The Research On Coal-bed Methane Well’s Parameter Prediction And Fault Diagnosis Based On Time Series Analysis

There is a huge demand for energy because of the fast development of industrialization in our country. In the context of sustainable development policy, the new environmentally friendly fuel or energy is advocated more by the government. Coal-bed methane, a new environmentally friendly type of energy, plays a very important role in the national resource structure. With the large-scale production, high security requirement is especially necessary to the development of coal-bed methane single well. We hope a rapid, correct prediction can be made about the state of the system parameters to let the field workers can make a good plan on the producing process, and also can make a quick fault diagnosis when the coal-bed single well has a fault to decrease the possibility of fault in the bud and protect the property and lives, the single well, as the basic part of the coal-bed gas producing system is crucial to the safe use of the equipment and even the subsequently transferring and supercharger of coal-bed methane. The traditional monitoring method depends on the labours’polling which is time-consuming and low efficiency, so the study of an intelligent algorithm is particularly important.According to the actual survey of the single-well field, we find that the data we can collect is only the producing parameters and it will have a trend change when there a fault, there exists some correlation between gas producing parameters and the parameters have slowly abnormal trend change in different fault status, so the traditional classification method based on the original data such as the Support Vector Mechine (SVM) and neural network and so on is not suitable for the fault diagnosis of the coal-bed Methane Well system, which requires us to find a new fault diagnosis method to solve this problem.The wells are widely distributed which means that there will be difference between different wells. The location of the wells are very complicated which means the producing system of different well are influenced by many factors, which involves both the known and the unknown, so the producing systems are considered to be a grey process. The many parameters’variety are thought to be also a grey process, so we need to come up with a method to deal with the grey characters.Aimed at the new problems arising in the producing process of the coal-bed methane well, this paper proposes a method on the parameters’prediction and fault diagnosis based on Grey theory and time series theory. First, we use the GM (1,1) model belong to the grey theory to build the grey model for each producing parameter to get the trend part of each parameter’s time series. Then, in order to improve the accuracy of prediction, we propose to build the AR model for the residual series of each parameter. At last, we put the two parts together to obtain the combined model of each parameter which is used to make the prediction and fault diagnosis. The innovation of this paper is the use of multi-parameters combined models to measure the working state of the single well and make the fault diagnosis with the characteristic parameters in the model. The simulation results show that this method has good practicality, and can be applied to both parameters’prediction and fault diagnosis of single well.