Research On Personalized Recommendation Pattern Of Electric Nursing-bed Movement Control

Personalized recommendation has always been the research focus of the data mining and analysis. It is an effective way to reduce data load, improve service quality and user experience. With the increase of the population of the old, the health care to the old is becoming one of a most important problem in the society. The typical manual health care method should be transformed under the pressure of the improvement of life quality. Using automatic and intelligent devices is a better way to solve the above problem. It is an effective way to introduce the personalized recommendation into the field of nursing-bed movement control. Aim to the weak intelligence and humanity of current electric nursing-bed control modes, the personalized recommendation movement control mode is proposed, which is based on the existing manual, timing and speech control pattern. It is a new method that could reduce the labor intension of nurses, increase the self-help capability of patients. This work was financially supported by Science&Technology Project of Guangdong Provincial (2010A030500006) and Science&Technology Project of Foshan.First, the paper describes the background knowledge, research significance, the research status of nursing-bed and personalized recommendation system.Second, according to analyze the principal recommendation algorithm in the traditional recommendation system, to find the differences between personalized recommended system of electric nursing-bed and traditional recommendation system. As its characteristics, selects the cluster analysis technology as the basis of the recommendation algorithm; Depth study of similarity distance calculation method of cluster analysis, according to the advantages, disadvantages and specific use of clustering algorithm, partition clustering is selected as a basic recommendation algorithm. Common clustering methods include:based on partition clustering, based on hierarchical clustering, density-based clustering, model-based and grid-based clustering. And a detailed analysis of several indicators of effectiveness, IGP index is chosen to evaluate the optimal number of clusters. It provides a theoretical basis for parameter selection of clustering algorithm. Third, two classical clustering algorithm are detailed analyzed in this paper: K-Means and Affinity Propagation. Randomly generated50data points as the experimental dataset, the clustering result of K-Means is variable, the different initial cluster centers lead to different result. In addition, it has not yet stabilized after200iterations. However, AP clustering algorithm does not need to pre-specify the initial clustering centers, it is stabilized only after39iterations. Then analyzes the impact of two key parameters on the clustering results, the experiments shows that AP clustering algorithm is significantly better than K-Means algorithm in the case of lack of prior knowledge. Then according to the mechanical and electric constraints, some rules are established to adjust the clustering centersr,and the specific steps of improved AP algorithm is given.Fourth, the nursing-bed is divided into four independent bed-panel to explain the type of movements, and then designs the whole idea of the nursing-bed control system. The drive system of nursing-bed including data acquisition, analysis and processing functions. ATmegal6as the core chip. And a complete set of protocols is build to achieve communication between the host computer (man-machine interface controller) and the slave machine (motor drive system). The slave machine takes the design idea of gradually refined and top-to-down, mainly provides data transceiver, data analysis, signal detection function. The software of host computer is designed as three architecture model, a generic entity class is created by LINQ to SQL to transfer data and manipulate databases. A complete system is established from hardware design to software development. The system workflow is user input-> depth analysis-process control signal-> linear motor motion. The control system of nursing-bed provides a necessary experimental platform for researching personalized recommended pattern of electric nursing-bed movement.Fifth, the control system of nursing-bed is tested, experiments show that the transformation between movements are accurate and reliable, the speed is suitable during executing movement (Pose transform takes about20-30seconds). And the recognition rate of speech is more than95%. the system has a high stability and effectiveness. The nursing-bed movement sequence can be obtained by improved AP cluster algorithm, which is logical. The rationality of the recommended movement sequence is analyzed according to the distribution characteristic of the dataset. The experimental results show that the recommended movement sequence can basically reflect the user’s habits, which is more intelligent and human than other control modes.Finally, summarizes the work carried out and gives suggestions for follow-up study.