A method for evaluating and predicting fish blocking effect of trash rack at intake of low-head power station

Abstract

The application discloses a method for evaluating and predicting fish blocking effect of a trash rack at an intake of a low-head power station, which comprises the following steps: step 1, monitoring fish cluster distribution near the trash rack; step 2, identifying fish acoustic and video monitoring behaviors; step 3, constructing a fish blocking effect evaluation index system; step 4, establishing a fish blocking effect evaluation model; step 5, identifying key influence factors of the fish blocking effect; and step 6, simulating and predicting the fish blocking effect of the trash rack. Based on fusion of acoustic image tracking and video recognition technologies, the fish activity behavior data before and after the trash rack are acquired, a fish blocking effect evaluation model coupled with fish behaviors and hydrodynamic environmental factors is constructed, the fish blocking probability and passing behavior characteristics of the trash rack can be quantitatively evaluated, and scientific decision-making bases for structure optimization and ecological friendly operation of the trash rack at the intake of the low-head power station are provided.

Description

Technical Field

[0001] This invention belongs to the field of ecological water conservancy engineering and fish protection technology, and specifically relates to a method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station. Background Technology

[0002] Low-head power stations are numerous and widely distributed. Their intake trash racks, as common components of hydraulic structures, primarily function to intercept floating debris and other contaminants to protect the safe operation of the generating units. Influenced by the induced flow at the intake and local hydrodynamic structures, fish may gather, avoid, or linger in front of the trash racks, or, under certain operating conditions, risk accidentally entering or being sucked in, resulting in fish damage, hindered population replenishment, and the spread of ecological impacts. To reduce the risk of fish accidentally entering the turbines while ensuring the safe operation of the power station, engineering practices often involve adjusting the trash rack spacing, inclination angle, cleaning methods, and unit operating conditions to improve fish interception and anti-sucking capabilities. However, the "fish interception effect" is influenced by multiple factors, including fish biological characteristics, spatiotemporal distribution, hydrodynamic environment, unit operating conditions, and structural parameters, exhibiting significant randomness and nonlinearity.

[0003] In existing technologies, the evaluation of the fish-blocking effect of debris barriers often relies on field experience or statistical results from single monitoring methods, which frequently suffers from the following shortcomings: First, there is a lack of a fusion mechanism for multi-source monitoring data such as acoustics and video, making it difficult to continuously and objectively acquire key behavioral events of fish in the vicinity of the barrier. Second, evaluation indicators are often based solely on "whether they pass" or "the number of fish passing through the barrier," lacking an indicator system that can be used for structural optimization and scheduling decisions, such as the probability of reaching the barrier, avoidance behavior, dwell time, and risk of contact with the barrier. Third, there is a lack of probabilistic models that couple fish biological factors with hydrodynamic environmental factors, making it difficult to achieve comparable and reproducible quantitative predictions under different operating scenarios and structural parameters. Fourth, model selection and key factor identification often adopt simple goodness-of-fit criteria, making it difficult to balance predictive ability and uncertainty expression, resulting in insufficient credibility of engineering decisions.

[0004] Therefore, there is an urgent need to propose a method for evaluating and predicting the fish-blocking effect of trash racks at the intake of low-head power stations, which can integrate multi-source monitoring data, establish an indicator system, and conduct scenario prediction, so as to achieve quantitative evaluation of the fish-blocking effect of trash racks, identification of key influencing factors, and support for structural and operational optimization. Summary of the Invention

[0005] The purpose of this invention is to provide a method for evaluating and predicting the fish-blocking effect of trash racks at the intake of low-head power stations. This method uses acoustic and video monitoring to obtain the distribution and behavioral events of fish near the trash rack, constructs an evaluation index system for the fish-blocking effect, establishes a probabilistic model of the fish-blocking effect that couples fish biological factors and hydrodynamic environmental factors, and simulates and predicts the fish-blocking effect based on model selection and key factor identification. This provides a scientific basis for optimizing the structural parameters of the trash rack and for eco-friendly operation scheduling.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for evaluating and predicting the fish-blocking effect of trash racks at the intake of low-head power stations includes the following steps: Step 1: Monitoring the distribution of fish schools near the debris barrier; Step 2: Fish acoustic and video monitoring behavior recognition; Step 3: Construct an evaluation index system for the effectiveness of fish blocking; Step 4: Establish a fish-blocking effect evaluation model; Step 5: Identify the key factors affecting the effectiveness of fish blocking; Step 6: Simulation and prediction of the fish-blocking effect of the debris barrier.

[0007] Preferably, step 1 is performed as follows: Fish detection equipment was deployed in the waters upstream of the trash rack at the power station's intake and in the adjacent waters. Portable fish finders were used to monitor the trash rack area around the clock and across the entire cross section to obtain information on fish population density, body length distribution, and spatial aggregation areas. At the same time, multi-parameter water quality monitors and portable flow meters were used to collect environmental and hydrodynamic data such as water temperature, dissolved oxygen, turbidity, flow velocity, reservoir water level, and unit operating conditions, thereby identifying key fish activity areas and forming a monitoring dataset.

[0008] Preferably, in step 2, dual-frequency identification sonar and underwater video equipment are deployed in the key activity areas identified in step 1. The target detection, body length inversion and trajectory tracking methods of acoustic images and video images are used to identify and continuously track fish moving targets. The fish crossing behavior is classified into at least three categories: approaching, staying, avoiding, attempting to cross the fence, successfully crossing the fence, and risk events of touching the fence and being pinched, and the time and spatial location of their occurrence are recorded.

[0009] Preferably, in step 3, the constructed fish-blocking effect evaluation index system includes at least four or more of the following: gate arrival rate, passage rate, blocking rate, avoidance rate, dwell time, and gate contact event rate, and supports grouped statistics by fish species and body length grade. Preferably, the evaluation indicators are defined as follows: (1) Reach rate, defined as the proportion of fish targets entering the evaluation and monitoring area that reach the near-grid area of ​​the debris barrier:

[0010] In the formula, To gate rate; The number of fish targets that have entered the evaluation and monitoring area and have been identified and tracked; The target number of fish reaching the vicinity of the trash rack; (2) Pass-through rate, defined as the proportion of fish targets that successfully pass through the debris barrier and appear on the downstream side of the near-barrier area:

[0011] In the formula, For the pass rate; The target number of fish that successfully pass through the gate; (3) Barrier rate, defined as the proportion of fish targets reaching the near-gate area that do not pass through the debris barrier:

[0012] In the formula, For blocking rate; (4) Avoidance rate, defined as the proportion of fish targets that turn away or move away from the debris barrier after reaching the near-barrier area:

[0013] In the formula, For avoidance rate; To avoid the number of fish targets corresponding to the event; (5) Dwell time, defined as the average dwell time of fish targets in the near-gate area:

[0014] In the formula, This represents the average length of stay. For the first The moment when the fish target first arrives near the fence area; The moment it leaves the near-grid area; (6) Fence contact rate, defined as the proportion of fish targets that have encountered, been trapped, or are suspected of being damaged when they reach the fence area:

[0015] In the formula, For the gate touch event rate; The number of fish targets corresponding to the risk events of touching the fence and being pinched.

[0016] Preferably, in step 4, the "whether the screen was successfully cleared" obtained in step 3 is used as the response variable, and fish biological factors and hydrodynamic environmental factors are used as covariates to establish a fish-blocking effect evaluation model. The fish biological factors include at least one or more of the following: fish body length and fish species. The hydrodynamic environmental factors include at least three or more of the following: flow velocity, flow rate or single unit flow rate, total head difference, reservoir water level, water temperature, dissolved oxygen, turbidity, diurnal rhythm, and screen structural parameters. The probability of screen clearance and its uncertainty range under given operating conditions are output.

[0017] Preferably, the fish-blocking effect evaluation model adopts a hierarchical Bayesian logistic regression model, the expression of which is as follows: ; In the formula, For the first The probability of a fish target successfully passing through the gate under a given working condition; For the intercept term; In order to be with the first Each sample corresponds to a grouping of units, boreholes, or cross-sections. The random effects term; … The values ​​for covariates include any number of items among fish body length, fish species, flow velocity, flow rate, reservoir water level, water temperature, dissolved oxygen, turbidity, diurnal rhythm, unit operating conditions, and the spacing and inclination angle of the trash rack. … The main effect coefficients of the covariates; These are the coefficients of the covariate interaction terms.

[0018] Preferably, in step 5, leave-one-out cross-validation (LOO-CV) is used to compare and select candidate fish-blocking effect evaluation models, determine the model with the best prediction performance, and screen and rank the key influencing factors affecting the fish-blocking effect based on the posterior effect size, posterior inclusion probability, or feature contribution of the optimal model, so as to obtain the threshold range or sensitive range of the key factors.

[0019] Preferably, the model selection criterion for leave-one-out cross-validation is to maximize the expected log prediction density, and its calculation formula is as follows:

[0020] In the formula, The expected log-prediction density for leave-one-out cross-validation; For sample size; For the first Observation results of one sample; Let be the set of all samples except the i-th sample; In order to remove the first After a sample, the model predicts the probability density of that sample; selects... The largest model is taken as the optimal model.

[0021] Preferably, in step 6, based on the optimal fish-blocking effect evaluation model determined in step 5, under the scenario combination of different fish group characteristics, power plant operating conditions, and trash rack structure parameters, the corresponding scenario outputs the indicators of the probability of passing through the rack, the probability of blocking, the probability of avoiding, the dwell time, and the rate of touching the rack. Response curves or probability distributions of key influencing factors and target indicators are plotted to achieve quantitative evaluation, scenario simulation, and prediction of the fish-blocking effect of the trash rack, thereby providing a decision-making basis for the optimization of the trash rack structure and eco-friendly operation scheduling.

[0022] Beneficial effects of this invention: Compared with the prior art, the present invention has the following beneficial effects: 1. Enables continuous monitoring and event-based representation through the fusion of acoustic and video sources, objectively depicting key behavioral processes of fish in the near-grid area; 2. Construct an indicator system that includes risks of encountering, avoiding, staying, and touching the fence, so that the evaluation of fish-blocking effect is more comprehensive, comparable, and can be used for structural and scheduling optimization; 3. By employing a hierarchical Bayesian logistic regression model that couples fish biological factors with hydrodynamic environmental factors, the probability of crossing the gate and the range of uncertainty can be output, thereby improving the credibility of engineering decisions. 4. The prediction criterion of leave-one-out cross-validation is used for model selection, emphasizing generalization prediction ability and avoiding the risk of overfitting caused by screening solely based on goodness of fit. 5. It can conduct probabilistic simulations and predictions under different structural parameters and operating scenarios, providing quantitative basis for the eco-friendly operation of trash racks in low-head power stations. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention. Figure 2 Schematic diagram of the monitoring area Figure 2 A schematic diagram showing the layout of monitoring sections and key activity areas in the trash rack area; Figure 3 A schematic diagram illustrating the acoustic and video fusion recognition and trajectory extraction of fish. Figure 4 A schematic diagram of the evaluation index system and calculation process for fish-blocking effectiveness; Figure 5 A schematic diagram illustrating the training of the fish-blocking effectiveness evaluation model, the selection of LOO-CV, and the output of scenario prediction. Figure 6 This is a schematic diagram of the key influencing factors and the cross-gate probability response curve. Detailed Implementation

[0024] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0025] Example 1: As Figures 1 to 6 As shown, this embodiment provides a method for evaluating and predicting the fish-blocking effect of trash racks at the intake of low-head power stations. The method aims to achieve quantitative evaluation and engineering decision support for the fish-blocking effect of trash racks by focusing on "full-section cluster monitoring—detailed identification of key areas—quantification of indicator system—coupling of probabilistic models—identification of key factors—scenario prediction output." The steps are as follows: Step 1: Monitoring the distribution of fish schools near the debris barrier An evaluation and monitoring zone and a near-gate zone boundary were set up in the waters upstream of the trash rack at the power station intake. Portable fish finders were deployed to conduct full-time, full-section monitoring to obtain fish population density, body length distribution, aggregation water layers, and spatial hotspot distribution. Simultaneously, multi-parameter water quality monitors and portable flow meters were deployed to collect environmental and hydrodynamic data such as water temperature, dissolved oxygen, turbidity, flow velocity, reservoir water level, and unit operating conditions. The fish monitoring data and environmental data were unified according to timestamps, and fish aggregation hotspot maps were generated using spatial interpolation or rasterization statistical methods to determine key activity areas and trash rack openings that require detailed monitoring.

[0026] Step 2: Fish Acoustic and Video Monitoring Behavior Recognition In the key activity areas identified in step 1, dual-frequency identification sonar and underwater video equipment are deployed. Target detection, body length inversion and trajectory tracking methods based on acoustic images and video images are used to identify and continuously track moving fish targets. Based on the trajectory and spatial boundaries, the types of fish behavior events are determined, and fish crossing the fence are classified into at least three categories: approaching, staying, avoiding, attempting to cross the fence, successfully crossing the fence, and risk events of touching the fence and being pinched. The time of occurrence, spatial location and corresponding environmental conditions of the events are recorded to form a modelable "behavioral event sample library".

[0027] Step 3: Construct an evaluation index system for fish-blocking effectiveness Based on the behavioral event samples from step 2, construct an indicator system that includes at least four or more of the following: gate penetration rate, pass rate, blockage rate, avoidance rate, dwell time, and gate contact rate. This system should also support grouped statistics by fish species and body length grade. The definitions of each evaluation indicator are as follows: (1) Reach rate, defined as the proportion of fish targets entering the evaluation and monitoring area that reach the near-grid area of ​​the debris barrier:

[0028] In the formula, To gate rate; The number of fish targets that have entered the evaluation and monitoring area and have been identified and tracked; The target number of fish reaching the vicinity of the trash rack; (2) Pass-through rate, defined as the proportion of fish targets that successfully pass through the debris barrier and appear on the downstream side of the near-barrier area:

[0029] In the formula, For the pass rate; The target number of fish that successfully pass through the gate; (3) Barrier rate, defined as the proportion of fish targets reaching the near-gate area that do not pass through the debris barrier:

[0030] In the formula, For blocking rate; (4) Avoidance rate, defined as the proportion of fish targets that turn away or move away from the debris barrier after reaching the near-barrier area:

[0031] In the formula, For avoidance rate; To avoid the number of fish targets corresponding to the event; (5) Dwell time, defined as the average dwell time of fish targets in the near-gate area:

[0032] In the formula, This represents the average length of stay. For the first The moment when the fish target first arrives near the fence area; The moment it leaves the near-grid area; (6) Fence contact rate, defined as the proportion of fish targets that have encountered, been trapped, or are suspected of being damaged when they reach the fence area:

[0033] In the formula, For the gate touch event rate; The number of fish targets corresponding to the risk events of touching the fence and being pinched.

[0034] Step 4: Establish a fish-blocking effect evaluation model Using "whether the fish successfully passes through the screen" as the response variable, and fish biological factors (including at least one of the fish body length and fish species) and hydrodynamic environmental factors (including at least three of the following: flow velocity, flow rate or unit single flow rate, total head difference, reservoir water level, water temperature, dissolved oxygen, turbidity, diurnal rhythm, unit operating conditions, screen spacing and tilt angle) as covariates, a fish-blocking effect evaluation model is established. The model adopts a hierarchical Bayesian logistic regression form (according to the formula (7) in the claim) to output the probability of passing through the screen under a given operating condition and its uncertainty range.

[0035] Step 5: Identify the key factors affecting the effectiveness of fish blocking A candidate model set containing different combinations of covariates and interaction terms is established. Leave-one-out cross-validation (LOO-CV) is used to compare and select candidate models. The model with the best predictive performance is determined based on the criterion of maximizing the expected log prediction density (ELPD_LOO). The posterior effect size, posterior inclusion probability, or feature contribution of the optimal model is used to screen and rank key influencing factors to obtain the threshold range or sensitive range of key factors. ELPD_LOO is calculated according to the following formula in the claims:

[0036] Step 6: Simulation and Prediction of the Fish-Blocking Effect of the Debris Barrier Based on the optimal model determined in step 5, under scenario combinations of different fish population characteristics, power plant operating conditions, and trash rack structural parameters, the output indicators include the probability of passing through the rack, the probability of being blocked, the probability of avoiding the rack, the dwell time, and the rate of rack contact events; and the response curves or probability distributions of key influencing factors and target indicators are plotted. Figure 6 This is used to propose suggestions for optimizing the structure of trash racks and for eco-friendly operation and scheduling.

[0037] Table 1. Data file variables for the hierarchical Bayesian logistic regression model

[0038] Example 2: This example uses the trash rack at the intake of the GZB power station as an example to provide a reproducible "data-modeling-result output" example to verify the operability and consistency of the method of the present invention. The monitoring area of ​​the GZB power station trash rack can cover the vicinity of the trash racks of the Dajiang and Erjiang power stations. Portable fish finders are used for full-section monitoring, and DIDSON dual-frequency identification sonar and underwater video equipment are deployed in key activity areas to carry out refined monitoring, forming a monitoring system of "full-area screening-key precision measurement". Figure 2 , Figure 3 The specific implementation process is as follows: 1. Monitoring Deployment and Database Construction An evaluation and monitoring area and a near-gate area boundary were set up in the waters upstream of the trash rack at the Gezhouba Hydropower Station intake. Portable fish finders were used to scan the trash rack area at all times and across the entire cross section to obtain the number of fish, their body length distribution, and spatial aggregation areas. Simultaneously, multi-parameter water quality monitors and flow meters were deployed to collect environmental and hydrodynamic data such as water temperature (T), dissolved oxygen (DO), turbidity (TU), reservoir water level (WL), and near-gate flow velocity (V). The unit group number (U) and the trash rack spacing (SP) were also recorded. All monitoring data were timestamped and quality controlled to form a dataset for model training (Table 2).

[0039] Table 2. Statistics on the range of monitoring data for trash racks at Gezhouba Hydropower Station (N=240)

[0040] 2. Acoustic and video fusion recognition and event sample extraction Based on the key fish activity areas identified in step 1, DIDSON dual-frequency identification sonar and underwater video equipment were deployed in front of the trash rack of a typical unit. Target detection, body length inversion, and trajectory tracking methods were used to identify and continuously track moving fish targets. Figure 3 Based on the trajectory and spatial boundary, the behavioral event type is determined, and fish behavior is divided into approach, stay, avoidance, attempt to cross the fence, successful crossing of the fence, and risk events of touching / clamping the fence. The time and spatial location of the event are recorded. The acoustic and video results are checked for consistency to form a "fish target-behavioral event" sample table that can be used to calculate indicators and model.

[0041] 3. Calculation of fish-blocking effect indicators Based on the event samples obtained in step 2, according to Figure 4 The indicator system shown calculates the grid ratio. pass rate Blocking rate Avoidance rate Average stay time and gate touch event rate The average stay time Calculate using the following formula:

[0042] It also supports grouping and statistical analysis by fish species and body length grade to obtain stratified evaluation results.

[0043] 4. Establish a fish-blocking effect evaluation model (hierarchical Bayesian logistic regression). Using "whether the sluice gate was successfully crossed" as the response variable Y (successful crossing = 1, failure = 0), fish biological factors BL, environmental hydrodynamic factors V, N, T, DO, TU, WL, and structural factor SP as covariates, and the group random effect u_{g(i)} corresponding to the unit group number U to characterize the differences in sluice gate location, a hierarchical Bayesian logistic regression model is established; the model expression is as follows: ; In the formula For the first The probability of a fish target successfully crossing the gate under given conditions; model output. And the range of uncertainty, which can be converted into the probability of obstruction ( (This is used to evaluate the effectiveness of fish-blocking.)

[0044] 5. LOO-CV Model Selection and Key Factor Identification A candidate model set is constructed and the expected log prediction density ELPD_LOO is calculated using leave-one-out cross-validation (LOO-CV) (according to formula (8)). The optimal model is selected based on maximizing ELPD_LOO, and key influencing factors are screened and ranked according to the posterior effect size (OR) and confidence interval of the optimal model. The ELPD_LOO calculation formula is as follows:

[0045] The candidate models are defined as follows (all include unit grouping random effects U and grid spacing SP): M1: BL + V + N + T + SP + U; M2: BL + V + N + T + DO + TU + WL + SP + U; M3: Add interactive items (BL×V) to M2.

[0046] Table 3. Model selection results based on leave-one-out cross-validation (ELPD_LOO)

[0047] Note: The model weights are calculated based on ΔELPD, representing the relative support of candidate models. Table 3 shows that M2 has the largest ELPD_LOO and the best predictive performance, thus it is determined as the optimal model for evaluating fish-blocking effectiveness. Figure 5 ).

[0048] The evaluation results of the key parameters obtained under the optimal model M2 are as follows:

[0049] Note: The effects of the random effect U of unit grouping and the effect term of the grid spacing SP are estimated simultaneously in the model to characterize the impact of aperture position differences and structural differences on the grid crossing probability.

[0050] 6. Scenario Simulation and Predictive Output Based on the optimal model M2 determined in step 5, under different combinations of input fish body length BL, near-gate flow velocity V, diurnal rhythm N, and scenarios such as DO, TU, WL, and gate spacing SP, the output gate crossing probability p and the blocking probability (1 / 2) are calculated. p), avoidance probability, dwell time RT, and fence contact rate And other indicators, and plot the response curves or probability distributions of key influencing factors and target indicators ( Figure 6 ).

[0051]

[0052] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station, characterized in that: It includes the following steps: Step 1: Monitoring the distribution of fish schools near the debris barrier; Step 2: Fish acoustic and video monitoring behavior recognition; Step 3: Construct an evaluation index system for the effectiveness of fish blocking; Step 4: Establish a fish-blocking effect evaluation model; Step 5: Identify the key factors affecting the effectiveness of fish blocking; Step 6: Simulation and prediction of the fish-blocking effect of the debris barrier.

2. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 1, characterized in that: The specific process of step 1 is as follows: Fish detection equipment was deployed in the waters upstream of the trash rack at the power station's intake and in the adjacent waters. Portable fish finders were used to monitor the trash rack area around the clock and across the entire cross section to obtain information on fish population density, body length distribution, and spatial aggregation areas. At the same time, multi-parameter water quality monitors and portable flow meters were used to collect environmental and hydrodynamic data such as water temperature, dissolved oxygen, turbidity, flow velocity, reservoir water level, and unit operating conditions, thereby identifying key fish activity areas and forming a monitoring dataset.

3. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station, as described in claim 2, is characterized in that: In step 2, dual-frequency identification sonar and underwater video equipment are deployed in the key activity areas identified in step 1. Target detection, body length inversion and trajectory tracking methods based on acoustic images and video images are used to identify and continuously track fish moving targets. Fish crossing behavior is classified into at least three categories: approaching, lingering, avoiding, attempting to cross the fence, successfully crossing the fence, and risk events of touching the fence and being pinched, and the time and spatial location of these events are recorded.

4. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 1, characterized in that: In step 3, the constructed fish-blocking effect evaluation index system shall include at least four or more of the following: gate arrival rate, passage rate, blocking rate, avoidance rate, dwell time, and gate contact event rate, and support grouped statistics by fish species and body length grade.

5. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 4, characterized in that: The evaluation indicators are defined as follows: (1) Reach rate, defined as the proportion of fish targets entering the evaluation and monitoring area that reach the near-grid area of ​​the debris barrier: In the formula, To gate rate; The number of fish targets that have entered the evaluation and monitoring area and have been identified and tracked; The target number of fish reaching the vicinity of the trash rack; (2) Pass-through rate, defined as the proportion of fish targets that successfully pass through the debris barrier and appear on the downstream side of the near-barrier area: In the formula, For the pass rate; The target number of fish that successfully pass through the gate; (3) Barrier rate, defined as the proportion of fish targets reaching the near-gate area that do not pass through the debris barrier: In the formula, For blocking rate; (4) Avoidance rate, defined as the proportion of fish targets that turn away or move away from the debris barrier after reaching the near-barrier area: In the formula, For avoidance rate; To avoid the number of fish targets corresponding to the event; (5) Dwell time, defined as the average dwell time of fish targets in the near-gate area: In the formula, This represents the average length of stay. For the first The moment when the fish target first arrives near the fence area; The moment it leaves the near-grid area; (6) Fence contact rate, defined as the proportion of fish targets that have encountered, been trapped, or are suspected of being damaged when they reach the fence area: In the formula, For the gate touch event rate; The number of fish targets corresponding to the risk events of touching the fence and being pinched.

6. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 5, characterized in that: In step 4, the "whether the screen was successfully cleared" obtained in step 3 is used as the response variable, and fish biological factors and hydrodynamic environmental factors are used as covariates to establish a fish-blocking effect evaluation model. The fish biological factors include at least one or more of the following: fish body length and fish species. The hydrodynamic environmental factors include at least three or more of the following: flow velocity, flow rate or single unit flow rate, total head difference, reservoir water level, water temperature, dissolved oxygen, turbidity, diurnal rhythm, and screen structural parameters. The probability of screen clearance and its uncertainty range under given operating conditions are output.

7. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 6, characterized in that: The fish-blocking effect evaluation model adopts a hierarchical Bayesian logistic regression model, the expression of which is as follows: ； In the formula, For the first The probability of a fish target successfully passing through the gate under a given working condition; For the intercept term; In order to be with the first Each sample corresponds to a grouping of units, boreholes, or cross-sections. The random effects term; … The values ​​for covariates include any number of items among fish body length, fish species, flow velocity, flow rate, reservoir water level, water temperature, dissolved oxygen, turbidity, diurnal rhythm, unit operating conditions, and the spacing and inclination angle of the trash rack. … The main effect coefficients of the covariates; These are the coefficients of the covariate interaction terms.

8. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 1, characterized in that: In step 5, leave-one-out cross-validation (LOO-CV) is used to compare and select candidate fish-blocking effect evaluation models, determine the model with the best prediction performance, and screen and rank the key influencing factors affecting the fish-blocking effect based on the posterior effect size, posterior inclusion probability, or feature contribution of the optimal model, so as to obtain the threshold range or sensitive range of the key factors.

9. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 8, characterized in that: The model selection criterion for leave-one-out cross-validation is to maximize the expected log prediction density, and its calculation formula is as follows: In the formula, The expected log-prediction density for leave-one-out cross-validation; For sample size; For the first Observation results of one sample; Let be the set of all samples except the i-th sample; In order to remove the first After a sample, the model predicts the probability density of that sample; selects... The largest model is taken as the optimal model.

10. The method for evaluating and predicting the fish-blocking effect of a trash rack at the intake of a low-head power station according to claim 9, characterized in that: In step 6, based on the optimal fish-blocking effect evaluation model determined in step 5, under the scenario combination of different fish group characteristics, power plant operating conditions, and trash rack structural parameters, the corresponding scenario outputs the indicators of the probability of passing through the rack, the probability of blocking, the probability of avoiding, the dwell time, and the rate of contact with the rack. Response curves or probability distributions of key influencing factors and target indicators are plotted to achieve quantitative evaluation, scenario simulation, and prediction of the fish-blocking effect of the trash rack, thereby providing a decision-making basis for the optimization of the trash rack structure and eco-friendly operation scheduling.

Images