Power generation equipment fault analysis method and system

Abstract

The invention discloses a power generation equipment fault analysis method and system. The method comprises the following steps: collecting current operation data of the power generation equipment in real time, and preprocessing the current operation data to obtain a current input vector; based on the current input vector, using an intelligent prediction model obtained by training based on a historical normal operation sample data set to obtain a prediction value of a target measurement point, and comparing the prediction value with a corresponding actual measurement value to obtain a residual value; and comparing the residual value with a residual threshold value obtained based on statistics of a historical normal operation sample data set, and when the residual value exceeds the residual threshold value, judging that the operation state of the power generation equipment is abnormal and triggering early warning. According to the invention, the technical problem of inaccurate monitoring in the operation monitoring of the power generation equipment in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and more specifically, to a method and system for analyzing faults in power generation equipment. Background Technology

[0002] Currently, the automated control systems (DCS, PLC, etc.) of thermal power plants have set alarm functions, capable of generating alarms when monitored parameters reach preset alarm values. Some power generation equipment fault prediction software systems can provide basic early warning functions, giving power plant operators time to intervene in advance. However, these systems generally have the following limitations: they use fixed threshold alarm mechanisms, which cannot adapt to different operating conditions; they lack correlation analysis between parameters, making it impossible to identify gradual fault trends; alarm lag or lead time is too short, usually triggering only when the fault has already occurred or is close to the critical state; and they cannot provide intelligent diagnostic analysis of the fault causes.

[0003] Therefore, power companies urgently need to establish more efficient intelligent monitoring and early warning technologies, and through in-depth mining of existing data resources, achieve proactive early warning of deaerator failures to ensure the safe and reliable operation of equipment. Summary of the Invention

[0004] This invention provides a method and system for analyzing power generation equipment faults, which at least solves the technical problem of inaccurate monitoring in the prior art for power generation equipment operation monitoring.

[0005] According to one aspect of the present invention, a method for fault early warning analysis of power generation equipment is provided, comprising: real-time acquisition of current operating data of power generation equipment; preprocessing the current operating data to obtain a current input vector; based on the current input vector, using an intelligent prediction model trained on a historical normal operation sample dataset to obtain a predicted value of a target measurement point, and comparing the predicted value with the corresponding actual measurement value to obtain a residual value; comparing the residual value with a residual threshold obtained statistically based on the historical normal operation sample dataset; and when the residual value exceeds the residual threshold, determining that the operating state of the power generation equipment is abnormal and triggering an early warning.

[0006] According to another aspect of the present invention, a power generation equipment fault early warning analysis system is also provided, comprising: a preprocessing module configured to collect current operating data of the power generation equipment in real time, preprocess the current operating data to obtain a current input vector; a residual determination module configured to obtain a predicted value of a target measuring point based on the current input vector using an intelligent prediction model trained on a historical normal operating sample dataset, and compare the predicted value with the corresponding actual measured value to obtain a residual value; and an early warning analysis module configured to compare the residual value with a residual threshold obtained statistically based on a historical normal operating sample dataset, and when the residual value exceeds the residual threshold, determine that the operating state of the power generation equipment is abnormal and trigger an early warning.

[0007] In this embodiment of the invention, real-time operating data of the power generation equipment is collected, and the current operating data is preprocessed to obtain a current input vector. Based on the current input vector, an intelligent prediction model trained on a historical normal operation sample dataset is used to obtain a predicted value for the target measurement point. The predicted value is then compared with the corresponding actual measured value to obtain a residual value. The residual value is compared with a residual threshold obtained statistically from the historical normal operation sample dataset. When the residual value exceeds the residual threshold, the operating status of the power generation equipment is determined to be abnormal, and an early warning is triggered. This solution solves the technical problem of inaccurate monitoring in existing power generation equipment operation monitoring technologies. Attached Figure Description

[0008] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart of an optional power generation equipment fault early warning analysis method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a power generation equipment fault early warning system based on an intelligent early warning model according to an embodiment of the present invention; Figure 3 This is a flowchart of another optional power generation equipment fault early warning analysis method according to an embodiment of the present invention; Figure 4 This is an optional data acquisition and preprocessing flowchart according to an embodiment of the present invention; Figure 5 This is an optional model selection measurement point according to an embodiment of the present invention; Figure 6 This is an optional model implementation flowchart according to an embodiment of the present invention; Figure 7This is an early warning display diagram of an optional intelligent monitoring and early warning model for deaerators before and after a fault occurs, according to an embodiment of the present invention. Figure 8 This is a structural diagram of an optional power generation equipment fault early warning and analysis system according to an embodiment of the present invention; Figure 9 A schematic diagram of the structure of a computer device suitable for implementing embodiments of the present disclosure is shown. Detailed Implementation

[0009] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0010] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0011] According to an embodiment of the present invention, a method embodiment of a power generation equipment fault early warning analysis method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0012] The core innovation of this invention lies in establishing a mathematical mapping model between the parameters of the deaerator measuring points through machine learning algorithms, enabling real-time intelligent monitoring of the operating system and providing early warning and auxiliary diagnosis before a fault occurs. Specifically, an SVM regression model is trained based on historical normal operating data, enabling anomaly detection without fault samples; residual analysis is used to identify equipment anomalies through the deviation between predicted and measured values; the model adapts to different operating conditions, overcoming the limitations of fixed threshold alarms; and intelligent diagnosis of fault causes is achieved through multi-measuring point correlation analysis.

[0013] Figure 1 This is a power generation equipment fault early warning analysis method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps: Step S102: Real-time acquisition of current operating data of the power generation equipment, preprocessing of the current operating data to obtain the current input vector.

[0014] From the historical operating data, data exceeding the physical range, data with consecutive identical values ​​exceeding a preset time threshold, and data with a rate of change exceeding the physical limit are removed to obtain cleaned data. Based on the unit load conditions, data within the rated load stable operating range are selected from the cleaned data to obtain normal operating condition data. The normal operating condition data is processed using a normalization method to obtain the historical normal operating sample dataset.

[0015] Step S104: Based on the current input vector, use the intelligent prediction model trained on the historical normal operation sample dataset to obtain the predicted value of the target measurement point, and compare the predicted value with the corresponding actual measurement value to obtain the residual value.

[0016] The process involves collecting historical operating data from the power generation equipment, cleaning the historical operating data, filtering for normal operating conditions, and normalizing the data to obtain a historical normal operating sample dataset. This historical operating data includes at least one of the following measurement parameters: deaerator water level, pressure, temperature, condensate flow rate, heating steam flow rate, and valve opening. Based on this historical normal operating sample dataset, a support vector machine regression algorithm is used to determine the mapping relationship between an input vector composed of multiple related measurement parameters and an output composed of target measurement parameters, thereby obtaining the intelligent prediction model describing the intrinsic correlation between the measurement parameters.

[0017] In some embodiments, a support vector machine regression algorithm is used to determine the mapping relationship between an input vector composed of multiple associated measurement point parameters and an output composed of target measurement point parameters. This includes: selecting multiple associated measurement point parameters that have a physical relationship with the target measurement point parameters from the multiple measurement point parameters, constructing an input vector composed of the multiple associated measurement point parameters, and using the target measurement point parameters as the output; constructing a kernel function using a Gaussian radial basis function kernel function, and modeling the relationship between the input vector and the target measurement point parameters based on the kernel function, Lagrange multipliers, and bias terms, thereby obtaining the mapping relationship.

[0018] In some embodiments, selecting multiple associated measuring point parameters that have a physical relationship with the target measuring point parameter from multiple measuring point parameters includes: selecting multiple associated measuring point parameters that have a physical relationship with the target measuring point parameter based on at least one of the following: there is a mass balance relationship between the deaerator water level and the condensate flow rate and the feedwater pump outlet flow rate; there is a thermodynamic relationship between the deaerator pressure and the heating steam flow rate and the valve opening.

[0019] Step S106: Compare the residual value with the residual threshold obtained from the statistical analysis of historical normal operation sample datasets. When the residual value exceeds the residual threshold, determine that the operating status of the power generation equipment is abnormal and trigger an early warning.

[0020] After determining that the operating status of the power generation equipment is abnormal and triggering an early warning, the method further includes: comprehensively analyzing the residual changes of multiple input measurement points that have a process relationship with the target measurement point, determining the abnormal source measurement point and its corresponding equipment or process link, and obtaining the fault auxiliary diagnosis result.

[0021] This application provides a method for analyzing the faults of deaerators in thermal power units, applied to a power equipment fault early warning system based on an intelligent early warning model. The system, as follows... Figure 2 As shown, a two-stage architecture of "offline training + online prediction" is adopted. The offline training stage is executed before system deployment and during periodic updates. Its core task is to train the prediction model using the Support Vector Machine (SVM) algorithm based on historical normal operation data and establish the mathematical mapping relationship between each measurement point. The online prediction stage is executed in real time during system operation. Its core task is to obtain the predicted value by inputting real-time data and to perform anomaly detection based on residual analysis. Once an anomaly is detected, an alarm is triggered and diagnostic analysis is assisted.

[0022] like Figure 3 As shown, the method for analyzing the deaerator faults of this thermal power unit includes the following steps: Step S302, Data Acquisition and Preprocessing.

[0023] Conduct big data collection, organization, cleaning, and screening of historical operating data of on-site deaerators: 1) Measurement data should retain a complete historical record of all process parameters; 2) Measurement data should contain the inherent correlation between parameters; 3) Use the cleaned data as the basis for automatic modeling.

[0024] Specifically, the data preprocessing steps are as follows: Figure 4 As shown, it includes: 1) Data Collection Data from relevant measuring points of the deaerator are collected from the DCS / PLC system via the OPC interface, including: deaerator water level, pressure, temperature, condensate flow rate, heating steam flow rate, and valve opening. The typical sampling period is 1~10 seconds.

[0025] 2) Data cleaning Abnormal data are removed, including: (1) data that exceeds the physical range; (2) data freeze (the same value is continuously exceeded for a set time); and (3) data jump (the rate of change exceeds the physical limit).

[0026] 3) Normal data filtering Only data from normal operating conditions is retained, excluding data from start-up, shutdown, faults, and maintenance periods. Judgment criterion: The unit operates stably within the range of 30% to 100% of its rated load.

[0027] 4) Data normalization Min-Max standardization is used: X_norm = (X - X_min) / (X_max - X_min), which maps the data to the [0,1] interval to eliminate the influence of dimensions.

[0028] Step S304: Determine the inherent correlation between the measuring point parameters.

[0029] The inherent relationships in this invention have a clear mathematical definition: Definition: Intrinsic correlation refers to the mathematical mapping function between measurement point parameters. Given a vector X = (x1, x2, ..., xn) consisting of n input measurement point parameters, the intrinsic correlation is represented by a mapping function that can predict the target measurement point parameter Y: Y = f(X).

[0030] Mathematical Expression: This invention uses an SVM (Support Vector Machine) regression model to achieve the above mapping relationship, and its mathematical form is as follows: Ypred = Σi (αi - αi ) · K(Xi, X) + b Wherein, K(Xi, X) is the kernel function, and this invention uses the Gaussian radial basis function (RBF): K(x, x') = exp(-γ||x-x'||2); αi, αi is the Lagrange multiplier; b is the bias term. These parameters are automatically learned through the training process.

[0031] Measuring point selection principle: Based on the deaerator process mechanism, select a combination of measuring points with physical correlation, such as... Figure 5 As shown. For example, there is a mass balance relationship between the deaerator water level and the condensate flow rate and the feedwater pump outlet flow rate; there is a thermodynamic relationship between the deaerator pressure and the heating steam flow rate and the valve opening.

[0032] This invention enables comprehensive and reliable online monitoring of deaerator systems. Unlike traditional DCS / PLC systems that monitor only individual measuring points, this invention establishes a mathematical mapping model between measuring point parameters. This allows for simultaneous monitoring of all key parameters, including deaerator water level, pressure, temperature, condensate flow rate, heating steam flow rate, and regulating valve opening, and comprehensively analyzes the correlations between these parameters. The system performs real-time data acquisition and intelligent prediction at second-level intervals (typically 1-10 seconds), ensuring real-time and continuous monitoring.

[0033] Step S306: Training the intelligent monitoring model.

[0034] Based on big data mining and cleaning, a smart monitoring and early warning model for deaerators was designed, trained, and deployed. Model training, such as... Figure 6 As shown, the following steps are performed: 1) Construct the training dataset Training samples are constructed from cleaned historical normal data. Each sample contains: an input vector X = (x1, x2, ..., xn), which is the normalized measurement value of n related measurement points at the same time; and an output Y, which is the normalized measurement value of the target measurement point.

[0035] 2) Dataset partitioning The dataset is divided into training and validation sets in a 7:3 ratio. The training set is used for model parameter learning, and the validation set is used for performance evaluation and hyperparameter tuning.

[0036] 3) SVM model training The ε-SVR algorithm was used for training. The RBF kernel was selected as the kernel function; the penalty coefficient C was searched in the range of {0.1, 1, 10, 100} using a grid search; the kernel parameter γ was optimized in the range of {0.001, 0.01, 0.1, 1}; the training objective was to minimize the ε-insensitive loss function.

[0037] 4) Model Validation and Optimization The RMSE (Root Mean Square Error) is calculated using the validation set, and it is required that the RMSE be less than 1 / 3 of the normal fluctuation range of the target measurement point. If this condition is not met, return to T3 to adjust the hyperparameters and retrain.

[0038] 5) Model solidification and deployment Save the trained model parameters (support vectors, Lagrange multipliers, bias terms, etc.) to the model library for online prediction.

[0039] This invention utilizes an intelligent early warning model for in-depth early warning and auxiliary diagnostic analysis. When the system triggers an early warning, it not only reports the abnormal measuring point but also automatically analyzes the residual state of other measuring points physically related to that point. Through multi-measuring point correlation analysis, it can distinguish between measuring instrument failures and process equipment failures, pinpointing the specific location and possible causes of the failure. For example, if only the position feedback signal is abnormal while the control command is normal, it can be determined that the feedback device is faulty; if multiple thermodynamic parameters are abnormal simultaneously, it may be a problem with the upstream steam supply or the downstream feedwater system. This auxiliary diagnostic function significantly improves the efficiency of fault diagnosis.

[0040] Step S308, Online prediction and fault analysis process.

[0041] During system operation, the following online prediction and anomaly detection process is executed: 1) Real-time data acquisition: The current measurement values ​​of relevant measuring points of the deaerator are acquired in real time through the OPC interface, and the sampling period is consistent with the training data.

[0042] 3) Data preprocessing: Perform the same normalization process on the real-time data as in the training phase.

[0043] 3) Online monitoring: Input the normalized input vector Xcurrent into the trained SVM model and calculate the target measurement point prediction value Ypred = f(Xcurrent).

[0044] 4) Residual Analysis: The residual is calculated as Residual = Yactual - Ypred. The residual reflects the degree of deviation between the actual operating state and the normal mode. This invention employs an anomaly detection method based on residual analysis, which can promptly report abnormal operating behavior before equipment failure occurs, avoiding unexpected downtime. Traditional over-limit alarms only trigger when parameters exceed fixed thresholds, at which point the fault has often already occurred or is close to a critical state. This invention, however, triggers an early warning when the residual exceeds the 3σ threshold by comparing the deviation (residual) between the actual measured value and the intelligent predicted value. This can identify early anomalies where parameters, although not exceeding limits, have deviated from the normal operating mode. Engineering practice shows that this method can detect fault signs several minutes to tens of minutes earlier than traditional alarms, providing operators with a valuable intervention window.

[0045] 5) Anomaly Detection and Alarm: An alarm threshold of Threshold = 3σ (σ is the standard deviation of the residuals during the training phase) is set. When |Residual|>Threshold, an anomaly is detected, triggering an early warning. Unlike traditional fixed-threshold alarms, the intelligent prediction model of this invention can adapt to different operating conditions. By learning the relationship between parameters under various operating conditions in historical normal operating data, the model can dynamically adjust the predicted value according to the operating conditions during changes in unit load, adjustments to operating modes, etc. This adaptive capability solves the problem of high false alarm rates or missed alarms in traditional systems under varying operating conditions, ensuring reliable monitoring across the entire operating range of 30%~100% rated load.

[0046] 6) Auxiliary Diagnosis: When an alarm is triggered, the residuals of all relevant input measurement points are analyzed to identify the root cause of the anomaly. If multiple related measurement points are abnormal at the same time, the scope of the fault can be further narrowed down.

[0047] This invention can promptly report abnormal operating behaviors that lead to performance degradation in deaerator equipment, thus preventing economic losses. Gradual faults such as decreased deaerator heat exchange efficiency, valve internal leakage, and sensor drift are often difficult to detect in traditional monitoring systems. This invention continuously tracks the changing trend of residuals; when the residuals show a continuous shift or a gradual increase, it can identify early signs of equipment performance degradation. This degradation monitoring capability helps optimize equipment maintenance strategies, allowing for maintenance measures to be taken before performance degradation affects unit thermal efficiency and coal consumption rate.

[0048] The engineering verification will be described in detail below.

[0049] Under the steam turbine deaeration system engineering, relevant measuring points such as deaerator water level, pressure, temperature, condensate flow rate, heating steam flow rate, regulating valve opening degree, and position feedback are selected to build a deaerator system equipment operation monitoring and early warning model. Historical data during normal unit operation are selected as training samples, and SVM regression algorithm is used for machine learning modeling. The model is optimized by configuring parameter filtering strategies, optimizing alarm thresholds, defining operating condition switching scenarios, and adjusting training samples.

[0050] The intelligent monitoring and early warning system detected an anomaly: the position feedback signal of the electric regulating valve for auxiliary steam supply to the deaerator abruptly changed to 121.5% (normal range 0~100%), and the system immediately triggered an early warning. The system automatically performed correlation analysis and found that the control command signal for the regulating valve was normal during the same period (except for one occasional jump, it remained near zero), and other operating parameters such as deaerator water level, pressure, and temperature showed no significant abnormalities. Based on the complete operating condition performance of the correlated parameters in the intelligent early warning system, combined with process mechanism analysis, it was initially determined that this anomaly alarm was due to a malfunction in the electric regulating valve's stroke position feedback device (servo control unit circuit board), rather than a malfunction in the valve body or control system.

[0051] Upon verification with on-site personnel, the above-mentioned anomaly was discovered by the operating staff during routine inspections, and maintenance personnel were subsequently notified for investigation. Maintenance confirmed that the cause of the fault was a malfunction in the circuit board of the auxiliary steam to deaerator heating electric control valve, which was completely consistent with the diagnostic judgment of the intelligent early warning system.

[0052] The verification results of this case study indicate that: Early warning timeliness: The intelligent early warning system detected the problem 15 minutes and 58 seconds earlier than the routine detection by operators, fully demonstrating the timeliness value of real-time monitoring.

[0053] Early warning accuracy: The system not only accurately identified abnormal signals, but also correctly located the cause of the fault (circuit board failure rather than other causes) through correlation parameter analysis, verifying the effectiveness of the auxiliary diagnostic function.

[0054] Safety Value: The electric regulating valve for auxiliary steam supply to the deaerator is a critical control device in the deaerator system, and its reliable operation has a significant impact on the safe and stable operation of the entire system. This early warning provided valuable time for operators to respond and prevented potentially greater losses.

[0055] By acquiring real-time signals from instruments and meters at the power plant, and based on the technical solution proposed in this invention, data mining, processing, and analysis of operating conditions are conducted. Machine learning is performed using artificial intelligence algorithms such as SVM to create a real-time intelligent monitoring and early warning system model for the deaerator. This system can detect parameter anomalies and issue alarms in the early stages before a fault occurs. During a fault, it effectively identifies safety hazards in critical control equipment and performs comprehensive analysis based on operating conditions and process mechanisms, verifying the accuracy of the early warning and the significant timeliness value of the technical solution proposed in this invention.

[0056] This application also provides a power generation equipment fault early warning analysis system, such as Figure 8 As shown, it includes: a preprocessing module 82, configured to collect the current operating data of the power generation equipment in real time, preprocess the current operating data to obtain a current input vector; a residual determination module 84, configured to obtain a predicted value of the target measurement point based on the current input vector using an intelligent prediction model trained on a historical normal operation sample dataset, and compare the predicted value with the corresponding actual measurement value to obtain a residual value; and an early warning analysis module 86, configured to compare the residual value with a residual threshold obtained statistically based on a historical normal operation sample dataset, and when the residual value exceeds the residual threshold, determine that the operating status of the power generation equipment is abnormal and trigger an early warning.

[0057] It should be noted that the power generation equipment fault early warning analysis system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the equipment can be divided into different functional modules to complete all or part of the functions described above. In addition, the power generation equipment fault early warning analysis system and the power generation equipment fault early warning analysis method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process is detailed in the method embodiments, which will not be repeated here.

[0058] This application has the following beneficial effects: 1. Avoid unplanned downtime. With early warning capabilities, operators can detect problems before they escalate to a forced shutdown, allowing for timely maintenance during planned outage periods. Unplanned shutdowns not only result in direct power generation losses but can also cause secondary damage to equipment due to emergency shutdowns. For example, avoiding an unplanned shutdown for a 1000MW unit can reduce direct economic losses by hundreds of thousands to millions of yuan.

[0059] 2. Shorten planned maintenance cycles and reduce maintenance costs. Continuous monitoring and early warning of deterioration in the deaerator can optimize maintenance plans, shifting from scheduled maintenance to condition-based maintenance. Maintenance intervals can be appropriately extended when equipment is in good condition, and maintenance can be scheduled in advance when deterioration trends are detected, achieving optimal maintenance cycles. This condition-based maintenance strategy can reduce maintenance frequency, decrease spare parts consumption, and shorten maintenance time, resulting in an overall reduction in maintenance costs of 15% to 30%.

[0060] 3. Proactively arrange maintenance plans. Predictive maintenance, enabled by early warning systems, transforms reactive emergency repairs into proactive maintenance. Maintenance personnel can prepare spare parts, allocate personnel, and coordinate downtime windows based on early warning information, avoiding the high costs and long lead times associated with emergency repairs. Predictive maintenance significantly improves the planning and controllability of maintenance work.

[0061] 4. Improve the economic efficiency of unit operation. By monitoring the performance degradation of the deaerator, problems affecting thermal efficiency can be identified and resolved promptly. A 1% decrease in deaerator heat exchange efficiency may lead to an increase in unit coal consumption rate of 0.1~0.2 g / kWh. The degradation monitoring function of this invention can detect problems before efficiency losses accumulate, maintaining optimal equipment performance through timely maintenance. Furthermore, the system can provide data support for operational optimization, assisting operators in selecting the most economical combination of operating parameters.

[0062] 5. Improve equipment reliability and unit safety level The stable operation of the deaerator directly affects the safety of the condensate system and the entire thermal cycle. This invention, through intelligent monitoring technology, elevates equipment reliability management from reactive to proactive prevention, significantly reducing equipment failure rates. Simultaneously, the auxiliary diagnostic function helps maintenance personnel quickly locate problems, reducing erroneous operations caused by misjudgments, and overall improving the safe operation level of the unit.

[0063] Figure 9 A schematic diagram of a computer device suitable for implementing embodiments of the present disclosure is shown. It should be noted that... Figure 9 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments disclosed herein.

[0064] like Figure 9 As shown, the computer device includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage section 1008 into a random access memory (RAM) 1003. The RAM 1003 also stores various programs and data required for system operation. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0065] The following components are connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 1010 as needed so that computer programs read from it can be installed into storage section 1008 as needed.

[0066] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for early warning analysis of power generation equipment faults, characterized in that, include: Real-time acquisition of current operating data of power generation equipment; preprocessing of the current operating data to obtain the current input vector; Based on the current input vector, the intelligent prediction model trained on the historical normal operation sample dataset is used to obtain the predicted value of the target measurement point, and the predicted value is compared with the corresponding actual measurement value to obtain the residual value; The residual value is compared with the residual threshold obtained from the statistical analysis of historical normal operation sample datasets. When the residual value exceeds the residual threshold, the operating status of the power generation equipment is determined to be abnormal and an early warning is triggered.

2. The method according to claim 1, characterized in that, The intelligent prediction model is obtained through the following: Historical operating data of the power generation equipment is collected, and the historical operating data is cleaned, normal operating condition filtered, and normalized to obtain a historical normal operating sample dataset. The historical operating data includes at least one of the following measurement parameters: deaerator water level, pressure, temperature, condensate flow rate, heating steam flow rate, and valve opening. Based on the historical normal operation sample dataset, the support vector machine regression algorithm is used to determine the mapping relationship between the input vector composed of multiple related measurement point parameters and the output composed of target measurement point parameters, so as to obtain the intelligent prediction model describing the intrinsic correlation of measurement point parameters.

3. The method according to claim 2, characterized in that, The support vector machine regression algorithm is used to determine the mapping relationship between the input vector composed of multiple associated measurement point parameters and the output composed of target measurement point parameters, including: Select multiple related measurement point parameters that have a physical relationship with the target measurement point parameter from multiple measurement point parameters, construct an input vector composed of the multiple related measurement point parameters, and use the target measurement point parameter as the output; A kernel function is constructed using the Gaussian radial basis function. Based on the kernel function, Lagrange multipliers, and bias terms, a support vector machine regression algorithm is used to model the relationship between the input vector and the target measurement point parameters, thereby obtaining the mapping relationship.

4. The method according to claim 3, characterized in that, Multiple related measuring point parameters that have a physical correlation with the target measuring point parameter are selected from multiple measuring point parameters, including: based on at least one of the following: there is a mass balance relationship between the deaerator water level and condensate flow rate and the feedwater pump outlet flow rate; there is a thermodynamic relationship between the deaerator pressure and the heating steam flow rate and the valve opening.

5. The method according to claim 2, characterized in that, After determining that the operating status of the power generation equipment is abnormal and triggering an early warning, the method further includes: comprehensively analyzing the residual changes of multiple input measurement points that have a process relationship with the target measurement point, determining the abnormal source measurement point and its corresponding equipment or process link, and obtaining the fault auxiliary diagnosis result.

6. The method according to claim 2, characterized in that, The historical operating data is cleaned, filtered for normal operating conditions, and normalized to obtain a historical normal operating sample dataset, including: From the historical operating data, data that exceeds the physical range, data with consecutive identical values ​​exceeding a preset time threshold, and data with a rate of change exceeding the physical limit are removed to obtain cleaned data. Based on the unit load conditions, data within the rated load stable operating range are selected from the cleaned data to obtain normal operating condition data. The normal operating condition data is processed using a normalization method to obtain the historical normal operating sample dataset.

7. A fault early warning and analysis system for power generation equipment, characterized in that, include: The preprocessing module is configured to collect the current operating data of the power generation equipment in real time, preprocess the current operating data, and obtain the current input vector. The residual determination module is configured to obtain the predicted value of the target measurement point based on the current input vector using an intelligent prediction model trained on a historical normal operation sample dataset, and compare the predicted value with the corresponding actual measurement value to obtain the residual value. The early warning analysis module is configured to compare the residual value with a residual threshold obtained from the statistical analysis of historical normal operation sample datasets. When the residual value exceeds the residual threshold, the module determines that the operating status of the power generation equipment is abnormal and triggers an early warning.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 6.

9. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor performs the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images