Runoff forecasting performance improvement method, device and equipment considering medium and long term-short term water regimen coupling

Abstract

The invention provides a runoff forecasting performance improvement method, device and equipment considering medium and long term-short term water regimen coupling, and the method comprises the steps: calculating the gray correlation degree and Chatterj coefficient of a research region according to multi-time scale runoff data; performing state division on the multi-time-scale runoff data according to a distance-average percentage method to obtain a ten-day-month runoff state and determine a ten-day-month scale runoff state transition probability matrix, and determining ten-day-month coupling strength of the research area based on a grey correlation degree, a Chatterj coefficient and the ten-day-month runoff state transition probability matrix; according to a monthly scale runoff forecasting result, monthly runoff features are extracted in a clustering mode, and the monthly runoff features serve as trend reference factors to correct errors of the initial short-term runoff forecasting model; and according to a daily scale runoff forecasting result, aggregating and generating a ten-day runoff characteristic, and adding the ten-day runoff characteristic as an additional forecasting factor into the initial medium and long term runoff forecasting model to realize runoff forecasting performance improvement considering medium and long term-short term water regimen coupling.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of runoff forecasting performance improvement technology, and include, but are not limited to, a method, apparatus and equipment for improving runoff forecasting performance considering the coupling of medium- and long-term and short-term hydrological conditions. Background Technology

[0002] Runoff forecasting is a technical system that calculates changes in flow at the watershed outlet by analyzing rainfall or snowmelt processes within a watershed and using runoff generation and confluence models. As an important research topic in hydrology, watershed runoff forecasting is of great significance in mitigating floods under extreme conditions and in optimizing watershed water resource management (such as hydropower generation, drought relief, water resource allocation, and environmental protection), providing key technical support for ensuring watershed water security.

[0003] Most existing runoff forecasts are based on a single time scale. While this method can reflect runoff changes, it has significant limitations. Runoff formation is extremely complex, influenced by multiple factors that have different effects at different time scales. For example, precipitation may trigger flash floods on a short time scale, while determining seasonal water distribution on a long time scale. Studying only from a single time scale cannot fully consider the interactions between these factors, nor can it achieve accurate predictions of runoff changes. Therefore, constructing multi-time-scale coupled runoff forecasts is crucial. It can comprehensively reflect runoff change patterns and provide strong support for water resource management. In flood prevention and disaster reduction, multi-time-scale coupled runoff forecasts can provide early warnings of floods, rationally schedule flood control projects, reduce flood threats, analyze flood patterns, and improve watershed flood control capabilities. Furthermore, facing the increasing complexity of runoff spatiotemporal distribution caused by climate change and intensified human activities, multi-time-scale coupled runoff forecasts can better adapt to changes, improve forecast accuracy and reliability, and represent an important direction for future runoff forecasting research and application.

[0004] However, existing technologies have the following problems: First, although existing studies have revealed the characteristics of runoff variation in the Yellow River source area at different time scales, research on the interdependence and coupling laws between runoff at different time scales is still relatively scarce; Second, in the field of runoff forecasting, although medium- and long-term and short-term forecasting technologies have formed a relatively complete methodological system, a cross-scale coupling correction mechanism has not yet been formed. Summary of the Invention

[0005] Based on the problems in related technologies, embodiments of the present invention provide a method, apparatus and equipment for improving runoff forecasting performance that considers the coupling of medium- and long-term and short-term hydrological conditions.

[0006] The technical solution of this invention is implemented as follows:

[0007] This invention provides a method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions and short-term hydrological conditions. The method includes:

[0008] Acquire runoff data for the study area at multiple time scales;

[0009] Based on the multi-timescale runoff data, the grey relational degree and Chatterjee coefficient of the study area were calculated;

[0010] The multi-timescale runoff data are classified into states according to the anomaly percentage method to obtain the ten-day-month runoff state.

[0011] Based on the described ten-day-month runoff status, determine the ten-day-month runoff status transition probability matrix;

[0012] The ten-day-month coupling strength of the study area is determined based on the grey relational degree, the Chatterjee coefficient, and the ten-day-monthly runoff state transition probability matrix.

[0013] Based on the monthly runoff forecast results predicted by the pre-constructed initial short-term runoff forecast model, monthly runoff features are extracted by clustering, and the monthly runoff features are used as trend reference factors to correct the error of the initial short-term runoff forecast model.

[0014] Based on the daily runoff forecast results predicted by the pre-constructed initial medium- and long-term runoff forecast model, the ten-day runoff characteristics are aggregated and generated. These ten-day runoff characteristics are then added to the initial medium- and long-term runoff forecast model as additional forecasting factors to improve the runoff forecast performance by considering the coupling of medium- and long-term and short-term hydrological conditions.

[0015] This invention provides a device for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions and short-term hydrological conditions. The device includes:

[0016] The acquisition module is used to acquire runoff data at multiple time scales for the study area;

[0017] The calculation module is used to calculate the grey relational degree and Chatterjee coefficient of the study area based on the multi-timescale runoff data;

[0018] The segmentation module is used to segment the multi-timescale runoff data into states based on the anomaly percentage method to obtain the ten-day-month runoff state.

[0019] The determination module is used to determine the ten-day-monthly runoff state transition probability matrix based on the ten-day-monthly runoff state.

[0020] The determining module is further configured to determine the ten-day-month coupling strength of the study area based on the grey relational degree, the Chatterjee coefficient, and the ten-day-monthly runoff state transition probability matrix.

[0021] The correction module is used to cluster and extract monthly runoff features based on the monthly runoff forecast results predicted by the pre-built initial short-term runoff forecast model, and use the monthly runoff features as a trend reference factor to correct the error of the initial short-term runoff forecast model.

[0022] An addition module is used to aggregate and generate ten-day runoff features based on the daily runoff forecast results predicted by the pre-constructed initial medium- and long-term runoff forecast model, and add the ten-day runoff features as additional forecasting factors to the initial medium- and long-term runoff forecast model to improve the runoff forecast performance considering the coupling of medium- and long-term and short-term hydrological conditions.

[0023] In some embodiments, the calculation module is further configured to select a monthly runoff sequence as a reference sequence from the multi-timescale runoff data, denoted as X0={x0(t1),x0(t2),...,x0(t... n From the multi-timescale runoff data, a ten-day runoff sequence is selected as a comparison sequence, denoted as X. i ={x i (t1),x i (t2),...,x i (t n )}; where i = 1, 2, ..., m, m is the number of comparison sequences; n is the sequence length; calculate the sequence difference, maximum difference, and minimum difference between the reference sequence and the comparison sequence; the formula for calculating the sequence difference, the maximum difference, and the minimum difference is: Δ t (i,0)=|x i (t)-x0(t)|;Δ max =max{Δ t (i,0)};Δ min =min{Δ t (i,0)}; Based on the sequence difference, the maximum difference, and the minimum difference, calculate the correlation coefficient between the reference sequence and the comparison sequence at each observation point; the formula for calculating the correlation coefficient is: In the formula, ξ(t) is the correlation coefficient at time t; ρ is the resolution coefficient; based on the correlation coefficient, the grey correlation degree is calculated; the formula for calculating the grey correlation degree is: In the formula, n is the sequence length; the monthly runoff sequence and the ten-day runoff sequence are merged into a data pair, and sorted according to the value of the ten-day runoff sequence to obtain the sorted monthly runoff sequence; the rank of each value in the sorted value sequence is calculated, and the absolute values ​​of the differences between all adjacent ranks are summed; the Chatterjee coefficient is calculated based on the summed value.

[0024] In some embodiments, the segmentation module is further configured to segment the multi-timescale runoff data into states according to the anomaly percentage method to obtain the ten-day-month runoff state; the ten-day-month runoff state includes exceptionally high, moderately high, normal, moderately low, and exceptionally low; the formula for calculating the anomaly percentage is: In the formula, P is the percentage of anomaly; r is the actual observed runoff value within a certain period, in meters. 3 / s; The historical average runoff over the same period, in meters. 3 / s.

[0025] In some embodiments, the determining module is further configured to calculate the transition frequency from the first state to the second state in the ten-day-monthly runoff state; the formula for calculating the transition frequency is: In the formula, f ij N is the transfer frequency; ij N is the number of transitions from state i to state j; i Let i be the total number of times state i occurs; calculate the state transition probability based on the transition frequency; the formula for calculating the state transition probability is: In the formula, p ij Let l be the state transition probability; l be the total number in the state space; and determine the decadal-monthly runoff state transition probability matrix based on the state transition probabilities.

[0026] In some embodiments, the correction module is further configured to perform cluster analysis on the monthly runoff sequence using the K-means clustering algorithm to obtain monthly runoff characteristics; and determine a clustering loss function based on the monthly runoff characteristics; the calculation formula for the clustering loss function is: In the formula, J is the clustering loss function; x i Let there be i data samples; c k C is the center of cluster k; k The set of all data points belonging to cluster k; the flow of each cluster of the monthly runoff characteristics is used as the trend reference factor and input into the calibration model to correct the error of the initial short-term runoff forecast model; wherein, the calibration model includes LSTM model and Transformer model.

[0027] This invention provides a device for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions, comprising: a memory for storing executable instructions; and a processor for executing the executable instructions stored in the memory to implement the aforementioned method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions.

[0028] This invention provides a computer-readable storage medium storing executable instructions, which, when executed by a processor, implement the aforementioned method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions.

[0029] The present invention provides a method, apparatus, and equipment for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions. By constructing an index evaluation system for the interdependence and coupling strength of ten-day to monthly runoff, the present invention accurately derives the state transition probability matrix of ten-day to monthly runoff, systematically elucidating the multi-scale coupling law of medium- and long-term hydrological conditions. Simultaneously, based on the in-depth mining and integration of runoff information at different time scales, the present invention proposes a forecasting performance improvement technology for the coupling of medium- and long-term hydrological conditions, which can significantly improve the accuracy and reliability of hydrological forecasts. Furthermore, this technical solution is not limited by specific watershed geographical conditions or hydrological characteristics, and is applicable to forecasting performance improvement technology research in various watersheds, possessing strong versatility. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating a method for improving runoff forecasting performance that considers the coupling of medium- and long-term hydrological conditions, provided by the present invention.

[0031] Figure 2 This is a flowchart illustrating another method for improving runoff forecasting performance that considers the coupling of medium- and long-term hydrological conditions, provided by the present invention.

[0032] Figure 3 A schematic diagram of the composition structure of the runoff forecasting performance improvement device considering the coupling of medium- and long-term and short-term hydrological conditions provided by the present invention;

[0033] Figure 4 This is a schematic diagram of the composition structure of the electronic device provided by the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] In the following description, references to "some embodiments" refer to a subset of all possible embodiments; however, it is understood that "some embodiments" may be the same or different subsets of all possible embodiments and may be combined with each other without conflict. Unless otherwise defined, all technical and scientific terms used in the embodiments of the invention have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiments of the invention pertain. The terminology used in the embodiments of the invention is for the purpose of describing the embodiments of the invention only and is not intended to limit the invention.

[0036] The following describes an exemplary application of the runoff forecasting performance enhancement device considering the coupling of medium- and long-term hydrological conditions according to embodiments of the present invention. This device can be implemented as a terminal or a server. In one implementation, the device can be implemented as a laptop, tablet, desktop computer, mobile device, or other types of terminal. In another implementation, it can also be implemented as a server. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiments of the present invention. The following will illustrate an exemplary application of a runoff forecasting performance enhancement device that considers the coupling of medium- and long-term hydrological conditions with short-term hydrological conditions when implemented as a server.

[0037] This invention provides a method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions and short-term hydrological conditions. (See also...) Figure 1 , Figure 1 This is a flowchart illustrating a method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions, provided by an embodiment of the present invention. Figure 1 The steps shown are explained.

[0038] Step S110: Obtain runoff data for the study area at multiple time scales.

[0039] In some embodiments, multi-timescale runoff data refers to runoff data recorded at different time intervals within the study area, covering daily, decadal, and monthly scales. Daily runoff data consists of daily runoff records, decadal runoff data consists of runoff records every ten days, and monthly runoff data consists of runoff records every month. This data forms the basis for runoff forecasting and related analyses, comprehensively reflecting runoff changes across different time dimensions within the study area.

[0040] Step S120: Calculate the grey relational degree and Chatterjee coefficient of the study area based on the multi-timescale runoff data.

[0041] In some embodiments, grey relational degree is an indicator used to measure the degree of correlation between two factors. In this method, it is used to quantify the correlation between runoff data at different time scales within the study area. Its basic principle is to determine the tightness of the correlation by calculating the geometric similarity between the reference and comparison sequences; the higher the value, the stronger the correlation.

[0042] In some embodiments, the Chatterjee coefficient is a statistic that measures the nonlinear correlation between two variables. In this invention, it is used to assist grey relational analysis, enabling a more comprehensive analysis of the dependencies between runoff data at different time scales. It has particular advantages in handling nonlinear relationships and can compensate for the shortcomings of grey relational analysis in certain nonlinear scenarios.

[0043] In this invention, based on the acquired multi-timescale runoff data, decadal-scale runoff data and monthly-scale runoff data are selected as the research objects. For the calculation of grey relational degree, a reference sequence (e.g., using monthly-scale runoff data as the reference sequence) and a comparison sequence (e.g., using decadal-scale runoff data as the comparison sequence) are first determined. Then, the data are dimensionless, the correlation coefficient is calculated, and finally, the grey relational degree is obtained by averaging. For the calculation of the Chatterjee coefficient, it is performed using decadal-scale and monthly-scale runoff data according to its specific calculation formula to quantify the degree of nonlinear correlation between the two.

[0044] Step S130: The multi-timescale runoff data is divided into states according to the anomaly percentage method to obtain the ten-day-month runoff state.

[0045] In some embodiments, the anomaly percentage method is a method for analyzing the degree to which data deviates from the average level. In this method, the state of runoff data is classified by calculating the percentage deviation of multi-timescale runoff data from the multi-year average. For example, it can be classified into states such as exceptionally high, moderately high, normal, moderately low, and exceptionally low, in order to more clearly measure the changing trends of runoff data.

[0046] In some embodiments, the ten-day-monthly runoff status is obtained by classifying the runoff data at the ten-day and monthly scales using the anomaly percentage method, resulting in runoff statuses for each ten-day and month, such as exceptionally abundant, moderately abundant, normal, moderately scarce, and exceptionally scarce. These statuses can intuitively reflect the relative abundance or scarcity of runoff at different time scales.

[0047] In this invention, the calculation results of both grey relational degree and Chatterjee coefficient are combined, and an appropriate method is used to determine the ten-day-monthly coupling strength. For example, by setting certain weights, the two indicators can be combined to obtain a comprehensive value that reflects the tightness of the coupling between ten-day and monthly runoff; this value is the ten-day-monthly coupling strength. The higher the coupling strength, the more significant the interaction between ten-day and monthly runoff.

[0048] Step S140: Determine the ten-day-monthly runoff state transition probability matrix based on the ten-day-monthly runoff state.

[0049] In some embodiments, the decadal-monthly runoff state transition probability matrix is ​​used to describe the probability of transitioning between decadal and monthly runoff states. The elements in the matrix represent the likelihood of transitioning from one state to another, and this matrix allows us to understand the evolution patterns and trends of runoff states at different time scales.

[0050] Step S150: Determine the ten-day-month coupling strength of the study area based on the grey relational degree, the Chatterjee coefficient, and the ten-day-month runoff state transition probability matrix.

[0051] In some embodiments, the decadal-monthly coupling strength refers to the degree of interaction and mutual influence between decadal-scale runoff and monthly-scale runoff. It is determined based on a combination of grey relational analysis, Chatterjee coefficient, and decadal-monthly runoff state transition probability matrix. This strength value reflects the tightness of the coupling relationship between decadal and monthly runoff, providing an important reference for subsequent runoff forecasting.

[0052] In this invention, the ten-day-monthly runoff state transition probability matrix is ​​calculated by combining the grey relational degree, the Chatterjee coefficient, and the results, and an appropriate method is used to determine the ten-day-monthly coupling strength. For example, the two indicators can be combined by setting certain weights and then combined with the same-level state transition probabilities in the transition probability matrix to obtain a comprehensive value that reflects the tightness of the coupling between ten-day and monthly runoff. This value is the ten-day-monthly coupling strength. The higher the coupling strength, the more significant the interaction between ten-day and monthly runoff.

[0053] Step S160: Based on the monthly runoff forecast results predicted by the pre-constructed initial short-term runoff forecast model, the monthly runoff features are extracted by clustering, and the monthly runoff features are used as trend reference factors to correct the error of the initial short-term runoff forecast model.

[0054] In some embodiments, the initial short-term runoff forecasting model is a pre-built model for forecasting runoff in the short term (e.g., on a daily scale). This model is built based on certain hydrological principles, data characteristics, and algorithms, and can preliminarily predict short-term runoff conditions, but may contain certain errors.

[0055] In some embodiments, the monthly runoff forecast result refers to the result obtained by the initial short-term runoff forecast model to forecast the monthly runoff, which reflects information such as the size of the monthly runoff predicted by the model.

[0056] In some embodiments, monthly runoff characteristics refer to representative feature information extracted from monthly runoff forecast results through cluster analysis, such as the mean, peak, and trend of runoff. These features can reflect the overall characteristics of monthly runoff.

[0057] In some embodiments, the trend reference factor uses extracted monthly runoff characteristics as a reference factor to correct errors in the initial short-term runoff forecast model. Since monthly runoff characteristics reflect runoff trends over longer timescales, using them as a trend reference factor allows short-term forecasts to better align with overall runoff change trends.

[0058] In this invention, a pre-constructed initial short-term runoff forecasting model is first used to forecast monthly runoff, yielding monthly runoff forecast results. Then, a suitable clustering algorithm (such as K-means clustering) is employed to perform cluster analysis on these forecast results, grouping monthly runoff forecasts with similar characteristics into one category. Monthly runoff features representative of each category are extracted, such as the mean and peak range for each category. Finally, the extracted monthly runoff features are used as trend reference factors, and combined with the error of the initial short-term runoff forecasting model, an error correction model is established to correct the forecast error of the model, thereby improving the forecast accuracy of the initial short-term runoff forecasting model.

[0059] Step S170: Based on the daily runoff forecast results predicted by the pre-constructed initial medium- and long-term runoff forecast model, aggregate and generate ten-day runoff features, and add the ten-day runoff features as additional forecasting factors to the initial medium- and long-term runoff forecast model to improve the runoff forecast performance considering the coupling of medium- and long-term and short-term hydrological conditions.

[0060] In some embodiments, the initial medium- to long-term runoff forecasting model refers to a pre-constructed model used to forecast runoff in the medium to long term (e.g., on a decadal or monthly scale). Similar to the initial short-term runoff forecasting model, it is also constructed based on relevant principles and methods and is used to preliminarily predict medium- to long-term runoff conditions.

[0061] In some embodiments, the daily runoff forecast results refer to the results obtained by the initial medium- and long-term runoff forecast model in forecasting daily runoff, reflecting information such as the magnitude of daily runoff predicted by the model.

[0062] In some embodiments, the ten-day runoff characteristics refer to the characteristic information that represents the ten-day runoff characteristics obtained by aggregating daily runoff forecast results, such as the total amount and average value of runoff within a ten-day period.

[0063] In some embodiments, additional forecasting factors refer to adding ten-day runoff characteristics as an additional factor to the initial medium- and long-term runoff forecasting model to enrich the model's input information and improve the model's accuracy in forecasting medium- and long-term runoff.

[0064] In this invention, a pre-constructed initial medium- to long-term runoff forecasting model is first used to forecast daily runoff, yielding daily runoff forecast results. Next, the daily runoff forecast results for every ten days are aggregated, such as calculating the total runoff and average runoff for each ten-day period, to generate ten-day runoff characteristics that represent the characteristics of ten-day runoff. Finally, these ten-day runoff characteristics are added as additional forecasting factors to the input of the initial medium- to long-term runoff forecasting model, retraining the model to comprehensively consider hydrological information at different time scales. This improves the forecasting performance of the initial medium- to long-term runoff forecasting model, achieving a performance enhancement in runoff forecasting that considers the coupling of medium- to long-term and short-term hydrological conditions.

[0065] The present invention provides a method, apparatus, and equipment for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions. By constructing an index evaluation system for the interdependence and coupling strength of ten-day to monthly runoff, the present invention accurately derives the state transition probability matrix of ten-day to monthly runoff, systematically elucidating the multi-scale coupling law of medium- and long-term hydrological conditions. Simultaneously, based on the in-depth mining and integration of runoff information at different time scales, the present invention proposes a forecasting performance improvement technology for the coupling of medium- and long-term hydrological conditions, which can significantly improve the accuracy and reliability of hydrological forecasts. Furthermore, this technical solution is not limited by specific watershed geographical conditions or hydrological characteristics, and is applicable to forecasting performance improvement technology research in various watersheds, possessing strong versatility.

[0066] In some embodiments, step S120 can be implemented by the following steps S121 to S127:

[0067] Step S121: Select a monthly runoff sequence as a reference sequence from the multi-timescale runoff data, denoted as X0={x0(t1),x0(t2),...,x0(t... n From the multi-timescale runoff data, a ten-day runoff sequence is selected as a comparison sequence, denoted as X. i ={x i (t1),x i (t2),...,x i (t n )};In the formula, i=1,2,…,m,m is the number of comparison sequences; n is the sequence length.

[0068] Step S122: Calculate the sequence difference, maximum difference, and minimum difference between the reference sequence and the comparison sequence; the formula for calculating the sequence difference, maximum difference, and minimum difference is: Δ t (i,0)=|x i (t)-x0(t)|;Δ max =max{Δ t (i,0)};Δ min =min{Δ t (i,0)}.

[0069] Step S123: Calculate the correlation coefficient between the reference sequence and the comparison sequence at each observation point based on the sequence difference, the maximum difference, and the minimum difference; the formula for calculating the correlation coefficient is: In the formula, ξ(t) is the correlation coefficient at time t; ρ is the resolution coefficient.

[0070] Step S124: Calculate the grey relational degree based on the correlation coefficient; the formula for calculating the grey relational degree is: In the formula, n is the sequence length.

[0071] Step S125: The monthly runoff sequence and the ten-day runoff sequence are merged into a data pair, and sorted according to the values ​​of the ten-day runoff sequence to obtain a sorted value sequence.

[0072] Step S126: Calculate the rank of each value in the sorted value sequence, and sum the absolute values ​​of the interpolations of all adjacent ranks.

[0073] Step S127: Calculate the Chatterjee coefficient based on the summed values.

[0074] In some embodiments, step S140 can be implemented by step S141:

[0075] Step S141: The multi-timescale runoff data is classified into states according to the anomaly percentage method to obtain the ten-day-month runoff state; the ten-day-month runoff state includes exceptionally high, moderately high, normal, moderately low, and exceptionally low; the formula for calculating the anomaly percentage is: In the formula: P is the percentage of anomaly; r is the actual observed runoff value within a certain period, in meters. 3 / s; The historical average runoff over the same period, in meters. 3 / s.

[0076] In some embodiments, step S141 above can be implemented by the following:

[0077] First, calculate the transition frequency from the first state to the second state in the ten-day-monthly runoff state; the formula for calculating the transition frequency is: In the formula, f ij N is the transfer frequency; ij N is the number of transitions from state i to state j; i Let i be the total number of times state i occurs; then, calculate the state transition probability based on the transition frequency; the formula for calculating the state transition probability is: In the formula, p ij Let l be the state transition probability; l be the total number in the state space; finally, based on the state transition probabilities, determine the decadal-monthly runoff state transition probability matrix.

[0078] In some embodiments, step S160 can be implemented by the following steps S161 to S163:

[0079] Step S161: The K-means clustering algorithm is used to perform cluster analysis on the monthly runoff sequence to obtain the monthly runoff characteristics.

[0080] Step S162: Based on the monthly runoff characteristics, determine the clustering loss function; the calculation formula for the clustering loss function is as follows: In the formula, J is the clustering loss function; x i Let there be i data samples; c k C is the center of cluster k; k Let be the set of all data points belonging to cluster k.

[0081] Step S163: The flow rates of each cluster of the monthly runoff characteristics are input into the calibration model as the trend reference factors to correct the error of the initial short-term runoff forecast model; wherein, the calibration model includes an LSTM model and a Transformer model.

[0082] The following will describe an exemplary application of the embodiments of the present invention in a practical application scenario.

[0083] 1) A research method for medium- and long-term-short-term hydrological coupling based on grey relational analysis and Chatterjee correlation coefficient, the specific steps of which are as follows:

[0084] First, runoff data at multiple time scales were collected in the study area. Based on grey relational analysis and Chatterjee correlation coefficient, the interdependence and coupling strength between decadal and monthly runoff in the study area were quantified.

[0085] 1.1) The steps for calculating grey relational degree are as follows:

[0086] 1.1.1) Determine the sequence: Determine the initial reference object for comparison, and select the monthly runoff sequence of the study area as the mother sequence (reference sequence), denoted as X0={x0(t1),x0(t2),...,x0(t... n The runoff sequences for the first, middle, and last ten days of each month in the study area were selected as subsequences (comparison sequences), denoted as X. i ={x i (t1),x i (t2),...,x i (t n )}, where i = 1, 2, ..., m, and m is the number of comparison sequences. In this invention, m is 3.

[0087] 1.1.2) Calculate the difference: This step mainly measures the similarity between the parent sequence and the child sequence. The formulas for calculating the sequence difference, maximum difference, and minimum difference are as follows:

[0088] Δ t (i,0)=|x i (t)-x0(t)|;

[0089] Δ max =max{Δ t (i,0)};

[0090] Δ min =min{Δ t (i,0)};

[0091] 1.1.3) Calculate the correlation coefficient: As the core indicator of grey relational analysis, the correlation coefficient quantitatively characterizes the dynamic correlation strength between the comparison sequence and the reference sequence at each observation point. The formula for calculating the correlation coefficient is as follows:

[0092]

[0093] In the formula, ξ(t) is the correlation coefficient at time t; ρ is the resolution coefficient, which is usually taken as 0.5 and is used to adjust the significance of the difference in the correlation coefficient.

[0094] 1.1.4) Calculate the grey relational degree: The relational degree is a comprehensive reflection of the relational coefficient over time series, and the calculation formula is as follows:

[0095]

[0096] In the formula, n is the sequence length;

[0097] 1.1.5) Results Analysis: Based on the calculated grey relational degree R... i，0 It can compare and analyze the correlation between each comparison sequence and the reference sequence. R i，0 The larger the value, the stronger the correlation between the two sequences.

[0098] 1.2) The steps for calculating the Chatterjee coefficient are as follows:

[0099] 1.2.1) Data Preparation: The monthly runoff sequence is denoted as X0, and the runoff sequences for the first, middle, and last ten days of each month are denoted as X1, X2, and X3, respectively. Ensure that the sequences are of equal length and arranged chronologically.

[0100] 1.2.2) Sort: Sort the runoff sequence X for each ten-day period i (i = 1, 2, 3) are combined with the monthly runoff sequence X0 to form a data pair (X). i (X0), then according to X i Sort the values ​​of X0 in ascending order to obtain the sorted sequence of X0 values, denoted as X0.

[0101] 1.2.3) Calculate the difference: Calculate rank of each value That is, the position of the value in the sequence after sorting, starting from 1;

[0102] 1.2.4) Summation: Summing the absolute values ​​of the differences between all adjacent ranks;

[0103] 1.2.5) Calculate the Chatterjee correlation coefficient: The formula for calculating the Chatterjee correlation coefficient is as follows:

[0104] In the formula, X0 is the monthly runoff sequence; X i The runoff sequences for the first, middle, and last ten days of each month; To make X0 according to the corresponding X i The sorted sequence; for The rank of the j-th value.

[0105] 1.3) Based on the anomaly percentage method, the decadal and monthly runoff data within the study area were divided into five categories: exceptionally abundant (P≥20%), moderately abundant (10%≤P≤20%), normal (-10%≤P≤10%), moderately scarce (-20%≤P≤-10%), and exceptionally scarce (P≤-20%). The formula for calculating the anomaly percentage is as follows:

[0106]

[0107] In the formula, P is the percentage of anomaly; r is the actual observed runoff value within a certain period, in meters. 3 / s; The historical average runoff over the same period, in meters. 3 / s.

[0108] 1.4) The probability matrix for runoff state transition on a decadal-monthly scale is derived based on Markov chains. The calculation steps are as follows:

[0109] 1.4.1) Data collection and state space determination: Collect the state data of the system at different points in time to form a time series.

[0110] 1.4.2) Calculate the transition frequency: Count the frequency of transitions from one state to another. For each state i, calculate the number N of transitions from state i to state j. ij The transfer frequency f ij The calculation formula is as follows:

[0111]

[0112] 1.4.3) Normalization: Divide each element in the transition frequency matrix by the sum of the elements in that row, so that the sum of the elements in each row is 1, to obtain the state transition probability matrix. The state transition probability p from state i to state j is... ij The calculation formula is as follows:

[0113]

[0114] In the formula, p ij is the state transition probability; l is the total number of elements in the state space;

[0115] 1.4.4) Constructing the state transition probability matrix: The state transition probability matrix is ​​obtained by filling the normalized transition probabilities into the state transition probability matrix P.

[0116] 2) Technology for improving forecasting performance by coupling medium- and long-term hydrological information with short-term information, the specific steps of which are as follows:

[0117] 2.1) Error correction for short-term runoff forecasts based on monthly trend factors:

[0118] 2.1.1) K-means clustering algorithm is used to perform cluster analysis on the monthly runoff series. The density of data points within clusters is iteratively optimized to minimize the distance between samples within a cluster and maximize the distance between clusters. The sum of squared errors (SSE) is typically used to measure this, and the calculation formula is as follows:

[0119]

[0120] In the formula, J is the clustering loss function; x 0，i Let there be i data samples; c k C is the center of cluster k; k Let be the set of all data points belonging to cluster k.

[0121] 2.1.2) Set the initial search range for the number of clusters to [2, 10] and plot the trend of the squared error (SSE) as the number of clusters (K) increases. As the number of clusters increases, SSE gradually decreases, but after a certain value, the rate of decrease in SSE slows down significantly, forming an "elbow" inflection point. The cluster value corresponding to this inflection point is usually considered to be the most suitable number of clusters because it finds a reasonable trade-off between clustering quality, model complexity, and computational efficiency.

[0122] 2.1.3) The flow of each cluster after the monthly runoff sequence is clustered is used as the trend factor of the current month and input into the calibration model to correct the short-term runoff forecast error. The selected calibration models include LSTM model and Transformer model.

[0123] 2.2) Medium- and long-term model updates that integrate short-term forecast information:

[0124] Daily runoff forecasts are aggregated into decadal runoff, which is then incorporated as an additional forecasting factor into medium- and long-term runoff forecasting models to improve the accuracy of medium- and long-term model forecasts.

[0125] This invention also provides another technology for improving runoff forecasting performance that considers the coupling of medium- and long-term hydrological conditions with short-term conditions, characterized by comprising the following steps:

[0126] Step 1: Study on the multi-scale coupling law of medium- and long-term and short-term hydrological conditions: construct an index evaluation system for the interdependence and coupling strength of ten-day to monthly runoff, calculate the correlation of ten-day to monthly runoff sequences, and derive the ten-day to monthly runoff state transition probability matrix.

[0127] Step 2: Make full use of runoff information at different time scales to propose a forecasting performance enhancement technology that combines medium- and long-term hydrological information with short-term hydrological information, which can significantly improve the accuracy and reliability of hydrological forecasts.

[0128] Research on the forecast performance improvement technology of medium- and long-term - short-term hydrological coupling: 1) Long-term guides short-term: extract the monthly runoff trend reference factor based on the monthly forecast results, and use it as a dynamic constraint to correct the error of the short-term runoff forecast model; 2) Short-term corrects long-term: generate ten-day runoff characteristics by aggregating daily forecasts, and add them to the medium- and long-term runoff forecast model in the form of additional forecast factors to update the monthly model forecast results.

[0129] Figure 3 This is a schematic diagram of the composition of the runoff forecasting performance improvement device considering the coupling of medium- and long-term and short-term hydrological conditions provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the runoff forecasting performance enhancement device 300 considering the coupling of medium- and long-term and short-term hydrological conditions includes: an acquisition module 301 for acquiring multi-timescale runoff data of the study area; a calculation module 302 for calculating the grey relational degree and Chatterjee coefficient of the study area based on the multi-timescale runoff data; a partitioning module 303 for partitioning the multi-timescale runoff data into states according to the anomaly percentage method to obtain the ten-day-month runoff state; and a determination module 304 for determining the ten-day-month runoff state transition probability matrix based on the ten-day-month runoff state. The determination module 304 is further used to determine the grey relational degree and Chatterjee coefficient based on the Chatterjee coefficient. The coefficients and the ten-day-monthly runoff state transition probability matrix are used to determine the ten-day-monthly coupling strength of the study area; the correction module 305 is used to extract monthly runoff features by clustering based on the monthly runoff forecast results predicted by the pre-constructed initial short-term runoff forecast model, and use the monthly runoff features as trend reference factors to correct the error of the initial short-term runoff forecast model; the addition module 306 is used to aggregate and generate ten-day runoff features based on the daily runoff forecast results predicted by the pre-constructed initial medium- and long-term runoff forecast model, and add the ten-day runoff features as additional forecast factors to the initial medium- and long-term runoff forecast model to achieve improved runoff forecast performance considering medium- and long-term-short-term hydrological coupling.

[0130] It should be noted that the description of the apparatus in this embodiment is similar to that of the method embodiment described above, and has similar beneficial effects, therefore it will not be repeated. For technical details not disclosed in this apparatus embodiment, please refer to the description of the method embodiment of this invention for understanding.

[0131] It should be noted that, in the embodiments of the present invention, if the above-mentioned method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present invention, or the part that contributes to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a terminal to execute all or part of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of the present invention are not limited to any specific hardware and software combination.

[0132] Correspondingly, embodiments of the present invention provide an electronic device, Figure 4 This is a schematic diagram of the composition structure of the electronic bit device provided in the embodiment of the present invention, such as... Figure 4 As shown, the electronic device 400 includes at least a processor 401 and a computer-readable storage medium 402 configured to store executable instructions, wherein the processor 401 generally controls the overall operation of the electronic device 400. The computer-readable storage medium 402 is configured to store instructions and applications executable by the processor 401, and may also cache data to be processed or processed by various modules in the processor 401 and the electronic device 400, and may be implemented using flash memory or random access memory (RAM).

[0133] This invention provides a storage medium storing executable instructions. When these executable instructions are executed by a processor, they cause the processor to perform the method provided in this invention, for example... Figure 1 The method shown.

[0134] In some embodiments, the storage medium may be a computer-readable storage medium, such as a ferromagnetic random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), flash memory, magnetic surface memory, optical disc, or a compact disk-read-only memory (CD-ROM); or it may be a device that includes one or any combination of the above-mentioned memories.

[0135] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0136] As an example, executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file containing other programs or data, for example, in one or more scripts within a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files storing one or more modules, subroutines, or code sections). As an example, executable instructions may be deployed to execute on a single electronic device, or on multiple electronic devices located in one location, or on multiple electronic devices distributed across multiple locations and interconnected via a communication network.

[0137] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of the present invention are included within the scope of protection of the present invention.

[0138] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of the invention, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the invention. The sequence numbers of the above-described embodiments of the invention are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0139] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. In the several embodiments provided by this invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components may be combined, or integrated into another system, or some features may be ignored or not performed.

[0140] The above description is merely an embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions with short-term hydrological conditions, characterized in that, The method includes: Acquire runoff data for the study area at multiple time scales; Based on the multi-timescale runoff data, the grey relational degree and Chatterjee coefficient of the study area were calculated; The multi-timescale runoff data are classified into states according to the anomaly percentage method to obtain the ten-day-month runoff state. Based on the described ten-day-month runoff status, determine the ten-day-month runoff status transition probability matrix; The ten-day-month coupling strength of the study area is determined based on the grey relational degree, the Chatterjee coefficient, and the ten-day-monthly runoff state transition probability matrix. Based on the monthly runoff forecast results predicted by the pre-constructed initial short-term runoff forecast model, monthly runoff features are extracted by clustering, and the monthly runoff features are used as trend reference factors to correct the error of the initial short-term runoff forecast model. Based on the daily runoff forecast results predicted by the pre-constructed initial medium- and long-term runoff forecast model, the ten-day runoff characteristics are aggregated and generated. These ten-day runoff characteristics are then added to the initial medium- and long-term runoff forecast model as additional forecasting factors to improve the runoff forecast performance by considering the coupling of medium- and long-term and short-term hydrological conditions.

2. The method according to claim 1, characterized in that, The calculation of the grey relational degree and Chatterjee coefficient of the study area based on the multi-timescale runoff data includes: From the multi-timescale runoff data, a monthly runoff sequence is selected as a reference sequence, denoted as X0={x0(t1),x0(t2),...,x0(t3)}. n From the multi-timescale runoff data, a ten-day runoff sequence is selected as a comparison sequence, denoted as X. i ={x i (t1),x i (t2),...,x i (t n In the formula, i = 1, 2, ..., m, m is the number of comparison sequences; n is the sequence length. Calculate the sequence difference, maximum difference, and minimum difference between the reference sequence and the comparison sequence; the formulas for calculating the sequence difference, maximum difference, and minimum difference are: Δ t (i,0)=|x i (t)-x0(t)|;Δ max =max{Δ t (i,0)};Δ min =min{Δ t (i,0)}; Based on the sequence difference, the maximum difference, and the minimum difference, the correlation coefficient between the reference sequence and the comparison sequence at each observation point is calculated; the formula for calculating the correlation coefficient is: In the formula, ξ(t) is the correlation coefficient at time t; ρ is the resolution coefficient; Based on the correlation coefficient, the grey correlation degree is calculated; the formula for calculating the grey correlation degree is: In the formula, n is the sequence length; The monthly runoff sequence and the ten-day runoff sequence are merged into a data pair, and sorted according to the value of the ten-day runoff sequence to obtain the sorted monthly runoff sequence; Calculate the rank of each value in the sorted monthly runoff sequence, and sum the absolute values ​​of the differences between all adjacent ranks; The Chatterjee coefficient is calculated based on the summed values; the formula for calculating the Chatterjee coefficient is as follows: In the formula, ξ n (X i X0) represents the Chatterjee coefficients of decadal and monthly runoff; X0 is the monthly runoff sequence; X i It is a ten-day runoff sequence; The rank of the j-th value after sorting the monthly runoff sequence according to the values ​​of the ten-day runoff sequence.

3. The method according to claim 1, characterized in that, The process of classifying the multi-timescale runoff data into states based on the anomaly percentage method to obtain the decadal-monthly runoff states includes: The multi-timescale runoff data is classified into states according to the anomaly percentage method to obtain the ten-day-month runoff states; the ten-day-month runoff states include exceptionally high, moderately high, normal, moderately low, and exceptionally low; the formula for calculating the anomaly percentage is: In the formula: P is the percentage of anomaly; r is the actual observed runoff value within a certain period, in meters. 3 / s; The historical average runoff over the same period, in meters. 3 / s.

4. The method according to claim 3, characterized in that, The step of determining the ten-day-monthly runoff state transition probability matrix based on the ten-day-monthly runoff state includes: Calculate the transition frequency from the first state to the second state in the aforementioned ten-day-monthly runoff state; the formula for calculating the transition frequency is: In the formula, f ij N is the transfer frequency; ij N is the number of transitions from state i to state j; i Let i be the total number of times state i occurs. The state transition probability is calculated based on the transition frequency; the formula for calculating the state transition probability is: In the formula, p ij is the state transition probability; l is the total number of elements in the state space; Based on the state transition probabilities, the decadal-monthly runoff state transition probability matrix is ​​determined.

5. The method according to claim 1, characterized in that, The step of clustering and extracting monthly runoff features from the monthly runoff forecast results based on the pre-constructed initial short-term runoff forecast model, and using these monthly runoff features as trend reference factors to correct the errors of the initial short-term runoff forecast model, includes: The K-means clustering algorithm was used to perform cluster analysis on the monthly runoff sequences to obtain the monthly runoff characteristics; Based on the monthly runoff characteristics, a clustering loss function is determined; the formula for calculating the clustering loss function is as follows: In the formula, J is the clustering loss function; x i Let there be i data samples; c k C is the center of cluster k; k Let k be the set of all data points belonging to cluster k. The flow rates of each cluster of monthly runoff characteristics are used as trend reference factors and input into the calibration model to correct the error of the initial short-term runoff forecast model; wherein, the calibration model includes an LSTM model and a Transformer model.

6. A device for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions with short-term hydrological conditions, characterized in that, The device includes: The acquisition module is used to acquire runoff data at multiple time scales for the study area; The calculation module is used to calculate the grey relational degree and Chatterjee coefficient of the study area based on the multi-timescale runoff data; The segmentation module is used to segment the multi-timescale runoff data into states based on the anomaly percentage method to obtain the ten-day-month runoff state. The determination module is used to determine the ten-day-monthly runoff state transition probability matrix based on the ten-day-monthly runoff state. The determining module is further configured to determine the ten-day-month coupling strength of the study area based on the grey relational degree, the Chatterjee coefficient, and the ten-day-month runoff state transition probability matrix; The correction module is used to cluster and extract monthly runoff features based on the monthly runoff forecast results predicted by the pre-built initial short-term runoff forecast model, and use the monthly runoff features as a trend reference factor to correct the error of the initial short-term runoff forecast model. An addition module is used to aggregate and generate ten-day runoff features based on the daily runoff forecast results predicted by the pre-constructed initial medium- and long-term runoff forecast model, and add the ten-day runoff features as additional forecasting factors to the initial medium- and long-term runoff forecast model to improve the runoff forecast performance considering the coupling of medium- and long-term and short-term hydrological conditions.

7. A device for improving runoff forecasting performance considering the coupling of medium- and long-term hydrological conditions with short-term conditions, characterized in that, include: Memory, used to store executable instructions; The processor, when executing executable instructions stored in the memory, implements the runoff forecasting performance improvement method considering medium- and long-term hydrological coupling as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing executable instructions for causing a processor to execute the executable instructions to implement the runoff forecasting performance improvement method considering medium- and long-term hydrological coupling as described in any one of claims 1 to 5.

Images