Multi-stage Optimal Scheduling Method and System for Seaport Logistics-Energy Coupling System

Abstract

A multi-stage optimal scheduling method and system for a seaport logistics-energy coupling system. After establishing a logistics system model, an optimal logistics-power flow model, and an uncertainty model, a method based on a hybrid binary and continuous affine strategy is used to solve for the logistics scheduling and energy scheduling schemes under the coordination of logistics and energy. On the basis of meeting the requirements of logistics operations, the overall operating cost of the system is reduced, and the renewable energy consumption rate is increased. The present invention also establishes an optimal logistics-power flow model for the spatial and temporal attributes of ship berth allocation and quay crane scheduling. By adjusting the ship docking time and location and the operating state of the quay crane, and cooperating with energy scheduling, the logistics flexibility is fully utilized in port energy scheduling, promoting the economic and efficient operation of the port energy system.

Description

Technical Field

[0001] The present invention relates to a technology in the field of seaport logistics, specifically a multi-stage optimal scheduling method and system for a seaport logistics-energy coupling system considering uncertainty. Background Art

[0002] With the deepening of port electrification, the port logistics system and the energy system are coupled with each other. Ships and handling machinery become part of the energy system through onshore power supply technology and electric energy substitution technology respectively. After port electrification, the logistics operation plan will have an inestimable impact on the energy system scheduling. However, the current logistics scheduling plan at ports still adopts the traditional method, that is, only focusing on logistics efficiency and ignoring the impact of logistics operations on the energy system, resulting in low overall energy efficiency and high operating costs in electrified ports. The current research deficiencies specifically include:

[0003] 1) Currently, ports still adopt the traditional logistics scheduling plan, that is, only maximizing logistics efficiency under the constraints of logistics operations is considered in the optimization model, without considering the coupling relationship between logistics operations and energy scheduling and the impact of logistics operations on energy scheduling, resulting in the inability to fully exert the advantages of the coordinated operation of the port logistics system and the energy system.

[0004] 2) When the logistics system and the energy system are coupled, the uncertainties from the two systems are also coupled, and the impacts of these two types of uncertainties on the operation of the coupled system need to be considered collaboratively. However, the current research on logistics uncertainty mainly focuses on the maritime field, and no research has considered the impact of logistics uncertain factors on the operation of the energy system. In addition, the traditional two-stage robust optimization method for dealing with uncertainty violates the non-anticipated logistics operations in decision-making. When the actual arrival time of the ship deviates from the expected value, the pre-determined logistics operation decision will become infeasible, reducing the effectiveness of system decision-making.

[0005] 3) The multi-stage robust optimization method is usually used to solve the non-anticipated problem of the two-stage robust method. However, the traditional solution methods for multi-stage robust problems (such as robust dual dynamic programming, linear affine strategies, etc.) can only solve models constructed by continuous variables. However, the port logistics system model consists of a large number of discrete variables, resulting in the current solution methods being unable to be used for the port multi-stage logistics-energy collaborative scheduling problem with mixed integer properties.

[0006] The scheduling objective of the existing port logistics scheduling method based on the Dijkstra algorithm is to maximize logistics efficiency (such as minimizing the docking duration of ships). The scheduling model only includes logistics operation constraints and does not consider the impact of logistics scheduling on the port energy system. With the deepening of port electrification, the logistics side and the energy side of the port are gradually coupled. For example, shore power supply technology enables ships to be connected to the port power grid, and ships become part of the port power grid load; electric energy substitution technology enables traditional fuel-driven handling machinery to be converted into electric drive, and the electrified handling machinery also becomes part of the port power grid load. Therefore, in an electrified port, the logistics scheduling plan directly determines the spatio-temporal distribution of the logistics power load and affects the port energy system. If only the single scheduling of the logistics system is considered and only the maximization of logistics transportation efficiency is pursued at this time, the flexibility of logistics operations cannot be fully utilized to cooperate with the energy system scheduling, resulting in unnecessary energy waste and high energy consumption. Summary of the Invention

[0007] Aiming at the deficiencies of the prior art in the coordination of port logistics operations and energy scheduling, only considering the impact of the uncertainty of ship arrivals on logistics scheduling, the impact of the uncertainty of renewable energy generation on energy scheduling, and the lack of research considering the impact of dual uncertainties on the logistics-energy coupling system, the present invention proposes a multi-stage optimal scheduling method and system for a seaport logistics-energy coupling system. At the same time, an optimal logistics-power flow model is established for the spatial and temporal attributes of ship berth allocation and quay crane scheduling. By adjusting the ship docking time, docking location, and the operating state of quay cranes, it cooperates with energy scheduling, so as to give full play to the logistics flexibility in port energy scheduling and promote the economic and efficient operation of the port energy system.

[0008] The present invention is realized through the following technical solutions:

[0009] The present invention relates to a multi-stage optimal scheduling method for a seaport logistics-energy coupling system. After establishing a logistics system model, an optimal logistics-power flow model, and an uncertainty model, a method based on a hybrid binary and continuous affine strategy is used to solve for the logistics scheduling and energy scheduling plans under the coordination of logistics and energy. On the basis of meeting the requirements of logistics operations, the overall operating cost of the system is reduced, and the renewable energy consumption rate is increased.

[0010] The present invention relates to a system for implementing the above method, including: a data acquisition module, a model calculation module, and an instruction execution module, where: the data acquisition module collects parameters of the logistics system and the energy system, predicts the output of new energy, dynamically updates the ship arrival time, and inputs the collected data into the model calculation module; the model calculation module takes the collected and predicted data as input, solves the port logistics-energy multi-stage robust dynamic optimization model, and outputs the optimal logistics and energy scheduling control instructions; the instruction execution module issues the scheduling instructions to the port control room to execute the obtained optimal scheduling plan.

[0011] Technical effects

[0012] The present invention fully exploits the flexibility of logistics operations through the logistics-energy coupling modeling technology, enabling two independent systems to operate in coordination under a unified scheduling framework and time scale; through the logistics-energy multi-stage dynamic robust optimization technology, considering the uncertainty of ship arrivals from the logistics system and the uncertainty of renewable energy generation from the energy system, through multi-stage dynamic robust optimization, the feasibility and economy of the scheduling strategy can be achieved under any uncertainty conditions; for the model conversion and solution technology of multi-stage logistics-energy scheduling, based on mixed affine decisions, through a series of derivations and conversions, the original complex model with a multi-layer nested structure is converted into a single-layer mixed integer optimization model that is easy to solve, enabling the proposed logistics-energy multi-stage rolling robust optimization scheduling technology to be applied to the port scheduling system, and realizing the actual implementation and application of complex theoretical methods. Description of the drawings

[0013] Figure 1 It is a flow chart of the multi-stage dynamic robust optimization of port logistics-energy;

[0014] Figure 2 It is a schematic diagram of the dynamic update of ship arrival time;

[0015] Figure 3 It is a schematic diagram of the topology of the Rizhao Port power system;

[0016] Figure 4 It is a schematic diagram of ship docking decision-making;

[0017] Figure 5 It is a schematic diagram of sensitivity analysis;

[0018] Figure 6 It is a schematic diagram of calculation time comparison. Detailed implementation manners

[0019] As Figure 1 shown, this embodiment relates to a multi-stage optimal scheduling method for a seaport logistics-energy coupling system, including:

[0020] Step 1. Establish a logistics system model: Logistics operations include berth allocation and quay crane allocation. The arriving ships first wait in the anchorage. After receiving the dispatching signal from the seaport, they start to berth at the allocated berths and allocate quay cranes to the berthed ships for cargo handling. On the basis of meeting the logistics operation constraints, by adjusting the logistics operation plan, the load transfer can be fully realized by making use of the spatio-temporal flexibility on the logistics side.

[0021] The described logistics system model includes: ship berthing status, berth allocation constraints, and quay crane allocation constraints, where:

[0022] Three-dimensional binary ship berthing status When ship s berths at berth b at time t, then X bst is 1, otherwise it is 0;

[0023] The berth allocation constraints include: berthing duration and berthing position restrictions, ship berth restrictions, specifically: Among them: and respectively represent the arrival time, berthing start time and departure time of ship s, is the maximum berthing duration of ship s, B s is the berth number where ship s docks, B max is the total number of onshore power berths at the port terminal.

[0024] The quay crane allocation constraints include: quay crane - ship restrictions, total quay crane restrictions, quay crane restrictions for cargo handling tasks, idle quay crane restrictions, quay crane position restrictions and quay crane cross - over restrictions, specifically:

[0025] Among them: ω qst is a binary variable, ω qst = 1 indicates that quay crane q provides loading and unloading services for ship s at time t, Q st represents the total number of quay cranes serving ship s at time t, and Q max is the total number of quay cranes at the port terminal, represents the minimum number of quay cranes that ship s can bear due to the hull length limit, represents the maximum number of quay cranes that ship s can bear due to the hull length limit, η is the container loading and unloading efficiency of a single quay crane, TEU s is the number of standard containers on ship s, TEU is the measurement unit of standard containers, B qt is the position of quay crane q at time t. If quay crane q serves ship s at time t, its position is the same as that of ship s, δ qt is also a binary variable, δqt = 1 indicates that the quay crane q is in the working state at time t, and

[0026] Step 2: Establish an optimal logistics-power flow model to find the best logistics and energy scheduling decisions under logistics and power dispatch constraints. The goal is to minimize the power purchase cost of the main grid, that is The constraint conditions include linear DistFlow equations, load constraints of ships and quay cranes, renewable energy generation constraints, line power flow constraints, and node voltage magnitude constraints, specifically:

[0027] Where: is the power purchase price of the main grid, is the power purchase quantity from the main grid, P ij,t , Q ij,t and P jk,t , Q jk,t are the active power flow and reactive power flow from node i to node j and the active power flow and reactive power flow from node j to node k at time t, respectively. Ω(j) and Θ(j) are the upstream node and downstream node connected to node j, respectively. If and only if node j is connected to the grid node, are the active load and reactive load of node j at time t, respectively. U i,t is the voltage magnitude of node i at time t, r ij , x ij are the resistance and reactance of line ij, respectively. are the ship power and quay crane power on node j at time t, respectively. are the rated power of ship s and the rated power of quay crane q, respectively. Π(j) and Φ(j) are the sets of all shore power berths and quay cranes electrically connected to node j, respectively. is the conventional power load of node j. is the actual output power of renewable energy unit m at time t. is the predicted output power of renewable energy unit m at time t.

[0028] ξ m,t is the deviation value between the actual power and the predicted power of renewable energy unit m at time t. U i,t is the voltage magnitude of node i at time t. is the rated capacity of line ij. are the minimum and maximum values of the voltage magnitude of node j, respectively.

[0029] The compact form of the optimal logistics-power flow model described above is Where: x t is a vector representing binary variables, including {X bst , ω qst}, y t is a vector representing continuous variables, including z t is a vector representing integer variables, including {B qt , Q st , δ st}, I is an integer variable independent of time, including ξ t represents an uncertain variable, indicating the deviation of new energy output.

[0030] Step 3: Establish an uncertainty model, including: ship arrival uncertainty model, renewable energy uncertainty model, and multi-stage robust dynamic optimization model.

[0031] The ship arrival uncertainty model mentioned above means that for the ships arriving in the current stage, their arrival times are exactly known. For the ships arriving in the remaining stages, their arrival times are estimated by the ship drivers according to the navigation situation and conveyed to the port. Based on the dynamic update of the ship arrival time, a feasible logistics operation plan can always be obtained at each stage regardless of the future ship arrival times.

[0032] For example: The example of the dynamic update process of the ship arrival time is as Figure 2 shown. In stage 1, the black and blue ships are estimated to arrive at times 2 and 3 respectively. In the 2nd stage, the black and blue ships update their estimated arrival times, but no ships arrive at the port in this stage. In stages 3 and 4, the accurate arrival times of the black and blue ships can be obtained respectively.

[0033] The renewable energy uncertainty model mentioned above means that: Where: ξ t is the deviation of new energy output, ξ m,t , are the lower and upper limits of the deviation of new energy output respectively, and Γ is a parameter controlling the conservatism of the uncertainty set.

[0034] The multi-stage robust dynamic optimization model mentioned above means that the multi-stage robust dynamic optimization model of the optimal logistics-power flow model makes sequential decisions dynamically within the scheduling period to minimize the total cost. The logistics operation plan is made according to the accurate ship arrival time and the worst case of renewable power generation at the beginning of stage t; after determining the logistics operation plan, the optimal power flow calculation is carried out according to the actual renewable power generation. The decision-making process of each stage will be executed step by step until stage T, specifically:

[0035] Step 4. Solve the model established in Steps 1 - 3 by using a method based on a hybrid binary and continuous affine strategy, specifically including:

[0036] 4.1) Equivalent transformation of the objective function, including: And further obtain the following equivalent compact model:

[0037]

[0038] HI ≤ d, where: The variables x τ (ξ τ-1 ), z τ (ξ τ-1 ) are determined before the uncertainty variable ξ τ is revealed, so they are regarded as affine functions of the uncertainty variable ξ τ-1 . The variable y τ (ξ τ ) is determined after the uncertainty variable ξ τ is revealed, so it is regarded as an affine function of the uncertainty variable ξ τ .

[0039] 4.2) Apply the linear and binary affine strategy to further map the decision variables x τ (ξ τ-1 ), z τ (ξ τ-1 ), y τ (ξ τ ):

[0040] Where: ξ t = (ξ 0,t , …, ξ M,t ), G(ξ t ) = (G(ξ 0,t ), …, G(ξ M,t )), ξ 0,t = 1, G(ξ 0,t ) = 1 are constant terms in the affine strategy, and Y t , X t , Z t are new decision variables to be solved in the affine strategy.

[0041] Apply the above affine strategy to obtain the following equivalent model:

[0042] 4.3) The affine strategy introduces the non - linear affine function G(ξ t), for the linearized model, it is necessary to construct the lifted uncertainty set and its convex hull as follows: where: Ξ′ t is the lifted uncertainty set, ξ′ t is the lifted uncertain variable, conv(Ξ′ t ) is the convex hull of the lifted uncertainty set, ε t is the parameter for constructing the convex hull, and W t , V t , h t are all coefficient matrices.

[0043] Substituting the above into the formula, we get:

[0044] 4.4) To obtain the scheduling strategy in the worst case of the uncertain variable, the dual theory is used to transform the model into a single-layer mixed-integer optimization model: where: Λ, ψ are dual variables, and L is a coefficient matrix.

[0045] The above dualized model is a single-layer mixed-integer optimization model, and a commercial solver (such as GUROBI) can be directly used for solving. The solution process adopts the form of rolling optimization, that is, let t be equal to 1,..., T in turn, and the model is solved in each period. After solving the model, the x t , z t , Y t at the current stage can be obtained. Among them, x t , z t are logistics operation scheduling variables. Before the uncertain variable ξ t is revealed, x t , z t can be directly used for the logistics operation scheduling decision at the current stage. After the uncertain variable ξ t is revealed, according to y t (ξ t ) = Y t ξ t and the value of Y t , the power scheduling plan y t can be obtained.

[0046] In this embodiment, taking the Rizhao Port logistics operation system and the power distribution system supporting the operation of the logistics system as examples, the data of Rizhao Port is used to verify the proposed method. The topological structure of the distribution network is as Figure 3As shown in the figure, there are 6 berths, 12 quay cranes and 3 photovoltaic units. The connections of berths, quay cranes, photovoltaic units to the distribution network are shown in Table I. Table II shows the ship parameters. Each quay crane processes 50 TEUs per hour and has a rated power of 0.3 MW. The main grid electricity price is $0.127 per kWh. The voltage amplitude is limited to 0.95 to 1.05 p.u. The scheduling period is 48 hours, i.e., T = 48, and the model time resolution is 1 hour.

[0047] In this embodiment, the following three cases are set for comparison.

[0048] Case 1: Multi-stage dynamic optimization without considering logistics flexibility. The logistics operation plan is determined independently of the power dispatch in each stage to minimize the total berthing time. Then, the optimal power flow is implemented with the determined logistics operation plan.

[0049] Case 2: Two-stage robust optimization of the optimal logistics-power flow model. In the first stage, the logistics operation plan is determined according to the estimated ship arrival time, and in the second stage, the power dispatch plan is determined according to the photovoltaic power generation.

[0050] Case 3: The proposed multi-stage robust dynamic optimization method for the optimal logistics-power flow.

[0051] Table I Corresponding relationship between electrical nodes and connected devices

[0052]

[0053]

[0054] Table II Ship parameters

[0055] Vessel Number Expected Arrival Time / hour Actual Arrival Time / hour Number of Containers / TEU Power Demand / MW 1 5 5 1035 3.0 2 8 14 1200 3.5 3 13 9 980 3.0 4 15 15 850 2.5 5 17 17 870 4.0 6 27 29 1245 3.5 7 29 27 960 2.5 8 36 36 1470 4.0 9 38 34 1025 3.5 10 42 39 830 2.0

[0056] First, the feasibility of the scheduling strategy is analyzed. For Case 1 and Case 3, the berthing decisions of all ships are feasible because the logistics operation plan is dynamically determined according to the actual ship arrival time in each stage. However, in Case 2, the day-ahead logistics operation plan is determined according to the estimated ship arrival time. Due to the deviation between the actual arrival time and its estimate, the decisions of some ships cannot be executed. In Figure 4 (b), Ship 2 is scheduled to start berthing at 10:00 according to the expected arrival time (8:00). However, Ship 2 actually arrives at 14:00, which makes the day-ahead logistics operation plan infeasible. To obtain a feasible solution, Ship 2 is rescheduled to start berthing at 14:00. Although the scheduling strategy becomes feasible after rescheduling, due to the mutual influence of ships, this plan is not optimal. Specifically, as Figure 4(c) As shown, it is optimal for Ship 3 to start berthing at 10:00, but this cannot be achieved in Case 2 because Ship 2 is scheduled to occupy Berth 2. Therefore, Case 2 cannot fully utilize the logistics flexibility, thereby reducing the benefits of the optimal logistics-power flow model for the power system.

[0057] To evaluate the economics of Cases 1 - 3, 1000 random scenarios of PV output are generated as uncertainty realizations. The average values of the optimization results are listed in Table III.

[0058] Table III Comparison of Optimization Results

[0059]

[0060]

[0061] In Case 1, the logistics operation plan is not coordinated with the power dispatch. All ships start berthing immediately after arriving at the seaport and depart as soon as the cargo handling tasks are completed. Therefore, the loads of the ships and quay cranes cannot match the PV power generation, resulting in a higher electricity purchase cost. In Case 2, the inaccurate ship arrival times reduce the optimality of the day-ahead logistics operation plan, leading to an increase in costs. Compared with Cases 1 - 2, Case 3 can fully exploit the demand response potential on the logistics side through the multi-stage dynamic optimization of the optimal logistics-power flow model, achieving better PV accommodation and economic costs.

[0062] Further explore the influence of the electricity price model and PV unit capacity on the optimization results. First, the time-of-use electricity price is adopted, with the prices set at $0.068, $0.116, and $0.163 per kWh during low (1:00 - 7:00, 24:00), normal (8:00 - 10:00, 16:00 - 18:00, 22:00 - 23:00), and peak (11:00 - 15:00, 19:00 - 21:00) hours respectively.

[0063] As Figure 4 (a)- Figure 4(c) shows the ship scheduling results under the scheduling schemes of Case 1 - Case 3. It can be seen that in Case 1, the traditional scheduling method (where the logistics system and the energy system operate independently), the ship scheduling efficiency is the highest. The ship docks immediately after arriving at the port and departs from the port as soon as possible. However, in this case, considering the coordination with the energy system scheduling, the photovoltaic power generation cannot be effectively utilized to reduce the electricity purchase cost. Therefore, the overall operating cost is the highest. In Case 2, the dynamic arrival of ships is not considered. Therefore, when there is a deviation between the actual arrival time of the ship and the predicted arrival time, the day-ahead logistics scheduling scheme becomes infeasible, reducing the optimality of the scheduling decision. Case 3 is the scheduling method proposed in this method. It can not only ensure the feasibility of the logistics scheduling strategy but also effectively coordinate the logistics system and the energy system, effectively reducing the overall operating cost of the system while meeting the requirements of ship operations.

[0064] As Figure 5 (a) shows, the advantages of the model of the present invention are more obvious under the time-varying electricity price mode. The power loads of ships and quay cranes can be distributed during periods of high photovoltaic output or low electricity prices. Therefore, the logistics flexibility is improved, resulting in an increased gap in economic costs among Case 1 - Case 2 and Case 3.

[0065] As Figure 5 (b) shows, the optimization results for different capacities of photovoltaic units. When the single photovoltaic unit capacity is low, the economic benefits of the proposed method are not obvious. However, as the capacity of the photovoltaic unit increases, more photovoltaic power generation can be used for load transfer. Therefore, the economic performance of Case 3 gradually becomes prominent.

[0066] Continue to compare the computational efficiency of Case 1 and Case 3. Since the optimization models of Case 1 and Case 3 are solved dynamically at each stage, the computational time at each stage is as Figure 6 shown. The model of Case 2 is solved by a one-time calculation, and its computational time is also added Figure 6 for comparison. As the problem scale decreases, the computational times of Case 1 and Case 3 decrease exponentially. For Case 1, the logistics operations (i.e., integer programming) and power dispatching (i.e., linear programming) are solved independently. Therefore, each stage can be quickly completed within 5 seconds. The model of Case 2 is a mixed integer linear programming and can be quickly solved by a commercial solver. In Case 3, the model scale is larger than that of Case 1 - Case 2. However, the longest solution time for a single stage is much less than the model time resolution (i.e., 1 hour). Although the solution time of Case 3 is longer than that of Case 1 - Case 2, the proposed model can still be solved within an acceptable time range.

[0067] Compared with the existing port logistics scheduling methods that only consider logistics efficiency and do not consider the impact of energy consumption generated by logistics operations on the operating cost of the energy system, this method models the logistics system and the energy system under a unified scheduling framework. The optimization model includes both logistics operation constraints and energy scheduling constraints. By minimizing the total system operating cost, the collaborative optimization of logistics and energy is achieved. For example, by adjusting the docking time of ships, the transfer of ship power load on the time scale can be realized. If the ship is made to dock during the period with higher renewable energy generation or lower grid electricity price, the power supply cost of the ship can be effectively reduced, achieving the effect of reducing the system operating cost while meeting the on-time completion of logistics operations. Therefore, compared with the traditional port logistics scheduling methods in the listed method technologies, this method extends the single logistics scheduling to logistics-energy collaborative scheduling, which can not only meet the constraint conditions required by logistics operations but also achieve the spatio-temporal transfer of logistics load through the collaboration of the two, achieving the effects of reducing the system operating cost and increasing the renewable energy consumption rate.

[0068] The above specific implementation can be locally adjusted in different ways by those skilled in the art without departing from the principles and purposes of the present invention. The protection scope of the present invention is subject to the claims and is not limited by the above specific implementation. All implementation solutions within its scope are subject to the present invention.

Claims

1. A multi-stage optimization scheduling method for a seaport logistics-energy coupling system, characterized in that: include: Step 1: Establish a logistics system model: Logistics operations include berth allocation and quay crane allocation. Arriving ships first wait at the anchorage. After receiving the port's dispatch signal, they begin berthing at the assigned berth and quay cranes are assigned to the berthed ship for cargo loading and unloading. Based on meeting logistics operation constraints, the logistics operation plan is adjusted to fully utilize the temporal and spatial flexibility of the logistics side to achieve load transfer. Step 2: Establish an optimal logistics-power flow model to find the best logistics and energy scheduling decisions under the constraints of logistics and power scheduling. The goal is to minimize the power purchase cost of the main grid, that is, The constraints include the linear DistFlow equation, ship and quay crane load constraints, renewable energy generation constraints, line flow constraints, and node voltage amplitude constraints. Specifically: U i,t -U j,t =r ij P ij,t +x ij Q ij,t , in: The electricity price for the main grid is is the amount of electricity purchased from the main network, P ij,t , Q ij,t and P jk,t , Q jk,t are the active power flow and reactive power flow from node i to node j and the active power flow and reactive power flow from node j to node k at time t, Ω(j) and Θ(j) are the upstream and downstream nodes connected to node j, respectively. If and only if node j is connected to a grid node, are the active load and reactive load of node j at time t, U i,t is the voltage amplitude of node i at time t, r ij 、x ij are the resistance and reactance of line ij respectively, are the ship power and quay crane power at node j at time t, are the rated power of ship s and the rated power of quay crane q, Π(j) and Φ(j) are the set of all shore power berths and quay cranes electrically connected to node j, respectively. is the predicted output power of renewable energy unit m at time t, ξ m,t is the deviation between the actual power and the predicted power of renewable energy unit m at time t, is the rated capacity of line ij, are the minimum and maximum values of the voltage amplitude at node j, respectively; Step 3: Establish uncertainty models: including ship arrival uncertainty model, renewable energy uncertainty model and multi-stage robust dynamic optimization model; Step 4: Solve the model established in Steps 1 to 3 using a method based on mixed binary and continuous affine strategies; The ship arrival uncertainty model is as follows: for ships arriving in the current phase, their arrival times are precisely known; for ships arriving in the remaining phases, their arrival times are estimated by the ship's pilot based on navigation conditions and communicated to the port. Based on the dynamic update of ship arrival times, a feasible logistics operation plan can always be obtained at each phase, regardless of the future arrival time of the ship. The renewable energy uncertainty model is: Where: t For the output deviation of new energy, m,t 、 are the lower and upper limits of the new energy output deviation, respectively, and Γ is the parameter for controlling the conservative type of the uncertainty set; The multi-stage robust dynamic optimization model is a multi-stage robust dynamic optimization model of the optimal logistics-power flow model. It dynamically makes sequential decisions within the scheduling cycle to minimize the total cost. The logistics operation plan is made at the beginning of stage t based on the accurate ship arrival time and the worst-case scenario of renewable power generation. After the logistics operation plan is determined, the optimal power flow is calculated based on the actual renewable power generation. The decision-making process of each stage will be executed step by step until stage T, specifically:

2. The multi-stage optimization scheduling method for the seaport logistics-energy coupling system according to claim 1 is characterized in that: The logistics system model includes: ship berthing status, berth allocation constraints, and quay crane allocation constraints, among which: Three-dimensional binary ship berthing status When ship s berths at berth b at time t, then X bst is 1, otherwise 0; Berth allocation constraints include: berthing duration and berthing position restrictions, and ship berth restrictions, specifically: in: and They represent the arrival time, berthing start time and departure time of ship s respectively, is the maximum port stay of ship s, B s is the berth number of ship s, B max is the total number of shore power berths at the port terminal; The quay crane allocation constraints include: quay crane vessel restrictions, total number of quay crane restrictions, cargo loading and unloading task quay crane restrictions, idle quay crane restrictions, quay crane location restrictions, and quay crane crossing restrictions. Specifically: Among them: binary variable ω qst =1 means that the quay crane q provides loading and unloading services for the ship s at time t, Q st represents the total number of quay cranes serving ship s at time t, and Q max is the total number of quay cranes at the port terminal, It represents the minimum number of quay cranes that the ship can bear due to the limitation of the hull length. They represent the maximum number of quay cranes that ship s can bear due to the limitation of hull length, η is the container loading and unloading efficiency of a single quay crane, TEU s is the number of standard containers on ship s, TEU is the unit of measurement for standard containers, B qt is the position of the quay crane q at time t. If the quay crane q serves the ship s at time t, then its position is the same as that of the ship s. The binary variable δ qt =1 means that the quay crane q is in working state at time t, and 3. The multi-stage optimization scheduling method for the seaport logistics-energy coupling system according to claim 1 is characterized in that: The compact form of the optimal logistics-power flow model is Where: x t is a vector representing binary variables, including {X bst ,ω qst },y t is a vector representing continuous variables, including z t is a vector representing integer variables, including {B qt , Q st , δ st }, I represents an integer variable that is independent of time, including ξ t Represents an uncertain variable, indicating the deviation of new energy output.

4. The multi-stage optimization scheduling method for the seaport logistics-energy coupling system according to claim 1 is characterized in that: The step 4 specifically includes: 4.1) Equivalent transformation of objective function, including: And further obtain the following equivalent compact model: Among them: variable x τ (ξ τ-1 ), z τ (ξ τ-1 ) in the uncertain variable ξ τ Determined before revealing, so it is considered as an uncertain variable ξ τ-1 The affine function of y τ (ξ τ ) in the uncertain variable ξ τ Determined after being revealed, so it is considered as an uncertain variable ξ τ Affine function of ; 4.2) Apply linear and bivariate affine strategies to transform the decision variable x τ (ξ τ-1 ), z τ (ξ τ-1 ),y τ (ξ τ ) for further mapping: Where: t =(ξ 0,t ,…,ξ M,T ), G(ξ t )=(G(ξ 0,t ),…,G(ξ M,t )),ξ 0,t =1, G(ξ 0,t )=1 is the constant term in the affine strategy, Y t 、X t , Z t is the new decision variable to be sought in the affine strategy; Applying the above affine strategy, we get the following equivalent model: 4.3) The affine strategy introduces a nonlinear affine function G(ξ t ), for the linearized model, it is necessary to construct the lifting uncertainty set and its convex envelope as shown below: Where: Ξ′ t To improve the uncertainty set, ξ′ t To improve the uncertainty variables, conv(Ξ′ t ) is the convex envelope of the lifting uncertainty set, ε t To construct the parameters of the convex envelope, W t 、V t 、h t are coefficient matrices; Substituting into the above formula, we get: 4.4) To obtain the worst-case scheduling strategy for uncertain variables, the duality theory is used to transform the model into a single-layer mixed integer optimization model: stΛh+ψμ≥0,ΛW+ψU=L,ΛV≤0,ψ≥0,HI≤d, where: Λ and ψ are dual variables, and L is the coefficient matrix.

5. The multi-stage optimization scheduling method for the seaport logistics-energy coupling system according to claim 4 is characterized in that: The dual model is a single-layer mixed integer optimization model, which is solved directly using a commercial solver: the solution process adopts a rolling optimization form, that is, let t be equal to 1,…,T in sequence, and solve the model in each cycle; after solving the model, the current stage x is obtained. t ,z t ,Y t , where x t ,z t is the logistics operation scheduling variable, and the uncertain variable ξ t Before revealing, x t ,z t It can be directly used for the logistics operation scheduling decision at the current stage. t After the disclosure, according to y t (ξ t )=Y t ξ t and Y t The value of can be used to obtain the power dispatching plan y t .

6. A system for implementing the multi-stage optimization scheduling method for the seaport logistics-energy coupling system according to any one of claims 1 to 5, characterized in that: include: Data acquisition module, model calculation module and instruction execution module, among which: the data acquisition module collects parameters of the logistics system and energy system, predicts new energy output, dynamically updates ship arrival time, and inputs the collected data into the model calculation module; The model calculation module uses the collected and predicted data as input to solve the port logistics-energy multi-stage robust dynamic optimization model and output the optimal logistics and energy scheduling control instructions; the instruction execution module issues the scheduling instructions to the port control room to execute the obtained optimal scheduling plan.

Images