Storm-based coal mining machine operation state real-time prediction method

Abstract

The invention relates to the technical field of coal mine safety, and discloses a Storm-based coal mining machine operation state real-time prediction method, a Storm-based coal mining machine operation state data distributed real-time prediction framework is combined with a Hadoop Database database, and operation data of a time sequence in a coal mining machine in practical production is simulated. And on the basis of Storm, various data predicted by the prediction model GRU is combined with a threshold value of each data to realize early warning of a data state. And after the GRU is trained in a training set, the performance of the GRU in a test set is evaluated from three aspects of RMSE, MAE and R2, and it is proved that the prediction model GRU can be suitable for prediction of coal mining machine state data. According to the method, a traditional monitoring method for the running state of the coal mining machine is broken through, a prediction and early warning method for the real-time running state of the coal mining machine is integrated, and a Storm distributed real-time processing framework has great practical value in parallel processing of mass coal mine data.

Description

technical field [0001] This application relates to the technical field of coal mine safety, in particular to a Storm-based method for real-time prediction of the operating status of a coal shearer. Background technique [0002] As one of the three fully-mechanized mining machines, the shearer has a complex working environment. With the development of intelligent coal mines, the shearer itself is equipped with multiple sensors, and the sampling frequency is high. The data collected every day increases by the order of PB. Real-time collection and analysis of operating status data, predicting data changes at the next moment and whether the status of the shearer is abnormal can not only make effective use of data, but also ensure the safety of the shearer and personnel to a certain extent, so the status of the shearer can be realized The real-time monitoring and prediction of data is of great significance to the intelligent operation of coal mining machines. [0003] In the pas...

AI Technical Summary

Problems solved by technology

[0003] In the past, in the process of studying the operation state of the shearer, most of the monitoring data of the shearer were simply analyzed through mathematical statistics, and there were few studies on how to predict the operation state of the shearer in real time, which had the following disadvantages: (1) Practical process It is difficult to use the traditional preprocessing method to judge whether the abnormal data is useful data that can extract equipment status information or useless data that can be cleaned, and it cannot adapt to the dynamic characteristics of time series data. It is necessary to establish a preprocessing cleaning method that can adapt to the characteristics of time series The model can not only identify useless outliers of coal shearers, but also dynamically repair them; (2) Traditional machine learning methods such as ARIMA, SVR, hidden Markov model and other algorithms can predict time series data, although the accuracy can meet requirements, but cannot meet the requirements of big data and is not easy to apply in practice

Although the deep learning method performs well in big data, it is rarely applied to coal mine data. The coal mine data is complex and changeable, and there are a lot of unreal data. Reasonable preprocessing is required to train and predict through the deep learning model

(3) The data volume of the coal mining machine is huge. Realizing the intelligentization of the coal mine needs to realize the parallel real-time processing of the sensor data. The existing coal mine data research is based on offline data processing, and the hysteresis of data processing leads to a sharp drop in the value of the data.

[0004] Storm is an open source distributed real-time computing framework. It can easily process unlimited data streams. It is widely used in power grids and big data, but it is rarely used in real-time processing of coal mine data. In terms of processing coal mine equipment data, Ma Hongwei Aiming at the large amount of coal mine fully-mechanized mining equipment operation status data, data noise and missing values, etc., a MapReduce-based coal mine fully-mechanized mining equipment operation status big data cleaning model was established, but Hadoop's MapReduce could not meet the real-time requirements; Cao Xiangang Aiming at the problems of noise points and missing values ​​in the operating state data of coal mining machines, a Storm-based real-time data cleaning platform was established, but the prediction and early warning of the state data of coal mining machines were not realized through Storm

Images