio of log was calculated.The quantitative relationship between individual acoustic parameter and internal defect ratio of log was studied,and the influence of tree species,defect distribution and defect type on acoustic parameters was analyzed.The regression results showed that the determining coefficients between the first-order damping,sound velocity,time center,the second-order damping ratio and the defect ratio were 0.65,0.72,0.87 and 0.92,respectively.The quantitative relationship between log defect ratio and multiple acoustic parameters was studied by using multiple regression method and established the optimal acoustic parameter model for defect ratio prediction.The determination coefficient and the mean square error of the predicted model is0.95 and 6.47,respectively.The quasi-defect signal characteristics filtered by AR-MED were extracted to realize accurate prediction of log quality.The quantitative relationship between internal defect ratio of four species of hardwood logs and kurtosis extracted by AR-MED method was studied,and the kurtosis based prediction model of defect ratio,the determination coefficient of 0.89,was established by regression analysis.A spectral kurtosis method based on complex Morlet continuous wavelet transform was proposed to further determine the main types and primary and secondary of internal defects in logs.Then,the selection method of the optimal wavelet center frequency for defect signal decomposition was studied and the selection principle of frequency band of defect components based on kurtogram was discussed.The characteristic frequency band of defect signal based on spectral kurtosis was extracted,and the corresponding relationship between signal frequency bands and defect types was analyzed,and then the accurate identification of main defect types in logs was realized.Finally,the limitation of global parameters（kurtosis,damping ratio,time centroid,acoustic velocity,etc.）in defect detection was analyzed,and the advantages of spectral kurtosis parameters in defect identification were obtained compared with global acoustic parameters.