中文摘要：

以港口岸边集装箱起重机装卸操作过程为研究背景，基于隐马尔科夫模型（HMM），建立双层HMM模型，运用修改后的Forward-Backward算法，计算操作层各个HMM模型的似然度，选择似然度最大的模型作为操作行为的识别结果，组成观察序列串后，送入意图层HMM模型集，进行桥吊司机操作意图的识别。最后，采用Matlab环境实现HMM模型，通过现场统计确定案例基础数据，验证了模型的有效性。结果表明，该模型可准确识别岸边集装箱起重机司机的操作意图，对于研究港口机械智能辅助驾驶系统具有一定的意义。

英文摘要：

With the background of loading and unloading process of quayside container crane, upon Hidden Markov Model, double HMM model is established The algorithm of revised Forward-Backward is applied to calculate each likelihood of HMM in operation layer, the model of the largest likelihood is selected to be the identify result of operation behavior. After combining them to constitute the observation sequence bunch, it will be sent to the intention layer of HMM to conduct the identification of operation intention of crane driver. Finally, HMM is realized by Matlab. By means of field statistics, the basic data can be determined and effectiveness is also verified. It turns out that this model can accurately identify the operational intention of quayside container crane drivet, which is of great significance for studying intelligent auxiliary drive system of port machinery.