A method and device for smooth optimization distribution of multi-unit thermal load in a thermal power plant

Abstract

The application discloses a kind of thermal power plant multi-unit thermal load smooth optimization distribution method and device, belong to computer software and computer application field.By constructing the optimization model including time sequence dynamic constraint, limit the change rate of electric, thermal load in adjacent dispatching period of unit;Set the highest proportion of main unit in total load, and embed it as strategy constraint into model;Based on the model, the electric load and heat supply steam flow of multi-unit are jointly optimized using intelligent optimization algorithm, the total load balance, dynamic climbing and main unit proportion constraint are satisfied simultaneously in iteration;Finally, the optimal load distribution scheme meeting the requirement of dynamic stability is output, the operation economy of whole plant and the long-term reliability of equipment are optimized by suppressing short-term drastic fluctuation and frequent switching of load.The application effectively improves the stability and overall operation efficiency of load distribution.

Description

Technical Field

[0001] This invention relates to the field of computer software and computer applications, and in particular to a method, apparatus, equipment and storage medium for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant. Background Technology

[0002] With increasingly severe load fluctuations in the power system and the continuous expansion of new energy grid connection, the task of thermal power units participating in grid peak shaving and maintaining real-time power balance is becoming increasingly arduous. Against this backdrop, the heat and power load allocation scheme for dual units within a thermal power plant directly affects the overall economic efficiency of the plant operation, the speed of response to dispatch instructions, and the long-term reliability of equipment operation. Currently, heat and power load optimization allocation methods mainly revolve around the static economic objective under a given total load, aiming to determine the optimal combination of electrical and thermal output for each unit through optimization algorithms. Traditional heat and power load allocation methods mainly include the constant incremental rate method, linear programming method, and optimization methods based on intelligent algorithms (such as genetic algorithms and particle swarm optimization). These methods share the common characteristic of using the total heat and power load demand at a fixed time segment as boundary conditions, and solving for the single objective of minimizing the plant's operating cost or coal consumption. Existing research shows that unit coal consumption characteristics, power generation efficiency, and the thermoelectric coupling relationship are key factors affecting the optimization results. Therefore, effectively modeling the thermoelectric coupling characteristics of the units becomes the core of improving the accuracy of static economic optimization.

[0003] However, while current load allocation methods based on static operating conditions can guarantee optimal economic efficiency at specific operating points, they generally neglect the dynamic correlation and constraints between the operating conditions of the units at different time points. For example, conventional optimization models do not incorporate the actual ramp-up capability of the units, i.e., the maximum allowable rate of change of load over time, as a necessary constraint. This can lead to optimization results indicating drastic load fluctuations in adjacent scheduling periods, exceeding the actual allowable load change rates of the unit's boilers, turbines, and other main and auxiliary equipment. Furthermore, existing methods typically place two units in a completely equal competitive position, lacking a strategic distinction between the main operating units and auxiliary regulating units. This results in frequent switching of load allocation commands between units, which not only exacerbates mechanical stress and wear on equipment but also makes strict execution difficult due to ramp-up rate limitations during actual operation, ultimately affecting the overall controllability and operational stability of the power plant. Therefore, the existing technology has obvious shortcomings: First, the optimization model lacks time-series dynamic constraints, resulting in the solution lacking stability and engineering feasibility in the time dimension; Second, the optimization strategy lacks strategic specification of the operating role, resulting in a contradiction between economic goals and the long-term operational stability of the equipment. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the related art.

[0005] To address this, the present invention proposes a method for the stable optimization allocation of thermal and electrical loads of multiple units in a thermal power plant. By constructing an optimization model that includes time-series dynamic constraints and strategic load ratio constraints, and using an intelligent optimization algorithm based on this model to jointly optimize the electrical and thermal loads, the optimal allocation scheme that can suppress short-term load fluctuations can be output under the premise of satisfying multiple operational constraints, thereby achieving a synergistic improvement in the overall plant's operational economy and the long-term reliability of equipment.

[0006] Another objective of this invention is to provide a device for the stable and optimized distribution of thermal and electrical loads of multiple units in a thermal power plant.

[0007] The third objective of this invention is to provide a computer device.

[0008] A fourth objective of this invention is to provide a non-transitory computer-readable storage medium.

[0009] To achieve the above objectives, this invention proposes a method for the stable and optimized allocation of thermal and electrical loads across multiple units in a thermal power plant, comprising: S1. Construct an optimization model that includes time-series dynamic constraints. The time-series dynamic constraints are used to limit the rate of change of the electrical load and heating steam extraction flow of each unit in adjacent scheduling periods to not exceed the preset ramp-up capacity threshold. S2, set the highest load percentage of the main generating unit with better economic efficiency in the total load, and embed the control logic of the highest load percentage as a strategic constraint into the optimization model; S3, based on the optimization model, uses an intelligent optimization algorithm to jointly optimize the electrical load and heating steam extraction flow of multiple units. The intelligent optimization algorithm simultaneously satisfies the total load balance constraint, dynamic ramp constraint and main unit strategy constraint during the iteration process. S4 outputs the optimal load allocation scheme that meets the requirements of dynamic stability. The optimal load allocation scheme achieves synergistic optimization of the plant's overall operating economy and the long-term reliability of equipment by limiting the short-term drastic fluctuations and frequent switching of unit loads.

[0010] The method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant according to an embodiment of the present invention may also have the following additional technical features: In one embodiment of the present invention, constructing an optimization model including time-series dynamic constraints includes: S11, Set dynamic ramping constraints: For each unit, the ratio of the change in electrical load to the maximum electrical load in adjacent scheduling periods shall not exceed the preset electrical load ramping capacity threshold; the ratio of the change in heating steam extraction flow rate to the maximum heating steam extraction flow rate in adjacent scheduling periods shall not exceed the preset heat load ramping capacity threshold; wherein, the absolute values ​​of the change in electrical load and the change in heating steam extraction flow rate are used in the calculation, and the electrical load ramping capacity threshold and the heat load ramping capacity threshold are upper limits preset according to the actual operating characteristics of the units; S12, the threshold values ​​for electrical load ramping capability and thermal load ramping capability are dynamically adjusted based on the real-time mechanical stress characteristics of the main and auxiliary equipment of the unit. The threshold parameters are corrected online through preset correction coefficients to adapt to the differentiated limitation requirements on the rate of change of unit load under different operating conditions.

[0011] In one embodiment of the present invention, setting the highest load percentage of the main generating unit with better economic efficiency in the total load includes: S21, Load Ratio Control Logic: The electrical load value of the main generating unit during the scheduling period shall not exceed the product of the total electrical load value of the entire plant in the same period and the preset load ratio upper limit parameter; the preset load ratio upper limit parameter is a proportional coefficient pre-set according to the economic advantages of the main generating unit, and the corresponding value range is 0.8 to 0.95; S22, the value of the load ratio upper limit parameter α is determined based on the relative economy between the main unit and the other units: compare the slope of the heat consumption characteristic curve of the main unit at the current operating point with the slope of the heat consumption characteristic curve of the second most economical unit at the corresponding operating point; if the slope of the curve of the main unit is less than the slope of the curve of the second most economical unit, then α is set to a first preset value; otherwise, α is set to a second preset value; wherein, the first preset value is 0.85 and the second preset value is 0.9.

[0012] In one embodiment of the present invention, an intelligent optimization algorithm is used to jointly optimize the electrical load and heating steam extraction flow of multiple units, including: S31, the population of the intelligent optimization algorithm is initialized using a matrix structure; each individual corresponds to a unit load allocation scheme; for the k-th individual in the population, the corresponding initial electrical load and initial thermal load of the unit are generated by arithmetically dividing the total load demand value and superimposing random disturbance components to ensure that the initial solution has sufficient dispersion in the preset solution space under the premise of satisfying the total load balance constraint. S32, the selection operation uses a formulaic roulette wheel selection method, where the probability of an individual being selected is... Its fitness value The relationship is: ; in, Population size.

[0013] In one embodiment of the present invention, it further includes: S5. Based on the real-time load fluctuation of the power grid, dynamically adjust the weight coefficient of the penalty term related to the load change rate in the optimization model: establish an adaptive correction mechanism for the penalty coefficient, based on the benchmark penalty coefficient, combined with the ratio of the fluctuation amplitude of the total load of the power grid to the corresponding maximum load in the current scheduling period, and adjust it through preset correlation parameters, and calculate the real-time penalty coefficient for this period according to the power function relationship; wherein, the larger the fluctuation amplitude of the power grid load, the larger the calculated real-time penalty coefficient will be, so as to strengthen the suppression of drastic changes in unit load.

[0014] To achieve the above objectives, another aspect of the present invention proposes a device for the stable and optimized distribution of thermal and electrical loads across multiple units in a thermal power plant, comprising: The time-series dynamic constraint modeling module is used to construct an optimization model that includes time-series dynamic constraints. The time-series dynamic constraints are used to limit the rate of change of the electrical load and heating steam extraction flow of each unit in adjacent scheduling periods to not exceed the preset ramp-up capacity threshold. The main unit strategy constraint module is used to set the highest load percentage of the main unit with better economic efficiency in the total load, and embed the control logic of the highest load percentage as a strategy constraint into the optimization model. The joint optimization execution module is used to jointly optimize the electrical load and heating steam extraction flow of multiple units based on the optimization model and using intelligent optimization algorithms. The intelligent optimization algorithm simultaneously satisfies the total load balance constraint, dynamic ramp constraint and main unit strategy constraint during the iteration process. The dynamic stability output module is used to output the optimal load allocation scheme that meets the dynamic stability requirements. The optimal load allocation scheme achieves synergistic optimization of the plant's overall operating economy and the long-term reliability of equipment by limiting the short-term drastic fluctuations and frequent switching of unit load.

[0015] In one embodiment of the present invention, it further includes: The penalty factor dynamic adjustment module is used to dynamically adjust the weight coefficient of the penalty term related to the load change rate in the optimization model according to the real-time load fluctuation of the power grid. An adaptive correction mechanism for the penalty coefficient is established. Based on the benchmark penalty coefficient, the ratio of the fluctuation range of the total load of the power grid to the corresponding maximum load in the current scheduling period is combined and adjusted through preset correlation parameters. The real-time penalty coefficient for this period is calculated according to the power function relationship. The larger the fluctuation range of the power grid load, the larger the calculated real-time penalty coefficient will be, so as to strengthen the suppression of drastic changes in unit load.

[0016] This invention provides a method and apparatus for the stable optimization allocation of thermal and electrical loads across multiple units in a thermal power plant. By embedding time-series dynamic constraints and strategic constraints of the main generating units into the optimization model, and employing intelligent optimization algorithms for the coordinated optimization of electrical and thermal loads, it effectively solves the core problems of traditional load allocation methods, such as the tendency for short-term drastic fluctuations in unit loads and the difficulty in balancing economic objectives with the requirements for stable equipment operation. It achieves automated optimization under multiple complex constraints, including overall plant load balance, unit ramp-up capabilities, and strategic operation, significantly improving the dynamic stability of load allocation, overall operational economy, and long-term equipment reliability. This enhances the adaptive optimization capabilities and comprehensive operational benefits of thermal power plants under complex grid dispatching demands.

[0017] To achieve the above objectives, a third aspect of this application provides a computer device, including a processor and a memory; wherein the processor reads executable program code stored in the memory to run a program corresponding to the executable program code, for implementing a method for the smooth and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant as described in the first aspect embodiment.

[0018] To achieve the above objectives, the fourth aspect of this application proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant as described in the first aspect embodiment.

[0019] Additional aspects and advantages of the invention 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 the invention. Attached Figure Description

[0020] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant according to an embodiment of the present invention; Figure 2 This is a flowchart of a genetic algorithm-based optimization method for the stable optimization allocation of thermal and electrical loads of multiple units in a thermal power plant, according to an embodiment of the present invention. Figure 3 This is a heat consumption curve of Unit 1, which is a method for the stable and optimized allocation of heat and power load of multiple units in a thermal power plant according to an embodiment of the present invention. Figure 4 This is a heat consumption curve of Unit 2 of a method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant according to an embodiment of the present invention. Figure 5This is a diagram illustrating the optimized distribution effect of thermal power load according to an embodiment of the present invention, which describes a method for the stable and optimized distribution of thermal power load across multiple units in a thermal power plant. Figure 6 This is a schematic diagram of the structure of a device for the stable and optimized distribution of thermal and electrical loads of multiple units in a thermal power plant according to an embodiment of the present invention; Figure 7 It is a computer device according to an embodiment of the present invention. Detailed Implementation

[0021] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] The following description, with reference to the accompanying drawings, describes a method, apparatus, equipment, and storage medium for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant, according to an embodiment of the present invention.

[0024] The core idea of ​​this invention is to construct an optimization model system that integrates time-series dynamic constraints and main unit strategy constraints. This system incorporates multiple factors, such as unit dynamic ramp-up capability, real-time operating status, and relative economy, into a unified optimization framework, achieving synergistic quantification of load allocation economy and stability. Based on this model, intelligent optimization algorithms are used to jointly optimize electrical and thermal loads. This adaptively satisfies multiple complex constraints, including total load balance, dynamic ramp-up, and strategic guidance. Thus, the traditional load allocation process, centered on static economy, is transformed into an intelligent decision-making closed loop that deeply perceives unit dynamic characteristics, strategically guides load distribution, and continuously iterates and optimizes. Ultimately, while outputting the optimal solution, it achieves synergistic optimization of plant-wide operational economy and long-term equipment reliability.

[0025] Example 1 To achieve the above invention, embodiments of the present invention provide a method for the stable and optimized allocation of thermal and electrical loads across multiple units in a thermal power plant, such as... Figure 1 As shown, it includes: S1. Construct an optimization model that includes time-series dynamic constraints. The time-series dynamic constraints are used to limit the rate of change of the electrical load and heating steam extraction flow of each unit in adjacent scheduling periods to not exceed the preset ramp-up capacity threshold.

[0026] Specifically, the constraints in this step are a key element in achieving the dynamic stability of the optimization scheme, and have significant engineering implications, especially in scenarios involving the coordinated operation of multiple units.

[0027] Specifically, the time-series dynamic constraints introduce upper and lower limits on the load change rate between adjacent scheduling periods into the optimization model, in the following form: ; in, and They represent the first Taiwanese unit in Electrical load and heating steam extraction flow rate for each scheduling period and The maximum allowable ramp rate for the unit in terms of electrical load and heating steam extraction flow is typically determined by the dynamic response capabilities of key equipment such as the turbine and boiler, and is measured in MW / min and t / h / min, respectively. These parameters need to be calibrated based on the technical specifications or historical operating data provided by the unit manufacturer to ensure that the model matches the actual operating capabilities.

[0028] Furthermore, this constraint is embedded as a hard constraint in the optimization process of the genetic algorithm, meaning that in each generation of the population, only individuals that satisfy the aforementioned rate of change constraint are retained. In this way, the optimization process can avoid generating load allocation schemes that are discontinuous and unexecutable in the time dimension, thereby improving the executability of scheduling instructions and the stability of equipment operation.

[0029] Specifically, this step applies to the coordinated scheduling of multiple units in thermal power plants when participating in grid peak shaving and responding to sudden load changes. For example, in scenarios where grid load decreases or increases rapidly, without setting timing constraints, the optimization results may lead to a unit significantly reducing or increasing its load in a short period, exceeding its ramp-up capacity, causing equipment stress concentration or response delay. After introducing this constraint, the optimization model will automatically adjust the load allocation strategy, ensuring that the units meet the total load demand while keeping their output changes within the allowable range of the equipment.

[0030] Specifically, by incorporating dynamic response capabilities into the optimization model, drastic fluctuations in unit load were effectively suppressed, improving the temporal continuity of dispatch instructions and the safety of equipment operation. Simultaneously, by combining the strategic constraints of the main generating units, the economic efficiency and stability of the optimization scheme in long-term operation were further enhanced, providing solid technical support for thermal power plants to achieve efficient and safe cogeneration dispatch in the new power system.

[0031] Furthermore, S1 includes: S11, Set dynamic ramping constraints: For each unit, the ratio of the change in electrical load to the maximum electrical load in adjacent scheduling periods shall not exceed the preset electrical load ramping capacity threshold; the ratio of the change in heating steam extraction flow rate to the maximum heating steam extraction flow rate in adjacent scheduling periods shall not exceed the preset heat load ramping capacity threshold; wherein, the absolute values ​​of the change in electrical load and the change in heating steam extraction flow rate are used in the calculation, and the electrical load ramping capacity threshold and the heat load ramping capacity threshold are upper limits preset according to the actual operating characteristics of the units.

[0032] Specifically, this constraint is expressed in the form of time-differentiation, and is specifically applied to the electrical load of each unit. and heat load During adjacent scheduling periods and The rate of change of the variation between the two values ​​should satisfy the following constraint: ; in, and They represent the generating units. The maximum ramp rate in both electrical and thermal load directions is measured in MW / min or t / h / min, with specific values ​​set according to the load change rate limits of key equipment such as the turbine and boiler of the unit. For example, for a typical 300MW thermal power unit, its electrical load ramp rate... Usually not exceeding The ramp rate of the heating steam extraction flow rate This is set according to the dynamic response characteristics of the steam extraction system, generally in... Within the range.

[0033] Furthermore, this constraint is embedded into the optimization model of the genetic algorithm, serving as a hard constraint for each unit in each scheduling period. During the crossover and mutation operations of the genetic algorithm, if the generated offspring violate this climbing constraint, their fitness value will be significantly reduced or they will be directly eliminated, thus guiding the algorithm to converge towards a solution space that satisfies dynamic stability. This step plays a crucial role in the multi-unit thermal power load allocation in thermal power plants, effectively avoiding problems such as equipment mechanical stress concentration and thermal system imbalance caused by sudden load changes, and improving the response quality and operational safety of scheduling commands. Furthermore, this constraint works synergistically with the "main unit strategy constraint" to ensure that the more economical unit assumes a stabilizing role in load adjustments, thereby achieving the dual optimization goals of economy and stability in dynamic operation.

[0034] S12, the threshold values ​​for electrical load ramping capability and thermal load ramping capability are dynamically adjusted based on the real-time mechanical stress characteristics of the main and auxiliary equipment of the unit. The threshold parameters are corrected online through preset correction coefficients to adapt to the differentiated limitation requirements on the rate of change of unit load under different operating conditions.

[0035] Specifically, the system collects characteristic signals reflecting the mechanical stress state of the main equipment and key auxiliary equipment of the unit in real time through an online monitoring system. Based on a preset stress-safety margin mapping model, the safe upper limit of the load change rate that the equipment can withstand under the current operating conditions is calculated. The preset correction coefficients are essentially one or more sets of scaling factors or bias terms that are strongly correlated with the stress state. The original threshold parameters are checked and updated online in real time and continuously through an embedded control algorithm, thereby directly coupling the physical state of the equipment to the constraint boundary of the optimization model.

[0036] Furthermore, this mechanism manifests as a closed-loop adaptive adjustment process. Operators or the automatic control system can dynamically relax or tighten the restrictions on the unit's load change rate based on the actual health status and operating conditions of the equipment. For example, when the equipment is in a well-cooled state and the stress level is low, the ramp-up threshold can be appropriately increased to enhance the unit's response flexibility; conversely, when local overheating or increased vibration is detected, the threshold is automatically lowered to prioritize equipment safety.

[0037] Furthermore, the "electrical load ramping capability threshold" and "thermal load ramping capability threshold" are defined as the upper limit of the standardized relative rate of change (e.g., percentage change per minute). Their dynamic adjustment is not unbounded, but rather within a feasible range based on the design value and bounded by the real-time bearing capacity of the equipment. The "preset correction coefficient" is a set of parameters that have been safety-assessed and experimentally calibrated to ensure that the adjusted thresholds always remain within the safe allowable range of the equipment's mechanical strength and fatigue life.

[0038] Specifically, this method is particularly suitable for complex operational phases with special requirements for load change rates, such as unit startup, shutdown, large-scale load tracking, recovery from abnormal equipment conditions, and emergency grid frequency regulation. It enables the optimization model to impose differentiated rate limits based on the inherent safety requirements of different scenarios, thereby achieving a fine balance between safety and economy in the control strategy.

[0039] Specifically, the core effect of this technical solution lies in its transformation of static, conservative ramp-up constraints into dynamic, intelligent, and adaptive constraints. This not only effectively avoids the waste of peak-shaving capacity caused by overly strict fixed threshold restrictions, but also fundamentally prevents the risk of mechanical damage induced under weak equipment conditions due to overly lenient threshold settings. By incorporating real-time equipment status feedback into the optimization boundary, the physical feasibility and long-term operational reliability of the plant-wide load distribution scheme are significantly improved during actual implementation.

[0040] S2, set the highest load percentage of the main generating unit with better economic efficiency in the total load, and embed the control logic of the highest load percentage as a strategic constraint into the optimization model.

[0041] Specifically, the core of this step lies in strategically guiding the optimization process so that the more economical units can assume a stable base load in load distribution, thereby reducing frequent switching between units and drastic load fluctuations, and improving the executability of system operation and equipment lifespan.

[0042] Specifically, this step first identifies the most economical main generating units based on the heat consumption characteristic curves of thermal power units. Typically, these units have lower specific coal consumption. This means that under the same electrical and thermal load conditions, its heat consumption value is lower. A maximum load percentage for it in the total load is set; for example, the upper limit for the electrical load percentage of unit 1 is set to [value missing]. This value can be set based on the power plant's actual operating experience, equipment ramp-up capability, and the stability requirements of dispatch instructions. A typical value range is... This constraint can be expressed as: ; in, The total electrical load demand for the current scheduling period. This is the electrical load allocation value for Unit 1. This constraint is embedded in the optimization model of the genetic algorithm and participates as a strategic constraint in the fitness evaluation and constraint satisfaction judgment of individuals.

[0043] Furthermore, The settings need to be combined with the unit's climbing ability. With scheduling cycle This ensures that load changes within adjacent time periods do not exceed the equipment's permissible ramp rate. For example, if the scheduling cycle is 15 minutes and Unit 1's ramp rate is... If the load ratio is adjusted within this range, the adjustment range should be controlled to avoid stress concentration in the equipment due to sudden load changes.

[0044] Specifically, this strategic constraint applies to dynamic load allocation scenarios where thermal power plants participate in grid peak shaving and respond to load fluctuations. By prioritizing the load stability of the main generating units in the optimization model, the frequent start-up and shutdown of auxiliary units and load adjustments can be effectively reduced, thereby reducing equipment wear and improving operating efficiency.

[0045] Specifically, by introducing strategic load ratio constraints, the optimization model can guide the main generating units to undertake basic loads while meeting the total load demand, reducing the uncertainty and dynamic fluctuations of load allocation, significantly improving the stability and executability of the optimization results in actual operation, and providing strong support for the safe and stable operation of thermal power plants under the new power system.

[0046] Furthermore, S2 includes: S21, Load Ratio Control Logic: The electrical load value of the main generating unit during the scheduling period shall not exceed the product of the total electrical load value of the entire plant in the same period and the preset load ratio upper limit parameter; the preset load ratio upper limit parameter is a proportional coefficient pre-set according to the economic advantages of the main generating unit, and the corresponding value range is 0.8 to 0.95.

[0047] Specifically, within each scheduling period, the optimization algorithm must enforce this constraint when solving for the power output allocation of each unit. This is typically defined as a set of hard constraints during the model building phase, ensuring that no feasible solution causes the output value of the main unit to exceed the dynamic upper limit determined by the total plant load and preset upper limit parameters. Mathematically, this logic can be expressed as a direct constraint on the decision variables, and its implementation depends on the optimization solver's ability to identify and process constraints.

[0048] Specifically, when the total load demand changes, the optimization model, in seeking the economically optimal solution, automatically allocates load to each unit according to this constraint, ensuring that the main generating units always bear the core base load within the preset economic operating range, and do not overload due to local economic optimization, thereby reserving the necessary regulation capacity for other peak-shaving units. This process does not require real-time manual intervention, realizing closed-loop execution and dynamic adjustment of the strategy.

[0049] Furthermore, the "preset load percentage upper limit parameter" is a key strategic parameter, with its value limited to between 0.8 and 0.95. This range is set based on statistical analysis of the typical thermodynamic characteristic curves of the main generating units, aiming to ensure that they operate within a highly efficient and flat range where heat or coal consumption rates are insensitive to load changes. The specific value of the parameter, such as 0.85 or 0.9, needs to be finely calibrated through economic analysis of historical operating data to achieve the best balance between leveraging the economic advantages of the main generating units and maintaining the necessary flexibility of the system.

[0050] Furthermore, this control logic is primarily applied to thermal power plants containing multiple heterogeneous units (such as high-efficiency supercritical units and lower-parameter units coexisting). When undertaking base load or participating in grid peak shaving, this logic can effectively prevent large and frequent adjustments in the output of high-efficiency units due to short-term fluctuations in market signals or dispatch instructions. It is particularly suitable for complex scenarios in the electricity market environment where both economic pricing and unit operational stability must be considered simultaneously.

[0051] Specifically, this not only ensures that the plant's overall operating economy tends to be optimal within a single scheduling cycle, but more importantly, by limiting the output fluctuations of the main generating units, it reduces equipment wear and tear, and improves long-term reliability and life-cycle economics. At the same time, it also provides other units with clear and stable operating expectations and reasonable load margins, enhancing the plant's overall coordination and stability in responding to load changes at the system level.

[0052] S22, the value of the load ratio upper limit parameter α is determined based on the relative economy between the main unit and the other units: compare the slope of the heat consumption characteristic curve of the main unit at the current operating point with the slope of the heat consumption characteristic curve of the second most economical unit at the corresponding operating point; if the slope of the curve of the main unit is less than the slope of the curve of the second most economical unit, then α is set to a first preset value; otherwise, α is set to a second preset value; wherein, the first preset value is 0.85 and the second preset value is 0.9.

[0053] Specifically, the core of this step lies in establishing a parameter adaptive adjustment mechanism based on the real-time relative economics among generating units. This mechanism acquires or calculates the first derivative (i.e., slope) of the heat consumption characteristic curves of the main generating unit and other units at the current planned load point online and performs real-time comparisons. The slope directly reflects the heat consumption increment caused by increasing the unit load at this operating point, i.e., the marginal heat consumption rate. The comparison logic is designed as a binary decision-maker: if the marginal heat consumption rate of the main generating unit is lower than that of the second-best economically, it indicates that its economic advantage is significant, and a higher load share upper limit (second preset value 0.9) can be assigned; conversely, it indicates that its economic advantage is narrowing, and a slightly more conservative upper limit (first preset value 0.85) should be adopted to guide the load to other units appropriately to optimize the overall economy. This process is automatically completed by an algorithm integrated into the optimization preprocessing module.

[0054] Specifically, this method transforms static parameter settings into a dynamic process linked to operational status. Before each load allocation optimization, the system recalculates and determines the α value used in the current optimization based on the latest unit performance data and predicted load points. This allows the output ceiling of the main generating units to flexibly respond to the dynamic changes in their own and other units' economics due to factors such as load rate, equipment status, and environmental conditions, achieving precise matching between strategy constraints and current operating conditions.

[0055] Furthermore, the first preset value of 0.85 and the second preset value of 0.9 are key thresholds calibrated through quantitative analysis and engineering experience. The difference between them (0.05) reflects the sensitivity and adjustment strength to the degree of economic difference. This setting ensures the core load-carrying status of the main generating units while reserving necessary flexibility for system adjustment. The discretization of the threshold itself (rather than a continuous function) balances the clarity, robustness, and ease of engineering implementation of the strategy.

[0056] Specifically, when a power plant is in a phase of rapid load increase or decrease, or when a unit's heat consumption characteristic curve shifts due to equipment performance degradation, a fixed upper limit for the proportion may lead to misjudgment of economic efficiency or insufficient regulation capacity. This method can automatically adapt to these changes and is particularly suitable for complex operating scenarios involving deep peak shaving, responding to large fluctuations in renewable energy grid connection, or inconsistent unit health conditions, ensuring that economic strategies remain reasonable and effective under any operating condition.

[0057] Specifically, it introduces a refined decision-making dimension of "marginal comparison" of economic efficiency, upgrading the load allocation control strategy from "experience-based" to "dynamically optimized." This not only improves the overall economic operation level of the entire plant under varying operating conditions and avoids the potential loss of economic benefits due to rigid parameters, but also gently guides load allocation through dynamic constraints, helping to delay the performance degradation of the main generating units under adverse operating conditions. It achieves a more intelligent and adaptive balance between short-term economic efficiency, long-term equipment status, and overall plant operational flexibility.

[0058] S3, based on the optimization model, uses an intelligent optimization algorithm to jointly optimize the electrical load and heating steam extraction flow of multiple units. The intelligent optimization algorithm simultaneously satisfies the total load balance constraint, dynamic ramp constraint and main unit strategy constraint during the iteration process.

[0059] Specifically, the preferred intelligent optimization algorithm is the Genetic Algorithm (GA), which has advantages such as strong global search capability, high robustness, and applicability to nonlinear optimization problems with multiple variables and multiple constraints. It is particularly suitable for handling complex optimization tasks of thermoelectric coupling systems in thermal power plants.

[0060] Specifically, the genetic algorithm first sets an initial solution set using a population initialization matrix, where the initial electrical and thermal load values ​​for each unit are set to the average of the output values ​​of the thermal power plant's operating plan. During the iteration process, the algorithm uses roulette wheel selection to choose individuals, with the selection probability proportional to the fitness value. The fitness function is defined as the total heat consumption... The reciprocal or negative value of the integer part is used to minimize the objective. The crossover operation employs an arithmetic crossover strategy, generating offspring solutions through linear combinations of parent individuals, where... The crossover coefficient typically ranges from [0.5, 0.8] to balance exploration and development capabilities. The mutation operation employs a Gaussian-based mutation operator, with the mutation probability of its probability density function generally set to... Standard deviation of variation It can be adaptively adjusted according to the range of variables to maintain population diversity and prevent premature convergence.

[0061] Furthermore, the algorithm's termination condition is set to reaching the maximum number of generations. Or the fitness value converges to a stable state, usually Values Specifically, this depends on the problem complexity and computational resources. During the optimization process, dynamic ramp constraints limit the rate of change of electrical load and heating steam extraction flow within adjacent scheduling periods, with a maximum permissible rate of change. and The settings are usually based on the ramp-climbing capabilities of key equipment such as the turbine and boiler of the generating unit, for example... Rated load per minute. The main unit strategy constraint sets an upper limit on the load share of the more economical units. ,like This is to ensure that it can bear the basic load and reduce equipment wear caused by frequent adjustments.

[0062] Specifically, this step is applicable to thermal power plants operating under conditions of frequent grid peak shaving and load fluctuations, achieving a stable distribution of electrical and thermal loads through dynamic optimization. For example, when there are sudden changes in grid load or fluctuations in renewable energy output, the algorithm can smooth the unit output curve while maintaining overall load balance, avoiding equipment over-limit operation or response delays caused by sudden load changes.

[0063] Specifically, by introducing dynamic ramp-up constraints and main unit strategy constraints, the drastic fluctuations and frequent switching of unit loads are effectively suppressed, thereby improving the executability and dynamic stability of the optimization scheme in actual operation. Simulation results show that this method has significant advantages in reducing unit coal consumption, with an average reduction in power supply coal consumption reaching [percentage missing]. This provides reliable technical support for the safe, stable, and economical operation of thermal power plants under the new power system.

[0064] Furthermore, S3 includes: S31, the population of the intelligent optimization algorithm is initialized using a matrix structure; each individual corresponds to a unit load allocation scheme; for the k-th individual in the population, the corresponding initial electrical load and initial thermal load of the unit are generated by arithmetically dividing the total load demand value and superimposing random disturbance components to ensure that the initial solution has sufficient dispersion in the preset solution space under the premise of satisfying the total load balance constraint.

[0065] Specifically, the entire population is organized as a two-dimensional matrix, where each row corresponds to an individual (i.e., a candidate load allocation scheme), and each column corresponds to an optimization variable (such as the electrical or thermal load value of a specific unit). For each individual in the population, the initial load of each unit is not generated completely randomly, but follows the principle of "baseline allocation plus random disturbance": First, the total electrical and thermal load demand of the scheduling period is arithmetically divided according to preset rules (such as proportional or capacity-based weights) to form a basic allocation scheme that satisfies the total load balance constraint; then, a random disturbance component conforming to a specific distribution is superimposed on this base value. The amplitude of this disturbance is limited to the feasible output range of the units to ensure that all initial individuals are within the preset solution space, thereby achieving sufficient diversity of the initial population within the solution space while strictly satisfying the core equality constraints.

[0066] Furthermore, this method ensures that the optimization algorithm has a set of high-quality and widely distributed starting points at the beginning of the iterative search. This effectively avoids the large number of invalid solutions (such as solutions that severely violate load balance) that may be generated by completely random initialization, reducing the overhead of repair and elimination in the early stages of the algorithm. By controlling the intensity of random perturbation, a balance can be achieved between the two strategies of "local exploration around a relatively optimal benchmark" and "widespread distribution of points globally," laying a good foundation for subsequent evolutionary operations such as selection, crossover, and mutation.

[0067] Specifically, key parameters include: population size (number of rows in the matrix), total number of optimization variables (number of columns in the matrix), weighting rules for arithmetic partitioning, and the probability distribution type (e.g., uniform distribution, normal distribution) and amplitude range of random perturbation components. These parameters need to be configured jointly based on the specific number of power plant units, load range, and characteristics of the intelligent optimization algorithm used to ensure that the initialization process can efficiently generate feasible solutions while providing sufficient exploration breadth for the algorithm.

[0068] Specifically, this initialization method is particularly suitable for large-scale, multi-constraint complex optimization problems. For example, in optimization scenarios involving multiple thermoelectric coupled units that simultaneously satisfy load balancing on both the electrical and thermal sides, as well as ramp-up constraints, completely random initialization is extremely difficult to generate feasible solutions. This method rapidly constructs a baseline by deterministic partitioning based on total demand, and then applies bounded random perturbations, which can efficiently generate a batch of initial solutions that are both feasible and diverse, significantly improving the startup efficiency and reliability of the algorithm in complex scenarios.

[0069] Specifically, by introducing prior knowledge and structured randomness, it significantly improves the initialization quality of intelligent optimization algorithms for the specific problem of load allocation in thermal power plants. This not only accelerates the initial convergence speed of the algorithm but also effectively reduces the risk of the algorithm getting trapped in local optima by enhancing the dispersion of the initial population. As a result, it improves the overall quality and global optimality of the final load allocation scheme, ensuring the economic and safety benefits of the optimization results.

[0070] S32, the selection operation uses a formulaic roulette wheel selection method, where the probability of an individual being selected is... Its fitness value The relationship is: ; in, Population size.

[0071] Specifically, this method maps fitness values ​​to a probability distribution, giving individuals with higher fitness a higher probability of being selected, thereby preserving superior genes in the population and driving the algorithm toward a better solution.

[0072] In practical implementation, individuals Probability of being selected Defined as: ; in, Indicates the first The fitness value of each individual This represents the total number of individuals in the current population. This formula reflects the "fitness-proportional selection" mechanism of the roulette wheel selection method, that is, the higher the fitness value of each individual, the larger the "sector" area it occupies in the probability distribution, and thus the more likely it is to be selected.

[0073] Specifically, first, the fitness value of all individuals is calculated, and then the selection probability of each individual is calculated. Subsequently, a random number is generated within the interval [0, 1] using a random number generator. This is then compared with a cumulative probability distribution to determine the selected individuals. This process can be repeated multiple times to build a new parent population for crossover operations.

[0074] Furthermore, the roulette wheel selection method plays a crucial role in this invention. Due to the complex characteristics of multi-unit thermal power load allocation in thermal power plants, including multiple variables, multiple constraints, and nonlinearity, the rationality of the selection operation directly affects the algorithm's convergence speed and global search capability. By introducing a fitness-proportional selection mechanism, this invention can prioritize retaining load allocation schemes with higher fitness while satisfying dynamic stability constraints, thereby accelerating algorithm convergence and improving optimization quality.

[0075] Specifically, this method offers good flexibility in parameter settings. The design of the fitness function needs to be closely related to the optimization objective (such as minimizing total heat consumption) and must consider the penalty term of the constraints. In this invention, the fitness value... Usually related to the objective function Inversely proportional, that is This ensures that individuals with lower heat consumption have higher fitness. The calculation process for selection probability follows the standard genetic algorithm procedure, conforms to the definition of selection strategy in the IEEE standard, and has good engineering feasibility and portability.

[0076] S4 outputs the optimal load allocation scheme that meets the requirements of dynamic stability. The optimal load allocation scheme achieves synergistic optimization of the plant's overall operating economy and the long-term reliability of equipment by limiting the short-term drastic fluctuations and frequent switching of unit loads.

[0077] Specifically, this step aims to output the optimal load allocation scheme that meets the requirements of dynamic stability. Its technical implementation principle is based on an optimization strategy combining a genetic algorithm (GA) and multi-dimensional constraint modeling. In the multi-unit operation scenario of a thermal power plant, short-term drastic fluctuations and frequent switching of unit loads can significantly increase the mechanical stress and fatigue damage of equipment, reducing operational reliability. Therefore, this invention introduces dynamic ramping constraints and main unit strategy constraints into the optimization model to achieve synergistic optimization of the plant's overall operational economy and long-term equipment reliability.

[0078] Furthermore, this step uses a genetic algorithm to analyze the unit's electrical load. and heating steam extraction flow rate Perform a global optimization. The optimization objective is to minimize the total heat consumption of the entire plant. The calculation formula is as follows: ; in, and Let be the heat consumption functions of Unit 1 and Unit 2 under specific electrical and thermal loads, respectively. To ensure dynamic stability, a time-series constraint is introduced into the optimization model, which limits the rate of change of the unit's electrical load and heating extraction steam flow rate within adjacent scheduling periods to not exceeding its ramp-up capacity. For example, if a certain unit... The electrical load during the period is In the The time period is Then it must satisfy: ; in, For the unit The maximum permissible ramp rate is typically determined by the operating specifications provided by the equipment manufacturer, with a typical value of 5% of rated load per minute. Furthermore, this can be achieved by designating the most economical unit as the primary unit and limiting its load share to no more than [a certain percentage]. This can avoid frequent switching and improve operational stability.

[0079] Specifically, this step is applicable to scenarios such as thermal power plants participating in grid peak shaving and dynamic load adjustment of combined heat and power (CHP). By introducing dynamic stability constraints, the optimization results significantly reduce the fluctuation amplitude and switching frequency of load commands while meeting total load demand, thereby improving equipment lifespan and dispatch response quality. Simulation verification shows that this method can reduce the average coal consumption for unit power supply by [percentage missing]. It has significant economic and engineering value.

[0080] S5. Based on the real-time load fluctuation of the power grid, dynamically adjust the weight coefficient of the penalty term related to the load change rate in the optimization model: establish an adaptive correction mechanism for the penalty coefficient, based on the benchmark penalty coefficient, combined with the ratio of the fluctuation amplitude of the total load of the power grid to the corresponding maximum load in the current scheduling period, and adjust it through preset correlation parameters, and calculate the real-time penalty coefficient for this period according to the power function relationship; wherein, the larger the fluctuation amplitude of the power grid load, the larger the calculated real-time penalty coefficient will be, so as to strengthen the suppression of drastic changes in unit load.

[0081] Specifically, the core of this step lies in constructing a flexible objective function shaping method that is linked in real time with the macroscopic operating status of the power grid. In practice, load fluctuation information from the power grid dispatching terminal is acquired in real time through a data interface. The absolute value of the total load fluctuation amplitude within the current dispatching period is calculated and normalized to a preset maximum load reference value for that period, resulting in a relative index characterizing the severity of the fluctuation. Using this index as input, a mathematical model with a benchmark penalty coefficient as the starting point, preset correlation parameters as adjustment factors, and a power function as the calculation relationship is used to solve online for the real-time penalty coefficient applicable to the current period, which is then updated to the objective function of the optimization model in real time. This mechanism achieves automated mapping and closed-loop correction of external power grid dynamic characteristics to internal optimization model parameters.

[0082] Furthermore, this mechanism manifests as the optimization model's intelligent adaptability to the power grid operating environment. When the power grid load is stable, the system automatically reduces the penalty weight for the unit's load change rate, allowing the optimization algorithm to pursue economic optimization more freely, and the units can adjust more flexibly. Once a large or frequent fluctuation in the power grid is detected, the system rapidly increases the penalty coefficient, causing the optimization algorithm to favor solutions with gradual load changes during optimization, thus proactively responding to the power grid's stability requirements at the overall plant output level, acting as a "damper."

[0083] Furthermore, key parameters include the baseline penalty coefficient, correlation parameters, and power exponent. The baseline penalty coefficient represents the basic penalty intensity of the power grid under typical steady-state conditions; the correlation parameters determine the sensitivity of variability in converting into penalty increments; the introduction of the power exponent makes the regulation relationship non-linear, enabling a faster increase in suppression after variability exceeds a certain threshold. These parameters need to be jointly tuned based on power grid safe operation procedures, unit regulation performance test data, and historical optimization cases to ensure that the regulation is both effective and not excessive.

[0084] Specifically, this mechanism is particularly effective in addressing the intermittency and volatility caused by a high proportion of renewable energy integration. For example, when a sudden drop in wind and solar power output leads to a power deficit in the grid requiring rapid load increases from thermal power plants, this mechanism can automatically increase the penalty coefficient based on the extent of the deficit. This guides the optimization model to prioritize sharing the load impact through the gentle coordination of multiple generating units, rather than causing drastic changes in a single unit, while meeting frequency regulation requirements. This ensures both grid safety and the operational stability of the power plant's internal equipment.

[0085] Specifically, by quantifying and internalizing the real-time volatility of the power grid as an adjustment factor for the optimization objective, a dynamic unification between the power plant's internal economic optimization objectives and the external power grid's stable operation requirements is achieved. This not only enhances the proactive stability and coordination of the plant's output response to grid dispatch, but also reduces equipment wear and risks that may be caused by the transmission of large-scale load fluctuations within the plant from a global optimization perspective, thereby improving the safety support value and overall operational resilience of thermal power plants as an important flexible resource in complex power grid environments.

[0086] This invention provides a method for the stable optimization and allocation of thermal and electrical loads across multiple units in a thermal power plant. By embedding time-series dynamic constraints, main unit strategy constraints, and an adaptive penalty mechanism into the optimization model, and employing intelligent algorithms for collaborative optimization, this method effectively solves the core problems of traditional methods, such as the difficulty in coordinating the optimization of short-term drastic load fluctuations, operational economy, and long-term equipment reliability. This method achieves automated and precise optimization under multiple complex conditions, including overall plant load balance, dynamic unit ramp-up capabilities, and grid fluctuation perception. It significantly improves the dynamic stability of load allocation, overall operational economy, and proactive adaptation to grid demand, thereby enhancing the comprehensive operational benefits and safety support value of thermal power plants in the new power system.

[0087] Example 2 To achieve the above invention, embodiments of the present invention also provide another method for the stable and optimized allocation of thermal and electrical loads across multiple units in a thermal power plant, comprising: The process of calculating the heat and power load allocation involves using a genetic algorithm to optimize the electrical load, first-stage extraction steam load, and second-stage extraction steam load for each unit. The goal is to find the operating state that minimizes total heat consumption while meeting the overall load demand. For the electrical load of each unit... / and its heating steam extraction flow rate / Then, the corresponding heat consumption is obtained through the heat consumption curve. / The optimization target is the total heat consumption of thermal power plants. As shown in formula (1). Assuming that the economy of unit 1 is better than that of unit 2, in order to achieve the optimization of the heat and power load distribution of multiple units in a thermal power plant with dynamic stability, the constraints of the optimization process are as shown in formula (2). The constraint system of this invention is closely built around the core goal of "dynamic stability", specifically including three key levels: First, the basic balance constraint ensures that the sum of the electrical load and heat load allocated to each unit is strictly equal to the total load command issued by the power grid and the heating network, which is the basic premise for the optimization to be feasible; Second, the dynamic ramp constraint is innovatively introduced, which clearly limits the rate of change of the electrical load and the steam extraction flow of each unit in adjacent scheduling periods, thereby directly incorporating the dynamic factor of the actual ramp capacity of the unit into the optimization model, and fundamentally avoiding the drastic fluctuation of the load command; Finally, the main unit strategy constraint is uniquely designed. By setting the highest proportion of the load of the main unit with better economy in the total load, the load distribution is strategically guided in the optimization process to ensure that the main unit undertakes the basic load, thereby significantly improving the overall stability of the plant operation in the time series. These three constraints work together to ensure that the optimization results not only satisfy static economics, but also possess the dynamic stability necessary for actual operation.

[0088] (1) (2) This invention uses a genetic algorithm as the core algorithm for optimization, and its specific process is as follows: Figure 2 As shown. In the algorithm initialization stage, the population initialization matrix is ​​as shown in formula (3). The initial values ​​of the electrical load and heat load of each unit are set as the average of the output values ​​of the thermal power plant operation plan, and optimization is performed starting from this. In the genetic evolution process, the selection operation adopts the roulette wheel selection method. The probability of an individual being selected is proportional to its fitness value. The specific calculation is shown in formula (4). The crossover operation adopts the arithmetic crossover strategy. The new offspring individuals are generated by linearly combining the parent individuals. Its mathematical model is shown in formula (5). The mutation operation adopts the mutation operator based on Gaussian distribution. The population diversity is maintained by introducing random perturbation. Its probability density function is shown in formula (6). The termination condition of the algorithm is set to reach the maximum number of generations or the fitness value converges to a stable state. At this time, the optimal load allocation scheme is output. In order to verify the optimization effect, this invention is based on the specific heat consumption characteristic curves of the two units (as shown in formula (4)). Figure 3 , Figure 4 Simulation calculations were performed (as shown), and the optimized load allocation results are as follows: Figure 5 As shown in the figure. Performance calculations show that, compared with traditional optimization methods, the implementation of this scheme reduces the average coal consumption for power generation by 2.635 g / kWh and the average coal consumption for power supply by 2.938 g / kWh, significantly improving the economic efficiency of power plant operation.

[0089] (3) (4) (5) (6) This invention provides another method for the stable optimization and allocation of thermal and electrical loads across multiple units in a thermal power plant. By constructing an optimization model that integrates dynamic ramping and main unit strategy constraints, and employing an intelligent genetic algorithm for collaborative optimization of electrical and thermal loads, this method effectively solves the core problem of balancing economic objectives and operational stability in traditional load allocation. This method automates the entire process from constraint modeling and intelligent optimization to scheme generation, significantly improving the dynamic stability and overall operational economy of load allocation. It reduces unit coal consumption while ensuring long-term equipment reliability, enhancing the adaptive optimization capabilities and comprehensive operational benefits of thermal power plants under complex scheduling scenarios.

[0090] Example 3 To achieve the above invention, such as Figure 6 As shown, this embodiment also provides a device 10 for the stable and optimized distribution of thermal and electrical loads of multiple units in a thermal power plant. The device 10 includes: The time-series dynamic constraint modeling module 100 is used to construct an optimization model that includes time-series dynamic constraints. The time-series dynamic constraints are used to limit the rate of change of the electrical load and heating steam extraction flow of each unit in adjacent scheduling periods to not exceed the preset ramp-up capacity threshold.

[0091] The main unit strategy constraint module 200 is used to set the highest load ratio of the main unit with better economic efficiency in the total load, and embeds the control logic of the highest load ratio as a strategy constraint into the optimization model.

[0092] The joint optimization execution module 300 is used to jointly optimize the electrical load and heating steam extraction flow of multiple units based on the optimization model and using intelligent optimization algorithms. During the iteration process, the intelligent optimization algorithm simultaneously satisfies the total load balance constraint, dynamic ramp constraint, and main unit strategy constraint.

[0093] The dynamic stability output module 400 is used to output the optimal load allocation scheme that meets the dynamic stability requirements. The optimal load allocation scheme achieves synergistic optimization of the plant's overall operating economy and the long-term reliability of equipment by limiting short-term drastic fluctuations and frequent switching of unit load.

[0094] In one embodiment of the present invention, it further includes: a penalty factor dynamic adjustment module, used to dynamically adjust the weight coefficient of the penalty term related to the load change rate in the optimization model according to the real-time load fluctuation of the power grid: establishing an adaptive correction mechanism for the penalty coefficient, based on the benchmark penalty coefficient, combined with the ratio of the fluctuation amplitude of the total load of the power grid to the corresponding maximum load in the current scheduling period, and adjusted through preset correlation parameters, and calculating the real-time penalty coefficient for this period according to the power function relationship; wherein, the greater the fluctuation amplitude of the power grid load, the greater the calculated real-time penalty coefficient, so as to strengthen the suppression of drastic changes in unit load.

[0095] This invention provides a multi-unit thermal power load balancing optimization device for thermal power plants. Through the synergy of time-series dynamic constraint modeling, main unit strategy constraints, and an intelligent joint optimization module, it effectively solves the core problem of unifying economic objectives and operational stability in traditional optimization. This device achieves closed-loop optimization throughout the entire process, from constraint embedding and dynamic optimization to scheme generation, significantly improving the dynamic stability of load allocation and the overall plant's operational economy. It suppresses short-term load fluctuations while ensuring long-term reliable equipment operation, enhancing the adaptive decision-making capabilities and comprehensive operational efficiency of thermal power plants in complex grid dispatch environments.

[0096] To implement the methods of the above embodiments, the present invention also provides a computer device, such as... Figure 7As shown, the computer device 600 includes a memory 601 and a processor 602; wherein, the processor 602 reads the executable program code stored in the memory 601 to run a program corresponding to the executable program code, so as to implement the various steps of the above-described method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant.

[0097] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant as described in the foregoing embodiments.

[0098] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. 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.

[0099] 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 invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for the stable and optimized allocation of thermal and electrical loads across multiple units in a thermal power plant, characterized in that, include: S1. Construct an optimization model that includes time-series dynamic constraints. The time-series dynamic constraints are used to limit the rate of change of the electrical load and heating steam extraction flow of each unit in adjacent scheduling periods to not exceed the preset ramp-up capacity threshold. S2, set the highest load percentage of the main generating unit with better economic efficiency in the total load, and embed the control logic of the highest load percentage as a strategic constraint into the optimization model; S3, based on the optimization model, uses an intelligent optimization algorithm to jointly optimize the electrical load and heating steam extraction flow of multiple units. The intelligent optimization algorithm simultaneously satisfies the total load balance constraint, dynamic ramp constraint and main unit strategy constraint during the iteration process. S4 outputs the optimal load allocation scheme that meets the requirements of dynamic stability. The optimal load allocation scheme achieves synergistic optimization of the plant's overall operating economy and the long-term reliability of equipment by limiting the short-term drastic fluctuations and frequent switching of unit loads.

2. The method as described in claim 1, characterized in that, Construct an optimization model that includes time-series dynamic constraints, including: S11, Set dynamic ramping constraints: For each unit, the ratio of the change in electrical load to the maximum electrical load in adjacent scheduling periods shall not exceed the preset electrical load ramping capacity threshold; the ratio of the change in heating steam extraction flow rate to the maximum heating steam extraction flow rate in adjacent scheduling periods shall not exceed the preset heat load ramping capacity threshold; wherein, the absolute values ​​of the change in electrical load and the change in heating steam extraction flow rate are used in the calculation, and the electrical load ramping capacity threshold and the heat load ramping capacity threshold are upper limits preset according to the actual operating characteristics of the units; S12, the threshold values ​​for electrical load ramping capability and thermal load ramping capability are dynamically adjusted based on the real-time mechanical stress characteristics of the main and auxiliary equipment of the unit. The threshold parameters are corrected online through preset correction coefficients to adapt to the differentiated limitation requirements on the rate of change of unit load under different operating conditions.

3. The method as described in claim 1, characterized in that, Set the highest load percentage for the most economical main generating units in the total load, including: S21, Load Ratio Control Logic: The electrical load value of the main generating unit during the scheduling period shall not exceed the product of the total electrical load value of the entire plant in the same period and the preset load ratio upper limit parameter; the preset load ratio upper limit parameter is a proportional coefficient pre-set according to the economic advantages of the main generating unit, and the corresponding value range is 0.8 to 0.95; S22, the value of the load ratio upper limit parameter α is determined based on the relative economy between the main unit and the other units: compare the slope of the heat consumption characteristic curve of the main unit at the current operating point with the slope of the heat consumption characteristic curve of the second most economical unit at the corresponding operating point; if the slope of the curve of the main unit is less than the slope of the curve of the second most economical unit, then α is set to a first preset value; otherwise, α is set to a second preset value; wherein, the first preset value is 0.85 and the second preset value is 0.

9.

4. The method as described in claim 1, characterized in that, Intelligent optimization algorithms are used to jointly optimize the electrical load and heating steam extraction flow of multiple units, including: S31, the population of the intelligent optimization algorithm is initialized using a matrix structure; each individual corresponds to a unit load allocation scheme; for the k-th individual in the population, the corresponding initial electrical load and initial thermal load of the unit are generated by arithmetically dividing the total load demand value and superimposing random disturbance components to ensure that the initial solution has sufficient dispersion in the preset solution space under the premise of satisfying the total load balance constraint. S32, the selection operation uses a formulaic roulette wheel selection method, where the probability of an individual being selected is... Its fitness value The relationship is: ； in, Population size.

5. The method as described in claim 1, characterized in that, Also includes: S5. Based on the real-time load fluctuation of the power grid, dynamically adjust the weight coefficient of the penalty term related to the load change rate in the optimization model: establish an adaptive correction mechanism for the penalty coefficient, based on the benchmark penalty coefficient, combined with the ratio of the fluctuation amplitude of the total load of the power grid to the corresponding maximum load in the current scheduling period, and adjust it through preset correlation parameters, and calculate the real-time penalty coefficient for this period according to the power function relationship; wherein, the larger the fluctuation amplitude of the power grid load, the larger the calculated real-time penalty coefficient will be, so as to strengthen the suppression of drastic changes in unit load.

6. A device for the stable and optimized distribution of thermal and electrical loads across multiple units in a thermal power plant, characterized in that, include: The time-series dynamic constraint modeling module is used to construct an optimization model that includes time-series dynamic constraints. The time-series dynamic constraints are used to limit the rate of change of the electrical load and heating steam extraction flow of each unit in adjacent scheduling periods to not exceed the preset ramp-up capacity threshold. The main unit strategy constraint module is used to set the highest load percentage of the main unit with better economic efficiency in the total load, and embed the control logic of the highest load percentage as a strategy constraint into the optimization model. The joint optimization execution module is used to jointly optimize the electrical load and heating steam extraction flow of multiple units based on the optimization model and using intelligent optimization algorithms. The intelligent optimization algorithm simultaneously satisfies the total load balance constraint, dynamic ramp constraint and main unit strategy constraint during the iteration process. The dynamic stability output module is used to output the optimal load allocation scheme that meets the dynamic stability requirements. The optimal load allocation scheme achieves synergistic optimization of the plant's overall operating economy and the long-term reliability of equipment by limiting the short-term drastic fluctuations and frequent switching of unit load.

7. The apparatus as claimed in claim 6, characterized in that, Also includes: The penalty factor dynamic adjustment module is used to dynamically adjust the weight coefficient of the penalty term related to the load change rate in the optimization model according to the real-time load fluctuation of the power grid. An adaptive correction mechanism for the penalty coefficient is established. Based on the benchmark penalty coefficient, the ratio of the fluctuation range of the total load of the power grid to the corresponding maximum load in the current scheduling period is combined and adjusted through preset correlation parameters. The real-time penalty coefficient for this period is calculated according to the power function relationship. The larger the fluctuation range of the power grid load, the larger the calculated real-time penalty coefficient will be, so as to strengthen the suppression of drastic changes in unit load.

8. An electronic device, comprising: processor; The memory stores executable instructions; when the processor executes the instructions, it implements the method for stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant as described in any one of claims 1-5.

9. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements a method for the stable and optimized allocation of thermal and electrical loads of multiple units in a thermal power plant as described in any one of claims 1-5.

Images