A method for predicting upstream water level of canals and ponds based on data mechanism fusion model

Abstract

This invention discloses a method for predicting upstream water levels in canals and ponds based on a data mechanism fusion model. It includes the following steps: Step 1: Constructing a set of similar prediction historical sequences using historical datasets; Step 2: Constructing an integral time-delay model to predict downstream water levels in future periods; Step 3: Using the downstream water level sequence obtained from the integral time-delay prediction model as the correlated downstream water level sequence of the similarity analysis model, constructing a similarity prediction model using Manhattan distance, predicting the most similar downstream water level in history, determining the historical contemporaneous scenario, and predicting the upstream water level in the same historical scenario. This invention overcomes the uncertainty of the correlated sequence bias in the similarity prediction model and significantly improves the prediction accuracy of upstream water levels in canals and ponds.

Description

Technical Field

[0001] This invention relates to the field of hydrological prediction technology, and in particular to a method for predicting upstream water levels in canals and ponds based on a data mechanism fusion model. Background Technology

[0002] Water level prediction in canals is crucial to the safety and efficiency of water conveyance scheduling in rivers and canals. High-precision water level prediction models can assist scheduling decision-makers in formulating favorable measures in advance, avoiding potential water conveyance safety hazards, and improving water supply efficiency. In recent years, numerous mechanism-driven and data-driven prediction models have been applied to river and canal water level prediction. Among them, mechanism-driven prediction models, represented by one-dimensional hydrodynamic models, are widely used for river and canal water surface line prediction; however, their construction process is complex, and their accuracy is limited by parameter calibration. Data-driven models, represented by deep learning, are also increasingly being used for river and canal water level prediction. They offer fast prediction speeds but lack interpretability, and model training time is typically long. These two types of models usually appear as single models, although some studies have adopted coupled models. While this further improves water level prediction accuracy and mitigates the lack of interpretability in data-driven models, the model construction and training process is more complex and often consumes significant computational resources. Summary of the Invention

[0003] The purpose of this invention is to overcome the above-mentioned shortcomings and provide a method for predicting upstream water levels in canals and ponds based on a data mechanism fusion model. In view of the deficiencies of existing river and canal water level prediction models, this invention considers using a computationally efficient integral time delay model as a mechanism-driven model to predict downstream water levels in the river and canal and use it as a correlation downstream water level sequence for similarity prediction. A time series similarity analysis model is then used to further predict upstream water level sequences, thereby improving the interpretability and prediction accuracy of the prediction model.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for predicting upstream water levels in canals and ponds based on a data mechanism fusion model, which includes the following steps:

[0005] Step 1: Construct a set of similar prediction historical sequences using historical datasets;

[0006] Step 2: Construct an integral time-delay model to predict the downstream water level of the canal and pond in future time periods;

[0007] Step 3: Use the downstream water level sequence obtained from the integral time delay prediction model as the correlation downstream water level sequence of the similarity analysis model. Construct a similarity prediction model using Manhattan distance to predict the most similar downstream water level in history, determine the historical same-period scenario, and predict the upstream water level of the canal under the same historical scenario.

[0008] Furthermore, step 1 specifically includes the following:

[0009] First, the historical water conveyance data of the river and canal is cleaned, including the upstream flow, downstream flow, diversion flow, and downstream water level data of the river and canal, to ensure the integrity and accuracy of the data at each time point, and to obtain a set of upstream flow, downstream flow, diversion flow, and downstream water level sequences with a total time length of L.

[0010] Define the relevant sequence time length of the historical similar scene in the similarity prediction model as p. Select the long sequence historical data of upstream and downstream water levels of the canal as the similarity prediction historical dataset. Divide the long sequence historical data of upstream and downstream water levels into N similar prediction historical sequence sets according to p, where each sequence set contains an upstream water level sequence and a downstream water level sequence of length p.

[0011] The total number of similar predicted sequence groups is calculated as follows:

[0012] N = L - p + 1;

[0013] In the formula, L represents the total time length of the historical dataset.

[0014] Furthermore, step 2 specifically includes the following:

[0015] Using the upstream flow rate, downstream flow rate, diversion flow rate, and initial downstream water level for p+1 future time periods as input, the downstream water levels for the next P periods are predicted, resulting in a downstream water level sequence of length p for the canal and pool. The specific prediction formula is as follows:

[0016]

[0017] In the formula, s(t) represents the water depth H downstream of the canal at time t. d (t) relative to the downstream water depth H at the initial time d The increment of (0), q in (t-τ) represents the inflow rate Q of the channel at time (t-τ). in (t-τ) relative to the inflow rate q at the initial time in The increment of (0), q out (t) represents the outflow rate Q from the canal at time t. out (t) relative to the outflow rate Q at the initial time out The increment of (0), d(t) represents the increment of the flow rate D(t) at time t of the channel pool at time t relative to the flow rate D(0) at the initial time. The parameters backwater area A and lag time τ are obtained by the hydrodynamic model.

[0018] Furthermore, step 3 specifically includes the following:

[0019] The downstream water level sequence predicted by the integral time delay model is used as the relevance downstream water level sequence data for the similarity analysis model. The Manhattan distance between the relevance downstream water level sequence and each group of downstream water level sequences in the similar prediction historical sequence set is calculated. The historical downstream water level sequence with the smallest Manhattan distance is taken as the most similar historical downstream water level sequence. The Manhattan distance calculation formula is as follows:

[0020]

[0021] In the formula, l d,i Let l' be the downstream water level at time node i in the sequence of integral time delay results. d,i Let i be the downstream water level at time i in any sequence group within the historical sequence set.

[0022] The most similar upstream water level sequence in history is calculated using the most similar downstream water level sequence in history. The calculation formula is as follows:

[0023]

[0024] In the formula, Let i be the downstream water level at the i-th time point of the most similar downstream water level sequence in the historical sequence set. The upstream water level at the i-th time point of the most similar upstream water level sequence in the historical sequence set;

[0025] The upstream water level of a river or canal can be predicted using the most similar historical upstream water level. The calculation formula is as follows:

[0026]

[0027] In the formula, l u,i β represents the predicted upstream water level of the river channel, where β is the prediction correction coefficient for the upstream water level.

[0028] Furthermore, the formula for calculating the prediction correction coefficient of the upstream water level is as follows:

[0029]

[0030] The beneficial effects of this invention are as follows: Addressing the shortcomings of existing river and canal water level prediction models, this invention considers using a computationally efficient integral time-delay model as a mechanism-driven model to predict downstream water levels and use this as a correlated downstream water level sequence for similarity prediction. A time series similarity analysis model is then used to further predict upstream water level sequences. This fully leverages the predictive power of the integral time-delay model and the efficiency of the similarity prediction model, overcoming the uncertainty of correlation sequence bias in similarity prediction models, and significantly improving the prediction accuracy of upstream water levels in canals and ponds. Attached Figure Description

[0031] Figure 1This is a flowchart illustrating a method for predicting upstream water levels in a canal based on a data mechanism fusion model.

[0032] Figure 2 This is a comparison chart of the RMSE results of the method proposed in this invention and the similarity prediction method;

[0033] Figure 3 This is a comparison chart of the MAPE results of the method proposed in this invention and the similarity prediction method;

[0034] Figure 4 This is a comparison chart of the MSE results of the method proposed in this invention and the similarity prediction method;

[0035] Figure 5 This is a comparison chart of the MAE results of the method proposed in this invention and the similarity prediction method. Detailed Implementation

[0036] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0037] Example 1: Figure 1 As shown, a method for predicting upstream water levels in a canal or reservoir based on a data mechanism fusion model includes the following steps:

[0038] Step 1: Construct a set of similar prediction historical sequences using historical datasets;

[0039] First, the historical water conveyance data of the river and canal is cleaned, including the upstream flow, downstream flow, diversion flow, and downstream water level data of the river and canal, to ensure the integrity and accuracy of the data at each time point, and to obtain a set of upstream flow, downstream flow, diversion flow, and downstream water level sequences with a total time length of L.

[0040] Define the relevant sequence time length of the historical similar scene in the similarity prediction model as p. Select the long sequence historical data of upstream and downstream water levels of the canal as the similarity prediction historical dataset. Divide the long sequence historical data of upstream and downstream water levels into N similar prediction historical sequence sets according to p, where each sequence set contains an upstream water level sequence and a downstream water level sequence of length p.

[0041] The total number of similar predicted sequence groups is calculated as follows:

[0042] N = L - p + 1;

[0043] In the formula, L represents the total time length of the historical dataset.

[0044] Step 2: Construct an integral time-delay model to predict the downstream water level of the canal and pond in future time periods;

[0045] Using the upstream flow rate, downstream flow rate, diversion flow rate, and initial downstream water level for p+1 future time periods as input, the downstream water levels for the next P periods are predicted, resulting in a downstream water level sequence of length p for the canal and pool. The specific prediction formula is as follows:

[0046]

[0047] In the formula, s(t) represents the water depth H downstream of the canal at time t. d (t) relative to the downstream water depth H at the initial time d The increment of (0), q in (t-τ) represents the inflow rate Q of the channel at time (t-τ). in (t-τ) relative to the inflow rate q at the initial time in The increment of (0), q out (t) represents the outflow rate Q from the canal at time t. out (t) relative to the outflow rate Q at the initial time out The increment of (0), d(t) represents the increment of the flow rate D(t) at time t of the channel pool at time t relative to the flow rate D(0) at the initial time. The parameters backwater area A and lag time τ are obtained by the hydrodynamic model.

[0048] Step 3: Use the downstream water level sequence obtained from the integral time delay prediction model as the correlation downstream water level sequence of the similarity analysis model. Construct a similarity prediction model using Manhattan distance to predict the most similar downstream water level in history, determine the historical same-period scenario, and predict the upstream water level of the canal under the same historical scenario.

[0049] The downstream water level sequence predicted by the integral time delay model is used as the relevance downstream water level sequence data for the similarity analysis model. The Manhattan distance between the relevance downstream water level sequence and each group of downstream water level sequences in the similar prediction historical sequence set is calculated. The historical downstream water level sequence with the smallest Manhattan distance is taken as the most similar historical downstream water level sequence. The Manhattan distance calculation formula is as follows:

[0050]

[0051] In the formula, l d,i Let l' be the downstream water level at time node i in the sequence of integral time delay results. d,i Let i be the downstream water level at time i in any sequence group within the historical sequence set.

[0052] The most similar upstream water level sequence in history is calculated using the most similar downstream water level sequence in history. The calculation formula is as follows:

[0053]

[0054] In the formula, Let i be the downstream water level at the i-th time point of the most similar downstream water level sequence in the historical sequence set. The upstream water level at the i-th time point of the most similar upstream water level sequence in the historical sequence set;

[0055] The upstream water level of a river or canal can be predicted using the most similar historical upstream water level. The calculation formula is as follows:

[0056]

[0057] In the formula, l u,i β represents the predicted upstream water level of the river channel, where β is the prediction correction coefficient for the upstream water level.

[0058] Furthermore, the formula for calculating the prediction correction coefficient of the upstream water level is as follows:

[0059]

[0060] Example 2: This example uses the Xuhong River section between Suining Station and Pizhou Station in Jiangsu Province, part of the South-to-North Water Diversion Project (Eastern Route), as an example. Following the method described in the patent, the upstream water level of this section is predicted. To demonstrate the superiority of the proposed upstream water level prediction method, a fusion prediction model combining a similarity prediction model and an integral time-delay model is constructed. A general similarity prediction model is selected as a comparison method for this example. By comparing the effects of the general similarity prediction model and the proposed method, the implementation process and effects of the invention are described.

[0061] (1) Data source and parameter settings for upstream water level prediction in the canal pool

[0062] The historical upstream and downstream water level monitoring series uses hourly data from May 16, 2023 to March 17, 2023, and the test set uses hourly data from April 11, 2023 to April 26, 2023. The results after implementing the technical solution are shown in Table 1. Figure 1 , Figure 2 , Figure 3 , Figure 4 .

[0063] (2) Water level prediction upstream of the canal pool

[0064] The steps for predicting the upstream water level of the canal using the proposed method are as follows:

[0065] (2-1) Construct a set of similar prediction historical sequences using historical datasets;

[0066] (2-2) Construct an integral time delay model to predict the downstream water level of the canal and pool in future time periods;

[0067] (2-3) The downstream water level sequence obtained by the integral time delay prediction model is used as the correlation downstream water level sequence of the similarity analysis model. The similarity prediction model is constructed by Manhattan distance to predict the upstream water level of the canal under the same historical scenario.

[0068] (3) Comparative analysis of similar prediction and the upstream water level prediction results of the proposed method

[0069] As shown in Table 1, the average RMSE, average MAPE, average MSE, and average MAE of the proposed method for predicting the upstream water level of the canal are all lower than those of similar prediction methods, indicating that the average level of the prediction results of the proposed method for the upstream water level of the canal is smaller and the average accuracy is higher.

[0070] Depend on Figure 2 , 3 As can be seen from points 4 and 5, the RMSE, MAPE, MSE, and MAE of the water level prediction results of the proposed method for upstream water level in the canal are generally lower than those of similar prediction methods, indicating that the overall deviation of the proposed method for predicting the water level upstream of the canal is smaller and the overall accuracy is higher.

[0071] Compared with similar prediction methods, the proposed method uses the prediction results of the integral time delay model as the relevant sequence of the similar prediction model, which makes full use of the predictive power of the integral time delay model and the efficiency of the similar prediction model, overcomes the uncertainty of the relevant sequence bias of the similar prediction model, and significantly improves the prediction accuracy of the upstream water level of the canal.

[0072] Table 1 compares the prediction results of the proposed method and the similarity prediction method.

[0073]

[0074] 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 predicting upstream water levels in canals and ponds based on a data mechanism fusion model, characterized in that: It includes the following steps: Step 1: Construct a set of similar prediction historical sequences using historical datasets; Step 2: Construct an integral time-delay model to predict the downstream water level of the canal and pond in future time periods; Step 3: Use the downstream water level sequence obtained from the integral time delay prediction model as the correlation downstream water level sequence of the similarity analysis model. Construct a similarity prediction model through Manhattan distance to predict the most similar downstream water level in history, determine the historical same-period scenario, and predict the upstream water level of the canal under the same historical scenario. Step 3 specifically includes the following: The downstream water level sequence predicted by the integral time delay model is used as the relevance downstream water level sequence data for the similarity analysis model. The Manhattan distance between the relevance downstream water level sequence and each group of downstream water level sequences in the similar prediction historical sequence set is calculated. The historical downstream water level sequence with the smallest Manhattan distance is taken as the most similar historical downstream water level sequence. The Manhattan distance calculation formula is as follows: ； In the formula, Let be the downstream water level at the i-th time node in the sequence of integral time delay results. Let i be the downstream water level at time i in any sequence group within the historical sequence set. The most similar upstream water level sequence in history is calculated using the most similar downstream water level sequence in history. The calculation formula is as follows: ； In the formula, Let i be the downstream water level at the i-th time point of the most similar downstream water level sequence in the historical sequence set. The upstream water level at the i-th time point of the most similar upstream water level sequence in the historical sequence set; The upstream water level of a river or canal can be predicted using the most similar historical upstream water level. The calculation formula is as follows: ； In the formula, The upstream water level prediction results for the river canal. This is the prediction correction coefficient for the upstream water level.

2. The method for predicting upstream water level of a canal / pond based on a data mechanism fusion model according to claim 1, characterized in that: Step 1 specifically includes the following: First, the historical water conveyance data of the river and canal is cleaned, including the upstream flow, downstream flow, diversion flow, and downstream water level data of the river and canal, to ensure the integrity and accuracy of the data at each time point, and to obtain a set of upstream flow, downstream flow, diversion flow, and downstream water level sequences with a total time length of L. Define the relevant sequence time length of the historical similar scene in the similarity prediction model as p. Select the long sequence historical data of upstream and downstream water levels of the canal as the similarity prediction historical dataset. Divide the long sequence historical data of upstream and downstream water levels into N similar prediction historical sequence sets according to p, where each sequence set contains an upstream water level sequence and a downstream water level sequence of length p. The total number of similar predicted sequence groups is calculated as follows: ； In the formula, L represents the total time length of the historical dataset.

3. The method for predicting upstream water level of a canal / pond based on a data mechanism fusion model according to claim 1, characterized in that: Step 2 specifically includes the following: Using the upstream flow rate, downstream flow rate, diversion flow rate, and initial downstream water level for p+1 future time periods as input, the downstream water levels for the next P periods are predicted, resulting in a downstream water level sequence of length p for the canal and pool. The specific prediction formula is as follows: ； In the formula, Let t be the water depth downstream of the canal pool. Relative to the downstream water depth at the initial moment The increment, for Inflow rate of the canal pool relative to the inflow rate at the initial moment The increment, This represents the outflow rate of the channel / pool at time t. relative to the outflow rate at the initial moment The increment, This represents the flow rate at the canal's water distribution outlet at time t. Relative to the initial flow rate at the water distribution point The increment, parameter return water area With lag time It was identified by the hydrodynamic model.

4. The method for predicting upstream water level of a canal / pond based on a data mechanism fusion model according to claim 1, characterized in that: The formula for calculating the prediction correction coefficient for the upstream water level is as follows: 。

Images