采矿环境下深海钴结壳声学探测与识别方法研究

The surface of cobalt crust ore bed in deep sea is complex and undulating，the coverage rate of cobalt crust reaches about 50%. So in order to realize effective collection, the critical technology is real time measurement of the ore bed surface. A kind of new submarine cobalt crust acoustic detection and recognition method is researched in this project. The main research contents are as follow: the research on acoustic transmission characteristics in turbid water areas in mining environment and putting up acoustic detecting experiment system which is suitable for micro-topography detection and identification of cobalt crust in turbid water areas; the research on reverberation model of acoustic detection in short range based on ray theory; the research on target echoes detector in the non-Gaussian background by using Gaussianized reverberation technology; the research on anti-reverberation and target echoes detection method based on chaos theory; the research on the distinguishing method、elimination method and vacant point prediction theory of abnormal elevation value; the research on stable echo feature abstraction methods in time domain、frequency-domain、time-frequency domain and stylebook domain which is used in the complicated and difficult case, such as the case that the surface of cobalt crust is undulating、broken and crushing; the research on feature fusion theory and the best cobalt crust echo feature fusion method; the research on nonlinear dimensionality reduction and recognition method based on manifold learning combined with kernelized learning and cobalt crust echo feature dimensionality reduction and recognition model. The problem of improving effective collection rate of cobalt crust in deep sea will be solved by studying on the relative theories、simulations and experiments in this project.

深海钴结壳矿层表面起伏不平,覆盖率只有50％左右,对其实时测定是实现有效采集的关键技术。本项目提出采矿环境下基于声学方法进行海底钴结壳探测与识别方法。研究采矿环境混浊水域声学特性，建立混浊水域声学探测系统；基于射线理论，构建采矿环境下近程声学探测混响模型；应用高斯化混响技术，建立非高斯背景下的目标回波检测器，利用混沌理论,建立非线性动力学预测模型，对比分析确定最佳回波检测方法；研究采矿环境下异常高程值判别与剔除方法及空缺点预测理论，构建采矿环境下地形重构模型。通过对钴结壳、基岩声学特性测定，从时域、频域、时频域、样本域角度研究钴结壳表面起伏、破碎风化等复杂情况下回波特征提取问题；研究特征融合机理，探索钴结壳识别的最优特征融合方法；研究流形学习并结合核空间处理技术的非线性降维与分类方法，建立适合于钴结壳的非线性降维与识别模型。通过理论、仿真、实验研究，解决提高深海钴结壳有效采集率的关键技术。

钴结壳作为稀有金属，具有广泛的应用价值。本项目围绕深海钴结壳的超声探测进行研究，证明了深海采矿时的超声环境具有混沌属性，提出了利用伪逆法求解Volterra滤波器的权向量，基于混响的混沌属性构建PIM-Volterra预测模型，并对混响背景下的目标回波进行了检测；利用小波降噪法，对信号中所含的干扰噪声进行了抑制，以信号间Itakura距离为评价参数，论述了模拟采矿环境下混响的局部平稳性，并确定了以分段预白化匹配滤波器为采矿环境下目标回波信号的检测方法，对比了两种目标回波预测方法的优劣势；以数字微地形的重构为研究内容，确定以渐近采样方式的采样策略，并采用25mm的规则方形网格对地表进行采样，建立了基于统计原理的高程数据粗差剔除方法；基于混响背景下的主动声呐方程，分析了声呐系统参数对声呐方程各参数的影响，建立了模拟采矿环境下的声传播模型、混响级模型和海底反射损失模型，并确定了声呐系统频率及脉冲宽度；研制了超声换能器，并通过仿真和实验证明了KZK方程在描述混浊水域中换能器轴向声场的有效性；在分析常用声呐探测波形的抗混响性能的基础上，利用S•Kay理论设计了不同混响环境下的探测波形；从时域、频域、样本域提取了信号的多类特征；研究了非线性降维方法的性能，选择KFDA对特征进行降维，并利用遗传算法对特征进行选择，设计了多个非线性分类器，并利用改进的DS理论对多分类器进行了融合，从而建立了钴结壳的识别模型。项目组的研究成果对于深海钴结壳的开采有重要的研究意义。.项目在国内外重要期刊或会议上发表或录用学术论文6篇，申请实用新型或发明专利4项，培养硕士研究生5名。