1, boiler fault prediction related knowledge
Artificial intelligence fault diagnosis and prediction technology is a new technology with the rapid development of modern science and technology and economy. It can identify whether the state of equipment is normal, find and determine the location and nature of faults and put forward corresponding countermeasures, thus improving the reliability of equipment operation, prolonging its service life and reducing the life cycle cost of equipment. Fault prediction technology can realize early detection of faults, predict their future development trend, facilitate timely adjustment of thermal power units, avoid the occurrence of vicious accidents, make the units run safely and reliably, and improve the economy of the units.
Fault prediction can be divided into three types according to the length of prediction period. Long-term prediction, which is used to make long-term maintenance plan and maintenance decision of boiler units, usually takes more than one month, and the prediction accuracy is low. Medium-term prediction is to predict the state of boiler units in a long period of time in the future, which serves the medium-term maintenance plan and maintenance decision of units. The time is generally about one week, so the prediction accuracy is low. Short-term prediction takes about one day to predict the recent development of boiler units, which requires high prediction accuracy. For medium and long-term forecasting, because the accuracy requirement is not high, we can consider adopting a simple forecasting model and establishing a univariate time series model for forecasting. For short-term forecasting, it requires high accuracy, and at the same time, because various related factors have great influence on the state value at that time, in short-term forecasting, besides the time series itself, other related factors should be considered, which requires the establishment of a multivariate time series model for forecasting to meet the accuracy requirements of short-term forecasting.
2, the accuracy of fault prediction requirements
Intelligent fault diagnosis and prediction technology is a new technology developed with the rapid development of modern science and technology and economy. Because it can identify whether the state of the equipment is J-normal, find and determine the location and nature of the fault, and put forward corresponding countermeasures to improve the reliability of the equipment operation, prolong its service life and reduce the life cycle cost of the equipment. Using the fault prediction technology, it can also realize the early detection of faults and predict their future development trend, so as to adjust the J -2 unit and related equipment in time, avoid the occurrence and expansion of vicious accidents, ensure safe and reliable operation and improve the operation economy of the unit. As far as the prediction period is concerned, fault prediction can generally be divided into long-term prediction, medium-term prediction and short-term prediction according to the different prediction periods.
2. 1 Long-term forecast
It is a prediction of making a long-term maintenance plan and maintenance decision for boiler units, which usually takes more than one month, so the accuracy of the prediction is low.
2.2 Medium-term forecast
It predicts the state of boiler units in a long period of time in the future, and serves for the medium-term maintenance plan and maintenance decision of units. Generally, the time is about one week, and its prediction accuracy is also low.
2.3 Short-term forecast
It generally takes about one day to predict the recent development of boiler units, and its prediction accuracy is required. For medium and long-term forecasting with low accuracy, we can consider adopting simple forecasting model and establishing univariate time series model for forecasting. For short-term forecasting with high accuracy, various related factors have great influence on the state value at that time. In short-term forecasting, other related factors should be considered besides the time series itself. Therefore, it is necessary to establish a multivariate time series model plow for forecasting to meet the accuracy requirements of short-term forecasting.
3. Common boiler fault prediction methods
In recent years, many researchers use linear regression analysis, time series analysis, grey model prediction, expert system, artificial neural network and other methods to study the fault diagnosis of boiler equipment, in order to explore fast and effective fault diagnosis and prediction methods. Commonly used forecasting methods are:
3. 1 linear regression analysis method
Regression analysis is a method to find the mathematical relationship between several uncertain variables and make statistical inference. The simplest of these relationships is linear regression analysis.
3.2 Time series analysis method
Time series refers to a group of data arranged in time sequence, and time series analysis refers to a data processing method that uses parameter model to analyze and process the observed ordered random data. Time series analysis methods mainly include curve fitting, exponential smoothing, seasonal model and linear stochastic model, which are mainly suitable for single factor prediction. However, when the boiler fault prediction is a multi-factor prediction with both deterministic trend and certain randomness, it is necessary to separate the deterministic trend and the calculation is complicated. At the same time, it is necessary to assume the zero mean and stability of the separation residual, and its prediction accuracy is not high.
3.3 Grey model prediction method
Grey model prediction method is to establish a prediction model based on grey system theory. It establishes a general grey differential equation according to the universal development law of the system, and then obtains the coefficients of the differential equation by fitting the data sequence, thus obtaining the grey prediction model equation. There are two main methods to apply the grey system theory to fault prediction, one is the grey prediction model based on the dynamic equation gm (or dm) of the grey system, and the other is the residual identification prediction model based on the residual information data sequence. Among them, gm prediction model is a differential equation with 1 order and 1 variables, which is commonly described as a grey model. From a mathematical point of view, the solution of grey prediction is equivalent to the superposition of power series, which contains the contents of general linear regression and power series regression, so the grey prediction model is better than general linear regression or exponential curve fitting, and also better than deterministic time series analysis.
3.4 Expert system
Expert system can successfully solve problems in some specialized fields and has many advantages, but after years of practice, it is always far from reaching the level of experts, and sometimes it is not as good as a beginner in some issues. Analyzing the reasons, there are mainly the following aspects: the "bottleneck" problem of knowledge acquisition; The limitation of single reasoning mechanism that simulates the expert's thinking process; The system lacks self-learning ability.
3.5 artificial neural network prediction method
There are many problems in neural network fault diagnosis. It can't make good use of the accumulated experience and knowledge of experts in the field, only uses some clear fault diagnosis examples, and needs a certain number of samples to learn. After training, it finally gets some threshold matrices and weight matrices, rather than logical reasoning like expert experience and knowledge, so it lacks the ability to explain the diagnosis results, can not be applied to real-time diagnosis, and can only process historical data.
3.6 Combination of Expert System and Artificial Neural Network
The combination of expert system and artificial neural network is a hot research topic at present. Neural network expert system, which is composed of neural network and expert system, can use the characteristics of large-scale parallel distributed processing and knowledge acquisition automation of neural network to solve the knowledge acquisition problems such as weak reasoning ability, poor fault tolerance and difficulty in dealing with large-scale problems, realize parallel association and adaptive reasoning, improve the intelligence level of the system, and make the system have real-time processing ability and high stability. Compared with the traditional expert system, the expert system based on neural network has the following advantages: it has a unified internal knowledge representation, and any knowledge rule can be stored in each connection right of the same neural network through learning examples, which is convenient for the organization and management of knowledge base and has strong universality; Knowledge capacity is large, and a lot of knowledge can be stored in a much smaller neural network; It is convenient to acquire knowledge automatically and can adapt to the change of environment; The reasoning process is a parallel numerical calculation process, which avoids the problems of slow reasoning speed and low efficiency. Have the ability of thinking in images such as association, memory and analogy. , and can work outside the scope of knowledge; Knowledge representation, storage and reasoning are integrated, that is, they are all realized by a neural network.
Concluding remarks
In short, because most workers have been working underground for a long time, they have to take a bath, and besides heating in winter, except those who can use the waste heat of power plants, they are generally inseparable from steam boilers, and some use water-heating boilers for heating. And the number is still large, many medium-sized coal mines have as many as seven or eight boilers. Therefore, it is very important to keep the normal operation of the boiler and ensure the normal production. It is necessary to predict the boiler failure.
For more information about project/service/procurement bidding, and to improve the winning rate, please click on the bottom of official website Customer Service for free consultation:/#/? source=bdzd