Prediction method for events influencing cascade reservoir dispatching operation

Abstract

The invention relates to a method for predicting events influencing cascade reservoir dispatching operation, which comprises the following steps: acquiring real-time monitoring data of a cascade reservoir to be detected, inputting the real-time monitoring data of the cascade reservoir to be detected into a water conservancy professional model, and acquiring prediction data corresponding to the real-time monitoring data of the cascade reservoir to be detected; obtaining meteorological prediction data, inputting the meteorological prediction data and prediction data corresponding to the real-time monitoring data of the cascade reservoir to be detected into the trained data analysis model, and obtaining a monitoring result; the monitoring result comprises a key factor, an event tag and an event occurrence probability; the data analysis model comprises a key factor determination unit, an event classification determination unit and an event occurrence probability determination unit; and performing event early warning based on the monitoring result and a preset early warning rule. According to the invention, the safety and reliability of reservoir dispatching operation can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir dispatching and operation, and in particular to a method for predicting events affecting the dispatching and operation of cascade reservoirs. Background Art

[0002] During the operation of cascade reservoirs, reservoir management departments often face challenges from various emergencies, such as natural disasters and equipment failures. The occurrence of these events is often difficult to predict, and the existing handling methods mainly rely on post-event recording and archiving, lacking systematic management and analysis methods. This passive response method not only fails to effectively prevent the occurrence of events, but may also lead to a delay in scheduling decisions, increasing the risk and uncertainty of reservoir scheduling.

[0003] At present, the prediction of some events mainly relies on time factors, such as estimating based on seasonal changes or historical data. However, this prediction method is not very accurate and cannot timely and accurately predict the specific time and impact range of emergencies, making it difficult to provide strong support for emergency preparedness arrangements.

[0004] Therefore, there is an urgent need for a prediction method for events that affect the operation of cascade reservoirs. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method for predicting events that affect the scheduling and operation of cascade reservoirs, which solves the technical problems that the prior art cannot effectively prevent the occurrence of events and may also cause delays in scheduling decisions, thereby increasing the risks and uncertainties in the scheduling and operation of cascade reservoirs.

[0007] (II) Technical solution

[0008] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0009] The embodiment of the present invention provides a method for predicting events that affect the operation of cascade reservoirs, including:

[0010] S100, obtaining real-time monitoring data of the cascade reservoirs to be detected, inputting the real-time monitoring data of the cascade reservoirs to be detected into a water conservancy professional model, and obtaining prediction data corresponding to the real-time monitoring data of the cascade reservoirs to be detected;

[0011] The real-time monitoring data includes: hydrological data, power data and engineering safety monitoring data; the water conservancy professional model includes: hydrological and hydrodynamic prediction sub-model, business scenario forecast sub-model, power market power load forecast sub-model and wind and solar pumped storage forecast dispatch sub-model;

[0012] S200, obtaining meteorological forecast data, inputting the meteorological forecast data and forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into a trained data analysis model, and obtaining monitoring results;

[0013] The monitoring results include: key factors, event labels and event occurrence probability; the data analysis model includes: a key factor determination unit, an event classification determination unit and an event occurrence probability determination unit;

[0014] S300: Based on the monitoring results and pre-set warning rules, an event warning is issued.

[0015] Optionally, before S100, the step further includes:

[0016] S000. Collect historical event data, use the historical event data to train the data analysis model, and obtain a trained data analysis model.

[0017] Optionally, the historical event data includes:

[0018] Event type, time of event, location of event, scope of event impact, control measures for the event, corresponding meteorological data, hydrological data, power data and engineering safety monitoring data when the event occurred.

[0019] Optionally, the S000 specifically includes:

[0020] S010, classifying the historical event data according to event type, event object, and event type name, and assigning a label to each historical event data based on the classification result;

[0021] S020. Input each data in the classified historical event data into the corresponding data analysis model, use the cross entropy function and Adam optimizer, with a learning rate of 0.001, until the cross entropy regression loss function is minimized, and obtain the trained data analysis model.

[0022] Optionally, the S200 includes:

[0023] S21 0. Input the meteorological forecast data and the forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into the trained key factor determination unit to obtain the corresponding key factors;

[0024] S220, inputting the corresponding key factors into the event classification determination unit to obtain an event label; the event label includes: event type, event object, and event type name;

[0025] S230: Input the corresponding key factors and event labels into an event occurrence probability determination unit to obtain the event occurrence probability.

[0026] Optionally, in S210, the corresponding key factor is:

[0027]

[0028] Among them, Key Factors is the key factor, X n For each prediction data, is the estimated value of the regression coefficient for each prediction data, is the standard error of each predicted data; T is the confidence level.

[0029] Optionally, in S220,

[0030] The event classification determination unit is a convolutional neural network model; the event classification determination unit includes: an input layer, a hidden layer, and an output layer; wherein the input layer inputs the key factors corresponding to the predicted data; the hidden layer includes multiple neurons, which are used to obtain the event labels corresponding to the predicted data through the sigmoid activation function; and the output layer outputs the event labels corresponding to the predicted data.

[0031] Optionally, in S230,

[0032] The event occurrence probability unit includes a plurality of event occurrence probability subunits;

[0033] Wherein, according to the event label, the corresponding event occurrence probability subunit is obtained; the event occurrence probability subunit is:

[0034]

[0035] Among them, β0 is the intercept constant, β i is the i-th key factor K i The coefficient of i is the i-th key factor K i The quadratic coefficient, δ ij is the interaction coefficient between the i-th and j-th key factors, u is the random effect constant, K i is the i-th key factor, K j is the jth key factor, m is the total number of key factors, P(Y=1|K1,K2,…,K m ) is the probability of the event occurring.

[0036] Optionally, the method further comprises:

[0037] Regular warnings are set according to the periodic events of the dry season, the flood season and the normal water season; the regular warnings are determined according to the start and end times of the dry season, the flood season and the normal water season.

[0038] (III) Beneficial effects

[0039] The beneficial effects of the present invention are as follows: the method for predicting events that affect the scheduling and operation of cascade reservoirs of the present invention, due to the adoption of a key factor determination unit, an event classification determination unit, and an event occurrence probability determination unit, can forecast and warn possible events in advance compared to the prior art, providing management personnel with sufficient events for emergency preparation, thereby reducing the impact of events on the scheduling and operation of reservoirs, and improving the safety and reliability of the scheduling and operation of cascade reservoirs; at the same time, by using historical data for subsequent model training, a systematic and scientific event identification and forecasting method is provided for reservoir management departments, which has important practical value and promotion significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic flow chart of a method for predicting events affecting the scheduling and operation of cascade reservoirs according to Embodiment 1 of the present invention;

[0041] Figure 2 This is a schematic diagram of different time components of task execution in periodic warning in Example 2 of the present invention. DETAILED DESCRIPTION

[0042] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0043] The embodiment of the present invention proposes a method for predicting events that affect the scheduling and operation of cascade reservoirs, which solves the technical problems that the prior art cannot effectively prevent the occurrence of events and may also cause delays in scheduling decisions, increasing the risks and uncertainties in the scheduling and operation of cascade reservoirs. It improves the safety and reliability of the scheduling and operation of cascade reservoirs, and also provides a systematic and scientific time identification and forecasting method for reservoir management departments, which has important use value and promotion significance.

[0044] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0045] Specific embodiment description part

[0046] Example 1

[0047] See also Figure 1 A method for predicting events affecting the operation of cascade reservoirs according to an embodiment of the present invention includes:

[0048] Step S100, obtaining real-time monitoring data of the cascade reservoirs to be detected, inputting the real-time monitoring data of the cascade reservoirs to be detected into a water conservancy professional model, and obtaining prediction data corresponding to the real-time monitoring data of the cascade reservoirs to be detected;

[0049] The real-time monitoring data includes: hydrological data, power data and engineering safety monitoring data; the water conservancy professional model includes: hydrological and hydrodynamic prediction sub-model, business scenario forecast sub-model, power market power load forecast sub-model and wind and solar pumped storage forecast dispatch sub-model;

[0050] Step S200, obtaining meteorological forecast data, inputting the meteorological forecast data and the forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into the trained data analysis model, and obtaining the monitoring results;

[0051] The monitoring results include: key factors, event labels and event occurrence probability; the data analysis model includes: a key factor determination unit, an event classification determination unit and an event occurrence probability determination unit;

[0052] Step S300: Provide event warning based on monitoring results and pre-set warning rules.

[0053] In the specific implementation process, hydrological data include water level, flow rate monitored by hydrological stations, inflow and outflow, water quality parameters, etc., power data include power generation, power generation efficiency, transmission capacity, load conditions, etc., and engineering safety monitoring data include displacement data, seepage data, crack data, stress and strain data, vibration data, etc. The above data are collected and obtained through various sensors and monitoring equipment installed in the cascade reservoir basin.

[0054] Meteorological forecast data includes forecast data of temperature, humidity, wind speed, wind direction, rainfall, etc., which are obtained through weather forecast.

[0055] This embodiment obtains a data analysis model including a key factor determination unit, an event classification determination unit, and an event occurrence probability determination unit by training historical data. By inputting the prediction data corresponding to the real-time monitoring data into the data analysis model, the events that affect the scheduling and operation of cascade reservoirs are predicted. It can forecast the probability of event occurrence in advance. When the probability of event occurrence is too high, a corresponding response plan can be quickly formulated so that it can be dealt with quickly and effectively when the event occurs, thereby reducing the impact of emergencies on reservoir scheduling and operation and improving the safety and reliability of cascade reservoir scheduling.

[0056] Example 2

[0057] A method for predicting events that affect the operation of cascade reservoirs in this embodiment includes:

[0058] Step S100, obtaining real-time monitoring data of the cascade reservoirs to be detected, inputting the real-time monitoring data of the cascade reservoirs to be detected into a water conservancy professional model, and obtaining prediction data corresponding to the real-time monitoring data of the cascade reservoirs to be detected;

[0059] The real-time monitoring data includes: hydrological data, power data and engineering safety monitoring data; the water conservancy professional model includes: hydrological and hydrodynamic prediction sub-model, business scenario forecast sub-model, power market power load forecast sub-model and wind and solar pumped storage forecast dispatch sub-model;

[0060] Step S200, obtaining meteorological forecast data, inputting the meteorological forecast data and the forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into the trained data analysis model, and obtaining the monitoring results;

[0061] The monitoring results include: key factors, event labels and event occurrence probability; the data analysis model includes: a key factor determination unit, an event classification determination unit and an event occurrence probability determination unit;

[0062] Step S300: Provide event warning based on monitoring results and pre-set warning rules.

[0063] In this embodiment, before step S100, the following steps are also included:

[0064] Step S000: collect historical event data, use the historical event data to train the data analysis model, and obtain the trained data analysis model.

[0065] In the specific implementation process, historical event data includes:

[0066] Event type, time of event, location of event, scope of event impact, control measures for the event, corresponding meteorological data, hydrological data, power data and engineering safety monitoring data when the event occurred.

[0067] In this embodiment, after the historical event data is collected, the historical event data is input into a pre-set database to form a historical event database.

[0068] In the specific implementation process, the event can be traced back based on the selected historical event, and the changing rules of the event process can be intuitively displayed with icons; after selecting an event, the meteorological data, hydrological data, power data and engineering safety monitoring data corresponding to the event period can be viewed.

[0069] In this embodiment, step S000 specifically includes:

[0070] Step S010: classify the historical event data according to the event type, the object to which the event belongs, and the event type name, and assign a label to each historical event data based on the classification result;

[0071] In the specific implementation process, a three-level tree is used to classify events, and event types are divided into typical events and daily events. The object to which the event belongs is the reservoir where the event occurs, and the event type name is the specific occurrence at the time. After the classification is completed, the event type, the object to which the event belongs, and the event type name are all marked in the form of labels on the corresponding historical event data.

[0072] For example, suppose that doors and windows downstream of a reservoir vibrate. At this time, the event type is first determined as a typical event, then the object described by the event is determined, that is, the location where the event occurred, and finally the event type name is determined as door and window vibration. At this time, the meteorological data, hydrological data, power data and engineering safety monitoring data corresponding to the event are all marked with the labels "typical data, a reservoir (location of the event), downstream door and window vibration".

[0073] Step S020: Input each data in the classified historical event data into the corresponding data analysis model, use the cross entropy function and Adam optimizer, with a learning rate of 0.001, until the cross entropy regression loss function is minimized, and obtain a trained data analysis model.

[0074] In this embodiment, step S200 includes:

[0075] Step S210, inputting the meteorological forecast data and the forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into the trained key factor determination unit to obtain the corresponding key factors;

[0076] Step S220: inputting the corresponding key factors into the event classification determination unit to obtain an event label;

[0077] Event tags include: event type, event object, and event type name;

[0078] Step S230: input the corresponding key factors and event tags into the event occurrence probability determination unit to obtain the event occurrence probability.

[0079] In this embodiment, in step S210, the corresponding key factors are:

[0080]

[0081] Among them, Key Factors is the key factor, Xn For each prediction data, is the estimated value of the regression coefficient for each prediction data, is the standard error of each predicted data; T is the confidence level.

[0082] Among them, the standard error of each prediction data is:

[0083]

[0084] in, is the standard error of each predicted data, σ 2 is the residual variance, X n For each real-time prediction data, X n is the sample mean corresponding to each prediction data.

[0085] For example, for each prediction data X n , calculate the estimated value of its regression coefficient and standard error If in the forecast data, the estimated regression coefficient of water level is The standard error is And the confidence level T is set to 1.96 (corresponding to a 95% confidence interval), then The water level is the key factor at this time.

[0086] In step S220,

[0087] The event classification determination unit is a convolutional neural network model; the event classification determination unit includes: an input layer, a hidden layer, and an output layer; wherein the input layer inputs key factors corresponding to the predicted data; the hidden layer includes multiple neurons, which are used to obtain event labels corresponding to the predicted data through a sigmoid activation function; and the output layer outputs the event labels corresponding to the predicted data.

[0088] In this embodiment, in step S230, the event occurrence probability unit includes a plurality of event occurrence probability sub-units;

[0089] Among them, according to the event label, the corresponding event occurrence probability subunit is obtained; the event occurrence probability subunit is:

[0090]

[0091] Among them, β0 is the intercept constant, β i is the i-th key factor K i The coefficient of i is the i-th key factor K i The quadratic coefficient, δ ij is the interaction coefficient between the i-th and j-th key factors, u is the random effect constant, Ki is the i-th key factor, K j is the jth key factor, m is the total number of key factors, P(Y=1|K1,K2,…,K m ) is the probability of the event occurring.

[0092] In this embodiment, the pre-set warning rules are:

[0093] When the probability of an event occurring is greater than or equal to 80%, a first-level warning is issued; when the probability of an event occurring is greater than or equal to 50% and less than 80%, a second-level warning is issued; when the probability of an event occurring is less than 50%, no warning is issued; the first-level warning is an acoustic and visual warning, and the second-level warning is a sound warning.

[0094] For example, suppose that the key factors of an event have been obtained through the trained key factor determination unit as the water level change rate, the rainfall at the hydrological station in the catchment area, and the upstream water flow, and the corresponding water level change rate is 0.03m / h, the upstream rainfall is 30mm / h, and the upstream water flow is 120m 3 / s, that is, K1 = 30mm / h, K2 = 30mm / h, K3 = 120m 3 / s;

[0095] At the same time, the following parameters have been obtained through historical data analysis and model training:

[0096] The intercept constant β0 = -4; the coefficients of the first-order terms are β1 = 0.03, β2 = 0.03, β3 = 0.0001; the coefficients of the second-order terms are γ1 = 0.0003, γ2 = 0.0003, γ3 = 0.000002; the interaction coefficient δ 12 =0.00003,δ 13 =0.0000002,δ 23 =0.0000002; the random effect constant u can be ignored or set to 0;

[0097] Substituting the above parameters into the formula, the final probability of the event occurring is 79%, at which time a second-level warning is issued.

[0098] In the specific implementation process, the degree of harm of the incident also needs to be considered;

[0099] If the probability of an event is greater than 80%, the hazard level of the event is considered; in specific applications, the hazard level of the event is obtained by multiplying the probability of the event, the exposure frequency, and the severity score. The hazard level of the event is divided into four levels, namely red warning, orange warning, yellow warning, and blue warning, corresponding to different colors of flash warnings.

[0100] In a specific implementation process, the method of the present invention further includes:

[0101] Regular warnings are set according to the periodic events of the dry season, the flood season and the normal water season; the regular warnings are determined according to the start and end times of the dry season, the flood season and the normal water season.

[0102] In this embodiment, the time warning is configured according to the cron expression specification to configure the frequency of scheduled task execution. A standard cron expression contains up to 6 fields, which represent different time components of character execution, such as Figure 2 shown.

[0103] Each field has a set of allowed values ​​that indicate the time unit in which the action will be executed. You can specify it using a single value, a series of values, or wildcards. For example, the allowed values ​​for the minute field are 0 to 59, the allowed values ​​for the hour field are 0 to 23, the allowed values ​​for the day field are 1 to 31, and the allowed values ​​for the month field are 1 to 12.

[0104] In addition, in cron expressions, "*" represents all possible values. For example, setting "*" in the minute field means that it will be triggered every minute.

[0105] "," can be used to separate multiple values. For example, setting "1, 3, 4, 7" in the hour field means that it will be triggered once at 1:00, 3:00, 4:00, and 7:00.

[0106] “-” can be used to indicate a range. For example, setting “5-10” on the day field means that the event will be triggered every day from the 5th to the 10th.

[0107] " / " can be used to indicate the step size. For example, setting "0 / 15" in the minute field means triggering every 15 minutes, while "* / 15" means triggering every 15 minutes starting from 0.

[0108] “?” can only be used in date and week fields, indicating that no value is specified and it can be left blank. It is actually equivalent to a placeholder.

[0109] “#” can only be used in the week field to indicate the day of the week, for example, “6#3” means the third Friday (6 means Friday, 3 means the third).

[0110] For example, the official flood season of a reservoir lasts from May 1 to October 31 every year. At 8 a.m. on May 1 every year, the staff can be reminded that the flood season has begun through sound and light warnings, text messages, emails, etc.; during the flood season, the staff can be reminded through the above methods at 8 a.m. every day to check the safety status and monitoring data of the reservoir; at 8 a.m. on October 31 every year, the staff can be reminded again through the above methods that the flood season is over and the work results during the flood season can be summarized.

[0111] By setting up a regular early warning mechanism, it is possible to ensure that staff can take necessary preventive measures and preparatory work at critical time points, promptly discover and resolve potential problems, reduce the probability of accidents, and effectively improve the safety and operational efficiency of cascade reservoirs.

[0112] The present embodiment provides a method for predicting events that affect the operation of cascade reservoirs. By summarizing and organizing historical events and historical event data, and using the historical event data to train a data analysis model, it can achieve unified management and analysis of events, and at the same time can significantly provide a scientific data analysis model to identify and forecast events, so as to quickly make forecasts before events occur, provide staff with sufficient events for emergency preparation, and reduce the impact of events on reservoir operation. It has important use value and promotion significance.

[0113] In the description of the present invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0114] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0115] In the present invention, unless otherwise clearly specified and limited, when a first feature is “on” or “below” a second feature, it may be that the first and second features are in direct contact, or that the first and second features are in indirect contact through an intermediate medium. Moreover, when a first feature is “above”, “above” or “above” a second feature, it may be that the first feature is directly above or obliquely above the second feature, or it may simply mean that the first feature is higher in level than the second feature. When a first feature is “below”, “below” or “below” a second feature, it may be that the first feature is directly below or obliquely below the second feature, or it may simply mean that the first feature is lower in level than the second feature.

[0116] In the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0117] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may alter, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for predicting events affecting the operation of cascade reservoirs, characterized in that: include: S100, obtaining real-time monitoring data of the cascade reservoirs to be detected, inputting the real-time monitoring data of the cascade reservoirs to be detected into a water conservancy professional model, and obtaining prediction data corresponding to the real-time monitoring data of the cascade reservoirs to be detected; The real-time monitoring data includes: hydrological data, power data and engineering safety monitoring data; the water conservancy professional model includes: hydrological and hydrodynamic prediction sub-model, business scenario forecast sub-model, power market power load forecast sub-model and wind and solar pumped storage forecast dispatch sub-model; S200, obtaining meteorological forecast data, inputting the meteorological forecast data and forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into a trained data analysis model, and obtaining monitoring results; The monitoring results include: key factors, event labels and event occurrence probability; the data analysis model includes: a key factor determination unit, an event classification determination unit and an event occurrence probability determination unit; S300: Based on the monitoring results and pre-set warning rules, an event warning is issued.

2. The method for predicting events affecting the operation of cascade reservoirs according to claim 1, characterized in that: The S100 also includes: S000. Collect historical event data, use the historical event data to train the data analysis model, and obtain a trained data analysis model.

3. The method for predicting events affecting the operation of cascade reservoirs according to claim 2, characterized in that: The historical event data includes: Event type, time of event, location of event, scope of event impact, control measures for the event, corresponding meteorological data, hydrological data, power data and engineering safety monitoring data when the event occurred.

4. The method for predicting events affecting the operation of cascade reservoirs according to claim 2, characterized in that: The S000 specifically includes: S01 0. Classify the historical event data according to event type, event object, and event type name, and assign a label to each historical event data based on the classification result; S020. Input each data in the classified historical event data into the corresponding data analysis model, use the cross entropy function and Adam optimizer, with a learning rate of 0.001, until the cross entropy regression loss function is minimized, and obtain the trained data analysis model.

5. The method for predicting events affecting the operation of cascade reservoirs according to claim 1, characterized in that: The S200 includes: S21 0. Input the meteorological forecast data and the forecast data corresponding to the real-time monitoring data of the cascade reservoirs to be detected into the trained key factor determination unit to obtain the corresponding key factors; S220, inputting the corresponding key factors into the event classification determination unit to obtain an event label; the event label includes: event type, event object, and event type name; S230: Input the corresponding key factors and event labels into an event occurrence probability determination unit to obtain the event occurrence probability.

6. The method for predicting events affecting the operation of cascade reservoirs according to claim 5, characterized in that: In S210, the corresponding key factors are: Among them, Key Factors is the key factor, X n For each prediction data, is the estimated value of the regression coefficient for each prediction data, is the standard error of each predicted data; T is the confidence level.

7. The method for predicting events affecting the operation of cascade reservoirs according to claim 5, characterized in that: In the S220, The event classification determination unit is a convolutional neural network model; the event classification determination unit includes: an input layer, a hidden layer, and an output layer; wherein the input layer inputs the key factors corresponding to the predicted data; the hidden layer includes multiple neurons, which are used to obtain the event labels corresponding to the predicted data through the sigmoid activation function; and the output layer outputs the event labels corresponding to the predicted data.

8. The method for predicting events affecting the operation of cascade reservoirs according to claim 5, characterized in that: In the S230, The event occurrence probability unit includes a plurality of event occurrence probability subunits; Wherein, according to the event label, the corresponding event occurrence probability subunit is obtained; the event occurrence probability subunit is: Among them, β0 is the intercept constant, β i is the i-th key factor K i The coefficient of i is the i-th key factor K i The quadratic coefficient, δ ij is the interaction coefficient between the i-th and j-th key factors, u is the random effect constant, K i is the i-th key factor, K j is the jth key factor, m is the total number of key factors, P(Y=1|K1,K2,…,K m ) is the probability of the event occurring.

9. The method for predicting events affecting the operation of cascade reservoirs according to claim 1, characterized in that: The method further comprises: Regular warnings are set according to the periodic events of the dry season, the flood season and the normal water season; the regular warnings are determined according to the start and end times of the dry season, the flood season and the normal water season.

Images