Method suitable for multi-energy complementary linkage economy evaluation

Abstract

The invention relates to the technical field of multi-energy complementary energy systems, in particular to a method suitable for multi-energy complementary linkage economy evaluation, which comprises the following steps: S1, constructing a three-dimensional dynamic index system, and designing a time sequence weight matrix based on the peak-flat-valley period operation difference of an energy system; s2, according to the time sequence weight matrix, weight pre-judgment adjustment is conducted on the three-dimensional dynamic index system through LSTM, and a real-time dynamic index system is output. Through the combined design of the three-dimensional dynamic index and the time sequence weight, the total cost and total income of multi-energy linkage are comprehensively covered; the problem that a traditional static index cannot adapt to a dynamic linkage scene is effectively solved, and the scene adaptability is improved; a real-time feedback correction mechanism is introduced through a double-stage evaluation algorithm, the evaluation error rate is reduced conveniently, multi-energy complementary systems of different scales and different energy combinations can be flexibly adapted, only basic cost parameters and a weight matrix need to be adjusted, and the application range is widened.

Description

Technical Field

[0001] This invention relates to the field of multi-energy complementary energy system technology, and specifically to a method for evaluating the economic efficiency of multi-energy complementary linkage. Background Technology

[0002] Against the backdrop of advancing the "dual carbon" goal, multi-energy complementary systems that integrate intermittent renewable energy and improve energy utilization efficiency (such as wind power-photovoltaic-energy storage-gas peak shaving systems) are the core path for energy transformation. Whether the system can be successfully implemented depends on whether the economic evaluation, which serves as the fundamental basis for planning and investment decisions, is scientific.

[0003] Patent publication number CN111585305A discloses a method for economic evaluation of multi-energy complementary linkage, comprising the following steps: constructing an objective function for the economic evaluation of distributed power sources; building a system model of distributed power sources; initializing external factors and inputting typical annual meteorological data; calling annual load data and hybrid microgrid dispatching strategies, and calculating the charging and discharging models of batteries and pumped storage; calculating the annual investment cost using the objective function and considering constraints; constructing a system evaluation framework using a nonlinear particle swarm optimization algorithm, and outputting the optimization results. This invention can significantly save time in the economic evaluation of multi-energy complementary linkage technology, increase evaluation efficiency, not only improve the overall power generation efficiency of microgrid power generation systems and rationally allocate the number of distributed power sources, but also maximize the power generation efficiency of multi-energy complementary power generation systems and reduce system power generation costs under the same environmental, temperature, and illumination conditions.

[0004] While existing technologies have the advantages mentioned above, their disadvantages are as follows: they generally adopt fixed economic-energy efficiency dual indicators that only focus on equipment investment and annual power generation, or use fixed objective functions, without considering the benefits of cross-energy forms and dynamic risk costs. This results in a static indicator system that cannot adapt to multi-energy real-time linkage scenarios. In addition, existing technologies rely on a single nonlinear particle swarm optimization algorithm or are limited to static game optimization, without establishing a feedback correction mechanism between evaluation results and real-time operating data, leading to a large deviation between the evaluation results and the actual operating conditions.

[0005] In conclusion, developing a method suitable for evaluating the economic efficiency of multi-energy complementary systems remains a critical issue that urgently needs to be addressed in the field of multi-energy complementary energy system technology. Summary of the Invention

[0006] The purpose of this invention is to address the problems in existing technologies, which generally adopt fixed binary indicators of economy and energy efficiency, focusing only on equipment investment and annual power generation, or using fixed objective functions without considering the linkage benefits and dynamic risk costs across energy forms. This results in a static indicator system that cannot adapt to real-time linkage scenarios involving multiple energy sources. In addition, existing technologies rely solely on a single nonlinear particle swarm optimization algorithm or are limited to static game optimization, without establishing a feedback correction mechanism between evaluation results and real-time operating data, leading to significant deviations between evaluation results and actual operating conditions.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a method for evaluating the economic efficiency of multi-energy complementary linkage, comprising the following steps: S1. Construct a three-dimensional dynamic indicator system and design a time-series weight matrix based on the differences in operation of the energy system during peak, flat and valley periods; S2. Based on the time-series weight matrix, the weights of the three-dimensional dynamic index system are then adjusted using LSTM to output the real-time dynamic index system. S3. Based on the real-time dynamic index system, a two-stage algorithm is designed, including a first-stage Pareto game optimization solution and a second-stage real-time data feedback correction.

[0008] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: The three-dimensional dynamic indicator system includes: a basic cost layer, a linkage benefit layer, and a risk cost layer. The basic cost layer includes: equipment investment cost, fuel procurement cost, and operation and maintenance service cost. The equipment investment cost includes: the purchase and installation costs of core equipment, including but not limited to photovoltaic modules, wind turbine generators, and energy storage batteries, amortized over the equipment's depreciation period to an average annual cost, expressed as: In the formula, This represents the total annual depreciation cost of all core equipment in a multi-energy complementary system. Indicates the first The unit purchase price of core equipment Indicates the first The unit installation cost of core equipment, Indicates the first The design depreciation period of core equipment. Total number of core equipment types; The fuel procurement cost is calculated by differentiating between peak-shaving-valley gas prices and power generation for fossil energy equipment such as gas turbines, and combining the actual power generation of the gas turbines to calculate the total annual fuel expenditure. The maintenance service cost includes a device health discount factor. The baseline maintenance cost is dynamically adjusted when the battery's state of charge is below 20%. When the wear rate of the wind turbine blades exceeds 5%, .

[0009] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: The three-dimensional dynamic indicator system includes: a basic cost layer, a linkage benefit layer, and a risk cost layer. The linkage benefit layer includes: revenue from abandoned electricity recovery, revenue from the cascade utilization of waste heat, and revenue from ancillary services. The revenue from the recovery of abandoned electricity refers to the revenue generated when wind and solar power output is excessive, which is then discharged and fed back to the grid after being charged and stored through an energy storage system. The revenue from the cascade utilization of waste heat: the revenue generated from recovering waste heat from cogeneration units through a thermal storage system to supply energy to users, expressed as: In the formula, This indicates the revenue from the cascade utilization of waste heat. This indicates the annual heating price for local residents and industrial users. Indicates the first Electricity generation of the gas turbine during a given period For the power generation efficiency of gas turbines, This indicates the efficiency of the waste heat recovery system. The energy conversion factor is expressed in kJ / kWh. The annual heating load per unit area; The ancillary service revenue refers to the subsidy revenue obtained from participating in power grid peak shaving and frequency regulation.

[0010] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: The three-dimensional dynamic indicator system includes: a basic cost layer, a linkage benefit layer, and a risk cost layer. The risk cost layer includes: source load uncertainty risk, policy change risk, and ultimate risk cost. The source-load uncertainty risk is calculated using Monte Carlo simulation of ±20% wind power output fluctuation and ±15% user load change, to determine the additional cost of standby gas turbines. The expression is: In the formula, This indicates the cost of source-load uncertainty risk. This represents the expectation operator. Indicates the first Natural gas prices during the period Indicates the first User load fluctuation values ​​over a period of time Indicates the first The fluctuation value of renewable energy output over a period of time; The policy change risk mentioned: simulate carbon trading price fluctuations and calculate carbon trading profits and losses; The final risk cost is a combination of the two types of risk costs, with a risk premium coefficient introduced.

[0011] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: It should be noted that the time-series weight matrix is ,satisfy The peak-shaving-valley periods of the energy system are defined according to the guidelines of the China Power Grid. For the weight of linkage benefits, Risk cost weighting, Based on cost weights, , and The allocation rules are based on time period importance scoring. Confirmed, expression: In the formula, Indicates the first Importance rating for each time period The weighting coefficients for the load metric dimension are empirical parameters set manually. Indicates the first Average load over the period This represents the sum of the average loads during the peak, average, and trough periods. This represents the weighting coefficient for the electricity price as a percentage of total cost. Indicates the first Average electricity price during the period This represents the sum of the average electricity prices over three time periods. This represents the weighting coefficient for the proportion of risk volatility. Indicates the first Standard deviation of source load fluctuation over a period of time This represents the sum of the standard deviations of the source load fluctuations over the three time periods. Indicates the first Risk volatility percentage over a given period; Peak The focus is on the synergistic effects: , , ; usually Then, a balanced distribution will occur: , , ; Gu Shi The focus is then on basic cost control: , , .

[0012] Further, in step S2, based on the time-series weight matrix, the method for performing weight prediction and adjustment on the three-dimensional dynamic index system using LSTM to output the real-time dynamic index system is as follows: The LSTM method collects historical data from the coverage area of ​​the multi-energy complementary system. This historical data includes user load data and related impact data. The related impact data includes temperature, humidity, wind speed, date characteristics, holiday / holiday status, and electricity price period indicators. Outliers in the user load data are removed using the 3σ criterion, and missing values ​​are filled using linear interpolation. Finally, Min-Max normalization is used to map the user load data and related impact data to... The interval is used to construct the input feature vector and output label of the LSTM. The input feature vector is taken from the past N=24 hours and normalized by Min-Max and mapped to... User load data and related impact data are used as input sequences, and the output label is the normalized load forecast value for the next 24 hours. The input sequence is divided into training, validation, and test sets in a 7:2:1 ratio. The LSTM adopts a structure of 1 input layer + 2 LSTM hidden layers + 1 fully connected output layer, and captures long-term temporal dependencies of load data through a gating mechanism. The LSTM uses root mean square error to measure the deviation between the predicted and actual values ​​for training and optimization. Expression: In the formula, This represents the root mean square error. To determine the number of samples in the validation set, It is the first The first verification sample, the first Hourly load forecast It is the first The first verification sample, the first The actual hourly load value, the LSTM optimizer, using the Adam optimizer, with the initial learning rate set to... The training employs a learning rate decay strategy: the learning rate is decayed to 0.8 times the original value every 50 epochs to avoid overfitting; training is stopped when the RMSE does not decrease for 10 consecutive epochs or reaches the maximum epoch, the optimal LSTM parameters are saved, and the trained LSTM is obtained.

[0013] Further, in step S2, based on the time-series weight matrix, the method for performing weight prediction and adjustment on the three-dimensional dynamic index system using LSTM to output the real-time dynamic index system is as follows: It is important to note that the LSTM predicts user load data changes over the next 24 hours and adjusts the weights of the three-dimensional dynamic indicator system 4 hours in advance. The input feature vector at the current moment is input into the trained LSTM to obtain the normalized user load data prediction value for the next 24 hours. Then, inverse normalization is used to restore the actual user load data prediction value. Simultaneously, the user load data fluctuation coefficient for each time period within the next 24 hours is calculated. Based on the user load data fluctuation coefficient... Construct weight adjustment coefficients ,expression: In the formula, Indicates the future number Hourly load fluctuation coefficient Indicates the first The hourly weighting adjustment coefficient, combined with the peak-shaving-valley time period division of the Chinese power grid, is used to make targeted adjustments to the weights of each time period in the next 24 hours, with the original peak hour weights remaining unchanged. The adjusted weights are: In the formula, Indicates the first Time-series weight matrix adjusted for hourly peak periods. This indicates the initial weight of the linkage effect during the peak period. This indicates the initial risk cost weight during peak hours. This indicates the initial base cost weight during peak hours. Indicates the first The hourly weighting adjustment factor, in At times, the weighting of synergistic benefits and risk costs increases to adapt to the needs of energy coordination and risk control under high loads; under normal circumstances, the original weighting applies. Valley Original Weight The adjusted weights are: In the formula, Indicates the first Time-series weight matrix adjusted for hourly valley periods. This indicates the initial weight of the linkage effect during the trough period. This indicates the initial risk cost weight during the trough period. This represents the initial base cost weight during the trough period. At that time, the weight of basic costs increases to meet the cost control needs under low load conditions.

[0014] Furthermore, in step S3, based on the real-time dynamic index system, the two-stage algorithm, including the first stage of Pareto game optimization and the second stage of real-time data feedback correction, is designed as follows: In the first stage of Pareto game optimization, each energy unit in the multi-energy complementary system is regarded as an independent decision-making unit. Based on the real-time dynamic index system, the game objective function of the basic model for economic evaluation of the multi-energy complementary system is constructed and solved by an improved nonlinear particle swarm algorithm. The objective function of the game theory aims to maximize the net profit after risk hedging, and its expression is: In the formula, This represents the maximize operator. Represents a vector of decision variables The objective function, The time-series weighting coefficient represents the difference between the linkage benefits and the basic costs. Represents a vector of decision variables The linkage benefit function, Represents a vector of decision variables The basic cost function, The time-series weighting coefficients representing risk costs. Represents a vector of decision variables The risk cost function is substituted into the three-dimensional indicator system and time-series weight matrix of the real-time dynamic indicator system. Each particle is set to correspond to a set of decision variables, the population size is 50, the number of iterations is 100, and the initial parameters are based on the equipment investment cost. In each iteration, particles that do not meet the requirements of having a basic cost lower than the industry benchmark by 10% and a linkage benefit higher than the industry benchmark by 5% are removed, and non-dominated solutions are retained. After the iteration converges, the basic model for the economic evaluation of the multi-energy complementary system outputs the initial evaluation results, which include, but are not limited to, the optimal equipment configuration, the average annual net income, and the investment payback period.

[0015] Furthermore, in step S3, based on the real-time dynamic index system, the two-stage algorithm, including the first stage of Pareto game optimization and the second stage of real-time data feedback correction, is designed as follows: It is important to note that in the second stage, real-time data feedback correction involves setting up edge computing nodes to collect real-time operational data. Based on the real-time dynamic indicator system, the initial evaluation results are corrected in two dimensions: technical parameters and profit deviation. The corrected parameters are then fed back to the real-time dynamic indicator system to form a closed loop. The edge computing node collects real-time operating data. Through IoT sensors, it collects key data, including but not limited to energy storage charging and discharging efficiency and photovoltaic irradiance, at a frequency of 15 minutes per time. The parameter types of the collected key data correspond one-to-one with the indicator dimensions of the three-dimensional dynamic indicator system. The collected real-time operating data is subjected to outlier removal and time alignment.

[0016] Furthermore, in step S3, based on the real-time dynamic index system, the two-stage algorithm, including the first stage of Pareto game optimization and the second stage of real-time data feedback correction, is designed as follows: The dual-dimensional correction automatically adjusts the corresponding indicator parameters in the three-dimensional dynamic indicator system when the actual data deviates from the preset parameters of the economic evaluation model of the multi-energy complementary system. Simultaneously, it compares the linked revenue of the economic evaluation model of the multi-energy complementary system with the actual settlement revenue to calculate the revenue deviation rate. ,expression: In the formula, Indicates the profit deviation rate. This indicates the predicted associated returns. Indicates the actual linked benefits. This represents the difference between the predicted and actual linked returns, expressed as a deviation rate. When the benefit rate is greater than 5%, the benefit coefficient in the game objective function is adjusted using a PID controller; expression: In the formula, Indicates the first The curtailment benefit factor after time correction This represents the initial coefficient for the revenue from abandoned electricity. This represents the proportionality coefficient. Indicates the first The deviation at time. Represents the integral coefficient. Denotes the differential coefficient. Indicates the first The derivative of the time deviation with respect to time is used to rerun the first-stage algorithm every 24 hours based on the corrected index parameters, updating the initial evaluation results.

[0017] Beneficial effects Compared with known public technologies, the technical solution provided by this invention has the following beneficial effects: When in use, this invention comprehensively covers the total cost and total benefit of multi-energy linkage through a combination of three-dimensional dynamic indicators and time-series weights, effectively solving the problem that traditional static indicators cannot adapt to dynamic linkage scenarios, and is conducive to improving scenario adaptability. By introducing a real-time feedback correction mechanism through a two-stage evaluation algorithm, it is easy to reduce the evaluation error rate and can flexibly adapt to multi-energy complementary systems of different scales and energy combinations. Only the basic cost parameters and weight matrix need to be adjusted, which is conducive to improving the scope of application. Attached Figure Description

[0018] Figure 1 This is a flowchart of a method for evaluating the economic efficiency of multi-energy complementary linkage according to the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but includes other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] The present invention will now be described in further detail with reference to the accompanying drawings: Example: like Figure 1 As shown, this invention provides a method for evaluating the economic efficiency of multi-energy complementary linkages, comprising the following steps: S1. Construct a three-dimensional dynamic indicator system and design a time-series weight matrix based on the differences in operation of the energy system during peak, flat and valley periods; Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: The three-dimensional dynamic indicator system includes: a basic cost layer, a linkage benefit layer, and a risk cost layer. The basic cost layer includes: equipment investment cost, fuel procurement cost, and operation and maintenance service cost. The equipment investment cost includes: the purchase and installation costs of core equipment, including but not limited to photovoltaic modules, wind turbine generators, and energy storage batteries, amortized over the equipment's depreciation period to an average annual cost, expressed as: In the formula, This represents the total annual depreciation cost of all core equipment in a multi-energy complementary system. Indicates the first The unit purchase price of core equipment Indicates the first The unit installation cost of core equipment, Indicates the first The design depreciation period of core equipment. Total number of core equipment types; The fuel procurement cost is calculated by differentiating between peak-shaving-valley gas prices and power generation for fossil energy equipment such as gas turbines, and combining the actual power generation of the gas turbines to calculate the total annual fuel expenditure. The maintenance service cost includes a device health discount factor. The baseline maintenance cost is dynamically adjusted when the battery's state of charge is below 20%. When the wear rate of the wind turbine blades exceeds 5%, .

[0022] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: The three-dimensional dynamic indicator system includes: a basic cost layer, a linkage benefit layer, and a risk cost layer. The linkage benefit layer includes: revenue from abandoned electricity recovery, revenue from the cascade utilization of waste heat, and revenue from ancillary services. The revenue from the recovery of abandoned electricity refers to the revenue generated when wind and solar power output is excessive, which is then discharged and fed back to the grid after being charged and stored through an energy storage system. The revenue from the cascade utilization of waste heat: the revenue generated from recovering waste heat from cogeneration units through a thermal storage system to supply energy to users, expressed as: In the formula, This indicates the revenue from the cascade utilization of waste heat. This indicates the annual heating price for local residents and industrial users. Indicates the first Electricity generation of the gas turbine during a given period For the power generation efficiency of gas turbines, This indicates the efficiency of the waste heat recovery system. The energy conversion factor is expressed in kJ / kWh. The annual heating load per unit area; The ancillary service revenue refers to the subsidy revenue obtained from participating in power grid peak shaving and frequency regulation.

[0023] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: The three-dimensional dynamic indicator system includes: a basic cost layer, a linkage benefit layer, and a risk cost layer. The risk cost layer includes: source load uncertainty risk, policy change risk, and ultimate risk cost. The source-load uncertainty risk is calculated using Monte Carlo simulation of ±20% wind power output fluctuation and ±15% user load change, to determine the additional cost of standby gas turbines. The expression is: In the formula, This indicates the cost of source-load uncertainty risk. This represents the expectation operator. Indicates the first Natural gas prices during the period Indicates the first User load fluctuation values ​​over a period of time Indicates the first The fluctuation value of renewable energy output over a period of time; The policy change risk mentioned: simulate carbon trading price fluctuations and calculate carbon trading profits and losses; The final risk cost is a combination of the two types of risk costs, with a risk premium coefficient introduced.

[0024] Furthermore, in step S1, a three-dimensional dynamic index system is constructed, and based on the differences in operation during peak-shaving-valley periods of the energy system, the method for designing the time-series weight matrix is ​​as follows: It should be noted that the time-series weight matrix is ,satisfy The peak-shaving-valley periods of the energy system are defined according to the guidelines of the China Power Grid. For the weight of linkage benefits, Risk cost weighting, Based on cost weights, , and The allocation rules are based on time period importance scoring. Confirmed, expression: In the formula, Indicates the first Importance rating for each time period The weighting coefficients for the load metric dimension are empirical parameters set manually. Indicates the first Average load over the period This represents the sum of the average loads during the peak, average, and trough periods. This represents the weighting coefficient for the electricity price as a percentage of total cost. Indicates the first Average electricity price during the period This represents the sum of the average electricity prices over three time periods. This represents the weighting coefficient for the proportion of risk volatility. Indicates the first Standard deviation of source load fluctuation over a period of time This represents the sum of the standard deviations of the source load fluctuations over the three time periods. Indicates the first Risk volatility percentage over a given period; Peak The focus is on the synergistic effects: , , ; usually Then, a balanced distribution will occur: , , ; Gu Shi The focus is then on basic cost control: , , .

[0025] In this embodiment, by incorporating key factors such as equipment health and policy risks into the evaluation, the system's economic calculations are made more aligned with actual operational scenarios. This breaks through the limitations of traditional single-dimensional approaches that only focus on cost or benefit, facilitating a more comprehensive coverage of indicators. A dynamic weight matrix enables differentiated evaluation priorities during peak, flat, and valley periods. During peak hours, priority can be given to capturing energy storage curtailment revenue and peak-shaving subsidies, while during valley periods, cost control for equipment depreciation and fuel can be strengthened. This enhances the targeting of operational strategy adjustments and improves the accuracy of time-period adaptation. The indicator calculations are all based on industry-standard parameters and collectable data, and the weight allocation rules are clear and replicable. This makes the system applicable not only to residential communities but also to different scenarios such as industrial parks and remote villages with slight parameter adjustments. This provides a reliable evaluation basis for the planning, design, and operational optimization of multi-energy complementary systems, enhancing the practicality of the project.

[0026] S2. Based on the time-series weight matrix, the weights of the three-dimensional dynamic index system are then adjusted using LSTM to output the real-time dynamic index system. Further, in step S2, based on the time-series weight matrix, the method for performing weight prediction and adjustment on the three-dimensional dynamic index system using LSTM to output the real-time dynamic index system is as follows: The LSTM method collects historical data from the coverage area of ​​the multi-energy complementary system. This historical data includes user load data and related impact data. The related impact data includes temperature, humidity, wind speed, date characteristics, holiday / holiday status, and electricity price period indicators. Outliers in the user load data are removed using the 3σ criterion, and missing values ​​are filled using linear interpolation. Finally, Min-Max normalization is used to map the user load data and related impact data to... The interval is used to construct the input feature vector and output label of the LSTM. The input feature vector is taken from the past N=24 hours and normalized by Min-Max and mapped to... User load data and related impact data are used as input sequences, and the output label is the normalized load forecast value for the next 24 hours. The input sequence is divided into training, validation, and test sets in a 7:2:1 ratio. The LSTM adopts a structure of 1 input layer + 2 LSTM hidden layers + 1 fully connected output layer, and captures long-term temporal dependencies of load data through a gating mechanism. The LSTM uses root mean square error to measure the deviation between the predicted and actual values ​​for training and optimization. Expression: In the formula, This represents the root mean square error. To determine the number of samples in the validation set, It is the first The first verification sample, the first Hourly load forecast It is the first The first verification sample, the first The actual hourly load value, the LSTM optimizer, using the Adam optimizer, with the initial learning rate set to... The training employs a learning rate decay strategy: the learning rate is decayed to 0.8 times the original value every 50 epochs to avoid overfitting; training is stopped when the RMSE does not decrease for 10 consecutive epochs or reaches the maximum epoch, the optimal LSTM parameters are saved, and the trained LSTM is obtained.

[0027] Further, in step S2, based on the time-series weight matrix, the method for performing weight prediction and adjustment on the three-dimensional dynamic index system using LSTM to output the real-time dynamic index system is as follows: It is important to note that the LSTM predicts user load data changes over the next 24 hours and adjusts the weights of the three-dimensional dynamic indicator system 4 hours in advance. The input feature vector at the current moment is input into the trained LSTM to obtain the normalized user load data prediction value for the next 24 hours. Then, inverse normalization is used to restore the actual user load data prediction value. Simultaneously, the user load data fluctuation coefficient for each time period within the next 24 hours is calculated. Based on the user load data fluctuation coefficient... Construct weight adjustment coefficients ,expression: In the formula, Indicates the future number Hourly load fluctuation coefficient Indicates the first The hourly weighting adjustment coefficient, combined with the peak-shaving-valley time period division of the Chinese power grid, is used to make targeted adjustments to the weights of each time period in the next 24 hours, with the original peak hour weights remaining unchanged. The adjusted weights are: In the formula, Indicates the first Time-series weight matrix adjusted for hourly peak periods. This indicates the initial weight of the linkage effect during the peak period. This indicates the initial risk cost weight during peak hours. This indicates the initial base cost weight during peak hours. Indicates the first The hourly weighting adjustment factor, in At times, the weighting of synergistic benefits and risk costs increases to adapt to the needs of energy coordination and risk control under high loads; under normal circumstances, the original weighting applies. Valley Original Weight The adjusted weights are: In the formula, Indicates the first Time-series weight matrix adjusted for hourly valley periods. This indicates the initial weight of the linkage effect during the trough period. This indicates the initial risk cost weight during the trough period. This represents the initial base cost weight during the trough period. At that time, the weight of basic costs increases to meet the cost control needs under low load conditions.

[0028] In this embodiment, by using multi-dimensional correlation data fusion and LSTM gating mechanism, the load forecasting error is reduced compared to the traditional ARIMA model, providing a reliable basis for weight adjustment and facilitating improved load forecasting accuracy. By adapting to load changes 4 hours in advance, the linkage benefit weight can be prioritized during peak hours to capture peak-shaving subsidies, while the basic cost weight can be strengthened during off-peak hours to control gas consumption. This enhances the dynamism of weight adjustment and breaks through the limitations of traditional fixed weights.

[0029] S3. Based on the real-time dynamic index system, a two-stage algorithm is designed, including a first-stage Pareto game optimization solution and a second-stage real-time data feedback correction. Furthermore, in step S3, based on the real-time dynamic index system, the two-stage algorithm, including the first stage of Pareto game optimization and the second stage of real-time data feedback correction, is designed as follows: In the first stage of Pareto game optimization, each energy unit in the multi-energy complementary system is regarded as an independent decision-making unit. Based on the real-time dynamic index system, the game objective function of the basic model for economic evaluation of the multi-energy complementary system is constructed and solved by an improved nonlinear particle swarm algorithm. The objective function of the game theory aims to maximize the net profit after risk hedging, and its expression is: In the formula, This represents the maximize operator. Represents a vector of decision variables The objective function, The time-series weighting coefficient represents the difference between the linkage benefits and the basic costs. Represents a vector of decision variables The linkage benefit function, Represents a vector of decision variables The basic cost function, The time-series weighting coefficients representing risk costs. Represents a vector of decision variables The risk cost function is substituted into the three-dimensional indicator system and time-series weight matrix of the real-time dynamic indicator system. Each particle is set to correspond to a set of decision variables, the population size is 50, the number of iterations is 100, and the initial parameters are based on the equipment investment cost. In each iteration, particles that do not meet the requirements of having a basic cost lower than the industry benchmark by 10% and a linkage benefit higher than the industry benchmark by 5% are removed, and non-dominated solutions are retained. After the iteration converges, the basic model for the economic evaluation of the multi-energy complementary system outputs the initial evaluation results, which include, but are not limited to, the optimal equipment configuration, the average annual net income, and the investment payback period.

[0030] Furthermore, in step S3, based on the real-time dynamic index system, the two-stage algorithm, including the first stage of Pareto game optimization and the second stage of real-time data feedback correction, is designed as follows: It is important to note that in the second stage, real-time data feedback correction involves setting up edge computing nodes to collect real-time operational data. Based on the real-time dynamic indicator system, the initial evaluation results are corrected in two dimensions: technical parameters and profit deviation. The corrected parameters are then fed back to the real-time dynamic indicator system to form a closed loop. The edge computing node collects real-time operating data. Through IoT sensors, it collects key data, including but not limited to energy storage charging and discharging efficiency and photovoltaic irradiance, at a frequency of 15 minutes per time. The parameter types of the collected key data correspond one-to-one with the indicator dimensions of the three-dimensional dynamic indicator system. The collected real-time operating data is subjected to outlier removal and time alignment.

[0031] Furthermore, in step S3, based on the real-time dynamic index system, the two-stage algorithm, including the first stage of Pareto game optimization and the second stage of real-time data feedback correction, is designed as follows: The dual-dimensional correction automatically adjusts the corresponding indicator parameters in the three-dimensional dynamic indicator system when the actual data deviates from the preset parameters of the economic evaluation model of the multi-energy complementary system. Simultaneously, it compares the linked revenue of the economic evaluation model of the multi-energy complementary system with the actual settlement revenue to calculate the revenue deviation rate. ,expression: In the formula, Indicates the profit deviation rate. This indicates the predicted associated returns. Indicates the actual linked benefits. This represents the difference between the predicted and actual linked returns, expressed as a deviation rate. When the benefit rate is greater than 5%, the benefit coefficient in the game objective function is adjusted using a PID controller; expression: In the formula, Indicates the first The curtailment benefit factor after time correction This represents the initial coefficient for the revenue from abandoned electricity. This represents the proportionality coefficient. Indicates the first The deviation at time. Represents the integral coefficient. Denotes the differential coefficient. Indicates the first The derivative of the time deviation with respect to time is used to rerun the first-stage algorithm every 24 hours based on the corrected index parameters, updating the initial evaluation results.

[0032] In this embodiment, unreasonable configurations are eliminated by Pareto non-dominated solution screening, and real-time data is used for dual-dimensional correction, thereby reducing the deviation between the evaluation results and actual operation and improving the evaluation accuracy. By collecting data at high frequency through edge computing nodes and updating in a closed loop 24 hours a day, it is easy to respond in a timely manner to fluctuations in actual operating conditions such as equipment efficiency decline and irradiance changes, avoiding the problem of traditional static evaluation being "detached from the field" and enhancing dynamic adaptability.

[0033] like Figure 1 As shown, this invention provides a method for evaluating the economic efficiency of multi-energy complementary linkages, comprising the following steps: A1. Implementation preparation includes: basic data collection and indicator system initialization; A1.1 Equipment parameters for the implementation scenario: Photovoltaic module unit price 3 yuan / W, 2.5MW wind turbine unit price 5000 yuan / kW, lithium iron phosphate energy storage battery unit price 1.2 yuan / Wh; Price data for the implementation scenario: Local power grid peak electricity price 0.75 yuan / kWh, off-peak electricity price 0.35 yuan / kWh, peak natural gas price 3.5 yuan / m³, off-peak natural gas price 2.2 yuan / m³, carbon trading price 60 yuan / ton; Meteorological and load data for the implementation scenario: Typical annual wind speed and solar irradiance data obtained from the local meteorological bureau; Actual electricity consumption data of 200 households in the area over the past year, with a peak maximum load of 1200kW and an off-peak minimum load of 500kW; A1.2. Based on equipment parameters, price data, and meteorological and load data, the initial settings of the three-dimensional dynamic indicator system include: basic cost layer, linkage benefit layer, and risk cost layer. The basic cost layer is as follows: photovoltaic modules are depreciated over 25 years, with an average annual depreciation cost of 120,000 yuan; energy storage batteries are depreciated over 10 years, with an average annual depreciation cost of 240,000 yuan; and the system's baseline operation and maintenance cost is 500,000 yuan per year. The linkage benefit layer is as follows: the annual power curtailment compensation generation is set at 200,000 kWh, and the annual peak shaving subsidy revenue is 150,000 yuan; The risk cost layer is as follows: the number of Monte Carlo simulations is set to 1000, and the initial risk premium coefficient is... .

[0034] A2. The implementation steps include: calculating dynamic indicators and solving the two-stage algorithm. A2.1 The dynamic indicators are calculated based on the characteristics of the "peak-flat-valley" time periods and combined with the time-series weight matrix. The dynamic indicators for each time period are then calculated as follows: Peak time: weighting , , Revenue from abandoned electricity (150,000 yuan) + revenue from peak shaving (50,000 yuan) = 200,000 yuan. Gas cost (80,000 yuan) + Operation and maintenance cost (50,000 yuan) = 130,000 yuan. Backup gas cost (30,000 yuan) + carbon cost (20,000 yuan) = 50,000 yuan; Valley Time: Weight , , Revenue from abandoned electricity (50,000 yuan). Gas cost (30,000 yuan) + Operation and maintenance cost (30,000 yuan) = 60,000 yuan. Backup gas cost (10,000 yuan).

[0035] A2.2 The two-stage algorithm solution follows a "game theory optimization - real-time correction" process to complete the system's economic evaluation, including: Stage 1: The improved nonlinear particle swarm optimization algorithm converges after 80 iterations, outputting the optimal equipment configuration: 10MW photovoltaic modules + 2 x 5MW wind turbines + 20MWh energy storage battery, with an initial investment payback period of 6.8 years; Stage 2: Edge computing nodes collect one month of real-time operating data, discovering that the actual energy storage charging and discharging efficiency is 85% (lower than the preset 90%), initiating a two-dimensional correction, including: Technical parameter correction: Energy storage depreciation cost increased to 254,000 yuan / year, and abandoned power recovery revenue reduced to 170,000 yuan / year; Revenue deviation correction: Calculating the linkage revenue deviation rate. By adjusting the curtailment revenue factor to 0.9 using a PID controller, the system's investment payback period is 7.0 years after correction.

[0036] In this embodiment, the application of the method of the present invention improves economic indicators, energy utilization efficiency, and risk resistance capabilities.

[0037] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method suitable for multi-energy complementary linkage economy evaluation, characterized in that, The method comprises the following steps: S1, constructing a three-dimensional dynamic index system, and designing a time sequence weight matrix based on the peak-flat-valley period operation difference of an energy system; S2, according to the time sequence weight matrix, adjusting the weight of the three-dimensional dynamic index system through LSTM, and outputting a real-time dynamic index system; S3, based on the real-time dynamic index system, designing a two-stage algorithm including a first-stage Pareto game optimization solution and a second-stage real-time data feedback correction.

2. The method for evaluating the economy of multi-energy complementary linkage according to claim 1, characterized in that, In step S1, the method for constructing a three-dimensional dynamic index system and designing a time sequence weight matrix based on the peak-flat-valley period operation difference of an energy system is as follows: The three-dimensional dynamic index system includes a basic cost layer, a linkage benefit layer, and a risk cost layer. The basic cost layer includes equipment investment cost, fuel procurement cost, and operation and maintenance service cost. The equipment investment cost includes the purchase cost and installation engineering cost of core equipment such as photovoltaic components, wind turbine generators, and energy storage batteries, which are evenly distributed to the annual average cost according to the equipment depreciation period, and the expression is as follows: wherein, represents the total annual depreciation cost of all core equipment of the multi-energy complementary system, represents the unit purchase price of the core equipment of the first class, represents the unit installation engineering cost of the core equipment of the first class, represents the design depreciation life of the core equipment of the first class, is the total number of core equipment types; The fuel procurement cost: for fossil energy equipment such as gas turbines, the peak-flat-valley period gas price and power generation are distinguished, and the annual total fuel expenditure is calculated combined with the actual power generation of the gas turbine. The operation and maintenance service cost: introduce equipment health degree discount coefficient The benchmark operation and maintenance cost is dynamically corrected when the battery state of charge is lower than 20%, and the operation and maintenance service cost is dynamically corrected when the battery state of charge is lower than 20%. When the wear rate of the fan blade is more than 5%, the benchmark operation and maintenance cost is dynamically corrected. .

3. The method for evaluating the economy of multi-energy complementary linkage according to claim 2, characterized in that, In step S1, the method for constructing a three-dimensional dynamic index system and designing a time sequence weight matrix based on the peak-flat-valley period operation difference of an energy system is as follows: The three-dimensional dynamic index system includes a basic cost layer, a linkage benefit layer, and a risk cost layer. The linkage benefit layer includes abandoned electricity recovery income, waste heat cascade utilization income, and auxiliary service income. The abandoned electricity recovery income: when the output of wind power and photovoltaic power is excessive, the income generated by discharging the stored electricity back to the grid after charging through the energy storage system; The waste heat cascade utilization income: the income generated by supplying energy to users through the waste heat recovery system of cogeneration units, and the expression is as follows: wherein, represents the profit of the waste heat cascade utilization, represents the annual heating price of local residents and industrial users, represents the first represents the power generation of the period gas turbine, represents the power generation efficiency of the gas turbine, represents the efficiency of the waste heat recovery system, represents the energy unit conversion coefficient, the unit is kJ / kWh, represents the annual heating load per unit area; The auxiliary service income: the subsidy income obtained by participating in grid peak regulation and frequency modulation.

4. The method for evaluating the economy of multi-energy complementary linkage according to claim 3, characterized in that, In step S1, the method for constructing a three-dimensional dynamic index system and designing a time sequence weight matrix based on the peak-flat-valley period operation difference of an energy system is as follows: The three-dimensional dynamic index system includes a basic cost layer, a linkage benefit layer, and a risk cost layer. The risk cost layer includes source and load uncertainty risk, policy change risk, and final risk cost. The source and load uncertainty risk: ±20% wind power output fluctuation and ±15% user load change are simulated by Monte Carlo simulation, and the additional cost of standby gas turbines is calculated, and the expression is as follows: wherein, represents the source load uncertainty risk cost, represents the expected value operator, represents the natural gas price of the time period, represents the user load fluctuation value of the time period, represents the renewable energy output fluctuation value of the time period; The policy change risk: the carbon trading price fluctuation is simulated, and the carbon trading profit and loss is calculated; The final risk cost: the two types of risk costs are integrated, and a risk premium coefficient is introduced.

5. The method for evaluating the economy of multi-energy complementary linkage according to claim 4, characterized in that, In step S1, the method for constructing a three-dimensional dynamic index system and designing a time sequence weight matrix based on the peak-flat-valley period operation difference of an energy system is as follows: It should be noted that the timing weight matrix is , satisfying ; the peak-valley period of the energy system is divided according to the guidance of China's power grid, is the linkage benefit weight, is the risk cost weight, is the basic cost weight, , and The allocation rules are determined by the importance score of the period , expression: In the formula, represents the importance score of the th period, represents the weight coefficient of the load proportion dimension, which is an empirical parameter set by a person, represents the average load of the th period, represents the sum of the average loads of the peak, flat, and valley periods, represents the weight coefficient of the electricity price proportion dimension, represents the average electricity price of the th period, represents the sum of the average electricity prices of the three periods, represents the weight coefficient of the risk fluctuation proportion dimension, represents the source-load fluctuation standard deviation of the th period, represents the sum of the source-load fluctuation standard deviations of the three periods, represents the risk fluctuation proportion of the th period. Peak hours Then focus on the linkage benefits: , , ; Ordinary time Then the equal distribution: , , ; Valley time Then focus on the basic cost control: , , .

6. The method for evaluating the economy of multi-energy complementary linkage according to claim 5, characterized in that, In step S2, according to the time sequence weight matrix, the method for adjusting the weight of the three-dimensional dynamic index system through LSTM and outputting a real-time dynamic index system is as follows: The LSTM collects historical data of the multi-capacity complementary system coverage area, the historical data including user load data and associated influence data, the associated influence data including temperature, humidity, wind speed, date characteristics, whether a holiday, and power price period identification, 3σ criterion is used to eliminate outliers in the user load data, and linear interpolation is used to fill in missing values, and then Min-Max normalization is used to map the user load data and the associated influence data to an interval, input feature vectors of the LSTM and output labels are constructed, the input feature vectors taking past N=24 hours of the user load data and the associated influence data mapped to an interval as input sequences, the output labels being normalized load prediction values for future 24 hours, the input sequences being divided into a training set, a validation set and a test set according to a ratio of 7:2:1; the LSTM adopts a structure of 1 layer of input layer+2 layers of LSTM hidden layer+1 layer of fully connected output layer, and captures long time sequence dependence of the load data through a gating mechanism, the LSTM adopts root mean square error to measure deviation of the prediction value from the true value for training and optimization, and the expression is: In the formula, This represents the root mean square error. To determine the number of samples in the validation set, It is the first The first verification sample, the first Hourly load forecast It is the first The first verification sample, the first The actual hourly load value, the LSTM optimizer, using the Adam optimizer, with the initial learning rate set to... The training employs a learning rate decay strategy: the learning rate is decayed to 0.8 times the original value every 50 epochs to avoid overfitting; training is stopped when the RMSE does not decrease for 10 consecutive epochs or reaches the maximum epoch, the optimal LSTM parameters are saved, and the trained LSTM is obtained.

7. The method for evaluating the economy of multi-energy complementary linkage according to claim 6, characterized in that, In step S2, according to the timing weight matrix, the three-dimensional dynamic index system is further adjusted by the LSTM, and the method for outputting the real-time dynamic index system is: It should be noted that the LSTM predicts the future 24-hour user load data change, and adjusts the weight of the three-dimensional dynamic index system 4 hours in advance. The input feature vector at the current time is input into the trained LSTM to obtain the normalized user load data prediction value for the next 24 hours. Then, through inverse normalization, the actual user load data prediction value is obtained. At the same time, the user load data fluctuation coefficient of each period in the future 24 hours is calculated. Based on the user load data fluctuation coefficient , the weight adjustment coefficient is constructed, and the expression is: In the formula, Indicates the future number Hourly load fluctuation coefficient Indicates the first The hourly weighting adjustment coefficient, combined with the peak-shaving-valley time period division of the Chinese power grid, is used to make targeted adjustments to the weights of each time period in the next 24 hours, with the original peak hour weights remaining unchanged. The adjusted weights are: In the formula, represents the first hour peak period adjusted timing weight matrix, represents the initial linkage benefit weight of the peak period, represents the initial risk cost weight of the peak period, represents the initial basic cost weight of the peak period, represents the first hour weight adjustment coefficient, when , the linkage benefit and risk cost weight proportion is improved, adapting to the energy synergy and risk prevention and control demand under high load; in ordinary times, according to the original weight , the original weight in the valley , the adjusted weight is: In the formula, represents the first hour valley period adjusted timing weight matrix, represents the initial linkage benefit weight of the valley period, represents the initial risk cost weight of the valley period, represents the initial basic cost weight of the valley period, and when the basic cost weight ratio is increased, the cost control demand under low load is adapted.

8. The method for evaluating the economy of multi-energy complementary linkage according to claim 7, characterized in that, In step S3, based on the real-time dynamic index system, a two-stage algorithm including a first-stage Pareto game optimization solution and a second-stage real-time data feedback correction method is designed: Wherein, in the first-stage Pareto game optimization solution, each energy unit in the multi-energy complementary system is regarded as an independent decision unit, the game objective function of the multi-energy complementary system economic evaluation basic model is constructed based on the real-time dynamic index system, and the improved nonlinear particle swarm algorithm is used for solution; The game objective function takes the maximization of net income after risk hedging as the target, and the expression is: In the formula, represents a maximization operator, represents an objective function about the decision variable vector , represents a time sequence weight coefficient of the linkage benefit and the basic cost difference, represents a linkage benefit function about the decision variable vector , represents a basic cost function about the decision variable vector , represents a time sequence weight coefficient of the risk cost, represents a risk cost function about the decision variable vector , by substituting the three-dimensional index system of the real-time dynamic index system and the time sequence weight matrix, setting each particle to correspond to a set of decision variable combinations, the population size is 50, the iteration number is 100, and the initialization parameters are based on the equipment investment cost; in each iteration, particles that do not meet the conditions of basic cost being lower than the industry benchmark by 10% and linkage benefit being higher than the industry benchmark by 5% are removed, and non-inferior solutions are retained; after iteration convergence, the multi-energy complementary system economic evaluation basic model outputs an initial evaluation result, and the initial evaluation result includes but is not limited to optimal equipment configuration, annual average net income, and investment recovery period.

9. The method for evaluating the economy of multi-energy complementary linkage according to claim 8, characterized in that, In step S3, based on the real-time dynamic index system, a two-stage algorithm including a first-stage Pareto game optimization solution and a second-stage real-time data feedback correction method is designed: It should be noted that in the second-stage real-time data feedback correction, the edge computing node collects real-time operation data, performs two-dimensional correction of technical parameters and income deviation on the initial evaluation result based on the real-time dynamic index system, and calls back the corrected parameters to the real-time dynamic index system to form a closed loop; The edge computing node collects real-time operation data, collects key data including but not limited to energy storage charging and discharging efficiency and photovoltaic irradiance through Internet of Things sensors at a frequency of 15 minutes / time, the parameter types of the collected key data correspond to the index dimensions of the three-dimensional dynamic index system one by one, and the collected real-time operation data is subjected to outlier elimination and time alignment.

10. The method for evaluating the economy of multi-energy complementary linkage according to claim 8, characterized in that, In step S3, based on the real-time dynamic index system, a two-stage algorithm including a first-stage Pareto game optimization solution and a second-stage real-time data feedback correction method is designed: The double-dimension correction automatically adjusts the corresponding index parameters in the three-dimensional dynamic index system when the preset parameters of the economic evaluation basic model of the multi-energy complementary system deviate from the actual data; meanwhile, the yield deviation rate is calculated by comparing the linkage yield of the economic evaluation basic model of the multi-energy complementary system with the actual settlement yield , expression: In the formula, represents the yield deviation rate, represents the predicted linkage yield, represents the actual linkage yield, represents the difference between the predicted linkage yield and the actual linkage yield, and when the deviation rate is greater than 5%, the benefit coefficient in the game target function is corrected by a PID controller. Expression: wherein represents the first corrected at the time represents the initial coefficient of the curtailment revenue, represents the proportional coefficient, represents the first deviation at the time represents the integral coefficient, represents the derivative coefficient, represents the first derivative of the deviation at the time with respect to time, and based on the corrected indicator parameter every 24 hours, the first stage algorithm is re-run and the initial evaluation result is updated.

Images