Knowledge graph-driven line storage space joint optimization scheduling model construction method

Abstract

The invention discloses a knowledge graph-driven line storage space joint optimization scheduling model construction method. The method comprises the following steps: determining an optimal scheduling target and an overall target function of row storage space joint flood control scheduling, and setting constraint conditions; under a knowledge graph theoretical framework, constructing a response relationship between single-class line storage space scheduling and a river flood discharge state; selecting an excess peak water level exceeding a river embankment control water level as a coordination variable, decomposing the overall objective function into system association constraint equations of a river embankment, a reservoir, a flood storage and detention area and a flood drainage pump station, and reducing a multi-class line storage space joint scheduling problem into a problem of optimization scheduling of each single-class line storage space scheduling; and respectively solving by adopting an improved particle swarm algorithm to obtain an optimal scheduling process of each single-class line storage space. The method not only solves the problem that global optimum is difficult to ensure by directly applying a traditional optimization algorithm due to many row storage space optimization targets and a competitive relationship, but also reduces the calculation time of the model.

Description

Technical Field

[0001] The present invention relates to the technical field of joint scheduling of flood control projects, and specifically to a method for constructing a knowledge graph-driven joint optimization scheduling model for row and storage spaces. Background Art

[0002] Storage and storage primarily refers to the flood control engineering system comprised of rivers, reservoirs, and flood storage and detention areas, which function to "store, release, and distribute" floodwaters. Scientifically and rationally utilizing this engineering system to minimize flood losses is the core of flood prevention in a river basin. Currently, research on joint optimization and scheduling technologies for flood control projects primarily focuses on key flood control projects such as reservoir clusters and flood storage and detention areas. Research on the joint optimization and scheduling of complex flood control engineering systems encompassing river embankments, reservoirs, flood storage and detention areas, and drainage pumping stations is relatively limited.

[0003] The joint optimization and scheduling of complex flood control engineering systems faces difficulties such as the huge scale of flood control projects, complex topological structure, random water inflow patterns of main and tributary rivers, numerous flood protection objects with different flood control standards, and prominent contradictions between flood control and benefits.

[0004] In existing technologies, the joint optimization scheduling of watershed storage space consisting of river embankments, reservoirs, flood storage areas, and drainage pumping stations has the following technical problems:

[0005] First, flood evolution simulations are generally carried out using hydrodynamic methods based on traditional mechanism models. However, these methods are computationally time-consuming and cannot meet the needs of real-time flood control and scheduling decisions.

[0006] Second, complex flood control engineering systems have multiple and competing optimization and scheduling objectives, increasing the complexity of problem solving and exponentially increasing computational complexity. This makes it impossible to decouple optimization objectives for river embankments, reservoirs, flood storage areas, and drainage pumping stations.

[0007] Third, for solving the joint optimization scheduling model of complex flood control engineering systems, traditional optimization algorithms are difficult to guarantee global optimality, and the convergence of optimization algorithms needs to be improved. Summary of the Invention

[0008] In response to the shortcomings of the existing technology, the present invention proposes a knowledge graph-driven method for constructing a joint optimization scheduling model for flow and storage space, which realizes the joint optimization scheduling of complex flow and storage space composed of river embankments, reservoirs, flood storage areas, drainage pumping stations, etc., and obtains the scheduling process of various flow and storage spaces. It not only reduces the calculation time of the model, but also solves the problems of multiple optimization objectives and competitive relationships in flow and storage space, and the difficulty in ensuring global optimality by directly applying traditional optimization algorithms.

[0009] To achieve the above objectives, the present invention designs a knowledge graph-driven method for constructing a joint optimization scheduling model for rows, storages, and spaces. The method is particularly characterized in that it includes the following steps:

[0010] S1) Clarify the categories, scope, and scale of the storage space involved in joint flood control scheduling, determine the optimal scheduling objectives and overall objective function of the joint flood control scheduling of the storage space, and set constraints;

[0011] The storage space includes river embankments, reservoirs, flood storage areas, and drainage pumping stations.

[0012] S2) Under the framework of knowledge graph theory, construct the response relationship between single-category storage space scheduling and river flood discharge status, specifically including the response relationship between reservoir scheduling and river flood discharge status, the response relationship between flood storage and detention area scheduling and river flood discharge status, and the response relationship between drainage pump station scheduling and river flood discharge status;

[0013] S3) The excess flood peak water level exceeding the river embankment control water level is selected as the coordination variable, and the overall objective function is decomposed into the system-related constraint equations of the reservoir, the flood storage area, and the drainage pump station. The Lagrange multiplier vector is then introduced to further decompose the system-related constraint equations. Finally, information is transmitted between different levels based on the selected coordination variables and the optimization scheduling objectives to form a three-level hierarchical decomposition coordination model. This achieves the decoupling of the joint optimization scheduling of the river embankment, reservoir, flood storage area, and drainage pump station, thereby reducing the joint scheduling problem of multiple types of storage spaces to the optimization scheduling problem of each single type of storage space.

[0014] S4) using an improved particle swarm optimization algorithm to solve the optimization scheduling problem of each single type of row and storage space, and obtaining the optimization scheduling process of each single type of row and storage space as the joint optimization scheduling solution of the row and storage space;

[0015] Among them, the specific calculation steps of using the improved particle swarm algorithm to solve the optimization scheduling problem of each single type of row storage space are as follows:

[0016] S41) Initialize particles;

[0017] S42) evaluating the fitness value of each particle, and updating the initialized global optimal position, and updating the position of each particle after initialization to the historical optimal position;

[0018] S43) updating the speed and position of each particle;

[0019] S44) In order to avoid the algorithm falling into the local optimum and speed up the particle convergence, a particle local mutation strategy is introduced. The position of the newly generated particle can be determined by the following formula:

[0020] New x i (t+1)=x i(t+1)·(1+η·r3)

[0021] Where x i is the particle position, t is the number of iterative calculation steps, r3 is a random number in the interval [0,1], and η is the adjustment factor, which can be expressed as:

[0022] η=1-e (1-T / (t+1))

[0023] Where T is the total number of iterations of the particle swarm optimization algorithm;

[0024] S45) Calculate the fitness value of each particle. If the updated fitness value of the particle is lower than the fitness value of the historical optimal position, update the current position of the particle to the historical optimal position; if the updated fitness value of the particle is lower than the fitness value corresponding to the global optimal position of all particles, update the global optimal position of the particle;

[0025] S46) Determine whether the iteration stop condition is met.

[0026] Furthermore, in S1), the optimization scheduling objectives include minimizing the use of reservoir flood control capacity and minimizing flood diversion losses in the flood storage and detention area.

[0027] Furthermore, in S1), the constraints include river constraints, reservoir constraints, flood storage area constraints, and drainage pump station constraints.

[0028] Furthermore, in S1), the river embankment constraint includes the river water level constraint expressed by the following formula (1):

[0029]

[0030] Where,

[0031] The natural water level process of the river control section without considering the impact of storage space scheduling,

[0032] is the water level response process of the i-th reservoir at time t to intercept floods and stagger the peak flow in the downstream river control section,

[0033] The water level response process of the downstream river control section when the flood is diverted from the i-th flood storage area at time t.

[0034] The water level response process of the drainage pump station i at time t in the downstream river control section.

[0035] Z A It controls the safe operating water level of the downstream river section.

[0036] Furthermore, in S1), the reservoir constraints include the operating water level constraint expressed by the following formula (2), the discharge capacity constraint expressed by the following formula (3), the outflow amplitude constraint expressed by the following formula (4), and the water balance constraint expressed by the following formula (5),

[0037]

[0038] |q R,i (t)-q R,i (t-1)|≤Δq R,i (4)

[0039] V R,i (t) = V R,i (t-1)+

[0040] [Q R,i (t)+Q R,i (t-1)-q R,i (t)-q R,i (t-1)]Δt / 2 (5)

[0041] Where,

[0042] Z R,i (t) is the operating water level of the reservoir at time t,

[0043] is the minimum allowable operating water level of the reservoir,

[0044] is the maximum allowable operating water level of the reservoir,

[0045] q R,i (t) is the discharge flow of the i-th reservoir at time t,

[0046] is the minimum allowable discharge flow of the i-th reservoir,

[0047] is the maximum allowable discharge of the i-th reservoir at time t,

[0048] Δq R,i is the maximum allowable fluctuation of the reservoir outflow,

[0049] V R,i (t) the storage capacity of the i-th reservoir at time t,

[0050] V R,i (t-1) The storage capacity of the i-th reservoir at time t-1,

[0051] Q R,i (t) is the inflow of reservoir i at time t,

[0052] QR,i (t-1) is the inflow of reservoir i at time t-1,

[0053] q R,i (t-1) is the discharge flow of the i-th reservoir at time t-1,

[0054] Δt is the calculation time interval.

[0055] Furthermore, in S1), the flood storage and detention area constraints include the design diversion flow constraint expressed by the following formula (6), the design flood storage capacity constraint expressed by the following formula (7), and the flood storage and detention area water balance constraint expressed by the following formula (8).

[0056]

[0057] V D,i (t) = V D,i (t-1)+[q D,i (t)+q D,i (t-1)]Δt / 2 (8)

[0058] Where,

[0059] q D,i (t) is the diversion flow of the i-th flood storage area at time t,

[0060] is the designed diversion flow of the flood storage area,

[0061] V D,i (t) is the flood storage capacity of the i-th flood storage area at time t,

[0062] The designed flood storage capacity of the flood storage area,

[0063] V D,i (t-1) is the flood storage capacity of the i-th flood storage area at time t-1,

[0064] q D,i (t-1) is the flood diversion flow of the i-th flood storage area at time t-1,

[0065] Δt is the calculation time interval.

[0066] Furthermore, in S1), the drainage pump station constraint includes the design drainage flow constraint expressed by the following formula (9):

[0067]

[0068] Where,

[0069] q P,i (t) is the drainage flow of the i-th drainage pump station at time t,

[0070] It is the designed drainage flow of the drainage pump station.

[0071] Furthermore, in S2), the construction of the response relationship between the single-type flow storage and air conditioning mode and the river flood discharge state follows the following ideas: flood scenario simulation generation → flood control scheduling sample generation → historical flood control scheme knowledge extraction → historical flood control scheme knowledge base construction → flood control scheduling knowledge base correction.

[0072] Furthermore, in S3), the system-related constraint equation (1) of river embankments, reservoirs, flood storage areas, and drainage pumping stations can be further decomposed into the following equations (10) and (11):

[0073]

[0074] Where,

[0075] The natural water level process of the river control section without considering the impact of storage space scheduling,

[0076] Z A To control the safe operating water level of the downstream river section,

[0077] is the water level response process of the i-th reservoir at time t to intercept floods and stagger the peak flow in the downstream river control section,

[0078] n is the number of reservoirs,

[0079] is the water level response process of the drainage pump station i at time t in the downstream river control section,

[0080] k is the number of drainage pumping stations,

[0081] Z D (t) is the excess peak water level at each river section control station exceeding the dike control water level at time t,

[0082] The water level response process of the downstream river control section when the flood is diverted from the i-th flood storage area at time t.

[0083] m is the number of flood storage areas.

[0084] Furthermore, it also includes step S5), using the hydrodynamic mechanism model to calculate the joint optimization scheduling plan of the row and storage space in step S4), and reviewing whether the calculation result meets the optimization scheduling target in step S1). If it does, the joint optimization scheduling plan of the row and storage space is determined to be the optimal scheduling plan; if it does not, it is necessary to correct the response relationship between the scheduling of each single type of row and storage space and the river flood discharge state constructed in step S2), until the review calculation result of the hydrodynamic mechanism model meets the optimization scheduling target in step S1).

[0085] The advantages of the present invention are:

[0086] 1. Based on the collection of data on hydrology, river networks, river topography, and flood control projects, the proposed method can achieve joint optimal scheduling of complex water storage spaces consisting of river embankments, reservoirs, flood storage areas, and drainage pumping stations. The method can obtain the scheduling processes of various water storage spaces, such as the outflow process of reservoirs, the outflow process of flood storage areas, and the drainage process of drainage pumping stations in different river sections.

[0087] 2. Based on the concept of large-scale system decomposition and coordination, this invention decomposes the overall objective function into system-related constraint equations for river embankments, reservoirs, flood storage areas, and drainage pumping stations. This reduces the complexity of problem solving while ensuring the algorithm's global optimization capability, and decouples the optimization objectives of the river embankments, reservoirs, flood storage areas, and drainage pumping stations. This solves the problem of multiple and competing optimization objectives in the river embankment, reservoirs, flood storage areas, and drainage pumping stations, and the difficulty of directly applying traditional optimization algorithms to achieve global optimization.

[0088] 3. The present invention adopts an improved particle swarm algorithm to solve the optimization problem of the underlying subsystem of each single-class row storage space separately, thereby improving the global convergence of the particle swarm algorithm;

[0089] 4. Based on the theoretical framework of knowledge graph, this invention constructs the response relationship between the spatial scheduling of each type of storage and the flood discharge state of the river, avoiding the problem of long calculation time using traditional hydrodynamic methods;

[0090] The knowledge graph-driven method for constructing a joint optimization scheduling model for row-storage space in the present invention not only solves the problems of multiple row-storage space optimization objectives with competitive relationships and the difficulty in ensuring global optimality by directly applying traditional optimization algorithms, but also reduces the calculation time of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 A roadmap for the construction method of the present invention;

[0092] Figure 2a Schematic diagram of testing the improved particle swarm algorithm through verification function 1 in the embodiment;

[0093] Figure 2bSchematic diagram of testing the improved particle swarm algorithm through verification function 2 in the embodiment;

[0094] Figure 2c Schematic diagram of testing the improved particle swarm algorithm through verification function 3 in the embodiment;

[0095] Figure 3 Optimizing the operation of the Three Gorges Reservoir under the 1954 300-year design flood;

[0096] Figures 4a to 4d They are the optimization scheduling process of the flood storage areas in Jingjiang, Chenglingji, Wuhan and Hukou areas of the Yangtze River Basin under the 300-year design flood in 1954. DETAILED DESCRIPTION

[0097] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0098] This example studies a complex flood control engineering system consisting of 25 control reservoirs in the upper reaches of the Yangtze River and 42 national-level flood storage and detention areas in the middle and lower reaches. Using the super-standard flood in the middle and lower reaches of the Yangtze River—the 1954 300-year design flood—as the water regime input, the optimized flood control scheduling process for the Three Gorges Reservoir and the flood storage and detention areas in Jingjiang, Chenglingji, Wuhan, and Hukou regions is calculated.

[0099] like Figure 1 As shown, the present invention provides a knowledge graph-driven method for constructing a row-storage-space joint optimization scheduling model, including the following steps:

[0100] S1) Clarify the categories, scope, and scale of the storage spaces involved in joint flood control scheduling, determine the optimal scheduling objectives and overall objective function of the joint flood control scheduling of storage spaces, and set constraints.

[0101] The storage space includes river embankments, reservoirs, flood storage areas, and drainage pumping stations.

[0102] For complex watershed storage spaces comprised of river embankments, reservoirs, flood storage areas, and drainage pumping stations, the safe flow of floodwaters in rivers is ensured through the combined flood control and scheduling of reservoirs and flood storage areas, combined with the channel storage function of rivers. In terms of flood control, reservoirs regulate and store their own incoming floodwaters, altering the downstream flow process to effectively reduce peak flows and flood volumes in downstream rivers. River embankments primarily smooth flood waves through channel storage. Flood storage areas serve as temporary storage areas when floods cannot be fully controlled by reservoirs and rivers. Different flood control systems have different roles and significance in flood control and scheduling, and therefore the objectives considered in the combined optimized scheduling of complex flood control systems also differ.

[0103] The main scheduling objectives considered in the joint scheduling of flood storage space include three aspects: 1. Minimizing the use of reservoir flood control capacity (or maximizing the reservoir peak reduction rate); 2. Controlling the river flood flow rate below the river's safe discharge capacity (or keeping the river water level below the guaranteed water level, minimizing the duration of the river water level exceeding the warning level, etc.) to ensure the safety of the river flood flow; and 3. Minimizing the flood diversion losses in the flood storage area. Among them, there is a competitive relationship between objectives 1, 2, and 3. In the actual flood control scheduling decision-making process, to ensure the flood control safety of the flood protection objects, objective 2 has the highest priority. To achieve the decoupling of the river embankment scheduling optimization objectives from reservoirs and flood storage areas, this paper uses objective 2 as a constraint condition and only considers 1 and 3 in the model objective function, resulting in a multi-objective optimization scheduling model.

[0104] In this embodiment, the categories and scopes of the storage spaces participating in the joint flood control operation are as follows:

[0105] ① Reservoirs: Taking into account the construction progress, flood control capacity, control function, operation status and other factors of the reservoir, 25 flood control reservoirs in the upper reaches of the Yangtze River were included in this study, with a total flood control storage capacity of 48.898 billion m 3 The Three Gorges Dam is located in Sanduping Town, Yichang City, Hubei Province, and the reservoir controls a drainage area of approximately 1 million km. 2 , flood control storage capacity 22.15 billion m 3 .

[0106] ② Flood storage and detention areas: According to the overall layout of flood control in the Yangtze River Basin, a total of 42 national-level flood storage and detention areas have been arranged in the Jingjiang area of the middle and lower reaches of the Yangtze River, the area near Chenglingji, the area near Wuhan, and the area near Hukou. Among them, there are 4 in the Jingjiang area with a flood storage capacity of 6.9 billion m 3 ; 27 areas near Chenglingji, with a flood storage capacity of 33.823 billion m 3 ; 6 areas near Wuhan, with a flood storage capacity of 12.527 billion m 3 ; 5 areas near the lake mouth, with a flood storage capacity of 5.021 billion m 3 .

[0107] The optimization scheduling objectives of the joint flood control scheduling of the storage space are to maximize the peak reduction rate of the upstream reservoir group and minimize the flood diversion losses in the flood storage and detention area.

[0108] To ensure that the model calculation results are consistent with the actual operation of river embankments, reservoirs, and flood storage areas, corresponding constraints need to be set. These constraints include river constraints, reservoir constraints, flood storage area constraints, and drainage pump station constraints.

[0109] Specifically, the river embankment constraint includes the river water level constraint expressed by the following formula (1):

[0110]

[0111] Where,

[0112] The natural water level process of the river control section without considering the impact of storage space scheduling,

[0113] is the water level response process of the i-th reservoir at time t to intercept floods and stagger the peak flow in the downstream river control section,

[0114] The water level response process of the downstream river control section when the flood is diverted from the i-th flood storage area at time t.

[0115] The water level response process of the drainage pump station i at time t in the downstream river control section.

[0116] Z A It controls the safe operating water level of the downstream river section.

[0117] Specifically, the reservoir constraints include the minimum and maximum operating water level constraints expressed by the following formula (2), the minimum discharge and discharge capacity constraints expressed by the following formula (3), the outflow amplitude constraint expressed by the following formula (4), and the water balance constraint expressed by the following formula (5).

[0118]

[0119] |q R,i (t)-q R,i (t-1)|≤Δq R,i (4)

[0120] V R,i (t) = V R,i (t-1)+

[0121] [Q R,i (t)+Q R,i (t-1)-q R,i (t)-q R,i (t-1)]Δt / 2 (5)

[0122] Where,

[0123] Z R,i (t) is the operating water level of the reservoir at time t,

[0124] The lowest allowable operating water level of the reservoir, usually taken as the flood limit water level of the reservoir,

[0125] The highest permissible operating water level of the reservoir, usually taken as the highest flood control level of the reservoir.

[0126] qR,i (t) is the discharge flow of the i-th reservoir at time t,

[0127] is the minimum allowable discharge flow of the i-th reservoir, which is generally taken as the discharge flow of the reservoir's ecological, power generation, navigation and other functions.

[0128] is the maximum allowable discharge of the i-th reservoir at time t,

[0129] Δq R,i is the maximum allowable fluctuation of the reservoir outflow,

[0130] V R,i (t) the storage capacity of the i-th reservoir at time t,

[0131] V R,i (t-1) The storage capacity of the i-th reservoir at time t-1,

[0132] Q R,i (t) is the inflow of reservoir i at time t,

[0133] Q R,i (t-1) is the inflow of reservoir i at time t-1,

[0134] q R,i (t-1) is the discharge flow of the i-th reservoir at time t-1,

[0135] Δt is the calculation time interval.

[0136] For flood storage areas equipped with diversion gates, the diversion process can be controlled, and the constraints mainly include design flood flow, flood storage capacity and water balance constraints.

[0137] Specifically, the flood storage and detention area constraints include the design diversion flow constraint expressed by the following formula (6), the design flood storage capacity constraint expressed by the following formula (7), and the flood storage and detention area water balance constraint expressed by the following formula (8).

[0138]

[0139] V D,i (t) = V D,i (t-1)+[q D,i (t)+q D,i (t-1)]Δt / 2 (8)

[0140] Where,

[0141] q D,i (t) is the diversion flow of the i-th flood storage area at time t,

[0142] is the designed diversion flow of the flood storage area,

[0143] V D,i (t) is the flood storage capacity of the i-th flood storage area at time t,

[0144] The designed flood storage capacity of the flood storage area,

[0145] V D,i (t-1) is the flood storage capacity of the i-th flood storage area at time t-1,

[0146] q D,i (t-1) is the flood diversion flow of the i-th flood storage area at time t-1,

[0147] Δt is the calculation time interval.

[0148] Specifically, the drainage pump station constraint includes the design drainage flow constraint expressed by the following formula (9):

[0149]

[0150] Where,

[0151] q P,i (t) is the drainage flow of the i-th drainage pump station at time t,

[0152] It is the designed drainage flow of the drainage pump station.

[0153] In this embodiment, the constraints for the joint flood control scheduling of the flow and storage space can be given according to the "2024 Yangtze River Basin Water Project Joint Scheduling and Utilization Plan" and will not be described in detail here.

[0154] S2) Under the theoretical framework of knowledge graph, the response relationship between single-type storage space scheduling and river flood discharge status is constructed, specifically including the response relationship between reservoir scheduling and river flood discharge status, the response relationship between flood storage and detention area scheduling and river flood discharge status, and the response relationship between drainage pump station scheduling and river flood discharge status.

[0155] The present invention adopts the response relationship between the single-type air storage and flood discharge state of the river to quickly obtain the change of the hydrodynamic state of the downstream river, that is, to calculate the above formula (1) and Z P,i (t).

[0156] Specifically, the construction of the response relationship between a single type of flow and storage air conditioning and the river flood discharge state follows the following ideas: flood scenario simulation generation → flood control scheduling sample generation → historical flood control scheme knowledge extraction → historical flood control scheme knowledge base construction → flood control scheduling knowledge base correction.

[0157] The detailed process is as follows:

[0158] Flood scenario simulation generation: Sampling historical typical design flood method or multi-station flood random simulation method randomly simulates flood processes composed of different frequencies and different regions at multiple stations.

[0159] Flood control scheduling sample generation: Based on different flood processes and in combination with existing scheduling rules, different reservoir scheduling plans are formulated, and different inflow background fields and model initial boundary fields are comprehensively generated. Basin flood control scheduling simulations are conducted using reservoir optimization scheduling models and river flood evolution models. The water level and flow changes at different control stations under different flood processes and reservoir operation conditions are calculated to obtain a large number of scheduling samples.

[0160] Knowledge extraction of historical flood control plans: Analyze historical scheduling plans, extract the scheduling objectives, water conditions, engineering conditions, reservoir activation combinations, reservoir scheduling methods, corresponding scheduling effects, and other factors that need to be considered in the scheduling process, including the occurrence time, impact space, quantity involved, and impact degree, to form basic knowledge points of flood control.

[0161] Construction of a knowledge base of historical flood control schemes: The extracted knowledge element points are connected in series, and the obtained scheduling relationship models of different reservoirs to different control sites are used as relationship attributes to link the reservoir nodes and control station nodes to build a knowledge base of historical flood control scheme scheduling cases.

[0162] Correction of the flood control and operation knowledge base: Historical operation cases are replayed, and the operation-response relationship is repeatedly trained based on operation errors. For example, an error autoregressive model is directly established on the model output to correct the results. When errors are significant, historical cases need to be reconstructed and analyzed, and relevant factor features are re-screened. At the same time, the reservoir operation plan and operation results are verified for errors.

[0163] S3) The excess flood peak water level exceeding the control water level of the river embankment is selected as the coordination variable, and the overall objective function is decomposed into the system-related constraint equations of reservoirs, flood storage areas, and drainage pumping stations. The Lagrange multiplier vector is then introduced to further decompose the system-related constraint equations. Finally, information is transmitted between different levels based on the selected coordination variables and the optimization scheduling objectives to form a three-level hierarchical decomposition coordination model. This realizes the decoupling of the joint optimization scheduling of river embankments, reservoirs, flood storage areas, and drainage pumping stations, thereby reducing the joint scheduling problem of multiple types of storage spaces to the optimization scheduling problem of each single type of storage space.

[0164] Specifically, the system-related constraint equation (1) of river embankments, reservoirs, flood storage areas, and drainage pumping stations can be further decomposed into the following equations (10) and (11):

[0165]

[0166] Where,

[0167] The natural water level process of the river control section without considering the impact of storage space scheduling,

[0168] Z A To control the safe operating water level of the downstream river section,

[0169] is the water level response process of the i-th reservoir at time t to intercept floods and stagger the peak flow in the downstream river control section,

[0170] n is the number of reservoirs,

[0171] is the water level response process of the drainage pump station i at time t in the downstream river control section,

[0172] k is the number of drainage pumping stations,

[0173] Z D (t) is the excess peak water level at each river section control station exceeding the dike control water level at time t,

[0174] The water level response process of the downstream river control section when the flood is diverted from the i-th flood storage area at time t.

[0175] m is the number of flood storage areas.

[0176] Specifically, the target coordination method of the large system decomposition coordination method is applied, and the Lagrange multiplier vector λ is introduced R (t) and λ D (t) Continue to decompose the above constraints separately. After decomposing and coordinating the large system, the reservoir, flood storage area, and drainage pump station subsystems can be optimized and solved separately.

[0177] S4) An improved particle swarm algorithm is used to solve the optimization scheduling problem of each single type of row and storage space respectively, and the optimization scheduling process of each single type of row and storage space is obtained as the joint optimization scheduling solution of the row and storage space.

[0178] Among them, the specific calculation steps of using the improved particle swarm algorithm to solve the optimization scheduling problem of each single type of row storage space are as follows:

[0179] (1) An improved particle swarm algorithm is used to solve the single-class row-storage space joint optimization scheduling problem. To adapt to the solution paradigm of the particle swarm algorithm, the single-class row-storage space joint optimization scheduling problem needs to be encoded first. In practical applications, different variables to be solved can be selected as the particle positions to be optimized in the particle swarm algorithm according to the engineering characteristics.

[0180] 1) For reservoirs, the outflow process or the reservoir water level process can be selected as the particle location. In practical applications, if the reservoir capacity is large, a slight change in the reservoir water level will significantly change the outflow flow, so it is recommended to select the outflow process as the particle location. If the total number of reservoirs involved in the problem is D, the particle location can be encoded according to the following formula (12):

[0181]

[0182] Where x i is the particle position vector, is the discharge of reservoir A in the i-th particle at time t, and so on.

[0183] 2) For flood storage and detention areas, either the flood diversion flow process or the flood storage level process can be selected as the particle location. Similar to reservoirs, if the flood storage area has a large storage capacity, even slight changes in the flood storage level will significantly change the diversion flow. Therefore, it is recommended to use the diversion flow process as the particle location. The encoding method is the same as for reservoirs and will not be detailed here.

[0184] 3) For drainage pumping stations, the only available resource for basin flood control scheduling is the pumping station's discharge limit. In actual applications, considering the large number of drainage pumping stations, it is neither possible nor necessary to fine-tune the discharge limit of each drainage pumping station in actual scheduling decisions. Therefore, a river section can be processed and a unified discharge limit is assigned to all drainage pumping stations in this river section. In summary, if the total number of river sections involved in the problem is G, the particle position can be encoded according to the following formula (13):

[0185]

[0186] Where, is the limited discharge level of river section A in the i-th particle, and so on.

[0187] (2) An improved particle swarm optimization algorithm is used to solve the joint scheduling problem of each type of row storage space. The specific calculation process is as follows:

[0188] 1) Initialize particles, including setting the group size, the maximum number of iterations of particles, and the maximum allowed speed of particles Assigns random velocities to the initial particles.

[0189] 2) Calculate the fitness value of each particle according to the set objective function i ), evaluate the fitness value of each particle, and update the global optimal position after initialization, and update the position of each particle after initialization to the historical optimal position.

[0190] 3) Update the velocity v of each particle according to the following formula i (t) and position xi (t).

[0191] v i (t+1)=ωv i (t)+c1r1[pbest(t)-x i (t)]+c2r2[gbest(t)-x i (t)] (14)

[0192] x i (t+1)=x i (t)+v i (t+1) (15)

[0193] Where t is the number of iterative calculation steps, c1 and c2 are learning factor constants, r1 and r2 are random numbers in the interval [0,1], and ω is the inertia weight, which usually changes linearly with the number of iterative calculation steps as a variable.

[0194] 4) In order to avoid the algorithm from falling into local optimum and speed up the particle convergence, a particle local mutation strategy is introduced. The position of the newly generated particle can be determined using the following formula (16):

[0195] New x i (t+1)=x i (t+1)·(1+η·r3) (16)

[0196] Where x i is the particle position, t is the number of iterative calculation steps, r3 is a random number in the interval [0,1], and η is the adjustment factor, which can be expressed as:

[0197] η=1-e (1-T / (t+1)) (17)

[0198] Where T is the total number of iterations of the particle swarm optimization algorithm.

[0199] The above expression can ensure that η is large in the early stage of the algorithm, so that particles can escape the local area and continue searching in a larger range. In the later stage of the algorithm, η is small, which reduces the uncertainty of random numbers on particle convergence and retains the fast convergence characteristic of the particle swarm algorithm.

[0200] 5) Calculate the fitness value of each particle fitness (x i ), if the updated particle fitness value is lower than the fitness value of the historical optimal position, the current position of the particle is updated to the historical optimal position; if the updated particle fitness value is lower than the fitness value corresponding to the global optimal position of all particles, the global optimal position of the particle is updated.

[0201] 6) Determine whether the iteration stop condition is met.

[0202] In this paper, to verify the convergence of the improved particle swarm algorithm, three classic examples are used for verification. For each function, we use two algorithms, the improved particle swarm algorithm and the standard particle swarm algorithm, for optimization iteration. First, the necessary parameters are set: the initial population size is 100, the maximum number of iterations of the particle swarm algorithm is 1000, c1 = c2 = 2.0 in formula (14), and ω = 0.9 - t (0.9 - 0.4) / T.

[0203] In the iterative optimization process, if the particle position fitness (x i ) and the theoretical optimal solution f0 of the verification example is less than a certain small amount (|fitness(x i )-f0|≤∈), the iterative optimization is stopped and the total number of iterations is counted to reflect the speed at which the algorithm converges to the global optimal solution.

[0204] (1) Verification Example 1

[0205] Verification function 1 takes the following form:

[0206]

[0207] This function has many local maximum points, and the extreme position is (0,0). It obtains the maximum value near (0,0). Its geometric characteristics are as follows: Figure 2a shown.

[0208] (2) Verification Example 2

[0209] Verification function 2 takes the following form:

[0210]

[0211] This function is the famous Griewank function, whose global extreme point is surrounded by local extreme points, but this function has only one global extreme point; the global minimum point of this function is at x i =0(i=1,2,…,D), the minimum value of the function is 0, and it is difficult to find the global extreme point of the function. Its geometric characteristics are as follows Figure 2b shown.

[0212] (3) Verification Example 3

[0213] Verification function 3 takes the following form:

[0214]

[0215] This function is the famous Schaffer function. Its global maximum is (0,0). There are countless local maximums within a range of about π from the global maximum. Its geometric characteristics are as follows: Figure 2c shown.

[0216] For the above three test functions, the optimization calculation results of the improved particle swarm algorithm and the standard particle swarm algorithm are shown in Table 1 below.

[0217] Table 1 Comparison of optimization results between standard and improved particle swarm algorithm

[0218]

[0219] Table 1 shows that both the standard PSO and the improved PSO can find the global optimal solution for verification function 3. In fact, the standard PSO's initial convergence speed is faster than the improved PSO. However, for complex multimodal functions, the improved PSO achieves an optimal solution that is superior to the standard PSO. For verification functions 1 and 2, both algorithms reach the global optimal point, but the improved PSO converges significantly faster than the standard PSO, demonstrating the superiority of the improved algorithm in terms of convergence speed and global traversal capability.

[0220] In this embodiment, the optimization scheduling process of the Three Gorges Reservoir is detailed in Figure 3 The optimization scheduling process of flood storage and detention areas in Jingjiang, Chenglingji, Wuhan and Hukou areas is shown in Figure 2. Figures 4a to 4d .

[0221] Preferably, the knowledge graph-driven method for constructing a joint optimization scheduling model for row and storage space of the present invention also includes step S5), using a hydrodynamic mechanism model to calculate the joint optimization scheduling scheme for row and storage space in step S4), and reviewing whether the calculation result meets the optimization scheduling target in step S1); if so, the joint optimization scheduling scheme for row and storage space is determined to be the optimal scheduling scheme; if not, it is necessary to correct the response relationship between the scheduling of each single type of row and storage space and the flood discharge state of the river constructed in step S2), until the review calculation result of the hydrodynamic mechanism model meets the optimization scheduling target in step S1).

[0222] Specifically, in S5), the specific process is as follows:

[0223] S51), taking the optimized scheduling process of each single type of storage space obtained by optimization as input, a one-dimensional or two-dimensional coupled hydrodynamic model is used for calculation;

[0224] S52) The calculation result is brought into the overall objective function in step S1) to determine whether the inequality condition in the overall objective function holds. If not, the response relationship between the scheduling of each type of storage space and the river flood discharge state constructed in step S2) needs to be corrected, and steps S2) to S5) are repeated until the calculation result is verified to meet the optimization scheduling target in step S1).

[0225] In this embodiment, after calculation using the hydrodynamic model, the results show that each control station can meet the scheduling goal of ensuring that the maximum water level does not exceed the guaranteed water level.

[0226] The knowledge graph-driven method for constructing a joint optimization scheduling model for row-storage space in the present invention not only solves the problems of multiple row-storage space optimization objectives with competitive relationships and the difficulty in ensuring global optimality by directly applying traditional optimization algorithms, but also reduces the calculation time of the model.

[0227] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. A knowledge graph-driven method for constructing a joint optimization scheduling model for rows and storage spaces, characterized by: The steps include: S1) Clarify the categories, scope, and scale of the storage space involved in joint flood control scheduling, determine the optimal scheduling objectives and overall objective function of the joint flood control scheduling of the storage space, and set constraints; The storage space includes river embankments, reservoirs, flood storage areas, and drainage pumping stations; S2) Under the framework of knowledge graph theory, construct the response relationship between single-category storage space scheduling and river flood discharge status, specifically including the response relationship between reservoir scheduling and river flood discharge status, the response relationship between flood storage and detention area scheduling and river flood discharge status, and the response relationship between drainage pump station scheduling and river flood discharge status; S3) The excess flood peak water level exceeding the river embankment control water level is selected as the coordination variable, and the overall objective function is decomposed into the system-related constraint equations of the reservoir, the flood storage area, and the drainage pump station. The Lagrange multiplier vector is then introduced to further decompose the system-related constraint equations. Finally, information is transmitted between different levels based on the selected coordination variables and the optimization scheduling objectives to form a three-level hierarchical decomposition coordination model. This achieves the decoupling of the joint optimization scheduling of the river embankment, reservoir, flood storage area, and drainage pump station, thereby reducing the joint scheduling problem of multiple types of storage spaces to the optimization scheduling problem of each single type of storage space. S4) using an improved particle swarm optimization algorithm to solve the optimization scheduling problem of each single type of row and storage space, and obtaining the optimization scheduling process of each single type of row and storage space as the joint optimization scheduling solution of the row and storage space; Among them, the specific calculation steps of using the improved particle swarm algorithm to solve the optimization scheduling problem of each single type of row storage space are as follows: S41) Initialize particles; S42) evaluating the fitness value of each particle, and updating the initialized global optimal position, and updating the position of each particle after initialization to the historical optimal position; S43) updating the speed and position of each particle; S44) In order to avoid the algorithm falling into the local optimum and speed up the particle convergence, a particle local mutation strategy is introduced. The position of the newly generated particle can be determined by the following formula: newx i (t+1)=x i (t+1)·(1+η·r3) Where x i is the particle position, t is the number of iterative calculation steps, r3 is a random number in the interval [0,1], and η is the adjustment factor, which can be expressed as: n=1-e (1-T / (t+1)) Where T is the total number of iterations of the particle swarm optimization algorithm; S45) Calculate the fitness value of each particle. If the updated fitness value of the particle is lower than the fitness value of the historical optimal position, update the current position of the particle to the historical optimal position; if the updated fitness value of the particle is lower than the fitness value corresponding to the global optimal position of all particles, update the global optimal position of the particle; S46) Determine whether the iteration stop condition is met.

2. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 1 is characterized by: In S1), the optimization scheduling objectives include minimizing the use of reservoir flood control capacity and minimizing flood diversion losses in the flood storage and detention area.

3. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 2 is characterized by: In S1), the constraints include river constraints, reservoir constraints, flood storage area constraints, and drainage pump station constraints.

4. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 3 is characterized by: In S1), the river embankment constraint includes the river water level constraint expressed by the following formula (1): Where, The natural water level process of the river control section without considering the impact of storage space scheduling, is the water level response process of the i-th reservoir at time t to intercept floods and stagger the peak flow in the downstream river control section, The water level response process of the downstream river control section when the flood is diverted from the i-th flood storage area at time t. The water level response process of the drainage pump station i at time t in the downstream river control section. Z A It controls the safe operating water level of the downstream river section.

5. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 4 is characterized by: In S1), the reservoir constraints include the operating water level constraint expressed by the following formula (2), the discharge capacity constraint expressed by the following formula (3), the outflow amplitude constraint expressed by the following formula (4), and the water balance constraint expressed by the following formula (5). |q R,i (t)-q R,i (t-1)|≤Δq R,i (4) V R,i (t)=V R,i (t-1)+ [Q R,i (t)+Q R,i (t-1)-q R,i (t)-q R,i (t-1)]Δt / 2 (5) Where, Z R,i (t) is the operating water level of the reservoir at time t, is the minimum allowable operating water level of the reservoir, is the maximum allowable operating water level of the reservoir, q R,i (t) is the discharge flow of the i-th reservoir at time t, is the minimum allowable discharge flow of the i-th reservoir, is the maximum allowable discharge of the i-th reservoir at time t, Δq R,i is the maximum allowable fluctuation of the reservoir outflow, V R,i (t) the storage capacity of the i-th reservoir at time t, V R,i (t-1) The storage capacity of the i-th reservoir at time t-1, Q R,i (t) is the inflow of reservoir i at time t, Q R,i (t-1) is the inflow of reservoir i at time t-1, q R,i (t-1) is the discharge flow of the i-th reservoir at time t-1, Δt is the calculation time interval.

6. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 5, characterized in that: In S1), the flood storage and detention area constraints include the design diversion flow constraint expressed by the following formula (6), the design flood storage capacity constraint expressed by the following formula (7), and the flood storage and detention area water balance constraint expressed by the following formula (8). V D,i (t)=V D,i (t-1)+[q D,i (t)+q D,i (t-1)]Δt / 2 (8) Where, q D,i (t) is the diversion flow of the i-th flood storage area at time t, is the designed diversion flow of the flood storage area, V D,i (t) is the flood storage capacity of the i-th flood storage area at time t, The designed flood storage capacity of the flood storage area, V D,i (t-1) is the flood storage capacity of the i-th flood storage area at time t-1, q D,i (t-1) is the flood diversion flow of the i-th flood storage area at time t-1, Δt is the calculation time interval.

7. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 6, characterized in that: In S1), the drainage pump station constraint includes the design drainage flow constraint expressed by the following formula (9): Where, q P,i (t) is the drainage flow of the i-th drainage pump station at time t, It is the designed drainage flow of the drainage pump station.

8. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 1, characterized in that: In S2), the construction of the response relationship between the single-type flow storage and air conditioning mode and the river flood discharge state follows the following ideas: flood scenario simulation generation → flood control scheduling sample generation → historical flood control scheme knowledge extraction → historical flood control scheme knowledge base construction → flood control scheduling knowledge base correction.

9. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 1, characterized in that: In S3), the system-related constraint equation (1) of river embankments, reservoirs, flood storage areas and drainage pumping stations can be further decomposed into the following equations (10) and (11): Where, The natural water level process of the river control section without considering the impact of storage space scheduling, Z A To control the safe operating water level of the downstream river section, is the water level response process of the i-th reservoir at time t to intercept floods and stagger the peak flow in the downstream river control section, n is the number of reservoirs, is the water level response process of the drainage pump station i at time t in the downstream river control section, k is the number of drainage pumping stations, Z D (t) is the excess peak water level at each river section control station exceeding the dike control water level at time t, The water level response process of the downstream river control section when the flood is diverted from the i-th flood storage area at time t. m is the number of flood storage areas.

10. The method for constructing a knowledge graph-driven row-storage-space joint optimization scheduling model according to claim 1, characterized in that: It also includes step S5), using the hydrodynamic mechanism model to calculate the joint optimization scheduling plan of the flow and storage space in step S4), and reviewing whether the calculation result meets the optimization scheduling target in step S1). If it does, the joint optimization scheduling plan of the flow and storage space is determined to be the optimal scheduling plan; if it does not, it is necessary to correct the response relationship between the scheduling of each single type of flow and storage space and the river flood discharge state constructed in step S2), until the review calculation result of the hydrodynamic mechanism model meets the optimization scheduling target in step S1).

Images