Power plant energy-saving optimization control method

Abstract

The invention relates to the field of power plant energy saving, and discloses a power plant energy-saving optimization control method, which comprises the steps of calling historical efficiency data, calculating deviation based on an optimized actual energy efficiency value, quantifying energy-saving potential, generating a potential evaluation report, updating an optimized equipment parameter set on the basis of the potential evaluation report, and giving an adjustment suggestion. According to the method, global search and local optimization can be considered through combination of the hybrid genetic algorithm and the particle swarm algorithm, and limitation of a single algorithm, such as premature convergence of the genetic algorithm or falling into local optimization of the particle swarm algorithm, is avoided, so that optimization precision and efficiency are improved. According to the method, the optimization direction is accurately positioned, the defect that a key loss item is neglected in a traditional single balance method is overcome, the energy-saving potential is continuously excavated through closed-loop adjustment, it is ensured that the energy efficiency value continuously approaches the target, the optimization stagnation phenomenon is avoided, and the continuity and operability of energy-saving optimization of the power plant are improved.

Description

Technical Field

[0001] This invention relates to the field of power plant energy conservation, and more specifically, to a power plant energy conservation optimization control method. Background Technology

[0002] Power plant energy conservation refers to measures and methods that reduce energy consumption and improve energy efficiency by optimizing the internal production processes, equipment operation, and energy management of a power plant. This includes adopting efficient equipment and technologies, improving fuel management, optimizing the operation of boilers, turbines, and generator sets, and recovering and utilizing waste heat. Power plant energy conservation not only helps reduce operating costs but also reduces environmental pollution and enhances the economic benefits and sustainable development capabilities of enterprises. Through precise load management, energy monitoring and control, and the implementation of advanced automation systems, power plants can maximize energy savings and reduce unnecessary energy waste while ensuring a stable power supply.

[0003] Existing power plant energy-saving optimization control methods have several shortcomings in practical applications: First, single positive or negative balance methods often neglect key factors such as flue gas heat loss and mechanical incomplete combustion loss, leading to biases in energy efficiency assessment and affecting the accuracy of optimization decisions and overall effectiveness. Second, traditional single-objective optimization algorithms, such as the simplex method, have poor adaptability in high-dimensional, nonlinear parameter spaces, easily getting trapped in local optima and failing to achieve global optima, thus limiting the full exploitation of energy-saving potential. Third, many traditional optimization methods lack dynamic feedback mechanisms and cannot adjust based on real-time data. This "optimization stagnation" phenomenon makes the optimization process lack continuity and adaptability, unable to cope with complex and changing operating environments. Therefore, current power plant energy-saving optimization methods still face many challenges in theory and practice, requiring the integration of multi-objective optimization, dynamic adjustment feedback mechanisms, and equipment physical constraints to improve the comprehensiveness, operability, and sustainability of optimization effects. Summary of the Invention

[0004] To overcome the above-mentioned deficiencies of the prior art, embodiments of the present invention provide a power plant energy-saving optimization control method.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A power plant energy-saving optimization control method, the method comprising:

[0007] Collect operational data from power plants, preprocess the operational data to obtain a standardized dataset, and extract a structured dataset from the standardized dataset;

[0008] The structured dataset is preprocessed to obtain a calibrated parameter set. The positive balance efficiency value and the negative balance efficiency value are calculated. The negative balance efficiency value is corrected and then weighted averaged to obtain the actual energy efficiency value.

[0009] The target energy efficiency value and power plant constraints are obtained. The actual energy efficiency value, target energy efficiency value and constraints are searched using an improved genetic-particle swarm hybrid algorithm to find the optimized set of equipment parameters and obtain the optimized actual energy efficiency value.

[0010] Historical performance data is retrieved, deviations are calculated based on the optimized actual energy efficiency values, energy-saving potential is quantified, a potential assessment report is generated, the optimized equipment parameter set is updated based on the potential assessment report, and adjustment suggestions are given.

[0011] Further, the calculation of the positive balance efficiency value includes:

[0012] The effective heat utilization is calculated based on the first law of thermodynamics, and the input heat is calculated in combination with the fuel to obtain the positive balance efficiency value.

[0013] Furthermore, the calculation of the anti-balancing efficiency value and the correction of the anti-balancing efficiency value include:

[0014] The heat loss from flue gas, chemical incomplete combustion, mechanical incomplete combustion, heat dissipation, and physical heat loss from ash and slag were calculated separately. The five types of losses were corrected, and the corrected reverse balance efficiency value was obtained by subtracting the sum of the five types of losses after correction from 100%.

[0015] Furthermore, the improved genetic-particle swarm optimization algorithm searches for an optimized set of device parameters, including:

[0016] Based on the constraints, n sets of parameter combinations are randomly generated, with energy efficiency gap and energy consumption reduction as dual objectives, and weights are assigned to form a fitness function. The top 20% of optimal solutions are selected from the n sets of parameter combinations to obtain the parameter set of the initial population, which is regarded as the initial position of the particle swarm.

[0017] The parameters are optimized using the velocity formula of the particle swarm optimization algorithm to find the optimal solution and obtain the updated particle swarm position.

[0018] Select the top 30% of the best solutions for crossover and mutation operations, and retain the top 10% of the best solutions to obtain a new population parameter set;

[0019] By introducing a pre-trained Q-network model and outputting the new population parameter set to the pre-trained Q-network model, the optimized device parameter set is obtained.

[0020] Furthermore, the calculation of deviations based on optimized actual energy efficiency values ​​to quantify energy-saving potential includes:

[0021] The deviation is calculated using a dynamic weighted sliding window algorithm, and a historical correction factor is introduced to correct the deviation.

[0022] Establish a potential threshold, which includes a and b, where a < b = 0;

[0023] When the corrected deviation is ≤ a, it is considered to have high energy-saving potential, triggering a deep update;

[0024] When a ≤ the corrected deviation ≤ b, it is considered as medium energy saving potential and triggers a regular update;

[0025] If b ≤ the corrected deviation, then it is considered that there is no energy-saving space, and the optimized equipment parameter set is maintained.

[0026] Furthermore, the depth update includes:

[0027] Based on the constraints, m sets of parameter combinations are randomly generated. With energy efficiency gap and energy consumption reduction as dual objectives, new solutions are generated through crossover and mutation operations. The optimal solution is selected through non-dominated sorting to obtain a set of candidate parameters.

[0028] The optimal parameter set is obtained by verifying the data on the digital twin platform and then parsing the optimal parameter set through edge computing nodes.

[0029] Furthermore, the routine update includes:

[0030] A proportional-integral-derivative controller is used to calculate small adjustment amounts to obtain fine-tuned parameters. The model predictive controller is then used to optimize equipment parameter changes over several future periods.

[0031] Retrieve actual operating data, predict energy consumption trends based on fine-tuned parameters, compare with actual operating data, calculate similarity for verification, and obtain the fine-tuned equipment parameter set.

[0032] Furthermore, the calculation of small adjustment amounts using a proportional-integral-derivative controller includes:

[0033] Analyze the contribution of each parameter to energy efficiency, select three key parameters for adjustment, and prioritize adjusting the parameter with the highest contribution to generate an initial set of equipment parameters;

[0034] The initial set of equipment parameters was verified by combining the constraints, and the fine-tuned set of equipment parameters was obtained.

[0035] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the above-described power plant energy-saving optimization control method.

[0036] A computer-readable storage medium storing a computer program, which, when executed, implements the aforementioned power plant energy-saving optimization control method.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] This application discloses a power plant energy-saving optimization control method, comprising: obtaining a target energy efficiency value and power plant constraints; using an improved genetic-particle swarm optimization hybrid algorithm to search for an optimized set of equipment parameters based on the actual energy efficiency value, the target energy efficiency value, and the constraints, thereby obtaining the optimized actual energy efficiency value; retrieving historical energy efficiency data; calculating the deviation based on the optimized actual energy efficiency value; quantifying the energy-saving potential; generating a potential assessment report; updating the optimized set of equipment parameters based on the potential assessment report; and providing adjustment suggestions. This invention, by combining a hybrid genetic algorithm and a particle swarm optimization algorithm, can balance global search and local optimization, avoiding the limitations of a single algorithm, such as premature convergence in the genetic algorithm or getting trapped in local optima in the particle swarm optimization algorithm, thus improving optimization accuracy and efficiency. This method accurately locates the optimization direction, overcomes the shortcomings of traditional single balancing methods that ignore key loss terms, and continuously explores energy-saving potential through closed-loop calibration, ensuring that the energy efficiency value continuously approaches the target, avoiding optimization stagnation, and improving the sustainability and operability of power plant energy-saving optimization. Attached Figure Description

[0039] Figure 1 A flowchart of a power plant energy-saving optimization control method provided by the present invention;

[0040] Figure 2 A schematic diagram of the structure of an electronic device provided by the present invention;

[0041] Figure 3 A schematic diagram of the structure of a computer-readable storage medium provided by the present invention;

[0042] Figure 4 This invention provides a flowchart for optimizing equipment parameters. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0044] Example 1

[0045] Please see Figure 1 As shown in the figure, this embodiment discloses a power plant energy-saving optimization control method, the method comprising:

[0046] S101: Collect operational data from the power plant, preprocess the operational data to obtain a standardized dataset, and extract a structured dataset from the standardized dataset;

[0047] The power plant's operating data includes, but is not limited to, the following core parameters:

[0048] Boiler system: Key operating parameters such as boiler temperature, boiler pressure, and boiler steam flow.

[0049] Steam system: Indicators such as steam temperature and steam pressure reflect the thermal effect of steam production and pressure control status of the boiler.

[0050] Steam turbines and generators: Steam turbine speed, load, generator power, etc., are used to measure the operating status and power output of the generator set.

[0051] Cooling system: Parameters such as cooling tower water temperature and cooling water flow rate are monitored to assess the heat exchange efficiency and water circulation of the cooling system.

[0052] Flue gas emissions: Indicators such as oxygen content in flue gas are used to monitor boiler combustion efficiency and emissions.

[0053] It should be understood that: the operating data is automatically collected through different industrial automation systems and real-time monitoring platforms, such as sensor networks: widely distributed in various parts of the power plant, collecting precise physical quantities, such as temperature, pressure, flow rate, etc., and transmitting them to the central control system wirelessly or via wired. The collection method is existing technology, and will not be described in detail in this embodiment.

[0054] The preprocessing of operational data includes: data cleaning, normalization, and standardization of the acquired operational data;

[0055] As a concrete example, we first use the Z-score method: calculate the mean μ and standard deviation σ of each parameter column. If |Z|=(x-μ) / σ>3, it is judged as an outlier and is filled or deleted using the median. Then, we calculate the first quartile Q1 and the third quartile Q3. The range of outliers is [Q1-1.5IQR, Q3+1.5IQR]. Values ​​outside the range are filled with linear interpolation. Finally, we delete completely duplicate rows and take the mean of partially duplicated numerical fields.

[0056] In practice, the parameter values ​​are compressed to the [0, 1] interval and normalized first. Then, the parameter values ​​are converted into a normal distribution with a mean of 0 and a standard deviation of 1 for standardization, thus obtaining a standardized parameter set.

[0057] It is worth noting that the specific steps for extracting a structured dataset from a standardized dataset are as follows:

[0058] Correlation analysis was performed on the standardized dataset. Based on the results of the German-Austrian correlation analysis, features strongly correlated with the target variable were selected, redundant features were removed, and the variance thresholding method was used to remove features with variance close to 0, thereby reducing computational complexity.

[0059] S102: The structured dataset is calibrated to obtain the calibrated parameter set. The positive balance efficiency value and the negative balance efficiency value are calculated. The negative balance efficiency value is corrected and then weighted averaged to obtain the actual energy efficiency value.

[0060] It should be noted that although the structured dataset has undergone preliminary processing, factors such as sensor errors, equipment aging, and fluctuations in operating conditions can cause deviations between actual physical quantities and theoretical values. By introducing real-time correction coefficients (such as temperature compensation coefficients and pressure attenuation coefficients) through calibration, these deviations can be dynamically corrected to ensure that the input parameters (such as fuel flow rate and steam parameters) truly reflect the current operating status, thereby improving the accuracy and reliability of energy efficiency calculations and avoiding deviations from reality due to distortion of the original data.

[0061] The calibrated parameter set includes physical quantities such as fuel quantity B and exhaust gas temperature. Carbon content of ash Steam enthalpy / Feedwater enthalpy Steam output (m), lower heating value of fuel ;

[0062] Specifically, calculating the positive balance efficiency value includes:

[0063] The effective heat utilization is calculated based on the first law of thermodynamics, and the input heat is calculated in conjunction with the fuel to obtain the positive balance efficiency value. The calculation method is as follows:

[0064] Calculate the effective use of heat: ;

[0065] Calculate the heat input of fuel: ;

[0066] Calculate the positive balance efficiency: ;

[0067] For example, the calculation of the anti-equilibrium efficiency value and the correction of the anti-equilibrium efficiency value include:

[0068] The heat loss from flue gas, chemical incomplete combustion, mechanical incomplete combustion, heat dissipation, and physical heat loss from ash and slag were calculated separately. The five types of losses were corrected separately. The corrected reverse balance efficiency value was obtained by subtracting the sum of the five types of losses after correction from 100%.

[0069] During implementation, flue gas heat loss calculate:

[0070] Calculate the enthalpy difference of exhaust gas: ;

[0071] Calculate the heat loss per unit of fuel flue gas: ;

[0072] In the formula: For the boiler room ambient temperature, Theoretical air volume air density, Specific heat capacity of air at constant pressure;

[0073] in, At that time, adjustments are made through a Kalman filter to make error The corrected result is obtained. ;

[0074] Chemical incomplete combustion loss calculate:

[0075] Establishing the ANFIS nonlinear mapping: ;

[0076] Established based on historical data The relationship, output the current The numerical value is output through the ANFIS inference system. The value;

[0077] Mechanical incomplete combustion loss calculate:

[0078] Calculate the heat loss of ash and slag: ;

[0079] Calculate mechanical incomplete combustion losses: ;

[0080] In the formula: For ash and slag discharge, This indicates the lower heating value of ash and slag;

[0081] The correction factor is adjusted according to the particle size of the ash and slag; for example, when the proportion of fine ash increases... Increase;

[0082] Heat loss calculate:

[0083] Calculation of temperature difference driven: ;

[0084] Calculate the power loss due to heat dissipation: ;

[0085] Converted to heat loss rate: ;

[0086] Physical heat loss of ash calculate:

[0087] Calculate the enthalpy difference of ash and slag: ;

[0088] Calculate the physical heat loss rate: ;

[0089] Corrected anti-equilibrium efficiency value: .

[0090] As a specific example, the positive balance efficiency value and the corrected anti-balance efficiency value are fused using a weighted average method, with the weights determined through K-fold cross-validation of historical data. In this embodiment, , The actual energy efficiency value is obtained by the following formula:

[0091] ;

[0092] In this embodiment, as a specific example, we assume that the parameters are as follows:

[0093] Fuel consumption B: 13,000 tons / year (annual operating time T=8,000 hours, hourly consumption 1.625 tons).

[0094] Lower heating value of fuel 21,630 kJ / kg (calibrated value, with a moisture correction factor of 1.03 × 21,000 kJ / kg).

[0095] Main steam output D: 550 tons / hour, enthalpy h=3, 450kJ / kg (15MPa / 550℃);

[0096] Water supply enthalpy h_fw: 900 kJ / kg (180℃);

[0097] Smoke temperature 147℃ (calibrated value, including an ambient temperature correction factor of 0.98×150℃);

[0098] Carbon content of ash 8.5% (calibrated value, including fly ash resistivity correction);

[0099] Ambient temperature =25℃; flue gas volume =6.92m³ / kg;

[0100] air density =1.184 kg / m³; specific heat capacity of air =1.005kJ / kg·℃

[0101] Positive balance efficiency calculation:

[0102] Input energy: 13,000,000 × 21,630 = 2.8119 × 10¹¹ kJ / year;

[0103] Output energy: 550×8000h×(3450-900)×1000=1.122×10¹¹kJ / year;

[0104] Positive balance efficiency 1.122 × 10¹¹ / 2.8119 × 10¹¹ × 100% = 39.91%

[0105] The specific steps for calculating the anti-balancing efficiency and loss term are as follows:

[0106] Flue gas heat loss :

[0107] =(147-25)×1.005×1.184×6.92=122×8.00=976.0kJ / kg;

[0108] Unit fuel heat loss Initial value:

[0109] =(976.0) / (1,625×21,630)×3,600×100%=7.82%;

[0110] Dynamic correction trigger: If the temperature is 122℃ or higher than 15℃, activate the Kalman filter for correction.

[0111] State equation update: x_t = x_{t-1} + K_t(z_t - Hx_{t-1})

[0112] Covariance update: P_t=(I-K_tH)P_{t-1}

[0113] Revised With an error ≤ ±1.5%, the final q² = 7.96%;

[0114] Chemical incomplete combustion loss :

[0115] The measured CO content in the flue gas was 0.05%, and the H2 content was 0.03%.

[0116] =7.5×0.05+3.6×0.03=0.48%

[0117] Mechanical incomplete combustion loss :

[0118] =3,370 × 8.5% / 21,630 = 1.32%

[0119] Heat loss :

[0120] Based on a boiler surface area of ​​1000m², and a temperature difference of 300℃ - 25℃ = 275℃,

[0121] =0.5×1000×275 / 2.8119×10¹¹×100%≈0.49%

[0122] Physical heat loss of ash :

[0123] The ash content is 10%, the specific heat is 1.0 kJ / kg·K, and the temperature is 200℃.

[0124] =(0.1×13,000,000×1.0×200) / 2.8119×10¹¹×100%=0.92%

[0125] Anti-balance efficiency 100% - (7.96% + 0.48% + 1.32% + 0.49% + 0.92%) = 88.83%;

[0126] Comprehensive calculation of actual energy efficiency value

[0127] Weighted average method (positive balance weight 40%, negative balance weight 60%):

[0128] =39.91%×0.4+88.83%×0.6=69.07%.

[0129] S103: Obtain the target energy efficiency value and power plant constraints. Use the improved genetic-particle swarm hybrid algorithm to search for the optimized set of equipment parameters and obtain the optimized actual energy efficiency value.

[0130] It should be noted that in this embodiment, the target energy efficiency value is assumed to be 75%, and the optimized actual energy efficiency value is 73.2%.

[0131] like Figure 4 As shown, the improved genetic-particle swarm optimization algorithm searches for an optimized set of device parameters, including:

[0132] Based on the constraints, n sets of parameter combinations are randomly generated, with n taking the value of 50. The dual objectives are energy efficiency gap and energy consumption reduction, and weights are assigned to form a fitness function. The top 20% of the optimal solutions are selected from the n sets of parameter combinations to obtain the parameter set of the initial population, which is regarded as the initial position of the particle swarm. In this embodiment, the energy efficiency gap is assigned the value of 0.7, and the energy consumption reduction is assigned the value of 0.3.

[0133] The parameters are optimized using the velocity formula of the particle swarm optimization algorithm to find the optimal solution and obtain the updated particle swarm position.

[0134] The top 30% of optimal solutions are selected for crossover and mutation operations, while the top 10% of optimal solutions are retained to obtain a new population parameter set. It should be noted that the crossover operation is performed with a probability of 0.8 for single-point crossover to generate a new solution, while the mutation operation is performed with a probability of 0.1 for Gaussian mutation to maintain population diversity.

[0135] By introducing a pre-trained Q-network model and outputting the new population parameter set to the pre-trained Q-network model, the optimized device parameter set can be obtained.

[0136] It's important to note that the Q-network model is a crucial method in reinforcement learning. It combines Q-learning and deep neural networks. Q-networks approximate the Q-value function using deep neural networks, mapping states and actions to Q-values, thus enabling the agent to make decisions in complex environments. By continuously training the neural network, Q-networks allow the agent to learn optimal policies based on experience.

[0137] Specifically, the Q-network model is trained using a deep Q-learning method. Through an experience replay mechanism, a four-tuple of state, action, reward, and next state is stored in a replay pool, and 32 batches are randomly selected for training. During each update, an ε-greedy strategy is used to select the optimal action (90% optimal action selection, 10% random exploration) to minimize the prediction error of the Q-value, gradually adjusting the network weights. The network adopts a three-layer fully connected structure (20 input nodes, 64 hidden nodes, and 4 output nodes), and the Q-value is updated through backpropagation, ultimately training a Q-network model capable of making decisions in real time.

[0138] S104: Retrieve historical performance data, calculate the deviation based on the optimized actual energy efficiency value, quantify the energy-saving potential, generate a potential assessment report, update the optimized equipment parameter set based on the potential assessment report, and provide adjustment suggestions;

[0139] In this embodiment, it is assumed that the historical performance data retrieved shows a historical performance value of 72%.

[0140] The calculation of deviations based on optimized actual energy efficiency values ​​to quantify energy-saving potential includes:

[0141] The deviation is calculated using a dynamic weighted sliding window algorithm, and a historical correction factor is introduced to correct the deviation.

[0142] A potential threshold is defined, which includes a and b, where a < b = 0, and a and b take values ​​of -3% and 0, respectively.

[0143] When the corrected deviation is ≤ a, it is considered to have high energy-saving potential, triggering a deep update;

[0144] When a ≤ the corrected deviation ≤ b, it is considered as medium energy saving potential and triggers a regular update;

[0145] If b ≤ the corrected deviation, then it is considered that there is no energy-saving space, and the optimized equipment parameter set is maintained.

[0146] As a specific example, based on S103, the optimized actual efficiency value is 73.20%, and the target efficiency value is 75%, then the deviation = 73.2% - 75% = -1.8%;

[0147] Historical correction factor 73.2% / 72% ≈ 1.017;

[0148] In summary, the corrected deviation = ;

[0149] Compare the deviation with the potential threshold, i.e., -3% ≤ ≤0 is considered as medium-energy-saving potential, triggering a regular update;

[0150] The depth update includes:

[0151] Based on the constraints, m sets of parameter combinations are randomly generated. With energy efficiency gap and energy consumption reduction as dual objectives, new solutions are generated through crossover and mutation operations. The optimal solution is selected through non-dominated sorting to obtain a set of candidate parameters.

[0152] The optimal parameter set is obtained by verifying the data on the digital twin platform and then parsing the optimal parameter set through edge computing nodes.

[0153] It's important to note that a digital twin platform uses real-time simulation and modeling technologies to construct a virtual model corresponding to a physical system (such as a power plant or industrial equipment), reflecting the system's state and behavior in real time. Utilizing sensor data, computational models, and simulation techniques, it creates a digital twin capable of accurately simulating the physical system's operation, performance, and responses for prediction, optimization, and decision support. The key advantage of a digital twin platform lies in its ability to monitor, analyze, and optimize physical systems in real time without requiring excessive intervention from the actual equipment.

[0154] The training method for a digital twin platform involves the following steps: First, the platform needs to collect a large amount of historical and real-time sensor data to build a dynamic energy efficiency model for the system. Next, algorithms such as machine learning, deep learning, or reinforcement learning are used to train the model. For example, in reinforcement learning, the system simulates operations with different parameter combinations to evaluate their impact on system performance (such as energy efficiency and output) and continuously optimizes decision-making strategies. During training, techniques such as experience replay and Q-learning are typically used to continuously adjust the model's parameters to improve its predictive accuracy and decision-making capabilities. To ensure accuracy, the digital twin platform also needs to continuously update and self-optimize in real time to adapt to changes in equipment status and fluctuations in environmental factors. The trained digital twin model can be used for practical applications such as predicting equipment failures, optimizing operating strategies, and improving energy efficiency.

[0155] It should be understood that during the real-time deployment of edge computing, the system first receives the optimal parameter combination and the real-time status of the equipment (e.g., the current air-coal ratio of 1.1.5). Edge nodes parse the instructions transmitted via the OPCUA protocol and then send adjustment instructions to the actuators, such as the dampers of the boiler burner. Based on feedback, the edge computing real-time monitoring system responds to parameter adjustments, such as adjusting the damper opening from 85% to 90%. To avoid over-adjustment, a Kalman filter is used to suppress potential overshoot and ensure a smooth adjustment process. Within 30 seconds of deployment, the system continues to collect feedback data (e.g., boiler temperature change rate) to evaluate the adjustment effect in real time and dynamically optimize system operation based on the feedback data.

[0156] The regular updates include:

[0157] A proportional-integral-derivative controller is used to calculate small adjustment amounts to obtain fine-tuned parameters. The model predictive controller is then used to optimize equipment parameter changes over several future periods.

[0158] A proportional-integral-derivative (PID) controller is a feedback control strategy that calculates the current error in real time and adjusts the control input accordingly. It consists of three parts: the proportional component adjusts based on the current error; the integral component eliminates long-term small errors; and the derivative component predicts error changes to avoid overshoot. It is suitable for quickly and accurately correcting small-amplitude system deviations. In contrast, a model predictive controller (MPC) uses a dynamic model of the system to predict state changes over multiple future time periods. Based on these predictions, it optimizes the control strategy to ensure optimal overall system performance in the future. MPC considers long-term system behavior and is suitable for scenarios requiring long-term performance optimization, while the PID controller is better suited for real-time, local adjustments. Combining the two can achieve a balance between immediate adjustment and future optimization.

[0159] Retrieve actual operating data, predict energy consumption trends based on fine-tuned parameters, compare with actual operating data, calculate similarity for verification, and obtain the fine-tuned equipment parameter set.

[0160] The calculation of small adjustment amounts using a proportional-integral-derivative controller includes:

[0161] Analyze the contribution of each parameter to energy efficiency, select three key parameters for adjustment, and prioritize adjusting the parameter with the highest contribution to generate an initial set of equipment parameters;

[0162] As a concrete example, the "SHAP value interpretation model" is used to analyze the impact of various parameters on energy efficiency: the boiler air-to-coal ratio α contributes 60%, and the flue gas temperature... Contributes 25%, carbon content of ash slag Contribution 10% and turbine inlet valve opening β: contribution 5%;

[0163] Based on the contribution value, the top three key parameters were selected: boiler air-to-coal ratio α, and flue gas temperature. and the carbon content of ash ;

[0164] The following are the specific steps for rapid calibration using linear extrapolation based on historical data:

[0165] The extent to which it needs to improve energy efficiency ;

[0166] Allocation is based on contribution weight, see Table 1 for details:

[0167] Table 1: Weight Allocation Table

[0168] parameter Contribution Required contribution increase Boiler air-coal ratio 60% 0.72% Smoke temperature 25% 0.30% Carbon content of ash 10% 0.12% Steam turbine inlet valve opening 5% 0%

[0169] A1: Boiler air-coal ratio adjustment:

[0170] Its linear relationship: For every 0.01 increase, energy efficiency improves by 0.15%.

[0171] Adjustments required: =0.72%÷0.15%×0.01=0.048;

[0172] Optimized =1.198;

[0173] A2: Smoke exhaust temperature Adjustment:

[0174] Assume the optimized flue gas temperature is 145℃;

[0175] Its linear relationship: For every 5°C decrease, heat loss decreases by 0.10%.

[0176] Adjustment required: Δ =0.30%÷0.10%×5℃=15℃;

[0177] Optimized =145℃-15℃=130℃;

[0178] A3: Carbon content of ash Adjustment:

[0179] Historical data on carbon content in ash slag are as follows: =8.0%;

[0180] Its linear relationship: Each 1% decrease results in a 0.12% reduction in mechanical losses;

[0181] Adjustment required: Δ =0.12% ÷ 0.12% × 1% = 1%

[0182] After optimization =8.0% - 1% = 7.0%;

[0183] A4: Adjustment of turbine inlet valve opening β. Since the contribution is only 5% and the current value is close to the optimal value, the original value is maintained.

[0184] The optimized energy efficiency values ​​were verified:

[0185] =73.8% + 1.2(0.6% + 0.25% + 0.1%) = 74.94%;

[0186] The calculated difference from the target value is 0.2%, which can be further approximated by a second-order dynamic correction Kalman filter;

[0187] The initial set of equipment parameters was verified by combining the constraints, and the fine-tuned set of equipment parameters was obtained.

[0188] In this embodiment, the annual fuel savings can be calculated, and the specific steps are as follows:

[0189] ;

[0190] That is: ;

[0191] Example 2

[0192] Please see Figure 2 As shown, this embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the power plant energy-saving optimization control method provided by the above methods.

[0193] Since the electronic device described in this embodiment is the electronic device used to implement the power plant energy-saving optimization control method in the embodiments of this application, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the power plant energy-saving optimization control method described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the power plant energy-saving optimization control method in the embodiments of this application falls within the scope of protection of this application.

[0194] Example 3

[0195] Please see Figure 3 As shown, this embodiment discloses a computer-readable storage medium, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements the power plant energy-saving optimization control method provided by the above methods.

[0196] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters, weights, and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0197] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired or wireless network. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0198] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0199] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0200] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only one method, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0201] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0202] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0203] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0204] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power plant energy-saving optimization control method, characterized in that, The method includes: Collect operational data from power plants, preprocess the operational data to obtain a standardized dataset, and extract a structured dataset from the standardized dataset; The structured dataset is preprocessed to obtain a calibrated parameter set. The positive balance efficiency value and the negative balance efficiency value are calculated. The negative balance efficiency value is corrected and then weighted averaged to obtain the actual energy efficiency value. The target energy efficiency value and power plant constraints are obtained. The actual energy efficiency value, target energy efficiency value and constraints are searched using an improved genetic-particle swarm hybrid algorithm to find the optimized set of equipment parameters and obtain the optimized actual energy efficiency value. Historical performance data is retrieved, deviations are calculated based on the optimized actual energy efficiency values, energy-saving potential is quantified, a potential assessment report is generated, the optimized equipment parameter set is updated based on the potential assessment report, and adjustment suggestions are given.

2. The power plant energy-saving optimization control method according to claim 1, characterized in that, The calculation of the positive balance efficiency value includes: The effective heat utilization is calculated based on the first law of thermodynamics, and the input heat is calculated in combination with the fuel to obtain the positive balance efficiency value.

3. The power plant energy-saving optimization control method according to claim 1, characterized in that, The calculation of the anti-equilibrium efficiency value and the correction of the anti-equilibrium efficiency value include: The heat loss from flue gas, chemical incomplete combustion, mechanical incomplete combustion, heat dissipation, and physical heat loss from ash and slag were calculated separately. The five types of losses were corrected, and the corrected reverse balance efficiency value was obtained by subtracting the sum of the five types of losses after correction from 100%.

4. The power plant energy-saving optimization control method according to claim 1, characterized in that, The improved genetic-particle swarm optimization algorithm searches for an optimized set of device parameters, including: Based on the constraints, n sets of parameter combinations are randomly generated, with energy efficiency gap and energy consumption reduction as dual objectives, and weights are assigned to form a fitness function. The top 20% of optimal solutions are selected from the n sets of parameter combinations to obtain the parameter set of the initial population, which is regarded as the initial position of the particle swarm. The parameters are optimized using the velocity formula of the particle swarm optimization algorithm to find the optimal solution and obtain the updated particle swarm position. Select the top 30% of the best solutions for crossover and mutation operations, and retain the top 10% of the best solutions to obtain a new population parameter set; By introducing a pre-trained Q-network model and outputting the new population parameter set to the pre-trained Q-network model, the optimized device parameter set is obtained.

5. The power plant energy-saving optimization control method according to claim 1, characterized in that, The deviation calculated based on the optimized actual energy efficiency value, quantifying the energy-saving potential, includes: The deviation is calculated using a dynamic weighted sliding window algorithm, and a historical correction factor is introduced to correct the deviation. Establish a potential threshold, which includes a and b, where a < b = 0; When the corrected deviation is ≤ a, it is considered to have high energy-saving potential, triggering a deep update; When a ≤ the corrected deviation ≤ b, it is considered as medium energy saving potential and triggers a regular update; If b ≤ the corrected deviation, then it is considered that there is no energy-saving space, and the optimized equipment parameter set is maintained.

6. The power plant energy-saving optimization control method according to claim 5, characterized in that, The deep update includes: Based on the constraints, m sets of parameter combinations are randomly generated. With energy efficiency gap and energy consumption reduction as dual objectives, new solutions are generated through crossover and mutation operations. The optimal solution is selected through non-dominated sorting to obtain a set of candidate parameters. The optimal parameter set is obtained by verifying the data on the digital twin platform and then parsing the optimal parameter set through edge computing nodes.

7. The power plant energy-saving optimization control method according to claim 5, characterized in that, The regular updates include: A proportional-integral-derivative controller is used to calculate small adjustment amounts to obtain fine-tuned parameters. The model predictive controller is then used to optimize equipment parameter changes over several future periods. Retrieve actual operating data, predict energy consumption trends based on fine-tuned parameters, compare with actual operating data, calculate similarity for verification, and obtain the fine-tuned equipment parameter set.

8. The power plant energy-saving optimization control method according to claim 7, characterized in that, The calculation of small adjustment amounts using a proportional-integral-derivative controller includes: Analyze the contribution of each parameter to energy efficiency, select three key parameters for adjustment, and prioritize adjusting the parameter with the highest contribution to generate an initial set of equipment parameters; The initial set of equipment parameters was verified by combining the constraints, and the fine-tuned set of equipment parameters was obtained.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the power plant energy-saving optimization control method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed, implements the power plant energy-saving optimization control method according to any one of claims 1 to 8.

Images