An autonomous response method and system for a thermal power plant system architecture

Abstract

The application relates to the technical field of industrial control, and particularly discloses an autonomous response method and system for a thermal power plant system architecture, which comprises the following steps: converting a load response signal and a device risk signal of a thermal power plant into standardized data flow; multi-dimensionally monitoring a trigger stability mark and a deviation record in the standardized data flow, and combining a physical process characteristic quantity in the standardized data flow into a unified situation representation vector; inputting the unified situation representation vector into an online joint optimization rule library, rolling optimization is performed on an initial deviation sequence in the unified situation representation vector, and an optimal control target sequence is obtained; each target value in the optimal control target sequence is decomposed into a fuel regulating valve opening degree instruction, a steam turbine regulating valve opening degree instruction and a desuperheating water valve opening degree instruction, and is sent to a controller and a field bus device in the thermal power plant; and the application significantly improves the autonomous response accuracy, operation stability and economy of the thermal power unit under the deep peak regulation and flexible regulation scene.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to an autonomous response method and system for a thermal power plant system architecture. Background Technology

[0002] As the baseload and regulating power source of the power system, the autonomous response capability of thermal power generation control systems directly affects the safe and stable operation of the power grid. Currently, thermal power plants generally employ distributed control systems to achieve unit-level control, responding to grid dispatch commands through automatic generation control while relying on systems such as vibration monitoring and temperature inspection to collect equipment risk signals. However, the data formats, timestamps, and semantics of different signal sources are independent, making it difficult to form a standardized fused data stream. Existing technologies often use single thresholds to independently judge multi-dimensional triggering conditions such as changes in grid commands, deviations in operating parameters, and equipment trend evolution, without introducing hysteresis comparisons and deviation timing records. This easily leads to trigger oscillations or missed judgments, resulting in unreliable autonomous response initiation.

[0003] In existing technologies, the optimization decision-making and control command generation of thermal power units typically rely on static safety constraint tables and single technical objectives, lacking a systematic consideration of economic costs. There is a lack of joint invocation and coordinated tailoring mechanisms among safety constraints, regulation capacity, and economy, making it difficult to dynamically balance multiple objectives such as fuel cost, emission losses, and load tracking deviations while ensuring equipment safety. Therefore, there is an urgent need to develop an autonomous response method and system for thermal power plant system architecture to solve the technical problems of unified representation of multi-source signals, dynamic joint constraint tailoring, multi-objective rolling optimization of economy, and smooth command coordination, thereby improving the autonomous response accuracy, economy, and operational stability of thermal power units under complex operating conditions. Summary of the Invention

[0004] This invention provides an autonomous response method and system for thermal power plant system architecture to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an autonomous response method for a thermal power plant system architecture, comprising:

[0006] G1: Converts the load response signals and equipment risk signals of thermal power plants into standardized data streams;

[0007] G2: Multi-dimensional monitoring of the trigger stability flags and deviation records in the standardized data stream, and merging them with the physical process characteristics in the standardized data stream into a unified situational representation vector for the thermal power plant;

[0008] G3: Input the unified situation representation vector into the preset online joint optimization rule base, and perform rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence;

[0009] G4: Decompose each target value in the optimal control target sequence into fuel regulating valve opening command, turbine regulating valve opening command, and desuperheating water valve opening command, and send them to the controller and fieldbus equipment in the thermal power plant.

[0010] In a preferred embodiment, the conversion of the load response signal and equipment risk signal of the thermal power plant into a standardized data stream includes:

[0011] The load response signal and equipment risk signal of the thermal power plant are acquired. The load response signal includes the automatic generation control command value, the actual power generation value of the unit and the frequency regulation action flag. The equipment risk signal includes the bearing vibration amplitude, the bearing temperature change rate and the furnace pressure fluctuation amplitude.

[0012] Based on the deviation between the automatic generation control command value and the actual power generation value of the unit, a power deviation signal is determined, and the power deviation signal is aligned with the timestamp of the frequency regulation action flag bit and encapsulated into a load response data frame.

[0013] The vibration amplitude of the bearing is extracted by extracting the envelope to obtain the vibration trend feature value. The vibration trend feature value, the historical difference result of the bearing temperature change rate, and the furnace pressure fluctuation amplitude are combined into an equipment risk data frame.

[0014] The load response data frame and the equipment risk data frame are spliced ​​together using a timestamp reference and then subjected to rule verification to obtain a standardized data stream.

[0015] In a preferred embodiment, the multi-dimensional monitoring of the trigger stability flags and deviation records in the standardized data stream includes:

[0016] The grid dispatch command value, the measured value of key operating parameters of the generating unit, and the sequence of equipment status characteristics are parsed from the standardized data stream.

[0017] Based on the power command change amount of the automatic generation control target power in the power grid dispatch command value, determine the trigger valid flag and the trigger reset flag, and integrate the trigger valid flag and the trigger reset flag into a hysteresis status flag;

[0018] The instantaneous deviation of the measured value of the main steam pressure in the key operating parameters of the unit is read in real time. Based on the instantaneous deviation, a condition deviation flag is generated, and all instantaneous deviations are stored in chronological order as a pressure deviation record sequence.

[0019] The first-order difference operation is performed on the time series of the bearing vibration amplitude to obtain the vibration change rate sequence. Based on the sign judgment result of the vibration change rate sequence, the trend warning sign is confirmed. At the same time, the temperature change sign is confirmed according to the historical difference of the bearing temperature change rate.

[0020] The hysteresis status flag, the pressure deviation record sequence, the trend warning flag, and the temperature sudden change flag are timestamped and packaged to obtain the trigger stability flag and deviation record.

[0021] In a preferred embodiment, the merging of the physical process features in the standardized data stream into a unified situational representation vector for the thermal power plant includes:

[0022] Extract combustion-side features, soft drink-side features, and electrical-side features from the standardized data stream;

[0023] The instantaneous value of fuel quantity and the furnace pressure fluctuation amplitude in the combustion side characteristic quantities are summed by a first weight to obtain the combustion side situation component. The measured value of main steam pressure and the rate of change of main steam temperature in the steam-water side characteristic quantities are summed by a second weight to obtain the steam-water side situation component.

[0024] The absolute value of the difference between the actual generated power of the unit and the target power of the automatic generation control in the electrical side characteristic quantities is calculated to obtain the electrical side deviation component;

[0025] The pressure deviation recording sequence is weighted according to a preset time decay factor and then appended to the soda-side situation component. The trend warning flag is multiplied into the combustion-side situation component as a Boolean coefficient.

[0026] The combustion-side situational component after multiplication, the added steam-water-side situational component, the electrical-side deviation component, and the hysteresis state flag are integrated into a unified situational representation vector.

[0027] In a preferred embodiment, the preset online joint optimization rule base includes:

[0028] The online joint optimization rule base includes a security constraint matrix, an economic cost vector, and a flexibility adjustment capability mapping table;

[0029] Based on the load and pressure range where the thermal power plant is currently operating, the corresponding maximum load increase rate and minimum stable combustion load boundary value are read from the flexibility adjustment capability mapping table.

[0030] The power change rate limit and minimum load limit in the corresponding operating condition section of the safety constraint matrix are dynamically trimmed using the minimum stable combustion load boundary value.

[0031] The trimmed power change rate limit, minimum load limit, and untrimmed static safety limit are packaged together to obtain the joint safety boundary.

[0032] The joint security boundary and the economic cost vector are simultaneously input into the rolling optimization process to output the optimal control target sequence of the thermal power plant.

[0033] In a preferred embodiment, the step of inputting the unified situation representation vector into a preset online joint optimization rule base and performing rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence includes:

[0034] The unified situational representation vector is input into the online joint optimization rule base, and the target deviation value of the measured power value in multiple control cycles is read backward from the current time as the starting point.

[0035] The target deviation value and the pressure deviation record sequence in the unified situation characterization vector are aligned by time and merged into a joint deviation sequence.

[0036] Using the preset prediction time domain length as the window length and the joint deviation sequence as the initial condition, the rolling optimization process is invoked to generate a candidate control trajectory group;

[0037] The trajectory that minimizes the total economic cost is selected from the candidate control trajectory group, and the target values ​​of output power, main steam pressure and furnace outlet nitrogen oxide concentration in the trajectory are arranged in chronological order to obtain the optimal control target sequence.

[0038] In a preferred embodiment, the rolling optimization includes:

[0039] The last deviation value in the joint deviation sequence is used as the starting deviation;

[0040] The power adjustment step, pressure adjustment step, and nitrogen oxide adjustment step in the preset control step array are added to the actual values ​​of the current cycle to generate candidate values ​​for output power, main steam pressure, and furnace outlet nitrogen oxide concentration for the next control cycle. This operation is repeated until the entire prediction time domain is filled to obtain the candidate control trajectory.

[0041] Verify whether the candidate value of each control cycle in the candidate control trajectory satisfies all the constraints in the joint safety boundary;

[0042] The total economic cost of the candidate control trajectories that pass the verification is calculated, and the control trajectory with the minimum total economic cost is selected as the candidate control trajectory group.

[0043] In a preferred embodiment, the formula for calculating the total economic cost is:

[0044]

[0045] in, To predict the total number of control cycles in the time domain, For the first Candidate output power values ​​for each cycle For the first The power grid dispatch command value for each cycle. The weight of the power deviation penalty term. For the first Candidate values ​​for the main steam pressure in each cycle. Main steam pressure setpoint The weight of the pressure fluctuation penalty term, For the first Candidate values ​​for nitrogen oxide concentration in each period, For emission-adjusted cost weighting, For the first Fuel consumption per cycle Fuel cost weighting.

[0046] In a preferred embodiment, the step of decomposing each target value in the optimal control target sequence into fuel regulating valve opening commands, turbine regulating valve opening commands, and desuperheating water valve opening commands, and issuing them to the controller or fieldbus equipment in the power plant, includes:

[0047] The output power target value of the current control cycle in the optimal control target sequence is subjected to first-order inertial filtering to obtain the turbine control valve opening command;

[0048] The deviation between the current main steam pressure target value and the measured value is input together with the output power target value to the fuel-pressure coordination link of the thermal power plant, outputting the fuel quantity demand value, and converting the fuel quantity demand value into a fuel regulating valve opening command;

[0049] The deviation between the target value of nitrogen oxide concentration at the furnace outlet and the current measured nitrogen oxide concentration is input into the proportional-integral calculation unit of the thermal power plant to obtain the desuperheating water flow correction value, and the desuperheating water valve opening command is generated based on the desuperheating water flow correction value.

[0050] The steam turbine control valve opening command, the fuel regulating valve opening command, and the desuperheating water valve opening command are respectively sent to the corresponding controllers and fieldbus devices in the thermal power plant.

[0051] To address the aforementioned problems, the present invention also provides an autonomous response system for a thermal power plant system architecture, the system comprising:

[0052] The standardized data conversion module is used to convert the load response signals and equipment risk signals of thermal power plants into standardized data streams;

[0053] The multi-dimensional situation fusion module is used to monitor the trigger stability flags and deviation records in the standardized data stream from multiple dimensions, and merge them with the physical process feature quantities in the standardized data stream into a unified situation representation vector of the thermal power plant.

[0054] The rolling optimization module is used to input the unified situation representation vector into a preset online joint optimization rule base, and to perform rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence.

[0055] The instruction issuing module is used to decompose each target value in the optimal control target sequence into fuel regulating valve opening instructions, turbine regulating valve opening instructions, and desuperheating water valve opening instructions, and issue them to the controllers and fieldbus devices in the thermal power plant.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. This invention solves the inconsistency issues in timestamps, semantic labels, and formats of multi-source heterogeneous data by uniformly converting load response signals and equipment risk signals from thermal power plants into a standardized data stream, providing a reliable data foundation for subsequent decision-making. Based on this, a multi-dimensional parallel monitoring mechanism is adopted to simultaneously track changes in grid dispatch instructions, deviations in unit operating parameters, and equipment status trends. A hysteresis comparison method is introduced to generate a trigger stability flag, effectively avoiding frequent oscillations in the trigger signal. Simultaneously, pressure deviation sequences, sign reversal of vibration change rate, and historical differential results of temperature change rate are recorded, forming deviation records and trend warning flags with a time structure. Furthermore, the physical process characteristics of the combustion side, steam-water side, and electrical side are weighted and summed, and the pressure deviation sequence, trend warning flag, and hysteresis status flag are merged into a unified situational representation vector. This enables the system to comprehensively perceive the grid regulation requirements, internal operating condition deviations, and equipment health margins under current operating conditions, significantly improving the reliability of autonomous response initiation and the accuracy of situational representation.

[0058] 2. This invention constructs an online joint optimization rule base comprising a safety constraint matrix, an economic cost vector, and a flexibility adjustment capability mapping table. Based on the current load and pressure ranges, dynamic capability limits are read from the mapping table to prune the safety constraint matrix, generating a joint safety boundary. This achieves coordinated constraints on equipment physical safety and flexible adjustment capability. Furthermore, a rolling optimization method is employed. Measured deviation and pressure deviation records from multiple forward control cycles are aligned and merged into a joint deviation sequence. Using this joint deviation sequence as initial conditions, candidate control trajectories are generated through depth-first enumeration. Each cycle is checked to ensure compliance with all constraints within the joint safety boundary. A quantitative evaluation is performed according to a multi-objective economic cost formula encompassing power deviation, pressure fluctuation, emission costs, and fuel consumption. The globally optimal candidate control trajectory is selected, ensuring the optimality of the control objective sequence. In the command issuance stage, a first-order inertial filter is applied to the output power target value to obtain a smooth control command. The pressure deviation and power target are jointly mapped to fuel demand through the fuel-pressure coordination stage. Then, the desuperheating water flow rate is corrected according to the nitrogen oxide concentration deviation through the proportional-integral calculation stage. This effectively avoids sudden changes and oscillations in the actuator and improves the operational stability, economy and control accuracy of thermal power units in deep peak shaving and flexible adjustment scenarios. Attached Figure Description

[0059] Figure 1 A flowchart illustrating an autonomous response method for a thermal power plant system architecture according to an embodiment of the present invention;

[0060] Figure 2 A functional module diagram of an autonomous response system for a thermal power plant system architecture provided in an embodiment of the present invention;

[0061] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0062] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0063] This application provides an autonomous response method for a thermal power plant system architecture. The execution entity of this autonomous response method for a thermal power plant system architecture includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application embodiment: a server, a terminal, etc. In other words, the autonomous response method for a thermal power plant system architecture can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0064] Reference Figure 1 The diagram shown is a flowchart illustrating an autonomous response method for a thermal power plant system architecture according to an embodiment of the present invention. In this embodiment, the autonomous response method for a thermal power plant system architecture includes:

[0065] G1: Converts the load response signals and equipment risk signals of thermal power plants into standardized data streams;

[0066] In this embodiment of the invention, the conversion of the load response signal and equipment risk signal of a thermal power plant into a standardized data stream includes:

[0067] The load response signal and equipment risk signal of the thermal power plant are acquired. The load response signal includes the automatic generation control command value, the actual power generation value of the unit and the frequency regulation action flag. The equipment risk signal includes the bearing vibration amplitude, the bearing temperature change rate and the furnace pressure fluctuation amplitude.

[0068] Based on the deviation between the automatic generation control command value and the actual power generation value of the unit, a power deviation signal is determined, and the power deviation signal is aligned with the timestamp of the frequency regulation action flag bit and encapsulated into a load response data frame.

[0069] The vibration amplitude of the bearing is extracted by extracting the envelope to obtain the vibration trend feature value. The vibration trend feature value, the historical difference result of the bearing temperature change rate, and the furnace pressure fluctuation amplitude are combined into an equipment risk data frame.

[0070] The load response data frame and the equipment risk data frame are spliced ​​together using a timestamp reference and then subjected to rule verification to obtain a standardized data stream.

[0071] The system reads the automatic power generation control command value from the real-time database of the distributed control system. This command value is the target power value issued by the grid dispatch to the unit. At the same time, it reads the actual power value generated by the unit, i.e., the actual power value generated by the unit. Then, it reads the frequency regulation action flag bit from the primary frequency regulation action flag bit register. The value of the flag bit is true or false, indicating whether the unit is in the primary frequency regulation action state. Next, it reads the bearing vibration amplitude from the vibration channel of the vibration monitoring system, reads the bearing temperature change rate from the temperature change rate calculation unit of the temperature inspection system, and reads the furnace pressure fluctuation amplitude from the pressure fluctuation recording unit of the furnace pressure monitoring system.

[0072] The power deviation signal is obtained by subtracting the actual power output of the unit from the automatic generation control command value. The power deviation signal and the frequency regulation action flag are aligned with the same timestamp to ensure that the two data have the same time mark. The aligned power deviation signal and the frequency regulation action flag are then encapsulated into a complete load response data frame in a preset frame header, data segment, and frame tail order. This data frame contains the power deviation value, the frequency regulation action flag value, and the corresponding timestamp.

[0073] An envelope extraction operation is performed on the time series of bearing vibration amplitude. Specifically, the peak points within each vibration cycle are taken and connected by a smooth curve to obtain a curve reflecting the trend of vibration intensity change, i.e., the vibration trend characteristic value. Then, a historical difference operation is performed on the bearing temperature change rate series, which subtracts the temperature change rate from the previous time point to obtain the historical difference result. This result reflects the acceleration or deceleration trend of the temperature change rate. Finally, the vibration trend characteristic value, the historical difference result of the bearing temperature change rate, and the furnace pressure fluctuation amplitude are merged according to the same timestamp and encapsulated into an equipment risk data frame.

[0074] The load response data frame and the equipment risk data frame are aligned according to their respective timestamps to ensure a one-to-one correspondence between the two types of data frames at the same time. The aligned data frames are then merged from front to back into a single, complete data block, containing both load response and equipment risk information. Rule-based validation is then performed on the merged data block, checking that each value is within a preset reasonable range, that the timestamps are continuous and uninterrupted, and that the data frame format is complete and without missing elements. Data blocks that pass all validations are defined as standardized data streams.

[0075] The beneficial effects are as follows: By uniformly acquiring and processing load response signals and equipment risk signals, the deviation between the automatic power generation control command value and the actual power generation value of the unit is calculated to obtain the power deviation signal, which is then encapsulated into a load response data frame by aligning it with the timestamp of the frequency regulation action flag bit. At the same time, the envelope is extracted from the bearing vibration amplitude to obtain the vibration trend characteristic value. Combined with the historical differential results of the bearing temperature change rate and the furnace pressure fluctuation amplitude, an equipment risk data frame is formed. Then, the timestamps of the two types of data frames are spliced ​​and regular verification is performed to generate a standardized data stream. This completely solves the problems of inconsistent formats, misaligned timestamps, and inconsistent semantics of multi-source heterogeneous data, providing a reliable and unified data foundation for subsequent autonomous response and significantly improving the accuracy and real-time performance of data fusion.

[0076] G2: Multi-dimensional monitoring of the trigger stability flags and deviation records in the standardized data stream, and merging them with the physical process characteristics in the standardized data stream into a unified situational representation vector for the thermal power plant;

[0077] In this embodiment of the invention, the multi-dimensional monitoring of the trigger stability flags and deviation records in the standardized data stream includes:

[0078] The grid dispatch command value, the measured value of key operating parameters of the generating unit, and the sequence of equipment status characteristics are parsed from the standardized data stream.

[0079] Based on the power command change amount of the automatic generation control target power in the power grid dispatch command value, determine the trigger valid flag and the trigger reset flag, and integrate the trigger valid flag and the trigger reset flag into a hysteresis status flag;

[0080] The instantaneous deviation of the measured value of the main steam pressure in the key operating parameters of the unit is read in real time. Based on the instantaneous deviation, a condition deviation flag is generated, and all instantaneous deviations are stored in chronological order as a pressure deviation record sequence.

[0081] The first-order difference operation is performed on the time series of the bearing vibration amplitude to obtain the vibration change rate sequence. Based on the sign judgment result of the vibration change rate sequence, the trend warning sign is confirmed. At the same time, the temperature change sign is confirmed according to the historical difference of the bearing temperature change rate.

[0082] The hysteresis status flag, the pressure deviation record sequence, the trend warning flag, and the temperature sudden change flag are timestamped and packaged to obtain the trigger stability flag and deviation record.

[0083] The physical process features in the standardized data stream are combined to form a unified situational representation vector for the thermal power plant, including:

[0084] Extract combustion-side features, soft drink-side features, and electrical-side features from the standardized data stream;

[0085] The instantaneous value of fuel quantity and the furnace pressure fluctuation amplitude in the combustion side characteristic quantities are summed by a first weight to obtain the combustion side situation component. The measured value of main steam pressure and the rate of change of main steam temperature in the steam-water side characteristic quantities are summed by a second weight to obtain the steam-water side situation component.

[0086] The absolute value of the difference between the actual generated power of the unit and the target power of the automatic generation control in the electrical side characteristic quantities is calculated to obtain the electrical side deviation component;

[0087] The pressure deviation recording sequence is weighted according to a preset time decay factor and then appended to the soda-side situation component. The trend warning flag is multiplied into the combustion-side situation component as a Boolean coefficient.

[0088] The combustion-side situational component after multiplication, the added steam-water-side situational component, the electrical-side deviation component, and the hysteresis state flag are integrated into a unified situational representation vector.

[0089] Read the data blocks in the standardized data stream, extract the grid dispatch command value (i.e., the target power for automatic generation control) according to the preset field positions, extract the measured values ​​of the key operating parameters of the unit (i.e., the measured value of the main steam pressure), and then extract the equipment status characteristic sequence, including the bearing vibration amplitude time series and the bearing temperature change rate time series.

[0090] The power command change is obtained by subtracting the previous automatic generation control target power from the current automatic generation control target power. Pre-set lower and upper thresholds are determined by statistically analyzing the normal fluctuation range of the power command under stable operating conditions. The lower threshold is set to 80% of the normal fluctuation range, and the upper threshold is set to 1.2 times the normal fluctuation range. This change is compared with the lower and upper thresholds. A trigger validity flag is generated when the change rises from below the lower threshold to above the upper threshold, and a trigger reset flag is generated when the change falls from above the upper threshold to below the lower threshold. These two status values ​​are then combined into a joint status flag, namely the hysteresis status flag, which indicates the hysteresis interval in which the current power command change falls.

[0091] The measured values ​​of the main steam pressure are continuously read from the standardized data stream. The instantaneous deviation is obtained by subtracting the preset main steam pressure setpoint from the measured value. The main steam pressure setpoint is a fixed value determined based on the design parameters under rated operating conditions of the unit. The deviation threshold and the set time window length are determined by statistical analysis of pressure fluctuations in the unit's historical operating data. The deviation threshold is set to 50% of the normal pressure fluctuation amplitude, and the time window length is the average time required for the pressure to return to stability after a disturbance. When the absolute value of the instantaneous deviation continuously exceeds the deviation threshold and the duration exceeds the time window length, an operating condition deviation flag is generated. At the same time, the instantaneous deviation values ​​obtained at each sampling moment are stored sequentially in a storage area in chronological order, forming a pressure deviation record sequence.

[0092] The vibration rate of change sequence is obtained by subtracting the previous value from the subsequent value in the bearing vibration amplitude time series. The sign of each value in the vibration rate of change sequence is then determined sequentially. A preset rate of change threshold is obtained by statistically analyzing the distribution range of vibration rate of change during normal unit operation, using the value corresponding to the 95th percentile of the distribution range. When two adjacent signs change from positive to negative or vice versa, and the absolute value of both rates of change exceeds the preset rate of change threshold, a trend warning is confirmed, indicating a reversal of the vibration trend. Simultaneously, the bearing temperature rate of change value for the current period and the bearing temperature rate of change value stored in the previous period are read, and the two are subtracted to obtain the historical difference. A preset abrupt change threshold is determined through thermodynamic calculations based on the thermal expansion coefficient of the bearing material and the allowable rate of temperature change. If the absolute value of this historical difference exceeds the abrupt change threshold, a temperature abrupt change is confirmed.

[0093] The hysteresis status flag, pressure deviation record sequence, trend warning flag, and temperature change flag generated in the previous steps are collected according to the timestamp of the same moment. These data are then packaged into a complete data packet according to the preset packaging format. This data packet is the trigger stability flag and deviation record.

[0094] Read the data blocks in the standardized data stream, and extract the data belonging to the combustion side according to the preset characteristic field positions, including the instantaneous value of fuel quantity, the fluctuation range of furnace pressure, and the oxygen quantity deviation value. Extract the data belonging to the steam-water side, including the measured value of main steam pressure, the rate of change of main steam temperature, and the desuperheating water flow value. Extract the data belonging to the electrical side, including the actual power generation value of the unit, the target power of automatic power generation control, and the primary frequency regulation action flag.

[0095] The first and second weighting coefficients were obtained through multiple linear regression analysis of combustion stability test data of the unit under different loads, reflecting the relative contributions of fuel quantity and furnace pressure to the combustion status. The instantaneous fuel quantity is multiplied by the first weighting coefficient to obtain the first product, and the furnace pressure fluctuation amplitude is multiplied by the second weighting coefficient to obtain the second product. The first and second products are then added to obtain the combustion-side status component. The third and fourth weighting coefficients were determined after step response testing of the transfer function model of the steam-water system, respectively characterizing the influence of the main steam pressure and temperature change rate on the steam-water status. The measured main steam pressure is multiplied by the third weighting coefficient to obtain the third product, and the main steam temperature change rate is multiplied by the fourth weighting coefficient to obtain the fourth product. The third and fourth products are then added to obtain the steam-water status component.

[0096] The difference between the actual power output of the generating unit and the target power of the automatic generation control is obtained. The absolute value of this difference is the electrical side deviation component, which represents the degree of absolute deviation between the current output and the dispatch target.

[0097] The time decay factor is determined according to an exponential decay law; the further the deviation record is from the current time, the smaller the decay factor. The half-life of the decay factor is set to three control cycles based on the unit pressure dynamic response time constant. Each deviation value in the pressure deviation record sequence is read, and multiplied by the corresponding time decay factor according to its distance from the current time. All weighted deviation values ​​are summed and added to the steam-water side situation component to obtain the additional steam-water side situation component. The Boolean coefficient is directly determined by the trend warning flag; it is one when the trend warning flag is true and zero when it is false. The combustion-side situation component is multiplied by this coefficient to obtain the multiplied combustion-side situation component. This operation ensures that the combustion-side situation is included in the final vector only when a trend warning occurs; otherwise, the combustion-side situation component is zero.

[0098] The four values ​​obtained in the previous steps—the combustion-side situational component after multiplication, the added soda-liquid-side situational component, the electrical-side deviation component, and the previously generated hysteresis state flag—are arranged together in a preset order to form a multi-dimensional data structure, which is the unified situational representation vector. It carries the physical information of the combustion side, soda-liquid side, and electrical side, as well as the trigger state flag.

[0099] The beneficial effects are as follows: By parsing grid dispatch command values, measured unit parameter values, and equipment status sequences from standardized data streams, stability indicators are generated using hysteresis comparison of power command changes, avoiding signal oscillations. Real-time reading of instantaneous main steam pressure deviations generates operating condition deviation indicators and stores them in chronological order as a pressure deviation record sequence. First-order difference and sign reversal are performed on bearing vibration amplitudes to confirm trend warning indicators, and historical difference comparisons are used to confirm temperature mutation indicators for bearing temperature change rates. Furthermore, combustion-side, steam-water-side, and electrical-side feature quantities are extracted from the standardized data stream, weighted, and summed to obtain each situation component. The pressure deviation sequence is weighted and appended to the steam-water-side component, and the trend warning indicator is multiplied into the combustion-side component, ultimately integrating them into a unified situation representation vector. This achieves a multi-dimensional unified representation of grid regulation demands, operating condition deviations, and equipment health status, significantly improving the reliability of autonomous response initiation and the completeness of situational awareness.

[0100] G3: Input the unified situation representation vector into the preset online joint optimization rule base, and perform rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence;

[0101] In this embodiment of the invention, the preset online joint optimization rule base includes:

[0102] The online joint optimization rule base includes a security constraint matrix, an economic cost vector, and a flexibility adjustment capability mapping table;

[0103] Based on the load and pressure range where the thermal power plant is currently operating, the corresponding maximum load increase rate and minimum stable combustion load boundary value are read from the flexibility adjustment capability mapping table.

[0104] The power change rate limit and minimum load limit in the corresponding operating condition section of the safety constraint matrix are dynamically trimmed using the minimum stable combustion load boundary value.

[0105] The trimmed power change rate limit, minimum load limit, and untrimmed static safety limit are packaged together to obtain the joint safety boundary.

[0106] The joint security boundary and the economic cost vector are simultaneously input into the rolling optimization process to output the optimal control target sequence of the thermal power plant.

[0107] The step of inputting the unified situation representation vector into a preset online joint optimization rule base and performing rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence includes:

[0108] The unified situational representation vector is input into the online joint optimization rule base, and the target deviation value of the measured power value in multiple control cycles is read backward from the current time as the starting point.

[0109] The target deviation value and the pressure deviation record sequence in the unified situation characterization vector are aligned by time and merged into a joint deviation sequence.

[0110] Using the preset prediction time domain length as the window length and the joint deviation sequence as the initial condition, the rolling optimization process is invoked to generate a candidate control trajectory group;

[0111] The trajectory that minimizes the total economic cost is selected from the candidate control trajectory group, and the target values ​​of output power, main steam pressure and furnace outlet nitrogen oxide concentration in the trajectory are arranged in chronological order to obtain the optimal control target sequence.

[0112] The rolling optimization includes:

[0113] The last deviation value in the joint deviation sequence is used as the starting deviation;

[0114] The power adjustment step, pressure adjustment step, and nitrogen oxide adjustment step in the preset control step array are added to the actual values ​​of the current cycle to generate candidate values ​​for output power, main steam pressure, and furnace outlet nitrogen oxide concentration for the next control cycle. This operation is repeated until the entire prediction time domain is filled to obtain the candidate control trajectory.

[0115] Verify whether the candidate value of each control cycle in the candidate control trajectory satisfies all the constraints in the joint safety boundary;

[0116] The total economic cost of the candidate control trajectories that pass the verification is calculated, and the control trajectory with the minimum total economic cost is selected as the candidate control trajectory group.

[0117] The formula for calculating the total economic cost is as follows:

[0118]

[0119] in, To predict the total number of control cycles in the time domain, For the first Candidate output power values ​​for each cycle For the first The power grid dispatch command value for each cycle. The weight of the power deviation penalty term. For the first Candidate values ​​for the main steam pressure in each cycle. Main steam pressure setpoint The weight of the pressure fluctuation penalty term, For the first Candidate values ​​for nitrogen oxide concentration in each period, For emission-adjusted cost weighting, For the first Fuel consumption per cycle Fuel cost weighting.

[0120] The online joint optimization rule base pre-stores a safety constraint matrix, an economic cost vector, and a flexibility adjustment capability mapping table. The safety constraint matrix is ​​a table where rows correspond to different unit operating condition segments. These segments are multiple consecutive intervals divided according to the unit's load range; for example, 40% to 50% of rated load constitutes one segment, and 50% to 60% of rated load constitutes another. The columns correspond to the upper and lower limits of key safety parameters for the unit, including the upper limit of furnace pressure, the upper and lower limits of main steam temperature, the upper limit of bearing temperature, and the limit of generator power change rate. Each cell stores the maximum or minimum allowable value of that safety parameter within that operating condition segment.

[0121] The economic cost vector is a set of weighted values, each corresponding to the importance of a control objective. These weights are calibrated offline after economic analysis of historical power plant operating data. The flexibility adjustment capability mapping table is a two-dimensional lookup table structure. The rows of the table correspond to load intervals, and the columns correspond to pressure intervals. Each cell stores the maximum load increase rate and the minimum stable combustion load boundary value that the unit can achieve under the combination of that load interval and pressure interval. These values ​​are obtained through actual measurements during the unit's commissioning phase using load increase / decrease tests and stable combustion tests.

[0122] The system acquires the unit's current load and main steam pressure values ​​in real time. The load value is compared to preset load range boundaries to determine its corresponding load range; similarly, the main steam pressure value is compared to preset pressure range boundaries to determine its corresponding pressure range. The preset load range boundaries are determined during unit design based on equal division of the rated load, while the pressure range boundaries are defined based on the boiler's natural circulation pressure characteristics. Using these two ranges as indices, the system locates the corresponding cells in the flexibility adjustment capability mapping table and retrieves the maximum load increase rate and minimum stable combustion load boundary values ​​from those cells.

[0123] Locate the row in the safety constraint matrix that corresponds to the current operating condition segment. Replace the previously stored power change rate limit in that row with the maximum load increase rate read from the flexibility adjustment capability mapping table. Replace the previously stored minimum load limit in that row with the minimum stable combustion load boundary value. Other safety parameter limits, such as the upper limit of furnace pressure and the upper and lower limits of main steam temperature, remain unchanged.

[0124] The replaced power change rate limit and minimum load limit, together with the static safety limits in the safety constraint matrix such as the upper limit of furnace pressure, the upper and lower limits of main steam temperature, and the upper limit of bearing temperature, which have not been modified, are arranged and combined in a predetermined format to form a constraint set, which is called the joint safety boundary.

[0125] The rolling optimization process treats all constraints in the joint safety boundary as hard limitations, and any control scheme that does not meet these constraints is directly eliminated. Simultaneously, all weight values ​​in the economic cost vector are used as the basis for evaluating the importance of different control objectives, and the objectives are weighted and summed according to these weights during the optimization process. After optimization calculation, a set of target values ​​arranged continuously over time is output, including the output power target value, main steam pressure target value, and furnace outlet nitrogen oxide concentration target value for each control cycle. This set of target values ​​is called the optimal control objective sequence.

[0126] The electrical side deviation component is extracted from the unified situational awareness vector. This component is the deviation between the measured power value and the target value. Then, the same type of deviation values ​​stored in the historical memory area are retrieved sequentially from the most recent to the oldest data point. The number of cycles for rolling retrieval is preset, typically three to five cycles.

[0127] The newly read power target deviation value sequence is matched one-to-one with the pressure deviation record sequence carried in the unified situation characterization vector according to their respective timestamps, ensuring that two deviation values ​​at the same moment are paired together. Then, the paired deviation values ​​are arranged in chronological order to form a new sequence. Each element in this sequence contains both the power deviation and the pressure deviation at that moment. This new sequence is called the joint deviation sequence.

[0128] The prediction time domain length is a fixed value set based on the average time required for the unit to reach a new steady state after receiving the command, typically ten control cycles. The joint deviation sequence is used as the starting input for the rolling optimization process, which generates multiple possible future control trajectories. The set of these trajectories is called the candidate control trajectory group.

[0129] For each trajectory in the candidate control trajectory group, calculate its corresponding total economic cost. This total cost is obtained by multiplying the power deviation, pressure fluctuation, emission concentration, and fuel consumption of all control cycles in the trajectory by their respective weights and then summing them up. Compare the cost values ​​of all trajectories, find the trajectory with the minimum cost, and extract the target values ​​of output power, main steam pressure, and furnace outlet nitrogen oxide concentration for each control cycle of this trajectory in chronological order to form the optimal control target sequence.

[0130] The last deviation value in the joint deviation sequence is used as the starting deviation. The latest deviation value, which is the one closest to the current time, is taken from the joint deviation sequence and used as the starting point for rolling optimization, representing the deviation state at the current time.

[0131] The control step size array consists of three sets of adjustment step sizes determined in offline mode by analyzing the maximum permissible single-step adjustment of the unit. The power adjustment step size set contains multiple possible power adjustment increments, such as increments of plus or minus a few megawatts. The pressure adjustment step size set contains multiple possible pressure adjustment increments. The nitrogen oxide adjustment step size set contains multiple possible concentration adjustment increments.

[0132] In practice, a step size is selected from the power adjustment step size set and added to the actual output power value of the current cycle to obtain the candidate output power value for the next cycle; a step size is selected from the pressure adjustment step size set and added to the actual main steam pressure value of the current cycle to obtain the candidate main steam pressure value for the next cycle; a step size is selected from the nitrogen oxide adjustment step size set and added to the actual nitrogen oxide concentration value of the current cycle to obtain the candidate nitrogen oxide concentration value for the next cycle. Then, using the newly generated candidate value as the current value, the next set of step sizes is selected, and the above operation is repeated until all control cycles in the entire prediction time domain are filled, thus forming a complete candidate control trajectory. By enumerating different combinations of step sizes, multiple different candidate control trajectories can be generated.

[0133] For each control cycle in the candidate control trajectory, the following checks are performed sequentially: First, the candidate output power value for that cycle is checked to ensure it is not lower than the minimum load limit in the joint safety boundary and not higher than the rated load limit. Second, the power change between two adjacent control cycles divided by the cycle duration is checked to ensure it does not exceed the power change rate limit in the joint safety boundary. Third, the candidate main steam pressure value is checked to ensure it is between the upper and lower limits of the main steam pressure in the joint safety boundary. Fourth, the candidate furnace pressure value is checked to ensure it does not exceed the upper limit of the furnace pressure. Fifth, the candidate bearing temperature value is checked to ensure it does not exceed the upper limit of the bearing temperature. Only when all candidate values ​​for all control cycles satisfy all constraints can the candidate control trajectory pass the verification.

[0134] For each candidate control trajectory that passes the verification, the absolute value of the power deviation within each control cycle is multiplied by the power weight, the absolute value of the pressure fluctuation by the pressure weight, the nitrogen oxide concentration by the emission weight, and the fuel consumption by the fuel weight, according to a preset multi-objective cost accumulation rule. All these multiplications are then summed to obtain a total cost value. The total cost values ​​of all candidate control trajectories that pass the verification are compared, and the trajectories with the lowest total cost are identified and output as a group of candidate control trajectories.

[0135] The total number of control cycles in the prediction time domain is a fixed value preset based on the unit's dynamic response time, representing the number of control cycles predicted from the current moment forward. The candidate output power value for each cycle is the estimated unit output value generated during the rolling optimization process by adding the power adjustment step size in the control step size array to the actual value of the current cycle. The grid dispatch command value for each cycle is the target power value sent to the generating units by dispatching, parsed from the standardized data stream. The power deviation penalty term weight is a constant calibrated offline after performing economic analysis on historical power plant operating data; it is used to quantify the regulation cost caused by power deviation. The candidate values ​​for the main steam pressure in each cycle are estimated values ​​for the steam and water side pressures generated during the rolling optimization process. The setpoint for the main steam pressure is a fixed target pressure value determined based on the design parameters under rated operating conditions of the unit.

[0136] The pressure fluctuation penalty term weight is a constant determined through thermal stress analysis of the boiler's heating surfaces, used to quantify the lifespan loss caused by pressure deviation from the target value. The candidate values ​​for nitrogen oxide concentrations in each cycle are estimated values ​​of the reactor outlet emission levels generated during the rolling optimization process. The emission cost-based weight is a constant derived from a comprehensive conversion based on pollution discharge fees and non-penalty costs, used to quantify emission concentrations as economic costs. The fuel consumption for each cycle is an estimated fuel consumption obtained by mapping the output power candidate value through the fuel characteristic curve. The fuel cost weight is a constant determined based on the power plant's coal purchase price, used to quantify fuel consumption as an economic cost.

[0137] The calculation of the total economic cost involves summing the four costs for all control cycles within the prediction time domain. For each control cycle, the power deviation cost is calculated first by multiplying the weight of the power deviation penalty term by the absolute value of the difference between the candidate output power value and the grid dispatch command value. Then, the pressure fluctuation cost is calculated by multiplying the weight of the pressure fluctuation penalty term by the absolute value of the difference between the candidate main steam pressure value and the main steam pressure setpoint. Next, the emission cost is calculated by multiplying the emission cost weight by the candidate nitrogen oxide concentration value. Finally, the fuel cost is calculated by multiplying the fuel cost weight by the fuel consumption. The four values—power deviation cost, pressure fluctuation cost, emission cost, and fuel cost—calculated within the control cycle are summed to obtain the subtotal cost for that cycle. This calculation is repeated for all control cycles, with the subtotal cost for each cycle summed sequentially to obtain the total economic cost J for the entire prediction time domain. The smaller this sum, the better the overall performance of the candidate control trajectory.

[0138] The beneficial effects are as follows: By constructing an online joint optimization rule base containing a safety constraint matrix, an economic cost vector, and a flexibility adjustment capability mapping table, the maximum load increase rate and the minimum stable combustion load boundary values ​​are dynamically read based on the current load range and pressure range. The safety constraint matrix is ​​then trimmed to generate a joint safety boundary, achieving coordinated constraints on equipment physical safety and flexible adjustment capability. By merging the power deviation and pressure deviation record sequences in the unified situation representation vector into a joint deviation sequence, and using the joint deviation sequence as the initial condition, candidate control trajectories are generated using a step-size enumeration method. All constraints in the joint safety boundary are verified cycle by cycle, and the verified trajectories undergo multi-objective economic cost evaluation. The optimal trajectory is selected to form the optimal control target sequence, significantly improving the global optimality and safety of the control target sequence, ensuring that the target values ​​of output power, main steam pressure, and nitrogen oxide concentration achieve the best comprehensive performance while satisfying all constraints.

[0139] G4: Decompose each target value in the optimal control target sequence into fuel regulating valve opening command, turbine regulating valve opening command, and desuperheating water valve opening command, and send them to the controller and fieldbus equipment in the thermal power plant.

[0140] In this embodiment of the invention, the step of decomposing each target value in the optimal control target sequence into fuel regulating valve opening commands, turbine regulating valve opening commands, and desuperheating water valve opening commands, and issuing them to the controller or fieldbus device in the power plant, includes:

[0141] The output power target value of the current control cycle in the optimal control target sequence is subjected to first-order inertial filtering to obtain the turbine control valve opening command;

[0142] The deviation between the current main steam pressure target value and the measured value is input together with the output power target value to the fuel-pressure coordination link of the thermal power plant, outputting the fuel quantity demand value, and converting the fuel quantity demand value into a fuel regulating valve opening command;

[0143] The deviation between the target value of nitrogen oxide concentration at the furnace outlet and the current measured nitrogen oxide concentration is input into the proportional-integral calculation unit of the thermal power plant to obtain the desuperheating water flow correction value, and the desuperheating water valve opening command is generated based on the desuperheating water flow correction value.

[0144] The steam turbine control valve opening command, the fuel regulating valve opening command, and the desuperheating water valve opening command are respectively sent to the corresponding controllers and fieldbus devices in the thermal power plant.

[0145] The output power target value for the current control cycle is retrieved from the optimal control target sequence, and a first-order inertial filtering operation is performed on this output power target value. The filter coefficient of the first-order inertial filter is determined offline based on the mechanical inertial time constant of the turbine valve actuator. The turbine valve opening command value output after filtering in the previous control cycle is read, the output power target value of the current cycle is multiplied by the filter coefficient, and then the filtered output value of the previous cycle is multiplied by one and subtracted from the filter coefficient. The two products are added together to obtain a new filtering result. The filter coefficient is a fixed value between 0.1 and 0.3. The smaller the value, the stronger the filtering effect but the slower the response. The value obtained after the first-order inertial filtering is the turbine valve opening command.

[0146] The fuel-pressure coordination unit is a pre-set processing unit based on boiler combustion characteristics. It looks up a pre-calibrated power-reference fuel quantity curve based on the output power target value. This curve is obtained by fitting measured fuel consumption at multiple steady-state load points, with each power value on the curve corresponding to a reference fuel quantity. Simultaneously, the pressure deviation is calculated by comparing the target and measured main steam pressure values. The fuel correction amount is then determined using a pre-set pressure deviation-fuel correction coefficient table, which was established through step response testing during boiler disturbance experiments, based on the sign and magnitude of the pressure deviation. The reference fuel quantity and the fuel correction amount are added to obtain the fuel demand value. Then, based on the correspondence between fuel quantity and valve opening, this demand value is converted into a fuel regulating valve opening command. This correspondence is a linearized table obtained through valve flow characteristic tests. Each row in the table records the valve opening percentage corresponding to a fuel quantity value. Linear interpolation is used to find the opening value corresponding to the current fuel demand value.

[0147] The proportional-integral calculation stage comprises two branches: proportional action and integral action. The proportional coefficient is tuned offline based on the response sensitivity of the desuperheating water valve to changes in nitrogen oxide concentration, while the integral coefficient is determined based on the time constant required for the system to eliminate steady-state error.

[0148] The proportional action branch multiplies the nitrogen oxide concentration deviation by a proportional coefficient to obtain the proportional correction component. The integral action branch accumulates the nitrogen oxide concentration deviation for each control cycle into an integral register, then multiplies the sum by an integral coefficient to obtain the integral correction component. The proportional and integral correction components are added to obtain the desuperheating water flow rate correction value. After obtaining the correction value, a desuperheating water valve opening command is generated based on the correspondence between the desuperheating water flow rate and the valve opening. This correspondence is determined by the factory calibration curve of the desuperheating water regulating valve. The current desuperheating water flow rate setpoint is added to the correction value and substituted into the curve to obtain the corresponding valve opening percentage.

[0149] The turbine control valve opening command is transmitted to the control valve servo controller in the turbine control cabinet via an analog output channel in the form of a 4-20 mA current signal. The fuel regulating valve opening command is transmitted to the fuel regulating valve actuator intelligent positioner via a fieldbus network in the form of a digital communication message. The desuperheating water valve opening command is transmitted to the desuperheating water regulating valve electric actuator controller via a fieldbus network.

[0150] The beneficial effects are as follows: By performing first-order inertial filtering on the output power target value, a smooth turbine valve opening command is generated using the filtering coefficient determined based on the inertial time constant of the actuator, avoiding the impact of sudden valve changes on the turbine and power grid. Through the fuel-pressure coordination link, the pressure deviation and power target are jointly mapped to fuel demand using the power-reference fuel quantity curve and the pressure deviation-fuel correction coefficient table, and then converted into a fuel regulating valve opening command, realizing coordinated control of the boiler and turbine, improving the stability of load following and the accuracy of pressure control. Through the proportional-integral calculation link, the nitrogen oxide concentration deviation is processed using offline tuned proportional and integral coefficients to obtain the desuperheating water flow correction value and generate a valve opening command, effectively suppressing the periodic fluctuations and steady-state deviations of emission concentration. The three commands are sent to the corresponding controllers and fieldbus devices respectively, ensuring the relevance and reliability of command transmission, significantly improving the smoothness of actuator operation and the stability of overall control.

[0151] like Figure 2 The diagram shown is a functional block diagram of an autonomous response system for a thermal power plant system architecture provided in an embodiment of the present invention.

[0152] The autonomous response system 100 for a thermal power plant system architecture described in this invention can be installed in an electronic device. Depending on the functions implemented, the autonomous response system 100 for a thermal power plant system architecture may include a standardized data conversion module 101, a multi-dimensional situation fusion module 102, a rolling optimization module 103, and a command issuance module 104. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0153] In this embodiment, the functions of each module / unit are as follows:

[0154] The standardized data conversion module 101 is used to convert the load response signal and equipment risk signal of the thermal power plant into a standardized data stream;

[0155] The multi-dimensional situation fusion module 102 is used to monitor the trigger stability flags and deviation records in the standardized data stream in multiple dimensions, and merge them with the physical process feature quantities in the standardized data stream into a unified situation representation vector of the thermal power plant.

[0156] The rolling optimization module 103 is used to input the unified situation representation vector into a preset online joint optimization rule base, and to perform rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence.

[0157] The instruction issuing module 104 is used to decompose each target value in the optimal control target sequence into fuel regulating valve opening instructions, turbine regulating valve opening instructions and desuperheating water valve opening instructions, and issue them to the controller and fieldbus equipment in the thermal power plant.

[0158] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0159] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0160] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0161] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0162] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An autonomous response method for a thermal power plant system architecture, characterized in that, The method includes: G1: Converts the load response signals and equipment risk signals of thermal power plants into standardized data streams; G2: Multi-dimensional monitoring of the trigger stability flags and deviation records in the standardized data stream, and merging them with the physical process characteristics in the standardized data stream into a unified situational representation vector for the thermal power plant; G3: Input the unified situation representation vector into the preset online joint optimization rule base, and perform rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence; G4: Decompose each target value in the optimal control target sequence into fuel regulating valve opening command, turbine regulating valve opening command, and desuperheating water valve opening command, and send them to the controller and fieldbus equipment in the thermal power plant.

2. The autonomous response method for a thermal power plant system architecture as described in claim 1, characterized in that, The process of converting the load response signals and equipment risk signals of thermal power plants into standardized data streams includes: The load response signal and equipment risk signal of the thermal power plant are acquired. The load response signal includes the automatic generation control command value, the actual power generation value of the unit and the frequency regulation action flag. The equipment risk signal includes the bearing vibration amplitude, the bearing temperature change rate and the furnace pressure fluctuation amplitude. Based on the deviation between the automatic generation control command value and the actual power generation value of the unit, a power deviation signal is determined, and the power deviation signal is aligned with the timestamp of the frequency regulation action flag bit and encapsulated into a load response data frame. The vibration amplitude of the bearing is extracted by extracting the envelope to obtain the vibration trend feature value. The vibration trend feature value, the historical difference result of the bearing temperature change rate, and the furnace pressure fluctuation amplitude are combined into an equipment risk data frame. The load response data frame and the equipment risk data frame are spliced ​​together using a timestamp reference and then subjected to rule verification to obtain a standardized data stream.

3. The autonomous response method for a thermal power plant system architecture as described in claim 2, characterized in that, The multi-dimensional monitoring of the trigger stability flags and deviation records in the standardized data stream includes: The grid dispatch command value, the measured value of key operating parameters of the generating unit, and the sequence of equipment status characteristics are parsed from the standardized data stream. Based on the power command change amount of the automatic generation control target power in the power grid dispatch command value, determine the trigger valid flag and the trigger reset flag, and integrate the trigger valid flag and the trigger reset flag into a hysteresis status flag; The instantaneous deviation of the measured value of the main steam pressure in the key operating parameters of the unit is read in real time. Based on the instantaneous deviation, a condition deviation flag is generated, and all instantaneous deviations are stored in chronological order as a pressure deviation record sequence. The first-order difference operation is performed on the time series of the bearing vibration amplitude to obtain the vibration change rate sequence. Based on the sign judgment result of the vibration change rate sequence, the trend warning sign is confirmed. At the same time, the temperature change sign is confirmed according to the historical difference of the bearing temperature change rate. The hysteresis status flag, the pressure deviation record sequence, the trend warning flag, and the temperature sudden change flag are timestamped and packaged to obtain the trigger stability flag and deviation record.

4. The autonomous response method for a thermal power plant system architecture as described in claim 3, characterized in that, The physical process features in the standardized data stream are combined to form a unified situational representation vector for the thermal power plant, including: Extract combustion-side features, soft drink-side features, and electrical-side features from the standardized data stream; The instantaneous value of fuel quantity and the furnace pressure fluctuation amplitude in the combustion side characteristic quantities are summed by a first weight to obtain the combustion side situation component. The measured value of main steam pressure and the rate of change of main steam temperature in the steam-water side characteristic quantities are summed by a second weight to obtain the steam-water side situation component. The absolute value of the difference between the actual generated power of the unit and the target power of the automatic generation control in the electrical side characteristic quantities is calculated to obtain the electrical side deviation component; The pressure deviation recording sequence is weighted according to a preset time decay factor and then appended to the soda-side situation component. The trend warning flag is multiplied into the combustion-side situation component as a Boolean coefficient. The combustion-side situational component after multiplication, the added steam-water-side situational component, the electrical-side deviation component, and the hysteresis state flag are integrated into a unified situational representation vector.

5. The autonomous response method for a thermal power plant system architecture as described in claim 1, characterized in that, The preset online joint optimization rule base includes: The online joint optimization rule base includes a security constraint matrix, an economic cost vector, and a flexibility adjustment capability mapping table; Based on the load and pressure range where the thermal power plant is currently operating, the corresponding maximum load increase rate and minimum stable combustion load boundary value are read from the flexibility adjustment capability mapping table. The power change rate limit and minimum load limit in the corresponding operating condition section of the safety constraint matrix are dynamically trimmed using the minimum stable combustion load boundary value. The trimmed power change rate limit, minimum load limit, and untrimmed static safety limit are packaged together to obtain the joint safety boundary. The joint security boundary and the economic cost vector are simultaneously input into the rolling optimization process to output the optimal control target sequence of the thermal power plant.

6. The autonomous response method for a thermal power plant system architecture as described in claim 5, characterized in that, The step of inputting the unified situation representation vector into a preset online joint optimization rule base and performing rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence includes: The unified situational representation vector is input into the online joint optimization rule base, and the target deviation value of the measured power value in multiple control cycles is read backward from the current time as the starting point. The target deviation value and the pressure deviation record sequence in the unified situation characterization vector are aligned by time and merged into a joint deviation sequence. Using the preset prediction time domain length as the window length and the joint deviation sequence as the initial condition, the rolling optimization process is invoked to generate a candidate control trajectory group; The trajectory that minimizes the total economic cost is selected from the candidate control trajectory group, and the target values ​​of output power, main steam pressure and furnace outlet nitrogen oxide concentration in the trajectory are arranged in chronological order to obtain the optimal control target sequence.

7. The autonomous response method for a thermal power plant system architecture as described in claim 6, characterized in that, The rolling optimization includes: The last deviation value in the joint deviation sequence is used as the starting deviation; The power adjustment step, pressure adjustment step, and nitrogen oxide adjustment step in the preset control step array are added to the actual values ​​of the current cycle to generate candidate values ​​for output power, main steam pressure, and furnace outlet nitrogen oxide concentration for the next control cycle. This operation is repeated until the entire prediction time domain is filled to obtain the candidate control trajectory. Verify whether the candidate value of each control cycle in the candidate control trajectory satisfies all the constraints in the joint safety boundary; The total economic cost of the candidate control trajectories that pass the verification is calculated, and the control trajectory with the minimum total economic cost is selected as the candidate control trajectory group.

8. The autonomous response method for a thermal power plant system architecture as described in claim 7, characterized in that, The formula for calculating the total economic cost is as follows: in, To predict the total number of control cycles in the time domain, For the first Candidate output power values ​​for each cycle For the first The power grid dispatch command value for each cycle. The weight of the power deviation penalty term. For the first Candidate values ​​for the main steam pressure in each cycle. Main steam pressure setpoint The weight of the pressure fluctuation penalty term, For the first Candidate values ​​for nitrogen oxide concentration in each period, For emission-adjusted cost weighting, For the first Fuel consumption per cycle The weight is based on fuel cost.

9. The autonomous response method for a thermal power plant system architecture as described in claim 6, characterized in that, The process of decomposing each target value in the optimal control target sequence into fuel regulating valve opening commands, turbine regulating valve opening commands, and desuperheating water valve opening commands, and issuing them to controllers or fieldbus devices in the power plant, includes: The output power target value of the current control cycle in the optimal control target sequence is subjected to first-order inertial filtering to obtain the turbine control valve opening command; The deviation between the current main steam pressure target value and the measured value is input together with the output power target value into the fuel-pressure coordination link of the thermal power plant, outputting the fuel quantity demand value, and converting the fuel quantity demand value into a fuel regulating valve opening command; The deviation between the target value of nitrogen oxide concentration at the furnace outlet and the current measured nitrogen oxide concentration is input into the proportional-integral calculation unit of the thermal power plant to obtain the desuperheating water flow correction value, and the desuperheating water valve opening command is generated based on the desuperheating water flow correction value. The steam turbine control valve opening command, the fuel regulating valve opening command, and the desuperheating water valve opening command are respectively sent to the corresponding controllers and fieldbus devices in the thermal power plant.

10. An autonomous response system for a thermal power plant system architecture, characterized in that, An autonomous response method for implementing a thermal power plant system architecture as described in claim 1, the system comprising: The standardized data conversion module is used to convert the load response signals and equipment risk signals of thermal power plants into standardized data streams; The multi-dimensional situation fusion module is used to monitor the trigger stability flags and deviation records in the standardized data stream from multiple dimensions, and merge them with the physical process feature quantities in the standardized data stream into a unified situation representation vector of the thermal power plant. The rolling optimization module is used to input the unified situation representation vector into a preset online joint optimization rule base, and to perform rolling optimization on the initial deviation sequence in the unified situation representation vector to obtain the optimal control target sequence. The instruction issuing module is used to decompose each target value in the optimal control target sequence into fuel regulating valve opening instructions, turbine regulating valve opening instructions, and desuperheating water valve opening instructions, and issue them to the controllers and fieldbus devices in the thermal power plant.

Images