Deep learning-based water temperature ecological scheduling effect qualitative and quantitative evaluation method

Abstract

The invention discloses a water temperature ecological scheduling effect qualitative and quantitative evaluation method based on deep learning, and the method comprises the steps: calculating an equivalent water taking elevation through employing a segmented cubic Hermite interpolation method based on vertical temperature distribution data, and representing the water taking elevation positions in different scheduling modes; based on a random forest model, a mapping relation between the hydrodynamic parameters and the water taking elevation is trained, and the model is optimized based on MSE, NSE and MAE indexes; a Shapley value method is adopted to analyze key variables influencing the water taking elevation, and the contribution degree of each hydrodynamic force factor to ecological scheduling response is quantified; when other conditions in the ecological scheduling period are not changed, the stoplog gate state is set to be 0, and the variable quantity of the water taking elevation without ecological scheduling is calculated; according to a water taking elevation result without ecological scheduling, a predicted value of the tail water temperature under the anti-fact framework is obtained, and the improvement effect of ecological scheduling is quantitatively evaluated in combination with an uncertainty analysis method.

Description

Technical Field

[0001] The present invention relates to the technical fields of data science and statistics, and particularly to a qualitative and quantitative evaluation method for the effect of water temperature ecological regulation based on deep learning. Background Art

[0002] The evaluation of the effect of water temperature ecological regulation involves various complex non-linear relationships, and multiple data sources need to be comprehensively considered. The tail water temperature is only determined by the water intake elevation and has nothing to do with the absolute magnitude of the upstream temperature. Since the tail water intake elevation is affected by hydrodynamic conditions such as the water level and flow rate in front of the dam and the state of the stoplog gate, it has strong non-linearity and uncertainty. Traditional models are difficult to accurately describe the relationship between the water intake elevation and variables, while deep learning can capture complex non-linear relationships through a multi-layer network structure to meet the actual needs. In addition, the traditional method calculates the improvement effect of the stoplog gate by interpolation method, without considering the changes of other hydrodynamic conditions before and after, so there are certain errors in the result evaluation. Summary of the Invention

[0003] Therefore, the present invention provides a qualitative and quantitative evaluation method for the effect of water temperature ecological regulation based on deep learning. By applying the knowledge in the hydrological field and combining deep learning with optimization algorithms, the improvement effect of water temperature ecological regulation is qualitatively and quantitatively evaluated to solve the problems raised in the background art.

[0004] To achieve the above object, the present invention provides the following technical solution: A qualitative and quantitative evaluation method for the effect of water temperature ecological regulation based on deep learning, including: Step 1: Data collection: Obtain the time series of the water level in front of the dam, the time series of the flow rate, the vertical temperature distribution data, and the state information of the stoplog gate to form a reservoir hydrodynamic database; Step 2: Calculate the water intake elevation: Based on the vertical temperature distribution data, use the piecewise cubic Hermite interpolation method to calculate the equivalent water intake elevation to represent the water intake elevation position under different regulation methods; Step 3: Deep learning modeling: Based on the random forest model, train the mapping relationship between the hydrodynamic parameters and the water intake elevation, and optimize the model based on the mean square error MSE, Nash efficiency coefficient NSE, and mean absolute error MAE indicators; Step 4: Qualitative evaluation: Use the Shapley value method to analyze the key variables affecting the water intake elevation and quantify the contribution degree of each hydrodynamic factor to the ecological regulation response; Step 5: Counterfactual estimation: When other conditions remain unchanged during the ecological regulation period, set the state of the stoplog gate to 0 and calculate the change amount of the water intake elevation without ecological regulation.

[0005] Step 6: Quantitative evaluation: Based on the water intake elevation results under non-ecological operation, further obtain the predicted values of the tailwater temperature under the counterfactual framework, and combine the uncertainty analysis methods of mean, variance, and maximum probability point to quantitatively evaluate the improvement effect of ecological operation.

[0006] Preferably, the data collected in Step 1 are specifically: collect the time series of the water level in front of the dam , the time series of the flow rate , the vertical temperature distribution data , the tailwater temperature , , is the total number of time sampling points; where: represents the water level height corresponding to a certain moment in front of the dam; represents the inflow corresponding to a certain moment in front of the dam; represents the change of the water temperature distribution at different depths in front of the dam over time ; represents the water temperature corresponding to a certain moment of the tailwater; collect the ecological operation information of the water intake in front of the dam, that is, the status information of the stoplog , where: represents the stoplog is closed; represents the stoplog is open; Use Z-score to detect outliers: Use the standardization formula , where represents the time series of the water level in front of the dam at time , the time series of the flow rate , the vertical temperature distribution data , the tailwater temperature , where and respectively represent to the mean and standard deviation, represents the value of each item at time represents the value of each item at time; the outlier judgment method is ; Process missing values or outliers: For the monitoring data with missing or abnormal situations, perform imputation of missing values and outliers based on time series autoregressive analysis; use the following formula: ; where represents the time series of the water level in front of the dam , the time series of the flow rate , the vertical temperature distribution data , the tailwater temperature in the missing values or outliers at the time representing the water level in front of the dam , the flow time history , the vertical temperature distribution data , the tailwater temperature in the known monitoring values at the time, where are the autoregressive and moving average parameters; is the intercept term, is the autoregressive order, are the moving average parameters, is the error term.

[0007] Preferably, step 2 is specifically as follows: In the vertical temperature distribution data , let the known known depth points correspond to the water temperatures , where , are respectively the known depth and the temperature at the corresponding depth; The basic formula for piecewise cubic Hermite interpolation is as follows: ; where, , , , , , and are the derivatives at the interpolation points and and are calculated by finite differences; Cubic Hermite interpolation is used to establish the continuous mapping relationship between the water temperature in front of the dam and the depth : Set the tailwater temperature , and use the Newton iteration method to solve the elevation of the water intake in front of the dam corresponding to the tailwater, that is, ; where is the derivative of the cubic Hermite interpolation function, is t the elevation obtained at the -th iteration at the time; when , take

[0008] Preferably, step 3 is specifically as follows: Select the time series of the water level in front of the dam as the input variable of deep learning , the time series of flow , and the ecological regulation status , and the output variable is the elevation of the tail water intake ; Random Forest Regression (RF) is based on the decision tree integration method and is used to establish the mapping relationship between hydrodynamic conditions and intake elevation. Its basic model: ; Among them, represents the parameter set of the random forest model, including the number of trees , the maximum depth of the tree , and the training set division ratio ; The parameters are determined by the optimization method, and the evaluation indexes for optimization selection are: mean square error ; Nash efficiency coefficient , mean absolute error ; where is the actual value, is the predicted value, is the average value of the actual values, n is the number of samples, NSE The closer the value of ; to 1, the better the model fitting effect; The predicted intake elevation output of the random forest: is the prediction result of the th decision tree.

[0009] Preferably, the specific steps of step 4 are as follows: On the basis of training the deep learning random forest and obtaining the prediction result , for each input feature of the water level in front of the dam, the time series of flow, and the ecological regulation status, perform multiple random permutations to generate multiple different feature subsets; For each feature subset, use the trained model for prediction, and calculate its prediction result for the elevation of the tail water intake , where , is the total number of Monte Carlo simulations; For each input variable , represents one of the variables among the water level in front of the dam, flow, and ecological regulation status, calculate its Shapley value, represents the contribution of this variable to the prediction result of the elevation of the tail water intake: ; Among them: is the feature subset without for the th random sampling; is the model prediction result corresponding to the feature subset at time t, indicating the th random sampling of the feature subset containing .

[0010] Preferably, during the ecological regulation period , a counterfactual scenario is set, that is, it is assumed that there is no state of the stop logs during this period , while keeping the time history of the water level in front of the dam , the time history of the flow rate , etc. unchanged. Using the trained deep learning random forest model, under the new input conditions , with other conditions unchanged, substituting into the formula , recalculate the water intake elevation in the counterfactual situation, and record the calculation result as , that is: .

[0011] Preferably, in the case where the counterfactual water intake elevation result is known, substitute it into the above Hermite interpolation method result: ; Calculate the predicted value of the counterfactual tail water temperature, and calculate its difference: ; Use the mean value to measure the ecological regulation effect: ; If , it means that the water temperature without ecological regulation is higher than the current ecological regulation; Use the variance to measure the volatility of the water temperature difference: ; A smaller variance means that the volatility of the water temperature difference in time is smaller, reflecting the impact of ecological regulation on water temperature stability; Use the maximum probability point MPP to represent the most common value of the water temperature difference: ; Among them, is the probability density function PDF of the water temperature difference.

[0012] The present invention has the following advantages: The present invention establishes the mapping relationship between dynamic conditions and water intake elevation by constructing a random forest model. Through a single controlled variable, it can calculate the change in water intake elevation brought about by ecological regulation under the condition that other hydrodynamic conditions remain unchanged, and further analyze more precisely the impact of the stoplog gate on water temperature; at the same time, through the counterfactual prediction method combined with the uncertainty quantification analysis methods of mean, variance, and maximum probability point, it quantitatively evaluates the improvement effect of water temperature ecological regulation. Compared with traditional methods, it can adapt to high-dimensional and non-linear data, and the results are more accurate and objective. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is a flowchart provided by the present invention; Figure 2 is a graph of the fitting results of deep learning for Reservoir A in 2024; Figure 3 is a graph of SHAP value results for Reservoir A in 2024; Figure 4 is a comparison graph of counterfactual prediction data of the tail water temperature of Reservoir A in April 2024; Figure 5 is a comparison graph of counterfactual prediction data of the tail water temperature distribution of Reservoir A in April 2024. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The following specific embodiments illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0015] Embodiment: The aquatic biodiversity in the lower reaches of a certain river is relatively high, and the change in water temperature in the downstream river channel will interfere with fish reproduction and have an important impact on the river water ecological environment. The four cascade reservoirs planned for the lower reaches of a certain river - Reservoir A, Reservoir B, Reservoir C, and Reservoir D were impounded and put into operation in January 2020, April 2021, May 2013, and October 2012 respectively. The operation of the reservoirs will change the spatio-temporal distribution of the natural river runoff in the downstream, have an important impact on the water temperature distribution along the river channel and seasonal changes, resulting in obvious flattening and delay effects of the river water temperature, and the cumulative effect will also be produced due to the change in water temperature caused by cascade hydropower development.

[0016] Theoretically, by opening and closing the stop logs to draw water from the upper and middle layers, the adverse effects of the warming delay on fish species that produce pelagic eggs and fish species that produce adhesive and sinking eggs can be alleviated. The ecological regulation of Reservoir A, Reservoir B, and Reservoir C is to draw water from the upper and middle layers of the corresponding reservoirs by lowering the stop logs, increase the water temperature at the reservoir outlet, and enable the river section downstream of the dam to reach the water temperature suitable for the spawning and reproduction of fish species that produce pelagic eggs and fish species that produce adhesive and sinking eggs as soon as possible. The present invention models and analyzes the response relationship between hydrodynamic factors and the water intake elevation during the ecological regulation period, and uses counterfactual prediction to analyze the improvement effect of the stratified water intake of the stop logs on the water temperature during the ecological regulation period.

[0017] Therefore, the present invention provides a qualitative and quantitative evaluation method for the ecological regulation effect of water temperature based on deep learning, as Figure 1 shown, including the following steps: Step 1: Obtain the time series of the water level in front of the dam, the time series of the flow rate, the vertical temperature distribution data, and the status information of the stop logs to form a reservoir hydrodynamic database, including: Collect the time series of the water level in front of the dam , the time series of the flow rate , the vertical temperature distribution data , the tail water temperature , , where $N$ is the total number of time sampling points; among them: $H(t_i)$ represents the water level height corresponding to a certain moment $t_i$ in front of the dam ; $Q(t_i)$ represents the inflow corresponding to a certain moment $t_i$ in front of the dam ; $T(z, t)$ represents the change of the water temperature distribution at different depths $z$ in front of the dam over time $t$ ; $T_w(t_j)$ represents the water temperature corresponding to a certain moment $t_j$ in the tail water. Collect the ecological regulation information of the water intake in front of the dam, that is, the status information of the stop logs $S(t_k)$, where: $S = 0$ represents that the stop logs are closed; $S = 1$ represents that the stop logs are open, and $K$ is the total number of time sampling points.

[0018] Use the Z-score to detect outliers. Use the standardization formula $Z = \frac{X - \mu}{\sigma}$ , where $X$ represents the time series of the water level in front of the dam , the time series of the flow rate , the vertical temperature distribution data , the tail water temperature , where $\mu$ and $\sigma$ represent the mean and standard deviation of $X$ from $t_1$ to $t_N$, respectively, Represents the values at each moment, Represents the values at each moment. The method for determining outliers is .

[0019] Handle missing values or outliers. For missing or abnormal monitoring data, perform imputation of missing values and outliers based on time series autoregressive analysis. Use the following formula: ; where represents the time history of the water level in front of the dam , the time history of the flow rate , the vertical temperature distribution data , the tail water temperature at the moment; the missing value or outlier represents the time history of the water level in front of the dam , the time history of the flow rate , the vertical temperature distribution data , the tail water temperature at the moment; the known monitoring values , are autoregressive and moving average parameters; is the intercept term, is the autoregressive order, is the moving average parameter, is the error term.

[0020] Step 2: Based on the vertical temperature distribution data, use piecewise cubic Hermite interpolation to calculate the equivalent water intake elevation, which characterizes the water intake elevation position under different dispatching methods, including: In the water temperature data in front of the dam , assume that there are known known depth points corresponding to water temperatures , where , are the known depth and the temperature corresponding to the depth respectively. The basic formula for piecewise cubic Hermite interpolation is as follows: ; where , , , , , and is the interpolation point and the derivative at is calculated by finite difference.

[0021] Furthermore, cubic Hermite interpolation is adopted to establish the continuous mapping relationship between the water temperature in front of the dam and the depth : Set the tail water temperature , and use the Newton iteration method to solve the elevation of the water intake in front of the dam corresponding to the tail water , that is ; where is the derivative of the cubic Hermite interpolation function, is t the elevation obtained at the th iteration at time. When , take , take .

[0022] Step 3, using the deep learning random forest model method, establish the mapping relationship between hydrodynamic conditions and water intake elevation, and optimize the model based on indicators such as mean square error (MSE), Nash efficiency coefficient (NSE), and mean absolute error (MAE), including: Select the input variables of deep learning, the time series of the water level in front of the dam , the time series of flow , and the ecological regulation status , and the output variable is the tail water intake elevation ; Random Forest Regression (RF) is based on the decision tree ensemble method and is used to establish the mapping relationship between hydrodynamic conditions and water intake elevation. Its basic model: ; where, represents the parameter set of the random forest model, including the number of trees , the maximum depth of the tree and the training set division ratio . The parameters are determined by the optimization method. The optimization selection evaluation indicators are: mean square error MSE, Nash efficiency coefficient NSE, and mean absolute error MAE. Mean square error ; Nash efficiency coefficient , mean absolute error ; where is the actual value, is the predicted value, is the average value of the actual values, n is the number of samples, NSEThe closer the value is to 1, the better the model fitting effect. The predicted water intake elevation output of the random forest: ; where is the prediction result of the th decision tree.

[0023] Step 4: Use the Shapley value method to analyze the contributions of the water level time series before the dam, the flow rate time series, and the ecological regulation status to the tail water intake elevation, and reveal the key driving factors, including: On the basis of training the deep learning random forest and obtaining the prediction results ( , being the total number of time sampling points), for each input feature (water level before the dam, flow rate time series, ecological regulation status), perform multiple random permutations to generate multiple different feature subsets. For each feature subset, use the trained model to make predictions and calculate its prediction result for the tail water intake elevation , where , being the total number of Monte Carlo simulations.

[0024] For each input variable (representing one of the variables of the water level before the dam, flow rate, and ecological regulation status), calculate its Shapley value, indicating the contribution of this variable to the prediction result of the tail water intake elevation: ; where: is the feature subset without for the th random sampling; is the model prediction result corresponding to the feature subset at time t, represents the feature subset containing for the th random sampling.

[0025] Step 5: Conduct counterfactual estimation, including: During the ecological regulation period ( ), set the counterfactual scenario. That is, assume no flap gate state during this time period, while keeping the water level time series before the dam , the flow rate time series , etc. unchanged. Use the trained deep learning random forest model, under the new input conditions , with other conditions unchanged, substitute into the formula , and recalculate the water intake elevation in the counterfactual situation. The calculation result is recorded as , namely: .

[0026] Step 6, conduct a quantitative evaluation of the results, including: Based on the water intake elevation results under non-ecological scheduling, further obtain the predicted value of the tail water temperature in the counterfactual framework, and combine the uncertainty analysis methods of mean, variance, and maximum probability point to quantitatively evaluate the improvement effect of ecological scheduling. In the counterfactual water intake elevation results in the known situation, substitute it into the result of the Hermite interpolation method described in Step 3: ; Calculate the predicted value of the counterfactual tail water temperature , and calculate its difference: ; Use the mean to measure the effect of ecological scheduling: ; If , it indicates that the water temperature under non-ecological scheduling is higher than the current ecological scheduling.

[0027] Use the variance to measure the volatility of the water temperature difference:

[0028] A smaller variance means that the volatility of the water temperature difference over time is smaller, reflecting the impact of ecological scheduling on water temperature stability.

[0029] Use the maximum probability point (MPP) to represent the most common value of the water temperature difference: ; Among them, is the probability density function (PDF) of the water temperature difference.

[0030] Now use this invention to analyze the ecological scheduling data of Reservoir A.

[0031] Step s1, obtain the water level in front of the dam of Reservoir A, the inflow rate, the outflow rate, and the water temperature data per hour, and use to detect outliers and use the ARIMA model for interpolation; Step s2, use the obtained hydrodynamic data to establish and train a random forest regression model, evaluate the model, and obtain the fitting result; use the model to obtain the response relationship between each hydrodynamic parameter and the water intake elevation, and the mean square error MSE of the evaluation result is 0.009, and the R² value is 0.998, indicating that the model fitting effect is good.

[0032] Step s3, analyze the key variables affecting the water intake elevation by calculating the SHAP value; Comparison result between the predicted value and the true value of the random forest regression model. The horizontal axis represents the true value, and the vertical axis represents the predicted value of the model. The blue scatter points represent the predicted results of the model, and the red dashed line is the ideal prediction line. From Figure 2 it can be seen that the data points are distributed near the ideal prediction line, indicating that the predicted results of the random forest regression model are in good agreement with the true values. Figure 2

[0033] From Figure 3 it is known that the SHAP value distribution of the variable "whether to start scheduling" is relatively dense, and the median is greater than zero, that is, whether to carry out ecological scheduling has a significant impact on the water intake elevation. During the ecological scheduling period, the water intake elevation has a significant increase.

[0034] Step S4, through deep learning counterfactual prediction, set the stoplog gate state to 0, and use the output water intake elevation to inversely obtain the corresponding predicted value of the tail water temperature, and conduct a quantitative analysis on the improvement effect of ecological scheduling; From Figure 4 , the tail water temperature ranges from 14.5 degrees to 18 degrees. The ecological scheduling observed value (blue line) is slightly higher than the counterfactual predicted value (red line) as a whole, with an increase between 0.5 degrees and 1.0 degrees. That is, during the ecological scheduling period, the tail water temperature has increased significantly, which can effectively alleviate the adverse effects of warming delay on fish species that produce drifting eggs and fish species that produce sticky and sinking eggs.

[0035] Figure 5 shows the comparison of the water temperature distribution histogram and the kernel density estimation curve of the ecological scheduling observed value and the counterfactual predicted value. The water temperature distribution of the ecological scheduling observed value is slightly shifted to the right. The peak of the water temperature distribution of the counterfactual predicted value is relatively shifted to the left. It shows that under the ecological scheduling conditions, the peak of the probability density of the tail water temperature has increased significantly.

[0036] Step S5, combine the mean, variance, and maximum probability point uncertainty analysis methods to evaluate the effect during the ecological scheduling of Reservoir A; Table 1

[0037] From Table 1, it can be found from the data in the table that the present invention can evaluate the improvement effect from multiple dimensions. In terms of the mean, the water temperature improvement effect is significant; in terms of the variance, the water temperature improvement effect is more stable during the ecological scheduling period. It can be found that ecological scheduling can effectively increase the water temperature of the downstream river channel and improve the aquatic ecological environment of the river channel. Compared with the present invention, the traditional method can only give a general statement about the water temperature improvement effect before and after the presence or absence of the stoplog gate. Therefore, this method can not only improve the accuracy of result analysis, but also make up for the shortcoming of its single analysis angle.

[0038] The basic principle of the method is that the temperature of the tail water is only determined by the water intake elevation and has nothing to do with the absolute magnitude of the upstream temperature. Since the tail water intake elevation is affected by hydrodynamic conditions such as the water level and flow rate in front of the dam and the state of the stoplog gate, it has strong nonlinearity and uncertainty. It is difficult for traditional models to accurately describe the relationship between the water intake elevation and variables. Deep learning can capture complex nonlinear relationships through a multi-layer network structure to meet the actual needs. In addition, the traditional method calculates the improvement effect of the stoplog gate by interpolation method and fails to consider the changes in other hydrodynamic conditions before and after. Therefore, there are certain errors in the result evaluation. By establishing a model and controlling a single variable, the result is made more accurate and the evaluation method is further optimized. The present invention constructs the response relationship between each hydrodynamic parameter and the water intake elevation through deep learning, and then quantifies the contribution of each feature to the prediction result by the Shapley method. According to the counterfactual prediction, combined with the mean, variance, and maximum probability point uncertainty quantification, a quantitative evaluation method comprehensively evaluates the influence effect of the stoplog gate on the water intake elevation during the ecological regulation period.

[0039] Although the present invention has been described in detail with general descriptions and specific embodiments above, on the basis of the present invention, some modifications or improvements can be made, which are obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection required by the present invention.

Claims

1. A qualitative and quantitative evaluation method for water temperature ecological scheduling effect based on deep learning, characterized by: include: Step 1: Obtain the water level time history, flow time history, vertical temperature distribution data, and stoplog gate status information in front of the dam to form a reservoir hydrodynamic database; Step 2: Based on the vertical temperature distribution data, the piecewise cubic Hermite interpolation method is used to calculate the equivalent water intake elevation to characterize the water intake elevation position under different scheduling modes; Step 3: Based on the random forest model, the mapping relationship between hydrodynamic parameters and water intake elevation is trained, and the model is optimized based on the mean square error (MSE), Nash efficiency coefficient (NSE), and mean absolute error (MAE) indicators; Step 4: Use the Shapley value method to analyze the key variables affecting water intake elevation and quantify the contribution of each hydrodynamic factor to the ecological dispatch response; Step 5: When other conditions remain unchanged during the ecological dispatch period, set the stoplog gate state to 0 and calculate the change in water intake elevation without ecological dispatch; Step 6: Based on the water intake elevation results without ecological scheduling, the predicted value of tailwater temperature under the counterfactual framework is further obtained, and the improvement effect of ecological scheduling is quantitatively evaluated by combining uncertainty analysis methods.

2. The method for qualitative and quantitative evaluation of water temperature ecological scheduling effect based on deep learning according to claim 1 is characterized by: The data collected in step 1 are as follows: 、Flow time course , vertical temperature distribution data , Tailwater temperature , , is the total number of time sampling points; where: Indicates a certain time before the dam The corresponding water level height; Indicates a certain time before the dam The corresponding inflow rate; Indicates different depths in front of the dam The distribution of water temperature over time changes; Indicates the tailwater at a certain time Corresponding water temperature; Collect ecological dispatch information of the water intake in front of the dam, that is, the status information of the stoplog gate ,in: Indicates that the stopgate is closed; Indicates that the stopgate door is open; Detecting outliers using Z-score: Using the standardized formula ,in represent Water level in front of the dam at any given moment 、Flow time course , vertical temperature distribution data , Tailwater temperature ,in and Respectively to The mean and standard deviation of represent The values ​​of each moment, represent The values ​​of each moment; Processing of missing values ​​or outliers: For missing or abnormal monitoring data, interpolation of missing values ​​and outliers based on time series autoregression analysis is performed; the following formula is used: ; in Represents the water level time course in front of the dam 、Flow time course , vertical temperature distribution data , tailwater temperature Zhongzai Missing or outlier values ​​at moments; Represents the water level time course in front of the dam 、Flow time course , vertical temperature distribution data , Tailwater temperature Zhongzai The known monitoring value at time, , are the autoregressive and moving average parameters; is the intercept term, is the autoregressive order, is the moving average parameter, is the error term.

3. The method for qualitative and quantitative evaluation of water temperature ecological scheduling effect based on deep learning according to claim 1 is characterized by: Step 2 is as follows: vertical temperature distribution data In the equation, we assume that The water temperature corresponding to the known depth point ,in , are the known depth and the temperature at the corresponding depth respectively; the basic formula of piecewise cubic Hermite interpolation is as follows: ； in, ， ， ， ， ， and Interpolation point and The derivative at , computed by finite differences; The water temperature before the dam is established using cubic Hermite interpolation With Depth Continuous mapping relationship between: Setting tailwater temperature , use Newton iteration method to solve the dam front water intake elevation corresponding to the tailwater ,Right now ； in is the derivative of the cubic Hermitian interpolation function, for t Time iteration The corresponding elevation obtained.

4. The method for qualitative and quantitative evaluation of water temperature ecological scheduling effect based on deep learning according to claim 1 is characterized in that: Step 3 is as follows: Select the deep learning input variable dam water level time series 、Flow time course , Ecological Dispatch Status , the output variable is the tailwater intake elevation ; Random forest regression is based on the decision tree ensemble method and is used to establish the mapping relationship between hydrodynamic conditions and water intake elevation. Its basic model is: ； in, Represents the set of parameters for the random forest model, including the number of trees , the maximum depth of the tree And the training set division ratio ; The parameters are determined by optimization method, and the optimization evaluation index is: mean square error ; Nash efficiency coefficient , mean absolute error ;in is the actual value, is the predicted value, is the average of the actual values, n is the sample size, NSE The closer the value is to 1, the better the model fit is; the output of the predicted water intake elevation of random forest is: ； in It is The prediction results of the decision tree.

5. The method for qualitative and quantitative evaluation of water temperature ecological scheduling effect based on deep learning according to claim 4 is characterized by: Step 4 The specific steps are: Training deep learning random forest and obtaining prediction results On this basis, each input feature of the dam front water level, flow time course, and ecological dispatch status is randomly arranged multiple times to generate multiple different feature subsets; For each feature subset, use the trained model to predict and calculate its impact on the tailwater intake elevation. The prediction results are , is the total number of Monte Carlo simulations; For each input variable , represents one of the variables in the water level, flow and ecological dispatching state in front of the dam, and calculates its Shapley value. Indicates the contribution of this variable to the prediction result of tailwater intake elevation: ； in: For the The random sampling does not include A subset of features; For feature subset The model prediction result corresponding to time t is: Indicates The random sampling contains feature subset.

6. The method for qualitative and quantitative evaluation of water temperature ecological scheduling effect based on deep learning according to claim 5 is characterized by: In the ecological scheduling period , set the counterfactual scenario, that is, assume that there is no stoplog door state during this time period , and maintain the water level in front of the dam 、Flow time course No change, use the trained deep learning random forest model, under the new input conditions , when other conditions remain unchanged, substitute into the formula , recalculate the water intake elevation under the counterfactual situation The calculation result is recorded as ,Right now: 。 7. The method for qualitative and quantitative evaluation of water temperature ecological scheduling effect based on deep learning according to claim 6 is characterized by: Counterfactual water elevation results Given the known situation, substitute it into the result of the Hermite interpolation method described in claim 3: ； Calculate the predicted value of the counterfactual tailwater temperature , and calculate their difference: ； Use the mean to measure the ecological scheduling effect: ； if , indicating that the water temperature under no ecological regulation is higher than that under current ecological regulation; Use variance to measure the volatility of water temperature differences: ； Use the maximum probability point MPP to represent the most common value of the water temperature difference: ； in, is the probability density function PDF of the water temperature difference.

Images