Dynamic early warning method, system and equipment for induced draft fan parameters based on historical data mining and medium

Abstract

The invention discloses an induced draft fan parameter dynamic early warning method, system and device based on historical data mining and a medium, and relates to the technical field of coal-fired unit intelligent monitoring. The method comprises the steps that characteristic parameters of an induced draft fan are collected, and the characteristic parameters of the induced draft fan are preprocessed; according to the preprocessed data, an induced draft fan characteristic parameter prediction model is established, and an induced draft fan characteristic parameter prediction value is obtained; establishing an early warning threshold value according to the characteristic parameter predicted value of the induced draft fan, and dynamically adjusting the early warning threshold value; establishing an early warning model, and performing early warning on the abnormal characteristic parameters of the induced draft fan according to an early warning threshold value; according to the method, the characteristic parameters of the induced draft fan are collected, data preprocessing is carried out, the early warning threshold value is dynamically adjusted in combination with the predicted value and the environmental factors, the adaptability and accuracy of early warning are improved, the intelligent early warning model is constructed, and targeted fault diagnosis suggestions are provided; the monitoring precision, the early warning accuracy and the fault response efficiency of the operation state of the induced draft fan are effectively improved, and the operation safety and stability of a coal-fired unit are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent monitoring of coal-fired units, and specifically relates to a dynamic early warning method, system, equipment and medium for induced draft fan parameters based on historical data mining. Background Art

[0002] With the rapid development of the power industry, coal-fired units are still an important part of the global power supply. Especially in the context of energy structure adjustment and carbon emission reduction policies, improving the operating efficiency of coal-fired units, reducing energy consumption, and reducing pollutant emissions have become core issues of concern to the industry. As a key auxiliary equipment of coal-fired units, the operating status of the induced draft fan directly affects the combustion efficiency of the boiler and the stability of the smoke exhaust system. In actual operation, the changes in the induced draft fan parameters are affected by multiple factors such as load fluctuations, ambient temperature, and equipment aging. If the monitoring and early warning mechanism is not perfect, it may lead to a decrease in the fan operating efficiency, energy waste, and even cause serious equipment failures, affecting the safety and economy of the entire unit. Therefore, research on the monitoring and intelligent early warning of the operating status of the induced draft fan has received widespread attention in recent years.

[0003] Traditional induced draft fan monitoring methods mainly rely on fixed threshold settings and manual experience judgment, that is, fixed upper and lower thresholds are set according to the equipment manual or long-term operating experience, and an alarm is triggered once the parameter exceeds the threshold. Although this method can provide fault warnings to a certain extent, it is difficult for fixed thresholds to adapt to complex and changeable operating conditions, resulting in serious false alarms and missed alarms. For example, when the load is low, some parameter fluctuations may be normal, but when operating at high load, the same degree of parameter fluctuations may indicate potential faults. Therefore, fixed thresholds are difficult to accurately reflect the health status of the equipment.

[0004] The present invention collects the operating characteristic parameters of the induced draft fan, pre-processes the data by using methods such as data cleaning and outlier elimination to ensure the integrity and accuracy of the data, establishes a characteristic parameter prediction model of the induced draft fan based on the Gaussian mixture model (GMM), and estimates the parameters through the expectation maximization (EM) algorithm to adapt to multimodal data distribution and improve the prediction accuracy of the model. Different from the traditional fixed threshold method, the present invention dynamically calculates the warning threshold according to the prediction parameters, and adjusts the upper and lower limits in combination with the rated parameters of the equipment and the real-time operating status, so that the warning system can adapt to different operating conditions. Summary of the invention

[0005] In view of the above problems existing in the prior art, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is: the existing technology of the induced draft fan early warning method based on fixed thresholds has poor adaptability, high false alarm rate, lack of real-time performance, and is difficult to accurately reflect the operating status of the induced draft fan under different working conditions. It also solves the problem of how to realize adaptive dynamic threshold setting based on historical data mining to improve prediction accuracy and early warning accuracy.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a dynamic early warning method for induced draft fan parameters based on historical data mining, comprising: collecting characteristic parameters of the induced draft fan and preprocessing the characteristic parameters of the induced draft fan; establishing a prediction model for the characteristic parameters of the induced draft fan according to the preprocessed data to obtain predicted values ​​of the characteristic parameters of the induced draft fan; establishing a warning threshold according to the predicted values ​​of the characteristic parameters of the induced draft fan, and dynamically adjusting the warning threshold; establishing an early warning model, and issuing an early warning for abnormal characteristic parameters of the induced draft fan according to the early warning threshold; the prediction model for the characteristic parameters of the induced draft fan comprises constructing a data set based on historical data and preprocessing it, establishing a prediction model using a GMM model, estimating and training parameters through an EM algorithm, calculating the probability distribution of the characteristic parameters, extracting expected values ​​as prediction results, calibrating in combination with environmental variables and equipment aging factors, dynamically predicting and outputting parameter prediction values; dynamically adjusting the early warning threshold comprises calculating a real-time deviation range based on the predicted values ​​of the characteristic parameters of the induced draft fan, in combination with the rated parameters of the equipment and the operating conditions, and dynamically adjusting the early warning threshold; the early warning model comprises that when the unit is in a stable operating state and the induced draft fan is in an operating state, if the characteristic parameters exceed the early warning threshold, the system automatically triggers an early warning signal, and takes maintenance measures in response to the early warning signal.

[0008] As a preferred scheme of the dynamic early warning method of induced draft fan parameters based on historical data mining described in the present invention, wherein: the collection of induced draft fan characteristic parameters includes collecting induced draft fan current, induced draft fan bearing temperature, induced draft fan motor stator coil temperature, induced draft fan motor bearing temperature, and induced draft fan lubricating oil temperature.

[0009] As a preferred solution of the dynamic early warning method of induced draft fan parameters based on historical data mining described in the present invention, wherein: the preprocessing of the characteristic parameters of the induced draft fan includes value range cleaning of the characteristic parameters of the induced draft fan, dividing 70% of the cleaned data into a training set and 30% into a verification set.

[0010] As a preferred solution of the induced draft fan parameter dynamic early warning method based on historical data mining described in the present invention, wherein: the establishment of the induced draft fan characteristic parameter prediction model includes establishing the induced draft fan parameter dynamic early warning model through the GMM model, performing parameter estimation through the EM algorithm, training the induced draft fan characteristic parameter prediction model through the training set, inputting the verification set into the trained GMM model, calculating the mean square error, Score, mean absolute error, evaluates the prediction accuracy of the GMM model.

[0011] As a preferred solution of the dynamic early warning method of induced draft fan parameters based on historical data mining described in the present invention, wherein: the method of obtaining the predicted value of the characteristic parameter of the induced draft fan includes inputting the real-time characteristic parameter into the GMM model for operation, calculating the probability distribution of the characteristic parameter under the current working conditions, extracting the expected value of the characteristic parameter as the predicted value based on the probability distribution, and performing a secondary calibration on the predicted value in combination with the real-time ambient temperature and the equipment aging coefficient to obtain the real-time predicted value of the characteristic parameter.

[0012] As a preferred solution of the dynamic early warning method for induced draft fan parameters based on historical data mining described in the present invention, the early warning threshold is dynamically adjusted according to the predicted value of the characteristic parameter of the induced draft fan, including setting the characteristic parameter deviation according to the rated value and the real-time predicted value of the characteristic parameter as: , , in, is the upper limit of characteristic parameter deviation, is the real-time predicted value of the characteristic parameter, is the characteristic parameter rating, where is the lower limit of characteristic parameter deviation.

[0013] As a preferred solution of the dynamic early warning method for induced draft fan parameters based on historical data mining described in the present invention, the early warning of abnormal characteristic parameters of the induced draft fan according to the early warning threshold includes constructing an early warning trigger rule represented as: , in, is the real-time load of the unit, is the rated load of the unit, is the real-time current of the induced draft fan, is the rated current of the induced draft fan, is the real-time value of the characteristic parameter, It is the duration of feature parameter exceeding limit.

[0014] When the early warning trigger rules are met, the system automatically triggers an alarm signal and records abnormal data in real time. If the induced draft fan current exceeds the limit, check the motor load fluctuation, check the inverter working status, and adjust the moving blade opening. If the bearing temperature exceeds the upper limit, check the lubricating oil system, clean the bearing heat dissipation channel, monitor the vibration data, and check the stability of the mechanical structure. If the stator coil temperature exceeds the upper limit, reduce the unit load, check the cooling fan operating status, check the motor insulation aging and poor contact. If the lubricating oil temperature exceeds the upper limit, switch to the standby lubricating oil pump, check the heat dissipation efficiency of the heat exchanger, clean the dirt, replace the cooling medium, and replace the lubricating oil filter.

[0015] As a preferred solution of the dynamic early warning system of induced draft fan parameters based on historical data mining described in the present invention, it includes: a data acquisition and preprocessing module, a characteristic parameter prediction module, a dynamic threshold adjustment module, and an early warning model construction module; the data acquisition and preprocessing module includes a parameter acquisition unit and a data preprocessing unit, the parameter acquisition unit is used to collect the induced draft fan current, bearing temperature, stator coil temperature, and lubricating oil temperature, the data preprocessing unit is used to remove abnormal values ​​and normalize the collected data, and divide it into a training set and a verification set; the characteristic parameter prediction module includes a model training unit and a real-time prediction unit, the model training unit is used to The processed data is used to train a Gaussian mixture model (GMM), and the expectation maximization (EM) algorithm is used to optimize parameters. The real-time prediction unit is used to input the real-time operation data into the trained model to calculate the predicted value of the characteristic parameter of the induced draft fan; the dynamic threshold adjustment module is used to calculate the dynamic upper and lower limits of the parameters according to the predicted value and the equipment rated value; the early warning model construction module includes an abnormality detection unit and an alarm triggering unit. The abnormality detection unit is used to monitor the characteristic parameters in real time and determine the over-limit situation of the characteristic parameters. The alarm triggering unit is used to trigger a warning signal after the abnormal state continues for more than a set time, send an alarm message to the system, and take maintenance measures according to the alarm message.

[0016] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of any one of the methods for dynamic early warning of induced draft fan parameters based on historical data mining are implemented.

[0017] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of any one of the methods for dynamic early warning of induced draft fan parameters based on historical data mining are implemented.

[0018] Beneficial effects of the present invention: The present invention adopts data mining and machine learning technology, collects characteristic parameters of the induced draft fan and performs data preprocessing to ensure data integrity and accuracy, establishes a prediction model based on the Gaussian mixture model (GMM) combined with the expectation maximization (EM) algorithm, and realizes accurate prediction of the operating status of the induced draft fan. The warning threshold is dynamically adjusted in combination with the predicted value and environmental factors, breaking through the limitations of the fixed threshold method, improving the adaptability and accuracy of the warning, constructing an intelligent warning model, monitoring the changes in induced draft fan parameters in real time, and automatically triggering a warning when the abnormality exceeds the limit and lasts for a certain period of time, and providing targeted fault diagnosis suggestions. This method effectively improves the monitoring accuracy, warning accuracy and fault response efficiency of the induced draft fan operating status, reduces the risk of false alarms and missed alarms, and improves the safety and stability of the operation of the coal-fired unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0020] Figure 1 A schematic flow chart of a method for dynamic early warning of induced draft fan parameters based on historical data mining provided in the first embodiment of the present invention.

[0021] Figure 2 A historical data graph of characteristic parameters of an induced draft fan after cleaning according to a dynamic early warning method for induced draft fan parameters based on historical data mining provided in a second embodiment of the present invention.

[0022] Figure 3 A diagram of the dynamic upper and lower limit warning model of the induced draft fan motor current in a real-time monitoring system of the induced draft fan parameter dynamic warning method based on historical data mining provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0024] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a dynamic early warning method for induced draft fan parameters based on historical data mining, including:

[0025] S1: Collect characteristic parameters of the induced draft fan and pre-process the characteristic parameters of the induced draft fan.

[0026] It should be noted that the collection of characteristic parameters of the induced draft fan includes the collection of the induced draft fan current, the induced draft fan bearing temperature, the induced draft fan motor stator coil temperature, the induced draft fan motor bearing temperature, and the induced draft fan lubricating oil temperature.

[0027] Furthermore, the characteristic parameters of the induced draft fan are preprocessed, including range cleaning of the characteristic parameters of the induced draft fan, and 70% of the cleaned data are divided into a training set and 30% are divided into a validation set.

[0028] It should also be noted that all the collected characteristic data are cleaned to remove abnormal values ​​and invalid data to ensure the accuracy and completeness of the data. First, the value range of the unit load is cleaned to remove the data in the shutdown state and below 30% of the rated load of the unit to ensure that the unit is not in the shutdown state and deep adjustment operation state to avoid interference from non-steady-state conditions; secondly, the value range of all parameters is cleaned, and the upper and lower limits of the value range of each parameter are set according to the operating procedures to remove abnormal values ​​and invalid data.

[0029] The cleaned data is randomly divided into a training set (70%) and a validation set (30%) in a ratio of 7:3. This ratio balances the generalization ability of the model and the reliability of validation, avoiding the problems of overfitting (the proportion of the training set is less than 65%) or insufficient validation samples (the proportion of the training set is higher than 85%).

[0030] S2: Based on the preprocessed data, a prediction model for the characteristic parameters of the induced draft fan is established to obtain the predicted values ​​of the characteristic parameters of the induced draft fan.

[0031] It should be noted that the establishment of the induced draft fan characteristic parameter prediction model includes establishing the induced draft fan parameter dynamic warning model through the GMM model, estimating the parameters through the EM algorithm, training the induced draft fan characteristic parameter prediction model through the training set, inputting the validation set into the trained GMM model, and calculating the mean square error, Score, mean absolute error, evaluate the prediction accuracy of the GMM model. A preferred solution for establishing a prediction model for characteristic parameters of induced draft fans includes: extracting one year of historical characteristic parameter data of induced draft fans from a plant-level monitoring system (such as SIS), cleaning the data range, dividing the cleaned data into a training set (70%) and a validation set (30%) at a ratio of 7:3, establishing a dynamic warning model for induced draft fan parameters through a Gaussian mixture model, training the model through the training set, using the EM algorithm to solve the model parameters including mean, variance, and mixing weight, calculating the expected value of the probability distribution of the characteristic parameters as the prediction value, verifying the model accuracy through the validation set data, calculating the mean square error (MSE) to measure the overall prediction deviation, calculating the mean absolute error (MAE) to measure the average amount of each prediction deviation, and calculating the determination coefficient ( ) measures the model's ability to explain data fluctuations, and sets the optimal range of mean square error to , set the optimal range of mean absolute error to , set the optimal range of determination coefficient to , the model is verified by the validation set. If the mean square error, mean absolute error, and determination coefficient are within the optimal range, it means that the model accuracy is qualified and the model is available. In the predictive maintenance of the induced draft fan, the model prediction value error is required to be within 1% of the rated value, otherwise it may affect the safety of the equipment. If the MSE is too small, it may mean that the model overfits the training data, resulting in a decrease in generalization ability. If it is too low (such as lower than 0.997), it means that the model fails to fully capture the law of data changes. In the field of wind turbine intelligent monitoring, it is usually required that the prediction error is less than 1% of the measured value.

[0032] It should also be noted that based on all the feature data after cleaning, the machine learning algorithm Gaussian mixture model (GMM) is used to establish the dynamic warning model of induced draft fan parameters for parameter prediction, because it can effectively process multimodal distribution data (such as temperature changes under different loads) and adaptively fit the data distribution through the expectation maximization (EM) algorithm. After the training is completed, the split validation set part of the historical data is used to verify the model and evaluate the effect. The evaluation indicators are mean square error, The mean square error measures the overall deviation between the predicted value and the actual value. The smaller the value, the higher the model accuracy. The score reflects the model's ability to explain data variation. The closer it is to 1, the better the fitting effect. The mean absolute error (MAE) directly represents the absolute value of the prediction error, which is required to be less than 1% of the parameter rating to ensure engineering practicality. The validation set is used to calculate the MSE of each dependent variable parameter (induced draft fan motor current, induced draft fan bearing temperature, induced draft fan motor stator coil temperature, induced draft fan motor bearing temperature, induced draft fan lubricating oil temperature, etc.). with MAE.

[0033] Furthermore, the predicted values ​​of the characteristic parameters of the induced draft fan are obtained, including inputting the real-time characteristic parameters into the GMM model for operation, calculating the probability distribution of the characteristic parameters under the current working conditions, extracting the expected values ​​of the characteristic parameters as the predicted values ​​based on the probability distribution, and performing a secondary calibration on the predicted values ​​in combination with the real-time ambient temperature and the equipment aging coefficient to obtain the real-time predicted values ​​of the characteristic parameters.

[0034] It should also be noted that, under real-time working conditions, the independent variable data at the current moment is obtained and input into the trained mixed Gaussian model to calculate the probability that the dependent variable parameters under the working condition belong to different Gaussian distribution components, obtain the input data of the current working condition, extract the independent variable parameters, and calculate the Gaussian probability density value corresponding to the dependent variable parameters under the working condition. The probability density is determined by the mean and covariance matrix of each Gaussian distribution. For each Gaussian component, the posterior probability of the dependent variable parameter under the working condition is calculated. The posterior probability indicates the possibility that the data point belongs to each Gaussian distribution component. The probability is used to weight the predicted value, and the posterior probability calculated is normalized to ensure that the sum of the posterior probabilities of all components is one. Based on the calculated probability distribution of the dependent variable parameters, the predicted value is further calculated. Based on the mean and posterior probability of each Gaussian distribution component, the weighted sum is calculated, and the calculation result is used as the predicted value of the dependent variable parameter. The predicted value reflects the optimal estimated value of the dependent variable parameter under the current working conditions. The current ambient temperature is obtained and compared with the benchmark ambient temperature. The temperature correction value is calculated based on the correction factor fitted by historical data. The temperature correction value is added to the predicted value to obtain the temperature-corrected predicted value. The equipment operation time is obtained, and the equipment aging influencing factor is fitted based on historical data. The aging correction value is calculated and superimposed on the predicted value to obtain the final corrected predicted value, which is used to set the dynamic warning threshold.

[0035] S3: Establish a warning threshold value based on the predicted value of the characteristic parameter of the induced draft fan, and dynamically adjust the warning threshold value.

[0036] It should be noted that the warning threshold is dynamically adjusted according to the predicted value of the characteristic parameter of the induced draft fan, including setting the characteristic parameter deviation according to the rated value and the real-time predicted value of the characteristic parameter as follows: , , in, is the upper limit of characteristic parameter deviation, is the real-time predicted value of the characteristic parameter, is the characteristic parameter rating, where is the lower limit of characteristic parameter deviation.

[0037] It should also be noted that the present invention uses a dynamic threshold adjustment method based on historical data analysis and combines the induced draft fan characteristic parameter prediction model to calculate the upper limit value of the characteristic parameter deviation in real time. and lower limit The threshold calculation is based on the deviation ratio between the rated value of the characteristic parameter and the real-time predicted value. The statistics of historical fault cases in the past 10 years show that the parameter deviation exceeds ±1.2% in 95% of the abnormal events. Therefore, ±1% is used as the preferred threshold of our invention, which can warn of potential risks in advance and dynamically adjust the warning range, thereby reducing false alarms and missed alarms, improving the effectiveness of induced draft fan fault warnings, and providing accurate protection for the safe operation of equipment.

[0038] S4: Establish an early warning model and issue an early warning for abnormal characteristic parameters of the induced draft fan according to the early warning threshold.

[0039] It should be noted that the warning threshold is used to warn the abnormal characteristic parameters of the induced draft fan, including the construction of the warning trigger rule, which is expressed as: , in, is the real-time load of the unit, is the rated load of the unit, is the real-time current of the induced draft fan, is the rated current of the induced draft fan, is the real-time value of the characteristic parameter, It is the duration of feature parameter exceeding limit.

[0040] When the early warning trigger rules are met, the system automatically triggers an alarm signal and records abnormal data in real time. If the induced draft fan current exceeds the limit, check the motor load fluctuation, check the inverter working status, and adjust the moving blade opening. If the bearing temperature exceeds the upper limit, check the lubricating oil system, clean the bearing heat dissipation channel, monitor the vibration data, and check the stability of the mechanical structure. If the stator coil temperature exceeds the upper limit, reduce the unit load, check the cooling fan operating status, check the motor insulation aging and poor contact. If the lubricating oil temperature exceeds the upper limit, switch to the standby lubricating oil pump, check the heat dissipation efficiency of the heat exchanger, clean the dirt, replace the cooling medium, and replace the lubricating oil filter.

[0041] It should also be noted that when the unit is in a stable operating state and the induced draft fan is in a state, it is determined whether the characteristic parameters of the induced draft fan exceed the threshold value. If the characteristic parameters of the induced draft fan are greater than the upper limit value, or the characteristic parameters of the induced draft fan are less than the lower limit value, and the overlimit duration lasts for 60 seconds, an early warning is triggered, and abnormal data is recorded to achieve real-time monitoring and intelligent alarm. By correlating and analyzing key parameters such as induced draft fan current overlimit, bearing temperature overlimit, stator coil temperature overlimit and lubricating oil temperature abnormality, the system can intelligently identify the fault type and provide corresponding diagnosis and processing strategies, such as adjusting the fan operating conditions, optimizing the cooling system, switching the lubricating oil pump and cleaning the filter. Compared with traditional early warning methods, the present invention can accurately locate the source of the fault, reduce false alarms and missed alarms, improve the accuracy and response speed of the early warning, effectively improve the operating safety and maintenance efficiency of the induced draft fan system, and ensure the stable operation of the unit under different operating conditions.

[0042] Example 2, reference Figure 2-Figure 3 , which is an embodiment of the present invention, provides a dynamic early warning method for induced draft fan parameters based on historical data mining. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0043] A certain plant will obtain the historical data and real-time data of the characteristic parameters of the induced draft fan from the plant-level monitoring system in the three zones by matching the measurement point information from the real-time database, so as to facilitate the analysis and mining of the historical data of the characteristic parameters of the induced draft fan, train the model, and form the parameters of the induced draft fan. A certain plant obtains the historical data of the characteristic parameters of the induced draft fan in the past year through the SIS time series database, cleans all the collected characteristic data, removes abnormal values ​​and invalid data, and ensures the accuracy and completeness of the data. First, the value range of the unit load is cleaned to remove the data in the shutdown state and the data below 30% of the rated load of the unit to ensure that the unit is not in the shutdown state or the deep adjustment operation state; secondly, the value range of all parameters is cleaned, and the upper and lower limits of the value range of each parameter are set according to the operating procedures to remove abnormal values ​​and invalid data. The pre-processed data is as follows: Figure 2 shown.

[0044] All the feature data after cleaning are randomly split into 70% and 30% parts. 70% of the sample data is used as model training data, and 30% of the sample data is used as the data for model verification after model training. The machine learning algorithm Gaussian Mixture Model is used to establish the dynamic early warning model of induced draft fan parameters, and 70% of the sample data is input for model training.

[0045] After the model training is completed, 30% of the data is input into the trained model to predict the induced draft fan motor current, induced draft fan front bearing temperature, induced draft fan middle bearing temperature, induced draft fan rear bearing temperature, induced draft fan motor stator coil temperature, induced draft fan motor bearing temperature, induced draft fan lubricating oil temperature, and then the mean squared error (MSE), R² score (R²) and mean absolute error (MAE) are used to calculate the actual values ​​and predicted values ​​of the above parameters. The calculation results are shown in Table 1.

[0046] Table 1 Evaluation data of deviation between actual and predicted values ​​of characteristic parameters of induced draft fan Parameter name MSE <![CDATA[R 2 ]]> MAE Induced draft fan motor current 2.03001 0.9994 0.3593 ID fan front bearing temperature 0.00422 0.9988 0.0308 Bearing temperature of induced draft fan 0.00424 0.9986 0.0323 ID fan rear bearing temperature 0.00446 0.9986 0.0309 Induced draft fan motor A phase stator coil temperature 0.03289 0.9994 0.0686 Induced draft fan motor B phase stator coil temperature 0.03434 0.9994 0.0689 Induced draft fan motor C phase stator coil temperature 0.0345 0.9994 0.0695 Driving end bearing temperature of induced draft fan motor 0.00403 0.9979 0.0343 ID fan motor non-drive end bearing temperature 0.00323 0.9986 0.0286 ID fan lubricating oil temperature 0.01136 0.9983 0.0334

[0047] According to the operating procedures, equipment parameters and field operating experience, the dynamic upper and lower limits of the induced draft fan motor current, induced draft fan front bearing temperature, induced draft fan middle bearing temperature, induced draft fan rear bearing temperature, induced draft fan motor stator coil temperature, induced draft fan motor bearing temperature, and induced draft fan lubricating oil temperature are obtained, as shown in Table 2.

[0048] Table 2 Data table of upper and lower limits of characteristic parameter deviation of induced draft fan Parameter name Upper limit of deviation Deviation lower limit Induced draft fan motor current 8 -8 Induced draft fan front bearing temperature 5 -5 Bearing temperature of induced draft fan 5 -5 ID fan rear bearing temperature 5 -5 Induced draft fan motor A phase stator coil temperature 6 -6 Induced draft fan motor B phase stator coil temperature 6 -6 Induced draft fan motor C phase stator coil temperature 6 -6 Driving end bearing temperature of induced draft fan motor 6 -6 ID fan motor non-drive end bearing temperature 5 -5 ID fan lubricating oil temperature 5 -5

[0049] According to Table 1, the model prediction accuracy is excellent and the model is available. According to Table 2, the upper and lower limits of the deviation of the characteristic parameters of the induced draft fan can be obtained. Therefore, a real-time prediction model of the induced draft fan and a real-time early warning model of the characteristic parameters of the induced draft fan are established in the power plant monitoring system.

[0050] After the early warning model is built in the real-time monitoring system of the power plant, the operation status of the induced draft fan is monitored in real time. When the key parameters of the induced draft fan exceed the set dynamic upper and lower warning thresholds, the system automatically triggers the early warning signal and generates an alarm record. The operation and maintenance personnel take corresponding measures in time according to the early warning information to ensure the safe operation of the induced draft fan. Figure 3 shown.

[0051] Embodiment 3 is an embodiment of the present invention, which provides a dynamic warning system for induced draft fan parameters based on historical data mining, including a data acquisition and preprocessing module, a characteristic parameter prediction module, a dynamic threshold adjustment module, and a warning model construction module.

[0052] Among them, S5: data acquisition and preprocessing module includes a parameter acquisition unit and a data preprocessing unit. The parameter acquisition unit is used to collect the induced draft fan current, bearing temperature, stator coil temperature, and lubricating oil temperature. The data preprocessing unit is used to remove abnormal values ​​and normalize the collected data, and divide it into a training set and a validation set.

[0053] It should also be noted that the data acquisition and preprocessing module is used to obtain the key parameters of the induced draft fan operation and preprocess the data. The processed data is divided into a training set and a validation set, and input into the characteristic parameter prediction module for model training and real-time prediction.

[0054] S6: The characteristic parameter prediction module includes a model training unit and a real-time prediction unit. The model training unit is used to train the Gaussian mixture model GMM based on the preprocessed data and use the expectation maximization (EM) algorithm for parameter optimization. The real-time prediction unit is used to input the real-time operation data into the trained model to calculate the predicted value of the characteristic parameter of the induced draft fan.

[0055] It should also be noted that the prediction model optimized by the GMM model and the EM algorithm generates the predicted values ​​of the characteristic parameters of the induced draft fan in the real-time prediction unit, and transmits the predicted values ​​to the dynamic threshold adjustment module for calculating the dynamic upper and lower limits of the characteristic parameters.

[0056] S7: The dynamic threshold adjustment module is used to calculate the dynamic upper and lower limits of the parameters based on the predicted values ​​and the equipment rated values.

[0057] It should also be noted that the dynamic threshold adjustment module calculates the dynamic upper limit and dynamic lower limit of the parameter based on the predicted parameter value and the equipment rating, and uses the upper limit and lower limit as input conditions of the early warning model.

[0058] S8: The early warning model construction module includes an anomaly detection unit and an alarm triggering unit. The anomaly detection unit is used to construct an alarm to monitor characteristic parameters in real time and determine whether the characteristic parameters are out of limit. The alarm triggering unit is used to trigger an early warning signal when the abnormal state continues for more than a set time, send an alarm message to the system, and take maintenance measures according to the alarm message.

[0059] It should also be noted that the anomaly detection unit constructs early warning trigger rules through dynamic thresholds, compares and analyzes the real-time collected induced draft fan operating parameters, and determines whether the characteristic parameters exceed the dynamic threshold range. If the characteristic parameters trigger an alarm, the alarm trigger unit generates an early warning signal and sends an alarm message to the system, prompting maintenance personnel to take corresponding repair measures.

[0060] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0061] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0062] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk case (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0063] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logical function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A dynamic early warning method for induced draft fan parameters based on historical data mining, characterized by: include: Collect characteristic parameters of the induced draft fan and pre-process the characteristic parameters of the induced draft fan; According to the preprocessed data, a prediction model for the characteristic parameters of the induced draft fan is established to obtain the predicted values ​​of the characteristic parameters of the induced draft fan; Establish warning thresholds based on predicted values ​​of characteristic parameters of induced draft fans and dynamically adjust the warning thresholds; Establish an early warning model to issue early warnings for abnormal characteristic parameters of induced draft fans based on early warning thresholds; The characteristic parameter prediction model of the induced draft fan includes constructing a data set based on historical data and preprocessing it, using the GMM model to establish a prediction model, using the EM algorithm to estimate and train parameters, calculating the probability distribution of characteristic parameters, extracting expected values ​​as prediction results, combining environmental variables with equipment aging factors for calibration, and dynamically predicting and outputting parameter prediction values; Dynamic adjustment of the warning threshold includes calculating the real-time deviation range based on the predicted value of the characteristic parameters of the induced draft fan, combining the rated parameters of the equipment with the operating conditions, and dynamically adjusting the warning threshold; The early warning model includes that when the unit is in a stable operating state and the induced draft fan is in operation, if the characteristic parameters exceed the early warning threshold, the system automatically triggers the early warning signal and takes maintenance measures based on the early warning signal.

2. The method for dynamic early warning of induced draft fan parameters based on historical data mining according to claim 1, characterized in that: The collecting of characteristic parameters of the induced draft fan includes collecting the induced draft fan current, the induced draft fan bearing temperature, the induced draft fan motor stator coil temperature, the induced draft fan motor bearing temperature, and the induced draft fan lubricating oil temperature.

3. The dynamic early warning method for induced draft fan parameters based on historical data mining according to claim 1 or 2, characterized in that: The preprocessing of the characteristic parameters of the induced draft fan includes cleaning the value range of the characteristic parameters of the induced draft fan, dividing 70% of the cleaned data into a training set and dividing 30% into a verification set.

4. The method for dynamic early warning of induced draft fan parameters based on historical data mining according to claim 3, characterized in that: The establishment of the induced draft fan characteristic parameter prediction model includes establishing an induced draft fan parameter dynamic warning model through a GMM model, performing parameter estimation through an EM algorithm, training the induced draft fan characteristic parameter prediction model through a training set, inputting a validation set into the trained GMM model, calculating a mean square error, Score, mean absolute error, evaluates the prediction accuracy of the GMM model.

5. The method for dynamic early warning of induced draft fan parameters based on historical data mining according to any one of claims 1, 2 or 4, characterized in that: The method of obtaining the predicted value of the characteristic parameter of the induced draft fan includes inputting the real-time characteristic parameter into the GMM model for operation, calculating the probability distribution of the characteristic parameter under the current working condition, extracting the expected value of the characteristic parameter as the predicted value based on the probability distribution, and performing a secondary calibration on the predicted value in combination with the real-time ambient temperature and the equipment aging coefficient to obtain the real-time predicted value of the characteristic parameter.

6. The method for dynamic early warning of induced draft fan parameters based on historical data mining according to claim 5, characterized in that: The method of dynamically adjusting the warning threshold according to the predicted value of the characteristic parameter of the induced draft fan includes setting the characteristic parameter deviation according to the rated value and the real-time predicted value of the characteristic parameter as follows: ， ， in, is the upper limit of characteristic parameter deviation, is the real-time predicted value of the characteristic parameter, is the characteristic parameter rating, where is the lower limit of characteristic parameter deviation.

7. The method for dynamic early warning of induced draft fan parameters based on historical data mining according to any one of claims 1, 2, 4 or 6, characterized in that: The method of giving an early warning to abnormal characteristic parameters of the induced draft fan according to the early warning threshold value includes constructing an early warning trigger rule expressed as: ， in, is the real-time load of the unit, is the rated load of the unit, is the real-time current of the induced draft fan, is the rated current of the induced draft fan, is the real-time value of the characteristic parameter, The duration of feature parameter exceeding the limit; When the early warning trigger rules are met, the system automatically triggers an alarm signal and records abnormal data in real time. If the induced draft fan current exceeds the limit, check the motor load fluctuation, check the inverter working status, and adjust the moving blade opening. If the bearing temperature exceeds the upper limit, check the lubricating oil system, clean the bearing heat dissipation channel, monitor the vibration data, and check the stability of the mechanical structure. If the stator coil temperature exceeds the upper limit, reduce the unit load, check the cooling fan operating status, check the motor insulation aging and poor contact. If the lubricating oil temperature exceeds the upper limit, switch to the standby lubricating oil pump, check the heat dissipation efficiency of the heat exchanger, clean the dirt, replace the cooling medium, and replace the lubricating oil filter.

8. The dynamic early warning system of induced draft fan parameters based on historical data mining is characterized by: It includes data collection and preprocessing module, characteristic parameter prediction module, dynamic threshold adjustment module and early warning model construction module; The data acquisition and preprocessing module includes a parameter acquisition unit and a data preprocessing unit. The parameter acquisition unit is used to collect the induced draft fan current, bearing temperature, stator coil temperature, and lubricating oil temperature. The data preprocessing unit is used to remove abnormal values ​​and normalize the collected data, and divide it into a training set and a verification set. The characteristic parameter prediction module includes a model training unit and a real-time prediction unit. The model training unit is used to train a Gaussian mixture model (GMM) based on preprocessed data and use an expectation maximization (EM) algorithm to perform parameter optimization. The real-time prediction unit is used to input real-time operation data into the trained model to calculate the predicted value of the characteristic parameter of the induced draft fan. The dynamic threshold adjustment module is used to calculate the dynamic upper and lower limits of the parameter based on the predicted value and the device rated value; The early warning model building module includes an abnormality detection unit and an alarm triggering unit. The abnormality detection unit is used to monitor characteristic parameters in real time and determine whether the characteristic parameters are out of limit. The alarm triggering unit is used to trigger an early warning signal after the abnormal state continues for more than a set time, send an alarm information to the system, and take maintenance measures according to the alarm information.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images