Generator set operation index optimization method and device, equipment and medium

Abstract

The application relates to the technical field of intelligent control of the power generation industry, in particular to a generator set operation index optimization method, device, equipment and medium, which comprises the following steps: inputting multi-source operation data of a current generator set into a preset time sequence prediction model to perform operation parameter prediction, comparing the prediction result with historical optimal operation parameters under corresponding working conditions, generating an operation deviation report, and performing attribution analysis on the operation deviation report based on a preset SHAP algorithm; generating operation index optimization suggestions according to the attribution analysis result, optimizing the operation index of the current generator set, and obtaining the optimized operation index. Thus, the problem that the operation economy and safety of the generator set are poor due to the lack of dynamic prediction, interpretable strategy generation, safety boundary evaluation and experience closed-loop mechanism in the existing power generation technology is solved, dynamic intelligent optimization and safety regulation and control of the operation index of the generator set are realized, and the operation economy and safety are significantly improved.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology in the power generation industry, and in particular to a method, device, equipment and medium for optimizing the operating indicators of generator sets. Background Technology

[0002] Currently, generator sets face the dual challenges of optimizing economic efficiency and ensuring safe operation. Traditional methods, which rely mainly on manual experience or small-model-based automatic control systems for load regulation and parameter optimization, have significant limitations.

[0003] For example, load regulation relies on human experience, and overshoot or lag in automatic control systems can lead to deviations in economic efficiency and safety. While traditional small models can identify some operational deviations, they lack the ability to deeply attribute the causes of these deviations and cannot generate explainable and executable optimization strategies. Furthermore, existing systems generally lack closed-loop verification mechanisms, making it difficult to accumulate and reuse effective operational experience, thus limiting the system's ability to continuously optimize. In high-risk operating scenarios such as deep peak shaving and peak load operations, problems such as delayed risk identification and insufficient operational standardization are particularly prominent, seriously affecting the safety and economy of unit operation and urgently requiring solutions. Summary of the Invention

[0004] This application provides a method, apparatus, equipment, and medium for optimizing generator set operating indicators to solve the problems of poor economic efficiency and low safety of generator set operation caused by the lack of dynamic prediction, interpretable strategy generation, safety boundary assessment, and experience closed-loop mechanism in existing power generation technologies. It realizes dynamic intelligent optimization and safe control of generator set operating indicators, significantly improving operating economy and safety.

[0005] The first aspect of this application provides a method for optimizing the operating indicators of a generator set, comprising the following steps:

[0006] Acquire multi-source operating data of the current generator set, and input the multi-source operating data into a preset time series prediction model to predict operating parameters and obtain prediction results;

[0007] The predicted results are compared with the historical best operating parameters under the corresponding working conditions to generate an operating deviation report. Attribution analysis is then performed on the operating deviation report based on the preset SHAP algorithm to obtain the attribution analysis results.

[0008] Based on the attribution analysis results, optimization suggestions for operating indicators are generated, and the current operating indicators of the generator set are optimized according to the optimization suggestions to obtain the optimized operating indicators.

[0009] According to one embodiment of this application, after optimizing the current generator set operating indicators based on the operating indicator optimization suggestions to obtain the optimized operating indicators, the method further includes:

[0010] The current generator set is controlled to operate based on optimized operating indicators, and the target effective strategy is selected according to the operating results. The target effective strategy is then used to iteratively update the preset time series prediction model.

[0011] According to one embodiment of this application, controlling the current generator set to operate based on optimized operating indicators, and selecting target effective strategies based on the operating results, includes:

[0012] The actual coal consumption reduction and the estimated coal consumption reduction are determined based on the operational results.

[0013] Calculate the deviation rate between the actual decrease in coal consumption and the estimated decrease in coal consumption;

[0014] If the deviation rate is less than a preset deviation threshold, the optimization suggestions for the operating indicators corresponding to the operating results are marked as the target effective strategy.

[0015] According to one embodiment of this application, after generating the operational indicator optimization suggestions based on the attribution analysis results, the method further includes:

[0016] Calculate the multidimensional risk index based on the aforementioned operational indicator optimization suggestions;

[0017] The safety boundary of the current generator set when implementing the operation index optimization suggestions is dynamically assessed based on the multidimensional risk index, and control suggestions are generated based on the safety boundary.

[0018] According to one embodiment of this application, the multidimensional risk index is:

[0019] ;

[0020] in, As a multidimensional risk index, This is the real-time temperature value. To preset a safe temperature threshold, Weighted sum of multiple preset gas concentrations, These are the operating parameters of the target device.

[0021] According to one embodiment of this application, the preset SHAP algorithm is:

[0022] ;

[0023] in, For the first The SHAP value of each feature, For feature set, For the total number of features, This is the output function for the model.

[0024] According to one embodiment of this application, the multi-source operating data includes at least one of the following: daily load curve of the power grid, real-time operating parameters, fuel parameters, and pulverizing system combination mode; and the operating indicator optimization suggestions include at least one of the following: type label, specific action, economic estimate, and risk warning.

[0025] According to the generator set operation index optimization method provided in this application embodiment, the multi-source operation data of the current generator set is input into a preset time-series prediction model to predict operation parameters. The prediction results are compared with the historical best operation parameters under the corresponding operating conditions to generate an operation deviation report. Based on a preset SHAP algorithm, attribution analysis is performed on the operation deviation report. Based on the attribution analysis results, operation index optimization suggestions are generated, and the current generator set operation index is optimized to obtain the optimized operation index. This solves the problems of poor generator set operation economy and low safety caused by the lack of dynamic prediction, interpretable strategy generation, safety boundary assessment, and experience closed-loop mechanism in existing power generation technologies. It realizes dynamic intelligent optimization and safe control of generator set operation index, significantly improving operation economy and safety.

[0026] A second aspect of this application provides a device for optimizing the operating parameters of a generator set, comprising:

[0027] The prediction module is used to acquire multi-source operating data of the current generator set, and input the multi-source operating data into a preset time series prediction model to predict operating parameters and obtain prediction results.

[0028] The analysis module is used to compare the prediction results with the historical best operating parameters under the corresponding working conditions, generate an operating deviation report, and perform attribution analysis on the operating deviation report based on the preset SHAP algorithm to obtain the attribution analysis results.

[0029] The optimization module is used to generate operational index optimization suggestions based on the attribution analysis results, and to optimize the current generator set operational indexes based on the operational index optimization suggestions to obtain optimized operational indexes.

[0030] According to one embodiment of this application, after optimizing the current generator set operating indicators based on the operating indicator optimization suggestions to obtain the optimized operating indicators, the optimization module is further configured to:

[0031] The current generator set is controlled to operate based on optimized operating indicators, and the target effective strategy is selected according to the operating results. The target effective strategy is then used to iteratively update the preset time series prediction model.

[0032] According to one embodiment of this application, the optimization module is further configured to:

[0033] The actual coal consumption reduction and the estimated coal consumption reduction are determined based on the operational results.

[0034] Calculate the deviation rate between the actual decrease in coal consumption and the estimated decrease in coal consumption;

[0035] If the deviation rate is less than a preset deviation threshold, the optimization suggestions for the operating indicators corresponding to the operating results are marked as the target effective strategy.

[0036] According to one embodiment of this application, after generating the operational indicator optimization suggestions based on the attribution analysis results, the analysis module is further configured to:

[0037] Calculate the multidimensional risk index based on the aforementioned operational indicator optimization suggestions;

[0038] The safety boundary of the current generator set when implementing the operation index optimization suggestions is dynamically assessed based on the multidimensional risk index, and control suggestions are generated based on the safety boundary.

[0039] According to one embodiment of this application, the multidimensional risk index is:

[0040] ;

[0041] in, As a multidimensional risk index, This is the real-time temperature value. To preset a safe temperature threshold, Weighted sum of multiple preset gas concentrations, These are the operating parameters of the target device.

[0042] According to one embodiment of this application, the preset SHAP algorithm is:

[0043] ;

[0044] in, For the first The SHAP value of each feature, For feature set, For the total number of features, This is the output function for the model.

[0045] According to one embodiment of this application, the multi-source operating data includes at least one of the following: daily load curve of the power grid, real-time operating parameters, fuel parameters, and pulverizing system combination mode; and the operating indicator optimization suggestions include at least one of the following: type label, specific action, economic estimate, and risk warning.

[0046] The generator set operation index optimization device provided in this application embodiment inputs multi-source operation data of the current generator set into a preset time-series prediction model to predict operation parameters. The prediction results are compared with the historical best operation parameters under the corresponding operating conditions to generate an operation deviation report. Based on a preset SHAP algorithm, attribution analysis is performed on the operation deviation report. Based on the attribution analysis results, operation index optimization suggestions are generated, and the current generator set operation index is optimized to obtain the optimized operation index. This solves the problems of poor generator set operation economy and low safety caused by the lack of dynamic prediction, interpretable strategy generation, safety boundary assessment, and experience closed-loop mechanisms in existing power generation technologies. It achieves dynamic intelligent optimization and safe control of generator set operation indicators, significantly improving operation economy and safety.

[0047] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for optimizing generator set operating indicators as described in the above embodiments.

[0048] A fourth aspect of this application provides a computer-readable storage medium storing computer instructions for causing the computer to execute the generator set operation index optimization method as described in the above embodiments.

[0049] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0050] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0051] Figure 1 This is a flowchart of a method for optimizing generator set operating indicators according to an embodiment of this application;

[0052] Figure 2 This is a flowchart of the operation of a generator set dynamic optimization and safety control system based on a time-series large model according to an embodiment of this application;

[0053] Figure 3This is an optimization flowchart for a 300MW load steady-state coal consumption optimization scenario according to an embodiment of this application;

[0054] Figure 4 This is an optimization flowchart for a 600MW unit being deeply adjusted to a 140MW safety optimization scenario according to one embodiment of this application;

[0055] Figure 5 This is a block diagram of a generator set operation index optimization device according to an embodiment of this application;

[0056] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0057] Reference numerals: 10-Optimization device for generator set operating indicators, 100-Prediction module, 200-Analysis module, 300-Optimization module; 601-Memory, 602-Processor, 603-Communication interface. Detailed Implementation

[0058] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0059] The following description, with reference to the accompanying drawings, outlines a method, apparatus, equipment, and medium for optimizing generator set operating indicators according to embodiments of this application. Addressing the challenges mentioned in the background section regarding the difficulty of simultaneously resolving the four core issues of dynamic prediction, strategy interpretability, dynamic assessment of safety boundaries, and experience-based closed-loop management in existing power industry indicator optimization and safety control technologies, this application provides a method for optimizing generator set operating indicators. By introducing a large-scale time-series model, combining an expert knowledge base and a closed-loop optimization mechanism, a complete technical system for dynamic optimization and safety control of generator set operating indicators is constructed. This significantly improves operational economy and safety, and is applicable to the economic optimization and safety control of thermal power units under different operating conditions.

[0060] Specifically, Figure 1 This is a flowchart illustrating a method for optimizing generator set operating indicators provided in an embodiment of this application.

[0061] like Figure 1 As shown, the optimization method for the generator set's operating indicators includes the following steps:

[0062] In step S101, the multi-source operating data of the current generator set is obtained, and the multi-source operating data is input into the preset time series prediction model to predict the operating parameters and obtain the prediction results.

[0063] In some embodiments, the multi-source operating data includes at least one of the following: daily load curve of the power grid, real-time operating parameters, fuel parameters, and pulverizing system combination mode.

[0064] Preferably, the preset temporal prediction model in this application embodiment can adopt a large-scale temporal model based on the Transformer architecture finely tuned by LoRA technology, and construct a temporal feature extraction network based on the Encoder layer of the Transformer, with a hidden layer dimension of It can be set to 512, and the number of attention heads can be 8. By using multi-head attention, the long-term and short-term dependencies between load and equipment parameters (such as the hysteresis change of main steam pressure 5 minutes after the load rises) can be captured.

[0065] In addition, for the large-scale temporal model, LoRA fine-tuning is performed on the attention layer and feedforward network layer of the Transformer architecture. For the generator set-specific scenario, a low-rank LoRA matrix with rank r=8 is set respectively. , , , ), updated via parameters , To achieve model fine-tuning, the original Transformer parameters are fixed during the fine-tuning process, and only the low-rank matrix is ​​updated. That is, only the low-rank parameters of the attention layer and the feedforward network layer are updated, reducing the scale of the fine-tuning parameters to less than 10% of the original model and reducing the consumption of computing resources.

[0066] The preset time-series prediction model in this application embodiment can also use a sliding window mechanism to segment continuous time-series data, combined with sine-cosine position encoding. and Process time-series information and mark data time positions to prevent the model from losing time-series information. Among these, For time location index, For dimensional indexing, This represents the dimension of the model's hidden layers.

[0067] Furthermore, the daily load curve of the power grid (e.g., 96 points / day) reflects the periodic and trend changes in electricity demand over the next day. Real-time operating parameters include key operating variables of the generator units themselves, such as main steam pressure, main steam temperature, ammonia injection rate for denitrification, fan vibration, and coal consumption rate. Fuel parameters mainly refer to the characteristics of the coal, such as the calorific value, volatile matter, and sulfur content of the coal type. The pulverizing system combination refers to the combination of coal mills currently in operation (e.g., mill combination 1 / 3 / 4). Different combinations will affect the fineness of pulverized coal, ventilation volume, etc., and thus affect the combustion stability and efficiency of the boiler. In addition, real-time operating parameters may also include real-time environmental indicators (NOx, SO2) and equipment vibration sequence.

[0068] Furthermore, in this embodiment, the daily load curve of the power grid can be obtained in real time from the power grid dispatch center through the data interface; the operating data such as main steam pressure, temperature, and fan vibration can be collected in real time through the power plant's distributed control system; the calorific value and volatile matter of coal can be obtained from the fuel management system; and the operating status of the coal mill can be read through the power plant's plant-level monitoring information system, and the current operating combination mode can be automatically identified.

[0069] Furthermore, the multi-source operational data obtained above is input into a preset time series prediction model (such as a Transformer architecture time series model finely tuned by LoRA technology) to predict minute-level operational parameters. The model captures the long-term and short-term dependencies between load fluctuations and equipment parameters through an attention mechanism, and outputs a predicted sequence of key parameters (such as main steam temperature, coal consumption rate, and fan efficiency value) in the future short period (such as 15 minutes). The prediction error is controlled within 3%. By predicting the operating status in advance, a decision-making basis for subsequent optimization is provided.

[0070] In step S102, the prediction results are compared with the historical best operating parameters under the corresponding working conditions to generate an operating deviation report. Based on the preset SHAP algorithm, the operating deviation report is subjected to attribution analysis to obtain the attribution analysis results.

[0071] Specifically, the embodiments of this application calculate the Euclidean distance between the predicted value and the historical best value. Generate a deviation report, in which, For the first The predicted values ​​of each parameter, For the first The historical optimal values ​​of each parameter This represents the total number of parameters; it also distinguishes between different types of deviations, including economic deviations (coal consumption, NOx emissions) and safety deviations (vibration, temperature, load response), etc.

[0072] Furthermore, combined with the improved SHAP value algorithm Feature importance is ranked to achieve four types of attribution analysis, among which, For the first The SHAP value of each feature, For feature set, For the total number of features, The model output function is used to perform multi-dimensional attribution based on the context (coal type characteristics, equipment defects, and operating mode).

[0073] Optionally, typical cause analysis may include unsuitable coal type (fluctuation in calorific value, abnormal volatile matter), deviation of operating mode from the optimal combination, hidden defects in equipment (such as sensor drift, coal mill blockage), and operational risk interference (maintenance operations affect operating parameters, resulting in low induced draft).

[0074] In step S103, optimization suggestions for operating indicators are generated based on the attribution analysis results, and the operating indicators of the current generator set are optimized based on the optimization suggestions to obtain the optimized operating indicators.

[0075] The operational indicator optimization suggestions include at least one of the following: type label (such as operation mode / coal blending / equipment feedback), specific action (such as "adjust the opening of the hot air damper of No. 1 coal mill to 55%), estimated economic improvement (unit: g / kWh), and risk warning (such as "pay attention to sudden changes in flue gas temperature").

[0076] Specifically, a collaborative reasoning mechanism between an expert rule base and a large language model is constructed. This mechanism employs a hybrid strategy combining rule matching (e.g., directly invoking rules when confidence is ≥0.85) and model generation (initiating LLM inference when confidence is <0.85) to generate structured optimization suggestions with quantitative indicators. By invoking the expert rule base and semantic parsing model, executable optimization actions are generated, and a structured suggestion format (type label, specific action, benefit prediction, and risk warning) is output.

[0077] Furthermore, in some embodiments, after generating operational indicator optimization suggestions based on the attribution analysis results, the method further includes: calculating a multidimensional risk index based on the operational indicator optimization suggestions; dynamically assessing the safety boundary of the current generator set when implementing the operational indicator optimization suggestions based on the multidimensional risk index; and generating control suggestions based on the safety boundary.

[0078] For example, the security boundary deduction and control mechanism of this application embodiment can be applied to deep-tuning security optimization scenarios and peak security control scenarios.

[0079] For example, in deep peak-shaving scenarios with a load rate ≤40%, the safety boundary is dynamically assessed. Using ERP system defect data (three or more types of defects), work order system operation information, and a historical accident case database as input data, a multidimensional risk index is calculated based on a pre-defined time-series large-scale model.

[0080] ;

[0081] in, As a multidimensional risk index, This is the real-time temperature value. To preset a safe temperature threshold, Weighted sum of multiple preset gas concentrations, These are the key operating parameters of the target equipment.

[0082] The safety boundary is dynamically assessed based on a multidimensional risk index, and control recommendations are output. A Level 1 warning is triggered. Ultimately, the system outputs the minimum safe load boundary, upper limit of load change rate, oxygen control range and combustion mode recommendations, and high-risk operation restriction recommendations. For example, the defect classification assessment unit retrieves ERP defect information, associates three or more types of defects with in-depth risk assessment, and dynamically associates work orders through the work conflict check unit, generating restriction recommendations for high-risk operations. Furthermore, the historical case reasoning unit compares the similarity between the current operating conditions and accident cases, triggering a warning.

[0083] Optionally, during peak load periods, potential equipment safety hazards can be identified and mitigated. Real-time equipment status (such as main steam temperature and furnace positive pressure), combustion characteristics of high-calorific-value coal, and DCS high-limit protection values ​​are used as input data to calculate a multidimensional risk index based on the aforementioned pre-defined time-series large-scale model. Dynamically assess the safety boundary and output control recommendations, when A Level 1 warning is triggered. Ultimately, the system outputs equipment defect linkage alarms, real-time verification of protection logic, coal type adaptability alerts, and a whitelist of peak-stage operations. For example, a non-closed-loop defect in the boiler / main steam system is marked as a red alarm, and protection logic verification is performed. The main steam temperature is compared with the DCS high-limit protection value in real time, and coal type adaptability analysis is conducted, alerting users of high-calorific-value coal types to the risk of positive pressure in the furnace.

[0084] Furthermore, in some embodiments, after optimizing the current generator set's operating indicators based on the operating indicator optimization suggestions to obtain optimized operating indicators, the method further includes: controlling the current generator set to operate based on the optimized operating indicators, selecting target effective strategies based on the operating results, and iteratively updating the preset time series prediction model using the target effective strategies.

[0085] Furthermore, in some embodiments, controlling the current generator set to operate based on optimized operating indicators and selecting target effective strategies based on the operating results includes: determining the actual coal consumption reduction value and the estimated coal consumption reduction value based on the operating results; calculating the deviation rate between the actual coal consumption reduction value and the estimated coal consumption reduction value; and marking the operating indicator optimization suggestions corresponding to the operating results as target effective strategies when the deviation rate is less than a preset deviation threshold.

[0086] Specifically, the embodiments of this application design a closed-loop optimization and knowledge accumulation mechanism, through... Verify the effectiveness of the strategy execution, among which, This represents the actual decrease in coal consumption. This is a projected estimate of the decrease in coal consumption.

[0087] The closed-loop optimization mechanism includes verifying the optimization effect (e.g., the coal consumption reduction value ≥ 90% of the estimated value), generating a labeled strategy (e.g., "medium load + high ash coal optimization #042"), and writing it into the knowledge base for subsequent use and model fine-tuning.

[0088] Furthermore, embodiments of this application employ an incremental learning algorithm to apply effective strategies ( The data is encoded and written into the knowledge base and fine-tuning dataset (e.g., "#330MW_high ash coal_milling combination D"), and a fine-tuning iteration is performed on the preset time series large model every quarter.

[0089] Thus, through the aforementioned closed-loop optimization and knowledge accumulation mechanism, a closed loop of the entire process of "strategy generation - execution - feedback - iteration" is achieved, with more than 500 new structured cases added each year, and the strategy adoption rate continues to improve.

[0090] The following describes the generator set operation index optimization method of this application, which involves a dynamic optimization and safety control system for generator set operation index based on a time-series large model.

[0091] like Figure 2 As shown, the generator set operation index dynamic optimization and safety control system based on a time-series large model includes a dynamic prediction module, a deviation identification and attribution module, a strategy generation and output module, a safety boundary deduction module, and a closed-loop optimization and knowledge accumulation module.

[0092] The system comprises the following modules: a dynamic prediction module for real-time data collection of unit operation data, and a time-series large-scale model for minute-level predictions (within 15 minutes) based on multi-source operation data; a deviation identification and attribution module for identifying operational deviations, generating deviation reports by comparing them with historical best values, and performing attribution analysis based on an improved SHAP algorithm; a strategy generation and output module for generating structured strategy recommendations by combining an expert knowledge base and a large language model, and outputting structured optimization recommendations; a safety boundary deduction module for dynamically assessing safety boundaries through multi-dimensional risk indices and outputting control recommendations; and a closed-loop optimization and knowledge accumulation module for verifying execution effectiveness and accumulating effective strategies through an effect verification mechanism, encoding effective strategies into structured knowledge, and continuously optimizing the model and decision-making capabilities.

[0093] Furthermore, the generator set operation index dynamic optimization and safety control system of this application embodiment can be mounted on an electronic device to display a comparison chart of the predicted curve and the real-time operation curve on the business platform, and output a PDF / Word version operation guidance report, including a checklist of safety checks.

[0094] To make the objectives, technical solutions, and advantages of this invention clearer, the following is combined with... Figure 3 and Figure 4 The present invention will be further described in detail with reference to embodiments.

[0095] Example 1: Steady-state coal consumption optimization scenario for 300MW load. The specific process is as follows, combined with... Figure 3 Please provide an explanation.

[0096] (I) Deviation Identification and Context Analysis

[0097] (1) Deviation identification.

[0098] The optimization model detected that the current coal consumption of the 300MW load segment is 305g / kWh, which is 15g / kWh higher than the historical best benchmark value (290g / kWh), triggering an alarm in the main system.

[0099] (2) Context analysis.

[0100] Operating parameters: Stable operation at a load of 300MW, using a combination of coal mills 1 / 3 / 4, air-to-coal ratio of 2.8:1, and hot air damper opening of 60%;

[0101] Coal characteristics: Blended with 30% high-sulfur coal (calorific value 20.5 MJ / kg, sulfur content 1.8%).

[0102] Equipment status: A recent drift of ±3℃ was detected in the outlet temperature sensor of coal mill No. 2 (non-closed-loop defect);

[0103] Environmental factors: Ambient temperature 32℃, boiler back pressure +100Pa;

[0104] Operational interference: No maintenance work is scheduled.

[0105] (II) Intelligent Decomposition and Attribution Analysis of Influencing Factors

[0106] By calling the expert knowledge graph and rule set, a multi-dimensional attribution chain and credibility are generated, as shown in Table 1.

[0107] Table 1

[0108]

[0109] (III) Strategy Recommendation Generation

[0110] Based on the attribution result matching optimization strategy, specific suggestions are generated, as shown in Table 2.

[0111] Table 2

[0112]

[0113] (iv) Strategy Evaluation and Structured Output

[0114] The strategy is evaluated in three dimensions and an executable list is generated. Finally, a structured operation guidance report is output, as shown in Table 3.

[0115] Table 3

[0116]

[0117] Example 2: Deep adjustment of a 600MW unit to a 140MW safety optimization scenario. The specific process is as follows, combined with... Figure 4 Please provide an explanation.

[0118] (I) Multi-source data fusion and dynamic assessment of security boundaries

[0119] (1) Defect and fault early warning analysis.

[0120] Retrieving the ERP defect database revealed a non-closed-loop Class III defect: "High temperature alarm for induced draft fan bearing (code #DF_202405_07)".

[0121] Risk classification: Determined to be a critical risk point for deep adjustment (affecting the stability of wind turbines when the load factor is ≤40%).

[0122] Linked early warning: Load is prohibited from falling below 160MW (triggering the defect classification mechanism).

[0123] (2) Dynamic association of work risk tickets.

[0124] Access to the work permit system: There is currently a high-risk operation (hot work + confined space) involving "catalyst replacement in the denitrification system".

[0125] Conflict detection: Deep tuning leading to a drop in flue gas temperature will prolong catalyst activation time;

[0126] Alternative output options: Prioritize completing the denitrification operation before implementing the deep adjustment; if simultaneous operation is necessary, limit the load to ≥180MW (maintain flue gas temperature >300℃).

[0127] (3) Historical accident review and scenario reasoning.

[0128] Case database matching: On November 2, 2023, during a deep adjustment of the same 140MW unit, the vibration of the induced draft fan suddenly increased to 7.8mm / s, causing the bearing to seize.

[0129] Similarity analysis (92%): Same coal type (low volatile matter lean coal); same ambient temperature (-5℃);

[0130] Risk management recommendations: Increase the minimum safe load to 160MW; start the booster fan to assist in induced draft (a historically successful measure).

[0131] (4) Coal type matching and combustion boundary analysis, as shown in Table 4.

[0132] Table 4

[0133]

[0134] (5) Dynamic safety boundary integrated output.

[0135] Due to current limitations, the minimum safe load should be controlled to be greater than 160MW;

[0136] Based on historical case warning values, the load change rate should be controlled to ≤2.0MW / min;

[0137] The current combustion mode is as follows: the micro-oil stable combustion system is activated, and the oxygen content is controlled within the range of 3.8-4.2%.

[0138] (ii) Standardized Operation Guide Generation Module

[0139] (1) Task background and command information.

[0140] Dispatch instructions: From 22:00 to 06:00, adjust the power to 140MW;

[0141] Actual achievable target: 160MW (safety boundary limit);

[0142] Main adjustment directions: ensure stable combustion + prevent excessive equipment vibration.

[0143] (2) Pre-peak preparation items and safety checklist.

[0144] According to the pre-deep-tunnel safety checklist, the following conditions were confirmed: the temperature of the induced draft fan bearing is <65℃, the ignition test of the micro-oil stable combustion system has been completed, and the proportion of high-calorific-value coal blended is ≥30%.

[0145] (3) The combination of operating procedures and risk warnings is shown in Table 5.

[0146] Table 5

[0147]

[0148] (4) Guidance report output and closed loop

[0149] Issue an in-depth investigation operation guidance report (following the PDF / Word output sample), the main contents of which may include:

[0150] 1) Execution results: The actual deep adjustment reached 155MW (safety margin + 5MW); the vibration peak was 5.8mm / s (< warning threshold); coal consumption decreased by 1.2g / kWh compared with the conventional deep adjustment;

[0151] 2) Forming a strategy tag: #Deep Adjustment 160MW_Low Volatile Coal_Defect Association 07;

[0152] 3) Add to the case library: "Deduction of safe load boundary under induced draft fan bearing defects".

[0153] Therefore, this application uses a time-series large-scale model to lead the entire process of accurate prediction, deviation attribution, and safety simulation. By integrating an expert knowledge base and a large language model, it generates structured strategy recommendations that are both interpretable and operable. A complete closed-loop optimization mechanism has been established, realizing a closed loop throughout the entire process from strategy generation, on-site execution, effect feedback to model iteration. At the same time, a dynamic risk field model has been constructed. By integrating multi-dimensional environmental data such as gas concentration, noise, and temperature, a DRF risk index model has been formed to achieve quantitative assessment of safety boundaries. Finally, it achieves multi-scenario adaptability and can effectively cover the optimization and safety control needs under complex operating conditions such as steady-state operation, deep peak shaving, and peak load.

[0154] The intelligent optimization method for multi-scenario operation indicators of generator sets, proposed in this application, integrates time-series large-scale model prediction, large-language model decision-making, and expert knowledge base accumulation. It inputs multi-source operation data of the current generator set into a preset time-series prediction model to predict operation parameters. The prediction results are compared with the historical best operation parameters under the corresponding operating conditions to generate an operation deviation report. Based on a preset SHAP algorithm, attribution analysis is performed on the operation deviation report. Based on the attribution analysis results, optimization suggestions for operation indicators are generated, and the current generator set operation indicators are optimized to obtain the optimized operation indicators. This solves the problems of poor economic efficiency and low safety of generator set operation caused by the lack of dynamic prediction, interpretable strategy generation, safety boundary assessment, and experience-based closed-loop mechanisms in existing power generation technologies. It achieves dynamic intelligent optimization and safe control of generator set operation indicators, significantly improving operational economy and safety.

[0155] Next, referring to the accompanying drawings, an optimization device for generator set operating indicators proposed according to an embodiment of this application is described.

[0156] Figure 5 This is a block diagram of a generator set operation index optimization device according to an embodiment of this application.

[0157] like Figure 5 As shown, the generator set operation index optimization device 10 includes: a prediction module 100, an analysis module 200, and an optimization module 300.

[0158] The prediction module 100 is used to acquire multi-source operating data of the current generator set and input the multi-source operating data into a preset time-series prediction model to predict operating parameters and obtain prediction results; the analysis module 200 is used to compare the prediction results with the historical best operating parameters under the corresponding operating conditions, generate an operating deviation report, and perform attribution analysis on the operating deviation report based on a preset SHAP algorithm to obtain attribution analysis results; the optimization module 300 is used to generate operating index optimization suggestions based on the attribution analysis results, and optimize the current generator set operating indexes based on the operating index optimization suggestions to obtain optimized operating indexes.

[0159] Furthermore, in some embodiments, after optimizing the current generator set's operating indicators according to the operating indicator optimization suggestions to obtain optimized operating indicators, the optimization module 300 is also used to: control the current generator set to operate based on the optimized operating indicators, and select target effective strategies based on the operating results, and use the target effective strategies to iteratively update the preset time series prediction model.

[0160] Furthermore, in some embodiments, the optimization module 300 is also used to: determine the actual coal consumption reduction value and the estimated coal consumption reduction value based on the operation results; calculate the deviation rate between the actual coal consumption reduction value and the estimated coal consumption reduction value; and mark the operation indicator optimization suggestions corresponding to the operation results as the target effective strategy if the deviation rate is less than a preset deviation threshold.

[0161] Furthermore, in some embodiments, after generating operational indicator optimization suggestions based on the attribution analysis results, the analysis module 200 is also used to: calculate a multidimensional risk index based on the operational indicator optimization suggestions; dynamically assess the safety boundary of the current generator set when implementing the operational indicator optimization suggestions based on the multidimensional risk index, and generate control suggestions based on the safety boundary.

[0162] Furthermore, in some embodiments, the multidimensional risk index is:

[0163] ;

[0164] in, As a multidimensional risk index, This is the real-time temperature value. To preset a safe temperature threshold, Weighted sum of multiple preset gas concentrations, These are the operating parameters of the target device.

[0165] Furthermore, in some embodiments, the preset SHAP algorithm is:

[0166] ;

[0167] in, For the first The SHAP value of each feature, For feature set, For the total number of features, This is the output function for the model.

[0168] Furthermore, in some embodiments, the multi-source operating data includes at least one of the following: the grid's daily load curve, real-time operating parameters, fuel parameters, and pulverizing system combination methods; and the operating indicator optimization suggestions include at least one of the following: type labels, specific actions, economic estimates, and risk warnings.

[0169] It should be noted that the explanation of the aforementioned method embodiment for optimizing generator set operating indicators also applies to the generator set operating indicator optimization device 10 of this embodiment, and will not be repeated here.

[0170] According to the generator set operation index optimization device 10 proposed in this application embodiment, the multi-source operation data of the current generator set is input into a preset time-series prediction model to predict the operation parameters. The prediction results are compared with the historical best operation parameters under the corresponding operating conditions to generate an operation deviation report. Based on a preset SHAP algorithm, the operation deviation report is subjected to attribution analysis. Based on the attribution analysis results, operation index optimization suggestions are generated, and the current generator set operation index is optimized to obtain the optimized operation index. This solves the problems of poor generator set operation economy and low safety caused by the lack of dynamic prediction, interpretable strategy generation, safety boundary assessment, and experience closed-loop mechanism in existing power generation technologies. It realizes dynamic intelligent optimization and safe control of generator set operation index, significantly improving operation economy and safety.

[0171] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0172] The memory 601, the processor 602, and the computer program stored on the memory 601 and capable of running on the processor 602.

[0173] When the processor 602 executes the program, it implements the method for optimizing the generator set operating indicators provided in the above embodiments.

[0174] Furthermore, electronic devices also include:

[0175] Communication interface 603 is used for communication between memory 601 and processor 602.

[0176] The memory 601 is used to store computer programs that can run on the processor 602.

[0177] The memory 601 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0178] If the memory 601, processor 602, and communication interface 603 are implemented independently, then the communication interface 603, memory 601, and processor 602 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0179] Optionally, in a specific implementation, if the memory 601, processor 602, and communication interface 603 are integrated on a single chip, then the memory 601, processor 602, and communication interface 603 can communicate with each other through an internal interface.

[0180] The processor 602 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0181] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-mentioned method for optimizing generator set operating indicators.

[0182] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0183] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0184] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0185] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0186] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0187] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0188] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0189] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method of optimizing an operating index of a generator set, characterized by, The method comprises the following steps: obtaining multi-source operation data of a current generator set and inputting the multi-source operation data into a preset time series prediction model to predict operation parameters and obtain a prediction result; comparing the prediction result with historical optimal operation parameters under a corresponding working condition, generating an operation deviation report, and performing attribution analysis on the operation deviation report based on a preset SHAP algorithm to obtain an attribution analysis result; generating an operation index optimization suggestion based on the attribution analysis result, and optimizing the operation index of the current generator set based on the operation index optimization suggestion to obtain an optimized operation index, wherein the operation index optimization suggestion comprises at least one of a type label, a specific action, an economic estimated value, and a risk prompt; wherein the preset time series prediction model adopts a Transformer architecture, and a LoRA low-rank matrix with a rank r=8 is set, original Transformer parameters are fixed during a fine-tuning process, and only low-rank parameters of attention layers and feedforward network layers are updated; after generating the operation index optimization suggestion based on the attribution analysis result, further comprising: calculating a multi-dimensional risk index based on the operation index optimization suggestion; dynamically evaluating a safety boundary of the current generator set when executing the operation index optimization suggestion based on the multi-dimensional risk index, and generating a regulation and control suggestion based on the safety boundary; the multi-dimensional risk index is: ； wherein, is a multi-dimensional risk index, is a real-time temperature value, is a preset safety temperature threshold value, is a plurality of preset gas concentration weightings, is an operating parameter of the target device.

2. The method of claim 1, wherein, after optimizing the operation index of the current generator set based on the operation index optimization suggestion to obtain the optimized operation index, further comprising: controlling the current generator set to operate based on the optimized operation index, and screening a target effective strategy based on an operation result, and iteratively updating the preset time series prediction model based on the target effective strategy.

3. The method of claim 2, wherein, the control of the current generator set based on the optimized operation index and the screening of the target effective strategy based on the operation result comprises: determining an actual coal consumption reduction value and a coal consumption reduction estimated value based on the operation result; calculating a deviation rate between the actual coal consumption reduction value and the coal consumption reduction estimated value; in a case where the deviation rate is less than a preset deviation threshold, marking the operation index optimization suggestion corresponding to the operation result as the target effective strategy.

4. The method of claim 1, wherein, the preset SHAP algorithm is: ； where, SHAP values for the th feature, set of features, total number of features, model output function.

5. The method according to any one of claims 1-4, characterized in that, the multi-source operation data comprises at least one of a power grid daily load curve, real-time operation parameters, fuel parameters, and a combination mode of a pulverizing system.

6. An apparatus for optimizing an operating index of a generator set, characterized by comprising: a generator set operating index optimization device according to any one of claims 1 to 5. comprises: a prediction module configured to obtain multi-source operation data of a current generator set and input the multi-source operation data into a preset time series prediction model to predict operation parameters and obtain a prediction result; an analysis module configured to compare the prediction result with historical optimal operation parameters under a corresponding working condition, generate an operation deviation report, and perform attribution analysis on the operation deviation report based on a preset SHAP algorithm to obtain an attribution analysis result; and An optimization module is configured to generate an operation index optimization suggestion based on the attribution analysis result, and optimize the current operation index of the power generator set based on the operation index optimization suggestion to obtain an optimized operation index, wherein the operation index optimization suggestion comprises at least one of a type label, a specific action, an economic estimation value, and a risk prompt; The preset time sequence prediction model adopts a Transformer architecture, and a LoRA low-rank matrix with a rank r=8 is set respectively, original Transformer parameters are fixed in a fine-tuning process, and only low-rank parameters of an attention layer and a feedforward network layer are updated; After the operation index optimization suggestion is generated based on the attribution analysis result, the optimization module is further configured to calculate a multidimensional risk index based on the operation index optimization suggestion, dynamically evaluate a safety boundary of the current power generator set when the operation index optimization suggestion is executed based on the multidimensional risk index, and generate a regulation and control suggestion based on the safety boundary. The multidimensional risk index is: ； wherein, is a multi-dimensional risk index, is a real-time temperature value, is a preset safety temperature threshold value, is a plurality of preset gas concentration weightings, is an operating parameter of the target device.

7. An electronic device, comprising: The computer program is executed by the processor to implement the optimization method of the operation index of the power generator set. The computer program is executed by the processor to implement the optimization method of the operation index of the power generator set.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, ​

Images