An integrated safety management system and method for thermal power plants

Abstract

This invention discloses an integrated safety management system and method for thermal power plants, relating to the field of thermal power production safety control technology. The system quantifies the collected real-time disturbance dataset using a preset coupling algorithm to generate a risk pre-coupling factor characterizing the risk of multi-source disturbance coupling. Combined with the current operating status set characterizing the equipment's operating condition, it generates predicted operating status for future time points. The predicted operating status is compared with a preset operating safety threshold to calculate the safety and stability margin, dynamically adjusting the unit's original control objectives. The system generates a prediction error by comparing the predicted operating status with subsequently collected actual operating statuses. This prediction error is used to correct the prediction model's correction parameters, achieving online adaptive adjustment of the prediction model. This enables a quantitative assessment of multiple disturbance coupling effects, providing a decision-making basis for predictive risk prevention and control.

Description

Technical Field

[0001] This invention relates to the field of thermal power production safety control technology, specifically to an integrated thermal power enterprise safety management system and method. Background Technology

[0002] As a crucial link in energy supply, the safety and stability of the operation of thermal power units are of paramount importance. Traditional safety management systems typically rely on threshold monitoring of key operating parameters, often viewing various disturbance sources in isolation. Response mechanisms are mostly passive, triggering alarms or protection actions only when parameters actually exceed preset safety limits, lacking foresight and failing to provide operators with sufficient response time, increasing the risk of unplanned shutdowns. Control strategies are generally rigid; when faced with safety risks, the control system often rigidly switches between ensuring unit safety and meeting grid dispatch instructions, lacking flexible adjustment mechanisms. Control and prediction models are mostly static; during long-term operation, factors such as wear and scaling of unit equipment can cause their dynamic characteristics to drift, making it difficult for static models to adapt to such changes, leading to a gradual decline in the accuracy of prediction and control.

[0003] Therefore, how to overcome the problems of slow response, one-sided risk assessment, rigid control strategies and static models in traditional safety management systems, and develop an integrated safety management system that can proactively assess coupled risks, flexibly adjust control objectives and has adaptive capabilities, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated safety management system for thermal power plants to solve the problems mentioned in the background art.

[0005] The technical solution of this invention is: an integrated safety management system for thermal power plants, comprising:

[0006] The risk quantification module is used to collect real-time disturbance datasets, quantify the real-time disturbance datasets according to a preset coupling algorithm, and generate a risk pre-coupling factor characterizing the risk of multi-source disturbance coupling; wherein, the real-time disturbance dataset includes load disturbance data characterizing changes in grid load and fuel disturbance data characterizing changes in fuel quality.

[0007] The vision prediction module is used to generate a predicted operating state for future time nodes based on the risk pre-coupling factor and the current operating state set representing the current operating status of the equipment, according to the prediction model; and to calculate the safety and stability margin by comparing the predicted operating state with a preset operating safety threshold.

[0008] The dynamic optimization module is used to dynamically adjust the original control target of the unit according to the safety and stability margin, and generate a safety-corrected control target. The safety-corrected control target is used to guide the equipment operation under the condition of meeting safety constraints.

[0009] An adaptive correction module is used to generate a prediction error by comparing the predicted operating state with the subsequently collected actual operating state; based on the prediction error, a prediction model correction parameter is generated to correct the prediction model, thereby realizing the online adaptive adjustment of the prediction model.

[0010] Preferably, the risk quantification module is further used to collect a combustion stability index characterizing the furnace combustion conditions; and the risk quantification module includes:

[0011] The load disturbance data is determined as the power grid load dispatch rate;

[0012] By performing differential calculations on the fuel moisture data entering the furnace, the fuel disturbance data characterizing the rate of change of fuel moisture is generated;

[0013] The risk pre-coupling factor is generated by weighting and summing the power grid load dispatch rate and the fuel moisture change rate, and then normalizing the sum according to the combustion stability index.

[0014] Preferably, the weighting coefficients used in the coupling algorithm to weight the power grid load dispatch rate and the fuel moisture change rate are obtained by performing multiple linear regression analysis on historical non-outage operation data.

[0015] Preferably, the current operating state set includes the current turbine vibration intensity; the horizon prediction module includes:

[0016] Using the prediction model, the current turbine vibration intensity and the risk pre-coupling factor are taken as inputs to generate the predicted vibration intensity as the predicted operating state.

[0017] The safety and stability margin is calculated by subtracting the predicted vibration intensity from the turbine vibration interlock protection value, which serves as the operational safety threshold.

[0018] Preferably, the dynamic optimization module includes:

[0019] Based on the aforementioned safety and stability margin, a boost rate suppression coefficient is calculated using a preset nonlinear function. When the safety and stability margin is greater than a high-level threshold, the boost rate suppression coefficient approaches 1. When the safety and stability margin is between the low-level and high-level thresholds, the boost rate suppression coefficient continuously varies between 0 and 1. When the safety and stability margin is lower than the low-level threshold, the boost rate suppression coefficient approaches 0.

[0020] The original control target is multiplied by the boost rate suppression coefficient to generate the safety-corrected control target.

[0021] Preferably, the original control target is the original grid scheduling pressure boosting target, and the safety-corrected control target is the safety pressure boosting rate; the dynamic optimization module is further used to generate a dynamic main steam pressure setpoint for distribution to the coordinated control system based on the safety pressure boosting rate.

[0022] Preferably, the adaptive correction module includes:

[0023] The prediction error is generated by subtracting the actual operating state from the predicted operating state.

[0024] The prediction error is multiplied by a preset feedback gain coefficient to generate an adjustment amount for the correction parameters of the prediction model;

[0025] By iteratively updating the correction parameters of the prediction model, online adaptive adjustment is achieved.

[0026] Preferably, an integrated safety management method for thermal power plants includes the following steps:

[0027] Collect real-time disturbance datasets, quantize the real-time disturbance datasets according to a preset coupling algorithm, and generate a risk pre-coupling factor that characterizes the risk of multi-source disturbance coupling.

[0028] Based on the aforementioned risk pre-coupling factors and the current operating status set representing the current operating status of the equipment, a predicted operating status for future time nodes is generated according to the prediction model; the predicted operating status is compared with a preset operating safety threshold to calculate the safety stability margin.

[0029] Based on the aforementioned safety and stability margin, the original control objectives of the unit are dynamically adjusted to generate safety-corrected control objectives.

[0030] By comparing the predicted operating state with the subsequently collected actual operating state, a prediction error is generated; based on the prediction error, prediction model correction parameters are generated to correct the prediction model, thereby realizing the online adaptive adjustment of the prediction model.

[0031] This invention provides an integrated safety management system and method for thermal power plants, which has the following beneficial effects:

[0032] (1) By integrating real-time disturbance data of multi-source heterogeneous grid load changes and fuel quality fluctuations, the system generates risk pre-coupling factors based on the coupling algorithm, realizes the quantitative assessment of the coupling effects of various disturbances, shifts the starting point of safety management from post-event response to pre-event assessment, and provides a decision-making basis for predictive risk prevention and control.

[0033] (2) Based on the risk pre-coupling factor and the current operating state set, the system predicts key parameters such as the turbine vibration intensity at future time nodes and compares them with the operating safety threshold to calculate the safety stability margin. This enables the system to proactively identify potential risks, quantify the distance between the unit and the safety boundary, and realize the transformation from passive triggering protection to active avoidance.

[0034] (3) Based on the safety and stability margin, the system generates the boost rate suppression coefficient through a nonlinear function, and makes a flexible and smooth adjustment to the original control target to generate a safety-corrected control target. Under the premise of ensuring safety constraints, it can maximize the satisfaction of the grid dispatching requirements and enhance the unit's operational flexibility and control flexibility under complex operating conditions.

[0035] (4) By continuously comparing the predicted operating status with the actual operating status, the system uses the prediction error to make online adaptive adjustments to the correction parameters of the prediction model, which solves the problem that the traditional static model gradually fails due to its inability to adapt to changes in the unit. Attached Figure Description

[0036] Figure 1 This is a flowchart of the system of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0038] Example 1

[0039] This embodiment provides an integrated safety management system for thermal power plants. Through the synergistic effect of a risk quantification module and a vision prediction module, the system achieves real-time perception and forward-looking early warning of operational risks in thermal power plants, thereby effectively improving the safety and stability of unit operation.

[0040] In practical applications, the risk quantification module measures the power grid load dispatch rate. Online data on moisture content of fuel entering the furnace Real-time data collection is performed, combined with the furnace combustion stability index. The risk pre-coupling factor F is generated through quantitative processing, and the calculation formula is as follows:

[0041] ;

[0042] in, For the power grid load dispatch rate weighting coefficient, For power grid load dispatch rate, The weighting coefficient for the rate of change in moisture content of fuel entering the furnace. The change rate of moisture content in fuel entering the furnace, The combustion stability index in the furnace;

[0043] and The weighting coefficients, obtained through multiple linear regression analysis of historical non-outage operation data, reflect the relative importance of grid load disturbances and fuel disturbances on system vibration. The combustion stability index is a dimensionless index that characterizes the stability of the combustion process. The closer its value is to 1, the more stable the combustion state. When the moisture content of the fuel is reduced, the risks associated with changes in fuel moisture content are amplified, which is highly consistent with the risks of thermal shock and thermal stress coupling caused by unstable combustion in the actual operation of thermal power plants.

[0044] For example, in scenarios where the grid load fluctuates rapidly or the fuel quality suddenly declines, the risk quantification module can quickly capture these disturbances and quantify them into a single risk pre-coupling factor, providing a unified risk assessment basis for subsequent prediction and control.

[0045] This proactive risk assessment enables the system to shift from a reactive, alarm-based mode to a proactive, pre-assessment mode, allowing valuable time for unit response and adjustments.

[0046] The horizon prediction module receives the risk pre-coupling factor. and the current turbine vibration intensity Using the current operating state set as input, and based on a preset prediction model, the predicted vibration intensity within the horizon in the short term is generated. Predicting vibration intensity The calculation formula is:

[0047] ;

[0048] in, The current vibration intensity of the steam turbine, To predict the time step, It is an increment function. The current main steam pressure, To correct parameters for the prediction model. It is an incremental function fitted using historical data, and its internal key coefficients are adjusted by the parameters of the prediction model. Online dynamic adjustments are made to reflect real-time changes in unit characteristics and ensure the accuracy of forecasts;

[0049] The horizon prediction module will predict the vibration intensity. Interlock protection value with preset turbine vibration By comparing the results, the safety and stability margin was calculated. The calculation formula is: ( This is the value for turbine vibration interlock protection. (For predicting vibration intensity), safety and stability margin It can be directly quantified as the safe distance between the current operation and the triggering of emergency shutdown protection. For example, when the safety and stability margin is small, it indicates that the unit's operating status is close to the safety threshold. The system can issue an early warning and trigger the dynamic optimization module to intervene.

[0050] This proactive avoidance logic based on predictive safety margins enables the system to move from passively waiting for vibration values ​​to approach the interlocking protection boundary to proactively predicting future vibration trends and calculating the gap between them and the safety boundary, thus transforming from passive protection response to predictive proactive protection.

[0051] Example 2

[0052] This embodiment further refines the functions of the risk quantification module, enabling it to generate risk pre-coupling factors more accurately.

[0053] The risk quantification module directly determines the power grid load dispatch rate by analyzing load disturbance data. This can directly reflect rapid changes in the power grid's load commands to the generating units. Simultaneously, by analyzing fuel moisture data... Perform differential calculations to generate the fuel moisture change rate. This is a better indicator than a simple fuel moisture content of the thermal shock to the combustion process caused by drastic fluctuations in fuel quality. For example, when fuel moisture increases rapidly within a short period of time... The risk will increase significantly, and the risk quantification module can immediately identify this potential risk of combustion instability;

[0054] The risk quantification module also collects combustion stability indices that characterize furnace combustion conditions. This index is determined by the furnace temperature. Furnace pressure and oxygen content in flue gas Several key parameters are calculated using a weighted model, for example: ( These are the weighting coefficients corresponding to furnace temperature, furnace pressure, and flue gas oxygen content, respectively. For the normalization function, For furnace temperature, For furnace pressure, Oxygen content in flue gas, combustion stability index The closer the value is to 1, the closer the combustion state is to the ideal stable state.

[0055] The risk quantification module will determine the grid load dispatch rate. With the change rate of fuel moisture Perform a weighted summation based on the combustion stability index. Normalization is performed to generate risk pre-coupling factors. ;

[0056] This precise calculation of the risk pre-coupling factor enables the system to more sensitively capture potential risks of thermal shock and thermal stress coupling, providing more reliable input for subsequent prediction and control, and effectively avoiding the limitation that a single disturbance index is insufficient to fully reflect the risks under complex working conditions.

[0057] Example 3

[0058] This embodiment further illustrates the method for determining the weighting coefficients used to weight the power grid load dispatch rate and the fuel moisture change rate in the coupled algorithm.

[0059] The weighting coefficients in the coupling algorithm are determined by collecting historical data on the quasi-steady-state operation of the generating units over the past 1 to 2 years, including grid load dispatch rates. Online data on moisture content of fuel entering the furnace Furnace combustion stability index and the actual vibration intensity of the steam turbine During the data screening process, data segments under abnormal operating conditions such as unit start-up and shutdown, load shedding, overhaul, and major equipment defects are strictly excluded to ensure that the data used for analysis can truly reflect the dynamic response characteristics of the unit within the normal operating boundary. The data synchronization sampling frequency is usually set to 1 minute / time to capture the effective dynamic process of load and fuel changes.

[0060] After normalization, the grid load dispatch rate L and the combustion stability index S are identified as key influencing factors of fuel moisture change rate. These two parameters (primary independent variables) are fitted to the turbine vibration change (dependent variable). The weighting coefficients and calibration process employ a statistical method of multiple linear regression. The objective is to minimize the sum of squared errors between the actual vibration intensity and the model prediction. The mathematical expression is:

[0061] ;

[0062] in, For power grid load dispatch rate weighting coefficient; The weighting coefficient for the rate of change in moisture content of fuel entering the furnace; Let be the actual change in the vibration intensity of the steam turbine caused by the disturbance at the i-th data point; Let be the power grid load dispatch rate for the i-th data point; Let be the rate of change of fuel moisture content at the i-th data point; Let be the furnace combustion stability index for the i-th data point; This represents the total number of data points.

[0063] The optimal fit coefficient can be obtained through this regression analysis. and For this specific unit, these two coefficients are accurately quantified as the contributions of grid load disturbance and fuel moisture disturbance to vibration risk, respectively. This method, based on historical data calibration, enables the risk quantification module to more accurately assess the risk of multi-source disturbance coupling, avoiding biases caused by subjective experience.

[0064] Example 4

[0065] This embodiment further clarifies the input and output of the horizon prediction module and elaborates in detail the calculation process of predicting vibration intensity and safety and stability margin;

[0066] The horizon prediction module uses the current turbine vibration intensity and risk-preceding coupling factors As the core input, combined with the current main steam pressure And the prediction model correction parameters from the adaptive correction module Using a pre-set prediction model, the predicted vibration intensity can be forecast within the next 3-5 minutes. The calculation formula is:

[0067] ;

[0068] in, The current vibration intensity of the steam turbine; To predict the time step; It is an increment function; This is the current main steam pressure; Adjust parameters for the prediction model;

[0069] Increment function It is a multivariable nonlinear function, for example, it can be expressed in the form of a second-order polynomial:

[0070] ;

[0071] in, The coefficients are the increment function coefficients; the correction parameters are the prediction model parameters. It is a collection of all coefficients The parameter vector;

[0072] Fine-tuning is performed through an online calibration process to ensure the accuracy of predictions. For example, when the unit's operating characteristics experience slight drift, the prediction model's parameters are corrected. It can automatically adjust to predict vibration intensity. Always maintain a high degree of consistency with the actual situation;

[0073] The horizon prediction module will calculate the predicted vibration intensity. Interlock protection value for turbine vibration Compare and calculate the safety and stability margin. The calculation formula is: ( This refers to the turbine vibration interlock protection value. These are mandatory safety thresholds explicitly defined in the unit design specifications; (For predicting vibration intensity), safety and stability margin This is directly quantified as the safe distance between the current operation and triggering emergency shutdown protection. For example, when the safety and stability margin... A high level indicates that the unit is operating smoothly and has sufficient safety redundancy; while a low safety and stability margin indicates that the unit is operating smoothly and has sufficient safety redundancy. When the value gradually decreases, it indicates a potential risk, and the system can take timely intervention measures.

[0074] This precise calculation of predicted vibration intensity and safety margin enables thermal power plants to proactively assess unit operating risks, providing a basis for decision-making by the dynamic optimization module, thereby achieving accurate control over unit operating status and proactive safety management.

[0075] Example 5

[0076] This embodiment details how the dynamic optimization module dynamically adjusts the original control objectives of the unit based on the safety and stability margin.

[0077] The dynamic optimization module is used to calculate and generate a boost rate suppression coefficient based on a safety and stability margin using a preset nonlinear function. When the safety and stability margin is greater than a high-level threshold, the boost rate suppression coefficient approaches 1; when the safety and stability margin is between the low-level and high-level thresholds, the boost rate suppression coefficient continuously changes between 0 and 1; when the safety and stability margin is lower than the low-level threshold, the boost rate suppression coefficient approaches 0. The original control target is multiplied by the boost rate suppression coefficient to generate a safety-corrected control target.

[0078] The dynamic optimization module receives a safety and stability margin. As input, a boost rate suppression coefficient is calculated using a preset Sigmoid nonlinear function. The calculation formula is:

[0079] ;

[0080] in, This is the boost rate suppression coefficient; Parameters to suppress the steepness of the curve; For safety and stability margin; For the response threshold, the parameter and Determining the steepness of the suppression curve and the response threshold has clear engineering significance and is easy to tune online. Its initial value can be statistically analyzed based on the safety margin distribution before the protection action is triggered in the unit's historical operating data to ensure that the suppression curve takes effect smoothly within a reasonable safety margin range.

[0081] For example, when the safety and stability margin When the value is large, the boost rate suppression coefficient A value approaching 1 indicates that the unit is operating safely and that pressure increase is permitted according to the original control objectives; when the safety and stability margin is... Decrease and fall below the threshold At that time, the boost rate suppression coefficient The voltage rise rate rapidly approaches 0, thus strongly suppressing the rate of increase. When the safety margin M is between the high and low thresholds, the rate of increase suppression coefficient I smoothly transitions between 0 and 1. This intermediate zone is key to achieving flexible control; it ensures that control adjustments are a continuous and smooth process, rather than abrupt steps. Specifically, the response threshold... This point can be considered the center of the transition region. The system is most sensitive to changes in the safety margin near this point. The high and low thresholds can be understood as defining the main range of change in the Sigmoid curve from "virtually no suppression" (I≈1) to "strong suppression" (I≈0). Simultaneously, an alarm threshold can be set in the system, which can be set to the same value as the response threshold. Equal to or as set according to safety procedures, when the safety stability margin When the temperature falls below this alarm threshold, the system will trigger an audible and visual alarm to alert operators.

[0082] This nonlinear characteristic enables continuous and smooth adjustment of control silence in risk-free states, fine-tuning intervention in low-risk scenarios, and strong suppression control in high-risk conditions by constructing a risk-level adaptive adjustment mechanism. This avoids the secondary oscillations that may be caused by traditional rule-based step control adjustments, ensuring the flexibility and stability of control.

[0083] The dynamic optimization module then returns the original control target The calculated boost rate suppression coefficient Multiply to generate the safe boost rate as the control target after safety correction. , ( To ensure safety, the control target was corrected. The original control objective; (This refers to the boost rate suppression coefficient), and the control target after safety correction. Used to guide equipment operation under conditions that meet safety constraints, for example, in situations with low safety stability margins, the safety-corrected control objective. This will be significantly reduced, thus effectively preventing the unit from entering a dangerous area due to blindly pursuing high load;

[0084] This nonlinear dynamic flexible control enables thermal power plants to maximize the satisfaction of grid dispatching needs while ensuring unit safety, generating a flexible boost curve that can both maximize grid dispatching needs and ensure unit safety.

[0085] Example 6

[0086] This embodiment further clarifies the specific application of the control target after safety correction, specifying the safety boost rate. Steps applied to subsequent control.

[0087] To achieve precise control of the unit, the dynamic optimization module further bases its control on this safe boost rate. The dynamic main steam pressure setpoint is generated for distribution to the coordinated control system. The calculation formula is as follows:

[0088] ;

[0089] in, for The dynamic main steam pressure setpoint at any given time; for The dynamic main steam pressure setpoint at any given time; For safe boost rate; For time step;

[0090] Dynamic main steam pressure setpoint It can adjust the main steam pressure in real time to adapt to the unit's operating requirements under different safety margins. For example, when the unit's vibration risk increases, the dynamic main steam pressure setpoint will be adjusted to slow down the pressure rise rate and effectively suppress vibration deterioration.

[0091] This mechanism, which generates a dynamic main steam pressure setpoint based on a safe pressurization rate, enables more refined and intelligent control of the unit, achieving flexible control of unit operation and improving the operating efficiency and safety of thermal power plants.

[0092] Example 7

[0093] This embodiment details how the adaptive correction module achieves online adaptive adjustment of the prediction model by comparing the predicted operating state with the subsequently collected actual operating state;

[0094] The adaptive correction module calculates the predicted vibration intensity output by the horizon prediction module after each control cycle. The actual vibration intensity collected in real time by the sensor Prediction error The formula is: ( This represents the prediction error; This represents the actual vibration intensity. (To predict vibration intensity)

[0095] Prediction error It directly reflects the accuracy of the prediction model. For example, when the dynamic characteristics of a unit drift due to wear, scaling, or other reasons, the prediction error... The value will gradually increase, indicating that the prediction model needs to be revised.

[0096] Adaptive correction module based on prediction error With learning rate To generate the adjustment amount for correcting the parameters of the prediction model, we first define the increment function. The corresponding input feature vector:

[0097] ;

[0098] Subsequently, the least mean square algorithm is used to update the parameters of the prediction model. The update formula is as follows:

[0099] ;

[0100] in, for The parameter vector for the prediction model at each time step; for The parameter vector for the prediction model at each time step; The learning rate is used to control the speed and magnitude of adjustments. for The prediction error at any given time; for The input feature vector at time step;

[0101] Prediction model correction parameters It is an increment function The parameter vector of the model coefficients, with initial values It was obtained through offline historical data regression analysis, and the parameters of the prediction model were adjusted accordingly. Iterative updates are performed to enable online adaptive adjustment of the prediction model. For example, when the operating environment of the unit changes slowly, the adaptive correction module can continuously learn and correct the parameters of its own prediction model to ensure that the prediction model accurately reflects the actual operating status of the unit.

[0102] This prediction error-driven adaptive correction closed loop can solve the problem of traditional static models gradually becoming ineffective due to their inability to adapt to the drift of unit operating characteristics, and realize the continuous self-updating of the entire safety management system.

[0103] Example 8

[0104] This embodiment provides an integrated safety management method for thermal power plants, which includes the following steps:

[0105] Collect real-time disturbance datasets, quantize the real-time disturbance datasets according to a preset coupling algorithm, and generate risk pre-coupling factors that characterize the risk of multi-source disturbance coupling.

[0106] Based on the risk pre-coupling factors and the current operating status set representing the current operating status of the equipment, the predicted operating status at future time points is generated according to the prediction model; the safety and stability margin is calculated by comparing the predicted operating status with the preset operating safety threshold.

[0107] Based on the safety and stability margin, the original control objectives of the unit are dynamically adjusted to generate safety-corrected control objectives;

[0108] By comparing the predicted operating status with the actual operating status collected subsequently, a prediction error is generated. Based on the prediction error, prediction model correction parameters are generated to correct the prediction model, thereby realizing the online adaptive adjustment of the prediction model.

[0109] This method collects real-time disturbance datasets through a risk quantification module and generates risk pre-coupling factors based on a preset coupling algorithm. This step enables the system to proactively identify and quantify the coupling risks of multi-source disturbances, providing a basis for subsequent decision-making. For example, when the grid load changes drastically or fuel quality fluctuates, the system can quickly perceive and quantify these risks, avoiding the one-sided focus on a single disturbance source in traditional methods.

[0110] The visibility prediction module, based on the risk pre-coupling factor and the current operating state set, generates predicted vibration intensity for future time nodes according to the prediction model. It then compares the predicted vibration intensity with the turbine vibration interlock protection value to calculate the safety and stability margin. This step enables the prediction of the unit's future operating state and the assessment of its safety margin, allowing the system to proactively avoid potential risks. For example, when the safety and stability margin decreases, the system can provide early warning, buying time for intervention and preventing shutdown accidents caused by excessive vibration.

[0111] The dynamic optimization module calculates the boost rate suppression coefficient using a nonlinear function based on the safety and stability margin, and dynamically adjusts the original control target of the unit to generate a safe boost rate. This step enables flexible adjustment of the unit's control target, balancing safety and economy. For example, when the risk is high, the system will reduce the safe boost rate to ensure the safe operation of the unit; while when the risk is low, the unit is allowed to respond quickly to grid dispatching needs and improve operating efficiency.

[0112] The adaptive correction module generates a prediction error by comparing the predicted vibration intensity with the actual vibration intensity collected subsequently. Combined with the feedback gain coefficient, it generates prediction model correction parameters to correct the prediction model, realizing online adaptive adjustment of the prediction model. This step ensures the long-term accuracy of the prediction model and its adaptability to changes in unit characteristics. For example, when the unit's operating characteristics drift, the adaptive correction module can automatically correct the prediction model to ensure that the system continuously provides accurate risk assessment and control commands.

[0113] Through the coordinated execution of the above steps, this integrated safety management method for thermal power plants enables comprehensive, intelligent, and proactive management of operational risks.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An integrated safety management system for thermal power plants, characterized in that, include: The risk quantification module is used to collect real-time disturbance datasets, quantify the real-time disturbance datasets according to a preset coupling algorithm, and generate a risk pre-coupling factor characterizing the risk of multi-source disturbance coupling; wherein, the real-time disturbance dataset includes load disturbance data characterizing changes in grid load and fuel disturbance data characterizing changes in fuel quality. The vision prediction module is used to generate a predicted operating state for future time nodes based on the risk pre-coupling factor and the current operating state set representing the current operating status of the equipment, according to the prediction model; and to calculate the safety and stability margin by comparing the predicted operating state with a preset operating safety threshold. The dynamic optimization module is used to dynamically adjust the original control target of the unit according to the safety and stability margin, and generate a safety-corrected control target. The safety-corrected control target is used to guide the equipment operation under the condition of meeting safety constraints. An adaptive correction module is used to generate a prediction error by comparing the predicted operating state with the subsequently collected actual operating state; based on the prediction error, a prediction model correction parameter is generated to correct the prediction model, thereby realizing the online adaptive adjustment of the prediction model. The risk quantification module is also used to collect a combustion stability index characterizing the furnace combustion conditions; and the risk quantification module includes: The load disturbance data is determined as the power grid load dispatch rate; By performing differential calculations on the fuel moisture data entering the furnace, the fuel disturbance data characterizing the rate of change of fuel moisture is generated; The risk pre-coupling factor is generated by weighting and summing the power grid load dispatch rate and the fuel moisture change rate, and then normalizing the sum according to the combustion stability index. The current operating status set includes the current turbine vibration intensity; the horizon prediction module includes: Using the prediction model, the current turbine vibration intensity and the risk pre-coupling factor are taken as inputs to generate the predicted vibration intensity as the predicted operating state. The safety and stability margin is calculated by subtracting the predicted vibration intensity from the turbine vibration interlock protection value, which serves as the operating safety threshold. The dynamic optimization module includes: Based on the aforementioned safety and stability margin, a boost rate suppression coefficient is calculated using a preset nonlinear function. When the safety and stability margin is greater than a high-level threshold, the boost rate suppression coefficient approaches 1. When the safety and stability margin is between a low-level threshold and a high-level threshold, the boost rate suppression coefficient continuously varies between 0 and 1. When the safety and stability margin is lower than a low-level threshold, the boost rate suppression coefficient approaches 0. The original control target is multiplied by the boost rate suppression coefficient to generate the safety-corrected control target.

2. The integrated safety management system for thermal power plants according to claim 1, characterized in that, The weighting coefficients used in the coupling algorithm to weight the power grid load dispatch rate and the fuel moisture change rate are obtained by performing multiple linear regression analysis on historical non-outage operation data.

3. The integrated safety management system for thermal power plants according to claim 2, characterized in that, The original control target is the original grid scheduling pressure boosting target, and the safety-corrected control target is the safety pressure boosting rate; the dynamic optimization module is also used to generate a dynamic main steam pressure setpoint for distribution to the coordinated control system based on the safety pressure boosting rate.

4. The integrated safety management system for thermal power plants according to claim 1, characterized in that, The adaptive correction module includes: The prediction error is generated by subtracting the actual operating state from the predicted operating state. The prediction error is multiplied by a preset feedback gain coefficient to generate an adjustment amount for the correction parameters of the prediction model; By iteratively updating the correction parameters of the prediction model, online adaptive adjustment is achieved.

5. An integrated safety management method for thermal power plants, characterized in that: The method is performed by the system according to any one of claims 1-4, and includes the following steps: Collect real-time disturbance datasets, quantize the real-time disturbance datasets according to a preset coupling algorithm, and generate a risk pre-coupling factor that characterizes the risk of multi-source disturbance coupling. Based on the aforementioned risk pre-coupling factors and the current operating status set representing the current operating status of the equipment, a predicted operating status for future time nodes is generated according to the prediction model; the predicted operating status is compared with a preset operating safety threshold to calculate the safety stability margin. Based on the aforementioned safety and stability margin, the original control objectives of the unit are dynamically adjusted to generate safety-corrected control objectives. By comparing the predicted operating state with the subsequently collected actual operating state, a prediction error is generated; based on the prediction error, prediction model correction parameters are generated to correct the prediction model, thereby realizing the online adaptive adjustment of the prediction model.

Images