An ecological traffic optimization system and method integrating visual perception and artificial intelligence

Abstract

The present invention proposes an ecological flow optimization system and method integrating visual perception and artificial intelligence. In the present invention, a visual perception water level and flow measurement device collects flow and water level data within the sliding window of each cross-section in sequence through a sliding window acquisition method; a flow evolution simulation model for each cross-section is constructed through a Transformer neural network model, and the flow data within the sliding window of the previous cross-section is input into the flow evolution simulation model of the present cross-section to predict the flow data within the sliding window of the present cross-section; the flow evolution simulation models of multiple cross-sections are cascaded in sequence to construct a multi-cross-section continuous flow evolution simulation model; the ecological flow of the downstream ecological assessment cross-section is obtained through the water level sorting wetted perimeter calculation method combined with the discrete difference method; according to the flow discharged continuously at multiple moments from the upstream reservoir cross-section, the flow data of the downstream ecological assessment cross-section is predicted through the multi-cross-section continuous flow evolution simulation model, and the newly discharged flow of the upstream reservoir cross-section is further calculated in combination with the ecological flow of the downstream ecological assessment cross-section.

Description

Technical Field

[0001] The present invention belongs to the field of ecological flow calculation and scheduling, and particularly relates to an ecological flow optimization system and method integrating visual perception and artificial intelligence. Background Art

[0002] With the large-scale construction and operation of water conservancy projects represented by reservoirs, the original natural hydrological regime of rivers has changed, which will have an adverse impact on the dynamic balance of river ecosystems. Therefore, in order to maintain the ecological health of rivers, it is necessary to explore the change characteristics of river runoff processes, determine the river ecological flow that ensures the stability of river ecosystems, and accordingly guide the ecological regulation of reservoirs. At present, there are two major difficulties in ecological flow calculation and reservoir ecological regulation. The first difficulty is the calculation of ecological flow under changing environments. Commonly used river ecological flow calculation methods mainly rely on hydrological methods, mainly based on historical runoff data, to obtain static ecological flows at long time scales such as years and months, and cannot calculate dynamic ecological flows at smaller time scales considering changing environments such as reservoir regulation and rainfall runoff. The second difficulty is the real-time ecological regulation of reservoirs. Usually, reservoirs are built upstream of rivers, while ecological assessment sections are generally set downstream of rivers. Therefore, the flow process of ecological assessment sections needs to consider the evolution calculation of reservoir discharge and the influence of inflow in the river reach, which makes it difficult for reservoirs to conduct real-time ecological regulation. Therefore, proposing an ecological flow calculation method considering historical data and real-time monitoring data and conducting real-time ecological regulation of reservoirs based on this is of great significance for ensuring the ecological health of rivers. Summary of the Invention

[0003] The purpose of the present invention is to solve the difficulties in ecological flow calculation and real-time ecological regulation. By deploying visual perception water level and flow measurement devices, constructing a multi-section flow evolution simulation model, combining the channel wetted perimeter method and the reservoir ecological regulation feedback mechanism, an ecological flow optimization and regulation technology integrating visual perception and artificial intelligence is proposed, which can provide technical support for ensuring the health of river ecosystems.

[0004] To solve the above technical problems, the present invention proposes an ecological flow optimization system and method integrating visual perception and artificial intelligence.

[0005] The technical solution of the system of the present invention is an ecological flow optimization system integrating visual perception and artificial intelligence, including:

[0006] Multiple visual perception water level and flow measurement devices, a remote server;

[0007] The multiple visual perception water level and flow measurement devices are sequentially deployed at multiple sections;

[0008] The remote server is sequentially connected to the multiple visual perception water level and flow measurement devices.

[0009] The technical solution of the method of the present invention is an ecological flow optimization method integrating visual perception and artificial intelligence, including the following steps:

[0010] Step 1: The water level and flow measurement device collects the flow and water level data in the sliding window of each cross-section in turn through the sliding window acquisition method, and wirelessly transmits the flow and water level data in the sliding window of each cross-section to the remote server;

[0011] Step 2: Build a flow evolution simulation model for each cross-section through the Transformer neural network model, input the flow data in the sliding window of the previous cross-section into the flow evolution simulation model of this cross-section for prediction, obtain the predicted flow data in the sliding window of this cross-section, combine the flow data in the sliding window of this cross-section to build a loss function model, and use the genetic algorithm to optimize and train to obtain the optimized flow evolution simulation model for each cross-section;

[0012] Step 3: Cascade the flow evolution simulation models of each cross-section in turn to build a multi-cross-section continuous flow evolution simulation model;

[0013] Step 4: The water level data in the sliding window of the downstream ecological assessment cross-section is calculated by the water level sorting wetted perimeter calculation method to obtain the wetted perimeter data in the sliding window of the downstream ecological assessment cross-section, and the discrete difference method is used to calculate the curvature data in the sliding window of the downstream ecological assessment cross-section, and the ecological flow of the downstream ecological assessment cross-section is obtained by combining the curvature data in the sliding window of the downstream ecological assessment cross-section;

[0014] Step 5: Build the flow data in the sliding window of the upstream reservoir cross-section by continuously discharging the flow at multiple moments at the upstream reservoir cross-section, input the flow data in the sliding window of the upstream reservoir cross-section into the multi-cross-section continuous flow evolution simulation model for prediction, obtain the predicted flow data in the sliding window of the downstream ecological assessment cross-section, and further calculate the newly discharged flow of the upstream reservoir cross-section in combination with the ecological flow of the downstream ecological assessment cross-section.

[0015] Preferably, the flow data in the sliding window of each cross-section described in Step 1 is specifically defined as follows:

[0016] X m =[Q m (1 + τ(m - 1)), …, Q m (t + τ(m - 1)), … Q m (T + τ(m - 1))], m ∈ [1, M], t ∈ [1, T]

[0017] Among them, X m represents the flow data in the m-th cross-section sliding window; Q m(t + τ(m - 1)) represents the flow rate data at the t-th moment within the sliding window of the m-th cross-section, that is, the flow rate data at the (t + τ(m - 1))-th acquisition moment of the m-th cross-section. τ represents the flow evolution lag time between adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window;

[0018] The water level data within the sliding window of each cross-section described in step 1 is specifically defined as follows:

[0019] Y m = [Z m (1 + τ(m - 1)), …, Z m (t + τ(m - 1)), … Z m (T + τ(m - 1))], m ∈ [1, M], t ∈ [1, T]

[0020] Among them, Y m represents the water level data within the sliding window of the m-th cross-section; Z m (t + τ(m - 1)) represents the water level data at the t-th moment within the sliding window of the m-th cross-section, that is, the water level data at the (t + τ(m - 1))-th acquisition moment of the m-th cross-section. τ represents the flow evolution lag time between adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window;

[0021] Preferably, the loss function model described in step 2 is specifically as follows:

[0022]

[0023]

[0024] Among them, represents the predicted flow rate data within the sliding window of the m-th cross-section, that is, the predicted flow rate data within the sliding window of this cross-section, represents the predicted flow rate data at the t-th moment at the (t + τ(m - 1))-th moment within the sliding window of this cross-section, that is, the predicted flow rate data at the t-th moment at the (t + τ(m - 1))-th moment within the sliding window of the m-th cross-section. τ represents the flow evolution lag time between adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window;

[0025] Preferably, the downstream ecological assessment cross-section described in step 4 is the M-th cross-section among the multiple cross-sections described in step 1;

[0026] The water level data within the sliding window of the downstream ecological assessment cross-section described in step 4 is specifically defined as follows:

[0027] Y M = [Z M (1 + τ(M - 1)), …, Z M (t + τ(M - 1)), … Z M(T + τ(M - 1))], t ∈ [1, T]

[0028] Among them, Y M represents the water level data within the sliding window of the Mth cross-section; Z M (t + τ(M - 1)) represents the water level data at the t-th moment within the sliding window of the Mth cross-section, that is, the water level data at the (t + τ(M - 1))-th acquisition moment of the Mth cross-section. τ represents the flow evolution lag time between two adjacent cross-sections, and T represents the length of the sliding window;

[0029] The water level sorting and wetted perimeter calculation method described in step 4 is as follows:

[0030] Sort the water level data at T moments within the sliding window of the Mth cross-section from smallest to largest to obtain the sorted water level data within the sliding window of the Mth cross-section;

[0031] Count the serial numbers of the water level data at each moment within the sliding window of the Mth cross-section in the sorted water level data within the sliding window of the Mth cross-section, and further construct the ranking data within the sliding window of the Mth cross-section;

[0032] The ranking data within the sliding window of the Mth cross-section is specifically defined as follows:

[0033] S M = [s M (1 + τ(M - 1)), …, s M (t + τ(M - 1)), … s M (T + τ(M - 1))], t ∈ [1, T]

[0034] Among them, S M represents the ranking data within the sliding window of the Mth cross-section; s M (t + τ(M - 1)) represents the t-th ranking data within the sliding window of the Mth cross-section. τ represents the flow evolution lag time between two adjacent cross-sections, and T represents the length of the sliding window;

[0035] Calculate the wetted perimeter data within the sliding window of the Mth cross-section by using the wetted perimeter calculation method for each moment's ranking data within the sliding window of the Mth cross-section. The specific definition is as follows:

[0036] R M = [r M (1 + τ(M - 1)), …, r M (t + τ(M - 1)), … r M (T + τ(M - 1))], t ∈ [1, T]

[0037] Among them, R M represents the wetted perimeter data within the sliding window of the Mth cross-section; r M(t + τ(M - 1)) represents the wetted perimeter data of the t-th ranked data within the sliding window of the M-th cross-section. τ represents the flow routing lag time between two adjacent cross-sections, and T represents the length of the sliding window;

[0038] The wetted perimeter data within the sliding window of the M-th cross-section is the wetted perimeter data within the sliding window of the downstream ecological assessment cross-section described in step 4;

[0039] The curvature data within the sliding window of the downstream ecological assessment cross-section obtained by calculation using the discrete difference method in step 4 is as follows:

[0040] The wetted perimeter data within the sliding window of the downstream ecological assessment cross-section and the flow rate data within the sliding window of the downstream ecological assessment cross-section are calculated using the discrete difference method to obtain the curvature data within the sliding window of the downstream ecological assessment cross-section;

[0041] The flow rate data within the sliding window of the downstream ecological assessment cross-section is specifically defined as follows:

[0042] O M =[o M (1 + τ(M - 1)), …, o M (t + τ(M - 1)), … o M (T + τ(M - 1))], t ∈ [1, T]

[0043] Among them, O M represents the flow rate data within the sliding window of the M-th cross-section; o M (t + τ(M - 1)) represents the flow rate data corresponding to the t-th ranked data within the sliding window of the M-th cross-section. τ represents the flow routing lag time between two adjacent cross-sections, and T represents the length of the sliding window;

[0044] The curvature data within the sliding window of the downstream ecological assessment cross-section is specifically defined as follows:

[0045] K M =[k M (1 + τ(M - 1)), …, k M (t + τ(M - 1)), … k M (T + τ(M - 1))], t ∈ [1, T]

[0046] Among them, K M represents the curvature data within the sliding window of the M-th cross-section; k M (t + τ(M - 1)) represents the curvature data of the t-th ranked data within the sliding window of the M-th cross-section. τ represents the flow routing lag time between two adjacent cross-sections, and T represents the length of the sliding window;

[0047] The specific process of obtaining the ecological flow rate of the downstream ecological assessment cross-section in step 4 is as follows:

[0048] Among the curvature data at multiple moments within the sliding window of the downstream ecological assessment section, select the moment with the maximum curvature data, and use the flow rate data corresponding to the moment with the maximum curvature data within the sliding window of the downstream ecological assessment section as the ecological flow rate of the downstream ecological assessment section;

[0049] The ecological flow rate of the downstream ecological assessment section is defined as follows:

[0050] Q M (time + τ(M - 1))

[0051] where Q M (time + τ(M - 1)) represents the ecological flow rate of the downstream ecological assessment section, τ represents the flow rate evolution lag time between two adjacent sections, and time represents the moment with the maximum curvature data within the sliding window of the downstream ecological assessment section;

[0052] Preferably, the upstream reservoir section in step 5 is the first section among the multiple sections in step 1;

[0053] The predicted flow rate data within the sliding window of the downstream ecological assessment section in step 5 is specifically defined as follows:

[0054]

[0055] where X m p represents the flow rate data within the sliding window of the Mth section; Q m p (t + τ(M - 1)) represents the predicted flow rate data at the tth moment within the sliding window of the Mth section, τ represents the flow rate evolution lag time between two adjacent sections, M represents the number of sections, and T represents the length of the sliding window;

[0056] The further calculation of the additional discharge of the upstream reservoir section in step 5 by combining the ecological flow rate of the downstream ecological assessment section is as follows:

[0057] Compare the predicted flow rate data at each moment within the sliding window of the downstream ecological assessment section with the ecological flow rate of the downstream ecological assessment section in turn;

[0058] If the predicted flow rate data at each moment within the sliding window of the downstream ecological assessment section is greater than or equal to the ecological flow rate of the downstream ecological assessment section, the additional discharge of the upstream reservoir section is 0;

[0059] If there is a moment within the downstream ecological assessment section window where the predicted flow rate data is less than the ecological flow rate of the downstream ecological assessment section, increase the discharge of the upstream reservoir section through the upstream reservoir section discharge increment;

[0060] The increased discharge of the upstream reservoir section is calculated as follows:

[0061]

[0062] Among them, Q m p (t l +τ(M - 1)) represents the predicted flow data at the t l th moment within the sliding window of the Mth section. ΔQ(t l ) represents the increased discharge of the upstream reservoir section at the t l th moment. Q M (time + τ(M - 1)) represents the ecological flow of the downstream ecological assessment section. time represents the moment with the maximum curvature data within the sliding window of the downstream ecological assessment section. τ represents the flow evolution lag time between two adjacent sections. M represents the number of sections, and T represents the length of the sliding window.

[0063] Compared with the prior art, the present invention has the following advantages and effects:

[0064] The present invention combines visual perception, artificial intelligence methods with the channel wetted perimeter method, and for the first time proposes an ecological flow calculation method based on the point relationship between the wetted perimeter and flow of the historical monitoring data and real-time ecological assessment section, providing support for the real-time management and dynamic evaluation of river ecological health.

[0065] Based on the Transformer neural network and visual perception water level and flow measurement equipment, the present invention constructs a multi-section flow evolution simulation model for water, solves the evolution calculation problem of the upstream reservoir discharge to the downstream ecological section, and realizes the real-time ecological operation of the reservoir by introducing the reservoir ecological operation feedback mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 : Schematic diagram of the method flow of the embodiment of the present invention.

[0067] Figure 2 : Schematic diagram generalized with a certain river basin water system as an example in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0068] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0069] In specific implementation, the method proposed by the technical solution of the present invention can be automatically operated by those skilled in the art using computer software technology. The system device for realizing the method, such as a computer-readable storage medium storing the corresponding computer program of the technical solution of the present invention and a computer device including the operation of the corresponding computer program, should also be within the protection scope of the present invention.

[0070] The technical solution of the system in the embodiment of the present invention is an ecological flow optimization system integrating visual perception and artificial intelligence, including:

[0071] Multiple visual perception water level and flow measurement devices, a remote server;

[0072] Arrange multiple visual perception water level and flow measurement devices in multiple cross-sections in sequence;

[0073] The remote server is sequentially connected to the multiple visual perception water level and flow measurement devices;

[0074] The selected type of the visual perception water level and flow measurement device is the Wuda AiFlow video flow measurement product;

[0075] The selected type of the remote server is Dell PowerEdge T150;

[0076] Next, in conjunction with Figure 1 - Figure 2 Introduce an ecological flow optimization method integrating visual perception and artificial intelligence provided by the embodiment of the present invention, specifically as follows:

[0077] Step 1: The visual perception water level and flow measurement device sequentially collects the flow and water level data in the sliding window of each cross-section through the sliding window collection method, and wirelessly transmits the flow and water level data in the sliding window of each cross-section to the remote server;

[0078] The flow data in the sliding window of each cross-section described in Step 1 is specifically defined as follows:

[0079] X m =[Q m (1 + τ(m - 1)), …, Q m (t + τ(m - 1)), … Q m (T + τ(m - 1))], m ∈ [1, M], t ∈ [1, T]

[0080] Among them, X m represents the flow data in the m-th cross-section sliding window; Q m (t + τ(m - 1)) represents the flow data at the t-th moment in the m-th cross-section sliding window, that is, the flow data at the (t + τ(m - 1))-th collection moment of the m-th cross-section, τ = 3 represents the flow evolution lag time between adjacent cross-sections, M = 4 represents the number of cross-sections, and T = 50 represents the length of the sliding window;

[0081] The water level data within the sliding window of each cross-section described in Step 1 are specifically defined as follows:

[0082] Y m =[Z m (1 + τ(m - 1)), …, Z m (t + τ(m - 1)), … Z m (T + τ(m - 1))], m ∈ [1, M], t ∈ [1, T]

[0083] Where Y m represents the water level data within the sliding window of the m-th cross-section; Z m (t + τ(m - 1)) represents the water level data at the t-th moment within the sliding window of the m-th cross-section, that is, the water level data at the (t + τ(m - 1))-th acquisition moment of the m-th cross-section. τ = 3 represents the flow evolution lag time between adjacent cross-sections, M = 4 represents the number of cross-sections, and T = 50 represents the length of the sliding window;

[0084] Step 2: Construct a flow evolution simulation model for each cross-section through the Transformer neural network model. Input the flow data within the sliding window of the previous cross-section into the flow evolution simulation model of this cross-section for prediction to obtain the predicted flow data within the sliding window of this cross-section. Combine the flow data within the sliding window of this cross-section to construct a loss function model, and use the genetic algorithm to optimize and train to obtain the optimized flow evolution simulation model for each cross-section;

[0085] The loss function model described in Step 2 is specifically as follows:

[0086]

[0087]

[0088] Where represents the predicted flow data within the sliding window of the m-th cross-section, that is, the predicted flow data within the sliding window of this cross-section, represents the predicted flow data at the t-th moment at the (t + τ(m - 1))-th moment within the sliding window of this cross-section, that is, the predicted flow data at the t-th moment at the (t + τ(m - 1))-th moment within the sliding window of the m-th cross-section. τ = 3 represents the flow evolution lag time between adjacent cross-sections, M = 4 represents the number of cross-sections, and T = 50 represents the length of the sliding window;

[0089] Step 3: Cascade the flow evolution simulation models of the 1st cross-section, the 2nd cross-section, …, the last M-th cross-section in sequence to construct a multi-cross-section continuous flow evolution simulation model;

[0090] Step 4: The water level data within the sliding window of the downstream ecological assessment section is used to calculate the wetted perimeter data within the sliding window of the downstream ecological assessment section through the water level sorting wetted perimeter calculation method. The curvature data within the sliding window of the downstream ecological assessment section is calculated using the discrete difference method, and the ecological flow of the downstream ecological assessment section is obtained by combining the curvature data within the sliding window of the downstream ecological assessment section.

[0091] The downstream ecological assessment section in Step 4 is the Mth section among the multiple sections in Step 1.

[0092] The water level data within the sliding window of the downstream ecological assessment section in Step 4 is specifically defined as follows:

[0093] Y M =[Z M (1 + τ(M - 1)), …, Z M (t + τ(M - 1)), … Z M (T + τ(M - 1))], t ∈ [1, T]

[0094] where Y M represents the water level data within the sliding window of the Mth section; Z M (t + τ(M - 1)) represents the water level data at the t-th moment within the sliding window of the Mth section, that is, the water level data at the (t + τ(M - 1))-th acquisition moment of the Mth section. τ = 3 represents the flow evolution lag time between adjacent sections, and T = 50 represents the length of the sliding window.

[0095] The specific process of the water level sorting wetted perimeter calculation method in Step 4 is as follows:

[0096] Sort the water level data at T moments within the sliding window of the Mth section from smallest to largest to obtain the sorted water level data within the sliding window of the Mth section.

[0097] Count the serial numbers of the water level data at each moment within the sliding window of the Mth section in the sorted water level data within the sliding window of the Mth section, and further construct the ranking data within the sliding window of the Mth section.

[0098] The ranking data within the sliding window of the Mth section is specifically defined as follows:

[0099] S M =[s M (1 + τ(M - 1)), …, s M (t + τ(M - 1)), … s M (T + τ(M - 1))], t ∈ [1, T]

[0100] where S M represents the ranking data within the sliding window of the Mth section; s M(t + τ(M - 1)) represents the data at the t-th position within the sliding window of the M-th cross-section. τ = 3 represents the flow evolution lag time between two adjacent cross-sections, and T = 50 represents the length of the sliding window;

[0101] Calculate the data at each moment within the sliding window of the M-th cross-section through the wetted perimeter calculation method to obtain the wetted perimeter data within the sliding window of the M-th cross-section. The specific definition is as follows:

[0102] R M =[r M (1 + τ(M - 1)), …, r M (t + τ(M - 1)), … r M (T + τ(M - 1))], t ∈ [1, T]

[0103] Among them, R M represents the wetted perimeter data within the sliding window of the M-th cross-section; r M (t + τ(M - 1)) represents the wetted perimeter data of the data at the t-th position within the sliding window of the M-th cross-section. τ = 3 represents the flow evolution lag time between two adjacent cross-sections, and T = 50 represents the length of the sliding window;

[0104] The wetted perimeter data within the sliding window of the M-th cross-section is the wetted perimeter data within the sliding window of the downstream ecological assessment cross-section described in step 4;

[0105] The curvature data within the sliding window of the downstream ecological assessment cross-section is calculated using the discrete difference method as described in step 4, specifically as follows:

[0106] Calculate the curvature data within the sliding window of the downstream ecological assessment cross-section by using the discrete difference method for the wetted perimeter data and the flow data within the sliding window of the downstream ecological assessment cross-section;

[0107] The flow data within the sliding window of the downstream ecological assessment cross-section is specifically defined as follows:

[0108] O M =[o M (1 + τ(M - 1)), …, o M (t + τ(M - 1)), … o M (T + τ(M - 1))], t ∈ [1, T]

[0109] Among them, O M represents the flow data within the sliding window of the M-th cross-section; o M (t + τ(M - 1)) represents the flow data corresponding to the data at the t-th position within the sliding window of the M-th cross-section. τ = 3 represents the flow evolution lag time between two adjacent cross-sections, and T = 50 represents the length of the sliding window;

[0110] The curvature data within the sliding window of the downstream ecological assessment section is specifically defined as follows:

[0111] K M =[k M (1 + τ(M - 1)),…,k M (t + τ(M - 1)),…k M (T + τ(M - 1))], t ∈ [1, T]

[0112] Among them, K M represents the curvature data within the sliding window of the Mth section; k M (t + τ(M - 1)) represents the curvature data of the t-th ranked data within the sliding window of the Mth section, τ = 3 represents the flow evolution lag time between two adjacent sections, and T = 50 represents the length of the sliding window;

[0113] The specific process of obtaining the ecological flow of the downstream ecological assessment section described in step 4 is as follows:

[0114] Among the curvature data at multiple moments within the sliding window of the downstream ecological assessment section, select the moment with the largest curvature data, and use the flow data corresponding to the moment with the largest curvature data within the sliding window of the downstream ecological assessment section as the ecological flow of the downstream ecological assessment section;

[0115] The ecological flow of the downstream ecological assessment section is defined as follows:

[0116] Q M (time + τ(M - 1))

[0117] Among them, Q M (time + τ(M - 1)) represents the ecological flow of the downstream ecological assessment section, τ represents the flow evolution lag time between two adjacent sections, and time represents the moment with the largest curvature data within the sliding window of the downstream ecological assessment section;

[0118] Step 5: By continuously discharging the flow at multiple moments from the upstream reservoir section, construct the flow data within the sliding window of the upstream reservoir section, input the flow data within the sliding window of the upstream reservoir section into the multi-section continuous flow evolution simulation model for prediction, obtain the predicted flow data within the sliding window of the downstream ecological assessment section, and further calculate the newly discharged flow of the upstream reservoir section in combination with the ecological flow of the downstream ecological assessment section;

[0119] The upstream reservoir section described in step 5 is the first section among the multiple sections described in step 1;

[0120] The predicted flow data within the sliding window of the downstream ecological assessment section described in step 5 is specifically defined as follows:

[0121]

[0122] Among them, X m p represents the flow data in the sliding window of the Mth section; Q m p (t+τ(M-1)) represents the predicted flow data at the tth time in the sliding window of the Mth section, τ=3 represents the flow evolution lag between two adjacent sections, M=4 represents the number of sections, and T=50 represents the length of the sliding window;

[0123] In step 5, the newly added downstream discharge flow of the upstream reservoir section is further calculated in combination with the ecological flow of the downstream ecological assessment section, as follows:

[0124] The predicted flow data at each moment in the sliding window of the downstream ecological assessment section are compared with the ecological flow of the downstream ecological assessment section in turn;

[0125] If the predicted flow data at each moment in the sliding window of the downstream ecological assessment section is greater than or equal to the ecological flow of the downstream ecological assessment section, the newly added downstream flow of the upstream reservoir section is 0;

[0126] If there is a moment in the downstream ecological assessment section window when the predicted flow data is less than the ecological flow of the upstream ecological assessment section, the downstream flow of the upstream reservoir section is increased by the downstream flow increment of the upstream reservoir section;

[0127] The downstream discharge increment of the upstream reservoir section is specifically calculated as follows:

[0128]

[0129] Among them, Q m p (t l +τ(M-1)) represents the tth section in the Mth section sliding window l The traffic data predicted at each moment, ΔQ(t l ) represents the tth section of the upstream reservoir l Always increase the flow rate of discharge, Q M (time+τ(M-1)) represents the ecological flow of the downstream ecological assessment section, time represents the moment when the curvature data in the sliding window of the downstream ecological assessment section is the largest, τ=3 represents the flow evolution lag between two adjacent sections, M=4 represents the number of sections, and T=50 represents the length of the sliding window.

[0130] It should be understood that parts not elaborated in detail in this specification belong to the prior art.

[0131] Although the terms such as visual perception water level flow measurement device and remote server are used more frequently in this article, the possibility of using other terms is not excluded. The use of these terms is only for more convenient description of the essence of the present invention, and interpreting them as any additional restrictions is contrary to the spirit of the present invention.

[0132] It should be understood that the above description of the preferred embodiment is relatively detailed, and it should not be considered as a limitation on the protection scope of the invention patent of the present invention. Under the inspiration of the present invention, those of ordinary skill in the art can also make substitutions or deformations without departing from the scope protected by the claims of the present invention, and all fall within the protection scope of the present invention. The scope of protection requested by the present invention shall be subject to the appended claims.

Claims

1. An ecological flow calculation system integrating visual perception and artificial intelligence, characterized in that Including: Multiple visual perception water level and flow measurement devices, and a remote server; Sequentially arranging multiple visual perception water level and flow measurement devices at multiple cross-sections; The remote server is sequentially connected to the multiple visual perception water level and flow measurement devices; The visual perception water level and flow measurement devices sequentially collect the flow and water level data within the sliding windows of each cross-section through the sliding window acquisition method, and wirelessly transmit the flow and water level data within the sliding windows of each cross-section to the remote server; construct a flow evolution simulation model for each cross-section through the Transformer neural network model, and use the genetic algorithm to optimize and train to obtain the optimized flow evolution simulation model for each cross-section; sequentially cascade the flow evolution simulation models of each cross-section to construct a multi-cross-section continuous flow evolution simulation model; The water level data within the sliding window of the downstream ecological assessment cross-section is used to obtain the wetted perimeter data within the sliding window of the downstream ecological assessment cross-section through the water level sorting wetted perimeter calculation method, and the curvature data within the sliding window of the downstream ecological assessment cross-section is calculated using the discrete difference method, and the ecological flow of the downstream ecological assessment cross-section is obtained by combining the curvature data within the sliding window of the downstream ecological assessment cross-section; Based on the multi-cross-section continuous flow evolution simulation model, obtain the predicted flow data within the sliding window of the downstream ecological assessment cross-section, and further calculate the newly released flow of the upstream reservoir cross-section in combination with the ecological flow of the downstream ecological assessment cross-section.

2. A method for calculating ecological flow by fusing visual perception and artificial intelligence using the ecological flow calculation system for fusing visual perception and artificial intelligence according to claim 1, characterized in that, Including the following steps: Step 1: The visual perception water level and flow measurement devices sequentially collect the flow and water level data within the sliding windows of each cross-section through the sliding window acquisition method, and wirelessly transmit the flow and water level data within the sliding windows of each cross-section to the remote server; Step 2: Construct a flow evolution simulation model for each cross-section through the Transformer neural network model, input the flow data within the sliding window of the previous cross-section into the flow evolution simulation model of this cross-section for prediction to obtain the predicted flow data within the sliding window of this cross-section, construct a loss function model in combination with the flow data within the sliding window of this cross-section, and use the genetic algorithm to optimize and train to obtain the optimized flow evolution simulation model for each cross-section; Step 3: Sequentially cascade the flow evolution simulation models of each cross-section to construct a multi-cross-section continuous flow evolution simulation model; Step 4: The water level data within the sliding window of the downstream ecological assessment cross-section is used to obtain the wetted perimeter data within the sliding window of the downstream ecological assessment cross-section through the water level sorting wetted perimeter calculation method, and the curvature data within the sliding window of the downstream ecological assessment cross-section is calculated using the discrete difference method, and the ecological flow of the downstream ecological assessment cross-section is obtained by combining the curvature data within the sliding window of the downstream ecological assessment cross-section; Step 5: Continuously release the flow at multiple moments from the upstream reservoir cross-section to construct the flow data within the sliding window of the upstream reservoir cross-section, input the flow data within the sliding window of the upstream reservoir cross-section into the multi-cross-section continuous flow evolution simulation model for prediction to obtain the predicted flow data within the sliding window of the downstream ecological assessment cross-section, and further calculate the newly released flow of the upstream reservoir cross-section in combination with the ecological flow of the downstream ecological assessment cross-section.

3. The ecological flow calculation method integrating visual perception and artificial intelligence according to claim 2, characterized in that: The flow data within the sliding window of each cross-section described in Step 1 is specifically defined as follows: X m = [Q m (1 + τ(m - 1)), …, Q m (t + τ(m - 1)), … Q m (T + τ(m - 1))], m ∈ [1, M], t ∈ [1, T] Among them, X m represents the flow rate data within the sliding window of the m-th cross-section; Q m (t + τ(m - 1)) represents the flow rate data at the t-th moment within the sliding window of the m-th cross-section, that is, the flow rate data at the (t + τ(m - 1))-th acquisition moment of the m-th cross-section. τ represents the flow evolution lag time between adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window; The water level data within the sliding window of each cross-section described in Step 1 is specifically defined as follows: Y m = [Z m (1 + τ(m - 1)), …, Z m (t + τ(m - 1)), … Z m (T + τ(m - 1))], m ∈ [1, M], t ∈ [1, T] Among them, Y m represents the water level data within the sliding window of the m-th cross-section; Z m (t + τ(m - 1)) represents the water level data at the t-th moment within the sliding window of the m-th cross-section, that is, the water level data at the (t + τ(m - 1))-th acquisition moment of the m-th cross-section. τ represents the flow routing lag time between two adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window.

4. The ecological flow calculation method integrating visual perception and artificial intelligence according to claim 3, characterized in that: The loss function model described in Step 2 is specifically as follows: Among them, represents the predicted flow rate data within the m-th cross-section sliding window, that is, the predicted flow rate data within the sliding window of this cross-section. represents the predicted flow rate data at the t-th moment at the (t + τ(m - 1))-th moment within the sliding window of this cross-section, that is, the predicted flow rate data at the t-th moment at the (t + τ(m - 1))-th moment within the m-th cross-section sliding window. τ represents the flow evolution lag time between two adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window.

5. The ecological flow calculation method integrating visual perception and artificial intelligence according to claim 4, characterized in that: The downstream ecological assessment cross-section described in Step 4 is the Mth cross-section among the multiple cross-sections described in Step 1; The water level data within the sliding window of the downstream ecological assessment cross-section described in Step 4 is specifically defined as follows: Y M = [Z M (1 + τ(M - 1)), …, Z M (t + τ(M - 1)), … Z M (T + τ(M - 1))], t ∈ [1, T] Among them, Y M represents the water level data within the sliding window of the M-th cross-section; Z M (t + τ(M - 1)) represents the water level data at the t-th moment within the sliding window of the M-th cross-section, that is, the water level data at the (t + τ(M - 1))-th acquisition moment of the M-th cross-section. τ represents the flow evolution lag time between two adjacent cross-sections, and T represents the length of the sliding window; The water level sorting wetted perimeter calculation method described in Step 4 is specifically as follows: Sort the water level data at T moments within the sliding window of the Mth cross-section from small to large to obtain the sorted water level data within the sliding window of the Mth cross-section; Count the serial numbers of the water level data at each moment within the sliding window of the Mth cross-section in the sorted water level data within the sliding window of the Mth cross-section, and further construct the ranking data within the sliding window of the Mth cross-section; The ranking data within the sliding window of the Mth cross-section is specifically defined as follows: S M = [s M (1 + τ(M - 1)), …, s M (t + τ(M - 1)), … s M (T + τ(M - 1))], t ∈ [1, T] Among them, S M represents the ranking data within the sliding window of the M-th cross-section; s M (t + τ(M - 1)) represents the t-th ranking data within the sliding window of the M-th cross-section, τ represents the flow routing lag time between two adjacent cross-sections, and T represents the length of the sliding window; Calculate the ranking data at each moment within the sliding window of the Mth cross-section through the wetted perimeter calculation method to obtain the wetted perimeter data within the sliding window of the Mth cross-section, which is specifically defined as follows: R M = [r M (1 + τ(M - 1)), …, r M (t + τ(M - 1)), … r M (T + τ(M - 1))], t ∈ [1, T] Among them, R M represents the wetted perimeter data within the sliding window of the M-th cross-section; r M (t + τ(M - 1)) represents the wetted perimeter data of the t-th ranked data within the sliding window of the M-th cross-section, τ represents the flow routing lag time between two adjacent cross-sections, and T represents the length of the sliding window; The wetted perimeter data within the sliding window of the Mth cross-section is the wetted perimeter data within the sliding window of the downstream ecological assessment cross-section described in Step 4.

6. The ecological flow calculation method integrating visual perception and artificial intelligence according to claim 5, characterized in that: The curvature data within the sliding window of the downstream ecological assessment cross-section calculated by using the discrete difference method described in Step 4 is specifically as follows: Calculate the wetted perimeter data within the sliding window of the downstream ecological assessment cross-section and the flow data within the sliding window of the downstream ecological assessment cross-section by using the discrete difference method to obtain the curvature data within the sliding window of the downstream ecological assessment cross-section; The flow data within the sliding window of the downstream ecological assessment cross-section is specifically defined as follows: O M = [o M (1 + τ(M - 1)), …, o M (t + τ(M - 1)), … o M (T + τ(M - 1))], t ∈ [1, T] Among them, O M represents the flow rate data within the sliding window of the M-th cross-section; o M (t + τ(M - 1)) represents the flow rate data corresponding to the t-th ranked data within the sliding window of the M-th cross-section, τ represents the flow evolution lag time between two adjacent cross-sections, and T represents the length of the sliding window; The curvature data within the sliding window of the downstream ecological assessment cross-section is specifically defined as follows: K M = [k M (1 + τ(M - 1)), …, k M (t + τ(M - 1)), … k M (T + τ(M - 1))], t ∈ [1, T] Among them, K M represents the curvature data within the sliding window of the M-th cross-section; k M (t + τ(M - 1)) represents the curvature data of the t-th ranked data within the sliding window of the M-th cross-section. τ represents the flow evolution lag time between two adjacent cross-sections, and T represents the length of the sliding window.

7. The ecological flow calculation method integrating visual perception and artificial intelligence according to claim 6, characterized in that: The process of obtaining the ecological flow of the downstream ecological assessment cross-section described in Step 4 is specifically as follows: Select the moment with the largest curvature data from the curvature data at multiple moments within the sliding window of the downstream ecological assessment cross-section, and use the flow data corresponding to the moment with the largest curvature data within the sliding window of the downstream ecological assessment cross-section as the ecological flow of the downstream ecological assessment cross-section; The ecological flow of the downstream ecological assessment cross-section is defined as follows: Q M (time + τ(M - 1)) Among them, Q M (time + τ(M - 1)) represents the ecological flow at the downstream ecological assessment section. τ represents the flow evolution lag time between two adjacent sections, and time represents the moment when the curvature data within the sliding window of the downstream ecological assessment section is the largest.

8. The ecological flow calculation method integrating visual perception and artificial intelligence according to claim 7, characterized in that: The upstream reservoir cross-section described in Step 5 is the 1st cross-section among the multiple cross-sections described in Step 1; The predicted flow data within the sliding window of the downstream ecological assessment cross-section described in Step 5 is specifically defined as follows: Among them, X m p represents the flow rate data within the M-th cross-section sliding window; Q m p (t + τ(M - 1)) represents the predicted flow rate data at the t-th moment within the M-th cross-section sliding window, τ represents the flow evolution lag time between two adjacent cross-sections, M represents the number of cross-sections, and T represents the length of the sliding window; The process of further calculating the additional discharge of the upstream reservoir cross-section in combination with the ecological flow of the downstream ecological assessment cross-section described in Step 5 is specifically as follows: The predicted flow data at each moment within the sliding window of the downstream ecological assessment section are successively compared with the ecological flow of the downstream ecological assessment section; If the predicted flow data at each moment within the sliding window of the downstream ecological assessment section are all greater than or equal to the ecological flow of the downstream ecological assessment section, the additional discharge flow of the upstream reservoir section is 0; If there is a moment when the predicted flow data within the window of the downstream ecological assessment section is less than the ecological flow of the downstream ecological assessment section, the discharge flow of the upstream reservoir section is increased through the incremental discharge flow of the upstream reservoir section; The incremental discharge flow of the upstream reservoir section is specifically calculated as follows: Among them, Q m p (t l + τ(M - 1)) represents the predicted flow data at the t l -th moment within the sliding window of the M-th cross-section. ΔQ(t l ) represents the increased discharge at the upstream reservoir cross-section at the t l -th moment. Q M (time + τ(M - 1)) represents the ecological flow at the downstream ecological assessment cross-section. time represents the moment with the maximum curvature data within the sliding window of the downstream ecological assessment cross-section. τ represents the flow evolution lag time between two adjacent cross-sections. M represents the number of cross-sections, and T represents the length of the sliding window.

Images