Method, system, equipment and medium for selecting site for acoustic rainfall enhancement operation point for reservoir storage

Abstract

The invention relates to a method, system, equipment and medium for selecting a site for an acoustic rain enhancement operation point for increasing water storage in a reservoir. The method comprises the following steps: constructing the site selection problem for an acoustic rain enhancement operation point for increasing water storage in a reservoir as a multi-objective combination problem; obtaining basic influencing factors of the multi-objective combination problem based on the decision space of the multi-objective combination problem; constructing an objective function of the multi-objective combination problem, and converting the multi-objective combination problem into a Markov decision process using a multi-tuple based on each basic influencing factor and historical acoustic wave operation data of the reservoir increase; applying a multi-objective reinforcement learning algorithm to the Markov decision process, taking a control system of an acoustic rain enhancement device as an intelligent agent and a target reservoir range as an environment, obtaining the Pareto frontier of the intelligent agent in different states by using the multi-objective reinforcement learning algorithm, and outputting the address of the acoustic wave operation point of the target reservoir based on the Pareto frontier.

Description

Technical Field

[0001] The invention relates to a method, system, equipment and medium for selecting a site for an acoustic wave rainfall enhancement operation point for increasing water storage in a reservoir, and belongs to the technical field of acoustic wave rainfall enhancement operations. Background Art

[0002] Traditionally, reservoir storage is to increase the water level in the reservoir through artificial means, such as opening the gate to release water, closing the gate, etc., to increase the water storage capacity of the reservoir and ensure that the reservoir has sufficient reserve water in the dry season or during the period of water shortage. In recent years, the method of increasing reservoir storage through acoustic rainfall technology has gradually emerged.

[0003] Acoustic rain enhancement is a method of artificially influencing the weather by using acoustic wave technology to promote precipitation from clouds. Site selection is a very important link in acoustic rain enhancement. The rationality of site selection is directly related to the effect and safety of acoustic rain enhancement. Site selection should take into account factors such as the power, coverage, local meteorological conditions, and environmental protection of acoustic rain enhancement equipment.

[0004] Most of the existing facility site selection schemes are aimed at fixed facilities. For example, the Chinese invention patent with publication number CN114049031A discloses a method for optimizing the site selection of rainwater storage tanks. It uses the construction of an index system, SWMM model and hierarchical analysis method to determine the nodes with high risk of urban flooding, and determines the address of the rainwater storage tank with the corresponding nodes. In the above patent, urban construction and structure will not change for a long time, but once the urban structure changes, the nodes with high risk of urban flooding will also change accordingly, and the address of the storage tank set in the past will not be the optimal choice. The acoustic rain enhancement operation for reservoir storage is a process that requires frequent adjustments. In the short term, the cloud layer and air conditions will change greatly, and the preferred site selection of the acoustic rain enhancement operation will change accordingly. Therefore, the traditional facility site selection scheme is not applicable to the site selection problem of acoustic rain enhancement operations.

[0005] How to comprehensively consider factors such as reservoir storage, meteorological conditions, ecological environment, etc. and choose a suitable location for operations to achieve the best reservoir storage effect is an issue that urgently needs to be resolved. Summary of the invention

[0006] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a method, system, equipment and medium for site selection of acoustic wave rainfall enhancement operation points for reservoir storage.

[0007] The technical solution of the present invention is as follows:

[0008] In one aspect, the present invention provides a method for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir, comprising the following steps:

[0009] The problem of site selection for acoustic rainfall enhancement operation points for reservoir storage is constructed as a multi-objective combination problem.

[0010] Obtain the basic influencing factors of multi-objective combination problems based on their decision space;

[0011] The objective function of the multi-objective combination problem is constructed, and based on the basic influencing factors and the historical acoustic operation data of reservoir storage, the multi-objective combination problem is converted into a Markov decision process using multivariate groups;

[0012] A multi-objective reinforcement learning algorithm is applied to the above-mentioned Markov decision process. The control system of the acoustic rainmaking device is taken as the intelligent agent and the target reservoir range is taken as the environment. The multi-objective reinforcement learning algorithm is used to obtain the Pareto frontier of the intelligent agent in different states, and the address of the acoustic operation point of the target reservoir is output based on the Pareto frontier.

[0013] As a preferred implementation, the multi-objective combination problem includes optimization of reservoir capacity, optimization of rainfall enhancement effect and optimization of environmental impact.

[0014] As a preferred implementation, the objective function of constructing the multi-objective combination problem specifically includes:

[0015] Optimization objective function of reservoir capacity, optimization objective function of rainfall enhancement effect and optimization objective function of environmental impact;

[0016] The reservoir capacity optimization objective function is specifically:

[0017]

[0018] in, For the formulation The planned reservoir water storage increase during the period, for The estimated precipitation in the reservoir area during the period of time for the acoustic rainfall enhancement operation at the corresponding site, for The natural increase in water storage in the reservoir during the period, for The amount of water evaporation from the reservoir during the period;

[0019] The optimization objective function of the rainfall enhancement effect is specifically:

[0020]

[0021] in, for The number of predicted changes in rainfall type caused by acoustic rainfall enhancement operations at the corresponding site within the time period; For the The weight corresponding to the rainfall type, ∈0~1; For the The predicted duration of rainfall for each rainfall type;

[0022] The environmental impact optimization objective function is specifically:

[0023]

[0024] in, is the geomorphic coefficient of the corresponding site. The more convenient the geomorphic transportation of the corresponding site is, the smaller the geomorphic coefficient is; is the altitude coefficient of the corresponding site, and the altitude coefficient decreases with the altitude; is the acoustic wave impact coefficient of the corresponding site. The fewer the number of households in the preset range of the corresponding site, the smaller the acoustic wave impact coefficient; is the reservoir area, It is the overlapping area between the rainfall area and the reservoir area caused by the acoustic rainfall enhancement operation at the corresponding site.

[0025] As a preferred implementation, the step of converting the multi-objective combination problem into a Markov decision process using a multi-tuple is specifically as follows:

[0026] Using historical acoustic operation data of reservoir storage, using multivariate Construct a Markov decision process for selecting the site for acoustic rainfall enhancement operations to increase reservoir storage;

[0027] in, is the state space, , for At this moment, from the reservoir Status data obtained from each location, ,in for At this moment, from the reservoir range The first location obtained Observed values ​​of basic influencing factors; is the action space, The action of the agent at each moment ,action correspond The movement of the acoustic rainfall enhancement operation point at all times; is the state transfer matrix; is a reward function, which is specifically the objective function of the multi-objective combination problem.

[0028] As a preferred implementation, the steps of using a multi-objective reinforcement learning algorithm to obtain the Pareto frontier of the agent in different states are specifically as follows:

[0029] Establish the function for finding the maximum reward as shown below:

[0030]

[0031] in, is the cumulative reward of the Mth objective in the objective function of the multi-objective combination problem corresponding to the reward function;

[0032]

[0033] in, Considering the state data at time t and actions The cumulative reward calculation function of According to the state data at time t and actions The joint probability distribution of To find the optimal termination time, is the reward coefficient of the Mth target, Indicates the importance of the Mth target at time t, ∈0~1, and , Indicates The importance of a goal; According to the state data at time t and actions The calculated reward value of the Mth target;

[0034] Initialize the multi-objective reinforcement learning algorithm, iteratively perform the multi-objective reinforcement learning task through the historical acoustic operation data of reservoir storage, and put the obtained Pareto points into the Pareto set;

[0035] The Pareto set is standardized, and the standardized Pareto points in the standardized Pareto set are sorted to obtain the serialized optimal point set;

[0036] At each iteration, the reward coefficient of the Mth target at the next moment is determined based on the serialized optimal point set , according to the reward coefficient of the Mth target at the next moment Perform the reinforcement learning task at the next moment, obtain a new Pareto set, and reconstruct the Pareto frontier based on the new Pareto set.

[0037] As a preferred implementation, the Pareto set is standardized, and the standardized Pareto points in the standardized Pareto set are sorted to obtain a serialized optimal point set; in each iteration, the reward coefficient of the Mth target at the next moment is determined based on the serialized optimal point set. The specific steps are:

[0038] According to the Pareto set, the minimum value set and maximum value set in the Pareto point corresponding to each target are obtained, where the minimum value set is:

[0039]

[0040] in, Indicates The first Pareto optimal point; Indicates The minimum value among the Pareto optimal points of the objectives;

[0041] The maximum value set is:

[0042]

[0043] in, Indicates The maximum value among the Pareto optimal points of the objectives;

[0044] Standardize each Pareto optimal point in the Pareto set:

[0045]

[0046] in, is the Pareto optimal point after standardization;

[0047] The standardized Pareto optimal points are sorted by target to obtain the serialized optimal point set for each target:

[0048]

[0049] in, Sort the jth point in the sequenced optimal point set for the Mth goal, is the total number of sorts;

[0050] According to the serialized optimal point set of each target, the reward coefficient of the Mth target at the next moment is calculated by the following formula:

[0051]

[0052] in, represents the reward coefficient of the Mth target at the next moment, represents the two-norm.

[0053] As a preferred implementation, the basic influencing factors include:

[0054] Temperature, humidity, cloud cover, cloud echo intensity, cloud base height, cloud thickness, terrain factors and airflow influencing factors.

[0055] On the other hand, the present invention also provides a system for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir, comprising:

[0056] The target construction module is used to construct the site selection problem of acoustic rainfall enhancement operation points for reservoir storage as a multi-target combination problem;

[0057] The impact factor determination module obtains the basic impact factors of the multi-objective combination problem based on the decision space of the multi-objective combination problem;

[0058] The Markov decision module is used to construct the objective function of the multi-objective combination problem and convert the multi-objective combination problem into a Markov decision process using multivariate groups based on the basic influencing factors and the historical acoustic operation data of the reservoir storage.

[0059] The site selection optimization module is used to apply the multi-objective reinforcement learning algorithm to the above-mentioned Markov decision process. The control system of the acoustic rainmaking device is taken as the intelligent agent and the target reservoir range is taken as the environment. The multi-objective reinforcement learning algorithm is used to obtain the Pareto frontier of the intelligent agent in different states, and the address of the acoustic operation point of the target reservoir is output based on the Pareto frontier.

[0060] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for selecting a site for an acoustic wave rainfall enhancement operation point for increasing water storage in a reservoir as described in any embodiment of the present invention is implemented.

[0061] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the method for selecting a site for an acoustic wave rainfall enhancement operation point for reservoir storage as described in any embodiment of the present invention is implemented.

[0062] The present invention has the following beneficial effects:

[0063] 1. The present invention converts the site selection of acoustic rain enhancement operation points for reservoir storage into a multi-objective combination problem, collects the basic influencing factors affecting the site selection problem based on the decision space of the multi-objective combination problem, and constructs an objective function for optimizing multiple objectives. The Markov decision process is then used to convert the multi-objective optimization problem into a random dynamic process, and the multi-objective reinforcement learning method is used to output the Pareto frontier corresponding to the multi-objective optimization in this random dynamic process, and the site selection points of acoustic rain enhancement operation points that comprehensively consider multiple reservoir storage problems in the dynamic process are obtained, thereby realizing the intelligent site selection of acoustic rain enhancement operation points.

[0064] 2. The present invention comprehensively considers the storage capacity optimization, rainfall impact and environmental impact issues in the site selection of the acoustic rain enhancement operation point for reservoir storage, and constructs the objective function respectively, so that the Pareto frontier outputted later comprehensively considers the relevance of the acoustic rain enhancement and the planned storage, and the acoustic rain enhancement has the least impact on the natural environment and the geographical environment. While ensuring the completion of the acoustic rain enhancement operation, the possible threats caused by the acoustic rain enhancement operation are reduced.

[0065] 3. The present invention converts the Pareto set into a serialized optimal point set corresponding to each target with consistent scale by standardizing the Pareto set, and determines the reward coefficient of the target in the next reinforcement learning according to the two points with the largest distance in the serialized optimal point set of each target. The importance of different targets at each moment is introduced through the reward coefficient to obtain the maximum cumulative reward considering each target and approach the Pareto frontier. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 The figure is a schematic diagram of the method flow of the first embodiment of the present invention. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0069] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0070] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0071] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0072] Embodiment 1:

[0073] See also Figure 1 This embodiment provides a method for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir, comprising the following steps:

[0074] S100. The problem of site selection for acoustic rainfall enhancement operation points for increasing reservoir storage is constructed as a multi-objective combination problem. Specifically, in the present embodiment, the problem of site selection for acoustic rainfall enhancement operation points for increasing reservoir storage is split into reservoir storage increase target, rainfall increase impact target and environmental impact target, and a multi-objective combination problem with optimal reservoir storage increase, optimal rainfall increase impact and optimal environmental impact is constructed.

[0075] S200. Obtain basic influencing factors of the multi-objective combination problem based on the decision space of the multi-objective combination problem; for the multi-objective combination problem, there is a decision space for each sub-problem of each objective, and each decision space contains a number of decision variables that affect the corresponding objective. The decision variables in each decision space are repeatedly merged to obtain all basic influencing factors. Specifically, in this embodiment, the basic influencing factors include: air temperature, air humidity, cloud coverage, cloud echo intensity, cloud base height, cloud thickness, terrain factors, and airflow influencing factors, communication conditions, power supply conditions, etc., wherein the terrain factors are specifically characteristic values ​​set for different terrains and landforms, such as characteristic values ​​for mountainous terrain, characteristic values ​​for flat terrain, characteristic values ​​for water surface, etc.; the airflow influencing factors can be specifically set according to the distribution of rising airflow.

[0076] S300, constructing the objective function of the multi-objective combination problem, and based on each basic influencing factor and the historical acoustic wave operation data of the reservoir storage, the historical acoustic wave operation data of the reservoir storage provides detailed information about the historical process of the reservoir storage, including the basic influencing factor data of the historical acoustic wave rainfall enhancement operation site, and the rainfall duration data, rainfall data, reservoir expansion data, etc. caused by the acoustic wave operation. After clarifying each basic influencing factor and the historical acoustic wave operation data of the reservoir storage, the multi-objective combination problem is converted into a Markov decision process using a multivariate group. The Markov decision process is a mathematical model for processing random decision problems. This embodiment converts the multi-objective combination problem into a more intuitive and easy-to-handle mathematical problem through the Markov decision process. It is helpful to better understand and solve the multi-objective combination problem.

[0077] S400. Apply the multi-objective reinforcement learning algorithm to the above Markov decision process, take the control system of the acoustic rain-making device as the intelligent agent, and the target reservoir range as the environment. Use the multi-objective reinforcement learning algorithm to obtain the Pareto frontier of the intelligent agent in different states, and output the address of the acoustic operation point of the target reservoir based on the Pareto frontier. Pareto Frontier, also known as Pareto Boundary or Pareto Boundary Surface, is a set of optimal solutions found by optimization algorithms under multiple objective functions. Each solution in this set satisfies the following condition: it cannot be improved on any objective function without compromising the performance of another or more objective functions. In other words, the solutions on the Pareto frontier are mutually exclusive, and there is no room for Pareto improvement.

[0078] By obtaining a graphical representation of the Pareto front, the trade-offs between different optimization objectives in the acoustic operation site selection problem for different target reservoirs can be intuitively displayed. Decision makers can understand the trade-offs between different optimization objectives by observing the Pareto front, and find and output acoustic operation sites that meet their needs.

[0079] As a preferred implementation of this embodiment, the multi-objective combination problem includes reservoir capacity optimization, rainfall enhancement effect optimization and environmental impact optimization. Specifically, reservoir capacity optimization refers to minimizing the difference between the water volume change in the reservoir and the reservoir storage volume in the pre-formulated reservoir storage plan; rainfall enhancement effect optimization refers to minimizing the impact of the acoustic rainfall enhancement operation on the natural environment to prevent hazards such as landslides, water accumulation, and runoff flooding low-lying areas; environmental impact optimization refers to optimizing the impact of the acoustic rainfall enhancement operation site on the geographical environment, specifically including the influence of factors such as topography, altitude, rainfall area and reservoir overlap area.

[0080] As a preferred implementation of this embodiment, based on the above multi-objective combination problem including reservoir capacity optimization, rainfall enhancement effect optimization and environmental impact optimization, the objective function of constructing the multi-objective combination problem in this embodiment specifically includes:

[0081] Optimization objective function of reservoir capacity, optimization objective function of rainfall enhancement effect and optimization objective function of environmental impact;

[0082] The reservoir capacity optimization objective function is specifically:

[0083]

[0084] in, For the formulation The planned reservoir water storage increase during the period, for The estimated precipitation in the reservoir area during the period of time for the acoustic rainfall enhancement operation at the corresponding site, for The natural increase in water storage in the reservoir during the period, for The amount of water evaporation from the reservoir during the period;

[0085] It should be noted that The rainfall data of natural rainfall in the expected period can be obtained through the runoff and time of the reservoir's upstream, and the rainfall data of natural rainfall in the expected period can be accumulated. The rainfall data of natural rainfall in the expected period can be obtained through meteorological forecast data; The calculation of water evaporation involves multiple factors, including air temperature, relative humidity, wind speed, and the water surface area of ​​the reservoir. Based on the above factors, empirical formulas or models are used to estimate the water evaporation in the expected period.

[0086] Specifically, In order to combine the water storage increase of the acoustic rainfall enhancement operation with the water storage increase of the reservoir during the expected period of natural changes in the reservoir water volume, the result is combined with The difference between the planned reservoir water storage increases can be minimized, so that the reservoir capacity can be optimized.

[0087] The optimization objective function of the rainfall enhancement effect is specifically:

[0088]

[0089] in, for The number of predicted changes in rainfall type caused by acoustic rainfall enhancement operations at the corresponding site within the time period; For the The weight corresponding to the rainfall type, ∈0~1; For the The predicted duration of rainfall for this rainfall type.

[0090] It can be understood that the greater the rainfall and the longer the duration, the more likely it is to affect the natural environment. In this embodiment, different weights are set for different rainfall types, for example: light rain is 0.05, light rain is 0.05, moderate rain is 0.1, moderate rain is 0.2, heavy rain is 0.2, and rainstorm is 0.4. The greater the rainfall, the greater the impact, the higher the weight, and the greater the product with a certain rainfall duration; then the product of the weights of the rainfall types caused by all acoustic rain enhancement operations in the expected time period and the corresponding rainfall duration is minimized, so that the rainfall caused by the acoustic rain enhancement operations can minimize the impact on the natural environment.

[0091] The number of predicted changes in rainfall types caused by the acoustic rainfall enhancement operation can be predicted by a machine learning algorithm. This embodiment provides an implementation method, which specifically includes the following steps:

[0092] Obtain the acoustic rain enhancement historical operation data set, which contains factors that affect rainfall, such as the power data set by the acoustic rain enhancement operation device, temperature, humidity, cloud coverage, cloud echo intensity, cloud base height, cloud thickness, etc.; it also contains the rainfall type change curve caused by each acoustic rain enhancement operation;

[0093] Build a neural network structure consisting of an encoder network and a decoder network;

[0094] The power data, temperature, humidity, cloud coverage, cloud echo intensity, cloud base height, cloud thickness and other factor data in each historical operation data are used as the input of the encoder, and a feature value is output through the encoder. The feature value is input into the decoder network, and the feature curve is output through the decoder network; the loss function is constructed with the similarity between the feature curve and the rainfall type change curve as the target;

[0095] The hyperparameters of the encoder network and the decoder network are adjusted iteratively through the loss calculated by the loss function until the similarity between the output characteristic curve of the decoder network and the rainfall type change curve is greater than the preset threshold, and the model training is completed. The model can be used to predict the number of rainfall type changes and the specific change type according to the power set by the acoustic rain enhancement equipment and the observed values ​​of environmental factors.

[0096] The environmental impact optimization objective function is specifically:

[0097]

[0098] in, is the geomorphic coefficient of the corresponding site. The more convenient the geomorphic transportation of the corresponding site is, the smaller the geomorphic coefficient is; is the altitude coefficient of the corresponding site, and the altitude coefficient decreases with the altitude; is the acoustic wave impact coefficient of the corresponding site. The fewer the number of households in the preset range of the corresponding site, the smaller the acoustic wave impact coefficient; is the reservoir area, It is the overlapping area of ​​the rainfall area and the reservoir area caused by the acoustic rainfall enhancement operation at the corresponding site. , altitude coefficient and acoustic influence coefficient are all in the interval [0,1].

[0099] It should be noted that the geomorphic coefficient The smaller the coefficient, the more convenient the transportation of the corresponding location is, the less impact it will have on the surrounding environment, and the smaller the calculated value of the function is. The smaller the value, the higher the altitude of the corresponding site. The higher the altitude of the corresponding site, the closer the acoustic rain enhancement operation is to the clouds and the farther it is from the residents. The smaller the impact on the surrounding environment, the smaller the calculated value of the function. The number of households within the sound wave influence range of the corresponding site can be set. For example, the number of households within 500m∈(0,5] is 0.1, the number of households within 500m∈(5,10] is 0.3, and the number of households within 500m is greater than 10, which is set to 0.6. The sound waves emitted by the acoustic rain enhancement device are low-frequency and high-power sound waves, which will affect human health. Therefore, the fewer people within the sound wave influence range, the better the corresponding site selection, and the smaller the function calculation value. Reservoir area is an obtainable fixed value, and the overlapping area between the rainfall area caused by the acoustic rainfall enhancement operation and the reservoir area It can be obtained based on the basic influencing factors observed at the corresponding site, and the rainfall area can be determined based on the basic influencing factors through expert experience, thereby calculating the overlapping area. It can also be built through historical data to build a machine learning model, determine the rainfall area through the machine learning model, and then calculate the overlapping area.

[0100] As a preferred implementation of this embodiment, the step of converting the multi-objective combination problem into a Markov decision process using a multi-tuple is specifically as follows:

[0101] Using historical acoustic operation data of reservoir storage, using multivariate Construct a Markov decision process for selecting the site for acoustic rainfall enhancement operations to increase reservoir storage;

[0102] in, is the state space, , for At this moment, from the reservoir Status data obtained from each location, ,in for At this moment, from the reservoir range The first location obtained Observed values ​​of basic influencing factors; is the action space, The action of the agent at each moment ,action correspond The movement of the acoustic rainfall enhancement operation point at all times; is the state transfer matrix; is a reward function, which is specifically the objective function of the multi-objective combination problem.

[0103] As a preferred implementation of this embodiment, the steps of using a multi-objective reinforcement learning algorithm to obtain the Pareto frontier of the agent in different states are specifically as follows:

[0104] Establish the function for finding the maximum reward as shown below:

[0105]

[0106] in, is the cumulative reward of the Mth objective in the objective function of the multi-objective combination problem corresponding to the reward function;

[0107]

[0108] in, Considering the state data at time t and actions The cumulative reward calculation function of According to the state data at time t and actions The joint probability distribution of To find the optimal termination time, is the reward coefficient of the Mth target, Indicates the importance of the Mth target at time t, ∈0~1, and , Indicates The importance of a goal; According to the state data at time t and actions The calculated reward value of the Mth target;

[0109] Initialize the multi-objective reinforcement learning algorithm, iteratively perform the multi-objective reinforcement learning task through the historical acoustic operation data of reservoir storage, and put the obtained Pareto points into the Pareto set;

[0110] The Pareto set is standardized, and the standardized Pareto points in the standardized Pareto set are sorted to obtain the serialized optimal point set;

[0111] At each iteration, the reward coefficient of the Mth target at the next moment is determined based on the serialized optimal point set , according to the reward coefficient of the Mth target at the next moment Perform the reinforcement learning task at the next moment, obtain a new Pareto set, and reconstruct the Pareto frontier based on the new Pareto set.

[0112] As a preferred implementation of this embodiment, the set is standardized, and the standardized Pareto points in the standardized Pareto set are sorted to obtain a serialized optimal point set; at each iteration, the reward coefficient of the Mth target at the next moment is determined based on the serialized optimal point set. The specific steps are:

[0113] According to the Pareto set, the minimum value set and maximum value set in the Pareto point corresponding to each target are obtained, where the minimum value set is:

[0114]

[0115] in, Indicates The first Pareto optimal point; Indicates The minimum value among the Pareto optimal points of the objectives;

[0116] The maximum value set is:

[0117]

[0118] in, Indicates The maximum value among the Pareto optimal points of the objectives;

[0119] Standardize each Pareto optimal point in the Pareto set:

[0120]

[0121] in, is the Pareto optimal point after standardization;

[0122] The standardized Pareto optimal points are sorted by target to obtain the serialized optimal point set for each target:

[0123]

[0124] in, Sort the jth point in the sequenced optimal point set for the Mth goal, is the total number of sorts;

[0125] According to the serialized optimal point set of each target, the reward coefficient of the Mth target at the next moment is calculated by the following formula:

[0126]

[0127] in, represents the reward coefficient of the Mth target at the next moment, represents the two-norm.

[0128] Based on the contents of the above embodiments, this embodiment converts the site selection of acoustic rainfall enhancement operation points for reservoir storage into a multi-objective combination problem, aggregates the basic influencing factors affecting the site selection problem based on the decision space of the multi-objective combination problem, and constructs an objective function for optimizing multiple objectives. The objective function is solved by using the Markov decision process and multi-objective reinforcement learning method, and the Pareto frontier corresponding to the multi-objective optimization is output. The site selection points of the acoustic rainfall enhancement operation points that comprehensively consider the storage problems of multiple reservoirs are obtained, thereby realizing the intelligent site selection of the acoustic rainfall enhancement operation points.

[0129] This embodiment comprehensively considers the storage capacity optimization, rainfall impact and environmental impact issues in the site selection of the acoustic rain enhancement operation point for reservoir storage, and constructs the objective functions respectively, so that the Pareto frontier outputted later comprehensively considers the relevance of the acoustic rain enhancement and the planned storage increase, and the acoustic rain enhancement has the least impact on the natural environment and the least impact on the geographical environment. While ensuring the completion of the acoustic rain enhancement operation, the possible threats caused by the acoustic rain enhancement operation are reduced.

[0130] This embodiment converts the Pareto set into a scaled serialized optimal point set corresponding to each target by standardizing the Pareto set, determines the reward coefficient of the target in the next reinforcement learning according to the two points with the largest distance in the serialized optimal point set of each target, and introduces the importance of different targets at each moment through the reward coefficient to obtain the maximum cumulative reward considering each target and approach the Pareto frontier.

[0131] Embodiment 2:

[0132] This embodiment provides a system for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir, comprising:

[0133] A target construction module is used to construct the site selection problem of the acoustic rainfall enhancement operation point for reservoir storage into a multi-objective combination problem; this module is used to implement the function of step S100 in the first embodiment, which will not be described in detail here;

[0134] An impact factor determination module is used to obtain the basic impact factor of the multi-objective combination problem based on the decision space of the multi-objective combination problem; this module is used to implement the function of step S200 in the first embodiment, and will not be repeated here;

[0135] A Markov decision module is used to construct an objective function of a multi-objective combination problem, and based on each basic influencing factor and the historical acoustic operation data of the reservoir storage, the multi-objective combination problem is converted into a Markov decision process using a multi-tuple; this module is used to implement the function of step S300 in the first embodiment, which will not be described in detail here;

[0136] The site selection optimization module is used to apply the multi-objective reinforcement learning algorithm to the above-mentioned Markov decision process, taking the control system of the acoustic rainmaking device as the intelligent agent and the target reservoir range as the environment, and using the multi-objective reinforcement learning algorithm to obtain the Pareto front of the intelligent agent in different states, and output the address of the acoustic operation point of the target reservoir based on the Pareto front; this module is used to implement the function of step S400 in Example 1, which will not be repeated here.

[0137] Embodiment three:

[0138] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for selecting a site for an acoustic wave rainfall enhancement operation point for increasing water storage in a reservoir as described in any embodiment of the present invention is implemented.

[0139] Embodiment 4:

[0140] This embodiment proposes a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for selecting a site for an acoustic rainfall enhancement operation point for increasing water storage in a reservoir as described in any embodiment of the present invention is implemented.

[0141] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0142] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0143] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0144] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0145] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for selecting a site for acoustic rainfall enhancement operation for reservoir storage. It is characterized in that The following steps are involved: The problem of site selection for acoustic rainfall enhancement operation points for reservoir storage is constructed as a multi-objective combination problem. Obtain the basic influencing factors of multi-objective combination problems based on their decision space; The objective function of the multi-objective combination problem is constructed, and based on the basic influencing factors and the historical acoustic operation data of reservoir storage, the multi-objective combination problem is converted into a Markov decision process using multivariate groups; The multi-objective reinforcement learning algorithm is applied to the above Markov decision process. The control system of the acoustic rainmaking device is used as the intelligent agent, and the target reservoir range is used as the environment. The multi-objective reinforcement learning algorithm is used to obtain the Pareto frontier of the intelligent agent in different states, and the address of the acoustic operation point of the target reservoir is output based on the Pareto frontier. The multi-objective combination problem includes optimization of reservoir capacity, optimization of rainfall enhancement effect and optimization of environmental impact; The objective function of constructing the multi-objective combination problem specifically includes: Optimization objective function of reservoir capacity, optimization objective function of rainfall enhancement effect and optimization objective function of environmental impact; The reservoir capacity optimization objective function is specifically: in, For the formulation The planned reservoir water storage increase during the period, for The estimated precipitation in the reservoir area during the period of time for the acoustic rainfall enhancement operation at the corresponding site, for The natural increase in water storage in the reservoir during the period, for The amount of water evaporation from the reservoir during the period; The optimization objective function of the rainfall enhancement effect is specifically: in, for The number of predicted changes in rainfall type caused by acoustic rainfall enhancement operations at the corresponding site within the time period; For the The weight corresponding to the rainfall type, ∈0~1; For the The predicted duration of rainfall for each rainfall type; The environmental impact optimization objective function is specifically: in, is the geomorphic coefficient of the corresponding site. The more convenient the geomorphic transportation of the corresponding site is, the smaller the geomorphic coefficient is; is the altitude coefficient of the corresponding site, and the altitude coefficient decreases with the altitude; is the acoustic wave impact coefficient of the corresponding site. The fewer the number of households in the preset range of the corresponding site, the smaller the acoustic wave impact coefficient; is the reservoir area, It is the overlapping area between the rainfall area and the reservoir area caused by the acoustic rainfall enhancement operation at the corresponding site.

2. A method for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir according to claim 1, It is characterized in that The steps of converting the multi-objective combination problem into a Markov decision process using a multi-tuple are specifically as follows: Using historical acoustic operation data of reservoir storage, using multivariate Construct a Markov decision process for selecting the site for acoustic rainfall enhancement operations to increase reservoir storage; in, is the state space, , for At this moment, from the reservoir Status data obtained from each location, ,in for At this moment, the reservoir The first location obtained Observed values ​​of basic influencing factors; is the action space, The action of the agent at each moment ,action correspond The movement of the acoustic rain enhancement operation point at all times; is the state transfer matrix; is a reward function, which is specifically the objective function of the multi-objective combination problem.

3. A method for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir according to claim 2, It is characterized in that The steps of using the multi-objective reinforcement learning algorithm to obtain the Pareto frontier of the agent in different states are specifically as follows: Establish the function for finding the maximum reward as shown below: in, is the cumulative reward of the Mth objective in the objective function of the multi-objective combination problem corresponding to the reward function; in, Considering the state data at time t and actions The cumulative reward calculation function of According to the state data at time t and actions The joint probability distribution of To find the optimal termination time, is the reward coefficient of the Mth target, indicating the importance of the Mth target at time t, ∈0~1, and ; According to the state data at time t and actions The calculated reward value of the Mth target; Initialize the multi-objective reinforcement learning algorithm, iteratively perform the multi-objective reinforcement learning task through the historical acoustic operation data of reservoir storage, and put the obtained Pareto points into the Pareto set; The Pareto set is standardized, and the standardized Pareto points in the standardized Pareto set are sorted to obtain the serialized optimal point set; At each iteration, the reward coefficient of the Mth target at the next moment is determined based on the serialized optimal point set , according to the reward coefficient of the Mth target at the next moment Perform the reinforcement learning task at the next moment, obtain a new Pareto set, and reconstruct the Pareto frontier based on the new Pareto set.

4. A method for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir according to claim 3, It is characterized in that The Pareto set is standardized, and the standardized Pareto points in the standardized Pareto set are sorted to obtain a serialized optimal point set; in each iteration, the reward coefficient of the Mth target at the next moment is determined based on the serialized optimal point set. The specific steps are: According to the Pareto set, the minimum value set and maximum value set in the Pareto point corresponding to each target are obtained, where the minimum value set is: in, Indicates The first Pareto optimal point; Indicates The minimum value among the Pareto optimal points of the objectives; The maximum value set is: in, Indicates The maximum value among the Pareto optimal points of the objectives; Standardize each Pareto optimal point in the Pareto set: in, is the Pareto optimal point after standardization; The standardized Pareto optimal points are sorted by target to obtain the serialized optimal point set for each target: in, Sort the jth point in the sequenced optimal point set for the Mth goal, is the total number of sorts; According to the serialized optimal point set of each target, the reward coefficient of the Mth target at the next moment is calculated by the following formula: in, represents the reward coefficient of the Mth target at the next moment, represents the two-norm.

5. The method for selecting a site for an acoustic rainfall enhancement operation point for increasing the storage capacity of a reservoir according to claim 1, It is characterized in that The basic impact factors include: Temperature, humidity, cloud cover, cloud echo intensity, cloud base height, cloud thickness, terrain factors and airflow influencing factors.

6. A system for selecting sites for acoustic rainfall enhancement operations for reservoir storage. It is characterized in that include: The target construction module is used to construct the site selection problem of acoustic rainfall enhancement operation points for reservoir storage as a multi-target combination problem; The impact factor determination module obtains the basic impact factors of the multi-objective combination problem based on the decision space of the multi-objective combination problem; The Markov decision module is used to construct the objective function of the multi-objective combination problem and convert the multi-objective combination problem into a Markov decision process using multivariate groups based on the basic influencing factors and the historical acoustic operation data of the reservoir storage. The site selection optimization module is used to apply the multi-objective reinforcement learning algorithm to the above-mentioned Markov decision process, taking the control system of the acoustic rainmaking device as the intelligent agent and the target reservoir range as the environment, and using the multi-objective reinforcement learning algorithm to obtain the Pareto front of the intelligent agent in different states, and output the address of the acoustic operation point of the target reservoir based on the Pareto front; The multi-objective combination problem includes optimization of reservoir capacity, optimization of rainfall enhancement effect and optimization of environmental impact; The objective function of constructing the multi-objective combination problem specifically includes: Optimization objective function of reservoir capacity, optimization objective function of rainfall enhancement effect and optimization objective function of environmental impact; The reservoir capacity optimization objective function is specifically: in, For the formulation The planned reservoir water storage increase during the period, for The estimated precipitation in the reservoir area during the period of time for the acoustic rainfall enhancement operation at the corresponding site, for The natural increase in water storage in the reservoir during the period, for The amount of water evaporation from the reservoir during the period; The optimization objective function of the rainfall enhancement effect is specifically: in, for The number of predicted changes in rainfall type caused by acoustic rainfall enhancement operations at the corresponding site within the time period; For the The weight corresponding to the rainfall type, ∈0~1; For the The predicted duration of rainfall for each rainfall type; The environmental impact optimization objective function is specifically: in, is the geomorphic coefficient of the corresponding site. The more convenient the geomorphic transportation of the corresponding site is, the smaller the geomorphic coefficient is; is the altitude coefficient of the corresponding site, and the altitude coefficient decreases with the altitude; is the acoustic wave impact coefficient of the corresponding site. The fewer the number of households in the preset range of the corresponding site, the smaller the acoustic wave impact coefficient; is the reservoir area, It is the overlapping area between the rainfall area and the reservoir area caused by the acoustic rainfall enhancement operation at the corresponding site.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the method for selecting a site for an acoustic rainfall enhancement operation point for increasing reservoir storage as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the method for selecting a site for an acoustic rainfall enhancement operation point for increasing reservoir storage as described in any one of claims 1 to 5 is implemented.

Images