Port microgrid energy control strategy considering incomplete information

Abstract

The invention relates to a port micro-grid energy control strategy considering incomplete information, and the strategy is technically characterized in that the time-of-use electricity price of a power grid, the parameters of a micro-grid operator and the like are set, and a data source is determined; scene sampling is carried out, and uncertain information in micro-grid operation is converted into a computable deterministic scene; establishing a Bayesian-Stackelberg game model under incomplete information by taking a power transaction agent as a main body and a micro-grid operator as a slave body; a follower model constraint is established by considering port berth distribution, and a KKT condition is established for a follower model so that the double-layer game model can be converted into a single-layer optimization model; and solving the model by using a CPLEX solver to obtain an optimal electricity price and an energy control strategy of each scene. According to the method, the uncertainty of port micro-grid information is considered, port operation constraints, berth distribution strategies and the like are comprehensively considered, the economy of micro-grid operation is considered while the port logistics transportation efficiency is ensured, and a port micro-grid energy control strategy considering incomplete information is provided.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of port micro-grid energy control, and particularly relates to a port micro-grid energy control strategy considering incomplete information. BACKGROUND

[0002] With the promotion of port green low-carbon and intelligent transformation, the port micro-grid gradually becomes an important form to realize clean energy utilization and efficient energy scheduling. The port micro-grid integrates distributed energy, energy storage systems and shore power facilities to realize on-site consumption and optimal allocation of electric energy, improve port energy utilization rate and reduce carbon emissions. However, the port operating environment is complex, the ship operation, power load and energy supply are highly coupled, especially the berth allocation directly affects the shore power access timing and load distribution, which has a decisive role on the micro-grid scheduling strategy. If there is no effective coordination mechanism, it is easy to cause energy supply and demand imbalance, peak-valley load fluctuation and rising operating cost. The existing researches mostly adopt centralized optimization or hierarchical scheduling method, assuming that the information of port management, ship users and energy suppliers is completely available. However, in actual operation, there is significant information asymmetry among the subjects in terms of cost function, load demand, power generation capacity and arrival time; at the same time, uncertain factors such as renewable energy output fluctuation, ship delay and energy storage attenuation further aggravate the instability of the system.

[0003] To solve the above problems, the principal-agent game theory is widely used in energy system decision modeling. By establishing the hierarchical interaction relationship between the port management, ships and energy units, the system optimization considering economy and coordination can be realized. However, the existing researches are mostly limited to the complete information condition, and the influence of incomplete information on strategy selection and game equilibrium is not fully considered, and there is also a lack of comprehensive modeling method for dynamic coupling of berth allocation and energy control. Therefore, it is urgent to propose a port micro-grid energy control strategy considering incomplete information, which comprehensively considers the port operation constraints, berth allocation strategy and the like, so as to realize the coordinated decision and optimal operation of the micro-grid system under the condition of incomplete information. SUMMARY

[0004] The application aims to overcome the deficiencies of the prior art, and proposes a port micro-grid energy control strategy considering incomplete information, which ensures the port logistics transportation efficiency while considering the economy of micro-grid operation, and realizes the optimal scheduling and coordinated operation of the port energy system.

[0005] The application solves the technical problems by adopting the following technical solutions: A port micro-grid energy control strategy considering incomplete information comprises the following steps: Step 1, set the time-of-use price of the grid, microgrid operator parameters, etc., determine the data source, and generate the initial berthing time for each ship according to the arrival, departure, and operation time length, etc. Step 2, scene sampling is performed to convert uncertain information in microgrid operation into calculable deterministic scenarios. Step 3, establish a Bayesian-Nash game model and constraint conditions under incomplete information with the power trading agent as the leader and the microgrid operator as the follower. Step 4, consider the berth allocation in the port to establish the follower model constraint, combine the Bayesian-Nash game model under incomplete information in Step 3 to construct the KKT condition of the follower model as the constraint of the leader, and add it to the model to convert the double-layer game model into a single-layer optimization model. Step 5, directly solve the single-layer optimization model in Step 4 using the CPLEX solver to obtain the optimal electricity price and energy control strategy for each scenario. Step 1, set the time-of-use price of the grid, microgrid operator parameters, etc., determine the data source, and generate the initial berthing time for each ship according to the arrival, departure, and operation time length, etc. Step 2, in the port microgrid system, there are mainly uncertainties such as wind power output, photovoltaic output, and berth arrival time sequence. Through scene sampling, the uncertain information is converted into calculable deterministic scenarios. The probability distribution of wind turbine output power can be approximately subject to Weibull distribution, and its probability density function is shown in formula (1): (1) In the formula, is the wind speed probability density function; is the wind speed; is the shape parameter; is the scale parameter; The photovoltaic output power is approximately proportional to the solar irradiance, as shown in formula (2): (2) In the formula, is the photovoltaic output power; is the conversion efficiency; is the panel area; is the solar irradiance, which can be fitted as a Beta distribution from historical data; is the temperature coefficient; is the surface temperature of the solar cell panel; The ship arrival time and operation time length are affected by the channel, weather, and scheduling changes, and their error terms can be approximately subject to normal distribution, as shown in formula (3): (3) In the formula, This refers to the actual arrival time of the vessel. For the estimated arrival time of the vessel; This is the error time; It is a normal distribution function; Based on historical meteorological data and operational data, a joint probability distribution of the above uncertain events is constructed. The Latin hypercube sampling method is used to perform stratified uniform sampling on each random dimension to generate several scene sets as shown in equation (4): (4) In the formula, A collection of scenes; For the first The deterministic input matrix for each scenario; Total number of scenes; The deterministic input matrix for each scenario is shown in equation (5): (5) In the formula, For the first Wind power output in various scenarios; For the first Photovoltaic power generation in various scenarios; For the first Ship berthing load in various scenarios; For the first Basic load in each scenario; Step 3: Introduce Bayesian belief updates to replace the fixed probability assumptions. Based on historical operating data and expert experience, set the prior distribution of the microgrid type as shown in Equation (6): (6) In the formula, For microgrids The prior probability distribution of the type parameter; For microgrids Parameter space; The probability that a microgrid type falls into a certain parameter set reflects the leader's incomplete understanding of the behavior of the followers before the game begins. Each operating cycle collects microgrid response behavior data: power purchase and sale, electricity price response, energy storage charging and discharging decisions, etc., denoted as observation samples as shown in equation (7): (7) In the formula, For the first Observation samples for each operating cycle; For microgrids External input during this cycle; for its corresponding behavior response; The type distribution is updated using Bayes' rule as shown in equation (8). The updated The leader's belief distribution is used to compute the expected profit; (8) where, is the posterior probability distribution of the leader on the microgrid type after the th iteration; is the likelihood function, representing the probability of observing the data given that the microgrid type is ; and is the normalization constant. Let each microgrid have a private type , which is subject to a prior probability distribution and is not visible to the outside world but only known by itself. The overall type vector is shown in equation (9): (9) where, is the total number of microgrids; Leader's objective function: The leader is a power trading agent whose objective function is to maximize its expected revenue, as shown in equation (10): (10) where, is the price vector for buying and selling electricity; is the feasible strategy space; is the external environment random variable; is the joint distribution of the incomplete information and taking the mathematical expectation; is the leader's revenue function given the price , the body type , and the external disturbance ; and The price vector for buying and selling electricity by the power trading agent is shown in equation (11): (11) where, is the price for buying electricity at the power trading center in the time period; is the price for selling electricity at the power trading center in the time period; The leader's objective function given the belief distribution is shown in equation (12): (12) where, and for microgrid selling and buying electricity power from the trading center; for time step; for grid side cost; denotes taking expectation on the incomplete information of microgrid; for microgrid number; for time period; The revenue of power trading agent varies with the change of price and scenario, which is approximated by scenario sampling as shown in equation (13): (13) In the equation, for scenario weight; In order to ensure that the power trading agent can obtain profit, the purchase and sale electricity price constraints of the power trading agent are shown as equation (14): (14) Follower objective function: The follower is a microgrid aggregator, facing the given price strategy of the upper layer , selects the scheduling scheme that can minimize its total cost, as shown in equation (15): (15) In the equation, is the decision variable vector, including power flow, energy storage , flexible load response, etc.; is the feasible region of ; is the optimal decision variable vector; is the microgrid operation cost function; is the optimal decision operator, which represents taking to make ; The specific composition of the follower microgrid operation cost function is shown in equation (16): (16) In the equation, is the unit power generation cost of micro gas turbine; is the power generation of gas turbine; is the degradation cost function of energy storage charging and discharging; is the adjustment cost function of adjustable load; Step 4, combine the follower objective function in step 3 to build KKT conditions for the follower model as the constraints of the upper layer to the leader model, the port berth allocation constraints are shown as equations (17)-(19), and the port microgrid constraints are shown as equations (20)-(25): Constraint: The actual berthing time constraint of the ship is shown in equation (17): (17) In the equation, is the predicted arrival time of the ship; is the actual arrival time of the ship; is the scheduling period; The number of berthed ships constraint is shown in equation (18): (18) In the equation, is the total number of ships arriving at the port within the scheduling period; is the berthing state of the i-th ship; is the number of berths in the port; is the maximum berthing time; The port microgrid power balance constraint is shown in equation (20): (20) In the equation, is the wind power output; is the photovoltaic power output; is the energy storage discharging power; is the energy storage charging power; and are the power purchase and sale from the upper grid; is the port basic fixed load; is the berth shore power load; is the adjustable load; The energy storage SOC state of charge is shown in equation (21): (21) In the equation, is the state of charge of the energy storage at time period t; is the state of charge of the energy storage at time period t+1; and are the charging and discharging efficiencies; The energy storage SOC dynamic constraint is shown in equation (22): (22) In the equation, and are the minimum and maximum power boundaries allowed by the energy storage; (22) In the equation, and are the minimum and maximum power boundaries allowed by the energy storage; ​​The power purchase and sale power limit constraints are shown in equation (23): (23) where, and are the maximum power purchase and sale power; microgrid The micro-turbine output power and adjustable load constraints are shown in equations (24) and (25): (24) where, is the upper limit of the gas turbine output power; (25) where, is the upper limit of the adjustable load; The Lagrangian function for the follower model is shown in equation (26): (26) where, is the Lagrangian function of the microgrid under scenario ; is the Lagrangian multiplier corresponding to the equality constraint; is the equality constraint function; is the first inequality constraint function and ; ; is the Lagrangian multiplier corresponding to the first inequality constraint ; is the decision variable of the original problem; The optimization problem is replaced with the KKT necessary optimality conditions as shown in equation (27): (27) where, is the gradient of the cost function with respect to the decision variable ; is a set of real numbers; The complementary conditions are converted into linear constraints using the big M method, introducing a binary variable for each pair of complementary items as shown in equation (30): (28) where, is a binary variable used to switch the complementary relationship; is a constant large enough to represent the upper limit; The mathematical relationship is shown as formula (29): (29) The KKT condition is added to the model as the upper constraint, and the lower optimal solution Can be replaced by the variable Itself, without nesting solution, the double-layer game model is converted into a single-layer optimization model shown as formula (30): (30) Step 5, the joint optimization model in step 4 is directly solved by the CPLEX solver to obtain the optimal electricity price and the energy control strategy of each scene; The above port micro-grid energy control strategy considering incomplete information, the Bayesian belief update, the Latin hypercube sampling method, the KKT condition and the big M method are existing methods; The above port micro-grid energy control strategy considering incomplete information, the CPLEX solver is prior art and is well known to those skilled in the art; The above port micro-grid energy control strategy considering incomplete information, the Weibull distribution, the Beta distribution, the normal distribution, the prior distribution and the joint probability distribution are well known to those skilled in the art; The advantages and positive effects of the present application are: 1. The present application designs an energy control strategy under incomplete information, introduces incomplete information modeling and belief updating mechanism in the master-slave game framework, can accurately depict the cognitive difference and strategy uncertainty between the port management party and each distributed unit, improves the real applicability and robustness of the model, still realizes system-level coordination under the condition of asymmetric information, and enhances the adaptive decision-making ability of the port micro-grid; 2. The present application adopts incomplete information game modeling, introduces the Bayesian belief updating mechanism, realizes strategy learning and evolution under the condition of uncertain information, can adapt to the complex dynamic environment such as renewable energy output fluctuation and uncertain ship berthing time in the port scene, and improves the real-time adaptability and robustness of system operation; 3. The present application adopts the joint optimization method of double-layer to single-layer based on KKT condition, constructs the KKT condition of micro-grid aggregator considering berth allocation, linearly eliminates the nonlinear complementary items by using the big M method, realizes the conversion of the originally difficult-to-solve double-layer game optimization problem into single-layer model optimization, can quickly obtain the optimal solution satisfying the actual constraint, and effectively improves the economy of the port energy system; BRIEF DESCRIPTION OF DRAWINGS

[0006] The present application will be further described below in combination with the drawings and examples; Figure 1It is a flowchart of a port microgrid energy control strategy considering incomplete information. Figure 2 It is a master-slave game model structure diagram of a port microgrid considering incomplete information. Figure 3 It is a result diagram of the energy control strategy of the microgrid 1. Figure 4 It is a result diagram of the energy control strategy of the microgrid 2. DETAILED DESCRIPTION

[0007] Figure 1 The flow of the port microgrid energy control strategy considering incomplete information is: starting → initializing related parameters such as grid time-of-use electricity price, microgrid operator, ship berthing time, etc. → performing scene sampling → converting the uncertainties such as wind power output, photovoltaic output and berth berthing time sequence in the port microgrid into deterministic scenarios → introducing a Bayesian belief updating mechanism → establishing a Bayesian-Nash game model under incomplete information → converting the double-layer game model into a single-layer optimization model by constructing the KKT condition of the microgrid aggregator → solving the model by using a Cplex solver → judging whether the game reaches equilibrium or not → outputting the result if the equilibrium is reached, otherwise, re-solving → ending. Figure 2 The master-slave game model structure of the port microgrid considering incomplete information is: starting → initializing related parameters of power trading agents and berth allocation → initializing related parameters of the microgrid aggregator and berth allocation → solving the model of the microgrid aggregator by using a Cplex solver → calculating the objective function of the power trading agent → judging whether the game reaches equilibrium or not → outputting the result if the equilibrium is reached, otherwise, re-solving → ending. EMBODIMENT The present application proposes a port microgrid energy control strategy considering incomplete information, which adopts a PC as a platform for model building, wherein the CPU is i7-13650HX 2.60GHz, the installed memory is 8G, the operating system is Windows 64-bit, the MATLAB R2024b version is used, the flow of the game method is as shown in Figure 1 , and the game model structure is as shown in Figure 2 .

[0008] Step 1, set the grid time-of-use electricity price, microgrid operator parameters, etc., determine the data source, and generate the initial berthing time for each ship according to the information such as arrival, departure and operation time length; Step 2, perform scene sampling to convert the uncertain information in the microgrid operation into calculable deterministic scenarios; Step 3, take the power trading agent as the leader and the microgrid operator as the follower to establish a Bayesian-Nash game model and constraint conditions under incomplete information. Step 4, consider the port berth allocation to establish the follower model constraint, combine the Bayesian-Nash game model under incomplete information in step 3 to construct the KKT condition of the follower model as the constraint of the leader to add the model to make the double-layer game model into a single-layer optimization model; Step 5, directly solve the single-layer optimization model in step 4 with CPLEX solver to obtain the optimal electricity price and energy control strategy of each scenario; Step 1, set the time-of-use electricity price of the power grid, the microgrid operator parameters, etc., and generate the initial berthing time for each ship according to the arrival, departure and operation time length, etc. In (1)-(30) of the embodiment, the total number of microgrids , wherein the microgrid 1 has 3 berths, and the microgrid 2 has 2 berths; the scheduling period ; the step length ; the number of scenarios ; the marginal fuel cost ; the energy storage efficiency ; the time-of-use electricity price of the power grid is shown in Table 1; the microgrid aggregator parameters are shown in Table 2; Table 1 Time-of-use electricity price

[0009] Table 2 Microgrid aggregator parameters

[0010] According to the arrival, departure and operation time length, etc., the initial berthing time for each ship is generated, and the ship parameters are shown in Table 3: Table 3 Ship parameters

[0011] Step 2, scene sampling is performed, the probability density function of wind power output is established by formula (1); the photovoltaic output power function is established by formula (2); the ship arrival time function is established by formula (3); the scene set is established by formula (4); the deterministic input matrix under each scenario is established by formula (5); the parameter settings in scene sampling are shown in Table 4: Table 4 Parameter settings

[0012] Step 3, introduce Bayesian belief update instead of fixed probability assumption, establish the prior distribution of microgrid type by formula (6); represent the observation sample obtained in each operation period by formula (7); represent the updated microgrid type by formula (8); represent the microgrid type vector by formula (9); establish the Bayesian-Nash game model and constraint conditions under incomplete information with the power trading agent as the leader and the microgrid operator as the follower; Objective function: The objective function of the expected income of the power transaction agent of the game subject is established by using formula (10), the power purchase and sale price vector of the power transaction agent is represented by using formula (11), the objective function of the leader under the given belief distribution is established by using formula (12), the objective function obtained through scenario sampling is established by using formula (13), the dispatch scheme of the system is established by using formula (15), the micro-grid operation cost function is established by using formula (16), and then the constraint conditions are established: Constraint conditions: The power purchase and sale constraints of the power transaction agent are established by using formula (14); Step 4, the KKT conditions are constructed for the follower micro-grid model, the actual berthing time constraint of the ship is established by using formula (17), the number of berthed ships is established by using formula (18), the berthing time constraint is established by using formula (19), the power balance constraint is established by using formula (20), the energy storage SOC charge constraint is established by using formula (21), the energy storage SOC dynamic constraint is established by using formula (22), the energy storage SOC power limit constraint is established by using formula (23), the micro gas turbine output power constraint is established by using formula (24), and the adjustable load constraint is established by using formula (25); in the actual operation process of the port, the Lagrange function of the micro-grid is established by using formula (26); the necessary optimality condition is represented by using formula (27), the linear constraint of the complementary condition and the form of the binary variable are represented by using formula (28), and the complementary relationship is represented by using formula (29); the KKT conditions are taken as the constraint of the upper layer and added to the model, the leader target and all KKT conditions are integrated, the double-layer game model is converted into a single-layer optimization model, and the single-layer optimization model is established by using formula (32); the value range of M is shown in Table 5, the maximum value of the gas turbine output power, the maximum value of the port basic load; Table 5 M value range

[0013] Step 5, the single-layer model is directly solved by using the CPLEX solver, and the optimal electricity price is shown in Table 6; the energy control strategy results of the two micro-grids are shown in Figure 3 and Figure 4 ; Table 6 Power purchase and sale price of the power transaction agent

[0014] In this embodiment, in order to analyze the rationality and feasibility of the operation strategy of the application, three scenes I, II and III are set up for comparison, and the incomes of the power transaction agents and micro-grid aggregators of each system are calculated as shown in Table 7: (1) Scene I: without considering berth allocation, only the master-slave game of the port micro-grid is performed; (2) Scene II: master-slave game of port microgrid considering berth allocation; (3) Scene III: master-slave game of port microgrid considering berth allocation based on incomplete information; Table 7: Comparison results of different scenes

[0015] According to Table 7, the port microgrid energy control strategy considering incomplete information improves the income of the microgrid aggregator and also guarantees the income of the power transaction agent.

[0016] The Bayesian belief update, the Latin hypercube sampling method, the KKT condition and the big M method in the port microgrid energy control strategy considering incomplete information are all existing methods; the CPLEX solver is an existing technology; and the Weibull distribution, the Beta distribution, the normal distribution, the prior distribution and the joint probability distribution are all well known to those skilled in the art.

[0017] It should be emphasized that the embodiments described in the present application are illustrative rather than restrictive, and thus the present application is not limited to the embodiments described in the specific embodiments, and any other embodiments derived by those skilled in the art according to the technical solutions of the present application also belong to the scope of protection of the present application.

Claims

1. An energy control strategy for a port microgrid that considers incomplete information, characterized in that... Includes the following steps: Step 1: Set the time-of-use electricity price of the power grid, microgrid operator parameters, etc., determine the data source, and generate the initial berthing time for each ship based on information such as arrival, departure and operation duration; Step 2: Perform scenario sampling to transform the uncertain information in the operation of the microgrid into a computable deterministic scenario; Step 3: Using power trading agents as the main body and microgrid operators as the secondary body, establish a Bayesian-Stackelberg game model and its constraints under incomplete information. Step 4: Consider the port berth allocation and establish follower model constraints. Combine the Bayesian-Stackerberg game model with incomplete information in Step 3 to construct KKT conditions for the follower model as constraints for the leader and add them to the model to transform the two-level game model into a single-level optimization model. Step 5: Use the CPLEX solver to directly solve the single-layer optimization model in Step 4 to obtain the optimal electricity price and energy control strategies for each scenario. Step 1: Set the time-of-use electricity price of the power grid, microgrid operator parameters, etc., determine the data source, and generate the initial berthing time for each ship based on information such as arrival, departure and operation duration; Step 2: In the port microgrid system, there are uncertainties such as wind power output, photovoltaic power output, and berth berthing timing. Uncertain information is transformed into calculable deterministic scenarios through scenario sampling. The probability distribution of the wind turbine's output power can be approximated by a Weibull distribution, and its probability density function is shown in equation (1): (1) In the formula, Let be the wind speed probability density function; Wind speed; For shape parameters; For scale parameters; The photovoltaic output power is approximately proportional to the solar irradiance, as shown in equation (2): (2) In the formula, Photovoltaic output power; For conversion efficiency; The panel area; Solar irradiance can be fitted to a beta distribution using historical data; Temperature coefficient; Solar panel surface temperature; The arrival time and operation time of ships are affected by changes in waterways, weather and scheduling. The error term can be approximately distributed normally, as shown in equation (3): (3) In the formula, This refers to the actual arrival time of the vessel. For the estimated arrival time of the vessel; This is the error time; It is a normal distribution function; Based on historical meteorological data and operational data, a joint probability distribution of the above uncertain events is constructed. The Latin hypercube sampling method is used to perform stratified uniform sampling on each random dimension to generate several scene sets as shown in equation (4): (4) In the formula, A collection of scenes; For the first The deterministic input matrix for each scenario; Total number of scenes; The deterministic input matrix for each scenario is shown in equation (5): (5) In the formula, For the first Wind power output in various scenarios; For the first Photovoltaic power generation in various scenarios; For the first Ship berthing load in various scenarios; For the first Basic load in each scenario; Step 3: Introduce Bayesian belief updates to replace the fixed probability assumptions. Based on historical operating data and expert experience, set the prior distribution of the microgrid type as shown in Equation (6): (6) In the formula, For microgrids The prior probability distribution of the type parameter; For microgrids Parameter space; The probability that a microgrid type falls into a certain parameter set reflects the leader's incomplete understanding of the behavior of the followers before the game begins. Each operating cycle collects microgrid response behavior data: power purchase and sale, electricity price response, energy storage charging and discharging decisions, etc., denoted as observation samples as shown in equation (7): (7) In the formula, For the first Observation samples for each operating cycle; For microgrids External input during this cycle; The corresponding behavioral response; The type distribution is updated using Bayes' theorem as shown in equation (8). The distribution of a leader's beliefs is used to calculate expected profits; (8) In the formula, For the first The posterior probability distribution of the leader for the microgrid type after the next iteration; Let be the likelihood function, representing the type of microgrid. In this case, observation data The probability of occurrence; This is a normalization constant; Assume each microgrid has a private type It follows a prior probability distribution, is not visible to the outside world but is known only to itself, and is a vector of overall type. As shown in equation (9): (9) In the formula, Total number of microgrids; Leader objective function: The game leader is the electricity trading agent, whose objective function is to maximize their expected profit, as shown in equation (10): (10) In the formula, The electricity purchase and sale price vector; This represents the feasible strategy space. It is a random variable of the external environment; To address incomplete information and The joint distribution takes the expected value; To at a given price Body type External disturbances Below, the leader's payoff function; Electricity trading agent purchase and sale price vector As shown in equation (11): (11) In the formula, for Electricity purchase price at the time-of-use power trading center; for Electricity prices at the time-of-use power trading center; The objective function obtained by the leader under a given belief distribution is shown in equation (12): (12) In the formula, and For microgrids Electricity sold and bought at the trading center; For time step; For grid-side costs; This represents taking the expectation of incomplete information about the microgrid; Number the microgrid; For a period of time; The revenue of electricity trading agents varies with price and scenario, and can be approximated by scenario-based sampling as shown in equation (13): (13) In the formula, As scene weight; To ensure that electricity trading agents can make a profit, the electricity purchase and sale price constraints for electricity trading agents are as shown in equation (14): (14) Follower objective function: Followers are micro-network aggregators, subject to the pricing strategies given by the upper levels. When choosing a scheduling scheme that minimizes its total cost, as shown in equation (15): (15) In the formula, The decision variable vector includes power flow and energy storage. Flexible load response, etc.; for The feasible domain; The vector of optimal decision variables; This is a function of the microgrid operating cost. For the optimal decision operator, represents taking / making smallest ; The specific composition of the operating cost function of the follower microgrid is shown in equation (16): (16) In the formula, The unit power generation cost of a micro gas turbine; For the power generation of gas turbines; The energy storage charging and discharging degradation cost function; The adjustment cost function for adjustable load; Step 4: Combine the follower objective function from Step 3 to construct KKT conditions for the follower model and add them to the leader model as upper-level constraints. The port berth allocation constraints are shown in equations (17)-(19), and the port microgrid constraints are shown in equations (20)-(25). Constraints: The constraints on the actual berthing time of the ship are as shown in equation (17): (17) In the formula, The estimated arrival time of the vessel; This refers to the actual arrival time of the vessel. The scheduling period; The constraint on the number of ships at berth is shown in equation (18): (18) In the formula, This represents the total number of ships arriving at the port within the scheduling period. For the first The berthing status of the vessel; The number of port berths; The vessel waiting time constraint is shown in equation (19): (19) In the formula, This is the maximum waiting time. The power balance constraint of the port area microgrid is shown in equation (20): (20) In the formula, Powering wind power; Contribute to photovoltaic power; This refers to the energy storage discharge power; Power for energy storage charging; and This refers to the power capacity for purchasing and selling electricity from the upper-level power grid; For the port area's basic fixed load; For berth shore power load; For adjustable load; The state of charge (SOC) of the energy storage is shown in equation (21): (21) In the formula, For energy storage during the period The state of charge; For energy storage during the period The state of charge; and For charging and discharging efficiency; The dynamic constraints of energy storage SOC are shown in equation (22): (22) In the formula, and These are the minimum and maximum allowable energy boundaries for energy storage; The power purchase and sales limits are constrained as shown in equation (23): (23) In the formula, and This refers to the maximum power purchase and sales capacity. microgrids The output power and adjustable load constraints of the internal micro gas turbine are shown in equations (24) and (25): (24) In the formula, This is the upper limit of the gas turbine's output power; (25) In the formula, Adjustable load limit; The Lagrangian function for the follower model is constructed as shown in equation (26): (26) In the formula, For microgrids In the scene The Lagrangian function under the given conditions; These are the Lagrange multipliers corresponding to the equality constraints; This is the equality constraint function; For the first Inequality constraint functions and ; In order to be with the first Inequality constraints The corresponding Lagrange multipliers; These are the decision variables for the original problem; The optimization problem can be equivalently replaced by the KKT necessary optimality condition as shown in equation (27): (27) In the formula, Cost function For decision variables The gradient; It is the set of real numbers; The Big M method is used to transform the complementarity conditions into linear constraints, introducing binary variables for each pair of complementary terms. As shown in equation (30): (28) In the formula, It is a binary variable used to represent the complementary relationship between the switches; It is a sufficiently large constant, representing the upper bound; The mathematical relationship is shown in equation (29): (29) The KKT conditions are added to the model as upper-level constraints, and the optimal solution at the lower level is obtained. It can be replaced by a variable The two-layer game model is transformed into a single-layer optimization model as shown in equation (30), without the need for nested solutions. (30) Step 5: Use the CPLEX solver to directly solve the joint optimization model in Step 4 to obtain the optimal electricity price and energy control strategies for each scenario.

2. The energy control strategy for a port microgrid that considers incomplete information as described in claim 1, characterized in that: In the Bayesian belief update, the leader establishes a rolling update mechanism by dynamically sampling historical power trading data. When new data arrives, the Bayesian update formula is used to recalculate the posterior distribution of the microgrid type, thereby achieving real-time learning and optimization of the microgrid's operating characteristics.

Images