A multi-source and multi-dimensional dynamic configuration method for rural drinking water

Abstract

This invention discloses a multi-dimensional dynamic configuration method for multiple water sources in rural drinking water. This method, pertaining to the field of multi-water source configuration, includes collecting water source distribution information and village distribution information within a study area; determining the maximum coverage area of ​​each water source based on the water source distribution information; determining a water demand model for each village, used to describe the water supply conditions of each village, based on the village distribution information and the maximum coverage area of ​​each water source and the water demand of each village; and solving a water supply strategy for each water source to the village based on the maximum coverage area of ​​each water source and the water demand model of each village. This invention addresses the problem of existing methods that fail to consider construction costs and water supply efficiency from both the water supply and water demand sides.

Description

Technical Field

[0001] The present invention belongs to the field of multi-water source configuration, and in particular relates to a multi-dimensional dynamic configuration method for multi-water sources of rural drinking water. Background Art

[0002] With the development of the rural economy and the improvement of living standards, rural residents' demand for domestic water continues to increase. They not only require sufficient water quantity, but also have higher requirements for water quality.

[0003] Rural drinking water safety is crucial to the health and quality of life of rural residents. In rural areas, water resources are unevenly distributed, water demands are diverse, and water quality fluctuates. Single-source water supply models are unable to meet these growing and complex demands. Therefore, dynamic deployment of multiple water sources is becoming a crucial approach to ensuring rural drinking water safety.

[0004] Taking a comprehensive view of water resources from a macro perspective can better coordinate water use for life, production and ecology, and achieve sustainable use of water resources; it is conducive to giving full play to the comprehensive benefits of water conservancy projects and improving regional water resource allocation capabilities; it is of great significance for ensuring water supply security in large rural areas or ecologically sensitive areas. Summary of the Invention

[0005] In response to the above-mentioned deficiencies in the prior art, the present invention provides a multi-source and multi-dimensional dynamic configuration method for rural drinking water, which solves the problem that the prior method does not consider the construction cost and water supply efficiency from both the water supply end and the water demand end.

[0006] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a method for multi-source and multi-dimensional dynamic configuration of rural drinking water, comprising:

[0007] Collect information on water source distribution and village distribution in the study area;

[0008] Based on the water source distribution information, determine the maximum coverage area of ​​each water source;

[0009] Based on the village distribution information and the maximum coverage area of ​​each water source, a water demand model is determined for each village to describe the water supply conditions of the village.

[0010] Based on the maximum coverage area of ​​each water source and the water demand model of each village, the water supply strategy of each water source to the village is solved.

[0011] Furthermore, according to the village distribution information and the maximum coverage area of ​​each water source, a water demand model for each village used to describe the water supply condition of the village is determined based on the water demand of each village, specifically:

[0012] Collect historical data on water supply from various water sources and determine the historical water supply sources for each village;

[0013] Obtain water demand of each village;

[0014] Based on the maximum coverage area of ​​each water source, the historical water supply sources of each village and the water demand of each village, a water demand model was obtained to describe the water supply conditions of the villages.

[0015] Furthermore, the expression of the water demand model is:

[0016]

[0017] Among them, S i is the water demand model of the i-th village; M i is the historical water supply source set for the i-th village; is the set of alternative water sources for the i-th village; C i is the water demand of the i-th village; is a binary function used to determine whether to use alternative water sources. When it is 0, it is set to 0, indicating that the alternative water source does not need to be used; otherwise, it is set to 1, indicating that the alternative water source needs to be used; E i is the set of all water sources covering the i-th village; is the total water supply of the historical water supply source set; C m M i The water supply of the mth historical water supply source in |M i |For M i The number of historical water supply sources in the country.

[0018] Furthermore, according to the maximum coverage area of ​​each water source and the water demand model of each village, the water supply strategy of each water source to the village is solved, specifically:

[0019] According to the maximum coverage area of ​​each water source, the study area is divided into multiple water source supply areas and single water source supply areas;

[0020] Determine the remaining water supply of the water source in the single-source water supply area based on the water demand model of the villages in the single-source water supply area;

[0021] Determine the water supply capacity of each water supply source in the multi-source water supply area based on the remaining water supply capacity of the water supply source in the single-source water supply area;

[0022] Based on the water supply of each water source in the multi-source water supply area and the water demand model of each village in the multi-source water supply area, the water supply strategy of each water source to the village is solved.

[0023] Furthermore, according to the water supply of each water source in the multi-source water supply area and the water demand model of each village in the multi-source water supply area, the water supply strategy of each water source to the village is solved, specifically:

[0024] Based on the water supply volume of each water source in a multi-source water supply area and the water demand model of each village in the multi-source water supply area, each water source in the multi-source water supply area is regarded as an intelligent agent, and the state space, action space and reward are defined respectively. A deep reinforcement learning network model is used to solve the water supply strategy of each water source to the village.

[0025] Furthermore, the expression of the state space is:

[0026] s t ={M t ,S t}

[0027] Among them, s t is the state space at time t; M t is the remaining water supply of each water source in the multi-source water supply area at time t; S t It is the water demand model of the area supplied by multiple water sources at time t.

[0028] Furthermore, the expression of the action space is:

[0029] a t ={X(B:S B )}

[0030]

[0031] Among them, a t is the action space at time t; X(B:S B ) is the water supply village selection scheme for the water supply source with non-zero remaining water supply in the multi-source water supply area at time t; B is the set of water supply sources with non-zero remaining water supply in the multi-source water supply area at time t; S B The set of villages for which each water source in B is the historical water supply source; B is the action plan for the water supply village of the zth water supply source in the multi-source water supply area whose remaining water supply is not 0 at time t; z The zth water source whose remaining water supply in the multi-source water supply area is not 0 at time t; For B z A collection of villages that are historical water sources; f(S i (B z,t-1 )) is used to judge B z The binary function of whether the village supplied with water at the last moment meets the water demand. If so, it is 1, indicating that B z The next village needs to be selected for water supply, otherwise, it is 0, indicating B z The action remains unchanged and water supply continues to be provided to the village that was supplied with water at the last moment.

[0032] Furthermore, the reward is expressed as:

[0033] r t =α(1-MAN(f1))+β(1-MAN(f2))

[0034]

[0035] Among them, r t is the reward at time t; α is the weight of f1; MAN() is the normalization function; f1 is the first target; β is the weight of f2; f2 is the second target; D j is the number of water sources supplying the jth village in the multi-source water supply area at time t; K is the average number of water sources for each village in the multi-source water supply area where the number of water sources is not 0 at time t; p is the number of villages supplied by the pth water source in the multi-source water supply area at time t; The average number of villages supplied by a water source whose number of villages supplied by a water source is not zero in the multi-source water supply area at time t.

[0036] The beneficial effects of the present invention are as follows: the present invention fully takes into account the situation of insufficient water sources for water supply in rural areas, and takes single-source water supply areas into consideration in the overall water source allocation in rural areas. At the same time, in addition to this, for multi-source water supply areas, a deep reinforcement learning network model is used to construct two goals from the perspectives of the water supply source and the water-demanding villages. The first goal and the second goal jointly ensure the utilization rate of the water source and the stability of the water supply end of the water-demanding villages, and the reward function of the deep reinforcement learning network model is constructed based on the first goal and the second goal to guide the strategy solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0038] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0039] like Figure 1 As shown, in one embodiment of the present invention, a method for multi-dimensional dynamic configuration of rural drinking water with multiple water sources includes:

[0040] Collect information on water source distribution and village distribution in the study area;

[0041] Based on the water source distribution information, determine the maximum coverage area of ​​each water source;

[0042] Based on the village distribution information and the maximum coverage area of ​​each water source, a water demand model is determined for each village to describe the water supply conditions of the village.

[0043] Based on the maximum coverage area of ​​each water source and the water demand model of each village, the water supply strategy of each water source to the village is solved.

[0044] In this embodiment, taking into account the actual terrain factors, the area that each water supply source can cover is limited. Before configuring the water supply strategy for each village (water demand point), it is necessary to first investigate the maximum coverage range of each water supply source in order to maximize the utilization rate of the water supply source.

[0045] According to the village distribution information and the maximum coverage area of ​​each water source, the water demand model for each village used to describe the water supply conditions of the village is determined based on the water demand of each village. Specifically, it is:

[0046] Collect historical data on water supply from various water sources and determine the historical water supply sources for each village;

[0047] Obtain water demand of each village;

[0048] Based on the maximum coverage area of ​​each water source, the historical water supply sources of each village and the water demand of each village, a water demand model was obtained to describe the water supply conditions of the villages.

[0049] The expression of the water demand model is:

[0050]

[0051] Among them, S i is the water demand model of the i-th village; M i is the historical water supply source set for the i-th village; is the set of alternative water sources for the i-th village; C i is the water demand of the i-th village; is a binary function used to determine whether to use alternative water sources. When it is 0, it is set to 0, indicating that the alternative water source does not need to be used; otherwise, it is set to 1, indicating that the alternative water source needs to be used; E i is the set of all water sources covering the i-th village; is the total water supply of the historical water supply source set; C m M i The water supply of the mth historical water supply source in |M i |For M i The number of historical water supply sources in the country.

[0052] In this embodiment, the water demand model initially constructed is used to describe the historical water conditions of each village (water demand point). However, when solving the water supply source formula strategy, the water demand model is used to feedback environmental changes.

[0053] According to the maximum coverage area of ​​each water source and the water demand model of each village, the water supply strategy of each water source to the village is solved, specifically:

[0054] According to the maximum coverage area of ​​each water source, the study area is divided into multiple water source supply areas and single water source supply areas;

[0055] Determine the remaining water supply of the water source in the single-source water supply area based on the water demand model of the villages in the single-source water supply area;

[0056] Determine the water supply capacity of each water supply source in the multi-source water supply area based on the remaining water supply capacity of the water supply source in the single-source water supply area;

[0057] Based on the water supply of each water source in the multi-source water supply area and the water demand model of each village in the multi-source water supply area, the water supply strategy of each water source to the village is solved.

[0058] In this embodiment, the water source of the single-source water supply area will give priority to supplying water to the single-source water supply area. The water demand model of the villages in the single-source water supply area can provide feedback on whether the water supply resources in this area are sufficient. is an empty set and When it is 1, it indicates that a new water supply project is needed in this area.

[0059] The water supply strategy of each water source to the village is solved based on the water supply volume of each water source in the multi-source water supply area and the water demand model of each village in the multi-source water supply area, specifically:

[0060] Based on the water supply volume of each water source in a multi-source water supply area and the water demand model of each village in the multi-source water supply area, each water source in the multi-source water supply area is regarded as an intelligent agent, and the state space, action space and reward are defined respectively. A deep reinforcement learning network model is used to solve the water supply strategy of each water source to the village.

[0061] In this embodiment, the water supply strategy for each village is converted into a Markov decision process and solved using a deep reinforcement learning network model.

[0062] The expression of the state space is:

[0063] s t ={M t ,S t}

[0064] Among them, st is the state space at time t; M t is the remaining water supply of each water source in the multi-source water supply area at time t; S t It is the water demand model of the area supplied by multiple water sources at time t.

[0065] In this embodiment, two states need to be monitored in real time: the remaining water supply of each water supply source and the water availability of the water demand point. In this embodiment, the water supply mentioned is in daily units, including the maximum daily water supply and the remaining daily water supply.

[0066] The expression of the action space is:

[0067] a t ={X(B:S B )}

[0068]

[0069] Among them, a t is the action space at time t; X(B:S B ) is the water supply village selection scheme for the water supply source with non-zero remaining water supply in the multi-source water supply area at time t; B is the set of water supply sources with non-zero remaining water supply in the multi-source water supply area at time t; S B The set of villages for which each water source in B is the historical water supply source; B is the action plan for the water supply village of the zth water supply source in the multi-source water supply area whose remaining water supply is not 0 at time t; z The zth water source whose remaining water supply in the multi-source water supply area is not 0 at time t; For B z A collection of villages that are historical water sources; f(S i (B z,t-1 )) is used to judge B z The binary function of whether the village supplied with water at the last moment meets the water demand. If so, it is 1, indicating that B z The next village needs to be selected for water supply, otherwise, it is 0, indicating B z The action remains unchanged and water supply continues to be provided to the village that was supplied with water at the last moment.

[0070] In this embodiment, the action selection is based on the stability of water supply to avoid unnecessary facility construction. When the current water supply source is supplying water to the corresponding village, if there is still water supply from the current water supply source and the water demand of the corresponding village is not met, the current water supply source should maintain the current action.

[0071] The reward expression is:

[0072] r t=α(1-MAN(f1))+β(1-MAN(f2))

[0073]

[0074] Among them, r t is the reward at time t; α is the weight of f1; MAN() is the normalization function; f1 is the first target; β is the weight of f2; f2 is the second target; D j is the number of water sources supplying the jth village in the multi-source water supply area at time t; K is the average number of water sources for each village in the multi-source water supply area where the number of water sources is not 0 at time t; p is the number of villages supplied by the pth water source in the multi-source water supply area at time t; The average number of villages supplied by a water source whose number of villages supplied by a water source is not zero in the multi-source water supply area at time t.

[0075] In this embodiment, the first goal and the second goal are to ensure the stability of the water supply sources of each village and the maximum utilization rate of each water source respectively.

[0076] In this embodiment, the final solution is not the water supply sequence, but which villages each water source supplies water to and how much water is supplied per day.

[0077] According to the different water demands in different seasons, different water supply strategies can be solved.

[0078] Similarly, the water demand model of each village at the moment the deep reinforcement learning network model is solved can also reflect whether the corresponding area lacks water supply projects.

Claims

1. A multi-source multi-dimensional dynamic configuration method for rural drinking water, characterized in that: include: Collect information on water source distribution and village distribution in the study area; Based on the water source distribution information, determine the maximum coverage area of ​​each water source; Based on the village distribution information and the maximum coverage area of ​​each water source, a water demand model is determined for each village to describe the water supply conditions of the village. Based on the maximum coverage area of ​​each water source and the water demand model of each village, the water supply strategy of each water source to the village is solved. Specifically, based on the water supply volume of each water source in the multi-source water supply area and the water demand model of each village in the multi-source water supply area, each water source in the multi-source water supply area is used as an intelligent agent, and the state space, action space and reward are defined respectively. The water supply strategy of each water source to the village is solved using a deep reinforcement learning network model; The expression of the state space is: in, for The state space at the moment; for The remaining water supply of each water source in the multi-source water supply area at any given time; for Water demand model for areas with multiple water sources at all times; The expression of the action space is: in, for The action space at each moment; for Selection plan for water supply villages in water supply areas with multiple water sources and the remaining water supply is not zero at all times; for The set of water sources in the water supply area with multiple water sources whose remaining water supply is not 0 at any time; for The collection of villages whose water sources are historical water supply sources; for The remaining water supply in the multi-source water supply area is not 0 at the moment Water supply village action plan for each water supply source; for The remaining water supply in the multi-source water supply area is not 0 at the moment water supply sources; for a collection of villages that served as historical water sources; For judgment A binary function that determines whether the village supplied with water at the last moment meets the water demand. If so, it is 1, indicating Need to select the next village for water supply, otherwise, it is 0, indicating The action remains unchanged, and water supply continues to the villages that were supplied with water at the last moment; The reward expression is: in, for Rewards of the moment; for The weight of is the normalization function; As the first goal; for The weight of For the second goal; for The first Number of water supply sources per village; for The average number of water sources for each village in the area with multiple water sources at any given time where the number of water sources is not zero; for The first Number of villages supplied by the water source; for The average number of villages supplied by a water source in a water supply area with multiple water sources at any given time is not 0.

2. The multi-source multi-dimensional dynamic configuration method for rural drinking water according to claim 1 is characterized in that: According to the village distribution information and the maximum coverage area of ​​each water source, the water demand model for each village used to describe the water supply conditions of the village is determined based on the water demand of each village. Specifically, it is: Collect historical data on water supply from various water sources and determine the historical water supply sources for each village; Obtain water demand of each village; Based on the maximum coverage area of ​​each water source, the historical water supply sources of each village and the water demand of each village, a water demand model was obtained to describe the water supply conditions of the villages.

3. The multi-source multi-dimensional dynamic configuration method for rural drinking water according to claim 2 is characterized in that: The expression of the water demand model is: in, For the Water demand model for each village; For the A collection of historical water supply sources for villages; For the A collection of alternative water sources for each village; For the Water demand of a village; is a binary function used to determine whether to use alternative water sources. When it is 0, it is set to 0, indicating that the alternative water source does not need to be used. Otherwise, it is set to 1, indicating that the alternative water source needs to be used. To cover the The collection of all water sources in a village; The total amount of water supplied by the water sources in the historical water supply source set; for Middle The water supply of each historical water supply source; for The number of historical water supply sources in the country.

4. The multi-source multi-dimensional dynamic configuration method for rural drinking water according to claim 1 is characterized in that: According to the maximum coverage area of ​​each water source and the water demand model of each village, the water supply strategy of each water source to the village is solved, specifically: According to the maximum coverage area of ​​each water source, the study area is divided into multiple water source supply areas and single water source supply areas; Determine the remaining water supply of the water source in the single-source water supply area based on the water demand model of the villages in the single-source water supply area; Determine the water supply capacity of each water supply source in the multi-source water supply area based on the remaining water supply capacity of the water supply source in the single-source water supply area; Based on the water supply of each water source in the multi-source water supply area and the water demand model of each village in the multi-source water supply area, the water supply strategy of each water source to the village is solved.

Images