Thermal power plant monitoring alarm method and system based on state estimation and intelligent linkage

Abstract

The invention discloses a thermal power plant monitoring alarm method and system based on state estimation and intelligent linkage. According to the method, system state estimation is carried out through extended Kalman filtering, and fault diagnosis and sensor false alarm identification are realized through state chi-square detection; based on a diagnosis result, a first-out alarm identification and alarm storm suppression technology is adopted to intelligently filter redundant alarms; performing dynamic priority ranking on the filtered alarms in combination with safety, economy and equipment trend; and finally, according to the priorities, triggering intelligent voice alarms in a partitioned and graded manner, and automatically linking corresponding emergency disposal schemes. According to the invention, closed-loop linkage from accurate diagnosis and intelligent filtering to efficient disposal is realized, the problems of alarm flooding and false alarm are effectively solved, and the emergency disposal efficiency and the operation safety are improved.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation and fault diagnosis technology, specifically relating to an intelligent monitoring method and system for the centralized control center of a thermal power plant, and in particular to an alarm intelligent filtering, dynamic priority assessment and emergency response linkage method based on state estimation and fault diagnosis. Background Technology

[0002] As the backbone power source of the power system, the safe, stable, and economical operation of thermal power plants is of paramount importance. The centralized control center monitoring system is the "nerve center" of the power plant's operation, responsible for real-time monitoring and control of the operating status of all major equipment and systems. Alarm functionality, as one of the core functions of the monitoring system, is the first line of defense for on-duty personnel to promptly detect equipment anomalies, diagnose faults, and prevent the escalation of accidents.

[0003] However, with the continuous expansion of thermal power plants and the increasing complexity of their systems, traditional monitoring system alarm mechanisms have revealed numerous drawbacks and are no longer sufficient to meet the safety requirements of modern smart power plants. The alarm mechanisms of monitoring systems commonly used in the control centers of current thermal power plants have significant defects. The main problem is that alarm rules are based on single measurement points or simple logic settings. When equipment failure occurs, this easily triggers a chain reaction, resulting in a massive influx of irrelevant alarms, creating an "alarm storm." This not only overwhelms critical accident information but also renders voice alarms unreliable due to continuous false alarms. Furthermore, the alarm system and emergency response measures are independent of each other, requiring operators to rely on personal experience to find contingency plans, leading to slow responses and a high risk of accident escalation. Although improvements exist by increasing alarm delays or manual grouping, these methods cannot fundamentally achieve intelligent correlation analysis and accurate root cause determination of alarm signals. Therefore, there is an urgent need for an intelligent method capable of deeply identifying alarm logic, automatically filtering redundant information, and rapidly linking emergency response plans to improve the safety level of power plants. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage, which can effectively suppress redundant alarms, accurately identify the root cause of faults, and automatically associate and activate standardized emergency response plans.

[0005] Another objective of this invention is to provide a monitoring and alarm system for thermal power plants based on state estimation and intelligent linkage.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage includes the following steps:

[0008] Based on real-time data from the monitoring system of thermal power plants, system status estimation and fault diagnosis are performed.

[0009] Based on the results of the fault diagnosis, the alarm signals are intelligently filtered.

[0010] For the filtered alarm events, their priority level is determined through multi-factor dynamic risk assessment;

[0011] Based on the priority level, execute the corresponding intelligent voice broadcast and emergency response linkage;

[0012] Provides a human-computer interaction confirmation interface, which, in response to a confirmation operation, records the confirmation information and suppresses the repeated voice broadcast of the corresponding alarm.

[0013] Preferably, the system state estimation and fault diagnosis specifically include:

[0014] A stochastic state-space model of a discrete nonlinear system is constructed. The extended Kalman filter algorithm is used to linearize the stochastic state-space model of the nonlinear system, resulting in a linearized model of the system. The optimal estimation of the system state is achieved through state updates and measurement updates.

[0015] Based on the system state estimation results, the state chi-square test method is used for fault diagnosis, including: constructing a first state estimate that has only been updated and a second state estimate that has been updated by measurement, calculating a statistic of the difference between the two, and determining whether the system has failed based on the statistic.

[0016] Preferably, the intelligent filtering of alarm signals includes at least one of the following methods:

[0017] Identify and highlight the first alarm among multiple related alarms; merge or hierarchically suppress avalanche alarms caused by the same fault root cause; perform causal reasoning based on the time series of alarm signals; and suppress or mark false alarms triggered by faulty sensors based on fault diagnosis results.

[0018] Preferably, the step of determining the priority level through multi-factor dynamic risk assessment specifically involves assigning weight values ​​to the assessment factors, obtaining dynamic risk values ​​based on the calculated weight values, and determining the priority level from high to low based on the dynamic risk values.

[0019] Preferably, the evaluation factors include: impact on unit safety, impact on unit economy, equipment importance, failure development trend, and current operating conditions; the priority levels include: urgent, high, medium, and low.

[0020] Preferably, the execution of the corresponding intelligent voice broadcast specifically includes:

[0021] Based on the alarm type, priority, and associated device information, generate voice broadcast content; broadcast in zones according to the physical area where the alarm device is located; and broadcast in levels according to the priority level of the alarm.

[0022] Preferably, the execution of emergency response coordination includes:

[0023] In response to an alarm of a specific priority, the monitoring interface will display a corresponding emergency response card; the emergency response card dynamically integrates real-time operational data and presents standardized operation guidance steps.

[0024] A monitoring and alarm system for thermal power plants based on state estimation and intelligent linkage includes:

[0025] The data acquisition and assessment / diagnosis module is used to perform system status estimation and fault diagnosis based on real-time data from the thermal power plant monitoring system.

[0026] An alarm signal filtering module is used to intelligently filter alarm signals based on the results of the fault diagnosis.

[0027] The dynamic risk assessment module is used to determine the priority level of filtered alarm events through multi-factor dynamic risk assessment.

[0028] The voice broadcast and emergency response linkage module executes corresponding intelligent voice broadcast and emergency response linkage according to the priority level.

[0029] The human-computer interaction module provides a human-computer interaction confirmation interface. In response to the confirmation operation, it records the confirmation information and suppresses the repeated voice broadcast of the corresponding alarm.

[0030] Preferably, it also includes an operation log archiving module, which is used to continuously record the lifecycle information of alarm events, including event triggering, status changes, handling operations, personnel confirmation and signal reset, forming a traceable closed-loop management archive.

[0031] The advantages and effects of this invention are as follows: By integrating extended Kalman state estimation and state chi-square detection, this invention achieves accurate perception of system operating status and early accurate diagnosis of sensor and equipment faults, reducing false alarm rates from the source; it effectively suppresses alarm storms by using intelligent filtering strategies, compressing massive amounts of underlying signals into a few key events, greatly reducing the cognitive load on operators; it introduces a multi-factor dynamic risk assessment model, enabling alarm priority to truly and in real-time reflect risk levels, ensuring that critical alarms are handled first; through automatic linkage between alarms and handling plans, it transforms the traditional passive response relying on manual intervention into proactive standardized operations guided by system intelligence, improving response speed and standardization; ultimately, it forms an intelligent closed-loop management process from status monitoring, fault diagnosis, alarm prompts to handling confirmation, comprehensively enhancing the reliability and safety of thermal power plant operation. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0033] Figure 2 This is a block diagram of the system configuration of the present invention. Detailed Implementation

[0034] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, but the implementation of the present invention is not limited thereto.

[0035] Example 1

[0036] like Figure 1 As shown, the present invention provides a monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage, comprising the following steps:

[0037] Based on the real-time operation data of the DCS and SIS of the thermal power plant, and according to the main interlocking parameters, alarm settings, protection settings and control logic diagrams of the equipment, a comprehensive and standardized alarm signal source is established.

[0038] For units such as boilers, turbines, and electrical systems in thermal power plants, a stochastic state-space model of a discrete nonlinear system is constructed, and system noise and measurement noise are characterized as additive white Gaussian noise. An extended Kalman filter algorithm is employed, which linearizes the nonlinear function around the filter value to the first order and follows the basic framework of state and measurement updates to achieve optimal estimation of the nonlinear system state. The stochastic system state-space model is shown below:

[0039]

[0040] In the formula, This represents the n-dimensional state quantity at time k. Let f(g) and h(g) represent the m-dimensional measurement at time k, respectively, where f(g) and h(g) are the kinematic state equation and measurement equation following a Gaussian probability distribution, and w represents the system noise and measurement noise, respectively. k-1 and ν k It follows a Gaussian white noise distribution with zero mean.

[0041] (1) From the perspective of probability density, the state-space model can be restated as:

[0042]

[0043] In the formula, N(g) is the Gaussian distribution function.

[0044] (2)w k-1 and ν k It follows a Gaussian white noise distribution with zero mean, and the two are uncorrelated, E(w) k ) = 0, E(v)k ) = 0, Their variances are Q k-1 and R k .

[0045]

[0046] (3) At the initial time 0, the system state x0 and variance P0 satisfy the following equation:

[0047]

[0048] In practical applications, linear approximation of nonlinear system models is a common approach, although it introduces a small error. Extended Kalman filtering incorporates this idea, linearizing the nonlinear vector functions f(g) and h(g) in the stochastic nonlinear system model around the filter value, resulting in a linearized model of the system. Then, the basic framework of Kalman filtering is used to achieve the optimal estimation of the nonlinear system.

[0049] The extended Kalman filter procedure from time k-1 to k is shown below.

[0050] (1) Status update:

[0051] Calculate the one-step optimal state estimate at time k-1. The mean square error matrix of the state estimation is P k / k-1 ,Right now:

[0052]

[0053] (2) Measurement update:

[0054] Calculate the predicted observation y k With the observation mean square error P yy,k And the covariance matrix P between state one-step prediction and measurement one-step prediction xy,k ,Right now:

[0055]

[0056] Calculate the filter gain K k and state estimation at time k The corresponding mean square error is P k ,Right now:

[0057]

[0058] In the formula, P is the estimated value at time k. k Filtering variance. The most crucial aspect of the Extended Kalman Array (EKA) is solving for the linearized state transition matrix and the linearized measurement matrix, which are expressed using multivariate Jacobian matrices, i.e.:

[0059]

[0060] From the above extended Kalman derivation, it can be seen that, due to the first-order linearization of the state transition matrix and measurement matrix of the nonlinear system, the state at time k-1 is the optimal estimate.

[0061] Fault detection and diagnosis are important ways to improve the reliability of alarms in thermal power plants. Fault detection improves the reliability of the system from the overall design, and by detecting faults through system detection, necessary measures can be taken to isolate faulty devices and re-integrate normal components, so that the entire system can still work normally even when there are internal faults.

[0062] This embodiment constructs a Kalman filter framework for the subsystem states in a thermal power plant and employs a state-based chi-square method for fault identification. The state-based chi-square detection method primarily uses two state recursion methods to calculate the difference between them, constructing fault detection statistics for fault detection and isolation. Where x... 1,k This is a state variable updated after measurement, and it is affected by faults. Another state variable is x. 2,k It only performs state updates, recursively obtaining the state using prior indicators. By calculating the difference between the two estimates, the operating state of the system can be determined. Without loss of generality, the state estimate x... 1,k and x 2,k The calculation formula and process are shown below.

[0063]

[0064] The initial state is defined as a Gaussian random vector, so the state variable x k x 1,k and x 2,k All three are random vectors, and their relationships are determined by the state variable x. k Using this as a basis, we obtain the estimation error e. 1,k and e 2,k , as shown in the following formula.

[0065]

[0066] Define the difference between estimation errors as θ k .

[0067]

[0068] Its corresponding variance M k Described as follows.

[0069]

[0070] When the system is fault-free, θ kIt is a Gaussian random vector x 1,k and x 2,k A linear combination of θ. Therefore, θ k It is also a Gaussian vector, following a zero mean and variance of M. k The distribution of P 12,k =P 1,k Therefore, the above equation can be expressed as follows.

[0071] M k =P 2,k -P 1,k (20)

[0072] x 2,k This indicates that the state estimate obtained through recursive state updates is unaffected by measurement information, and therefore still satisfies E[e] when a fault occurs. 2,k ] = 0. Conversely, x 1,k The state estimate is obtained by filtering the measurement information. When the sensor malfunctions, the measurement information is contaminated, and its state estimate is no longer an unbiased estimate. Therefore, E[e 2,k ]≠0. When the sensor malfunctions, E[e 2,k ]≠0, the above is for selecting θ k This serves as the theoretical basis for the detection statistics used to determine whether a fault has occurred.

[0073] For θ k The following binary assumptions can be made:

[0074] H0 (No fault): E[θ] k ] = 0,

[0075] H1 (Faulty):

[0076] Based on the above mathematical statistics theory, the detection statistic function is constructed as follows.

[0077]

[0078] In the formula, the detection statistic λ k It follows a chi-square distribution with m degrees of freedom, i.e., λ k :χ 2 (m), and the fault judgment criteria are set as follows based on the above principles.

[0079] When λ k <T D When this happens, the system is deemed to be working normally;

[0080] When λ k ≥T D When this happens, the system is deemed to have malfunctioned.

[0081] Among them, TD The detection threshold can be obtained from the chi-square distribution table based on the set false alarm rate. Using the above method, the deviation between the state variables output by the prior state recursive model and the state variables output by the filter can be used to detect whether the sensor has malfunctioned, thus enabling fault detection and isolation.

[0082] The intelligent filtering of alarm signals in this embodiment is implemented as follows: Based on the above fault diagnosis results, one or more of the following intelligent filtering strategies are executed:

[0083] "First alarm" identification: When multiple related alarms are triggered simultaneously, the system can automatically identify and highlight the most fundamental and first alarm, which is the "first alarm", to help the duty officer quickly locate the root cause.

[0084] Alarm storm suppression: When a device malfunction triggers an avalanche of alarms, the system should be able to automatically merge similar alarms or only display the highest-level parent alarm to avoid the screen being overwhelmed.

[0085] Reasoning based on time sequence: Analyze the time sequence of alarm signal occurrences to determine causal relationships, thereby more accurately identifying the actual accident sequence.

[0086] False alarm identification based on fault detection: Using the measurement information generated in step three, the system determines that the sensor data is unreliable, automatically suppresses or significantly marks the alarms triggered by this sensor, and generates equipment calibration and maintenance work orders, greatly reducing the false alarm rate.

[0087] Through the above processing, the system can filter redundant, duplicate, secondary, and false alarms, and compress a series of alarms caused by the same fault root cause into a comprehensive alarm event.

[0088] In this embodiment, the dynamic prioritization of alarms is implemented as follows: For filtered alarm events, a dynamic risk assessment is performed based on a multi-factor weighted evaluation model.

[0089] The weighting factors are set by the shift operator. Evaluation factors include: Impact on unit safety (highest weight): including whether it is associated with protection tripping, the scope of impact, and the severity of consequences. Impact on unit economy: including impact on coal consumption for power supply, unit load, and plant power consumption rate. Importance of the equipment itself: main equipment has a higher weight than important auxiliary equipment, and important auxiliary equipment has a higher weight than general auxiliary equipment. Speed ​​and trend of fault development: including parameter change rate analysis and predictive diagnosis based on state estimation. Current operating condition of the unit: the same alarm has dynamic priority under different operating conditions.

[0090] Based on the calculated dynamic risk values, alarm events are categorized into four priority levels: P0 (Emergency / Accident), P1 (High / Warning), P2 (Medium / Abnormal), and P3 (Low / Reminder), generating an ordered alarm list. Specifically: P0 (Emergency / Accident): Alarms that may lead to unit tripping, damage to major equipment, or serious personal injury accidents. Immediate action is required. P1 (High / Warning): Indicates a serious equipment abnormality; without intervention, it is likely to develop into a P0-level accident. Operators need to pay close attention and prepare for immediate action. P2 (Medium / Abnormal): General equipment abnormalities affecting economic efficiency or long-term equipment health; action is required within a short time. P3 (Low / Reminder): Equipment status deviates from optimal values, or indicates items requiring periodic inspection.

[0091] Based on the priority level, execute the corresponding intelligent voice broadcast and emergency response linkage: This step includes two core linkage links.

[0092] The intelligent voice generation and broadcasting system, upon receiving high-priority alarm information processed in step five, first automatically generates standardized and concise voice broadcast statements based on the alarm type, priority, and associated equipment information. Then, according to the physical area where the alarm equipment is located (e.g., boiler area, turbine area, electrical unit), it broadcasts the alarm in a targeted, zoned manner using audio equipment deployed in the corresponding area to minimize interference with personnel in unrelated positions. Regarding the broadcasting strategy, the system implements a strict hierarchical broadcasting mechanism: general alarms (P3) only display text or emit a soft prompt tone on the workstation interface; early warnings (P2) are broadcast once in the central control center with a highlighted flashing interface; important alarms (P1) initiate multiple repeated voice broadcasts across the entire area to attract the attention of all on-duty personnel; while emergency / accident alarms (P0) activate a forced broadcast mode, interrupting all secondary voices and broadcasting repeatedly at the highest volume across the entire area to ensure that critical information is delivered.

[0093] In emergency response coordination, when a specific highest-priority alarm such as MFT or turbine trip is triggered, the system will activate a voice alarm, and the linkage control module will automatically display a unique "Emergency Response Card" on the main monitoring screen. This card is not static text but dynamically integrates real-time data, including the initial trip signal and key parameter trend charts, to help operators quickly diagnose the root cause of the fault. Simultaneously, the card presents a standard handling procedure in a clear, step-by-step format, guiding operators to perform confirmation, equipment isolation, auxiliary equipment startup / shutdown, reporting, and parameter adjustments in sequence. This effectively avoids misoperation and omission of key steps in emergency situations, improving the standardization and efficiency of emergency response.

[0094] Human-computer interaction and confirmation: Operators can quickly locate problems and execute actions on the handling card based on voice prompts and on-screen guidance. The system provides a "Confirmed" button. When the operator clicks "Confirm," the system records the confirmation time and personnel, and can stop the repeated voice broadcast of the alarm according to settings. However, the alarm text record will still be retained until the fault is completely eliminated.

[0095] Example 2

[0096] like Figure 2 As shown, the present invention discloses a monitoring and alarm system for thermal power plants based on state estimation and intelligent linkage, comprising:

[0097] The data acquisition and assessment / diagnosis module is used to perform system status estimation and fault diagnosis based on real-time data from the thermal power plant monitoring system.

[0098] An alarm signal filtering module is used to intelligently filter alarm signals based on the results of the fault diagnosis.

[0099] The dynamic risk assessment module is used to determine the priority level of filtered alarm events through multi-factor dynamic risk assessment.

[0100] The voice broadcast and emergency response linkage module executes corresponding intelligent voice broadcast and emergency response linkage according to the priority level.

[0101] The human-computer interaction module provides a human-computer interaction confirmation interface. In response to the confirmation operation, it records the confirmation information and suppresses the repeated voice broadcast of the corresponding alarm.

[0102] The operation log archiving module is used to continuously record the lifecycle information of alarm events, including event triggering, status changes, handling operations, personnel confirmation, and signal reset, forming a traceable closed-loop management archive.

[0103] The other contents are the same as in Example 1.

[0104] Specific Application Example 1

[0105] This embodiment provides a monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage, which is specifically implemented in the centralized control center monitoring system of a 300MW coal-fired unit.

[0106] This embodiment focuses on demonstrating the effectiveness of fault detection: fault detection and diagnosis of the main steam pressure sensor in the combustion system and emergency handling of primary air fan trip when the unit load is 80%.

[0107] S101. Alarm Signal Creation and Access: In this embodiment, the system uses the standard OPC protocol to collect approximately 5,000 key process variables in real time from the power plant's existing Distributed Control System (DCS) and System-Specific Information (SIS), including boiler main steam temperature, turbine speed, generator active power, coal feed rate, and fan current. Based on the unit's protection logic diagram, interlocking control diagram, and operating procedures, alarm and protection settings are set for these variables. This establishes a comprehensive and standardized alarm signal source database within the system.

[0108] S102. System State Estimation: The power plant boiler combustion system mainly consists of a main steam pressure sensor, a pulverizing system, and a load controller. The main steam pressure is adjusted to supply power to the pulverizing system, thereby outputting the grid-specified unit load. In this system, because the thermal system parameters change slowly, the following pure time-delay steady nonlinear discretized model can be used to represent it.

[0109]

[0110] In the formula, x represents the system state at time k, y represents the system output at time k, and f(g) and h(g) are nonlinear functions. Taking the main steam pressure state estimator 1 as an example: the inputs are the steam drum pressure, the main pipe pressure, and the output feedback of the main steam pressure sensor; the output is the output of the main steam pressure sensor. Similarly, the main pipe pressure state estimator 2 and the steam drum pressure state estimator 3 are constructed. Substituting into the extended Kalman filter algorithm mentioned above, in each sampling period, the system performs the following operations: predicting the state at the current time based on the state estimate value of the previous time and the combustion system model; comparing the predicted state with the actual sensor readings in the DCS, and correcting the predicted value using Kalman gain to obtain the optimal state estimate value. Through the extended Kalman algorithm, the system can effectively smooth measurement noise, identify and compensate for small sensor drifts, and output combustion system parameters that are closer to the true values.

[0111] S103. Fault Detection and Diagnosis, taking the main steam pressure state estimator 1 as an example. Main steam pressure state quantity one: obtained recursively only through the extended Kalman state update equation in S102, unaffected by the current pressure measurement value. Main steam pressure state quantity two: obtained after the extended Kalman measurement update in S102, dependent on the current pressure measurement value. The system calculates the difference between the above two state quantities in real time and constructs a chi-square detection statistic. According to system safety requirements, the false alarm rate α = 0.01 is set, and the detection threshold is obtained from the chi-square distribution table. When the detection statistic exceeds the threshold for three consecutive sampling periods, i.e., when λ k ≥T DAt this point, the system determines that the main steam pressure, steam drum pressure, and header pressure may all be malfunctioning. Ultimately, a malfunction in the main steam pressure sensor can be diagnosed through an exhaustive search method.

[0112] Action: Once a fault is diagnosed, the system automatically marks all alarms triggered by the sensor as "sensor fault, suspected false alarm" in S104, and generates a maintenance work order for "verifying the main steam pressure sensor" and pushes it to the equipment management system.

[0113] S104. Intelligent filtering of alarm signals: When a combustion system malfunction is not caused by a sensor malfunction false alarm but by a high-voltage trip caused by high vibration of a primary air fan bearing in the pulverizing system, the DCS system triggers more than 120 related alarms within 1 second.

[0114] First alarm identification: By analyzing the timing of the trip signal, the system accurately identifies "excessive vibration of primary fan bearing" as the first and most fundamental alarm, and highlights it in the alarm list.

[0115] Alarm storm suppression: The system automatically merges over a hundred secondary alarms, such as "primary air fan motor failure," "low primary air volume," and "abnormal furnace negative pressure," into a single parent alarm event called "primary air fan trip." The operator's interface prominently displays only this parent alarm, which can be expanded to view all details by clicking on it, freeing the screen from constant scrolling.

[0116] S105. Dynamic sorting of alarm priorities: For the "primary fan trip" alarm event obtained after filtering in S104, the system starts a multi-factor dynamic evaluation model to perform calculations.

[0117] Evaluation process:

[0118] Safety: A primary wind turbine tripping directly leads to a rapid reduction in unit load. If not handled properly, it may cause a main fuel trip, with serious consequences. Therefore, this item scores extremely high.

[0119] Economic efficiency: This caused the unit load to drop to 150MW instantly, and the coal consumption for power supply increased. This item scored highly.

[0120] Equipment importance: The primary air fan is an important auxiliary machine and has high weight.

[0121] Development trend: The negative pressure in the furnace is decreasing rapidly, which is a dangerous trend.

[0122] Operating conditions: The current unit load is 80% high load.

[0123] Result: After weighted calculation, the dynamic risk value of this alarm falls within the P0 (emergency / accident) range. The system displays it at the top and marks it in red.

[0124] S106. Multi-channel intelligent voice alarm and emergency response linkage, intelligent voice broadcast: Since this alarm is at the P0 level, the system activates the forced broadcast mode. All audio zones (boiler, turbine, electrical) in the central control center simultaneously broadcast at a high volume: "Emergency, No. 1 primary air fan tripped, the unit is rapidly reducing load," while interrupting other low-priority voice broadcasts.

[0125] Emergency Response Linkage: Simultaneously with the voice announcement, the linkage control module automatically displays a "Primary Air Fan Trip Emergency Response Card" on the operator's main monitoring screen. The top of the card displays the initial trip signal: high primary air fan bearing vibration, along with trend graphs of key parameters including: primary air fan bearing vibration, primary air fan current, furnace negative pressure, and unit load. Below, clear step-by-step instructions (1, 2, 3...) guide the operator through the process.

[0126] S107. Human-Machine Interaction and Confirmation: After hearing the voice alarm and seeing the handling card, the operator immediately follows the instructions on the card. Once the furnace negative pressure is confirmed to have stabilized initially, the operator clicks the "Confirmed" button on the handling card. The system then records the confirmation time and operator's employee number in the log. Repeated voice announcements regarding "Number One Primary Air Fan Tripped" cease to avoid continuous noise interference. The red text record of the alarm event remains at the top of the alarm list until the fault is completely cleared and the signal is reset.

[0127] This implementation case not only significantly reduced the risk of false alarms and effectively suppressed alarm storms, but also achieved precise location, intelligent guidance, and efficient handling when faults occur, thereby greatly improving the safety, economy, and emergency response efficiency of unit operation and providing a replicable and scalable advanced paradigm for intelligent operation and maintenance of power systems.

[0128] This invention also provides a storage medium storing a computer program. When executed by a processor, the computer program implements some or all of the steps in various embodiments of the thermal power plant monitoring and alarm method based on state estimation and intelligent linkage provided by this invention. The storage medium may be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0129] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of the present invention.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage, characterized in that, Includes the following steps: Based on real-time data from the monitoring system of thermal power plants, system status estimation and fault diagnosis are performed. Based on the results of the fault diagnosis, the alarm signals are intelligently filtered. For the filtered alarm events, their priority level is determined through multi-factor dynamic risk assessment; Based on the priority level, execute the corresponding intelligent voice broadcast and emergency response linkage; Provides a human-computer interaction confirmation interface, which, in response to a confirmation operation, records the confirmation information and suppresses the repeated voice broadcast of the corresponding alarm.

2. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 1, characterized in that, The system state estimation specifically refers to, A stochastic state-space model of a discrete nonlinear system is constructed. The extended Kalman filter algorithm is used to linearize the stochastic state-space model of the nonlinear system, resulting in a linearized model of the system. The optimal estimation of the system state is achieved through state updates and measurement updates.

3. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 2, characterized in that, The fault diagnosis specifically involves using the state chi-square test method based on the system state estimation results. This includes: constructing a first state estimate that has only been updated and a second state estimate that has been updated by measurement; calculating a statistic of the difference between the two; and determining whether the system has experienced a fault based on the statistic.

4. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 1, characterized in that, The intelligent filtering of alarm signals includes at least one of the following methods: Identify and highlight the first alarm among multiple related alarms; merge or hierarchically suppress avalanche alarms caused by the same fault root cause; perform causal reasoning based on the time series of alarm signals; and suppress or mark false alarms triggered by faulty sensors based on fault diagnosis results.

5. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 1, characterized in that, The process of determining the priority level through multi-factor dynamic risk assessment involves assigning weight values ​​to assessment factors, obtaining dynamic risk values ​​based on the calculated weight values, and determining the priority level from high to low based on the dynamic risk values.

6. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 5, characterized in that, The evaluation factors include: impact on unit safety, impact on unit economy, equipment importance, failure development trend, and current operating conditions; the priority levels include: emergency, high, medium, and low.

7. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 1, characterized in that, The execution of the corresponding intelligent voice broadcast specifically involves, Based on the alarm type, priority, and associated device information, generate voice broadcast content; broadcast in zones according to the physical area where the alarm device is located; and broadcast in levels according to the priority level of the alarm.

8. The monitoring and alarm method for thermal power plants based on state estimation and intelligent linkage according to claim 1, characterized in that, The aforementioned emergency response coordination includes: In response to an alarm of a specific priority, the monitoring interface will display a corresponding emergency response card; the emergency response card dynamically integrates real-time operational data and presents standardized operation guidance steps.

9. The system corresponding to the thermal power plant monitoring and alarm method based on state estimation and intelligent linkage according to any one of claims 1-8, characterized in that, include: The data acquisition and assessment / diagnosis module is used to perform system status estimation and fault diagnosis based on real-time data from the thermal power plant monitoring system. An alarm signal filtering module is used to intelligently filter alarm signals based on the results of the fault diagnosis. The dynamic risk assessment module is used to determine the priority level of filtered alarm events through multi-factor dynamic risk assessment. The voice broadcast and emergency response linkage module executes corresponding intelligent voice broadcast and emergency response linkage according to the priority level. The human-computer interaction module provides a human-computer interaction confirmation interface. In response to the confirmation operation, it records the confirmation information and suppresses the repeated voice broadcast of the corresponding alarm.

10. The thermal power plant monitoring and alarm system based on state estimation and intelligent linkage according to claim 9, characterized in that, It also includes an operation log archiving module, which continuously records the lifecycle information of alarm events, including event triggering, status changes, handling operations, personnel confirmation and signal reset, forming a traceable closed-loop management archive.

Images