融合成熟度无损检测的苹果最佳贮藏期预测模型与方法研究

The problem of lack the theoretical model and quantitative basis for the storage and preservation of apple in China was proposed leading to insufficient maturity or over- maturity of fruit before delivery of cargo from storage. Research on non - destructive maturity detection method is the key for achieving the optimal storage period and improving the quality of apple storage and storage. Fuji apple was used as the research object, the sampling method and physical and chemical properties of different maturity samples were designed. The non-destructive detection method of apple maturity with characteristic spectra of coupled multi-physiological indexes was explored fusion of sugar content, hardness, acidity, starch and other maturity associated with physiological indicators of spectral characteristics and discriminate model. Then, the maturity model of genetic support vector machine based on optical characteristics is established. Further, an experimental scheme was proposed with non-destructive method to carry out the storage period of apple multi-environmental factors nested acquiring the whole continuous data of storage maturity variety. The aim is to establish a time series prediction model based on nonlinear dynamic neural network and propose an optimal storage time prediction method based on wavelet transform. On the basis of, the test platform of multi-sensors information fusion is built and the model and method is verified and perfected. This study showed that the method and model of storage maturity prediction fusion non-destructive detection method providing the basic theory and decision-making method to enhance the efficiency of apple storage and preservation. In addition, it is an important significance for promote the precision management of modern fruit industry in China.

针对我国苹果贮藏保鲜缺乏理论模型与定量依据，导致出库果品成熟度不足或过熟的问题，研究成熟度无损检测方法，进而探明贮藏期成熟度变化规律是确定最佳贮藏期，提升苹果贮藏保鲜质量的关键。以富士苹果为研究对象，设计不同成熟度样品采样方法与理化特性试验，融合糖度、硬度、酸度、淀粉等成熟度关联生理指标光谱特性与判别模型，探寻耦合多生理指标特征光谱的苹果成熟度无损检测方法，建立基于光学特性的遗传支持向量机成熟度判别模型；开展苹果贮藏期多环境因子嵌套的贮藏期成熟度无损检测全程连续试验，建立基于非线性动态神经网络的贮藏期成熟度时序预测模型，提出基于小波变换的苹果最佳贮藏时间预测方法；在此基础上，构建多传感器信息融合的试验平台，进行模型与方法的验证完善。研究成果阐明融合无损检测方法的贮藏期成熟度预测模型与方法，为提升苹果贮藏保鲜效益提供基础理论与决策方法，对推动我国现代果业精准化管理具有重要意义。

采收成熟度和采后贮藏果实品质是决定苹果口感、风味和营养的关键因素，两者直接影响着苹果质量等级、商品价值和经济效益。如何准确预测苹果最佳贮藏时间，保证其商品性尤为重要。本项目主要围绕采前苹果成熟度分类模型方法和采后贮藏期果实最佳贮藏时间预测方法研究，并开展基于多传感器融合的智能决策算法研究。.基于可见和近红外光谱法，以淀粉指数作为成熟度指标来确定袋装富士苹果的成熟度，将846个苹果分为三个成熟度级别（未成熟，采收成熟度和食用成熟度）。通过提取光谱数据的特征波长，并利用五种机器学习算法建立校准模型。通过比较不同建模方法的结果，基于15个特征波长的RF-SPA-LSSVM模型的预测性能最佳，预测集的准确度为89.05％。结果表明，富士苹果的成熟度可以通过Vis-NIR光谱进行无损预测。依据成熟度模型判别结果对450个苹果样本进行分类贮藏，开展贮藏条件下包括温度、CO2、O2及相对湿度四个环境因子的苹果后熟过程嵌套实验，利用建立的光学无损检测系统对贮藏样品以7天间隔进行环境参数监测.以获取的全程贮藏期成熟度试验数据为样本，建立了基于动态神经网络的苹果贮藏期成熟度预测模型，预测集的预测精度为85.62%。利用曲率法对连续预测数据序列进行特征点提取，得到了最佳贮藏时间目标值。选择样本不变的最佳贮藏时间预测模型，得到了在40-100天内较高的最佳贮藏时间预测模型，在此区间准确率在90%以上。通过构建的多传感器融合系统，实现了对温度、湿度以及二氧化碳浓度的实时监测，环境因子对模型的建立具有意义。进行模型的植入与测试，项目成果，经过进一步的改善与验证，可为苹果成熟度的无损判别与贮藏过程中品质动态监控提供技术支撑。.