Power generation cost optimization method, system, equipment and medium

Abstract

The invention discloses a power generation cost optimization method, system and device and a medium. The method comprises the steps of obtaining target power generation parameters of a thermal power generating unit; inputting the target power generation parameter into a preset target power generation cost analysis model for processing to obtain a predicted marginal power generation cost of the thermal power generating unit; performing causal analysis based on the target power generation parameter and the predicted marginal power generation cost to obtain a cost influence reason of the thermal power generating unit; and adjusting the target power generation parameter based on the cost influence reason to obtain an optimized power generation parameter, and controlling the thermal power generating unit to generate power based on the optimized power generation parameter to reduce the power generation cost. The problem that the cost optimization precision is low due to the fact that an existing simple mathematical model is difficult to accurately reflect the cost change in the dynamic power generation process is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of thermal power generation, and particularly relates to a power generation cost optimization method, system, device and medium. BACKGROUND

[0002] With the construction of the electricity spot market, accurate marginal cost calculation has become a key basis for thermal power units to participate in market bidding. The marginal cost reflects the additional cost required for the unit to increase 1MW of power generation, and is an important basis for the unit to develop a reasonable bidding strategy. However, during the operation of the unit, multiple cost factors such as fuel, environmental protection, and start-stop are involved, and these factors dynamically change with the unit operating conditions, which brings great challenges to cost calculation.

[0003] At present, simple mathematical models are often used to calculate and optimize power generation costs. However, simple mathematical models cannot accurately reflect the cost changes in the dynamic power generation process, resulting in low cost optimization accuracy. SUMMARY

[0004] In order to overcome the problem that the existing simple mathematical model cannot accurately reflect the cost changes in the dynamic power generation process, resulting in low cost optimization accuracy, the present application provides a power generation cost optimization method, system, device and medium.

[0005] In a first aspect, in order to solve the above technical problems, the present application provides a power generation cost optimization method, comprising:

[0006] obtaining a target power generation parameter of a thermal power unit;

[0007] inputting the target power generation parameter into a preset target power generation cost analysis model for processing to obtain a predicted marginal power generation cost of the thermal power unit;

[0008] performing causal analysis based on the target power generation parameter and the predicted marginal power generation cost to obtain a cost influence reason of the thermal power unit;

[0009] adjusting the target power generation parameter based on the cost influence reason to obtain an optimized power generation parameter, and controlling the thermal power unit to generate power based on the optimized power generation parameter to reduce the power generation cost.

[0010] Further, obtaining the target power generation parameter of the thermal power unit comprises:

[0011] obtaining an initial power generation parameter of the thermal power unit;

[0012] performing abnormality analysis on the initial power generation parameter to obtain an abnormality index, the abnormality index including a data integrity index, an abnormal data proportion, a historical consistency index, and a sensor health status;

[0013] calculating the credibility of the initial power generation parameter based on a preset abnormality weight parameter and the abnormality index;

[0014] Adjust the initial power generation parameter based on the credibility to obtain the target power generation parameter of the thermal power generating unit.

[0015] Further, the training process of the target power generation cost analysis model is as follows:

[0016] Obtain the historical power generation parameter of the thermal power generating unit;

[0017] Input the historical power generation parameter into the initial model for training to obtain the historical predicted power generation cost of the thermal power generating unit, including the historical predicted fuel cost, the historical predicted environmental protection cost, the historical predicted maintenance cost and the historical predicted start-stop cost;

[0018] Calculate the historical predicted marginal power generation cost of the thermal power generating unit based on the historical predicted power generation cost;

[0019] Calculate the cost loss value based on the preset loss function for loss calculation of the historical predicted power generation cost;

[0020] Adjust the initial model based on the cost loss value to obtain the target power generation cost analysis model.

[0021] Further, input the target power generation parameter into the preset target power generation cost analysis model for processing to obtain the predicted marginal power generation cost of the thermal power generating unit, including:

[0022] Input the target power generation parameter into the preset target power generation cost analysis model, construct the cost causal structure equation based on the target power generation parameter, and calculate the predicted power generation cost;

[0023] Calculate the predicted marginal power generation cost of the thermal power generating unit based on the predicted power generation cost.

[0024] Further, construct the cost causal structure equation based on the target power generation parameter, and calculate the predicted power generation cost, including:

[0025] Determine the fuel cost variable, the environmental protection cost variable, the maintenance cost variable and the start-stop cost variable based on the target power generation parameter;

[0026] Construct the fuel cost causal structure equation based on the fuel cost variable, and calculate the predicted fuel cost;

[0027] Construct the environmental protection cost causal structure equation based on the environmental protection cost variable, and calculate the predicted environmental protection cost;

[0028] Construct the maintenance cost causal structure equation based on the maintenance cost variable, and calculate the predicted maintenance cost;

[0029] A start-stop cost causal structure equation is constructed based on the start-stop cost variable, and a predicted start-stop cost is calculated.

[0030] A predicted generation cost of the thermal power generating unit is calculated based on the predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost and the predicted start-stop cost.

[0031] Further, a causal analysis is performed based on the target generation parameter and the predicted marginal generation cost to obtain the cost influence reasons of the thermal power generating unit, including:

[0032] A causal relationship diagram is constructed based on the target generation parameter and the cost causal structure equation.

[0033] A causal analysis is performed on the predicted marginal generation cost based on the causal relationship diagram to obtain the cost influence reasons of the thermal power generating unit.

[0034] Further, the cost influence reasons include cost influence parameters and corresponding influence values, and the cost influence parameters correspond to parameters in the target generation parameter;

[0035] The target generation parameter is adjusted based on the cost influence reasons to obtain optimized generation parameters, so that the generation cost of the thermal power generating unit based on the optimized generation parameters is reduced, including:

[0036] A to-be-optimized parameter corresponding to the cost influence parameter is found out in the target generation parameter;

[0037] The to-be-optimized parameter is optimized based on the influence value to obtain an optimized parameter;

[0038] The optimized generation parameters of the thermal power generating unit are formed based on the optimized parameter and other generation parameters in the target generation parameter except the to-be-optimized parameter, so that the generation cost of the thermal power generating unit based on the optimized generation parameters is reduced.

[0039] In a second aspect, the present application further provides a generation cost optimization system, including:

[0040] An acquisition module is configured to acquire target generation parameters of a thermal power generating unit;

[0041] A parameter processing module is configured to input the target generation parameters into a preset target generation cost analysis model for processing to obtain a predicted marginal generation cost of the thermal power generating unit;

[0042] A causal analysis module is configured to perform a causal analysis based on the target generation parameters and the predicted marginal generation cost to obtain cost influence reasons of the thermal power generating unit;

[0043] A generation cost optimization module is configured to adjust the target generation parameters based on the cost influence reasons to obtain optimized generation parameters, and to control the thermal power generating unit to generate power based on the optimized generation parameters to reduce the generation cost.

[0044] In a third aspect, the present application provides a computing device, comprising a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor implements the steps of the power generation cost optimization method as described above when running the program.

[0045] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores instructions, and the instructions, when running on a terminal device, cause the terminal device to perform the steps of the power generation cost optimization method.

[0046] The present application has the following beneficial effects. Firstly, by inputting the target power generation parameters of the thermal power generating unit into the preset target power generation cost analysis model for processing, the predicted marginal power generation cost of the thermal power generating unit is obtained, so as to accurately reflect the cost change of the thermal power generating unit in the dynamic power generation process. Then, the causal analysis is performed based on the target power generation parameters and the predicted marginal power generation cost, so as to analyze the cost impact reasons causing the cost change, and the parameters in the target power generation parameters are adjusted based on the cost impact reasons, so as to obtain the optimized power generation parameters. The thermal power generating unit generates power based on the optimized power generation parameters, so as to reduce the power generation cost, thereby improving the accuracy of the cost optimization and providing a reliable cost reference for the thermal power enterprises participating in the electricity spot market. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 A flowchart of a power generation cost optimization method according to an example embodiment of the present application is shown.

[0048] Figure 2 A structural diagram of a power generation cost optimization system according to an example embodiment of the present application is shown. DETAILED DESCRIPTION

[0049] The following examples are further explanations and supplements of the present application and do not constitute any limitation on the present application.

[0050] At present, the existing marginal cost calculation methods mainly include the following types:

[0051] 1. Static calculation method based on standard coal consumption: this method simply calculates through the unit design coal consumption curve and the current coal price. The advantage is that the calculation is simple and intuitive, but it cannot reflect the cost change under the actual running state of the unit.

[0052] 2. Fitting method based on statistical regression: the cost curve is obtained by polynomial fitting of historical running data. This method considers the historical data, but the data quality requirement is high, and it is difficult to handle nonlinear relationships.

[0053] 3. Correction method based on empirical formula: superimpose various correction coefficients on the benchmark cost. This method is practical, but the determination of correction coefficients is more subjective and lacks scientific basis.

[0054] However, the existing marginal cost calculation methods have the following main defects:

[0055] 1. Lack of dynamic perception ability. The existing methods are mainly based on static models and cannot capture and respond to dynamic factors such as unit operating condition changes and environmental condition changes in real time, resulting in deviations between cost calculation and actual operating state. The present application realizes real-time perception and prediction of unit operating state through a deep learning model.

[0056] 2. Lack of causal analysis ability. The existing methods often treat each cost element separately, ignoring their interaction and causal relationship. For example, changes in fuel quality will affect both fuel cost and environmental cost, but the existing methods cannot capture this complex correlation. The present application introduces a causal reasoning framework to establish a causal graph between cost elements and achieve systematic analysis.

[0057] 3. Low data utilization efficiency. The existing methods mainly rely on simple mathematical statistics and cannot fully exploit the deep rules hidden in historical operation data. The present application uses deep learning technology to train an accurate cost prediction model through massive data.

[0058] 4. Single optimization dimension. The existing methods often only focus on fuel cost, ignoring other important dimensions such as environmental cost, start-stop cost, and maintenance cost. The present application constructs a full-dimensional cost model to achieve multi-dimensional collaborative optimization.

[0059] To solve the above problems, the embodiments of the present application provide a power generation cost optimization method, system, device and medium, which will be described in detail below.

[0060] The power generation cost optimization method provided by the embodiments of the present application can be specifically executed by a server. It should be noted that the server can be an independent server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms, etc. Basic cloud computing services, which are not limited here.

[0061] Please refer to Figure 1 , Figure 1 A power generation cost optimization method is shown in an exemplary embodiment of the present application, as shown in Figure 1 The present application provides a power generation cost optimization method, which includes:

[0062] S11, obtaining a target power generation parameter of the thermal power generating unit;

[0063] S12, inputting the target power generation parameter into a preset target power generation cost analysis model for processing to obtain a predicted marginal power generation cost of the thermal power generating unit;

[0064] S13, performing causal analysis based on the target power generation parameter and the predicted marginal power generation cost to obtain a cost influencing reason of the thermal power generating unit;

[0065] S14, adjusting the target power generation parameter based on the cost influencing reason to obtain an optimized power generation parameter, and controlling the thermal power generating unit to generate power based on the optimized power generation parameter to reduce the power generation cost.

[0066] The power generation cost optimization method provided in the embodiment can first input the target power generation parameter of the thermal power generating unit into the preset target power generation cost analysis model for processing to obtain the predicted marginal power generation cost of the thermal power generating unit, so as to accurately reflect the cost change of the thermal power generating unit in the dynamic power generation process. Then, the cost influencing reason causing the cost change is analyzed based on the target power generation parameter and the predicted marginal power generation cost, and the parameter in the target power generation parameter is adjusted based on the cost influencing reason to obtain the optimized power generation parameter. The thermal power generating unit is controlled to generate power based on the optimized power generation parameter to reduce the power generation cost, so as to improve the accuracy of cost optimization and provide a reliable cost reference for the thermal power enterprise to participate in the electricity spot market.

[0067] Optionally, the target power generation parameter of the thermal power generating unit is obtained, including:

[0068] obtaining an initial power generation parameter of the thermal power generating unit;

[0069] performing abnormality analysis on the initial power generation parameter to obtain abnormality indexes, the abnormality indexes including a data integrity index, an abnormal data proportion, a historical consistency index, and a sensor health state;

[0070] calculating a credibility of the initial power generation parameter based on preset abnormality weight parameters and the abnormality indexes;

[0071] adjusting the initial power generation parameter based on the credibility to obtain the target power generation parameter of the thermal power generating unit.

[0072] In the embodiment provided in the present application, the initial power generation parameters of the thermal power generating unit are analyzed for abnormalities to obtain abnormal indexes, the abnormal indexes include data integrity indexes, abnormal data proportions, historical consistency indexes, and sensor health statuses, and the credibility of the initial power generation parameters is calculated based on preset abnormal weight parameters and the abnormal indexes, the initial power generation parameters are adjusted to obtain target power generation parameters of the thermal power generating unit that can meet the credibility requirements, so that the data quality of the target power generation parameters can be effectively improved, and the accuracy of subsequent cost optimization of the thermal power generating unit based on the target power generation parameters can be further improved.

[0073] In the embodiment, the preset abnormal weight parameters include data integrity weights corresponding to the data integrity indexes, abnormal data weights corresponding to the abnormal data proportions, historical consistency weights corresponding to the historical consistency indexes, and health status weights corresponding to the sensor health statuses.

[0074] The data integrity index represents the ratio of the number of valid data points in the time window to the number of theoretical data points to be collected, for example, real-time control data (1 second time window): 60 points should be collected in theory (1 point per second), and 55 valid points are actually collected, so the data integrity index is 91.7%. The corresponding data integrity weight can be 40%.

[0075] The abnormal data proportion represents the proportion of data points marked as abnormal in the time window. For example, there are 60 points in a 1-minute time window, and 3 of them are abnormal points, so the abnormal data proportion is 5%. The corresponding abnormal data weight can be 35%.

[0076] The historical consistency index represents the cosine similarity between the current window data and the historical statistical characteristics of the same period. For example, the similarity between the current load rate average and the historical average is 0.85, so the historical consistency index is 85%. The corresponding historical consistency weight can be 15%.

[0077] The sensor health status represents the self-checking state of the sensor. For example, if the sensor self-checking state is normal, the sensor health status is 100%; if the sensor self-checking state is warning, the sensor health status is 50%; and if the sensor self-checking state is failure, the sensor health status is 0%. The corresponding health status weight can be 10%.

[0078] The credibility of the initial power generation parameters is calculated based on the preset abnormal weight parameters and the abnormal indexes, so the credibility can be: 91.7% x 40% + 5% x 35% + 85% x 15% + 100% x 10% = 148%.

[0079] In an example embodiment provided in the present application, the initial power generation parameters are adjusted based on the credibility to obtain the target power generation parameters of the thermal power generating unit, and the specific steps include:

[0080] dividing the credibility level based on the credibility: if the credibility is greater than or equal to 80%, the credibility level is high; if the credibility is 50% to 80%, the credibility level is medium; if the credibility is less than 50%, the credibility level is low;

[0081] when the credibility level is high, the initial power generation parameter is determined as the target power generation parameter of the thermal power generating unit;

[0082] when the credibility level is medium, the initial power generation parameter is manually reviewed or supplemented and corrected to obtain a corrected parameter, and the corrected parameter is determined as the target power generation parameter of the thermal power generating unit;

[0083] when the credibility level is low, a data alarm is triggered for the initial power generation parameter, a new initial power generation parameter of the thermal power generating unit is re-acquired, and the credibility is adjusted again to adjust the new initial power generation parameter to obtain the target power generation parameter of the thermal power generating unit.

[0084] In another example embodiment provided by the present application, the initial power generation parameter of the thermal power generating unit is acquired through the OPC UA (Open Platform Communications Unified Architecture) protocol and the power plant DCS (Distributed Control System) system and SIS (Safety Instrumented System) system docking to build a distributed data acquisition network.

[0085] When the initial power generation parameter is acquired, the hierarchical sampling frequency is set according to the data importance of each parameter in the initial power generation parameter, the key parameters are sampled every 1 second, the general parameters are sampled every minute, and the equipment state data is sampled every hour. Taking a 300 MW unit as an example, the key parameters such as boiler efficiency and steam turbine heat consumption are acquired every second, the general parameters such as coal quality and environmental protection facility operation are acquired every minute, and the equipment state data is acquired every hour.

[0086] In another example embodiment provided by the present application, the abnormal parameters in the initial power generation parameter are determined, and the proportion of abnormal data in the abnormal index is obtained by counting the proportion of the abnormal parameters in all parameters.

[0087] Then, the abnormal parameters in the initial power generation parameter are determined, and the specific steps can include:

[0088] The parameters in the initial power generation parameter are subjected to standardization calculation to obtain abnormal values, and the corresponding calculation formula is as follows:

[0089]

[0090] Z = (X - X) / St represents the abnormal value of the parameter at time t, X t represents the parameter (observation value) at time t, μ t represents the historical mean of the parameter, σ t represents the standard deviation of the parameter.

[0091] When |Z t |> set value, the corresponding parameter is determined as an abnormal parameter in the initial power generation parameter, and the set value can be 3.

[0092] For example, when the boiler efficiency suddenly drops from 93% to 75% at a certain time, the calculated |Z t | is -4.2, which exceeds the set threshold of ±3, triggering the abnormal processing procedure corresponding to the boiler efficiency to adjust the boiler efficiency.

[0093] In addition, the specific steps of determining the abnormal parameter in the initial power generation parameter can further include:

[0094] judging the abnormal situation of the parameter in the initial power generation parameter;

[0095] when the abnormal situation is any of the parameter abnormality, physical abnormality and rule abnormality, determining the parameter as an abnormal parameter.

[0096] wherein the parameter abnormality means that the abnormal value of the parameter is greater than the set value, the physical abnormality means that the parameter value of the parameter exceeds the reasonable range of equipment design or operation (such as negative coal consumption value, bearing temperature exceeding the safety threshold), and the rule abnormality means the abnormality determined by combining expert rules (such as parameter combination that should not appear under a specific working condition).

[0097] In another exemplary embodiment provided in the present application, if there is data missing in the parameter in the initial power generation parameter, a time series decomposition method is used to fill in the data of the same type to form complete initial power generation parameters, so as to effectively improve the data quality and feature expression ability of the subsequently obtained target power generation parameters.

[0098] The filling formula of the time series decomposition method for filling in the data of the same type is as follows:

[0099] X t = T t + S t + R t ;

[0100] wherein X t represents the observation value of the time series at time t, T t represents the trend item, S t represents the seasonal item, and R t represents the random item.

[0101] If the data missing is short-term missing (<1 hour), the trend item and seasonal item are mainly used for estimation; if the data missing is long-term missing (≥1 hour), the historical same period data is combined for filling.

[0102] In order to maintain the time consistency of the target power generation parameters, a time window alignment method is used to process the time consistency of different source data in the initial power generation parameters. For example, a 1-second window is used for real-time control data, a 1-minute window is used for slowly varying parameters, and the credibility of the data is established for subsequent modeling analysis.

[0103] Optionally, the training process of the target power generation cost analysis model is as follows:

[0104] Obtain historical power generation parameters of the thermal power generating unit;

[0105] Input the historical power generation parameters into the initial model for training to obtain historical predicted power generation costs of the thermal power generating unit, the historical predicted power generation costs including historical predicted fuel costs, historical predicted environmental protection costs, historical predicted maintenance costs and historical predicted start-stop costs;

[0106] Calculate the historical predicted marginal power generation costs of the thermal power generating unit based on the historical predicted power generation costs;

[0107] Loss calculation is performed on the historical predicted power generation costs based on a preset loss function to obtain a cost loss value;

[0108] The calculation formula of the cost loss value is as follows:

[0109] L=w f ·MSE(C f )+w e ·MSE(C e )+w m ·MSE(C m )+w s ·MSE(C s )+λ·R(θ);

[0110] Wherein, L represents the cost loss value, w f represents the fuel cost weight, C f represents the historical predicted fuel cost, w e represents the environmental protection cost weight, C e represents the historical predicted environmental protection cost, w m represents the maintenance cost weight, C m represents the historical predicted maintenance cost, w s represents the start-stop cost weight, C s represents the historical predicted start-stop cost, λ represents a regularization coefficient, and R(θ) represents a regularization term.

[0111] Adjust the initial model based on the cost loss value to obtain a target power generation cost analysis model.

[0112] In the embodiment provided in the present application, the historical predicted power generation cost of the thermal power generating unit is obtained by pre-inputting the historical power generation parameters of the thermal power generating unit into the initial model for training, and the initial model is adjusted based on the cost loss value obtained by loss calculation on the historical predicted power generation cost based on the preset loss function, so that the target power generation cost analysis model obtained by adjustment can meet the model loss requirement, the accuracy of the target power generation cost analysis model is improved, and the accuracy of subsequent cost optimization of the thermal power generating unit based on the target power generation cost analysis model is further improved.

[0113] In an example embodiment provided in the present application, the target power generation cost analysis model is a deep learning model based on the Transformer architecture, which includes a multi-layer encoder and a corresponding decoder structure. The encoder adopts a multi-head self-attention mechanism to process input features:

[0114]

[0115] MultiHead(Q,K,V)=Concat(head1,…,head h )W O ;

[0116] wherein Attention(Q,K,V) represents the attention mechanism, Q, K, and V represent the query matrix, the key matrix, and the value matrix respectively, softmax represents the normalization operation for converting the similarity into a probability distribution (attention weight with a sum of 1), T represents the transpose symbol, QK T represents finding the matching degree between the query matrix and the key matrix, highlighting the key features (such as high load), represents the square root of the dimension of the key matrix (such as 64 or 128), which is used to scale the dot product to avoid getting too large a value, MultiHead(Q,K,V) represents the output of the multi-head attention, and Concat represents the concatenation operation, for adaptive attention to different feature dimensions, W O represents the linear transformation weight matrix.

[0117] Optionally, the target power generation parameters are input into the preset target power generation cost analysis model for processing to obtain the predicted marginal power generation cost of the thermal power generating unit, including:

[0118] The target power generation parameters are input into the preset target power generation cost analysis model, a cost causal structure equation is constructed based on the target power generation parameters, and the predicted power generation cost is calculated;

[0119] The predicted marginal power generation cost of the thermal power generating unit is calculated based on the predicted power generation cost.

[0120] In this embodiment, the target power generation parameter is input into a preset target power generation cost analysis model, a cost causal structure equation is constructed based on the target power generation parameter, and a predicted power generation cost is calculated. The predicted marginal power generation cost of the thermal power generating unit is calculated based on the predicted power generation cost, which facilitates subsequent targeted causal analysis based on the predicted marginal power generation cost, so as to analyze the cost influence reasons causing the cost change, and to perform targeted cost optimization, thereby improving the accuracy of cost optimization.

[0121] In this embodiment, a cost causal structure equation is constructed based on the target power generation parameter by using a structural equation model (SEM).

[0122] In an example embodiment provided by the present application, the predicted marginal power generation cost of the thermal power generating unit is calculated based on the predicted power generation cost, which includes:

[0123] The predicted marginal fuel cost, the predicted marginal environmental protection cost, the predicted marginal maintenance cost, and the predicted marginal start-stop cost are calculated based on the predicted power generation cost.

[0124] The sum of the predicted marginal fuel cost, the predicted marginal environmental protection cost, the predicted marginal maintenance cost, and the predicted marginal start-stop cost is determined as the predicted marginal power generation cost.

[0125] For example, the predicted marginal fuel cost is 9.6 yuan / MWh, the predicted marginal environmental protection cost is 23.7 yuan / MWh, the predicted marginal maintenance cost is 10.4 yuan / MWh, and the predicted marginal start-stop cost is 5.5 yuan / MWh. Therefore, the predicted marginal power generation cost is 49.2 yuan / MWh.

[0126] Optionally, the cost causal structure equation is constructed based on the target power generation parameter, and the predicted power generation cost is calculated, which includes:

[0127] The fuel cost variable, the environmental protection cost variable, the maintenance cost variable, and the start-stop cost variable are determined based on the target power generation parameter.

[0128] The fuel cost causal structure equation is constructed based on the fuel cost variable, and the predicted fuel cost is calculated.

[0129] The environmental protection cost causal structure equation is constructed based on the environmental protection cost variable, and the predicted environmental protection cost is calculated.

[0130] The maintenance cost causal structure equation is constructed based on the maintenance cost variable, and the predicted maintenance cost is calculated.

[0131] The start-stop cost causal structure equation is constructed based on the start-stop cost variable, and the predicted start-stop cost is calculated.

[0132] The predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost and the predicted start-stop cost are calculated based on the predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost and the predicted start-stop cost.

[0133] In the embodiment provided in the present application, the fuel cost causal structure equation, the environmental protection cost causal structure equation, the maintenance cost causal structure equation and the start-stop cost causal structure equation are constructed based on the target power generation parameter, and the predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost and the predicted start-stop cost are calculated after the construction is completed, and the predicted power generation cost is calculated based on the four costs. In this way, the target power generation parameter and the final predicted power generation cost of the thermal power generating unit are dynamically associated through the causal structure equation, which facilitates the subsequent causal tracing of the cost influence reasons, so that the cost influence reasons can be intuitively and accurately determined, the optimization precision of the target power generation parameter is improved, and the precision of the cost optimization is further improved.

[0134] In the embodiment, the sum of the predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost and the predicted start-stop cost is equal to the predicted power generation cost.

[0135] In an example embodiment provided in the present application, the target power generation parameter includes a target unit operation parameter, a target fuel parameter, a target environmental protection parameter and a target equipment state parameter, the target unit operation parameter includes a real-time load, a power generation coal consumption curve, a boiler efficiency, a steam turbine heat consumption curve (including a unit heat consumption), a flue gas system resistance curve and a unit output (including a real-time output and a rated output), the target fuel parameter includes coal quality test indexes (including a received basis moisture, a received basis ash content, a received basis volatile matter and a received basis fixed carbon), a stock quantity and a market price (a coal price), the target environmental protection parameter includes a desulfurizing agent consumption, a denitration agent consumption (a denitration efficiency can be calculated), a dust removal efficiency, and the target equipment state parameter includes a bearing temperature, a vibration intensity, a pressure, an operation log (including an operation time accumulation), a fault frequency and a start-stop cost variable (a start-stop frequency, a start-stop time consumption and a start-stop energy consumption). In this way, the target power generation parameter can pay attention to multiple cost influence factor dimensions of the thermal power generating unit, the cost influence factors are comprehensively perceived, the feature expression capability of the target power generation parameter is improved, and the accuracy of the predicted marginal power generation cost of the thermal power generating unit based on the target power generation parameter is improved.

[0136] The fuel cost variable, the environmental protection cost variable, the maintenance cost variable and the start-stop cost variable are determined based on the target power generation parameter, and the specific steps include:

[0137] The unit load rate and the coal quality comprehensive index are calculated based on the target power generation parameter.

[0138] The calculation formula of the unit load rate a is: a = P / P n; wherein P represents the real-time output included in the target power generation parameter, P n represents the rated output included in the target power generation parameter;

[0139] The formula for calculating the coal quality comprehensive index λ is: λ = w1·M ar +w2·A ar +w3·V ar +w4·FC ar ; wherein w1 represents the received base moisture weight coefficient, M ar represents the received base moisture, w2 represents the received base ash weight coefficient, A ar represents the received base ash, w3 represents the received base volatile matter weight coefficient, V ar represents the received base volatile matter, and w4 represents the received base fixed carbon weight coefficient, FC ar represents the received base fixed carbon;

[0140] Based on the unit load rate, the coal quality comprehensive index, and the boiler efficiency, the unit heat consumption, and the coal price included in the target power generation parameter, a fuel cost variable is formed; as can be seen, the fuel cost variable is derived from the target fuel parameters and the unit operation data of the thermal power unit;

[0141] Based on the denitration efficiency, the dust removal efficiency, the desulfurizing agent consumption, and the denitration agent consumption included in the target power generation parameter, an environmental protection cost variable is formed; as can be seen, the environmental protection cost variable is derived from the target environmental protection parameters of the thermal power unit.

[0142] Based on the bearing temperature, the vibration intensity, the pressure, the running time accumulation, and the fault frequency included in the target power generation parameter, a maintenance cost variable is formed; as can be seen, the maintenance cost variable is derived from the target equipment state parameters of the thermal power unit.

[0143] Based on the start-stop frequency, the start-stop time consumption, and the start-stop energy consumption included in the target power generation parameter, a start-stop cost variable is formed; as can be seen, the start-stop cost variable is derived from the target equipment state parameters of the thermal power unit.

[0144] In this embodiment, through the coal quality comprehensive index λ, the influence of the coal quality on the power generation cost can be comprehensively evaluated, for example, when the coal M ar = 12%, A ar = 25%, V ar = 28%, and FC ar = 35% used by a certain thermal power unit, and the corresponding weight coefficients are w1 = 0.2, w2 = 0.3, w3 = 0.2, and w4 = 0.3, the coal quality comprehensive index λ = 0.85 is calculated, indicating that the current coal quality is at a good level.

[0145] At the same time, the unit load rate is 80% (in the economic output interval), and the coal quality comprehensive index is 0.85 (indicating that the coal quality is good, M ar= 12% leads to moderate dryness loss, A ar = 25% indicates ash loss in a controllable range.

[0146] In another example embodiment provided in the present application, first, a fuel cost causal structure equation is constructed based on fuel cost variables, which is expressed as:

[0147] (C f = P·h(P)·c coal + u3) or (C f = P·h(P)·c coal / 10 6 ); / / relationship between heat consumption and predicted fuel cost

[0148] or (h(P) = h0·(a·α 2 +b·α+c) / η b ); / / relationship between boiler efficiency and heat consumption

[0149] η b = β0+β1λ+u1; / / relationship between coal quality comprehensive index and boiler efficiency

[0150] wherein C f represents predicted fuel cost, P represents real-time output (MW), h(P) represents current unit heat consumption (kJ / kWh), c coal represents coal price (yuan / ton), 10 6 represents unit conversion coefficient, h0 represents theoretical minimum heat consumption under design conditions (kJ / kWh), α represents unit load rate (α = P / P n ), P n represents target power generation parameters including rated output, a, b, c represent steam turbine heat consumption curve coefficients (determined through experiments), η b represents current boiler efficiency, i.e., the ratio of boiler output heat to input fuel heat, λ represents coal quality comprehensive index, β0 represents baseline efficiency of the boiler, which is the boiler efficiency under standard coal quality conditions, β1 represents the influence coefficient of the coal quality comprehensive index, which is the sensitivity of coal quality change to boiler efficiency, and β2 represents efficiency influence coefficient. λ represents current coal quality comprehensive index, reflecting the quality of coal. u1 represents observed factor random error term, containing the influence of other unobserved factors, u2 represents heat consumption random error term, and u3 represents fuel cost random error term.

[0151] For example, under the baseline conditions of a 300 MW thermal power unit: β0 = 93%, β1 = 0.05, and λ = 0.85, the calculated boiler efficiency η b = 93% + 0.05 × 0.85 ≈ 93.4%. After the change of coal quality: λ drops to 0.80 (due to Mar Increased to 15%), the boiler efficiency η is calculated by substituting it into the formula b A decrease of 0.2 percentage points causes the heat consumption to increase by 8kJ / kWh, which ultimately leads to an increase of 12 yuan / MWh in the calculated predicted fuel cost, that is, the predicted marginal fuel cost is 12 yuan / MWh.

[0152] For example, a 300MW thermal power unit operates at 80% load rate, with basic parameters: real-time output P = 240MW, rated output P n =300MW, unit load factor α = 240 / 300 = 0.8, design theoretical minimum heat rate h0 = 8400kJ / kWh, turbine heat rate curve coefficients a = 0.15, b = -0.25, c = 1.1, then the current boiler efficiency η b =93.4% (calculated by the comprehensive coal quality index λ = 0.85), coal price c coal =750 yuan / ton. The calculation process is: load correction coefficient = a×α 2 +b×α+c=0.15×0.8 2 +(-0.25)×0.8+1.1=0.096-0.2+1.1=0.996, current unit heat consumption h(P)=8400×0.996 / 0.934=8960kJ / kWh; predicted fuel cost C f =240×8960×750 / 10^6=1612.8 yuan / hour. Therefore, the predicted marginal fuel cost (the additional fuel cost for an additional 1MW) is calculated as follows: When the real-time output increases from 240MW to 241MW: new unit load factor = 241 / 300 = 0.803, new unit heat rate = 8400×0.9965 / 0.934 = 8965kJ / kWh, new predicted fuel cost = 241×8965×750 / 10^6 = 1622.4 yuan / hour, so the predicted marginal fuel cost = 1622.4-1612.8 = 9.6 yuan / MWh.

[0153] Secondly, the causal structural equation of environmental protection cost is constructed based on the environmental protection cost variable and is expressed as follows:

[0154] C e =C 脱硫 +C 脱硝 +C 除尘 ;

[0155] C 脱硫 =P×gypsum production coefficient×gypsum processing cost+P×desulfurizer price;

[0156] Desulfurizer price = limestone consumption × limestone price;

[0157] C 脱硝=P×SCR catalyst depreciation+P×ammonia consumption×ammonia price;

[0158] C 除尘 = power consumption of dust collector × factory electricity price + P × amortization of filter bag replacement;

[0159] Among them, C e represents the predicted environmental protection cost, C 脱硫 represents the desulfurization cost, C 脱硝 represents the denitrification cost, C 除尘 represents the dust removal cost, and P represents the unit output.

[0160] Taking a 300MW thermal power unit as an example, the desulfurization cost calculation is: P = 240 tons, gypsum production coefficient = 0.03 tons / MWh (calculated based on coal ash content), gypsum processing fee = 150 yuan / ton, limestone consumption = 2.5 kg / MWh (based on SO2 removal), limestone price = 300 yuan / ton. Therefore, the desulfurization cost = 240 × 0.03 × 150 + 240 × 2.5 × 0.3 = 1080 + 180 = 1260 yuan / hour. The denitrification cost calculation is: P = 240 tons, SCR catalyst depreciation = 8 yuan / MWh, ammonia consumption = 1.8 kg / MWh (based on NOx removal), ammonia price = 2500 yuan / ton. Therefore, the denitrification cost = 240 × 8 + 240 × 1.8 × 2.5 = 1920 + 1080 = 3000 yuan / hour. Calculation of dust removal cost: Dust collector power consumption = 2MW, factory electricity price = 0.35 yuan / kWh, filter bag replacement amortization = 3 yuan / MWh, P = 240 tons, then dust removal cost = 2000×0.35+240×3=700+720=1420 yuan / hour.

[0161] Based on the above calculations, the environmental protection cost C is predicted e =1260+3000+1420=5680 yuan / hour. When the real-time output increases by 1MW, the processing capacity of the environmental protection facilities increases, and the predicted marginal environmental protection cost is about 23.7 yuan / MWh.

[0162] Again, based on the maintenance cost variable, the maintenance cost causal structural equation is constructed as follows:

[0163] C m =C 计划维护 +C 状态维护 ;

[0164] C 计划维护 =Σ(annual maintenance cost of equipment i / annual operating hours);

[0165] C 状态维护 =Σ(deterioration degree of equipment i × maintenance cost coefficient);

[0166] Among them, Cm represents the predicted maintenance cost, C 计划维护 represents the planned maintenance cost, C 状态维护 represents the state maintenance cost.

[0167] For example, the planned maintenance cost C 计划维护 of the main equipment includes: the annual maintenance cost of the boiler system is 8 million yuan, which is allocated to 913 yuan / hour per hour, the annual maintenance cost of the steam turbine system is 6 million yuan, which is allocated to 685 yuan / hour per hour, the annual maintenance cost of the generator system is 2 million yuan, which is allocated to 228 yuan / hour per hour, and the annual maintenance cost of auxiliary equipment is 4 million yuan, which is allocated to 456 yuan / hour per hour. The state maintenance cost is calculated based on monitoring data such as equipment vibration and temperature: the turbine bearing temperature is too high, the degradation coefficient is 1.2, and the additional cost is 100x1.2=120 yuan / hour; the boiler heating surface is coked, the degradation coefficient is 1.15, and the additional cost is 80x1.15=92 yuan / hour.

[0168] After the above calculation, the predicted maintenance cost C 计划维护 is 913+685+228+456+120+92=2494 yuan / hour, then the equipment wear and tear is positively correlated with the load, and the predicted marginal maintenance cost is about 10.4 yuan / MWh.

[0169] Then, the start-stop cost causal structure equation is constructed based on the start-stop cost variable, which is represented as:

[0170] C s =(start-up cost+shutdown cost)×start-stop frequency coefficient;

[0171] Wherein, C s represents the predicted start-stop cost, the start-up cost includes fuel consumption cost (fuel consumption cost during start-up process), auxiliary machine power consumption cost (auxiliary machine power consumption cost during start-up), and equipment wear cost (life loss cost of hot components such as main steam pipe caused by thermal stress), and the shutdown cost includes waste heat loss cost and equipment cooling loss.

[0172] For example, the start-up cost includes: fuel consumption cost = 25 tons x 750 yuan / ton = 187,500 yuan, auxiliary machine power consumption cost = 10 MWh x 0.35 yuan / kWh = 35,000 yuan, equipment wear cost = 80,000 yuan, and the single start-up cost = 1.875 + 3.5 + 8 = 13.375 million yuan. The shutdown cost includes: waste heat loss cost = 10 tons x 750 yuan / ton = 7,500 yuan, equipment cooling wear cost = 50,000 yuan, and the single shutdown cost = 0.75 + 5 = 5.75 million yuan. The start-stop frequency coefficient is calculated based on the unit operation mode and the grid scheduling demand: the start-stop frequency coefficient corresponding to the basic load operation = 0.1 (rare start-stop), the start-stop frequency coefficient corresponding to the peak regulation operation = 0.8 (frequent start-stop), and the start-stop frequency coefficient corresponding to the deep peak regulation operation = 1.2 (frequent start-stop).

[0173] According to the above calculation, assuming that the unit is in peak regulation operation mode and the monthly start-stop is 8 times, the predicted start-stop cost C s = (13.375 + 5.75) million x 8 x 0.8 / (30 x 24) = 1,700 yuan / hour. In the deep peak regulation period, the increase of 1 MW may affect the start-stop decision, and the predicted marginal start-stop cost is about 5.5 yuan / MWh.

[0174] Alternatively, the causal analysis is performed based on the target power generation parameter and the predicted marginal power generation cost to obtain the cost influence reasons of the thermal power unit, including:

[0175] A causal relationship diagram is constructed based on the target power generation parameter and the cost causal structure equation;

[0176] The causal relationship diagram is used to perform the causal analysis on the predicted marginal power generation cost to obtain the cost influence reasons of the thermal power unit.

[0177] In the embodiment provided in the present application, the causal relationship diagram constructed based on the target power generation parameter and the cost causal structure equation is used to perform the causal analysis on the predicted marginal power generation cost, and through this layer-by-layer causal analysis, the cost influence reasons of the thermal power unit are obtained. In this way, not only the direct reasons (for example, coal consumption increase) of the change of the predicted marginal power generation cost can be identified, but also the root causes (for example, coal quality change) can be traced back, so that accurate decision support can be provided for subsequent cost optimization, thereby improving the cost optimization precision.

[0178] In an example embodiment provided in the present application, the causal relationship diagram G is represented as C = (V, E), V represents the vertex set of the causal relationship diagram, and E represents the causal relationship between variables.

[0179] The vertex set V = {coal quality characteristics (M ar , A ar , V ar , FC ar ), environmental protection characteristics (ηs , η n , η d ), operating characteristics (α, η b , h), cost indicators (C f , C e , C m , C s )}, wherein M ar represents the received base moisture, A ar represents the received base ash, V ar represents the received base volatile matter, FC ar represents the received base fixed carbon, η s is the desulfurization efficiency, η n is the denitration efficiency, η d is the dust removal efficiency, α represents the unit load rate, η b represents the boiler efficiency, h represents the unit heat consumption, C f represents the predicted fuel cost, C e represents the predicted environmental protection cost, C m represents the predicted maintenance cost, and C s represents the predicted start-stop cost.

[0180] The causal relationship between variables includes a fuel cost related causal relationship chain, an environmental protection cost related causal relationship chain, a maintenance cost related causal relationship chain, and a start-stop cost related causal relationship chain. The fuel cost related causal relationship chain can be represented as: coal quality characteristics (M ar , A ar , V ar , FC ar )→coal quality comprehensive index (λ)→boiler efficiency (η b )→heat consumption (h(P))→fuel cost (C f ), for example, an increase of 5% in ash content leads to an increase of 1.2 yuan / MWh in fuel cost; the environmental protection cost related causal relationship chain can be represented as: unit output (P)→desulfurizer / ammonia water consumption→environmental protection cost (C e ), for example: an increase of 20 MW in output leads to an increase of 200 yuan / h in denitration cost; the maintenance cost related causal relationship chain can be represented as: equipment state (vibration, temperature)→deterioration coefficient→maintenance cost (C m ), for example: an increase of vibration from 2 mm / s to 4 mm / s leads to an increase of 15% in maintenance cost; and the start-stop cost related causal relationship chain can be represented as: start-stop frequency→thermal stress→start-stop cost (C s ), for example: 10 times of monthly start-stop leads to an increase of 30 yuan / h in start-stop cost.

[0181] In another example embodiment provided in the present application, the causal relationship graph and the relationship of quantifying causal effects based on a structural equation model (SEM) can be as follows:

[0182] Causal graph construction → Determine variable dependency → Define SEM equation → Parameter estimation (e.g. least squares) → Calculate causal effect strength. In simple terms, the causal graph is only a qualitative description of the dependency of variables, which defines the cost-related variables and their causal relationship structure, and the SEM equation quantitatively describes the relationship and gives each edge a mathematical expression.

[0183] Therefore, based on the structural equation model (SEM), the causal effect can be quantified:

[0184] x i =f i (pa i ,u i ),i=1,…,n;

[0185] where x i represents the variables in the vertex set of the causal graph (such as fuel cost), pa i represents the direct cause set of x i , f i represents the variable generating function of x i , u i represents the exogenous noise term (random error or unobserved factor), and n represents the total number of endogenous variables.

[0186] Exogenous noise term u i represents random factors in the model that are not observed or errors that are not explicitly modeled, including: unobserved variables (such as coal distribution uniformity, local coking of the burner, etc.), measurement errors (sensor accuracy limitations, data acquisition noise), and system inherent randomness (such as small fluctuations in coal composition). Exogenous noise term u i is used to quantify the part of the variable fluctuation that the model cannot fully explain, ensuring the statistical integrity of the structural equation. Exogenous noise term u i is not a fixed value, but a random variable that changes dynamically with working conditions and environment, and is calculated from the residual error after model fitting, and follows a normal distribution.

[0187] The direct cause set in the above formula can correspond to Table 1:

[0188] Table 1

[0189]

[0190] Optionally, the cost impact cause includes a cost impact parameter and a corresponding impact value, and the cost impact parameter corresponds to a parameter in the target power generation parameter;

[0191] Adjust the target power generation parameter based on the cost impact cause to obtain an optimized power generation parameter, so that the power generation cost of the thermal power generating unit based on the optimized power generation parameter is reduced, including:

[0192] find out the to-be-optimized parameter corresponding to the cost impact parameter in the target power generation parameter;

[0193] optimize the to-be-optimized parameter based on the impact value, to obtain an optimized parameter;

[0194] form optimized power generation parameters of the thermal power generating unit based on the optimized parameter and other power generation parameters in the target power generation parameter except the to-be-optimized parameter, so that the cost of power generation of the thermal power generating unit based on the optimized power generation parameters is reduced.

[0195] In the embodiment provided in the present application, firstly, since the cost impact reason includes the cost impact parameter and the corresponding impact value, and the cost impact parameter corresponds to the parameter in the target power generation parameter, the to-be-optimized parameter corresponding to the cost impact parameter is found out in the target power generation parameter, and the to-be-optimized parameter is optimized based on the impact value to obtain the optimized parameter, which can improve the accuracy of the optimized parameter. Secondly, the optimized power generation parameters of the thermal power generating unit are formed based on the optimized parameter and other power generation parameters in the target power generation parameter except the to-be-optimized parameter, so that the thermal power generating unit is optimized for power generation based on the optimized power generation parameters to reduce the power generation cost, thereby improving the accuracy of cost optimization.

[0196] In an example embodiment provided in the present application, the causal relationship diagram can be visually displayed, so that the predicted marginal power generation cost can be intuitively analyzed based on the causal relationship diagram to obtain the cost impact reason of the thermal power generating unit for display. For example, the cost impact reason obtained by performing causal composition analysis on the predicted marginal power generation cost of a certain thermal power generating unit is: the fuel cost accounts for 65% (of which, the coal consumption accounts for 40%, and the coal price accounts for 25%), the environmental protection cost accounts for 20% (mainly from desulfurization and denitrification agent consumption), and the maintenance cost accounts for 15%. Based on this analysis, parameter adjustment is preferentially performed from the aspects of coal consumption control and environmental protection agent consumption optimization to reduce the marginal cost, thereby improving the efficiency of cost optimization.

[0197] The power generation cost optimization method provided in the application realizes the intelligentization of the whole process from data acquisition to optimization decision based on the innovative integration of deep learning and causal reasoning. Through the distributed data acquisition network and multi-scale feature engineering, the unit operation state can be sensed in real time. By combining the deep learning model with the Transformer architecture and the causal reasoning, the deep understanding of the cost influence mechanism is realized. At the same time, a full-dimensional cost calculation and optimization system is constructed, which not only considers the traditional fuel cost, but also includes environmental protection cost, maintenance cost, start-stop cost, etc. in a unified framework. Through the multi-head attention mechanism, the dynamic correlation between each dimension is captured, and the cost influence path is found by using the structured causal model, which realizes the breakthrough from correlation analysis to causal relationship mining, greatly improves the accuracy and interpretability of cost prediction and optimization, and thus can improve the competitiveness of thermal power units participating in the electricity spot market.

[0198] Please refer to Figure 2 , Figure 2 A power generation cost optimization system is shown in an example embodiment of the application, as shown in Figure 2 The application provides a power generation cost optimization system 200, which includes:

[0199] The acquisition module 201 is configured to acquire the target power generation parameter of the thermal power unit.

[0200] The parameter processing module 202 is configured to input the target power generation parameter into a preset target power generation cost analysis model for processing to obtain the predicted marginal power generation cost of the thermal power unit.

[0201] The causal analysis module 203 is configured to perform causal analysis based on the target power generation parameter and the predicted marginal power generation cost to obtain the cost influence reason of the thermal power unit.

[0202] The power generation cost optimization module 204 is configured to adjust the target power generation parameter based on the cost influence reason to obtain the optimized power generation parameter, and control the thermal power unit to generate power based on the optimized power generation parameter to reduce the power generation cost.

[0203] The power generation cost optimization system 200 of the embodiment provided in the application firstly inputs the target power generation parameters of the thermal power generating unit acquired by the acquisition module 201 into the preset target power generation cost analysis model for processing by the parameter processing module 202, to obtain the predicted marginal power generation cost of the thermal power generating unit, so as to accurately reflect the cost change of the thermal power generating unit in the dynamic power generation process. Then, the cause and effect analysis module 203 performs cause and effect analysis based on the target power generation parameters and the predicted marginal power generation cost, to analyze the cost impact reasons causing the cost change, and the power generation cost optimization module 204 adjusts the parameters in the target power generation parameters based on the cost impact reasons, to obtain the optimized power generation parameters, controls the thermal power generating unit to generate power based on the optimized power generation parameters, to reduce the power generation cost, thereby improving the accuracy of cost optimization and providing reliable cost reference for the thermal power enterprises participating in the electricity spot market.

[0204] Optionally, the acquisition module 201 is specifically configured to:

[0205] acquire initial power generation parameters of the thermal power generating unit;

[0206] perform abnormality analysis on the initial power generation parameters, to obtain abnormality indexes, the abnormality indexes including a data integrity index, an abnormal data proportion, a historical consistency index and a sensor health status;

[0207] calculate a credibility of the initial power generation parameters based on preset abnormality weight parameters and the abnormality indexes;

[0208] adjust the initial power generation parameters based on the credibility, to obtain the target power generation parameters of the thermal power generating unit.

[0209] Optionally, the system further comprises a training module, the training module being specifically configured to:

[0210] acquire historical power generation parameters of the thermal power generating unit;

[0211] input the historical power generation parameters into an initial model for training, to obtain historical predicted power generation costs of the thermal power generating unit, the historical predicted power generation costs including a historical predicted fuel cost, a historical predicted environmental protection cost, a historical predicted maintenance cost and a historical predicted start-stop cost;

[0212] calculate a historical predicted marginal power generation cost of the thermal power generating unit based on the historical predicted power generation costs;

[0213] perform loss calculation on the historical predicted power generation costs based on a preset loss function, to obtain a cost loss value;

[0214] adjust the initial model based on the cost loss value, to obtain the target power generation cost analysis model.

[0215] Optionally, the parameter processing module 202 is specifically configured to:

[0216] input the target power generation parameter into a preset target power generation cost analysis model, construct a cost causal structure equation based on the target power generation parameter, and calculate a predicted power generation cost;

[0217] The predicted marginal power generation cost of the thermal power generating unit is calculated based on the predicted power generation cost.

[0218] Optionally, the parameter processing module 202 is specifically configured to:

[0219] determine a fuel cost variable, an environmental protection cost variable, a maintenance cost variable, and a start-stop cost variable based on the target power generation parameter;

[0220] construct a fuel cost causal structure equation based on the fuel cost variable, and calculate a predicted fuel cost;

[0221] construct an environmental protection cost causal structure equation based on the environmental protection cost variable, and calculate a predicted environmental protection cost;

[0222] construct a maintenance cost causal structure equation based on the maintenance cost variable, and calculate a predicted maintenance cost;

[0223] construct a start-stop cost causal structure equation based on the start-stop cost variable, and calculate a predicted start-stop cost;

[0224] The predicted power generation cost of the thermal power generating unit is calculated based on the predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost, and the predicted start-stop cost.

[0225] Optionally, the causal analysis module 203 is specifically configured to:

[0226] construct a causal relationship diagram based on the target power generation parameter and the cost causal structure equation;

[0227] perform causal analysis on the predicted marginal power generation cost based on the causal relationship diagram, to obtain a cost influence reason of the thermal power generating unit.

[0228] Optionally, the cost influence reason includes a cost influence parameter and a corresponding influence value, and the cost influence parameter corresponds to a parameter in the target power generation parameter;

[0229] The power generation cost optimization module 204 is specifically configured to:

[0230] adjust the target power generation parameter based on the cost influence reason, to obtain an optimized power generation parameter, so that the power generation cost of the thermal power generating unit based on the optimized power generation parameter is reduced, including:

[0231] finding out a to-be-optimized parameter corresponding to the cost influence parameter in the target power generation parameter;

[0232] The to-be-optimized parameter is optimized based on the influence value, to obtain an optimized parameter;

[0233] Based on the optimized parameter and other parameters in the target power generation parameter except the to-be-optimized parameter, an optimized power generation parameter of the thermal power generating unit is formed, so that the power generation cost of the thermal power generating unit based on the optimized power generation parameter is reduced.

[0234] It should be noted that the power generation cost optimization system provided in the above embodiments and the power generation cost optimization method provided in the above embodiments belong to the same concept, wherein the specific manner in which each module and unit performs operations has been described in detail in the method embodiments, which will not be described here. In actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the system is divided into different functional modules to complete all or part of the functions described above, and this is not limited herein.

[0235] The computing device provided in the embodiments of the present application includes a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, it implements some or all steps of the above power generation cost optimization method.

[0236] Correspondingly, the program is computer software, and the parameters and steps in the above computing device of the present application can refer to the parameters and steps in the above embodiments of the power generation cost optimization method, which will not be described here.

[0237] The computer readable storage medium in the embodiments of the present application stores instructions. When the instructions are executed, the steps of the above power generation cost optimization method are executed.

[0238] The computer readable storage medium can be a transient computer readable storage medium or a non-transient computer readable storage medium.

[0239] The technical solutions of the embodiments of the present disclosure can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes one or more instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method of the embodiments of the present disclosure. The aforementioned computer readable storage medium can be a non-transient computer readable storage medium, including a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes, or a transient computer readable storage medium.

[0240] The flow diagrams and block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of present application. It will be understood that each block of the flow diagrams and / or block diagrams, and combinations of blocks in the flow diagrams and / or block diagrams, can be implemented by

[0241] Those skilled in the art will understand that the application can be implemented as a system, method, or computer program product. Therefore, the present disclosure can be embodied in the form of a hardware-only implementation, a software-only implementation, or a combination of hardware and software implementations. The present disclosure can be embodied in the form of a computer program product that includes a computer readable storage medium having computer readable program code embodied in it. The computer readable storage medium can be, but is not limited to, a magnetic disk, optical disk, solid state drive, magnetic tape, or any other computer readable storage medium.

[0242] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the present application. In the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, different embodiments or examples described in the specification and the features of different embodiments or examples can be combined and combined by those skilled in the art without contradiction.

[0243] Although the embodiments of the present application have been shown and described above, it is to be understood that the above-described embodiments are exemplary, and are not to be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-described embodiments within the scope of the present application.

Claims

1. A method for optimizing power generation costs, characterized in that: include: Obtain target power generation parameters of thermal power units; Inputting the target power generation parameters into a preset target power generation cost analysis model for processing to obtain a predicted marginal power generation cost of the thermal power unit; Performing a cause-and-effect analysis based on the target power generation parameters and the predicted marginal power generation cost to obtain the cost impact factors of the thermal power unit; The target power generation parameters are adjusted based on the cost impact factors to obtain optimized power generation parameters, and the thermal power generation unit is controlled to generate electricity based on the optimized power generation parameters to reduce power generation costs.

2. The method according to claim 1, characterized in that The step of obtaining target power generation parameters of the thermal power generation unit includes: Obtain the initial power generation parameters of the thermal power unit; Performing an abnormality analysis on the initial power generation parameters to obtain abnormality indicators, wherein the abnormality indicators include a data integrity indicator, an abnormal data ratio, a historical consistency indicator, and a sensor health status; Calculating the credibility of the initial power generation parameters based on a preset abnormality weight parameter and the abnormality index; The initial power generation parameters are adjusted based on the credibility to obtain target power generation parameters of the thermal power unit.

3. The method according to claim 1, characterized in that The training process of the target power generation cost analysis model is as follows: Obtaining historical power generation parameters of the thermal power unit; Inputting the historical power generation parameters into an initial model for training to obtain the historical predicted power generation cost of the thermal power unit, wherein the historical predicted power generation cost includes the historical predicted fuel cost, the historical predicted environmental protection cost, the historical predicted maintenance cost, and the historical predicted start-up and shutdown cost; Calculating the historical predicted marginal power generation cost of the thermal power unit based on the historical predicted power generation cost; Performing loss calculation on the historical predicted power generation cost based on a preset loss function to obtain a cost loss value; The initial model is adjusted based on the cost loss value to obtain a target power generation cost analysis model.

4. The method according to claim 1, wherein The step of inputting the target power generation parameters into a preset target power generation cost analysis model for processing to obtain the predicted marginal power generation cost of the thermal power unit includes: Inputting the target power generation parameters into a preset target power generation cost analysis model, constructing a cost causal structure equation based on the target power generation parameters, and calculating a predicted power generation cost; The predicted marginal power generation cost of the thermal power generation unit is calculated based on the predicted power generation cost.

5. The method according to claim 4, characterized in that The constructing of a cost causal structure equation based on the target power generation parameters and calculating the predicted power generation cost includes: Determining a fuel cost variable, an environmental cost variable, a maintenance cost variable, and a start-stop cost variable based on the target power generation parameter; Constructing a fuel cost causal structural equation based on the fuel cost variable and calculating the predicted fuel cost; Constructing an environmental protection cost causal structural equation based on the environmental protection cost variable and calculating the predicted environmental protection cost; A maintenance cost causal structural equation is constructed based on the maintenance cost variable, and a predicted maintenance cost is calculated; A start-stop cost causal structural equation is constructed based on the start-stop cost variable, and a predicted start-stop cost is calculated; The predicted power generation cost of the thermal power unit is calculated based on the predicted fuel cost, the predicted environmental protection cost, the predicted maintenance cost and the predicted start-up and shutdown cost.

6. The method according to claim 4, characterized in that The performing of a causal analysis based on the target power generation parameters and the predicted marginal power generation cost to obtain the cost impact factors of the thermal power generation unit includes: Constructing a causal relationship diagram based on the target power generation parameters and the cost causal structure equation; Based on the cause-effect relationship diagram, a cause-effect analysis is performed on the predicted marginal power generation cost to obtain the cost impact reasons of the thermal power unit.

7. The method according to any one of claims 1 to 6, characterized in that The cost impact reasons include cost impact parameters and corresponding impact values, wherein the cost impact parameters correspond to parameters in the target power generation parameters; The adjusting the target power generation parameters based on the cost impact factors to obtain optimized power generation parameters, so as to reduce the power generation cost of the thermal power unit based on the optimized power generation parameters, includes: Searching for a parameter to be optimized corresponding to the cost-influencing parameter in the target power generation parameters; Optimizing the parameters to be optimized based on the influence values ​​to obtain optimized parameters; Based on the optimization parameters and other power generation parameters in the target power generation parameters except the parameters to be optimized, the optimized power generation parameters of the thermal power unit are formed to reduce the power generation cost of the thermal power unit based on the optimized power generation parameters.

8. A power generation cost optimization system, characterized in that: include: An acquisition module is used to obtain target power generation parameters of the thermal power unit; a parameter processing module, configured to input the target power generation parameters into a preset target power generation cost analysis model for processing to obtain a predicted marginal power generation cost of the thermal power unit; a cause-effect analysis module, configured to perform a cause-effect analysis based on the target power generation parameters and the predicted marginal power generation cost to obtain the cost impact factors of the thermal power unit; The power generation cost optimization module is used to adjust the target power generation parameters based on the cost influencing factors to obtain optimized power generation parameters, and control the thermal power unit to generate electricity based on the optimized power generation parameters to reduce power generation costs.

9. A computing device comprising a memory, a processor, and a program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the steps of the power generation cost optimization method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device executes the steps of a power generation cost optimization method according to any one of claims 1 to 7.

Images