A reservoir mutation early warning method based on time sequence water conservancy data

Abstract

This invention discloses a reservoir abrupt change early warning method based on time-series hydraulic data, belonging to the field of reservoir early warning technology. The method first decomposes the current and historical flow data of each tributary upstream of the reservoir to obtain current trend, current oscillation, historical annual trend, and historical annual oscillation terms. Then, it determines the baseline trend and baseline oscillation terms. Next, it calculates the differences between the current trend, current oscillation, and their corresponding baseline terms to obtain trend deviation and oscillation deviation terms, and acquires their respective risk accumulation amounts. Then, it combines the confluence delay to determine the confluence time, calculates the flow contribution weight, and weights and sums the two types of risk accumulation amounts to obtain a comprehensive trend and oscillation risk index. This comprehensive risk index is enhanced by risk trend factors and impact factors to obtain the trend and oscillation abrupt change intensity. Finally, it outputs the comprehensive early warning probability and early warning level. This invention improves the accuracy and reliability of reservoir abrupt change early warning.

Description

Technical Field

[0001] This invention relates to the field of reservoir early warning technology, specifically to a method for early warning of sudden changes in reservoirs based on time-series hydraulic data. Background Technology

[0002] As a core infrastructure for watershed water resource regulation and flood control, the dynamic changes in upstream inflows directly affect dam safety and downstream flood control scheduling. In actual operation, reservoirs typically have multiple tributaries upstream, and the flow processes of each tributary are influenced by various factors such as precipitation, snowmelt, and human activities, exhibiting obvious trends (e.g., seasonal fluctuations, long-term variations) and oscillations (e.g., short-term peaks caused by heavy rainfall, daily regulation fluctuations). When abnormal changes occur in upstream inflows, failure to provide timely warnings may lead to reservoirs operating above flood limits, delayed flood discharge scheduling, or even dam overtopping and other safety accidents.

[0003] Existing technologies include reservoir early warning methods based on time-series flow data. These methods typically identify abnormal flow increases by setting an absolute threshold for a single flow (such as the flood control limit flow) or by using a moving average method. These methods usually involve simply summing the flows of each tributary or taking the maximum value as a representative of the reservoir inflow, and then comparing it with a preset historical average or a fixed threshold. An early warning is triggered when the threshold is exceeded.

[0004] However, existing technologies have the following problems: First, these methods fail to distinguish between components of different physical origins (such as long-term trends and short-term oscillations) in the flow sequence, easily misjudging normal seasonal rises as sudden changes. Second, due to the varying distances of tributaries from the reservoir, there are significant differences in confluence delays. Existing technologies ignore this spatial heterogeneity, often applying the same response time to all tributaries, leading to a mismatch between the warning time and the actual arrival time of the sudden change, thus reducing the accuracy and timeliness of the warning. Therefore, existing technologies suffer from low accuracy in reservoir sudden change warnings. Summary of the Invention

[0005] To address the aforementioned shortcomings in existing technologies, this invention provides a reservoir abrupt change early warning method based on time-series water conservancy data, which solves the problem of low accuracy in reservoir abrupt change early warning in existing technologies.

[0006] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: a reservoir sudden change early warning method based on time-series hydraulic data, comprising the following steps: S1. Decompose the flow of each upstream tributary of the reservoir in the current time period and in the historical time period to obtain the current trend term, the current oscillation term, the historical year trend term, and the historical year oscillation term; S2. Take the average of the trend terms for each historical year to obtain the baseline trend term, and take the average of the oscillation terms for each historical year to obtain the baseline oscillation term; S3. Obtain the trend deviation term based on the difference between the current trend term and the benchmark trend term; obtain the oscillation deviation term based on the difference between the current oscillation term and the benchmark oscillation term. S4. Obtain the cumulative trend risk for the trend deviation item and the cumulative oscillation risk for the oscillation deviation item; S5. Determine the confluence time based on the confluence delay, calculate the flow contribution weight, and sum the cumulative amount of trend and oscillation risks to obtain the comprehensive risk index of trend and oscillation. S6. Based on the risk trend factor and the shock factor, the comprehensive risk index of trend and oscillation is enhanced to obtain the trend and oscillation change intensity. S7. Map the trend and the intensity of oscillation changes to corresponding probabilities, weight them to obtain the comprehensive early warning probability and output the early warning level.

[0007] Furthermore, S1 includes the following sub-steps: S11. Extract the flow rate of each upstream tributary of the reservoir in the current time period and in historical time periods to obtain the flow rate in the current time period and the flow rate in the same time period over many years in history; S12. Smooth the traffic flow in the current time period and the traffic flow in the same time period of each historical year to obtain the current trend item and the trend item of each historical year. S13. Subtract the current trend term from the current time period flow to obtain the current oscillation term; S14. Subtract the trend term of each historical year from the flow rate of the same time period in each historical year to obtain the oscillation term of each historical year.

[0008] Furthermore, the specific process of S3 includes: in the same upstream tributary, the current trend term... The first item in the time flow and baseline trend item Subtracting the flow rate at each moment, we obtain the trend term. The difference in flow rate at different times will be used as the trend item. The difference in flow rate at different times and the first term in the baseline trend term The ratio of flow rate at time step is used as the first... The trend deviation value at each time point constitutes the trend deviation term. Number the time; In the same tributary upstream, the th term in the current oscillation term will be... The first term in the time-current flow and reference oscillation term Subtracting the flow rates at each moment, we obtain the oscillation term. The difference in flow rate at different times will be the oscillation term. The difference in flow rate at time and the first term in the reference oscillation term The ratio of flow rate at time step is used as the first... The oscillation deviation value at each time point constitutes the oscillation deviation term.

[0009] Furthermore, S4 includes the following sub-steps: S41. Set the initial value of the trend risk accumulation to 0; [The sentence is incomplete and requires further context.] The cumulative amount of trend risk at time point and the trend deviation term in the first term The trend deviation values ​​at each time point are summed and subtracted from the trend decay threshold to obtain the value of the first time point. The initial trend accumulation at time step; determining the first... Is the initial trend accumulation at time step 0 greater than 0? If so, then... The initial trend accumulation at time 1 is given to the first The cumulative trend risk at any given moment; if not, assign 0 to the first... The cumulative amount of trend risk at any given moment. Number the time; S42. Set the initial value of the cumulative oscillation risk to 0, and determine the first oscillation deviation term. If the oscillation deviation value at time step is greater than 0, then the 1st step... The oscillation deviation value at any given time is retained; otherwise, 0 is assigned to the first value. The deviation value of the oscillation at time; will the first The cumulative oscillation risk at time point and the adjusted oscillation deviation term. The oscillation deviations at each time point are summed and subtracted from the oscillation decay threshold to obtain the first value. The initial cumulative oscillation at time t; determining the t... Is the initial cumulative oscillation at time step 0 greater than 0? If so, then... The initial oscillation accumulation at time t gives the first The cumulative oscillation risk at time point; if not, assign 0 to the first... The cumulative amount of oscillation risk at any given moment.

[0010] Furthermore, S5 includes the following sub-steps: S51. Calculate the confluence delay time of each upstream tributary to the reservoir; S52. Based on the confluence delay time, determine the confluence time corresponding to each tributary, and calculate the trend flow contribution weight and oscillation flow contribution weight of each upstream tributary at that confluence time. S53. The trend flow contribution weight and the cumulative trend risk are weighted and summed to obtain the trend comprehensive risk index. The oscillation flow contribution weight and the cumulative oscillation risk are weighted and summed to obtain the oscillation comprehensive risk index.

[0011] Furthermore, the process of calculating the trend flow contribution weight in S52 includes: summing the current trend items of all upstream tributaries at the confluence time to obtain the total flow, and the ratio of the current trend item flow of each tributary to the total flow is the trend flow contribution weight. The process of calculating the contribution weight of oscillating flow includes: summing the current oscillations of all upstream tributaries at the confluence time to obtain the total flow, and the ratio of the current oscillation of each tributary to the total flow is the contribution weight of oscillating flow. The expression for the trend composite risk index obtained in S53 is as follows: , in, For the first The overall risk index of the trend at any given moment. For the upstream The first tributary The weight of traffic contribution in real time trends For the upstream The first tributary The cumulative amount of trend risk at any given moment. The number of upstream tributaries, For the convergence moment, For the tributary number, Number the time; The expression for the oscillation comprehensive risk index obtained in S53 is as follows: , in, For the first The comprehensive risk index of oscillation at any given moment. For the upstream The first tributary The weight of the oscillating flow contribution at all times. For the upstream The first tributary Accumulated risk of constant oscillation.

[0012] Furthermore, S6 includes the following sub-steps: S61, according to the... The trend of the comprehensive risk index at any moment and the first The risk trend factor is calculated by taking the difference between the overall risk index at different times. Number the time; S62. Enhance the trend composite risk index using risk trend factors to obtain the trend change intensity; S63. Set a short-term impact window to filter the largest flow rate from the short-term impact window of the current oscillation term; S64. Add up the maximum flow rates of all upstream tributaries to obtain the total impact flow rate; S65. Calculate the impact factor based on the total impact flow rate; S66. The oscillation comprehensive risk index is enhanced by using the shock factor to obtain the oscillation mutation intensity.

[0013] Furthermore, the formula for calculating the risk trend factor is as follows: , in, For the first Risk trend factors at any time For the first The overall risk index of the trend at any given moment. For the first The overall risk index of the trend at any given moment. Here, | represents the time number, and | represents the absolute value operation; The formula for calculating the impact factor is: , in, For the first The impact factor of a moment For the first Total impact flow at any given moment This is the threshold for the total impact flow.

[0014] Furthermore, the expression for the intensity of the trend abrupt change is obtained as follows: , in, For the first The intensity of the trend change at any given moment. For the first The overall risk index of the trend at any given moment. For the first Risk trend factors at any given time; The expression for the intensity of the oscillation mutation is: , in, For the first The intensity of the oscillation change at time, For the first The comprehensive risk index of oscillation at any given moment. For the first The impact factor of a moment.

[0015] Furthermore, S7 includes the following sub-steps: S71. Map the intensity of trend mutation to the probability of trend mutation, and map the intensity of oscillatory mutation to the probability of oscillatory mutation; S72. Weight the probability of trend change and the probability of oscillation change to obtain the comprehensive warning probability, and output the warning level based on the comprehensive warning probability.

[0016] The beneficial effects of this invention are as follows: 1. This invention decomposes the flow sequence into trend terms and oscillation terms, and compares them with historical benchmarks for the same period. This can effectively distinguish between normal fluctuations such as seasonal floods and true abnormal changes, reduce false alarms and missed alarms, and significantly improve the accuracy of change warning.

[0017] 2. This invention considers the differences in the confluence delay of each tributary from the reservoir, determines the confluence time through the confluence delay and calculates the flow contribution weight, so that the warning trigger time matches the actual time when the sudden change arrives at the reservoir, thereby improving the timeliness and accuracy of the warning.

[0018] 3. This invention designs a risk trend factor for trend risk, utilizing the time-series difference of the trend comprehensive risk index to identify the accelerated process of risk accumulation, thereby enhancing sensitivity to "persistent anomalies." For oscillation risk, it designs an impact factor, independently selecting the maximum value of the oscillation term flow within the short-term impact window of each tributary and summing them to obtain the total impact flow, then calculating the impact factor to amplify the response to "short-term strong impacts." The differentiated design of these two types of enhancement factors enables the early warning system to have sensitive detection capabilities for both slowly developing trend-type abrupt changes and sudden oscillating abrupt changes.

[0019] 4. This invention combines the intensity of trend mutation and the intensity of oscillation mutation to obtain a comprehensive early warning probability, thereby improving the accuracy of reservoir mutation early warning. Attached Figure Description

[0020] Figure 1 This is a flowchart of a reservoir abrupt change early warning method based on time-series water conservancy data. Detailed Implementation

[0021] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0022] like Figure 1 As shown, a reservoir abrupt change early warning method based on time-series water conservancy data includes the following steps: S1. Decompose the flow of each upstream tributary of the reservoir in the current time period and in the historical time period to obtain the current trend term, the current oscillation term, the historical year trend term, and the historical year oscillation term; S2. Take the average of the trend terms for each historical year to obtain the baseline trend term, and take the average of the oscillation terms for each historical year to obtain the baseline oscillation term; S3. Obtain the trend deviation term based on the difference between the current trend term and the benchmark trend term; obtain the oscillation deviation term based on the difference between the current oscillation term and the benchmark oscillation term. S4. Obtain the cumulative trend risk for the trend deviation item and the cumulative oscillation risk for the oscillation deviation item; S5. Determine the confluence time based on the confluence delay, calculate the flow contribution weight, and sum the cumulative amount of trend and oscillation risks to obtain the comprehensive risk index of trend and oscillation. S6. Based on the risk trend factor and the shock factor, the comprehensive risk index of trend and oscillation is enhanced to obtain the trend and oscillation change intensity. S7. Map the trend and the intensity of oscillation changes to corresponding probabilities, weight them to obtain the comprehensive early warning probability and output the early warning level.

[0023] In this embodiment, S1 includes the following sub-steps: S11. Extract the flow rate of each upstream tributary of the reservoir in the current time period and in historical time periods to obtain the flow rate in the current time period and the flow rate in the same time period over many years in history; S12. Smooth the traffic flow in the current time period and the traffic flow in the same time period of each historical year to obtain the current trend item and the trend item of each historical year. S13. Subtract the current trend term from the current time period flow to obtain the current oscillation term; S14. Subtract the trend term of each historical year from the flow rate of the same time period in each historical year to obtain the oscillation term of each historical year.

[0024] In this embodiment, the length of the current time period and the historical time period is one week, and the sampling interval is half an hour per day. The current time period is July 8th to July 14th, 2024, and the historical time period is July 8th to July 14th of the same period in 2021 to 2023.

[0025] The expression for the current trend term in S12 is: ,in, For the upstream The current trend of the tributary Real-time flow For the upstream The tributary in the current time period Real-time flow For cumulative variables, For the tributary number, The length of the window. The time number, To round down, Take 48, which corresponds to a 24-hour range.

[0026] The expression for the current oscillation term in S13 is: ,in, For the upstream The current oscillation term of the branch Real-time flow For the upstream The tributary in the current time period Real-time traffic.

[0027] The expression for the historical year trend item in S12 is: ,in, For the upstream The first tributary The first historical year trend item Real-time flow For the upstream The first tributary The first historical year in the same period Real-time flow Used to number historical years.

[0028] The expression for the historical year oscillation term in S13 is: ,in, For the upstream The first tributary The first historical year oscillation item Real-time flow For the upstream The first tributary The first historical year in the same period Real-time traffic.

[0029] Each upstream tributary refers to all inflowing rivers that flow into the reservoir and have independent hydrological monitoring stations.

[0030] In this embodiment, an exponentially weighted moving average method can also be used for smoothing.

[0031] In this embodiment, the expression for the baseline trend term in S2 is: ,in, For the upstream The baseline trend term of the tributary Real-time flow This refers to the number of historical years.

[0032] In this embodiment, the expression for the reference oscillation term in S2 is: ,in, For the upstream The first reference oscillation term of the branch Real-time traffic.

[0033] In this embodiment, the specific process of S3 includes: in the same tributary upstream, the current trend item of the first... The first item in the time flow and baseline trend item Subtracting the flow rate at each moment, we obtain the trend term. The difference in flow rate at different times will be used as the trend item. The difference in flow rate at different times and the first term in the baseline trend term The ratio of flow rate at time step is used as the first... The trend deviation value at each time point constitutes the trend deviation term. The time is numbered, that is: ,in, For the upstream The trend deviation term of the tributary Deviation from trend at any time For the upstream The current trend of the tributary Real-time flow For the upstream The baseline trend term of the tributary Real-time flow The parameter is used to prevent division by zero in the denominator.

[0034] In the same tributary upstream, the th term in the current oscillation term will be... The first term in the time-current flow and reference oscillation term Subtracting the flow rates at each moment, we obtain the oscillation term. The difference in flow rate at different times will be the oscillation term. The difference in flow rate at time and the first term in the reference oscillation term The ratio of flow rate at time step is used as the first... The oscillation deviation value at each time point constitutes the oscillation deviation term, i.e.: ,in, For the upstream The oscillation deviation term of the branch Constant oscillation deviation value, For the upstream The current oscillation term of the branch Real-time flow For the upstream The first reference oscillation term of the branch Real-time traffic.

[0035] This invention constructs trend deviation and oscillation deviation terms by calculating the relative deviations between the current trend term, oscillation term and the corresponding historical benchmark term. This not only quantifies the degree of deviation of the current long-term flow trend from the same period in history, but also captures abnormal changes in short-term fluctuation amplitude.

[0036] In this embodiment, S4 includes the following sub-steps: S41. Set the initial value of the trend risk accumulation to 0; [The sentence is incomplete and requires further context.] The cumulative amount of trend risk at time point and the trend deviation term in the first term The trend deviation values ​​at each time point are summed and subtracted from the trend decay threshold to obtain the value of the first time point. The initial trend accumulation at time step; determining the first... Is the initial trend accumulation at time step 0 greater than 0? If so, then... The initial trend accumulation at time 1 is given to the first The cumulative trend risk at any given moment; if not, assign 0 to the first... The cumulative amount of trend risk at any given moment. Number the time; S42. Set the initial value of the cumulative oscillation risk to 0, and determine the first oscillation deviation term. If the oscillation deviation value at time step is greater than 0, then the 1st step... The oscillation deviation value at any given time is retained; otherwise, 0 is assigned to the first value. The deviation value of the oscillation at time; will the first The cumulative oscillation risk at time point and the adjusted oscillation deviation term. The oscillation deviations at each time point are summed and subtracted from the oscillation decay threshold to obtain the first value. The initial cumulative oscillation at time t; determining the t... Is the initial cumulative oscillation at time step 0 greater than 0? If so, then... The initial oscillation accumulation at time t gives the first The cumulative oscillation risk at time point; if not, assign 0 to the first... The cumulative amount of oscillation risk at any given moment.

[0037] In this embodiment, the expression for the cumulative trend risk is: , in, For the upstream The first tributary The cumulative amount of trend risk at any given moment. To obtain the maximum value, For the upstream The first tributary The cumulative amount of trend risk at any given moment. For the upstream The trend deviation term of the tributary Deviation from trend at any time This is the threshold for trend decay.

[0038] The expression for the cumulative oscillation risk is: , , in, For the upstream The oscillation deviation term after adjustment of the branch is the first Constant oscillation deviation value, For the upstream The first tributary Accumulated risk of constant oscillations For the upstream The first tributary Accumulated risk of constant oscillations This is the threshold for oscillation decay.

[0039] In this embodiment, the trend decay threshold and oscillation decay threshold Set as an empirical value, trend decay threshold Setting it to 0.015 allows trend risk to accumulate more persistently, setting a threshold for oscillation decay. Setting it to 0.02 allows small oscillation risks to dissipate more quickly.

[0040] This invention accumulates trend deviation values ​​and combines them with a trend decay threshold constraint, while accumulating oscillation deviation values ​​by retaining only the positive abnormal components and combining them with an oscillation decay threshold constraint, thus forming the accumulated amount of trend risk and the accumulated amount of oscillation risk respectively. The maximum value constraint is used to make the accumulated amount non-negative, avoiding interference from negative deviations in risk assessment, thereby achieving effective accumulation and appropriate decay of abnormal deviation signals. It can both continuously retain the cumulative effect of historical abnormal risks and filter out the ineffective accumulation caused by small random fluctuations through the decay threshold.

[0041] In this embodiment, S5 includes the following sub-steps: S51. Calculate the confluence delay time of each upstream tributary to the reservoir: ,in, For the upstream The confluence delay time of the tributaries For the upstream The distance from the monitoring point of the tributary to the reservoir, For the upstream The average flow velocity of the tributary The sampling interval; S52. Based on the confluence delay time, determine the confluence time corresponding to each tributary, and calculate the trend flow contribution weight and oscillation flow contribution weight of each upstream tributary at that confluence time. S53. The trend flow contribution weight and the cumulative trend risk are weighted and summed to obtain the trend comprehensive risk index. The oscillation flow contribution weight and the cumulative oscillation risk are weighted and summed to obtain the oscillation comprehensive risk index.

[0042] In this embodiment, the process of calculating the trend flow contribution weight in S52 includes: summing the current trend items of all upstream tributaries at the confluence time to obtain the total flow, and the ratio of the current trend item flow of each tributary to the total flow is the trend flow contribution weight. ,in, For the upstream The first tributary The weight of traffic contribution in real time trends For the upstream The current trend of the tributary Real-time flow The number of upstream tributaries; The process of calculating the oscillating flow contribution weight includes: summing the current oscillation terms of all upstream tributaries at the confluence time to obtain the total flow; the ratio of the current oscillation term flow of each tributary to the total flow is the oscillation flow contribution weight. ,in, For the upstream The first tributary The weight of the oscillating flow contribution at all times. For the upstream The current oscillation term of the branch Real-time traffic.

[0043] The expression for the trend composite risk index obtained in S53 is as follows: , in, For the first The overall risk index of the trend at any given moment. For the upstream The first tributary The weight of traffic contribution in real time trends For the upstream The first tributary The cumulative amount of trend risk at any given moment. The number of upstream tributaries, For the convergence moment, For the tributary number, Number the time; The expression for the oscillation comprehensive risk index obtained in S53 is as follows: , in, For the first The comprehensive risk index of oscillation at any given moment. For the upstream The first tributary The weight of the oscillating flow contribution at all times. For the upstream The first tributary Accumulated risk of constant oscillation.

[0044] In this embodiment, S6 includes the following sub-steps: S61, according to the... The trend of the comprehensive risk index at any moment and the first The risk trend factor is calculated by taking the difference between the overall risk index at different times. Number the time; S62. Enhance the trend composite risk index using risk trend factors to obtain the trend change intensity; S63. Set a short-term impact window to filter the largest flow rate from the short-term impact window of the current oscillation term; S64. Add up the maximum flow rates of all upstream tributaries to obtain the total impact flow rate; S65. Calculate the impact factor based on the total impact flow rate; S66. The oscillation comprehensive risk index is enhanced by using the shock factor to obtain the oscillation mutation intensity.

[0045] In this embodiment, for tributaries , its in The water flow that departs at any moment will Arriving at the reservoir in a timely manner, therefore, for tributaries tributary monitoring time Based on this, the maximum flow rate within the short-term impact window preceding the current moment is selected from the current oscillation term. The short-term impact window has a duration of 2 hours, corresponding to 4 flow rates, and the maximum flow rate within the current oscillation term is obtained. The invention first filters the maximum flow rate of each tributary to capture the cumulative impact effect of multiple tributaries reaching their peaks successively, while enhancing the sensitivity to extreme anomalies and avoiding peak value errors and risk underestimation.

[0046] In this embodiment, the formula for calculating the risk trend factor is: , in, For the first Risk trend factors at any time For the first The overall risk index of the trend at any given moment. For the first The overall risk index of the trend at any given moment. Here, | represents the time number, and | represents the absolute value operation; The formula for calculating the impact factor is: , in, For the first The impact factor of a moment For the first Total impact flow at any given moment This is a preset threshold for the total impact flow. For the first The total impact flow arriving at the reservoir at a given time is calculated as follows: for each upstream tributary, at its confluence time... The maximum value of the oscillation term of the tributary is selected within the previous short-term impact window, and then the maximum values ​​of all tributaries are summed to obtain the sum. . The threshold value is determined based on the upper limit of the total impact flow under normal fluctuation conditions in the same historical period.

[0047] In this embodiment, the expression for the trend mutation intensity is: , in, For the first The intensity of the trend change at any given moment. For the first The overall risk index of the trend at any given moment. For the first Risk trend factors at any given time; The expression for the intensity of the oscillation mutation is: , in, For the first The intensity of the oscillation change at time, For the first The comprehensive risk index of oscillation at any given moment. For the first The impact factor of a moment.

[0048] This invention constructs a normalized risk trend factor by introducing the time-series difference of the trend comprehensive risk index, which adaptively enhances the trend comprehensive risk index. This effectively highlights the abrupt changes in continuously rising risk and improves the sensitivity of identifying slowly developing abnormal trends. At the same time, by setting a short-term impact window to extract the maximum flow of tributaries and obtain the total impact flow, a normalized impact factor is constructed based on the total impact flow and a threshold value to enhance the oscillation comprehensive risk index and capture the drastic risk changes brought about by short-term sudden flow impacts. This invention can reflect the level of risk accumulation and amplify the magnitude of abnormal changes, making the characteristics of risk mutations more prominent.

[0049] when When the index is less than 0, the overall trend risk index shows a downward trend, which weakens the intensity of trend abrupt changes and reduces the tendency to issue early warnings; when... When the overall trend risk index is greater than 0, it shows an upward trend, which strengthens the intensity of trend abrupt changes and increases the likelihood of issuing a warning. When the total impact flow is less than 0, it is below the preset threshold, which weakens the intensity of the oscillation change and reduces the warning tendency; when When the total impact flow is greater than 0, it exceeds the preset threshold, which enhances the intensity of oscillation changes and increases the warning tendency.

[0050] In this embodiment, S7 includes the following sub-steps: S71. Map the intensity of trend mutation to the probability of trend mutation, and map the intensity of oscillatory mutation to the probability of oscillatory mutation; S72. Weight the probability of trend change and the probability of oscillation change to obtain the comprehensive warning probability, and output the warning level based on the comprehensive warning probability.

[0051] This invention uses the tanh function to map the intensity of trend mutations and the intensity of oscillatory mutations to the probabilities of trend mutations and oscillatory mutations, respectively: and ,in, For the first The probability of a sudden change in trend at any given time. For the first The probability of oscillation mutation at time t, For mapping functions, and The preset mapping steepness parameters are all greater than 0. ,in, For the first The overall probability of a warning at any given moment. The preset fusion weight coefficients, , and The specific setup should be based on experiments or experience. and The correlation between the intensity of trend / oscillation mutations in historical data and known risk events is used to calibrate the probability mapping to close to 0 during normal fluctuations and close to 1 during high risks. The value is determined based on the main risk type of the reservoir. If the risk is mainly continuous flood, a larger value (e.g., 0.7) is taken; if the risk is mainly short-term rainstorm impact, a smaller value (e.g., 0.3) is taken.

[0052] In this embodiment, 0 corresponds to no risk and 1 corresponds to extreme risk. The grading thresholds are set as shown in Table 1.

[0053] Table 1 This invention decomposes the flow sequence into trend terms and oscillation terms, and compares them with historical benchmarks for the same period. This effectively distinguishes normal fluctuations such as seasonal floods from true abnormal changes, reduces false alarms and missed alarms, and significantly improves the accuracy of abnormal change early warning.

[0054] This invention takes into account the differences in the confluence delay of each tributary from the reservoir, determines the confluence time by the confluence delay and calculates the flow contribution weight, so that the warning trigger time matches the actual time when the sudden change arrives at the reservoir, thereby improving the timeliness and accuracy of the warning.

[0055] This invention designs a risk trend factor for trend risk, utilizing the time-series difference of the trend comprehensive risk index to identify the accelerated process of risk accumulation, thereby enhancing sensitivity to "persistent anomalies." For oscillation risk, it designs an impact factor, independently selecting the maximum value of the oscillation term flow within the short-term impact window of each tributary and summing them to obtain the total impact flow, then calculating the impact factor to amplify the response to "short-term strong impacts." The differentiated design of these two types of enhancement factors enables the early warning system to have sensitive detection capabilities for both slowly developing trend-type abrupt changes and sudden oscillating abrupt changes.

[0056] This invention combines the intensity of trend mutations and the intensity of oscillation mutations to obtain a comprehensive early warning probability, thereby improving the accuracy of reservoir mutation early warning.

[0057] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for early warning of reservoir abrupt changes based on time-series hydraulic data, characterized in that, Includes the following steps: S1. Decompose the flow of each upstream tributary of the reservoir in the current time period and in the historical time period to obtain the current trend term, the current oscillation term, the historical year trend term, and the historical year oscillation term; S2. Take the average of the trend terms for each historical year to obtain the baseline trend term, and take the average of the oscillation terms for each historical year to obtain the baseline oscillation term; S3. Obtain the trend deviation term based on the difference between the current trend term and the benchmark trend term; obtain the oscillation deviation term based on the difference between the current oscillation term and the benchmark oscillation term. S4. Obtain the cumulative trend risk for the trend deviation item and the cumulative oscillation risk for the oscillation deviation item; S5. Determine the confluence time based on the confluence delay, calculate the flow contribution weight, and sum the cumulative amount of trend and oscillation risks to obtain the comprehensive risk index of trend and oscillation. S6. Based on the risk trend factor and the shock factor, the comprehensive risk index of trend and oscillation is enhanced to obtain the trend and oscillation change intensity. S7. Map the trend and the intensity of oscillation changes to corresponding probabilities, weight them to obtain the comprehensive early warning probability and output the early warning level.

2. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 1, characterized in that, S1 includes the following steps: S11. Extract the flow rate of each upstream tributary of the reservoir in the current time period and in historical time periods to obtain the flow rate in the current time period and the flow rate in the same time period over many years in history; S12. Smooth the traffic flow in the current time period and the traffic flow in the same time period of each historical year to obtain the current trend item and the trend item of each historical year. S13. Subtract the current trend term from the current time period flow to obtain the current oscillation term; S14. Subtract the trend term of each historical year from the flow rate of the same time period in each historical year to obtain the oscillation term of each historical year.

3. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 1, characterized in that, The specific process of S3 includes: in the same tributary upstream, the current trend term... The first item in the time flow and baseline trend item Subtracting the flow rates at different times, we obtain the trend term. The difference in flow rate at different times will be used as the trend item. The difference in flow rate at different times and the first term in the baseline trend. The ratio of flow rate at time step is used as the first... The trend deviation value at each time point constitutes the trend deviation term. Number the time; In the same tributary upstream, the th term in the current oscillation term will be... The first term in the time-current flow and reference oscillation term Subtracting the flow rates at each moment, we obtain the oscillation term. The difference in flow rate at different times will be the oscillation term. The difference in flow rate at time and the first term in the reference oscillation term The ratio of flow rate at time step is used as the first... The oscillation deviation value at each time point constitutes the oscillation deviation term.

4. The reservoir abrupt change early warning method based on time-series hydraulic data according to claim 1, characterized in that, S4 includes the following steps: S41. Set the initial value of the trend risk accumulation to 0; [The sentence is incomplete and requires further context.] The cumulative amount of trend risk at time point and the trend deviation term in the first term The trend deviation values ​​at each time point are summed and subtracted from the trend decay threshold to obtain the value of the first time point. The initial trend accumulation at time step; determining the first... Is the initial trend accumulation at time step 0 greater than 0? If so, then... The initial trend accumulation at time 1 is given to the first The accumulated trend risk at any given moment; if not, assign 0 to the first... The cumulative amount of trend risk at any given moment. Number the time; S42. Set the initial value of the cumulative oscillation risk to 0, and determine the first item in the oscillation deviation term. If the oscillation deviation value at time point is greater than 0, then the 1st... The oscillation deviation value at any given time is retained; otherwise, 0 is assigned to the first value. The deviation value of the oscillation at time; will the first The cumulative oscillation risk at time point and the adjusted oscillation deviation term. The oscillation deviations at each time point are summed and subtracted from the oscillation decay threshold to obtain the first value. The initial cumulative oscillation at time t; determining the t... Is the initial cumulative oscillation at time step 0 greater than 0? If so, then... The initial oscillation accumulation at time t gives the first The cumulative oscillation risk at time point; if not, assign 0 to the first... The cumulative amount of oscillation risk at any given moment.

5. The reservoir abrupt change early warning method based on time-series hydraulic data according to claim 1, characterized in that, S5 includes the following steps: S51. Calculate the confluence delay time of each upstream tributary to the reservoir; S52. Based on the confluence delay time, determine the confluence time corresponding to each tributary, and calculate the trend flow contribution weight and oscillation flow contribution weight of each upstream tributary at that confluence time. S53. The trend flow contribution weight and the cumulative trend risk are weighted and summed to obtain the trend comprehensive risk index. The oscillation flow contribution weight and the cumulative oscillation risk are weighted and summed to obtain the oscillation comprehensive risk index.

6. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 5, characterized in that, The process of calculating the trend flow contribution weight in S52 includes: summing the current trend items of all upstream tributaries at the confluence time to obtain the total flow, and the ratio of the current trend item flow of each tributary to the total flow is the trend flow contribution weight. The process of calculating the contribution weight of oscillating flow includes: summing the current oscillations of all upstream tributaries at the confluence time to obtain the total flow, and the ratio of the current oscillation of each tributary to the total flow is the contribution weight of oscillating flow. The expression for the trend composite risk index obtained in S53 is as follows: ， in, For the first The overall risk index of the trend at any given moment. For the upstream The first tributary The contribution weight of real-time trend traffic. For the upstream The first tributary The cumulative amount of trend risk at any given moment. The number of upstream tributaries, For the convergence moment, For the tributary number, Number the time; The expression for the oscillation comprehensive risk index obtained in S53 is as follows: ， in, For the first The comprehensive risk index of oscillation at any given moment. For the upstream The first tributary The contribution weight of oscillating flow at all times. For the upstream The first tributary Accumulated risk of constant oscillation.

7. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 1, characterized in that, S6 includes the following steps: S61, according to the... The trend of the comprehensive risk index at any moment and the first The risk trend factor is calculated by taking the difference between the overall risk index at different times. Number the time; S62. Enhance the trend composite risk index using risk trend factors to obtain the trend change intensity; S63. Set a short-term impact window to filter the largest flow rate from the short-term impact window of the current oscillation term; S64. Add up the maximum flow rates of all upstream tributaries to obtain the total impact flow rate; S65. Calculate the impact factor based on the total impact flow rate; S66. The oscillation comprehensive risk index is enhanced by using the shock factor to obtain the oscillation mutation intensity.

8. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 1 or 7, characterized in that, The formula for calculating the risk trend factor is: ， in, For the first Risk trend factors at any time For the first The overall risk index of the trend at any given moment. For the first The overall risk index of the trend at any given moment. Here, | represents the time number, and | represents the absolute value operation; The formula for calculating the impact factor is: ， in, For the first The impact factor of time, For the first Total impact flow at any given moment This is the threshold for the total impact flow.

9. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 1 or 7, characterized in that, The expression for obtaining the trend abrupt change intensity is: ， in, For the first The intensity of the trend change at any given moment. For the first The overall risk index of the trend at any given moment. For the first Risk trend factors at any given time; The expression for the intensity of the oscillation mutation is: ， in, For the first The intensity of the oscillation change at time, For the first The comprehensive risk index of oscillation at any given moment. For the first The impact factor of a moment.

10. The reservoir sudden change early warning method based on time-series hydraulic data according to claim 1, characterized in that, S7 includes the following steps: S71. Map the intensity of trend mutation to the probability of trend mutation, and map the intensity of oscillatory mutation to the probability of oscillatory mutation; S72. Weight the probability of trend change and the probability of oscillation change to obtain the comprehensive warning probability, and output the warning level based on the comprehensive warning probability.

Images