Safety Prediction Method for the Slope of the Reservoir Basin of a Daily Regulating Pumped-Storage Power Station

Abstract

The present invention belongs to the technical field of the operation safety of clean energy, and specifically discloses a method for predicting the safety of the reservoir basin slope of a daily-regulated pumped storage power station, which includes Step 1: Divide the operation cycle of the daily-regulated pumped storage power station into multiple rainless cycles and multiple rainfall cycles; Step 2: Determine a rainless prediction model according to the first monitoring data of the slope during the rainless cycle; Step 3: Determine a rainfall prediction model according to the rainfall data and the second monitoring data of the slope during the rainfall cycle; Step 4: Predict the safety of the reservoir basin slope of the daily-regulated pumped storage power station according to the rainless prediction model and the rainfall prediction model. The prediction method of the present invention can obtain accurate prediction data without a large amount of calculation, and can realize the rapid, economical and safe prediction of the stability of the reservoir basin slope.

Description

Technical Field

[0001] The present invention discloses a safety prediction method for the reservoir basin slope of a daily-regulated pumped-storage power station, belonging to the technical field of the operation safety of clean energy. Background Art

[0002] Pumped storage has three major functions: ensuring the safety of the large power grid, serving the consumption of clean energy, and promoting the optimal operation of the power system. It is currently the most mature, widely used, and economically optimal flexible regulating power source, playing an important role in the process of the clean and low-carbon transformation and development of energy. Pumped storage has entered a period of high-quality and rapid development.

[0003] A pumped-storage power station is a special form of hydropower station. Its reservoir basin is mostly formed by excavation and filling. After excavation, the stability problem of the reservoir basin slope is one of the key concerns. One of the major characteristics of a pumped-storage power station is that its water level rises and falls frequently and has a large change range. The frequently changing water level will have a great impact on the stability of the reservoir basin slope. When the reservoir water level rises, the water will generate buoyancy on the slope rock and soil mass, causing the soil body to soften and reducing the strength of the rock and soil mass; when the reservoir water level drops, it will generate dynamic water pressure on the slope rock and soil mass, thus affecting the stability of the reservoir basin slope. The factors affecting the stability of the pumped-storage reservoir basin slope include not only the amplitude and speed of the water level rise and fall, but also the reservoir basin leakage situation, the change of the slope angle, rainfall, etc. Moreover, the uncertainty of the geological parameters of the slope rock and soil mass makes the slope stability affected by the comprehensive influence of geological factors and engineering factors with uncertainty characteristics such as randomness, fuzziness, and variability.

[0004] Currently, when evaluating the slope stability, the more commonly used methods are based on traditional mechanical theory analysis methods or numerical analysis methods, but these methods all have problems such as "unclear mechanism", "inaccurate parameters", and "inaccurate models", and it is still difficult to accurately predict and forecast the slope stability. In addition, the traditional methods have a large amount of calculation, and the calculation process is long and cumbersome, and it is difficult to meet the long-term stability prediction of the reservoir basin slope involved in pumped-storage projects. Summary of the Invention

[0005] The purpose of the present invention is to provide a safety prediction method for the reservoir basin slope of a daily-regulated pumped-storage power station to solve the technical problems of large calculation amount and low accuracy rate existing in the existing reservoir basin slope prediction methods.

[0006] The present invention provides a safety prediction method for the reservoir basin slope of a daily-regulated pumped-storage power station, including:

[0007] Step 1: Divide the operation cycle of the daily-regulated pumped-storage power station into multiple rainless cycles and multiple rainfall cycles.

[0008] Step 2: Determine a rainless prediction model according to the first monitoring data of the slope within the rainless cycle.

[0009] Step 3: Determine a rainfall prediction model based on rainfall data and second monitoring data of the slope during the rainfall cycle.

[0010] Step 4: Predict the safety of the slope of the reservoir basin of the daily-regulated pumped-storage power station according to the rainless prediction model and the rainfall prediction model.

[0011] Preferably, the rainless cycle includes multiple stages divided according to the water level change trend, and step 2 specifically includes:

[0012] Step 2.1: Determine the characteristic values of the prediction model corresponding to each stage of each rainless cycle according to the first monitoring data, where the characteristic values include the influence parameters of the slope and the water level change speed on the prediction data.

[0013] Step 2.2: Determine the change trend function of the characteristic values in multiple consecutive rainless cycles, and determine the rainless prediction model according to the change trend function.

[0014] Preferably, the stage includes a water level change stage and a water level constant stage; the prediction model of the water level change stage is determined according to the first monitoring data and the water level change speed in the reservoir basin.

[0015] Preferably, the rainfall cycle includes multiple stages divided according to the water level change trend, and step 3 specifically includes:

[0016] Step 3.1: Determine the characteristic values of the prediction model corresponding to each stage of each rainfall cycle according to the rainfall data and the second monitoring data, where the characteristic values include the influence parameter of rainfall on the prediction data and the influence parameter of each change point of the second monitoring data on the prediction data.

[0017] Step 3.2: Determine the change trend function of the characteristic values in multiple consecutive rainfall cycles, and determine the rainfall prediction model according to the change trend function.

[0018] Preferably, step 3.2 specifically includes:

[0019] Determine the change trend function of the characteristic values in multiple consecutive rainfall cycles.

[0020] Determine the random fluctuation function according to the change trend function.

[0021] Determine the rainfall prediction model according to the random fluctuation function.

[0022] Preferably, step 4 specifically includes:

[0023] Step 4.1: Determine the prediction model to be used according to the weather conditions of the operation cycle to be predicted.

[0024] Step 4.2: Predict the safety of the reservoir slope of the daily-regulated pumped-storage power station according to the predicted value of the previous operation cycle corresponding to the operation cycle to be predicted and the prediction model determined in Step 4.1.

[0025] Preferably, both the first monitoring data and the second monitoring data include at least one of displacement monitoring data, anchor bolt stress monitoring data, and anchor bolt strain monitoring data.

[0026] Preferably, Step 1 specifically includes:

[0027] Step 1.1: Perform intelligent inspection on the monitoring data of each operation cycle of the daily-regulated pumped-storage power station. The intelligent inspection includes outlier rejection and missing data supplementation performed in sequence. The outlier rejection is based on the Grubbs criterion. The missing data supplementation is specifically to supplement the missing data using linear interpolation.

[0028] Step 1.2: Divide the operation cycles of the daily-regulated pumped-storage power station into multiple rainless cycles and multiple rainfall cycles, and obtain the monitoring data within each rainless cycle after intelligent inspection, denoted as the first monitoring data; obtain the monitoring data within each rainfall cycle after intelligent inspection, denoted as the second monitoring data.

[0029] Preferably, the intelligent inspection further includes abnormal trend rejection before missing data supplementation. The abnormal trend rejection specifically includes:

[0030] Obtain the monitoring data of different types at the same position and at the same moment.

[0031] Determine the change trend of each type of monitoring data.

[0032] Reject the monitoring data corresponding to the abnormal change trend.

[0033] Preferably, the intelligent inspection further includes continuity processing after missing data supplementation. The continuity processing specifically includes:

[0034] Obtain the mean value of the end data of the first operation cycle and the start data of the subsequent operation cycle in adjacent operation cycles of each monitoring section.

[0035] Use the mean value to replace the end data of the first operation cycle and the start data of the subsequent operation cycle.

[0036] The method for predicting the safety of the reservoir slope of the daily-regulated pumped-storage power station according to the present invention has the following beneficial effects compared with the prior art:

[0037] By decomposing data, the present invention can obtain the changes in the displacement, anchor stress, anchor strain, etc. of the slope during the water level decline - rise cycle in each operation period of the daily - regulated pumped - storage power station, clarify the influence of rainfall and reservoir water level changes on the slope, and then, in combination with future weather forecasts, judge the safety of the slope. It can also predict the slope deformation in multiple future operation periods and rainfall amounts through the model, and further predict the long - term stability of the reservoir area slope. The safety prediction method of the present invention has a small calculation amount and high accuracy, and can realize fast, economical, and safe prediction of the stability of the reservoir basin slope. Brief Description of the Drawings

[0038] Figure 1 It is a construction flow chart of the safety prediction method for the reservoir basin slope of the daily - regulated pumped - storage power station in the embodiment of the present invention. Detailed Embodiment

[0039] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present invention. However, those skilled in the art should clearly understand that the present invention can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well - known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present invention.

[0040] The embodiment of the present invention discloses a safety prediction method for the reservoir basin slope of a daily - regulated pumped - storage power station, as Figure 1 shown, including:

[0041] Step 1: Divide the operation period of the daily - regulated pumped - storage power station into multiple rain - free periods and multiple rainfall periods, specifically including:

[0042] Step 1.1: Conduct intelligent inspection on the monitoring data of each operation period of the daily - regulated pumped - storage power station.

[0043] One operation period of the daily - regulated pumped - storage power station in the embodiment of the present invention is the period from the reservoir level decline to the rise. Since the pumped - storage power station in the embodiment of the present invention is of daily - regulated type, one operation period is 1 day.

[0044] Before conducting safety prediction in the embodiment of the present invention, it is first necessary to construct a database, specifically as follows: The monitoring data of each cross - section of the reservoir basin slope is recorded as 1 database, and different types of monitoring data are respectively marked. The database includes displacement monitoring data, anchor stress monitoring data, anchor strain monitoring data, etc., which can be respectively recorded as: such databases, It is the slope section number. Since the data law characteristics are different during the water level rise and fall processes, and there are differences in the stage durations. For the convenience of applying the database data, in the embodiments of the present invention, the original database is uniformly intercepted starting from 0 o'clock after the first power generation, the day is the 1st operation cycle, 0 to 24 hours is a complete operation cycle, and each operation cycle has multiple stages.

[0045] To improve the quality of the monitoring data in the database, in the embodiments of the present invention, different databases are divided, and the data in the database is sequentially subjected to intelligent inspection (preprocessing) according to the operation cycle order to obtain effective databases etc.

[0046] Among them, the intelligent inspection includes outlier rejection and missing data supplementation performed in sequence.

[0047] In the embodiments of the present invention, each operation cycle is first regarded as an independent database, and the data within each operation cycle is corrected and supplemented. Specifically, the outliers in the monitoring data in the database are detected based on the Grubbs criterion. Each time an outlier is detected, the point is removed and the inspection continues in a loop until no more outliers can be detected.

[0048] : There are no outliers in the dataset; : There is exactly one outlier in the dataset.

[0049] Let the original data of a single database be , and the statistic of the Grubbs test is defined as shown in (1):

[0050] (1)

[0051] In the formula, is the th original data, is the total number of the original data, is the mean of the original data, is the standard deviation of the original data.

[0052] When the statistic satisfies formula (2), the original data corresponding to the statistic is defined as an outlier, and formula (2) is:

[0053] (2)

[0054] In the formula, represents the degrees of freedom and the critical value of the distribution with a significance level. Generally,

[0055] In the embodiment of the present invention, when supplementing missing data, specifically, the linear interpolation method is used to supplement the missing data, and the formula is as follows:

[0056] (3)

[0057] In the formula, respectively represent the data values before and after the data to be supplemented, represents the data to be supplemented, respectively represent the monitoring times of the corresponding data.

[0058] The intelligent inspection in the embodiment of the present invention further includes abnormal trend elimination before supplementing missing data; the abnormal trend elimination specifically includes: obtaining monitoring data of different types at the same moment at the same position. Exemplarily, the types of monitoring data include displacement monitoring data, anchor bolt stress monitoring data, and anchor bolt strain monitoring data; determining the change trend of each type of monitoring data; and eliminating the monitoring data corresponding to the abnormal change trend.

[0059] Exemplarily, the displacement, anchor bolt strain, and anchor bolt stress at the same position and at the same moment all have the same direction change trend, that is, they all increase (remain unchanged) or decrease (remain unchanged) together. For example, when the displacement increases, the anchor bolt strain increases or remains unchanged, and when the anchor bolt strain increases, the anchor bolt stress increases. Those that violate the rule are judged according to the rule of the adjacent moment, and those that do not conform to the change trend are eliminated. For example, compared with the previous moment, the displacement, anchor bolt strain, and anchor bolt stress are all in the growth process. If the displacement shows a decrease, it is eliminated.

[0060] The intelligent inspection in the embodiment of the present invention further includes continuity processing after supplementing missing data; the continuity processing specifically includes: obtaining the mean value of the end data of the first operation cycle and the start data of the subsequent operation cycle in adjacent operation cycles of each monitoring section; and using the mean value to replace the end data of the first operation cycle and the start data of the subsequent operation cycle.

[0061] Exemplarily, the database formed by the above steps is sorted in sequence according to the occurrence time; based on the principle of time continuity and slope displacement continuity, according to the occurrence time, section the last data retained in the database of the operation cycle and section the first data retained in the operation cycle are averaged and used to replace the original data, eliminating the deviation between units and ensuring the continuity of unit data.

[0062] Step 1.2: Divide the operation cycle of the daily-regulated pumped-storage power station into multiple rainless cycles and multiple rainfall cycles, and obtain the monitoring data in each rainless cycle after intelligent inspection, denoted as the first monitoring data, and obtain the monitoring data in each rainfall cycle after intelligent inspection, denoted as the second monitoring data.

[0063] After the intelligent inspection of the monitoring data, in the embodiments of the present invention, based on the effective database obtained after the intelligent inspection ( database), taking the operation cycle as the unit, 80% of the operation cycles of the effective monitoring data are used as data samples, and 20% are used as evaluation data; the operation cycles within the data samples are divided into rainless cycles and rainfall cycles , The serial number is represented by , The serial number is represented by . Correspondingly, in the embodiments of the present invention, the monitoring data within the rainless cycle is recorded as the first monitoring data, and the monitoring data within the rainfall cycle is recorded as the second monitoring data.

[0064] Step 2: Determine the rainless prediction model according to the first monitoring data of the slope within the rainless cycle.

[0065] In the embodiments of the present invention, first, a "zeroing" process is performed on each rainless cycle in the effective database: the data within the rainless cycle is subtracted from the data at the starting moment within the rainless cycle, that is, it is equivalent to taking the data at the starting moment of the rainless cycle as "0".

[0066] Each rainless cycle in the embodiments of the present invention is divided into multiple stages, and the multiple stages include a water level change stage and a water level constant stage. Among them, the water level change stage specifically includes a water level rising stage and a water level falling stage.

[0067] Exemplarily, each rainless cycle is divided into five stages: water level constant stage 1, water level rising stage, water level constant stage 2, water level falling stage, and water level constant stage 3.

[0068] Then step 2 of the embodiments of the present invention specifically includes:

[0069] Step 2.1: Determine the characteristic values of the prediction model corresponding to each stage of each rainless cycle according to the first monitoring data, and the characteristic values include the slope and the influence parameter of the water level change speed on the prediction data.

[0070] Exemplarily, taking time as the abscissa (0 - 24 hours), and the prediction data as the ordinate, the prediction data is displacement, anchor rod stress, or anchor rod strain. It is defined that each polyline consists of five parts: that is, corresponding to the 5 stages of each rainless cycle (water level constant stage 1, water level rising stage, water level constant stage 2, water level falling stage, and water level constant stage 3), the line segments of each stage of the rainless cycle are approximately monotonically linear, and the slopes of the 5 stages are defined as and the occurrence end times of each group are respectively , then the equation (prediction model) of the rainless cycle is defined as:

[0071] (4)

[0072] In the formula, is the prediction model for the -th rainless cycle, and this prediction model consists of 5 linear stages; is the slope of the first stage (constant water level stage 1) of the -th rainless cycle; is the end time of the first stage (constant water level stage 1) of the -th rainless cycle; is the influence factor of the rising water level of the -th rainless cycle on the slope prediction data, which is a function of the rising speed of the reservoir water level; is the influence factor of the falling water level of the -th rainless cycle on the slope prediction data, which is a function of the falling speed of the reservoir water level; In one aspect, the embodiment of the present invention considers the continuity of deformation, and in the other aspect, it considers the persistence of the influence of water level changes, and defines that and do not only exist in the water level change stage. The above and are determined according to formula (5) and formula (6) respectively.

[0073] (5)

[0074] (6)

[0075] In the formula, are respectively the influence parameters of the rising and falling water level speeds on the slope prediction data; is actually a function of , depending on the slope characteristics, including characteristics such as rock and soil materials and body shapes, and the seepage performance is different; are respectively the influence correction values of the rising and falling water level speeds on the slope prediction data; is the amount of water level rise, is the water level rise time, is the amount of water level fall, is the water level fall time.

[0076] The embodiment of the present invention records as the characteristic value of the rainless cycle. First, substitute each data sample (the value of the sample is known) into formula (4), and the of each rainless cycle and the corresponding characteristic value of the rainless cycle can be obtained. ( For each slope in the polyline, take the sample approximate slope within each stage).

[0077] Step 2.2: Determine the change trend function of the eigenvalue in multiple consecutive rainless periods, and determine the rainless prediction model according to the change trend function.

[0078] In the embodiment of the present invention, the eigenvalues corresponding to each stage of any selected number of consecutive rainless periods are required to have consistent operating conditions during the period (i.e., the change time and speed of the reservoir water level are the same), and the eigenvalues are sorted in the order of time occurrence. By analyzing the increasing and decreasing trends and periodicity of the sequence, the relationship between the eigenvalue of each rainless period and time is established, such as , indicating and time , indicating the time serial numbers corresponding to the occurrence of multiple rainless periods), is the change trend function of the slope in the first stage, and this change trend function can be obtained by fitting.

[0079] Substitute the change trend functions such as determined in the above step 2.2 into formula (4) to obtain the rainless prediction model

[0080] (7)

[0081] Step 3: Determine the rainfall prediction model according to the rainfall data and the second monitoring data of the slope within the rainfall period.

[0082] In the embodiment of the present invention, first perform a "zeroing" process on each rainfall period in the effective database: subtract the data at the starting moment within the rainfall period from the data within the rainfall period, that is, equivalent to taking the data at the starting moment of the rainfall period as "0".

[0083] The rainfall period in the embodiment of the present invention is the same as the rainless period, including multiple stages divided according to the water level change trend. The multiple stages include a water level change stage and a water level constant stage. Among them, the water level change stage specifically includes a water level rising stage and a water level falling stage.

[0084] Exemplarily, each rainfall period is divided into five stages: water level constant stage 1, water level rising stage, water level constant stage 2, water level falling stage, water level constant stage 3.

[0085] After the "zeroing" process, the above step 3 specifically includes:

[0086] Step 3.1: Determine the eigenvalues corresponding to each stage of each rainfall period for the prediction model according to the rainfall data and the second monitoring data. The eigenvalues include the influence parameter of rainfall on the prediction data and the influence parameter of each change point of the second monitoring data on the prediction data.

[0087] In the embodiment of the present invention, it is assumed that the rainfall cycle data = rainless cycle data + rainfall impact factor. Then, this data forms a periodic characteristic with the rainless cycle as the unit. At the same time, the rainfall impact factor is added to form random fluctuations. Based on the principle of the Prophet model, the rainfall equation (prediction model ) is defined as:

[0088] (8)

[0089] In the formula, is the rainfall prediction model for the th rainfall cycle, is the basic model for the th rainfall cycle, which is determined by combining the data samples and occurrence times of the th rainfall cycle and referring to the formula (7) of the rainless cycle.

[0090] The in the embodiment of the present invention is the random fluctuation of the th rainfall cycle, is the error term.

[0091] Among them is determined according to the formula (9):

[0092] (9)

[0093] In the formula, is the rainfall impact factor; is the first "change point" moment of the second monitoring data in the th rainfall cycle in the time series curve trend, ; is the total number of change point moments; is the th "change point" moment of the second monitoring data in the th rainfall cycle in the time series curve trend, ; is the th rainfall cycle time change factor function, indicating whether the time occurs after the change point, where , where ; represents the influence parameter of the th change point on the prediction data in the th rainfall cycle, that is, the change amount of growth, .

[0094] The above is determined according to the following formula (10):

[0095] (10)

[0096] In the formula, is the influence parameter of rainfall on the predicted data, is the rainfall influence correction value; is the rainfall condition in the th rainfall cycle, where is the th rainfall amount in the th rainfall cycle, is the th rainfall duration in the th rainfall cycle, is the number of rainfall times in the

[0097] Step 3.2: Determine the change trend function of the characteristic values in multiple consecutive rainfall cycles, and determine the rainfall prediction model according to the change trend function.

[0098] In any embodiment of the present invention, several consecutive rainfall cycles are arbitrarily selected, and it is required that the operating conditions during this period are the same (i.e., the change time and speed of the reservoir water level are the same), and they are sorted in the order of time occurrence. By analyzing the increasing and decreasing trends and periodicity of the sequence, the influence parameter of rainfall on the predicted data, the influence parameter of each change point of the second monitoring data on the predicted data, and the number of changes and time are obtained, and , are obtained, where represents the time serial number corresponding to multiple rainfall cycles. Correspondingly, formula (9) is transformed into the following form:

[0099] (11)

[0100] Step 4: Predict the safety of the basin slope of the daily-regulated pumped-storage power station according to the rainless prediction model and the rainfall prediction model, specifically including:

[0101] Step 4.1: Determine the prediction model to be used according to the weather conditions of the operation cycle to be predicted.

[0102] Step 4.2: Predict the safety of the basin slope of the daily-regulated pumped-storage power station according to the predicted value of the previous operation cycle corresponding to the operation cycle to be predicted and the prediction model determined in Step 4.1.

[0103] Based on the rainless prediction model and the rainfall prediction model obtained in Step 2 and Step 3, the present invention embodiment builds the for one operation cycle time prediction model, as shown in Formula (12) or Formula (13).

[0104] (12)

[0105] (13)

[0106] is the operation cycle, that is, one day is one operation cycle; is the th operation cycle (i.e., day) predicted value at time (0 - 24 hours); is the initial value, that is, the value at the initial time (0 hour) of the first operation cycle; is the th value at the end time (24 hours) of the operation cycle. In the above formula, is as shown in Formula (14):

[0107] (14)

[0108] In the formula, are the characteristic values of the rain - free cycle respectively, , are the influence parameters of the rising and falling water levels on the slope prediction data respectively, are the influence correction values of the rising and falling water levels on the slope prediction data respectively, and are all determined based on Step 2; are the end times corresponding to each stage of the constant - water - level stage 1, rising - water - level stage, constant - water - level stage 2, falling - water - level stage, and constant - water - level stage 3 within each operation cycle respectively, is the falling water level of the reservoir within the operation cycle, is the total falling - water - level time within the operation cycle, is the rising water level of the reservoir within the operation cycle, is the total rising - water - level time within the operation cycle, are all known.

[0109] In the above formula, is as shown in Formula (15):

[0110] (15)

[0111] In the formula, is the influence parameter of rainfall on the prediction data, is the rainfall influence correction value, is the error term, indicating the impact of the th change point in the operation cycle, that is, the change amount of growth, all determined based on Step 3; is the th rainfall amount within the operation cycle, is the th rainfall duration within the operation cycle, is the number of rainfall times within the operation cycle; is the moment of the first "change point" existing in the operation cycle in the time series curve trend, ; is the number of change point moments; is the moment of the th "change point" existing in the operation cycle in the time series curve trend, ; is the time variation factor function within the operation cycle, indicating whether the time occurs after the change point; the above parameters are all known.

[0112] in formula (15) is determined according to the following formula (16):

[0113] (16)

[0114] where .

[0115] in formula (15) is as shown in formula (17):

[0116] (17)

[0117] To verify the performance of the prediction model obtained by the present invention, the prediction model needs to be tested. Specifically: by inputting the data sample into the prediction model of Step 4 to obtain the prediction data, and using the root mean square error RMSE of it and the evaluation data as the objective function.

[0118] An embodiment of the present invention takes an existing complete cycle as the initial value and substitutes it into formula (13). Combining with the time to be predicted, using the number of cycles from the initial value as , the th cycle of can be predicted.; It can also be based on the predicted value, combined with the rainfall situation, to inversely deduce the operation cycle .

[0119] Through the decomposition of data, the present invention can obtain the changes in slope displacement, anchor stress, anchor strain, etc. during the water level decline - rise cycle in each operation cycle of the power station, clarify the influence of rainfall and reservoir water level changes on the slope, and then, in combination with future weather forecasts, judge the safety situation of the slope. It can also predict the slope deformation under future operation cycles and rainfall amounts through the model, and further predict the long-term stability of the slope in the reservoir area. The safety prediction method of the present invention has a small calculation amount and high accuracy.

[0120] The above are only several embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention is disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art, without departing from the scope of the technical solution of the present invention, making some changes or modifications using the disclosed technical content is equivalent to equivalent implementation cases and all fall within the scope of the technical solution.

Claims

1. A method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station, characterized in that: include: Step 1, dividing the operation cycle of the daily regulation pumped storage power station into a plurality of rainless cycles and a plurality of rainy cycles; Step 2: determining a rainless prediction model according to the first monitoring data of the slope in the rainless period; Step 3, determining a rainfall prediction model according to the rainfall data and the second monitoring data of the slope in the rainfall period; Step 4: predicting the safety of the reservoir basin slope of the daily regulation pumped storage power station according to the rainless prediction model and the rainfall prediction model; The rainless period includes multiple stages divided according to the water level change trend, and step 2 specifically includes: Step 2.1, determining the characteristic values ​​of the prediction model corresponding to each stage of each rainless period according to the first monitoring data, wherein the characteristic values ​​include the influencing parameters of the slope and the water level change speed on the prediction data; Step 2.2, determining a change trend function of characteristic values ​​in a plurality of consecutive rainless periods, and determining a rainless prediction model according to the change trend function; The rainless period includes a water level fluctuation stage and a water level unchanged stage; The prediction model for the water level change stage is determined based on the first monitoring data and the water level change speed in the reservoir basin; The rainfall cycle includes multiple stages divided according to the water level change trend, and step 3 specifically includes: Step 3.1, determining the characteristic value of the prediction model corresponding to each stage of each rainfall cycle according to the rainfall data and the second monitoring data, wherein the characteristic value includes the influence parameter of rainfall on the prediction data and the influence parameter of each change point of the second monitoring data on the prediction data; Step 3.2: Determine a change trend function of characteristic values ​​in a plurality of consecutive rainfall cycles, and determine a rainfall prediction model based on the change trend function.

2. The method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station according to claim 1 is characterized in that: Step 3.2 specifically includes: Determine the trend function of the characteristic values ​​in multiple consecutive rainfall cycles; Determine a random fluctuation function according to the change trend function; A rainfall prediction model is determined according to the random fluctuation function.

3. The method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station according to claim 1 is characterized in that: Step 4 specifically includes: Step 4.1, determine the prediction model to be used according to the weather conditions of the operation period to be predicted; Step 4.2: Predict the safety of the slope of the reservoir basin of the daily regulation pumped storage power station based on the predicted value of the previous operation cycle corresponding to the operation cycle to be predicted and the prediction model determined in step 4.

1.

4. The method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station according to claim 1 is characterized in that: The first monitoring data and the second monitoring data both include at least one of displacement monitoring data, anchor rod stress monitoring data, and anchor rod strain monitoring data.

5. The method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station according to claim 4 is characterized in that: Step 1 specifically includes: Step 1.1, performing intelligent inspection on the monitoring data of each operation cycle of the daily regulation pumped storage power station, the intelligent inspection includes outlier removal and missing data supplementation performed in sequence; the outlier removal is outlier removal based on the Grubbs criterion; the missing data supplementation specifically uses linear interpolation to supplement the missing data; Step 1.2, divide the operation cycle of the daily regulation pumped storage power station into multiple rainless cycles and multiple rainy cycles, and obtain the monitoring data in each rainless cycle after the intelligent inspection, recorded as the first monitoring data, and obtain the monitoring data of each rainy cycle after the intelligent inspection, recorded as the second monitoring data.

6. The method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station according to claim 5 is characterized in that: The intelligent inspection also includes the elimination of abnormal trends before the missing data is supplemented; The abnormal trend elimination specifically includes: Obtain different types of monitoring data at the same location and at the same time; Determine the changing trends of each type of monitoring data; The monitoring data corresponding to the abnormal change trend shall be eliminated.

7. The method for predicting the safety of the reservoir basin slope of a daily regulation pumped storage power station according to claim 6 is characterized in that: The intelligent inspection also includes continuity processing after missing data is supplemented; The continuity processing specifically includes: Obtain the mean of the end data of the first operation cycle and the start data of the subsequent operation cycle in the adjacent operation cycles of each monitoring section; The mean value is used to replace the end data of the first operation cycle and the start data of the subsequent operation cycles.

Images