A method, device and system for controlling a coal-fired power plant

Abstract

This application provides a method, apparatus, and system for downloading control strategies for coal-fired power plants. The method includes collecting multi-source production data from the power plant, performing heterogeneous adaptation and standardization processing to generate standard data frames; loading the control strategy to be implemented into a shadow running unit, running it in parallel based on the standard data frames, and obtaining verification results; performing gray-scale downloading of the control strategy to be implemented under preset gating conditions; and real-time monitoring and execution of tiered rollback operations in case of abnormal situations. This application solves the problem of differences between offline simulation and the real operating environment by setting up a shadow running unit to run the control strategy to be implemented in parallel; by performing gray-scale downloading, it avoids the sudden operational risks caused by full write; furthermore, by real-time monitoring of the operating status and execution of tiered rollback operations, it can automatically and quickly handle abnormal situations, improving the safety and reliability of downloading control strategies for coal-fired power plants and reducing operational risks.

Description

Technical Field

[0001] This application relates to the field of control technology, specifically to a method, device, and system for implementing control strategies in a coal-fired power plant. Background Technology

[0002] Coal-fired power plants are an important component of modern power systems, and their stable operation is a crucial foundation for ensuring the reliability of power supply.

[0003] In existing technologies, the distributed control system (DCS) of a coal-fired power plant is the control center of the plant, playing a crucial role in monitoring unit operating parameters, controlling equipment logic, and providing safety interlock protection. Updating and downloading control strategies for the DCS is a necessary operation in the daily operation and maintenance, performance optimization, and technical upgrades of coal-fired power plants, and its safety and reliability directly affect the unit's operating efficiency and production safety.

[0004] Currently, the download of control strategies for coal-fired power plants generally adopts an implementation method of directly writing the entire strategy after offline simulation verification. Technicians complete the writing and debugging of control logic in an offline environment, perform simulation verification based on historical operating data, and then write the complete control strategy into the distributed control system at once, making it effective immediately. Because the offline simulation environment differs from the real production environment, the above method cannot fully reproduce complex operating scenarios such as coal quality fluctuations, equipment characteristic drift, and dynamic switching of operating conditions, thus resulting in numerous potential control defects. Furthermore, existing download technologies lack a robust verification mechanism and automatic anomaly handling capabilities. The operational status after the control strategy is downloaded still mainly relies on manual real-time monitoring, which suffers from long response cycles and a high risk of human error.

[0005] In summary, existing control strategies for coal-fired power plants suffer from low safety, poor reliability, and uncontrollable operational risks. Summary of the Invention

[0006] This application addresses the problems existing in the prior art by providing a method, apparatus, and system for implementing control strategies in coal-fired power plants, thereby resolving the technical issues.

[0007] To achieve the above objectives, the technical solution adopted in this application is as follows: This application provides a method for downloading control strategies for coal-fired power plants, which includes: Collect multi-source production data from multiple heterogeneous production subsystems of a coal-fired power plant; The multi-source production data is subjected to heterogeneous adaptation and standardization processing to generate standardized data frames; Set up a shadow execution unit and load the control strategy to be implemented into the shadow execution unit; Based on the standardized data frame, the control strategy to be implemented and the currently implemented control strategy run in parallel, and generate shadow running results and actual running results; The verification results are obtained by comparing the shadow execution results with the actual execution results; If the verification results meet the preset gating conditions, the control strategy to be implemented will be downloaded in grayscale. In grayscale download, the operating status of the coal-fired power plant is monitored in real time; if an abnormal condition is detected in the operating status, a graded rollback operation is performed on the coal-fired power plant control system.

[0008] Optionally, collecting multi-source production data includes: Data from different sources in the coal-fired power plant is collected through an industrial protocol adapter interface to obtain initial collected data; The initial collected data is parsed and converted to obtain preprocessed data; Establish a synchronized time reference based on a globally unified clock; The preprocessed data is calibrated using the synchronized time reference, and the multi-source production data is obtained.

[0009] Optionally, heterogeneous adaptation and standardization processing of the multi-source production data includes: Generate a standardized data frame containing object identifier, variable identifier, timestamp, data value, engineering units, and quality status; In the standardized data frame, the data identified as continuously sampled variables are resampled in a time-synchronized manner and mapped to a globally unified time axis. In the standardized data frame, the data identified as the updated variable is subjected to the most recent valid value interpolation and preservation until new valid data is received; In the standardized data frame, the data identified as event-type variables are marked with an event timestamp, and the state data at the time of event triggering is stored.

[0010] Optionally, the control policy to be implemented and the currently implemented control policy receive standardized data frames from the same source; The comparison between the shadow running results and the actual running results includes: Extract the shadow control output sequence of the shadow operation result, and simultaneously extract the actual control output sequence of the actual operation result; Calculate the difference characteristics between the shadow control output sequence and the actual control output sequence; The verification result is generated based on the aforementioned difference features; The differential characteristics include output deviation index, number of constraint violations, anomaly trigger rate, and output volatility.

[0011] Optionally, the canary download of the control strategy to be implemented includes: The grayscale download process is executed sequentially, and each stage runs continuously to meet preset conditions and proceeds to the next stage without any abnormal triggering. During the grayscale download process, the adjustment increment of the control strategy to be implemented is written into the coal-fired power plant control system through the controlled interface of the coal-fired power plant control system. The adjustment increment is superimposed on the original control command of the coal-fired power plant control system; The adjustment increment includes at least one of the following: setpoint correction amount, control bias amount, or constraint boundary.

[0012] On the other hand, this application also provides a control strategy download device for a coal-fired power plant, which includes a data acquisition module, a data processing module, an online verification module, a grayscale download module, and an anomaly rollback module; The data acquisition module is used to collect multi-source production data from multiple heterogeneous production subsystems of a coal-fired power plant. The data processing module is connected to the data acquisition module and is used to perform heterogeneous adaptation and standardization processing on the multi-source production data to generate standardized data frames. The online verification module is connected to the data processing module and is used to set up a shadow running unit, load the control policy to be effective into the shadow running unit, so that the control policy to be effective and the currently effective control policy receive standardized data frames from the same source and run in parallel, generate shadow running results and actual running results, and compare the shadow running results and the actual running results to obtain verification results. The grayscale download module is connected to the online verification module and is used to perform grayscale download of the control strategy to be effective when the verification result meets the preset gating conditions, write the adjustment increment output by the control strategy to be effective into the coal-fired power plant control system and superimpose it onto the original control command. The abnormal rollback module is connected to the grayscale download module and is used to monitor the operating status of the coal-fired power plant in real time. When an abnormal situation is detected, the module performs a graded rollback operation on the coal-fired power plant control system according to a preset graded strategy.

[0013] Optionally, a storage module may also be included; The storage module is connected to the data processing module, the online verification module, and the grayscale download module, respectively. The storage module is used to store the standardized data frame, the control strategy to be implemented, the verification results, and the download record of the grayscale download module; The storage module is equipped with a version index unit, which is used to establish an association index between the control policy to be implemented, the verification result, and the download record.

[0014] In another aspect, this application provides a coal-fired power plant control strategy download system, which includes the aforementioned coal-fired power plant control strategy download device; The control strategy download system for the coal-fired power plant is deployed within the production control area of ​​the coal-fired power plant. The coal-fired power plant control strategy download system is connected to the coal-fired power plant control system through a controlled interface; A boundary isolation is set between the control strategy download system of the coal-fired power plant and the management information area of ​​the coal-fired power plant.

[0015] Optionally, the modules of the control strategy download system for the coal-fired power plant are isolated from each other; The controlled interface is an open, restricted write interface; The restricted write interface only allows the writing of the adjustment increment.

[0016] Compared with the prior art, this application has the following advantages: This application solves the problem of the difference between offline simulation and real operating environment by setting up a shadow running unit to run the control strategy to be effective in parallel; by performing gray-scale download, it avoids the sudden operation risk caused by full write; furthermore, by monitoring the operation status in real time and performing hierarchical rollback operation, it can automatically and quickly handle abnormal situations, improve the safety and reliability of the download of control strategy for coal-fired power plants, and reduce operation risks. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the method for downloading control strategies for coal-fired power plants in this application; Figure 2 This is a schematic diagram of the multi-source data heterogeneous adaptation and standardization processing flow in this application; Figure 3 This is a schematic diagram of the controlled write process in this application; Figure 4 This is a block diagram of the control strategy device for a coal-fired power plant in this application; Figure 5 This is a schematic diagram of the architecture of the secure download system for multi-source data control strategies in a coal-fired power plant, as described in a specific embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0021] It should be noted that, for ease of understanding, the method steps in the specific embodiments of this application are described in a certain order, but those skilled in the art can change the order of the steps according to actual needs, so this should not be used as a limiting condition; further, in the description of the following specific embodiments, the superscripts and subscripts of each parameter should be understood as distinguishing marks of similar identifiers in accordance with common interpretations unless otherwise specified, representing the parameters of the related or corresponding devices, and should not be understood as specific models or special marks.

[0022] like Figure 1 As shown, this application provides a method for downloading control strategies for coal-fired power plants, including the following steps: First, multi-source production data from multiple heterogeneous production subsystems of the coal-fired power plant is collected. During this process, data from different sources within the coal-fired power plant is acquired through an industrial protocol adaptation interface to obtain initial data. This initial data is then parsed and converted to obtain preprocessed data.

[0023] A synchronized time reference is established based on a globally unified clock; preprocessed data is calibrated using the synchronized time reference to obtain multi-source production data containing time information.

[0024] The multi-source production data includes at least process measurement point data and equipment status data, as well as environmental monitoring data, fuel test data, alarm data, operation log data, maintenance history data, and image or video inspection data.

[0025] Heterogeneous adaptation and standardization processing are performed on multi-source production data to generate standardized data frames. This process involves multi-source production data heterogeneous adaptation and standardization, such as... Figure 2As shown, the process includes protocol decoupling, time alignment, data cleaning, quality labeling, semantic mapping, and message distribution. Data cleaning identifies and handles missing values, out-of-bounds values, abrupt changes, and data with abnormal formats. Quality labeling adds a quality status identifier to each data entry. Semantic mapping maps measurement point names, equipment numbers, variable attributes, and engineering units from different systems to a unified data dictionary. Unified coding establishes unique object identifiers, unique variable identifiers, and version indexes. Message distribution sends standardized data to the corresponding modules according to subscription relationships.

[0026] Standardized data frames containing object identifiers, variable identifiers, timestamps, data values, engineering units, and quality status are generated from multi-source production data.

[0027] Standardized data frames adopt a unified structured format. The structure is as follows: ; In the formula, For object identification, For variable identification, For timestamps, For data values, For engineering units, This refers to the quality status.

[0028] Furthermore, the standardized data frames undergo secondary processing, including: In the standardized data frame, data with variables identified as continuously sampled variables are resampled in time and mapped to a global unified time axis. In a standardized data frame, the data marked as an update variable is subjected to the most recent valid value interpolation preservation until new valid data is received; In the standardized data frame, data identified as event-type variables are timestamped, and the state data at the time of event triggering is stored. In this embodiment, the updated variable specifically refers to low-frequency updated variables, i.e., non-real-time variables with a sampling period greater than 1 second.

[0029] Configure a shadow runtime unit, which is built based on digital twin technology and the current real runtime unit. Load the control policies to be implemented into the shadow runtime unit.

[0030] Both the control policy to be implemented and the currently effective control policy receive standardized data frames. Specifically, the control policy to be implemented and the currently effective control policy receive standardized data frames from the same source, and the two policies run in parallel to generate shadow execution results and actual execution results, such as... Figure 3 As shown.

[0031] The shadow operation results are reviewed by rule-based reasoning. Rule-based reasoning is used to review the consistency of prediction results, object status, interlock status, operation mode and adjustment increment based on a preset rule set.

[0032] Real-time prediction is used to generate trend prediction results for target variables or control correction strategies based on current data, historical window data, event marker data, and operating condition label data. Furthermore, real-time prediction is performed based on a real-time prediction unit, which receives an input feature set. and parameter set For the target variable Future prediction steps The predicted values ​​at each time point are calculated, and the relationship between the real-time prediction units is expressed as follows: ; In the formula, This includes current status variables, history window variables, event variables, and operating condition label variables. This represents the set of parameters corresponding to the current operating condition. This represents the prediction mapping relationship. When modeling and calculating, the prediction unit can use methods such as correlation analysis and mutual information analysis to select variables, and employ methods such as recursive identification and segmented identification to identify parameters.

[0033] Rule-based reasoning is implemented through rule-based reasoning units, which in turn are based on a set of conditions. and action set Establish explicit rule relationships , where the condition set This includes object state conditions, runtime boundary conditions, interlocking state conditions, manual authorization state conditions, and version state conditions; action sets. This includes approval, output limiting, output blocking, alarm triggering, policy freezing, and rollback execution. Only when the rule reasoning approval result is approved or the output limiting is approved will the shadow execution result proceed to the subsequent comparison stage.

[0034] Compare the results of the shadow run with the results of the actual run to obtain the verification results. The comparison steps include: Comparing the shadow operation results with the actual operation results includes extracting the shadow control output sequence from the shadow operation results and simultaneously extracting the actual control output sequence from the actual operation results; Calculate the difference characteristics between the shadow control output sequence and the actual control output sequence; Validation results are generated based on differential features; The differential characteristics include output deviation index, number of constraint violations, anomaly trigger rate, and output volatility.

[0035] The verification gate condition is expressed as ,in To comprehensively verify the indicators, one or more of the following can be combined: deviation indicators, constraint violation count, anomaly trigger rate, and output volatility. This is the preset allowed threshold.

[0036] If the verification results meet the preset gating conditions, the control policy to be implemented is gradually rolled out in a gray-scale manner. Gray-scale rollout is a security release mechanism that gradually expands the scope of application of the control policy by sequentially limiting the scope of effective objects, operating condition ranges, and output amplitude limits. It can also quickly roll back when anomalies are detected, avoiding systemic operational risks that may be caused by full rollout.

[0037] The canary deployment of the control strategy to be implemented includes sequentially executing each stage of the canary deployment, and continuously running in each stage while meeting preset conditions, and proceeding to the next stage without any abnormal triggering. Furthermore, the constraints of the canary deployment include at least load range limitations, variable whitelist limitations, maximum bias limits, and manual confirmation bypass limitations.

[0038] During the grayscale download process, the adjustment increment of the control strategy to be implemented is written into the coal-fired power plant control system through the controlled interface, thereby superimposing the adjustment increment with the original control command of the coal-fired power plant control system. The adjustment increment includes at least one of the following: setpoint correction, control bias, or constraint boundary.

[0039] The newly added module in this application does not directly replace the controller of the basic control system, nor does it directly acquire control of the actuator. It only outputs adjustment increments and acts on the basic control system of the coal-fired power plant when conditions are met through a controlled write interface.

[0040] In grayscale downloading, the operating status of the coal-fired power plant is monitored in real time. If an abnormal operating condition is detected, a tiered rollback operation is performed on the coal-fired power plant control system.

[0041] Performing tiered rollback operations on the control system of a coal-fired power plant includes: Level 1 rollback: Undoes the most recent adjustment increment written. Level 2 rollback: Freezes the newly added adjustment increments. Level 3 rollback: Switch to the control strategy that was most recently verified. Level 4 rollback resets the control strategy of the coal-fired power plant control system.

[0042] Furthermore, let the original output of the basic control system be... The newly added adjustment increment is Then the candidate output Represented as: ; In the formula, This represents the amplitude limiting function.

[0043] Let the maximum allowable change within a single sampling period be... Then the rate of change constraint satisfies When communication anomalies, sensor failures, data quality failures, prediction residual anomalies, interlocking triggers, output out-of-bounds errors, rate of change exceeding limits, version signature mismatches, or deployment failures are detected, a tiered rollback operation is triggered. Furthermore, the severity of anomalies in this application can be determined quantitatively, including but not limited to expert system scoring, preset quantitative standards, and weighted calculations.

[0044] On the other hand, such as Figure 4 As shown, this application also provides a control strategy download device for a coal-fired power plant, including a data acquisition module, a data processing module, an online verification module, a grayscale download module, and an anomaly rollback module.

[0045] The data acquisition module is used to collect multi-source production data from multiple heterogeneous production subsystems of a coal-fired power plant. It includes a protocol adaptation unit, an acquisition buffer unit, and a time synchronization unit. The protocol adaptation unit is used to adapt the interface to data from different sources. The acquisition buffer unit is used to cache and pre-sort real-time data. The time synchronization unit is used to establish a unified time reference for different data sources.

[0046] The data processing module is connected to the data acquisition module and is used to perform heterogeneous adaptation and standardization processing on multi-source production data to generate standardized data frames, including a data cleaning unit, a quality labeling unit, a semantic mapping unit, a unified encoding unit, and a message distribution unit.

[0047] The online verification module is connected to the data processing module and is used to set up a shadow running unit. The control strategy to be implemented is loaded into the shadow running unit, so that the control strategy to be implemented and the currently implemented control strategy receive standardized data frames from the same source and run in parallel, generating shadow running results and actual running results. Rule reasoning is performed on the shadow running results for review, and the shadow running results and actual running results are compared to obtain the verification results. The online verification module also includes a rule reasoning unit, which is used to perform consistency review on prediction results, object status, interlock status, running mode and adjustment increment according to a preset rule set.

[0048] The grayscale download module is connected to the online verification module. When the verification result meets the preset gating conditions, the control strategy to be effective is grayscale downloaded, and the adjustment increment output by the control strategy to be effective is written into the coal-fired power plant control system and superimposed on the original control command.

[0049] The abnormal rollback module is connected to the grayscale download module and is used to monitor the operating status of the coal-fired power plant in real time. When an abnormal condition is detected, it performs a graded rollback operation on the coal-fired power plant control system according to a preset graded strategy. It includes a communication monitoring unit, a data quality review unit, an output limiting unit, a rate of change constraint unit, an interlocking review unit, and a rollback control unit.

[0050] The control strategy download device for coal-fired power plants also includes a storage module. The storage module is connected to the data processing module, the online verification module, and the gray-scale download module. The storage module stores standardized data frames, control strategies to be implemented, verification results, and download records from the gray-scale download module. It includes a master data management unit, a metadata management unit, a real-time cache unit, a time-series data storage unit, a version index unit, and an audit traceability unit.

[0051] The device also includes a master data management unit, a metadata management unit, a real-time cache unit, a time-series data storage unit, a version index unit, and an audit traceability unit. The master data management unit maintains the relationships between measurement point objects, equipment objects, process section objects, control loop objects, operating condition label objects, and strategy version objects. The metadata management unit maintains variable types, engineering units, sampling frequencies, effective ranges, and quality rules. The real-time cache unit stores current operating data related to real-time control and online calculations. The time-series data storage unit stores historical process variable data, event data, prediction results, rule review results, control suggestion data, and actual written data; time-series data is organized according to object identifiers, variable identifiers, and time order, and queries are routed according to object range, time range, and version range. The version index unit establishes an index linking control strategies to be implemented, verification results, and download records. The audit traceability unit supports historical playback, strategy tracking, anomaly review, and accountability auditing, recording the anomaly trigger time, input data quality status, the current strategy version, the proposed adjustment increment, and the rollback execution result.

[0052] Furthermore, this application also provides a coal-fired power plant control strategy download system, including the coal-fired power plant control strategy download device as described above.

[0053] The control strategy download system for coal-fired power plants is deployed within the production control area of ​​the power plant. This system connects to the power plant's control system via a controlled interface. The control system includes a distributed control system (DCS) and a programmable logic controller (PLC). A lateral unidirectional isolation device provides boundary isolation between the control strategy download system and the power plant's management information area. When transmitting control commands, setpoint corrections, or important operational data, authentication, access control, and communication protection mechanisms are employed.

[0054] The modules of the control strategy system for coal-fired power plants are isolated from each other and deployed on isolated physical or virtual servers.

[0055] The aforementioned controlled interface in this application is an open, restricted write interface. Specifically, the restricted write interface only allows the writing of adjustment increments. The core of the controlled write mechanism lies in the fact that the newly added module in this application only outputs adjustment increments and does not directly bypass the basic control system to enter the actuator. All outputs intended to take effect must be verified through object whitelisting, variable boundaries, rate of change constraints, and interlocking conditions before they can be written to the basic control system interface.

[0056] This application applies to scenarios involving boiler combustion control, steam temperature control, steam pressure control, environmental protection island collaborative control, and multi-system coordinated optimization control. When the boiler, turbine, and environmental protection system operate collaboratively, the system, based on unified object semantics and a unified time reference, separately controls steam temperature, steam pressure, oxygen content, and NO. x A prediction and verification link is established for objects such as SO2 and slurry status, and then the adjustment increment is output to the basic control system through the controlled write interface, thereby realizing controlled autonomous optimization operation under all working conditions.

[0057] The following are specific embodiments of this application: Example 1; In this embodiment, the system is deployed within the production control area of ​​a coal-fired power plant. The system is as follows: Figure 5 As shown, this embodiment employs a multi-layered architecture, bypass computing, controlled writing, interlocking auditing, and tiered rollback deployment approach. The deployment objects in this embodiment include a distributed control system (DCS), a programmable logic controller (PLC), an environmental online monitoring system, an equipment status monitoring system, a laboratory data source, an inspection terminal, and an operation log system.

[0058] The system in this embodiment physically includes at least one data access server, one real-time computing server, one main data and time-series storage server, and one auditing and management server. In engineering implementation, these servers can be deployed independently or virtualized and integrated according to resource conditions, but at least the isolation between data access, real-time computing, and storage services must be ensured. The deployment steps include: Establish a data access list and determine the scope of data objects to be included in the system based on the unit's control requirements. This should include at least the boiler main steam temperature, reheat steam temperature, main steam pressure, feedwater flow rate, desuperheating water flow rate, total air volume, total coal volume, furnace negative pressure, furnace oxygen content, and NO₂ at the denitrification inlet. x Export NO x Ammonia injection flow rate, SO2 at desulfurization inlet, SO2 at desulfurization outlet, slurry pH value, slurry density, circulation pump status, equipment status parameters, alarm parameters, interlock status parameters, and mode switching parameters.

[0059] Establish an interface mapping table to correspond one-to-one with the original tag number in the field data source, forming an interface mapping table. The interface mapping table should at least include the original tag number, data source system name, object identifier, variable identifier, engineering unit, refresh cycle, allowable range, quality judgment rules, whether it is allowed to participate in control calculation, and whether it is allowed to participate in write verification.

[0060] It can access real-time data from multiple sources, periodically acquiring data from DCS, PLC and other systems through industrial protocol gateways or standard interfaces. High-frequency process variables are accessed using a fixed sampling period polling or subscription method, while low-frequency data is accessed using event triggering or periodic capture methods.

[0061] A unified time base is established, converting data from different sources into a standard timestamp format within the system. If there are millisecond or second-level deviations between different data sources, the unified clock within the production control area is used as the time reference, and time alignment is performed in the acquisition buffer unit.

[0062] After completing internal service connectivity, protocol-adapted data is sent to the real-time data platform layer. The platform performs data cleaning, quality labeling, semantic mapping, and unified encoding, then pushes the data to the master data and time-series storage layer, as well as the online verification module. The master data and time-series storage layer organizes the production system in an object-oriented manner, allowing the same equipment object to be associated with multiple measurement point objects, status objects, and control objects, and the same process segment object to be associated with multiple equipment objects and control loop objects. A bypass write link is established; the system does not directly connect to the actuators but writes adjustment increments through the controlled interface opened by the basic control system. Initially, this write link is configured to only calculate and not take effect.

[0063] In this embodiment, each piece of data entering the system is converted into a standard data frame. Continuously sampled variables are resampled according to a unified time axis. Low-frequency variables retain the most recent valid value; event quantities are organized using event time plus state value. For data that fails quality verification, its quality status is... Marked as unavailable, access is only permitted to the historical tracing link, not to the real-time control suggestion generation link. After deployment in this embodiment, the system will at least form a unified access and object-oriented representation of multi-source production data, an integrated data organization associated with controlled objects, process sections, and strategy versions, a data foundation available for online verification and controlled write calls, a bypass access capability that does not change the execution dominance of the basic control system, and a hierarchical task scheduling capability that prioritizes real-time control, rule review, and online tasks over offline tasks.

[0064] Example 2; This embodiment takes the update of the main steam temperature control strategy as an example to illustrate how the system performs shadow verification and generates adjustment increments based on unified data organization.

[0065] In this embodiment, the control strategy to be implemented is the optimized main steam temperature control strategy, while the currently implemented control strategy is the original main steam temperature control strategy of the basic control system. Both the control strategy to be implemented and the currently implemented control strategy receive standardized data frames from the same source. The input data includes at least the current main steam temperature, the historical window sequence of the main steam temperature, the current load, the desuperheating water flow rate, the feedwater flow rate, the total coal quantity, the total air volume, the furnace oxygen quantity, the current operating condition label, and the critical event marker.

[0066] The shadow execution steps include: Load the control policies that are yet to take effect into the shadow execution unit; Copy the same source input data to two control strategies; generate shadow execution results and actual execution results in parallel; The shadow operation results are reviewed by rule-based reasoning. The review includes whether all input data is of usable quality, whether the current object is in automatic mode, whether the current object is in interlock release mode, whether the current object is in the whitelist, whether the current adjustment increment exceeds the allowed range, and whether the current unit is in a state where automatic correction is prohibited. The approved shadow control output sequence and actual control output sequence are extracted. Calculate the output deviation index, constraint violation count, anomaly trigger rate, and output volatility between the two; generate a comprehensive verification index based on the above differences. When satisfied When the control strategy to be implemented is deemed to meet the conditions for entering the gray-scale download stage, then... This is the preset gate threshold.

[0067] The adjustment increment of the control strategy output to be effective is the setpoint correction amount or control bias amount, which is used to be superimposed on the original control command of the basic control system.

[0068] Example 3; This embodiment illustrates the entire process of a control strategy to be implemented, from successful verification to its gradual implementation in the field. The control strategy to be implemented must include at least the strategy version number, applicable scope, parameter set, activation conditions, exit conditions, rollback version, maximum allowed offset, and allowed effective operating condition range.

[0069] The grayscale download process must simultaneously meet the following mandatory constraints: load range limitation, variable whitelist limitation, maximum bias limitation, and manual confirmation bypass limitation.

[0070] Grayscale downloads should be implemented using a gradual rollout approach, including at least the following stages: The first phase is object-limited, and the new strategy is enabled only for one target object. For example, it can be enabled only for the main steam temperature offset. The second stage of operating condition limitation means that it can only be activated under a certain preset load range or a certain preset operating mode, such as only being activated within the steady-state load range. The third stage of amplitude limitation restricts the maximum bias of the new strategy output from not exceeding a preset limit. In the fourth stage, manual release is required. Only after the operators confirm that the unit status meets the trial operation conditions can the grayscale write take effect. The fifth stage involves expanding the scale, gradually increasing the applicable operating conditions and allowable offset range when the grayscale stage is running normally without any abnormal triggers.

[0071] If any of the following situations occur during the grayscale phase, the grayscale state will be immediately exited and the rollback process will begin: constraint violation count exceeds the threshold, interlocking is triggered, output volatility continues to increase, critical variable deviation worsens, communication is abnormal, or manual prohibition by operators. The key to this embodiment is that the control strategy to be implemented must first complete shadow verification under the same input source, and then undergo grayscale scaling with limited objects, limited operating conditions, and limited amplitudes; it cannot directly replace the current operating strategy in one go.

[0072] Example 4; This embodiment uses the writing of the main steam temperature control bias to the DCS as an example to illustrate the specific process of controlled writing and abnormal interlocking rollback. Candidate output generation: Assume the original output of the basic control system is... The newly added adjustment increment is The candidate output is ,in It is a limiting function, whose upper and lower limits are given by the allowable operating range of the target controlled object. The allowable operating range is preferably taken from the object boundary that has been put into operation by the basic control system or the operating boundary that has been debugged and confirmed.

[0073] The pre-writing verification steps include: Object whitelist verification: Check whether the target object belongs to the list of controlled objects that are allowed to be automatically corrected. If it does not belong, writing is prohibited in this cycle. Method verification checks whether the target object is in automatic mode and not manually locked. If it is in manual mode or manually disabled, writing is prohibited in this cycle. Boundary check, check candidate outputs If the value falls between the upper and lower limits of the allowable range, and exceeds the limits, then the range will be limited first before proceeding to the next step. Change rate check: Assume the maximum allowable change within a single sampling period is... Then it requires If the conditions are not met, the rate of change of the candidate output for this period will be truncated, or it will be directly determined that the download is prohibited for this period. Interlock verification checks the main protection, auxiliary protection, process interlocks, equipment failures, communication anomalies, and manual prohibition signs. Writing to the basic control system interface is only allowed when all release conditions are met. The write and confirmation process records the write value, write time, object identifier, policy version number, and audit result together in the audit traceability unit.

[0074] Abnormal interlocking and graded rollback: The rollback control unit is triggered when any of the following abnormalities occur: data acquisition link interruption, failure of quality bits of key input measurement points, prediction residuals continuously exceeding the threshold, output change rate continuously exceeding the limit, write interface confirmation failure, interlocking state triggering, and inconsistent signature of the current policy version.

[0075] After an exception is triggered, rollback will proceed in the following order: The first level cancels the newly added bias value, stops the system in this embodiment from writing new adjustment increments to the basic control system, and only retains the data acquisition and verification functions; The second-level freezing strategy is switched, prohibiting the new strategy from entering the gray-scale stage or the effective stage. The third level switches to the most recently verified stable version. If a verified stable version exists, the configuration of that version is restored. The fourth level resets the control strategy of the coal-fired power plant control system, reverts to the original control logic of the basic control system of the coal-fired power plant, completely exits the new correction function, and only retains the independent operation of the basic control system. The fifth level records abnormal audit information, including the time of the abnormality, input quality status, candidate outputs, interlock status, executed rollback level, and the current policy version, for subsequent review. The audit traceability unit also supports historical playback, policy tracking, and responsibility audit functions.

[0076] In this embodiment, the essence of controlled writing is not to send in the adjustment increment, but to allow limited writing only after the candidate output is constrained and reviewed layer by layer; the essence of rollback logic is not to simply disable, but to exit the new function according to the risk level.

[0077] Example 5; This embodiment takes the implementation of a coordinated optimization control strategy for the denitrification system and the desulfurization system as an example to illustrate the specific reuse method of the system in the environmental protection island scenario.

[0078] The objects to be controlled collaboratively include at least the denitrification ammonia injection control objects, the desulfurization slurry supply control objects, the absorber circulating pump combination objects, and the dust removal and ash conveying auxiliary operation objects. The data included in the collaborative control should at least include the current unit load and the denitrification inlet NO₂. x Export NO xAmmonia escape related state quantities, SO2 at the desulfurization inlet, SO2 at the outlet, pH of the absorber slurry, slurry density, liquid level, operating status of the circulating pump, flue gas flow rate and temperature, equipment failure and interlock status, and current environmental protection target boundaries.

[0079] The steps for downloading the collaborative control strategy include: Load the environmental island collaborative control strategy that is yet to take effect into the shadow operation unit; The collaborative control strategy to be implemented and the control strategies of each subsystem currently in effect receive standardized data frames from the same source and run in parallel. Perform collaborative rule reasoning review on the shadow operation results to check for conflicts between local adjustment increments; compare the shadow operation results with the actual operation results to generate comprehensive verification indicators. After the verification results meet the preset gate conditions, grayscale download is performed, and the incremental adjustment of denitrification ammonia injection, desulfurization slurry supply, and circulation pump combination is written into the basic control system in sequence. During the grayscale download process, the operating status is monitored in real time, and a tiered rollback operation is performed when an anomaly occurs. Before entering the basic control system, each adjustment increment of the environmental protection island still needs to pass the object whitelist, output limit, rate of change constraint, and interlock review separately. It is not allowed to bypass the single object safety verification under the name of combined adjustment increment.

[0080] Finally, it should be noted that the above content is only used to illustrate the technical solution of this application, and is not intended to limit the scope of protection of this application. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of this application shall not depart from the substance and scope of the technical solution of this application.

Claims

1. A method for downloading control strategies for a coal-fired power plant, characterized in that, include: Collect multi-source production data from multiple heterogeneous production subsystems of a coal-fired power plant; The multi-source production data is subjected to heterogeneous adaptation and standardization processing to generate standardized data frames; Set up a shadow execution unit and load the control strategy to be implemented into the shadow execution unit; Based on the standardized data frame, the control strategy to be implemented and the currently implemented control strategy run in parallel, and generate shadow running results and actual running results; The verification results are obtained by comparing the shadow execution results with the actual execution results; If the verification results meet the preset gating conditions, the control strategy to be implemented will be downloaded in grayscale. In grayscale download, the operating status of coal-fired power plants is monitored in real time; If an abnormal condition is detected in the operating state, a graded rollback operation is performed on the coal-fired power plant control system.

2. The method for downloading control strategies for coal-fired power plants according to claim 1, characterized in that, The collection of multi-source production data includes: Data from different sources in the coal-fired power plant is collected through an industrial protocol adapter interface to obtain initial collected data; The initial collected data is parsed and converted to obtain preprocessed data; Establish a synchronized time reference based on a globally unified clock; The preprocessed data is calibrated using the synchronized time reference, and the multi-source production data is obtained.

3. The method for downloading control strategies for coal-fired power plants according to claim 2, characterized in that, Heterogeneous adaptation and standardization processing of the multi-source production data includes: Generate a standardized data frame containing object identifier, variable identifier, timestamp, data value, engineering units, and quality status; In the standardized data frame, the data identified as continuously sampled variables are resampled in a time-synchronized manner and mapped to a globally unified time axis. In the standardized data frame, the data identified as the updated variable is subjected to the most recent valid value interpolation and preservation until new valid data is received; In the standardized data frame, the data identified as event-type variables are marked with an event timestamp, and the state data at the time of event triggering is stored.

4. The method for downloading control strategies for coal-fired power plants according to claim 1, characterized in that, The control strategy to be implemented and the currently implemented control strategy receive standardized data frames from the same source. The comparison between the shadow running results and the actual running results includes: Extract the shadow control output sequence of the shadow operation result, and simultaneously extract the actual control output sequence of the actual operation result; Calculate the difference characteristics between the shadow control output sequence and the actual control output sequence; The verification result is generated based on the aforementioned difference features; The differential characteristics include output deviation index, number of constraint violations, anomaly trigger rate, and output volatility.

5. The method for downloading control strategies for coal-fired power plants according to claim 1, characterized in that, The gray-scale download of the control strategy to be implemented includes: The grayscale download process is executed sequentially, and each stage runs continuously to meet preset conditions and proceeds to the next stage without any abnormal triggering. During the grayscale download process, the adjustment increment of the control strategy to be implemented is written into the coal-fired power plant control system through the controlled interface of the coal-fired power plant control system. The adjustment increment is superimposed on the original control command of the coal-fired power plant control system; The adjustment increment includes at least one of the following: setpoint correction amount, control bias amount, or constraint boundary.

6. The method for downloading control strategies for coal-fired power plants according to claim 1, characterized in that, Performing a graded rollback operation on the control system of the coal-fired power plant includes: Level 1 rollback: Undoes the most recent adjustment increment written. Level 2 rollback: Freezes the newly added adjustment increments. Level 3 rollback: Switch to the control strategy that was most recently verified. Level 4 rollback resets the control strategy of the coal-fired power plant control system.

7. A control strategy download device for a coal-fired power plant, characterized in that, It includes a data acquisition module, a data processing module, an online verification module, a grayscale download module, and an exception rollback module; The data acquisition module is used to collect multi-source production data from multiple heterogeneous production subsystems of a coal-fired power plant. The data processing module is connected to the data acquisition module and is used to perform heterogeneous adaptation and standardization processing on the multi-source production data to generate standardized data frames. The online verification module is connected to the data processing module and is used to set up a shadow running unit, load the control policy to be effective into the shadow running unit, so that the control policy to be effective and the currently effective control policy receive standardized data frames from the same source and run in parallel, generate shadow running results and actual running results, and compare the shadow running results and the actual running results to obtain verification results. The grayscale download module is connected to the online verification module and is used to perform grayscale download of the control strategy to be effective when the verification result meets the preset gating conditions, write the adjustment increment output by the control strategy to be effective into the coal-fired power plant control system and superimpose it onto the original control command. The abnormal rollback module is connected to the grayscale download module and is used to monitor the operating status of the coal-fired power plant in real time. When an abnormal situation is detected, the module performs a graded rollback operation on the coal-fired power plant control system according to a preset graded strategy.

8. The control strategy download device for coal-fired power plants according to claim 7, characterized in that, It also includes a storage module; The storage module is connected to the data processing module, the online verification module, and the grayscale download module, respectively. The storage module is used to store the standardized data frame, the control strategy to be implemented, the verification results, and the download record of the grayscale download module; The storage module is equipped with a version index unit, which is used to establish an association index between the control policy to be implemented, the verification result, and the download record.

9. A control strategy download system for a coal-fired power plant, characterized in that, Includes the control strategy installation device for coal-fired power plants as described in any one of claims 7-8; The control strategy download system for the coal-fired power plant is deployed within the production control area of ​​the coal-fired power plant. The coal-fired power plant control strategy download system is connected to the coal-fired power plant control system through a controlled interface; A boundary isolation is set between the control strategy download system of the coal-fired power plant and the management information area of ​​the coal-fired power plant.

10. The coal-fired power plant control strategy download system according to claim 9, characterized in that, The modules of the control strategy download system for the coal-fired power plant are isolated from each other. The controlled interface is an open, restricted write interface; The restricted write interface only allows the writing of the adjustment increment.

Images