Warning Algorithms And Applications Based On Relationship Recognition

With the development of the technology of cloud computing and networking technology, a variety of advanced technology and sensor have been widely used in data acquisition. This makes multi-source heterogeneous data continues to grow and accumulated exponentially, data have considerable size in volume and species, the type of data have also extended from structured to unstructured, and that the link between data has been more colorful. However, due to the complexity of the size, type and speed increasing of data, traditional analytical methods have been unable to analyze and process the useful information implied in the vast amounts data at present. On the one hand, modeling and analysis capabilities of traditional methods have lost value because the short of traditional methods lead to "multi-source heterogeneous" data can not be used for decision-making. On the other hand, the modeling of complex systems and physical processes are still difficult. If the modeling reduce variable, the model is too simplified; If the modeling increase variable, the model is too complex to be handled. In the context of such a big amount of information, the problem faced by all the industry is how to get the value from the large scale and variety of uncertain data.The paper start with the of complexity of big data analysis decision combined with traditional data mining method, proposed the early warning technology based on the correlation recognition, and analyzes big data in all industries. The main contents in the paper are as follows:(1) In the context of current big data, the paper research and refine scientific issues of big data, select the correlation recognition method by contrasting a variety of related relationships, and designs the early warning algorithm based on the correlation recognition.(2) For the current value of the low density of big data sets, the paper researches the method of data quality evaluation, establishes data quality evaluation model, and gives the implementation steps of the data quality evaluation. Meanwhile, the paper selects appropriate data to evaluate aimed at different the data characteristics of different industries.(3) The paper studies the industrial applications based on the correlation recognition forewarning algorithms, analyzes the logical relationship between the number of clues existed in big data by different types of data backgrounds, identify the key factor impacting the rate of battery failure and voltage fluctuations, and verifies the validity and accuracy of the algorithm by the applications in the actual system.The paper presents that correlation recognition forewarning technology can find hidden factor in the massive data, reveal regularity of data, explore the data value, identify early warning indicators, and form forewarning indicators-pattern recognition-warning. The early warning technology can provide a basis for business decisions, help to enhance the level of enterprise decision management, and make significant economic benefits.