Abnormal temperature detection and early warning method for bearing of induced draft fan

Abstract

The invention belongs to the technical field of equipment state monitoring, and particularly relates to an induced draft fan bearing temperature anomaly detection and early warning method, which comprises the following steps: dividing an induced draft fan bearing temperature time sequence into continuously overlapped sliding windows and carrying out data preprocessing; adopting an EGARCH model to fit fluctuation characteristics in each sliding window, and extracting three core parameters representing model parameter sensitivity; the three core parameters comprise a symmetric impact coefficient alpha, an asymmetric lever coefficient gamma and a durability coefficient beta; forming a three-dimensional time sequence matrix formed by n groups of parameters based on the three core parameters; and performing anomaly detection on the coefficient matrix by using an anomaly detection algorithm based on a Mahalanobis distance to realize temperature anomaly early warning. Through real-time monitoring, data analysis and early warning mechanisms, potential faults of the induced draft fan bearing are found in advance, and real-time diagnosis and fault early warning of the equipment health state are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment status monitoring, and in particular relates to a method for detecting and warning abnormal temperature of an induced draft fan bearing. Background Art

[0002] The induced draft fan (IDF) is a critical piece of equipment in large coal-fired power plants. It extracts flue gas from the furnace, maintains negative pressure, prevents flue gas leakage, and ensures a safe working environment. It also ensures complete fuel combustion, improves combustion efficiency, and adjusts air volume to accommodate fluctuating boiler loads. It also facilitates environmental protection, reduces pollutant emissions, and improves thermal efficiency through heat recovery. The stable operation of the IDF is fundamental to the stable energy supply of the power plant; any failure could impact the safety and efficiency of the boiler and the entire power plant. Therefore, the IDF plays a vital role in coal-fired power plants.

[0003] The bearings of induced draft fans (IDFs) in large coal-fired power plants are core components, responsible for supporting the high-speed rotation of the rotor. The operating condition of the bearings directly impacts the performance and reliability of the fans. Good bearings reduce friction, lower energy consumption, and ensure long-term, stable operation of the fans. At the same time, the health of the bearings also affects the maintenance costs and repair intervals. Bearing damage can lead to increased fan vibration and even serious equipment failure, disrupting normal power plant operations. Therefore, regular inspection, lubrication, and maintenance of IDF bearings are key measures to ensure efficient and safe operation of coal-fired power plants.

[0004] Detecting abnormal induced draft fan (IDF) bearing temperatures is crucial in large coal-fired power plants. Elevated bearing temperatures are often a precursor to failure, and timely detection can prevent bearing damage and avoid unplanned downtime. Abnormal temperature monitoring helps identify insufficient lubrication, excessive loads, or assembly issues, enabling timely maintenance or adjustments. This not only extends bearing life and reduces maintenance costs, but also ensures continuous and stable power supply to the power plant and reduces the risk of safety incidents. Therefore, a bearing temperature monitoring system is crucial for the healthy operation of IDF fans and significantly contributes to improving the overall efficiency and safety of power plants.

[0005] While temperature warnings for induced draft fan bearings in large coal-fired power plants are crucial for preventing failures, they have certain shortcomings. The warning logic can be overly simplistic. This simple logic typically determines bearing anomalies based on fixed temperature thresholds, ignoring the complexity and dynamic nature of bearing temperature fluctuations. This logic fails to intelligently analyze the actual operating status and temperature trends of the bearings, leading to false alarms and missed warnings.

[0006] Therefore, there is an urgent need for a method for detecting and warning abnormal temperature of induced draft fan bearings. By combining equipment sensing, thermodynamic analysis and big data technology, through real-time monitoring, data analysis and early warning mechanism, potential faults of induced draft fan bearings can be discovered in advance, and real-time diagnosis of equipment health status and fault warning can be achieved. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for detecting and warning abnormal temperature of an induced draft fan bearing, comprising the following steps:

[0008] The induced draft fan bearing temperature time series is divided into continuous overlapping sliding windows and data preprocessing is performed;

[0009] The EGARCH model is used to fit the volatility characteristics within each sliding window, and three core parameters that characterize the sensitivity of the model parameters are extracted; the three core parameters include: symmetric impact coefficient α, asymmetric leverage coefficient γ, and persistence coefficient β;

[0010] Based on the three core parameters, a three-dimensional time series matrix consisting of n groups of parameters is formed;

[0011] The anomaly detection algorithm based on Mahalanobis distance is used to detect anomalies in the coefficient matrix to achieve temperature anomaly warning.

[0012] The method of dividing the induced draft fan bearing temperature time series into continuous overlapping sliding windows and performing data preprocessing includes:

[0013] Set the sampling period Δt of the induced draft fan bearing temperature data; set the parameters of the dynamic sliding window: the number of sliding windows is n; set m, the sliding window length is m+1, that is, the sliding window contains m+1 induced draft fan bearing temperature time series data; the sliding step of the sliding window is Δt×s (s≥1); the initial time of collecting induced draft fan bearing temperature data is t0, and the start and end time periods of the i-th sliding window are: [t0+(i-1)×Δt×s, t0+(i-1)×Δt×s+m×Δt], where i=1,2,…,n;

[0014] Use dynamic sliding window to observe the time series data of induced draft fan bearing temperature;

[0015] The first-order difference processing is performed on the m+1 induced draft fan bearing temperature time series data in the sliding window to obtain the induced draft fan bearing temperature difference time series, and the length of the temperature difference time series is m.

[0016] The EGARCH model is used to fit the fluctuation characteristics, and the three core parameters that characterize the sensitivity of the model parameters are extracted, including:

[0017] In each sliding window, the generalized autoregressive conditional heteroskedasticity EGARCH (1,1) model is used to fit the differential time series of the induced draft fan bearing temperature. The three core parameters of the EGARCH (1,1) model variance equation are extracted, including: the symmetric impact coefficient α, which measures the impact of historical fluctuations on current fluctuations and reflects the fluctuation clustering effect; the asymmetric leverage coefficient γ, which captures the leverage effect; and the persistence coefficient β, which reflects the fluctuation memory. The closer β is to 1, the stronger the persistence of the impact of historical fluctuations on the current variance.

[0018] The forming of a three-dimensional time series matrix consisting of n groups of parameters based on the three core parameters includes:

[0019] The three core parameters of the variance equation of the EGARCH(1,1) model for the temperature difference time series within each sliding window and the three core parameters of the variance equation of the EGARCH(1,1) model for the difference time series within n sliding windows constitute the n×3 parameter matrix Θ;

[0020]

[0021] Where: the i-th row corresponds to the core coefficient of the i-th EGARCH(1,1) model; the first column: α i is the symmetric impact coefficient; the second column: γ i is the asymmetric leverage coefficient; the third column: β i is the persistence coefficient; i=1,2,…,n.

[0022] The method of using the Mahalanobis distance-based anomaly detection algorithm to perform anomaly detection on the coefficient matrix to implement temperature anomaly warning includes:

[0023] In three-dimensional space, the three core parameters of the variance equation of the EGARCH (1,1) model constitute a point; the n×3 parameter matrix Θ represents n points;

[0024] The anomaly detection algorithm based on Mahalanobis distance is used to detect anomalies of n points; if a point (α k , γ k , β k ) to the mean vector is greater than the threshold Threshold, the point is determined to be an abnormal point, k∈[1,2,…,n]; the warning is issued for abnormal temperature fluctuations within the start and end time periods of the sliding window associated with the point [t0+(k-1)×Δt×s, t0+(k-1)×Δt×s+m×Δt], k∈[1,2,…,n]; set Threshold=σ×3, σ is the value of all points (α i , γ i , β i) to the mean vector, i = 1, 2, …, n; k ∈ [1, 2, …, n].

[0025] Another object of the present invention is to disclose an induced draft fan bearing temperature anomaly detection and early warning system using the induced draft fan bearing temperature anomaly detection and early warning method according to the present invention, comprising: a bearing temperature acquisition device, a power plant production equipment SIS database, and an induced draft fan bearing temperature anomaly real-time monitoring and early warning server;

[0026] The bearing temperature collection device is used to collect the bearing temperature of the induced draft fan;

[0027] The power plant production equipment SIS database is used to obtain monitoring data;

[0028] The induced draft fan bearing temperature abnormality real-time monitoring and early warning server obtains monitoring data, inputs the monitoring data into an implementation monitoring program, and implements the induced draft fan bearing temperature abnormality detection and early warning method; the monitoring data includes: bearing temperature and other power plant equipment real-time operating data.

[0029] Another object of the present invention is to disclose a computer device comprising a memory and a processor, wherein the memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the method for detecting and warning abnormal temperature of an induced draft fan bearing according to the present invention.

[0030] Another object of the present invention is to disclose a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the induced draft fan bearing temperature abnormality detection and early warning method according to the present invention.

[0031] The beneficial effects of the present invention are:

[0032] The EGARCH time series model effectively captures the volatility characteristics of temperature fluctuations, particularly their clustering and asymmetry. By modeling the logarithm of the volatility index, the EGARCH model addresses the leverage effect in temperature data, specifically the asymmetric impact of the direction of temperature change on volatility. The EGARCH(1,1) model is concise and has few parameters, making it easy to estimate and interpret, making it suitable for volatility analysis of temperature data.

[0033] The Mahalanobis distance anomaly detection algorithm is an effective statistical method that is particularly suitable for multidimensional data. It considers the correlation between data and is used to identify outliers or outliers in the data set.

[0034] This paper combines the EGARCH model with the Mahalanobis distance anomaly detection algorithm to detect abnormal temperature fluctuations in induced draft fan bearings. This synergy significantly improves detection sensitivity, reduces false alarm rates, and provides efficient early warning before equipment failure, ensuring the safe and stable operation of thermal power production systems.

[0035] By combining equipment sensing, thermodynamic analysis and big data technology, through real-time monitoring, data analysis and early warning mechanisms, potential failures of induced draft fan bearings can be discovered in advance, achieving real-time diagnosis of equipment health status and fault warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A schematic flow chart of a method for detecting and warning abnormal temperature of an induced draft fan bearing according to the present invention;

[0037] Figure 2 Schematic diagram of the structure of an abnormal temperature monitoring and early warning system for an induced draft fan bearing according to an embodiment of the present invention;

[0038] Figure 3 This is a schematic diagram of time series data of the induced draft fan bearing temperature according to an embodiment of the present invention;

[0039] Figure 4 Schematic diagram of first-order difference of induced draft fan bearing temperature time series data according to an embodiment of the present invention;

[0040] Figure 5 Schematic diagram of the core parameter points of the EGARCH (1,1) model according to an embodiment of the present invention;

[0041] Figure 6 A schematic diagram of Mahalanobis distance and threshold distribution according to an embodiment of the present invention;

[0042] Figure 7 Schematic diagram of the core parameter outliers of the EGARCH (1,1) model according to an embodiment of the present invention;

[0043] Among them, 100-induced draft fan; 200-bearing temperature acquisition device; 300-power plant production equipment SIS database; 400-induced draft fan bearing temperature abnormality real-time monitoring and early warning server. DETAILED DESCRIPTION

[0044] The present invention provides a method for detecting and warning abnormal temperature of an induced draft fan bearing, which is further described in detail below with reference to the accompanying drawings.

[0045] like Figure 1 The embodiment of the present invention shown discloses a method for detecting and warning abnormal temperature of an induced draft fan bearing, comprising the following steps:

[0046] The induced draft fan bearing temperature time series is divided into continuous overlapping sliding windows and data preprocessing is performed;

[0047] The EGARCH model is used to fit the volatility characteristics within each sliding window, and three core parameters that characterize the sensitivity of the model parameters are extracted; the three core parameters include: symmetric impact coefficient α, asymmetric leverage coefficient γ, and persistence coefficient β;

[0048] Based on the three core parameters, a three-dimensional time series matrix consisting of n groups of parameters is formed;

[0049] The anomaly detection algorithm based on Mahalanobis distance is used to detect anomalies in the coefficient matrix to achieve temperature anomaly warning.

[0050] In this embodiment, the EGARCH model is combined with an anomaly detection algorithm based on the Mahalanobis distance to achieve dynamic monitoring and early warning of the operating status of the induced draft fan bearings in power plants. First, the induced draft fan bearing temperature time series is divided into continuous overlapping sliding windows. Within each window, the EGARCH model is used to fit the fluctuation characteristics. Three core parameters, α (symmetric impact coefficient), γ (asymmetric leverage coefficient), and β (persistence coefficient), which represent the sensitivity of the model parameters, are extracted to form a three-dimensional time series matrix composed of n groups of parameters. Anomaly detection and early warning are performed on this coefficient matrix using an anomaly detection algorithm based on the Mahalanobis distance. This method captures the clustering, asymmetry, and leverage effect of bearing temperature fluctuations through time-varying parameters, and uses the anomaly detection algorithm to achieve early fault warning, effectively addressing the limitations of traditional static threshold methods in nonlinear time-varying scenarios.

[0051] The following is a detailed explanation of each step:

[0052] Step 1: Sliding window division and data preprocessing.

[0053] Set the sampling period Δt for the induced draft fan bearing temperature data. Set the parameters for the dynamic sliding window: the number of sliding windows is n; set m, and the sliding window length is m+1, meaning that the sliding window contains m+1 induced draft fan bearing temperature time series data; the sliding window step size is Δt×s (s≥1). The initial time for collecting induced draft fan bearing temperature data is t0, and the start and end time periods of the i-th sliding window are [t0+(i-1)×Δt×s, t0+(i-1)×Δt×s+m×Δt], where i = 1, 2, …, n.

[0054] The dynamic sliding window is used to observe the time series data of the induced draft fan bearing temperature.

[0055] The first-order difference processing of the m+1 induced draft fan bearing temperature time series data in the sliding window is performed to obtain the induced draft fan bearing temperature difference time series, and the length of the temperature difference time series is m.

[0056] Step 2: EGARCH(1,1) model parameter estimation;

[0057] In each sliding window, the generalized autoregressive conditional heteroskedasticity EGARCH (1,1) model is used to fit the differential time series of the induced draft fan bearing temperature, and the three core parameters of the EGARCH (1,1) model variance equation are extracted: α (symmetric impact coefficient): measures the impact of historical fluctuations on current fluctuations and reflects the fluctuation clustering effect; γ (asymmetric leverage coefficient): captures the leverage effect; β (persistence coefficient): reflects the fluctuation memory. The closer β is to 1, the stronger the persistence of the impact of historical fluctuations on the current variance.

[0058] The EGARCH(1,1) model, or Exponential Generalized Autoregressive Conditional Heteroskedasticity model, is a variant of the GARCH model that is particularly suitable for analyzing time series data with asymmetric effects. Asymmetric effects refer to the different impacts of positive and negative shocks on volatility. The variance equation of the EGARCH(1,1) model is in logarithmic form and is as follows:

[0059] Variance equation: in is the conditional variance at time t, ω is a constant term, α measures the influence of past error terms on the current variance, γ is the asymmetric term coefficient, which is used to characterize the leverage effect, and β reflects the influence of past conditional variance on the current variance.

[0060] Step 3: Construct the parameter matrix.

[0061] The three core parameters of the variance equation of the EGARCH(1,1) model for the temperature difference time series within each sliding window and the three core parameters of the variance equation of the EGARCH(1,1) model for the difference time series within n sliding windows constitute the n×3 parameter matrix Θ.

[0062]

[0063] Note: The i-th row corresponds to the core coefficient of the i-th EGARCH(1,1) model; the first column: α i (Symmetrical impact coefficient); second column: γ i (asymmetric leverage coefficient); third column: β i (Persistence coefficient). Where i = 1, 2, …, n.

[0064] Step 4: Implementation of anomaly detection and early warning based on Mahalanobis distance

[0065] In three-dimensional space, the three core parameters of the variance equation of the EGARCH (1,1) model constitute a point; the n×3 parameter matrix Θ represents n points.

[0066] Use the anomaly detection algorithm based on Mahalanobis distance to detect anomalies of n points; set Threshold, if a point (α k , γ k , β k ) to the mean vector is greater than the threshold Threshold, the point is determined to be an abnormal point, k∈[1,2,…,n]; and an alert is issued for abnormal temperature fluctuations within the start and end time periods of the sliding window associated with the point [t0+(k-1)×Δt×s,t0+(k-1)×Δt×s+m×Δt], k∈[1,2,…,n].

[0067] The Mahalanobis distance calculation method is as follows.

[0068] 1. Covariance matrix estimation: Calculate the covariance matrix of the joint distribution of three-dimensional parameters.

[0069] 2. Distance calculation: Calculate the Mahalanobis distance from each data point to the mean vector using the formula: D 2 =(x-μ) T Σ -1 (x-μ) where x is a data point, μ is the mean vector, and Σ is the covariance matrix.

[0070] Abnormality judgment: Set a threshold (such as 3 times the standard deviation), and judge points whose distance exceeds the threshold as abnormal.

[0071] In this embodiment, by capturing abnormal changes in EGARCH model parameters of the induced draft fan bearing temperature difference series, various abnormal phenomena and faults of the bearing related to the abnormal changes in the induced draft fan bearing temperature are detected.

[0072] In this example, by identifying abnormal changes in the EGARCH model parameters of the differential temperature series of the induced draft fan bearing temperature, we can capture the time-varying characteristics and memory effects of temperature fluctuations, effectively distinguishing short-term anomalies caused by parameter mutations from long-term trend anomalies. The EGARCH time series model effectively captures the volatility characteristics of temperature changes, particularly the clustering and asymmetry of fluctuations. The EGARCH model effectively captures the asymmetric influence of the direction of temperature change on volatility, the so-called "leverage effect."

[0073] Compared with traditional technical methods, this method effectively and significantly improves the efficiency of detecting abnormal operating conditions of induced draft fan bearings, reduces the false alarm rate of bearing warnings, and provides quantitative decision support for preventive maintenance of thermal power production equipment.

[0074] In order to verify the effectiveness of the method for detecting and warning abnormal temperature of induced draft fan bearings disclosed in the present invention, a verification experiment was conducted. The specific verification process is as follows:

[0075] Step 1: Sliding window division and data preprocessing

[0076] Set the sampling period for the induced draft fan bearing temperature data to Δt = 60 seconds. Set the parameters for the dynamic sliding window: the number of sliding windows is n, set to n = 1000; m is set to 300, and the sliding window length is m + 1, meaning that the sliding window contains m + 1 induced draft fan bearing temperature time series data; the sliding window step is Δt × s (s ≥ 1), set to s = 10. The initial time for collecting induced draft fan bearing temperature data is t0, and the start and end time periods of the i-th sliding window are [t0 + (i - 1) × Δt × s, t0 + (i - 1) × Δt × s + m × Δt].

[0077] The dynamic sliding window is used to observe the time series data of the induced draft fan bearing temperature.

[0078] The first-order difference processing is performed on the m+1 induced draft fan bearing temperature time series data in the sliding window to obtain the induced draft fan bearing temperature difference time series. The length of the temperature difference time series is m, where m=300.

[0079] Step 2: EGARCH(1,1) model parameter estimation

[0080] In each sliding window, the generalized autoregressive conditional heteroskedasticity EGARCH (1,1) model is used to fit the differential time series of the induced draft fan bearing temperature, and the three core parameters of the EGARCH (1,1) model variance equation are extracted: α (symmetric impact coefficient): measures the impact of historical fluctuations on current fluctuations and reflects the fluctuation clustering effect; γ (asymmetric leverage coefficient): captures the leverage effect; β (persistence coefficient): reflects the memory of fluctuations. The closer β is to 1, the stronger the persistence of the impact of historical fluctuations on the current situation.

[0081] Step 3: Construct parameter matrix

[0082] The three core parameters of the variance equation of the EGARCH(1,1) model for the temperature difference time series within each sliding window and the three core parameters of the variance equation of the EGARCH(1,1) model for the difference time series within n sliding windows constitute the n×3 parameter matrix Θ.

[0083]

[0084] Note: The i-th row corresponds to the core coefficient of the i-th EGARCH(1,1) model; the first column: α i (Symmetrical impact coefficient); second column: γ i (asymmetric leverage coefficient); third column: β i (Persistence coefficient).

[0085] Where i = 1, 2,…, n.

[0086] Step 4: Implementation of anomaly detection and early warning based on Mahalanobis distance

[0087] In three-dimensional space, the three core parameters of the variance equation of the EGARCH (1,1) model constitute a point; the n×3 parameter matrix Θ represents n points.

[0088] The anomaly detection algorithm based on Mahalanobis distance is used to detect anomalies of n points; if a point (α k , γ k , β k ) to the mean vector is greater than the threshold Threshold, the point is determined to be an abnormal point, k∈[1,2,…,n]; an alert is issued for abnormal temperature fluctuations within the start and end time periods of the sliding window associated with the point [t0+(k-1)×Δt×s, t0+(k-1)×Δt×s+m×Δt], k∈[1,2,…,n]. Set Threshold=σ×3, σ is the value of all points (α i , γ i , β i ) to the mean vector, i = 1, 2, ..., n. Where k∈[1, 2, ..., n].

[0089] The Mahalanobis distance calculation method is as follows:

[0090] 1. Covariance matrix estimation: Calculate the covariance matrix of the joint distribution of three-dimensional parameters.

[0091] 2. Distance calculation: Calculate the Mahalanobis distance from each data point to the mean vector using the formula: D 2 =(x-μ) T Σ -1 (x-μ) where x is a data point, μ is the mean vector, and Σ is the covariance matrix.

[0092] Abnormality judgment: Set a threshold (such as 3 times the standard deviation), and judge points whose distance exceeds the threshold as abnormal.

[0093] The verification experiment results show that the application of the method for detecting and warning abnormal temperature of induced draft fan bearings disclosed in the present invention, combined with equipment sensing, thermodynamic analysis and big data technology, can detect potential faults of induced draft fan bearings in advance through real-time monitoring, data analysis and early warning mechanism, and realize real-time diagnosis of equipment health status and fault warning.

[0094] In existing technology, EGARCH (Exponential GARCH) is an important extension of the generalized autoregressive conditional heteroskedasticity (GARCH) model and is an extension of the GARCH model. Its main feature is that it allows the conditional variance to respond differently to positive and negative shocks, which is called the leverage effect. This differs from the traditional GARCH model, which uses square terms and treats positive and negative shocks symmetrically. Its core innovation lies in its exponential representation of the dynamic changes in the conditional variance, which can capture the "leverage effect" commonly seen in time series data. This is because negative shocks to time series data have a greater impact on the volatility of the time series than positive shocks.

[0095] In current engineering applications, when the EGARCH model is used for analysis, it is only used to analyze the leverage effect within a specific time period. The present invention is based on a continuous sliding window, using the three core parameters of the EGARCH model of different sliding windows to form a three-dimensional space. On this basis, the changes in the three core parameters of the EGARCH model are used to compare and analyze the leverage effect of fluctuations in different time windows. Further combined with an anomaly detection algorithm based on the Mahalanobis distance, anomaly detection of induced draft fan bearing temperatures is achieved. Because the method of the present invention combines the EGARCH model with an anomaly detection algorithm based on the Mahalanobis distance, it has better capabilities in distinguishing, comparing, analyzing and identifying the leverage effect of boiler wall temperature fluctuations within different time windows.

[0096] In a specific embodiment, the sampling period of the induced draft fan bearing temperature data is Δt=60 seconds. The parameters of the dynamic sliding window are set as follows: the number of sliding windows is n, set n=1000; set m=300, and the sliding window length is m+1, that is, the sliding window contains m+1 induced draft fan bearing temperature time series data; the sliding step of the sliding window is Δt×s (s≥1), set s=60. The induced draft fan bearing temperature time series data is as follows: Figure 3 The first-order difference of the induced draft fan bearing temperature time series data is as follows: Figure 4 shown.

[0097] In each sliding window, the generalized autoregressive conditional heteroskedasticity EGARCH (1,1) model is used to fit the differential time series of the induced draft fan bearing temperature, and the three core parameters of the EGARCH (1,1) model variance equation are extracted. The three core parameters include: the symmetric impact coefficient α, which measures the impact of historical fluctuations on current fluctuations and reflects the fluctuation clustering effect; the asymmetric leverage coefficient γ, which captures the leverage effect; and the persistence coefficient β, which reflects the fluctuation memory and indicates the persistence of the impact of historical fluctuations on the current variance.

[0098] The EGARCH time series model effectively captures the volatility characteristics of temperature fluctuations, particularly their clustering and asymmetry. By modeling the logarithm of the volatility index, the EGARCH model addresses the leverage effect in temperature data, specifically the asymmetric impact of the direction of temperature change on volatility. The EGARCH(1,1) model is concise and has few parameters, making it easy to estimate and interpret, making it suitable for volatility analysis of temperature data.

[0099] In three-dimensional space, the three core parameters of the variance equation of the EGARCH (1,1) model form a point; the n×3 parameter matrix Θ represents n points, n = 1000, such as Figure 5 As shown. Using the Mahalanobis distance anomaly detection algorithm, anomaly detection is performed on n points. The Mahalanobis distance and threshold distribution are as follows: Figure 6 As shown. The abnormal point (mark + point) is as follows Figure 7 As shown in Figure 2. The Mahalanobis distance anomaly detection algorithm is an effective statistical method that is particularly suitable for multidimensional data. It considers the correlation between data and is used to identify outliers or outliers in a data set.

[0100] This invention combines the EGARCH model with the Mahalanobis distance anomaly detection algorithm to detect abnormal temperature fluctuations in induced draft fan bearings. This synergy significantly improves detection sensitivity, reduces false alarm rates, and provides efficient early warning before equipment failure, ensuring the safe and stable operation of thermal power generation systems. By integrating equipment sensing, thermodynamic analysis, and big data technologies, this approach, through real-time monitoring, data analysis, and early warning mechanisms, can proactively detect potential induced draft fan bearing failures, enabling real-time diagnosis of equipment health and fault warnings.

[0101] like Figure 2 As shown, another embodiment of the present invention discloses an induced draft fan bearing temperature anomaly detection and early warning system using the induced draft fan bearing temperature anomaly detection and early warning method according to the present invention, comprising: a bearing temperature acquisition device 200, a power plant production equipment SIS database 300, and an induced draft fan bearing temperature anomaly real-time monitoring and early warning server 400;

[0102] In this embodiment, the induced draft fan 100 is used as the monitored object;

[0103] The bearing temperature collection device 200 is used to collect the bearing temperature of the induced draft fan 100;

[0104] The power plant production equipment SIS database 300 is used to obtain monitoring data;

[0105] The induced draft fan bearing temperature anomaly real-time monitoring and early warning server (400) acquires monitoring data, inputs the monitoring data into an implementation monitoring program, and implements the induced draft fan bearing temperature anomaly detection and early warning method; the monitoring data includes: bearing temperature and other power plant equipment real-time operating data.

[0106] In this embodiment, the bearing temperature acquisition device 200 is used to collect the bearing temperature of the induced draft fan 100; the power plant production equipment SIS database stores real-time operating data of the power plant equipment recorded in the Supervisory Information System (SIS). The induced draft fan bearing temperature anomaly real-time monitoring and early warning server 400 is installed with an application program that implements the induced draft fan bearing temperature anomaly detection and early warning method disclosed in the present invention. The application program obtains monitoring data and inputs the monitoring data into a monitoring program to implement the induced draft fan bearing temperature anomaly detection and early warning method; the monitoring data includes bearing temperature and real-time operating data of the power plant equipment.

[0107] Another embodiment of the present invention discloses a computer device, including a memory and a processor, wherein the memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the method for detecting and warning abnormal temperature of an induced draft fan bearing according to the present invention.

[0108] Another embodiment of the present invention discloses a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the method for detecting and warning abnormal temperature of an induced draft fan bearing according to the present invention.

[0109] The present invention discloses a method for detecting and warning abnormal temperature of an induced draft fan bearing. By capturing abnormal changes in the EGARCH model parameters of the induced draft fan bearing temperature difference sequence, various abnormal phenomena and faults of the bearing related to abnormal changes in the induced draft fan bearing temperature are detected. By identifying abnormal changes in the EGARCH model parameters of the induced draft fan bearing temperature difference sequence, the time-varying characteristics and memory effects of temperature fluctuations can be captured, and short-term abnormalities of parameter mutations can be effectively distinguished from long-term trend abnormalities. The EGARCH time series modeling model can effectively capture the volatility characteristics of temperature changes, especially the aggregation and asymmetry of fluctuations. The EGARCH model effectively captures the asymmetric influence of the direction of temperature change on volatility, the so-called "leverage effect". The present invention effectively and significantly improves the efficiency of detecting abnormal operating conditions of induced draft fan bearings, and can reduce the false alarm rate of bearing warnings. It provides quantitative decision support for preventive maintenance of thermal power production equipment.

Claims

1. A method for detecting and warning abnormal temperature of an induced draft fan bearing, characterized in that: The steps include: The induced draft fan bearing temperature time series is divided into continuous overlapping sliding windows and data preprocessing is performed; The EGARCH model is used to fit the fluctuation characteristics in each sliding window, and three core parameters that characterize the sensitivity of model parameters are extracted; The three core parameters include: symmetrical impact coefficient α, asymmetrical leverage coefficient γ and persistence coefficient β; Based on the three core parameters, a three-dimensional time series matrix consisting of n groups of parameters is formed; The anomaly detection algorithm based on Mahalanobis distance is used to detect anomalies in the coefficient matrix to achieve temperature anomaly warning.

2. The method for detecting and warning abnormal temperature of induced draft fan bearings according to claim 1, characterized in that: The method of dividing the induced draft fan bearing temperature time series into continuous overlapping sliding windows and performing data preprocessing includes: Set the sampling period Δt of the induced draft fan bearing temperature data; set the parameters of the dynamic sliding window: the number of sliding windows is n; set m, the sliding window length is m+1, that is, the sliding window contains m+1 induced draft fan bearing temperature time series data; the sliding step of the sliding window is Δt×s (s≥1); the initial time of collecting induced draft fan bearing temperature data is t0, and the start and end time periods of the i-th sliding window are: [t0+(i-1)×Δt×s, t0+(i-1)×Δt×s+m×Δt], where i=1,2,…,n; Use dynamic sliding window to observe the time series data of induced draft fan bearing temperature; The first-order difference processing is performed on the m+1 induced draft fan bearing temperature time series data in the sliding window to obtain the induced draft fan bearing temperature difference time series, and the length of the temperature difference time series is m.

3. The method for detecting and warning abnormal temperature of induced draft fan bearings according to claim 1, characterized in that: The EGARCH model is used to fit the fluctuation characteristics, and the three core parameters that characterize the sensitivity of the model parameters are extracted, including: In each sliding window, the generalized autoregressive conditional heteroskedasticity EGARCH (1,1) model is used to fit the differential time series of the induced draft fan bearing temperature. The three core parameters of the EGARCH (1,1) model variance equation are extracted, including: the symmetric impact coefficient α, which measures the impact of historical fluctuations on current fluctuations and reflects the fluctuation clustering effect; the asymmetric leverage coefficient γ, which captures the leverage effect; and the persistence coefficient β, which reflects the fluctuation memory. The closer β is to 1, the stronger the persistence of the impact of historical fluctuations on the current variance.

4. The method for detecting and warning abnormal temperature of induced draft fan bearings according to claim 1, characterized in that: The forming of a three-dimensional time series matrix consisting of n groups of parameters based on the three core parameters includes: The three core parameters of the variance equation of the EGARCH(1,1) model for the temperature difference time series within each sliding window and the three core parameters of the variance equation of the EGARCH(1,1) model for the difference time series within n sliding windows constitute the n×3 parameter matrix Θ; Where: the i-th row corresponds to the core coefficient of the i-th EGARCH(1,1) model; the first column: α i is the symmetric impact coefficient; the second column: γ i is the asymmetric leverage coefficient; the third column: β i is the persistence coefficient; i=1,2,…,n.

5. The method for detecting and warning abnormal temperature of induced draft fan bearings according to claim 1, characterized in that: The method of using the Mahalanobis distance-based anomaly detection algorithm to perform anomaly detection on the coefficient matrix to implement temperature anomaly warning includes: In three-dimensional space, the three core parameters of the variance equation of the EGARCH (1,1) model constitute a point; the n×3 parameter matrix Θ represents n points; The anomaly detection algorithm based on Mahalanobis distance is used to detect anomalies of n points; if a point (α k , γ k , β k ) to the mean vector is greater than the threshold Threshold, the point is determined to be an abnormal point, k∈[1,2,…,n]; the warning is issued for abnormal temperature fluctuations within the start and end time periods of the sliding window associated with the point [t0+(k-1)×Δt×s, t0+(k-1)×Δt×s+m×Δt], k∈[1,2,…,n]; set Threshold=σ×3, σ is the value of all points (α i , γ i , β i ) to the mean vector, i = 1, 2, …, n; k ∈ [1, 2, …, n].

6. An induced draft fan bearing temperature anomaly detection and early warning system according to the induced draft fan bearing temperature anomaly detection and early warning method according to any one of claims 1 to 5, characterized in that: include: A bearing temperature acquisition device (200), a power plant production equipment SIS database (300), and an induced draft fan bearing temperature abnormality real-time monitoring and early warning server (400); The bearing temperature collecting device (200) is used to collect the bearing temperature of the induced draft fan (100); The power plant production equipment SIS database (300) is used to obtain monitoring data; The induced draft fan bearing temperature abnormality real-time monitoring and early warning server (400) acquires monitoring data, inputs the monitoring data into an implementation monitoring program, and implements the induced draft fan bearing temperature abnormality detection and early warning method; The monitoring data includes: bearing temperature and other real-time operating data of power plant equipment.

7. A computer device, characterized in that: The invention comprises a memory and a processor, wherein a computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the induced draft fan bearing temperature abnormality detection and early warning method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the processor executes the method for detecting and warning abnormal temperature of an induced draft fan bearing according to any one of claims 1 to 5.

Images