Reservoir project dispatching management method

Abstract

The invention relates to a scheduling management method of a reservoir project, which is characterized in that: by generating a water flow mode network and establishing a target function, under the condition that the water consumption of a reservoir group system is limited, the optimal time and space distribution mode of the water quantity and a water head are sought, so that the generating capacity is maximized. According to the method, water energy resources are fully utilized, electric quantity generated by the hydropower station group system can be absorbed by the power system by default, and benefit maximization is achieved.

Description

Technical Field

[0001] The invention belongs to the field of water conservancy projects, and in particular relates to a dispatching and management method for reservoir projects. Background Art

[0002] Currently, reservoir operation and management primarily rely on empirical evidence and simple mathematical models, which fail to fully consider complex flow patterns and long-term benefits. Existing technologies often focus solely on short-term benefits and ignore long-term ones when determining year-end reservoir capacity, resulting in unscientific reservoir operation and a failure to maximize power generation. Furthermore, existing technologies lack systematic analysis of flow patterns and probabilistic calculations, making them unable to accurately predict and respond to various flow conditions. Summary of the Invention

[0003] The present invention aims to provide a reservoir project scheduling and management method. By generating a water flow pattern network and establishing an objective function, the method seeks the optimal temporal and spatial distribution of water volume and hydraulic head, while limiting the water consumption of a reservoir cluster system, to maximize power generation. This method fully utilizes hydropower resources and assumes that all electricity generated by the hydropower station cluster system can be absorbed by the power system, thereby maximizing efficiency. The specific technical solution is as follows: A water flow pattern network is generated: representative patterns under several water inflow categories are extracted from a medium- and long-term water flow sequence. The extracted patterns are iteratively modified to gradually reduce the distance between the patterns and the observed sample series, thereby seeking the pattern with the smallest distance to the observed sample series. During initialization, for each node in the pattern network, a record is randomly selected from historical water flow observations as the initialization node value. The Euclidean distance between each pattern in the pattern network and the randomly selected complete series is measured, and these distances are sorted in descending order, with the sorting sequence number recorded. The specific values ​​of each node in the water flow pattern are modified to gradually reduce the distance between the water flow pattern and the alternative series (i.e., the observed sample series), and the node is moved a certain distance toward the location of the alternative series. Calculate the probability of each flow pattern using the frequency of the flow cluster represented by the pattern. Use the average distance from the pattern network to the original flow cluster as a measure of the algorithm's average error. Establish an objective function: Optimize system power generation within the dispatch period. Calculate the average output per reservoir period and the total number of dispatch periods. Define the constraints: Water balance: The storage at the end of a reservoir period equals the storage at the beginning of the period plus the inflow during the period minus the outflow during the period. Hydraulic connection: The interval inflow of a reservoir equals the sum of the interval inflows of all upstream reservoirs with direct hydraulic connections. Output function: The average hydraulic head during a reservoir period is calculated using the hydropower station output characteristic function to obtain the net hydraulic head during the reservoir period. Reservoir storage: The storage at the end of the reservoir period must be within a given upper and lower limit. Outflow: The outflow during a reservoir period must be within a given upper and lower limit. Power station output: The output during a reservoir period must be within a given upper and lower limit. System load: The power system requires the output provided by the reservoir group to be within a given lower limit. Water level at the end of the scheduling period: For each reservoir, the water level at the end of the scheduling period needs to be within a given range. Beneficial effects: The present invention generates a water flow pattern network and establishes an objective function to seek the optimal time and space distribution of water volume and head under the condition of limiting the water consumption of the reservoir group system, so as to maximize the power generation. This method makes full use of water energy resources and assumes that the electricity generated by the hydropower station group system can be absorbed by the power system to maximize the benefits. Compared with the existing technology, the present invention can carry out reservoir scheduling more scientifically, improve power generation efficiency, and cope with various complex water flow conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0004] Figure 1 The present invention is a flow chart of a method for dispatching and managing a reservoir project. DETAILED DESCRIPTION

[0005] The year-end storage capacity is the boundary constraint of the reservoir operation model within the annual operation period, which limits the water supply within the study year. Since the year-end storage capacity affects both the short-term benefits within the operation period and the long-term benefits outside the operation period, the operation period for determining the year-end storage capacity must be longer than one year in order to directly calculate the short-term and long-term benefits. Therefore, in order to seek the lowest flow scenario, that is, the year-end storage capacity drawdown plan for each year in the continuous dry year group, this patent sets the operation period length to the longest continuous dry year number that has occurred in the study basin in history. See the process. Figure 1 .

[0006] 1. Generate water flow pattern network Medium- and long-term flow sequences are often assumed to be continuous random processes. The flow pattern network is a discretized description of this continuous random process, consisting of several typical flow patterns. , To include the number (Scale of the pattern network) A collection of patterns. Water flow patterns By several nodes Composition, representing the starting period Until the end period The interval inflow of each library in the system in each time period. Including reservoir group system in the model Next, period Inflow state vector: , .in, For the mode Next, period Inland Reservoir The interval inflow, is the total number of reservoirs. In particular, if the flow pattern With another water flow mode In the period Sharing the same node ,but , the water flow pattern network at the node (or ) branches out.

[0007] To generate a flow pattern network, typical representative patterns for several water inflow categories (e.g., high-flow years, normal-flow years, and low-flow years) are extracted from the original flow observation series. Specifically, an initial pattern is randomly generated. The resulting extracted pattern is iteratively altered, gradually reducing the distance from the observed sample series. Ultimately, the pattern with the smallest distance from the observed sample series is sought, and this result is the "center of gravity" of the dataset.

[0008] The main steps to generate a flow pattern network are as follows: (1) Initialization. For each node on the model network, a record is randomly selected from the historical water flow observation record as the initialization node value: ; Where: The historical observation series of natural water flow in each reservoir, with the dimension of [ K × M ], each row represents the monthly inflow sequence of each reservoir in a year, unit ;rand For interval A random number uniformly distributed within; is the total number of years in the annual streamflow history series, M is the total number of reservoirs, The total number of time periods in the scheduling period (such as the number of months).

[0009] For each node on the model network, a record is randomly selected from the historical water flow observation record as the initial value of the node. Specifically, for each node, a record is randomly selected from the historical natural water flow observation series of each reservoir as the initial value. Among them, the dimension of the historical natural water flow observation series of each reservoir is [ K × M ], each row represents the monthly inflow sequence of each reservoir in one year, and the unit is the total number of years in the annual water flow history series. M is the total number of reservoirs, and is the total number of time periods in the dispatching period (e.g., months).

[0010] (2) The distance between the model network and the water flow series. Before the node numerical iteration, Randomly select a complete series The following formula is used to measure the Euclidean distance of each pattern in the pattern network to the series: ; Where: For mode To series The Euclidean distance of Sort in descending order and use an array Record sorting sequence number.

[0011] Before iterating over the node values, randomly select a complete series. The Euclidean distance between each pattern in the pattern network and the series is measured. Specifically, the Euclidean distance between each pattern in the pattern network and the randomly selected complete series is calculated, and these distances are sorted in descending order, recording the sorting sequence.

[0012] (3) Iterative format. By changing the specific values ​​of each node in the water flow pattern, the water flow pattern and the alternative series are gradually reduced. Specifically, move the node a certain distance toward the location of the alternative series using the following formula:

[0013] ; Where: is the step size variable, which increases with the number of iterations The increase from Changes to is an exponential equation that sets the adjustment amplitude of the node value according to the distance between the flow pattern and the flow sequence; As its parameter, it is increased by Increase to . Initial step size Evidence range (0.5~1.0), final step length Value range (0.01~0.05), initial neighborhood radius 2. Final neighborhood radius The value range is (1~2), and the maximum number of iterations is .

[0014] Equation (2.1-3) indicates that the distance a node moves depends on a combination of the step size and the average distance of the node's model from the alternative series. Therefore, when the flow pattern network is far from the original flow sequence, the adjustment amplitude of each node on the model should be increased; conversely, the adjustment amplitude should be reduced. Furthermore, a gradually decreasing step size parameter is used to ensure iterative convergence. Therefore, with increasing iterations, the model network will gradually converge to the geometric center of gravity of the flow sequence. When the number of iterations reaches a given upper limit, the iterations are terminated and the corresponding model network is output.

[0015] By changing the specific values ​​of each node in the flow pattern, the distance between the flow pattern and the candidate series is gradually reduced. Specifically, the node is moved a certain distance in the direction of the candidate series. This distance depends on the step size and the average distance between the node's pattern and the candidate series. As the number of iterations increases, the pattern network gradually converges to the geometric center of gravity of the flow sequence. When the number of iterations reaches a given upper limit, the iterations are terminated and the corresponding pattern network is output.

[0016] (4) Probability of the pattern. Each water flow pattern is a typical representative of a series of observed water flow sequences. The probability of the pattern is the probability of occurrence of the water flow sequence it represents. Therefore, the probability of each water flow pattern is calculated based on the frequency of the water flow series cluster represented by the pattern: ; Where: For mode probability; is the counting function.

[0017] Each flow pattern is a typical representation of a series of observed flow sequences. The probability of a pattern is the probability of the occurrence of the flow sequence it represents. Therefore, the probability of each flow pattern is calculated using the frequency of the flow cluster represented by the pattern. Specifically, the probability of a pattern is equal to the frequency of occurrence of the flow cluster represented by the pattern.

[0018] (5) Average error. The average distance from the model network to the original water flow series is used as an indicator to measure the average error of the algorithm: ; The convergence process line of the average error can be used to set the maximum number of iterations allowed . That is, select is the number of iterations corresponding to two consecutive average errors tending to be stable ,For example The number of iterations when the threshold is less than 0.01.

[0019] The average distance from the model network to the original water flow series is used as an indicator to measure the average error of the algorithm. Specifically, the average distance from the model network to the original water flow series is calculated as an indicator to measure the average error of the algorithm. The convergence process line of the average error can be used to set the maximum number of iterations allowed. That is, the number of iterations corresponding to two consecutive average errors tending to be stable is selected. j , for example, the number of iterations when it is less than the threshold value 0.01.

[0020] Next, the "water-based electricity generation" model maximizes power generation by finding the optimal temporal and spatial distribution of water volume and hydraulic head within the reservoir system, while limiting water consumption. This model, based on the principle of fully utilizing hydropower resources, assumes that all electricity generated by the hydropower station system can be absorbed by the power system. Essentially, it addresses the problem of maximizing benefits, and the model structure is as follows.

[0021] 2. Establish the objective function The optimization goal is to maximize the system power generation during the dispatch period: ; Where: for Library The average output during the period, is the total number of time periods in the scheduling period; For a long period of time.

[0022] The optimization goal is to maximize the system power generation during the scheduling period. Specifically, the goal is to maximize the system power generation during the scheduling period, where N is the average output during the storage period, T is the total number of time periods in the scheduling period; Δt For a long period of time.

[0023] 3. Identify constraints (1) Water balance ; Where: They are Library Reservoir water storage at the beginning and end of the period, unit ; for Library Inbound flow during time period; for Library Outbound flow rate during the period, unit .

[0024] Specifically, for each reservoir and each time period, the water storage at the end of the reservoir period is equal to the water storage at the beginning of the reservoir period plus the inflow of the reservoir period minus the outflow of the reservoir period.

[0025] (2) Hydraulic connection ; Where: For The reservoir is a collection of upstream reservoirs with which it has direct hydraulic connections; for Library and Inflow between reservoirs.

[0026] Specifically, for each reservoir and each time period, the interval inflow of the reservoir is equal to the sum of the interval inflows of all reservoirs in the set of upstream reservoirs that have direct hydraulic connections with the reservoir.

[0027] (3) Output function ; Where: for Library Average head during the period; for Library Net water head during the period; The function is the output characteristic function of the hydropower station.

[0028] Specifically, for each reservoir and each time period, the average head of the reservoir period is calculated through the hydropower station output characteristic function to obtain the net head of the reservoir period.

[0029] (4) Reservoir water storage capacity ; Where: 、 Respectively Kudi The upper and lower limits of reservoir water storage at the end of the time period.

[0030] Specifically, for each reservoir and each time period, the water storage at the end of the reservoir period needs to be within the given upper and lower limits.

[0031] (5) Outbound flow ; Where: 、 Respectively Kudi The upper and lower limits of outbound flow during a time period.

[0032] Specifically, for each reservoir and each time period, the outflow flow during the reservoir period needs to be within the given upper and lower limits.

[0033] (6) Power station output ; Where: For the Kudi The upper and lower limits of output during a time period.

[0034] Specifically, for each reservoir and each time period, the output of the reservoir period needs to be within the given upper and lower limits.

[0035] (7) System load ; Where: The lower limit of the output required by the power system to be provided by the storage group.

[0036] Specifically, the power system requires that the output provided by the storage group must be within a given lower limit.

[0037] (8) Water level at the end of the dispatch period ; Where: For the Water level at the end of the reservoir scheduling period.

[0038] Specifically, for each reservoir, the water level at the end of the dispatch period needs to be within a given range.

[0039] The present invention generates a water flow pattern network and establishes an objective function to seek the optimal time and space distribution of water volume and head under the condition of limiting the water consumption of the reservoir group system, so as to maximize the power generation. This method makes full use of water energy resources and assumes that the electricity generated by the hydropower station group system can be absorbed by the power system to maximize the benefits. Compared with the existing technology, the present invention can carry out reservoir scheduling more scientifically, improve power generation efficiency, and cope with various complex water flow conditions. Although the present invention has disclosed the preferred embodiments as above, it is not intended to limit the present invention. Ordinary technicians in the technical field to which the present invention belongs can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be based on the definition of the previous claims.

Claims

1. A method for dispatching and managing a reservoir project, characterized in that The following steps are involved: Generate a water flow pattern network and extract typical representative patterns under several water inflow categories from medium- and long-term water flow sequences; Establish an objective function, take the maximum system power generation during the dispatch period as the optimization goal, calculate the average output of the storage period and the total number of dispatch periods; Determine the constraints, including water balance, hydraulic connection, output function, reservoir storage, outflow, power station output, system load and water level at the end of the dispatch period.

2. The method according to claim 1, characterized in that When generating a water flow pattern network, each node on the pattern network is initialized, and a record is randomly selected from the historical water flow observation records as the value of the initialized node.

3. The method according to claim 1, characterized in that When generating the water flow pattern network, the Euclidean distance of each pattern in the pattern network to the randomly selected complete series is measured, and these distances are arranged in descending order, and the sorting sequence number is recorded.

4. The method according to claim 1, wherein When generating the water flow pattern network, the distance between the water flow pattern and the alternative series is gradually reduced by changing the specific values ​​of each node in the water flow pattern, and the node is moved a certain distance toward the location of the alternative series.

5. The method according to claim 1, wherein When generating the flow pattern network, the probability of each flow pattern is calculated, and the probability of each flow pattern is calculated based on the frequency of the flow series cluster represented by the pattern.

6. The method according to claim 1, characterized in that When generating the water flow pattern network, the average distance from the pattern network to the original water flow series is used as an indicator to measure the average error of the algorithm.

7. The method according to claim 1, characterized in that When establishing the objective function, the maximum system power generation during the dispatch period is taken as the optimization goal, and the average output during the storage period and the total number of periods during the dispatch period are calculated.

8. The method according to claim 1, characterized in that When determining the constraints, they include water balance, hydraulic connection, output function, reservoir storage, outflow, power station output, system load and water level at the end of the dispatching period.

Images