Dam-passing facility optimization method based on multi-target fish descending law and flow field characteristics

Abstract

The invention discloses a dam-passing facility optimization method based on a multi-target fish descending rule and flow field characteristics. The method comprises the following steps: step 1, screening fish-passing objects; 2, a fish descending test in the dam river section reservoir area is carried out; 3, constructing a fish descending efficiency evaluation system; 4, diagnosing the problem of dam-passing facilities for preventing the fishes from smoothly descending; step 5, optimizing a descending dam-passing facility based on a fish descending rule and flow field characteristics; the method can be applied to water conservancy projects with built fish passing facilities, and reference methods and strategies are provided for smooth, safe and efficient descending of fishes under the engineering background that the water conservancy facilities are built to hinder fish migration.

Description

Technical Field

[0001] The invention relates to the technical field of safe descent of young fish across a dam, and in particular to a dam descent facility optimization method based on multi-target fish descent laws and flow field characteristics. Background Art

[0002] According to the specific life history of fish migration, young fish looking for fattening places need to cross the dam and migrate downstream to find suitable fattening places, thus forming a complete life cycle. Young fish that migrate upstream usually need to mature in the downstream habitat. If the downstream channel of young fish is blocked, they may not be able to survive upstream.

[0003] There are huge numbers of young fish in my country that have the need to migrate downstream, and the interference of dams on the descent of young fish has also attracted widespread attention. In recent years, my country has focused more on the research on fish upstreaming over dams, and the research on fish descent started relatively late. The decisive factor for adult fish to migrate upstream is the fish's ability to overcome the flow, while the descent of young fish is different from that of adult fish. During the descent of young fish over the dam, their survival will be threatened by passive suction. Therefore, the key to the descent of young fish lies in having a feasible descent channel to achieve efficient and safe descent over the dam. However, young fish that are descenting over the dam often fail to locate the entrance of the descent channel, and lose their way in the complex flow after entering the channel, which prolongs the descent time. They may be caught and unable to smoothly descend to find a fattening ground. The main reasons for the inability of young fish to smoothly descend are often environmental and hydraulic factors. Therefore, how to identify the conditions that affect the environment and hydraulics is the core issue in solving the efficient descent of multi-target fish. At present, there is no evaluation system for the downstream effect based on the behavioral laws of multi-target fish. There is a lack of key parameter identification methods that affect the multi-target downstream efficiency and a prediction model for the multi-target fish downstream efficiency that couples multiple factors. There is no report on the diagnosis and optimization method of multi-target fish downstream dam facilities. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a method for optimizing dam-passing facilities based on the descent rules of multi-target fish and flow field characteristics, which is applied to water conservancy projects with built fish-passing facilities, and provides reference methods and strategies for the smooth, safe and efficient descent of multi-target fish in the context of engineering projects where water conservancy facilities are built to hinder the migration of multi-target fish.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a multi-objective dam crossing facility optimization strategy based on multi-objective fish descending rules and flow field characteristics, which includes the following steps:

[0006] Step 1: Screening of fish objects;

[0007] Step 2: Conduct fish migration test in the reservoir area of ​​the river section above the dam;

[0008] Step 3: Construct a fish descent efficiency evaluation system;

[0009] Step 4: Diagnosis of problems with dam facilities that prevent fish from moving down smoothly;

[0010] Step 5: Optimize the downstream dam facilities based on the downward movement patterns of fish and flow field characteristics.

[0011] Preferably, in step 1, the method for screening fish passing objects is: for water conservancy projects that have built fish passing facilities, according to the flow of the river section, the operating conditions of the power station and the distribution of the river terrain, the DIDSON dual-frequency identification sonar or the EK80 split-beam fish finder is used in conjunction with the drift gillnet gear to obtain the distribution characteristics of fish resources; after understanding the distribution characteristics of fish resources in the reservoir area, a screening standard for fish passing objects is established with migratory habits, habitat adaptability, protection level, economic value, and historical status as scoring standards, and the grade score of each standard is 0, 1, 2, and those with a score ≥7 are set as the main fish passing objects.

[0012] Preferably, in step 2, the method for conducting a fish descent test in the reservoir area of ​​the dam section is as follows: using a fish movement behavior tracking method, firstly, an array of NexTrakR1 acoustic receivers are arranged in the dam section near the dam section and in the inlet, outlet and channel of the fish passing facilities to form a positioning network, a GPS receiver is used to accurately record the placement position of each receiver, and then a V9 acoustic tag is implanted into the test subject. During the formal test, different locations are selected in the reservoir area of ​​the dam section to conduct a fish descent test. When the fish marked with the acoustic tag is active in the positioning network, the receiver will receive the sound wave signal emitted by the acoustic tag, and use the underwater sound wave speed and the sound wave arrival time difference to calculate the position of the fish, and record the number of the marked fish and the exact arrival time.

[0013] Preferably, in step 3, the method for constructing a multi-objective fish descending efficiency evaluation system includes the following contents:

[0014] Step 3.1: Based on the data monitored by the fish movement behavior tracking technology, obtain indicators to evaluate the descent efficiency of multi-target fish, including attempt rate, import attraction rate, descent time and descent rate;

[0015] Step 3.2: Use the entropy weight method to assign weights to the multi-target fish descent efficiency evaluation indicators;

[0016] Step 3.3: For each sample i, use the weighted summation method to calculate the comprehensive score S of the i-th sample on the four indicators i , the calculation formula is as follows:

[0017] S i =w1X i ′1+w2X i′2+w3X i ′3+w4X i '4;

[0018] According to this method, the comprehensive score of each sample is calculated to obtain the comprehensive score matrix S = [S1, S2, ..., Sn], and the comprehensive score of downlink efficiency of all samples is obtained; according to the comprehensive score S i , the fish descent efficiency of each sample is ranked, the higher the score, the better the descent efficiency, and vice versa; the calculated comprehensive score is used to evaluate the fish descent efficiency as a whole, and the evaluation level standards are from good to bad, namely excellent, good, general, and poor. If the evaluation result is general or poor, it is necessary to diagnose the problems of the dam facilities.

[0019] Preferably, the evaluation indicators in step 3.1 are defined as follows:

[0020] (1) The attempt rate is defined as: when the test fish approaches the receiver at the inlet of the downlink channel and is detected for more than 5 seconds, it is considered that the test fish has made an attempt. The calculation formula is:

[0021] Where A represents the attempt rate (%), N1 represents the number of tagged fish detected by the receiver at the inlet of the downstream channel and monitored for more than 5 seconds, and N0 represents the number of tagged fish actually released in the reservoir area of ​​the upper reaches of the dam;

[0022] (2) The import attraction rate is defined as the ratio of the number of tagged fish detected by the receiver 3 m after the import of the downlink channel to the number of tagged fish that attempted to import the downlink channel. The calculation formula is:

[0023] Where E0 represents the attraction rate of the downlink channel import (%), N2 represents the number of tagged fish monitored by the downlink channel import receiver, and N1 represents the number of tagged fish monitored by the downlink channel import receiver and monitored for more than 5s;

[0024] (3) Downlink time is defined as the time interval between the time when the receiver at the outlet of the downlink channel recognizes the tagged fish and the time when the receiver at the inlet of the downlink channel recognizes the tagged fish. In order to select a standard to quantify the downlink time of fish, an average downlink time is selected here. The time exceeding the average downlink time is regarded as downlink delay. The downlink delay rate is proposed to characterize the speed of the tagged fish in the downlink channel. Its calculation formula is:

[0025] Where N3 represents the number of tagged fish with downlink delay, and N4 represents the number of tagged fish monitored by the receiver at the exit of the downlink channel;

[0026] (4) Downstream rate is defined as the percentage of the number of tagged fish detected by the downstream exit receiver to the number of tagged fish detected by the downstream entrance receiver. Its calculation is as follows:

[0027] Where R represents the fish descending rate (%), N4 represents the number of tagged fish monitored by the receiver at the exit of the downstream channel, and N1 represents the number of tagged fish monitored by the receiver at the import of the downstream channel and the monitoring time exceeds 5s.

[0028] More preferably, the process of entropy weight assignment in step 3.2 is as follows:

[0029] (1) Construct the original data matrix X, which contains n samples and the columns corresponding to the four evaluation indicators are as follows:

[0030]

[0031] Where X ij Represents the value of the i-th sample on the j-th index;

[0032] (2) Due to the different dimensions of each indicator, the Min-Max normalization method is used to normalize the original data. The calculation formula is as follows:

[0033]

[0034] Where min(X i ) is the minimum value of the jth index, max(X j ) is the maximum value of the jth index;

[0035] The data matrix X' after standardization is as follows:

[0036]

[0037] All elements of the standardized data matrix X' will fall within the interval [0,1];

[0038] (3) Calculate the proportion p of each sample in each indicator ij , and its calculation formula is as follows:

[0039]

[0040] In the formula It is the total amount of the index after the standardization of all samples under the jth index;

[0041] (4) For each indicator j, according to the proportion matrix p ij , calculate its information entropy H j , information entropy H jIt reflects the uncertainty of the jth indicator. The larger the entropy value, the smaller the amount of information of the indicator and the smaller the difference. The calculation formula is as follows:

[0042]

[0043] In the formula is a normalization constant used to ensure that the entropy value ranges between [0,1]; when p ij = 0, define p ij ln(p ij )=0 to avoid errors in logarithmic operations;

[0044] (5) According to the entropy value H of each indicator j Calculate the weight w j , the smaller the entropy value of an indicator, the greater its weight; its calculation formula is as follows:

[0045]

[0046] In the formula, m is the total number of indicators, here m = 4; the sum of the weights of the four indicators is 1, that is, w1+w2+w3+w4=1.

[0047] Preferably, step 4 includes: the diagnosis of dam facility problems is first based on the environmental and hydraulic parameters obtained from the field test, using the random forest RF model to identify the key factors affecting the descent efficiency of multi-target fish, providing ideas for the subsequent optimization plan; and then using FLOW-3D to construct a hydrological and hydrodynamic model to find the hydrological and hydrodynamic characteristics that multi-target fish need to face during the descent process, and accurately identify the hydraulic factors that hinder the smooth descent of multi-target fish.

[0048] More preferably, the specific process of diagnosing the problem of dam crossing facilities is as follows:

[0049] Step 4.1: In order to truly reflect the descending law of fish under the synergistic effect of multiple factors in actual field projects and find the key factors affecting the descending efficiency of multi-target fish under the interaction of multiple factors, a random forest RF model is constructed based on the biological and non-biological factors affecting the descending efficiency of multi-target fish. The above factors, that is, M features, are used as the feature variable matrix X, and whether the fish successfully descends is used as the target variable y to construct a binary classification model. The feature variables and the target variables form the data set D, and the single feature forms the feature A. In order to evaluate the performance of the model, the data set needs to be divided into a training set and a test set;

[0050] Set the number of decision trees in the random forest to T, introduce sample randomness and feature randomness to train the model, and perform the following steps for each node when building a decision tree:

[0051] (1) Randomly select feature subsets: At the current node, randomly select m features (m≤M) from all M features. For classification problems: For regression problems: m = M / 3;

[0052] (2) Select the best feature and split point: Among the selected m features, select a feature and its corresponding split point n to minimize the impurity of the data set. The range of the Gini index is [0,1]. The smaller the value, the higher the purity of the data set. In the node splitting process of the decision tree, select the feature with the smallest Gini index for classification to ensure that the purity of the child node is as high as possible. The formula of the Gini index is as follows:

[0053]

[0054] In the formula, Gini(D) is the Gini index, p k is the proportion of samples of the kth class in the dataset D, and K is the total number of classes;

[0055] Entropy: It is used to quantify the uncertainty of a data set. The larger the entropy, the higher the uncertainty of the data set. The formula for entropy is:

[0056]

[0057] Where H(D) is the entropy of the data set D, p k is the sample ratio of the kth class in the data set D; the data set is divided into several subsets D according to feature A v , the entropy formula can be used to calculate the entropy H(D v );

[0058] Information gain is used to measure the degree of reduction in impurity when dividing a data set on feature A. The greater the information gain, the better the effect of this feature on dividing data. The calculation formula is as follows:

[0059]

[0060] In the formula, IG(D,A) is the information gain, H(D) is the entropy of the data set D, A is the feature, and the data set is divided into V subsets D according to the value v of feature A. v ; It is a subset D v The proportion of H(D v ) is a subset D v Entropy of

[0061] (3) Divide the data set: divide the data set of the current node into two child nodes according to the selected features and split points; recursively construct child nodes: for each child node, repeat steps (1) to (3) until the depth of the tree reaches the preset maximum depth, the number of samples in the node is less than the preset threshold, and all samples in the node belong to the same category.

[0062] Based on the feature importance of impurity reduction, the impurity reduction of each feature in each tree is recorded. The higher the feature importance value, the greater the contribution of the feature in the random forest model and the more significant the impact on the final classification or regression results. The feature importance formula is as follows:

[0063]

[0064] Where T represents the total number of decision trees in the forest; is the set of nodes using feature A in the t-th tree; ΔI(n) is the reduction in impurity when feature A is used to partition node n;

[0065] Step 4.2: In order to accurately identify the hydraulic factors that hinder the smooth descent of multi-target fish, the truVOF method is used based on FLOW-3D to accurately capture the free liquid surface, simulate the flow field of the downward channel, and use the RNG k-ε turbulence model in the Reynolds time-averaged model to consider the influence of turbulent vortices. Under low Reynolds numbers, the equations are used to analyze the hydraulic characteristics of the near-wall area, deal with the influence of turbulence, improve the accuracy of model calculations, and find the hydraulic factors that hinder the smooth descent of multi-target fish and the preference rules of the downward motion trajectory of multi-target fish.

[0066] Preferably, the step 5 is specifically:

[0067] According to the key factors and hydrological and hydrodynamic preference rules that hinder the smooth descent of multi-target fish found in step 4, the cross-dam facilities that assist the descent of multi-target fish are optimized from three perspectives: structural transformation, power station operation scheduling, and flow control. After the optimization is completed, the fish descent test in the reservoir area above the dam and the descent efficiency of multi-target fish are evaluated again;

[0068] If the evaluation result is excellent, the optimization process is completed;

[0069] If the evaluation results are fair or poor, the problem diagnosis and optimization of the dam facilities will continue until the final optimization result makes the descent efficiency of multi-target fish reach an ideal state and then the optimization will be stopped.

[0070] Beneficial effects of the invention: The optimization method for dam-crossing facilities based on the descent rules and flow field characteristics of multi-target fish proposed in the invention can be applied to water conservancy projects with built fish-crossing facilities, providing theoretical support and optimization guidance for improving the efficiency of descent of multi-target fish across the dam, and providing reference methods and strategies for the smooth, safe and efficient descent of multi-target fish in the context of engineering projects where water conservancy facilities are built to hinder fish migration; the invention can effectively reduce the negative impact of water conservancy projects on fish ecosystems, optimize the descent conditions of multi-target fish while ensuring the benefits of water projects, and achieve the coordinated development of ecological protection and water project operation, which has significant ecological value and social significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 It is a flow chart of the steps of the present invention;

[0072] Figure 2 The distribution of environmental factors that affect the downward movement of fish;

[0073] Figure 3 To provide importance scores for key factors affecting the descent of multiple target fishes;

[0074] Figure 4 It is the flow field characteristics of the downlink facility. DETAILED DESCRIPTION

[0075] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0076] Embodiment 1: This embodiment discloses an optimization strategy for dam-crossing facilities based on the multi-target descent law of fish and flow field characteristics. It is a water conservancy project based on the built fish-crossing facilities. It uses fish movement behavior tracking technology to quantify the attempt rate, import attraction rate, descent time and descent rate of multi-target fish in the descent process, obtain the behavior law of multi-target fish completing the descent behavior from the river section above the dam, evaluate through the model, and then diagnose the problems of dam-crossing facilities in combination with the multi-target fish descent behavior law and the flow field characteristics of the descent channel, find the hydraulic factors that are not conducive to the descent of multi-target fish, and carry out optimization work. The specific steps are as follows:

[0077] Step 1: Screening of fish objects;

[0078] For water conservancy projects with built fish-passing facilities, we first select appropriate technical means to conduct fish resource surveys; then, after understanding the distribution characteristics of fish resources in the reservoir area, we establish a screening standard for fish-passing objects based on migratory habits, habitat adaptability, protection level, economic value, and historical status. The grade score for each standard is 0, 1, or 2 (Table 1), and those with a score ≥7 are set as the main fish-passing objects.

[0079] Table 1 Selection criteria for fish objects

[0080]

[0081] Step 2: Conduct fish migration test in the reservoir area of ​​the river section above the dam;

[0082] Using fish movement behavior tracking technology, an array of NexTrakR1 acoustic receivers were first arranged in the river section above the dam, near the dam, and in the inlet, outlet and channel of the fish passing facilities to form a positioning network. The V9 acoustic tag was implanted in the test subjects. During the formal test, the tagged fish were released at different locations in the reservoir area of ​​the river section above the dam. When the fish marked with the acoustic tag moved in the positioning network, the receiver would receive the sound wave signal emitted by the acoustic tag and use the underwater sound wave speed and the sound wave arrival time difference to calculate the fish's position. The system would also record the tagged fish number and the exact arrival time.

[0083] Step 3: Construct a multi-objective fish descent efficiency evaluation system;

[0084] Step 3.1: Based on the data monitored by the multi-target fish movement behavior tracking technology, indicators for evaluating the descent efficiency of multi-target fish are obtained, including the attempt rate, the import attraction rate, the descent time and the descent rate; the definitions of each evaluation indicator are as follows:

[0085] (1) The attempt rate is defined as: when the test fish approaches the receiver at the inlet of the downlink channel and is detected for more than 5 seconds, it is considered that the test fish has made an attempt. The calculation is as follows:

[0086] Where A represents the attempt rate (%), N1 represents the number of tagged fish detected by the receiver at the inlet of the downstream channel and monitored for more than 5 seconds, and N0 represents the number of tagged fish actually released in the reservoir area of ​​the river section above the dam.

[0087] (2) The import attraction rate is defined as the ratio of the number of tagged fish detected by the receiver 3 m after the import of the downlink channel to the number of tagged fish that attempted to import the downlink channel. The calculation formula is:

[0088] Where E0 represents the attraction rate of the downlink channel inlet (%), N2 represents the number of tagged fish monitored by the receiver 3 m after the downlink channel inlet, and N1 represents the number of tagged fish monitored by the receiver at the downlink channel inlet and the monitoring time exceeds 5 s.

[0089] (3) Downstream time is defined as the time interval between the time when the receiver at the outlet of the downstream channel recognizes the tagged fish and the time when the receiver at the inlet of the downstream channel recognizes the tagged fish. Since the time for fish to reach the outlet of the downstream channel from the river section above the dam varies and is unevenly distributed, an average downstream time is selected. Those exceeding the average downstream time are considered to be delayed downstream. The delayed downstream rate is proposed to characterize the speed of the tagged fish in the downstream channel. Its calculation formula is:

[0090] Where N3 represents the number of tagged fish delayed by the downlink, and N4 represents the number of tagged fish monitored by the receiver at the exit of the downlink channel.

[0091] (4) Downstream rate is defined as the percentage of the number of tagged fish detected by the downstream channel exit receiver to the number of tagged fish detected by the downstream channel entrance receiver. Its calculation is as follows:

[0092] Where R represents the fish descending rate (%), N4 represents the number of tagged fish monitored by the receiver at the exit of the downstream channel, and N1 represents the number of tagged fish monitored by the receiver at the import of the downstream channel and the monitoring time exceeds 5s.

[0093] Step 3.2: Use the entropy weight method to assign weights to the multi-target fish descent efficiency evaluation indicators. The entropy weight method weighting process is as follows:

[0094] (1) Construct the original data matrix X, which contains n samples and the columns corresponding to the four indicators (attempt rate, import attraction rate, downtime, and downtime rate) are as follows:

[0095]

[0096] Where X ij Represents the value of the i-th sample on the j-th index.

[0097] (2) Due to the different dimensions of each indicator, the Min-Max normalization method is used to normalize the original data. The calculation formula is as follows:

[0098]

[0099] Where min(X i ) is the minimum value of the jth index, max(X j ) is the maximum value of the j-th index.

[0100] The data matrix X' after standardization is as follows:

[0101]

[0102] All elements of the normalized data matrix X' will fall within the interval [0,1].

[0103] (3) Calculate the proportion p of each sample in each indicator ij , and its calculation formula is as follows:

[0104]

[0105] In the formula It is the total amount of the indicator after the standardization of all samples under the jth indicator.

[0106] (4) For each indicator j, according to the proportion matrix p ij , calculate its information entropy H j , information entropy H j It reflects the uncertainty of the jth indicator. The larger the entropy value, the smaller the amount of information of the indicator and the smaller the difference. The calculation formula is as follows:

[0107]

[0108] In the formula is a normalization constant used to ensure that the entropy value ranges between [0,1]; when p ij = 0, define p ij ln(p ij )=0 to avoid errors in logarithmic operations.

[0109] (5) According to the entropy value H of each indicator j Calculate the weight w j , the smaller the entropy value of an indicator, the greater its weight. The calculation formula is as follows:

[0110]

[0111] In the formula, m is the total number of indicators, here m = 4; the sum of the weights of the four indicators is 1, that is, w1+w2+w3+w4=1.

[0112] Step 3.3: For each sample i, use the weighted summation method to calculate the comprehensive score S of the i-th sample on the four indicators i , and its calculation formula is as follows:

[0113] S i =w1X i ′1+w2X i ′2+w3X i ′3+w4X i '4

[0114] According to this method, the comprehensive score of each sample is calculated to obtain the comprehensive score matrix S = [S1, S2, ..., Sn], and the comprehensive score of downlink efficiency of all samples is obtained. i , the fish descent efficiency of each sample is ranked, and the higher the score, the better the descent efficiency, and vice versa. The fish descent efficiency is evaluated overall by the calculated comprehensive score, and the evaluation level standard is excellent, good, general, and poor (Table 2). If the evaluation result is general or poor, it is necessary to diagnose the problem of the dam facilities.

[0115] Table 2 Evaluation results of fish descent efficiency

[0116]

[0117] Step 4: The above-mentioned diagnosis of the dam facility problem is to first use the random forest RF model to identify the key factors affecting the descent efficiency of multi-target fish based on the environmental and hydraulic parameters obtained from the field test, and provide ideas for the subsequent optimization plan; then use FLOW-3D to build a hydrological and hydrodynamic model to find the hydrological and hydrodynamic characteristics that fish need to face during the descent process, and accurately identify the hydraulic factors that hinder the smooth descent of fish. The specific diagnostic method is as follows:

[0118] Step 4.1: In order to truly reflect the downward movement of fish under the synergistic effect of multiple factors in actual field projects, and to find the key factors that affect the downward movement efficiency of multi-target fish under the interaction of multiple factors, a random forest RF model is constructed based on biological factors (fish species, fish physiological indicators, etc.) and non-biological factors (flow, dissolved oxygen, flow rate, water temperature, pH value, light, bottom quality, water transparency, circadian rhythm and other environmental factors) that affect the downward movement efficiency of multi-target fish. The above factors, that is, M features, are used as the feature variable matrix X, and whether the fish successfully descends is used as the target variable y to construct a binary classification model. The feature variables and the target variables form the data set D, and the single feature forms the feature A. In order to evaluate the performance of the model, the data set needs to be divided into a training set and a test set. The ratio of the training set to the test set is generally 8:2 or 7:3.

[0119] Set the number of decision trees in the random forest, T. Generally, the more trees there are, the more stable the model performance is, but the computational cost is also higher. Introduce sample randomness (bootstrap sampling) and feature randomness (randomly select features when splitting nodes) to train the model. When building a decision tree, perform the following steps for each node:

[0120] (1) Randomly select feature subsets: At the current node, randomly select m features (m≤M) from all M features. Usually: For classification problems: For regression problems: m = M / 3 (take the integer part).

[0121] (2) Select the best feature and split point: Among the selected m features, select a feature and its corresponding split point n to minimize the impurity of the data set (such as the Gini index, entropy). The range of the Gini index is [0,1]. The smaller the value, the higher the purity of the data set. In the process of node splitting in the decision tree, the feature with the smallest Gini index is usually selected for classification to ensure that the purity of the child node is as high as possible. The formula of the Gini index is as follows:

[0122]

[0123] In the formula, Gini(D) is the Gini index, p k is the proportion of samples of the kth class in the dataset D, and K is the total number of classes.

[0124] Entropy: It is used to quantify the uncertainty of a data set. The larger the entropy, the higher the uncertainty of the data set. The formula for entropy is:

[0125]

[0126] Where H(D) is the entropy of the data set D, p k is the sample ratio of the kth class in the dataset D. According to feature A, the dataset is divided into several subsets D v , the entropy formula can be used to calculate the entropy H(D v ).

[0127] Information gain is used to measure the degree to which impurity is reduced when the data set is divided on feature A. The greater the information gain, the better the effect of the feature in dividing the data. The calculation formula is as follows:

[0128]

[0129] In the formula, IG(D,A) is the information gain, H(D) is the entropy of the data set D, A is the feature, and the data set is divided into V subsets D according to the value v of feature A. v ; It is a subset D v The proportion of H(D v ) is a subset D v The entropy of .

[0130] (3) Divide the data set: divide the data set of the current node into two child nodes according to the selected features and split points; recursively construct child nodes: for each child node, repeat steps (1) to (3) until the depth of the tree reaches the preset maximum depth, the number of samples in the node is less than the preset threshold, and all samples in the node belong to the same category (for classification problems), which is the stopping condition.

[0131] Based on the feature importance of impurity reduction, the impurity reduction of each feature in each tree is recorded. The higher the feature importance value, the greater the contribution of the feature in the random forest model and the more significant the impact on the final classification or regression results. The feature importance formula is as follows:

[0132]

[0133] Where T represents the total number of decision trees in the forest; is the set of nodes using feature A in the t-th tree; ΔI(n) is the reduction in impurity when feature A is used to partition node n.

[0134] Step 4.2: In order to accurately identify the hydraulic factors that hinder the smooth descent of multi-target fish, the truVOF method is used based on FLOW-3D to accurately capture the free liquid surface, simulate the flow field of the downward channel, and use the RNG k-ε turbulence model in the Reynolds time-averaged model to consider the influence of turbulent vortices. Under low Reynolds number conditions, the equation can more accurately analyze the hydraulic characteristics of the near-wall area, better deal with the influence of turbulence, and significantly improve the accuracy of model calculations, in order to more accurately find the hydraulic factors that hinder the smooth descent of multi-target fish and the preference rules of the downward motion trajectory of multi-target fish.

[0135] Step 5: Based on the key factors and hydrological and hydrodynamic preference rules that hinder the smooth descent of multi-target fish found in step 4, the dam-crossing facilities that assist the descent of multi-target fish are optimized from three perspectives: structural transformation, power station operation scheduling, and flow control. After the optimization is completed, the fish descent test in the reservoir area of ​​the river section above the dam and the descent efficiency of multi-target fish are evaluated again.

[0136] If the evaluation result is excellent, the optimization process is completed;

[0137] If the evaluation result is average or poor, the problem diagnosis and optimization of the dam facilities will continue until the final optimization result makes the fish descent efficiency reach the ideal state and then the optimization will be stopped.

[0138] Embodiment 2: This embodiment relies on a fishway project, with the heavy-mouthed schizothorax, Songpan naked carp, Qikou schizothorax and Xichang Chinese loach as the main fish passing objects. Now a calculation example is provided to specifically illustrate the calculation method and process involved in the present invention, and the process is as follows:

[0139] The present invention relies on a fishway project and counts a total of 2,800 fishes through a fish resource survey, belonging to 2 orders, 3 families and 20 species. The 20 species of fish are scored as fish-passing objects, and according to the scoring criteria, the heavy-mouthed schizothorax (score 7), Songpan naked carp (score 8), Qikou schizothorax (score 7), and Xichang Chinese loach (score 7) with a total score ≥7 are selected as the main fish-passing objects.

[0140] For the selected fish, multi-target fish descent tests were conducted at three fish release locations in the reservoir area of ​​the upper dam section. The water temperature in the upstream river was 18.5℃ and the flow rate in the upstream river was 22.0m 3 / s, the fish movement behavior tracking technology was used to monitor the fish descent efficiency, and the various indicators of the four fish species were shown in Table 3:

[0141] Table 3 Values ​​of various indicators of test fish

[0142]

[0143] According to the above result data, construct the evaluation matrix X:

[0144]

[0145] Using the minimax normalization method, each column of data is standardized to obtain the standardized matrix Y:

[0146]

[0147] According to the formula, the weight of each sample under each indicator is calculated, and then the information entropy of each indicator is calculated. The information entropy of the four indicators finally obtained in this example is [0.676 0.654 0.500 0.708], and the weight of each indicator is [0.2220.237 0.342 0.200]. The final calculation of the comprehensive score S i , for the heavy-mouthed schizothorax, S1 = 0.222*1 + 0.237*0.297 + 0.342*1 + 0.200*0.323 = 0.70; the comprehensive scores of the four fish species were calculated according to the above formula as shown in Table 4:

[0148] Table 4 Comprehensive score of entropy weight method

[0149]

[0150] According to the established evaluation criteria, the descending effect of the Qikou schizothorax is excellent, the descending effect of the heavy-mouthed schizothorax is good, and the descending effect of the Songpan naked carp and the Xichang Chinese loach is poor, and their descending efficiency needs to be further improved.

[0151] Based on the above evaluation results, the descent efficiency of Songpan naked carp and Xichang Chinese loach is poor, so it is necessary to further determine what is preventing the two fish species from descenting smoothly. Therefore, according to the environmental and hydraulic parameters obtained in the field experiment ( Figure 2 ), using fish species, multi-target fish length, fish release location, water transparency, flow rate, and water temperature as feature variables, and whether the fish successfully descended as the target variable y to build a random forest RF binary classification model, and get the importance ranking of each feature in this case. The results show that: fish release location 39.56>flow rate 37.59>water temperature 30.25>water transparency 22.86>fish species 16.48>multi-target fish length 14.99( Figure 3 ).

[0152] Then FLOW-3D was used to simulate the flow field of the downwelling channel in this case, and the downwelling motion trajectories of multiple target fish were coupled with the flow field characteristics ( Figure 4) was used to explore the flow field characteristics inside the downlink facility. The results showed that the movement trajectories of the four fish species were disordered in the inlet and turning sections of the downlink channel, and the fish would stay in the inlet and turning sections for a long time. According to the flow field characteristics of the inlet and turning sections of the downlink facility, it can be found that the flow velocity and turbulent energy in most areas of the inlet and turning sections are too large, which hinders the smooth descent of the fish. When the fish pass through this area during the downward process, they may become lost. In order to maintain a stable swimming posture in this area, the descending fish will instinctively swim against the current, which may cause fatigue, resulting in a large amount of lactic acid accumulation in the fish body, thereby causing anaerobic exercise.

[0153] The above method found the key factors that affect the descent of fish (fish release position 39.56>flow 37.59>water temperature 30.25>water transparency 22.86>fish species 16.48>multi-target fish body length 14.99), and also found that when fish descend to the entrance and turning sections of the channel, they will be hindered by the high flow rate and high turbulent energy, thus delaying their descent. Based on the model results, the idea of ​​the subsequent optimization plan was obtained;

[0154] In order to improve the descent efficiency of fish, this case is based on the model results and combined with the flow field characteristics corresponding to the motion trajectories of multiple target fish. Finally, it is proposed to structurally transform the fish descent channel, change the structure of the inlet section to make the inlet section straighter, and add obstacles at the bottom of the turning section to create a flow state with low flow rate and low turbulent kinetic energy. Secondly, the fish release position is changed, and the downstream flow of the upstream power station is changed to change the high flow of the river.

[0155] For the optimized scheme, the fish release test and fish descending efficiency evaluation were carried out again. The results showed that after optimization, the various indicators of the four fish species were as shown in Table 5 below:

[0156] Table 5 Values ​​of various indicators of test fish

[0157]

[0158] According to the index results, the entropy weight method was used to evaluate the descent efficiency of multi-target fish, and the comprehensive scores of the four optimized fish species were shown in Table 6:

[0159] Table 6 Comprehensive score of entropy weight method

[0160]

[0161] According to the established evaluation standards, the descending efficiency of the above four types of fish is good, achieving the ideal fish passing efficiency and completing the optimization work.

[0162] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A multi-objective dam crossing facility optimization strategy based on multi-objective fish descending rules and flow field characteristics, characterized by: It includes the following steps: Step 1: Screening of fish objects; Step 2: Conduct fish migration test in the reservoir area of ​​the river section above the dam; Step 3: Construct a fish descent efficiency evaluation system; Step 4: Diagnosis of problems with dam facilities that prevent fish from moving down smoothly; Step 5: Optimize the downstream dam facilities based on the downward movement patterns of fish and flow field characteristics.

2. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 1 is characterized in that: In step 1, the method for screening fish-passing objects is as follows: for water conservancy projects with built fish-passing facilities, according to the flow of the river section, the operating conditions of the power station and the distribution of the river terrain, the DIDSON dual-frequency identification sonar or the EK80 split-beam fish finder is used in conjunction with drift gillnet gear to obtain the distribution characteristics of fish resources; after understanding the distribution characteristics of fish resources in the reservoir area, a screening standard for fish-passing objects is established with migratory habits, habitat adaptability, protection level, economic value and historical status as scoring criteria. The grade score of each standard is 0, 1, and 2, and those with a score ≥ 7 are set as the main fish-passing objects.

3. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 1 is characterized in that: In step 2, the method for the fish descent test in the reservoir area of ​​the Bashang River section is as follows: using the fish movement behavior tracking method, firstly, an array of NexTrakR1 acoustic receivers are arranged in the near-dam section of the Bashang River section and in the inlet, outlet and channel of the fish passing facilities to form a positioning network, and the placement position of each receiver is accurately recorded using a GPS receiver. Subsequently, the V9 acoustic tag is implanted into the test subject. During the formal test, different locations are selected in the reservoir area of ​​the Bashang River section to conduct fish descent tests. When the fish marked with the acoustic tag are active in the positioning network, the receiver will receive the sound wave signal emitted by the acoustic tag, and use the underwater sound wave speed and the sound wave arrival time difference to solve the position of the fish, and record the number of the marked fish and the precise arrival time.

4. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 1 is characterized in that: In step 3, the method of constructing a multi-objective fish descent efficiency evaluation system includes the following contents: Step 3.1: Based on the data monitored by the fish movement behavior tracking technology, obtain indicators to evaluate the descent efficiency of multi-target fish, including attempt rate, import attraction rate, descent time and descent rate; Step 3.2: Use the entropy weight method to assign weights to the multi-target fish descent efficiency evaluation indicators; Step 3.3: For each sample i, use the weighted summation method to calculate the comprehensive score S of the i-th sample on the four indicators i , the calculation formula is as follows: S i =w1X i ′1+w2X i ′2+w3X i ′3+w4X i ′4; According to this method, the comprehensive score of each sample is calculated to obtain the comprehensive score matrix S = [S1, S2, ..., Sn], and the comprehensive score of downlink efficiency of all samples is obtained; according to the comprehensive score S i , the fish descent efficiency of each sample is ranked, the higher the score, the better the descent efficiency, and vice versa; the calculated comprehensive score is used to evaluate the fish descent efficiency as a whole, and the evaluation level standards are from good to bad, namely excellent, good, general, and poor. If the evaluation result is general or poor, it is necessary to diagnose the problems of the dam facilities.

5. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 4 is characterized in that: The evaluation indicators in step 3.1 are defined as follows: (1) The attempt rate is defined as: when the test fish approaches the receiver at the inlet of the downlink channel and is detected for more than 5 seconds, it is considered that the test fish has attempted behavior. The calculation formula is: Where A represents the attempt rate (%), N1 represents the number of tagged fish detected by the receiver at the inlet of the downstream channel and monitored for more than 5 seconds, and N0 represents the number of tagged fish actually released in the reservoir area of ​​the upper reaches of the dam; (2) The import attraction rate is defined as the ratio of the number of tagged fish detected by the receiver 3 m after the import of the downlink channel to the number of tagged fish that attempted to import the downlink channel. The calculation formula is: Where E0 represents the attraction rate of the downlink channel import (%), N2 represents the number of tagged fish monitored by the downlink channel import receiver, and N1 represents the number of tagged fish monitored by the downlink channel import receiver and monitored for more than 5s; (3) Downlink time is defined as the time interval between the time when the receiver at the outlet of the downlink channel recognizes the tagged fish and the time when the receiver at the inlet of the downlink channel recognizes the tagged fish. In order to select a standard to quantify the downlink time of fish, an average downlink time is selected here. The time exceeding the average downlink time is regarded as downlink delay. The downlink delay rate is proposed to characterize the speed of the tagged fish in the downlink channel. Its calculation formula is: Where N3 represents the number of tagged fish with downlink delay, and N4 represents the number of tagged fish monitored by the receiver at the exit of the downlink channel; (4) Downstream rate is defined as the percentage of the number of tagged fish detected by the downstream exit receiver to the number of tagged fish detected by the downstream entrance receiver. Its calculation is as follows: Where R represents the fish descending rate (%), N4 represents the number of tagged fish monitored by the receiver at the exit of the downstream channel, and N1 represents the number of tagged fish monitored by the receiver at the import of the downstream channel and the monitoring time exceeds 5s.

6. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 4 is characterized in that: The process of entropy weight assignment in step 3.2 is as follows: (1) Construct the original data matrix X, which contains n samples and the columns corresponding to the four evaluation indicators are as follows: Where X ij Represents the value of the i-th sample on the j-th index; (2) Due to the different dimensions of each indicator, the Min-Max normalization method is used to normalize the original data. The calculation formula is as follows: Where min(X i ) is the minimum value of the jth index, max(X j ) is the maximum value of the jth index; The data matrix X' after standardization is as follows: All elements of the standardized data matrix X' will fall within the interval [0,1]; (3) Calculate the proportion p of each sample in each indicator ij , and its calculation formula is as follows: In the formula It is the total amount of the index after the standardization of all samples under the jth index; (4) For each indicator j, according to the proportion matrix p ij , calculate its information entropy H j , information entropy H j It reflects the uncertainty of the jth indicator. The larger the entropy value, the smaller the amount of information of the indicator and the smaller the difference. The calculation formula is as follows: In the formula is a normalization constant used to ensure that the entropy value ranges between [0,1]; when p ij = 0, define p ij ln(p ij )=0 to avoid errors in logarithmic operations; (5) According to the entropy value H of each indicator j Calculate the weight w j , the smaller the entropy value of an indicator, the greater its weight; its calculation formula is as follows: In the formula, m is the total number of indicators, here m = 4; the sum of the weights of the four indicators is 1, that is, w1+w2+w3+w4=1.

7. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 1 is characterized in that: The step 4 includes: the diagnosis of dam facility problems is to first use the random forest RF model to identify the key factors affecting the descent efficiency of multi-target fish based on the environmental and hydraulic parameters obtained from the field test, so as to provide ideas for the subsequent optimization plan; then use FLOW-3D to build a hydrological and hydrodynamic model to find the hydrological and hydrodynamic characteristics that multi-target fish need to face during their descent, and accurately identify the hydraulic factors that hinder the smooth descent of multi-target fish.

8. The method for optimizing dam crossing facilities based on the multi-target fish descending rules and flow field characteristics according to claim 7 is characterized in that: The specific process of diagnosing dam facility problems is as follows: Step 4.1: In order to truly reflect the descending law of fish under the synergistic effect of multiple factors in actual field projects and find the key factors affecting the descending efficiency of multi-target fish under the interaction of multiple factors, a random forest RF model is constructed based on the biological and non-biological factors affecting the descending efficiency of multi-target fish. The above factors, that is, M features, are used as the feature variable matrix X, and whether the fish successfully descends is used as the target variable y to construct a binary classification model. The feature variables and the target variables form the data set D, and the single feature forms the feature A. In order to evaluate the performance of the model, the data set needs to be divided into a training set and a test set; Set the number of decision trees in the random forest to T, introduce sample randomness and feature randomness to train the model, and perform the following steps for each node when building a decision tree: (1) Randomly select feature subsets: At the current node, randomly select m features (m≤M) from all M features. For classification problems: For regression problems: m = M / 3; (2) Select the best feature and split point: Among the selected m features, select a feature and its corresponding split point n to minimize the impurity of the data set. The range of the Gini index is [0,1]. The smaller the value, the higher the purity of the data set. In the node splitting process of the decision tree, select the feature with the smallest Gini index for classification to ensure that the purity of the child node is as high as possible. The formula of the Gini index is as follows: In the formula, Gini(D) is the Gini index, p k is the proportion of samples of the kth class in the dataset D, and K is the total number of classes; Entropy: It is used to quantify the uncertainty of a data set. The larger the entropy, the higher the uncertainty of the data set. The formula for entropy is: Where H(D) is the entropy of the data set D, p k is the proportion of samples of the kth class in the data set D; Divide the data set into several subsets D according to feature A v , the entropy formula can be used to calculate the entropy H(D v ); Information gain is used to measure the degree of reduction in impurity when dividing a data set on feature A. The greater the information gain, the better the effect of this feature on dividing data. The calculation formula is as follows: In the formula, IG(D,A) is the information gain, H(D) is the entropy of the data set D, A is the feature, and the data set is divided into V subsets D according to the value v of feature A. v ; It is a subset D v The proportion of H(D v ) is a subset D v Entropy of (3) Divide the data set: divide the data set of the current node into two child nodes according to the selected features and split points; recursively construct child nodes: for each child node, repeat steps (1) to (3) until the depth of the tree reaches the preset maximum depth, the number of samples in the node is less than the preset threshold, and all samples in the node belong to the same category. Based on the feature importance of impurity reduction, the impurity reduction of each feature in each tree is recorded. The higher the feature importance value, the greater the contribution of the feature in the random forest model and the more significant the impact on the final classification or regression results. The importance formula is as follows: Where T represents the total number of decision trees in the forest; is the set of nodes using feature A in the t-th tree; ΔI(n) is the reduction in impurity when feature A is used to partition node n; Step 4.2: In order to accurately identify the hydraulic factors that hinder the smooth descent of multi-target fish, the truVOF method is used based on FLOW-3D to accurately capture the free liquid surface, simulate the flow field of the downward channel, and use the RNGk-ε turbulence model in the Reynolds time-averaged model to consider the influence of turbulent vortices. Under low Reynolds numbers, the equations are used to analyze the hydraulic characteristics of the near-wall area, deal with the influence of turbulence, improve the accuracy of model calculations, and find the hydraulic factors that hinder the smooth descent of multi-target fish and the preference rules of the downward motion trajectory of multi-target fish.

9. The method for optimizing dam crossing facilities based on multi-target fish descending rules and flow field characteristics according to claim 1 is characterized in that: The step 5 is specifically as follows: According to the key factors and hydrological and hydrodynamic preference rules that hinder the smooth descent of multi-target fish found in step 4, the cross-dam facilities that assist the descent of multi-target fish are optimized from three perspectives: structural transformation, power station operation scheduling, and flow control. After the optimization is completed, the fish descent test in the reservoir area above the dam and the descent efficiency of multi-target fish are evaluated again; If the evaluation result is excellent, the optimization process is completed; If the evaluation results are fair or poor, the problem diagnosis and optimization of the dam facilities will continue until the final optimization result makes the descent efficiency of multi-target fish reach an ideal state and then the optimization will be stopped.

Images