Water regimen safety monitoring method based on digital twinning

Abstract

The invention relates to the technical field of environmental monitoring, and relates to a digital twinning-based water regimen safety monitoring method, which comprises the following steps of: acquiring historical water regimen data, and drawing a precipitation curve and a water level curve according to the historical water regimen data; determining an associated segment; lagging periods of the associated segments are calculated, and weighted summation is carried out on all the lagging periods belonging to positive numbers to obtain a target lagging period; and acquiring precipitation data at a set moment and at a target hysteresis stage before the set moment, and inputting the precipitation data into the trained ARIMA model to predict the water level at the set moment. Based on the complex lag relationship between the precipitation amount and the water level change, the water level change is predicted according to the historical precipitation data and the target lag stage, and the accuracy of water level prediction is improved.

Description

Technical Field

[0001] The present invention relates to the field of environmental monitoring technology. More specifically, the present invention relates to a water situation safety monitoring method based on digital twins. Background Art

[0002] Floods are common and destructive natural disasters. Based on conditions such as precipitation intensity and basin topography, floods can rise sharply in a short period of time, causing serious property losses and casualties. Water level monitoring can timely detect abnormal water level rise trends by collecting water level data in real time, thereby providing support for flood prevention warnings, drainage measures and risk avoidance decisions. Therefore, water level monitoring plays a vital role in water safety management.

[0003] The prior art usually monitors the water level of rivers, lakes, reservoirs and other places in real time, and issues an alarm when the water level exceeds the safety warning value, and takes measures such as flood discharge and personnel evacuation. The Chinese patent application document with publication number CN116524515A discloses a real-time water level monitoring method based on computer vision, which includes: photographing an image with a water gauge area in the monitored water area; performing HSV transformation on the photographed image with a water gauge area in the monitored water area to preliminarily obtain an image of the water gauge area; performing image tilt correction on the preliminarily obtained image of the water gauge area; performing image segmentation on the tilt-corrected image to extract an image containing only the water gauge area; performing character segmentation on the extracted image containing only the water gauge area to obtain a character image; and determining the height of the water level according to the character image.

[0004] Since precipitation is an important factor affecting water level changes, existing technologies also use neural network models to predict water levels based on precipitation. However, there is a certain time difference between the arrival of precipitation and water level changes. The increase in precipitation does not immediately lead to a rise in water levels. In some cases, even if the precipitation has ended, the water level in some areas of the basin may still be affected for a long time. This means that the impact of precipitation on water levels will not be immediately apparent, but as time goes by, the water flow gradually accumulates, and the water level will change significantly. Therefore, when using the ARIMA model to monitor and predict water levels, if the real-time precipitation is directly used as input data, the ARIMA model may make inaccurate predictions due to the lag in the impact of precipitation on water level changes. Summary of the invention

[0005] In order to solve the above-mentioned technical problem that directly using real-time precipitation as input data may lead to inaccurate prediction of the ARIMA model, the present invention provides a water situation safety monitoring method based on digital twins.

[0006] A water regime safety monitoring method based on digital twins includes: collecting historical water regime data, and drawing a precipitation curve and a water level curve according to the historical water regime data; using the previous data point of an inflection point between two adjacent peaks in the precipitation curve or the water level curve as a segmentation point, and dividing the corresponding curve into multiple segments; when the absolute value of the difference between the moment of the peak in any segment of the water level curve and the moment of the peak in a segment of the precipitation curve is less than a threshold, determining that the corresponding segments of the precipitation curve and the water level curve are mutually related segments; for all related segments of the segment, determining the related segment with the largest peak value as the associated segment of the segment; calculating the lag period of the associated segment G m , all positive lag periods are weighted summed to obtain the target lag period; , F m is the peak moment in the precipitation curve, E m is the moment of the inflection point closest to the peak in the precipitation curve, f m is the peak moment of the water level curve, e m is the moment of the inflection point closest to the peak in the water level curve, and E m < F m , e m < f m ; Obtain precipitation data at the set time and the target lag period before the set time, and input it into the trained ARIMA model to predict the water level at the set time.

[0007] The beneficial effects of the present invention are as follows: after segmenting the precipitation curve and the water level curve, the present invention matches the precipitation curve segments and the water level segment curves with similar characteristics according to the peak values ​​and specific inflection points of each segment, and calculates the time difference between each pair of matched segments (i.e., each pair of associated segments), so as to obtain the lag period of the precipitation effect on the water level through each segment, and then weighted sums the lag periods greater than 0 corresponding to each pair of matched segments to obtain the lag period of the precipitation effect on the water level as a whole, i.e., the target lag period. Based on this, the present invention can predict the water level according to the rainfall data at a certain time distance before the moment of the expected predicted water level to obtain a more accurate water level prediction result, wherein the certain time distance is the target lag period.

[0008] Preferably, performing weighted summation of all positive hysteresis periods to obtain a target hysteresis period comprises: taking an association segment whose hysteresis period is a positive number as a target association segment, calculating the weight of the hysteresis period of each pair of target association segments, wherein the weight of the hysteresis period of a pair of target association segments is proportional to the DTW distance between the two segments; performing weighted summation of the hysteresis period of each pair of target association segments: ,in G is the target lag period, For the i The hysteresis period for target-related segments, η i For the i The weight of the lag period for the target association segment, N The total number of pairs of segments associated with the target.

[0009] Preferably, calculate n The weight of the lag period of the target associated segment includes: calculating the DTW distance between the two segments in each pair of target associated segments; calculating the n Weights for the lag period of the target-related segment: ,in For the n The weight of the lag period for the target association segment, H n For the n For the DTW distance between associated segments, the exp() function is an exponential function with the natural constant e as the base. n Is a positive integer.

[0010] After determining the weights corresponding to each pair of target-associated segments according to the correlation between the two segments in each pair of target-associated segments (i.e., the DTW distance between the two segments), the present invention performs weighted summation on the lag period of each pair of target-associated segments to obtain the target lag period, and the stronger the correlation between the two segments in a pair of target-associated segments, that is, the smaller the DTW distance between the two segments, the greater the weight of the lag period of the pair of target-associated segments. Based on this, the target lag period can better reflect the time lag effect of precipitation on the water level.

[0011] Preferably, calculating the DTW distance between two segments in a pair of target-associated segments includes: obtaining precipitation data at multiple moments included in the segment corresponding to the precipitation curve in the pair of target-associated segments, recorded as a first sequence; obtaining water level data at multiple moments included in the segment corresponding to the water level curve in the pair of target-associated segments, recorded as a second sequence, wherein the data in the first sequence and the second sequence are arranged in time series; linearly scaling the sequence with fewer data in the first sequence and the second sequence so that the first sequence and the second sequence have the same number of data; after linearly scaling the first sequence or the second sequence, calculating the DTW distance between the first sequence and the second sequence to obtain the DTW distance between the two segments in the pair of target-associated segments.

[0012] The present invention makes the number of data points of the two sequences the same through linear scaling, thereby ensuring the alignment of precipitation and water level curves in time.

[0013] Preferably, drawing a precipitation curve includes: obtaining precipitation data in the historical water condition data after preprocessing the historical water condition data, wherein the preprocessing of the historical water condition data includes denoising the historical water condition data, removing outliers, and filling in missing values; fitting the precipitation data into a first curve through a fitting algorithm; and smoothing the first curve to obtain the precipitation curve.

[0014] The present invention obtains a precipitation curve after smoothing the first curve obtained by fitting precipitation data. The smoothing is performed to reduce noise and ensure the continuity of the precipitation curve.

[0015] Preferably, drawing a water level curve includes: obtaining water level data in the historical water situation data after preprocessing the historical water situation data, wherein the preprocessing of the historical water situation data includes denoising the water level data, removing outliers, and filling in missing values; fitting the water level data into a second curve through a fitting algorithm; and smoothing the second curve to obtain the water level curve.

[0016] The present invention obtains a water level curve after smoothing the first curve obtained by fitting the water level data. The smoothing is performed to reduce noise and ensure the continuity of the water level curve.

[0017] Preferably, constructing the ARIMA model includes: obtaining a training set, the training set includes multiple training data, wherein the collection time corresponding to the water level or precipitation in each training data is different, and the difference between the collection time of the water level and the collection time of the precipitation in the training data is equal to the lag period; obtaining an ARIMA initial model, wherein the parameters of the ARIMA initial model are all preset values; training the ARIMA initial model through the training set to obtain the ARIMA model.

[0018] Preferably, obtaining precipitation data at a set time and a target lag period before the set time, and inputting the data into a trained ARIMA model to predict the water level at the set time includes: determining the set time; after inputting the precipitation data at a target lag period before the set time into the ARIMA model, determining that the output of the ARIMA model is the result of predicting the water level at the set time.

[0019] The beneficial effects of the present invention are: The present invention matches precipitation and water level curves in segments, combines the dynamic time warping distance, i.e., DTW distance, to calculate the correlation of each pair of segments, thereby more accurately capturing the hysteresis effect of precipitation on water level. Based on the complex hysteresis relationship between precipitation and water level changes, the present invention predicts water level changes based on historical precipitation data and target hysteresis period, thereby improving the accuracy of water level prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a flowchart schematically showing the steps of a water situation safety monitoring method based on digital twin according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0022] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] Figure 1 It is a flowchart schematically showing the steps of a water situation safety monitoring method based on digital twin according to an embodiment of the present invention.

[0024] like Figure 1 As shown, a water situation safety monitoring method based on digital twins includes steps S1 to S5.

[0025] Step S1: Collect historical water regime data, and draw precipitation curves and water level curves based on the historical water regime data.

[0026] The historical water condition data includes precipitation data and water level data. In one embodiment, during the time period from the first predetermined time to the second predetermined time, the rainfall sensor detects the rainfall at a frequency of f 0 is recorded as precipitation data; during the time period from the first predetermined time to the second predetermined time, the water level sensor is at a frequency of f 0 is recorded as water level data. In this embodiment, the rainfall sensor and the water level sensor are located at the same position.

[0027] In one embodiment, drawing a precipitation curve includes: obtaining precipitation data in the historical water regime data after preprocessing the historical water regime data, wherein the preprocessing of the historical water regime data includes denoising the historical water regime data, removing outliers, and filling in missing values; fitting the precipitation data into a first curve through a fitting algorithm; and smoothing the first curve to obtain the precipitation curve. Drawing a water level curve includes: obtaining water level data in the historical water regime data after preprocessing the historical water regime data, wherein the preprocessing of the historical water regime data includes denoising the water level data, removing outliers, and filling in missing values; fitting the water level data into a second curve through a fitting algorithm; and smoothing the second curve to obtain the water level curve.

[0028] It should be noted that both the precipitation curve and the water level curve are time series curves. A time series curve represents a curve showing data changes over time, and a time series curve visualizes the numerical relationship between time (usually an independent variable) and a variable (precipitation or water level in this invention) at consecutive time points.

[0029] Furthermore, when fitting water level data and precipitation data, the same fitting algorithm is used. The fitting algorithm can be polynomial regression, spline fitting, etc. Among them, polynomial regression is a method of extending linear regression. Polynomial regression captures the nonlinear relationship of data (i.e., precipitation or water level) by raising the independent variable (i.e., precipitation or water level) to a high power. Spline curves are smooth curves generated by a given set of control points.

[0030] Step S2: Taking the previous data point of an inflection point between two adjacent peaks in the precipitation curve or the water level curve as a segmentation point, the corresponding curve is divided into a plurality of segments.

[0031] Wherein, the target curve is a precipitation curve or a water level curve. It should be noted that for a smooth curve, there is at least one inflection point between two peaks, and the inflection point is any inflection point between the two peaks. When the precipitation curve is segmented (that is, when the target curve is a precipitation curve), the target curve segmentation point is a point located on the precipitation curve at a moment before the inflection point. When the water level curve is segmented (that is, when the target curve is a water level curve), the target curve segmentation point is a point located on the upper water level curve at a moment before the inflection point.

[0032] In another embodiment, the segmentation point is a data point at a previous moment of an inflection point between two peaks and closest to the larger peak at the corresponding moment (ie, the previous data point).

[0033] Step S3: Determine the associated segments.

[0034] Wherein, determining the associated segmentation includes: taking the previous data point of an inflection point between two adjacent peaks in the precipitation curve or the water level curve as the segmentation point, dividing the corresponding curve into multiple segments; when the absolute value of the difference between the time of the peak in any segment of the water level curve and the time of the peak in a segment of the precipitation curve is less than a threshold, determining that the corresponding segments of the precipitation curve and the water level curve are mutually associated segments; for all the associated segments of the segmentation, determining the associated segment with the largest peak value as the associated segment of the segmentation. In one embodiment, the threshold is 5 days.

[0035] It should be noted that, in the correspondence process, multiple segments of precipitation curves may be related to the same segment of water level curve. Since low precipitation intensity has little effect on water level changes, the segment with the highest precipitation peak is selected to correspond to the segment of a water level curve, and the two are determined to be related segments. Similarly, if multiple segments of water level curves are related to the same segment of precipitation curve, the segment of the water level curve with the most obvious water level change, i.e. the largest peak, is selected to correspond to the segment of a precipitation curve, and the two are determined to be related segments.

[0036] Step S4: Calculate the lag period of the associated segment G m , all positive lags are weighted summed to obtain the target lag.

[0037] in, , F m is the peak moment in the precipitation curve, E m is the moment of the inflection point closest to the peak in the precipitation curve, f m is the peak moment of the water level curve, em is the moment of the inflection point closest to the peak in the water level curve, and E m < F m , e m < f m .

[0038] In one embodiment, G m For the m The lag period of the associated segment. m The associated segmentation includes two segments, one of which is obtained by segmenting the precipitation curve, so the segmentation belongs to the segmentation of the precipitation curve, and the other is obtained by segmenting the water level curve, so the segmentation belongs to the segmentation of the water level curve. When the curve (the curve is a precipitation curve or a water level curve) is segmented, the previous data point of the inflection point (any one) between two adjacent peaks is the segmentation point, so any segment of the curve includes a peak and at least one inflection point. The present invention selects the inflection point that is closest to the peak in time and is located before the peak in time sequence, and reflects the hysteresis of the influence of precipitation data on water level data through the average of the time difference between the selected inflection points of the two segments of the associated segmentation and the time difference between the peaks.

[0039] In one embodiment, all hysteresis periods that are positive are weighted and summed to obtain a target hysteresis period: the association segments whose hysteresis periods are positive are taken as target association segments, and the weights of the hysteresis periods of each pair of target association segments are calculated, wherein the weights of the hysteresis periods of a pair of target association segments are proportional to the DTW (Dynamic Time Warping) distance between the two segments; the weighted summation of the hysteresis periods of each pair of target association segments is performed: ,in G is the target lag period, For the i The lag period for target-related segmentation, η i For the i The weight of the lag period for the target association segment, N is the total number of pairs of target associated segments. In this embodiment, the calculation n The weight of the hysteresis period of the associated segments includes: calculating the DTW distance between the two segments in each pair of associated segments; calculating the n Weights for lags in associated segments: ,in For the n The weight of the lag period for the target association segment, H nFor the n For the DTW distance between associated segments, the exp() function is an exponential function with the natural constant e as the base. n Is a positive integer.

[0040] It should be noted that for a pair of associated segments, the DTW distance of the two segments it includes reflects the correlation between the two segments, and also reflects the correlation between the precipitation data and the water level data in the two segments. The larger the DTW distance between the two segments, the smaller the correlation between the two segments, the smaller the correlation between the precipitation data and the water level data in the two segments, and the smaller the weight corresponding to the pair of associated segments.

[0041] In another embodiment, the weights of the lag periods of each pair of target-associated segments are the same, i.e., , N is the total number of pairs of associated segments.

[0042] In another embodiment, the weights of the lag periods of each pair of target-associated segments are empirical values, and the sum of the weights of the lag periods of each pair of target-associated segments is equal to 1.

[0043] In one embodiment, calculating the DTW distance between associated segments includes: obtaining precipitation data at multiple moments included in the segment corresponding to the precipitation curve in the associated segment, recorded as a first sequence; obtaining water level data at multiple moments included in the segment corresponding to the water level curve in the associated segment, recorded as a second sequence, wherein the data in the first sequence and the second sequence are arranged in time series; linearly scaling the sequence with less data in the first sequence and the second sequence so that the first sequence and the second sequence have the same number of data; after linearly scaling the first sequence or the second sequence, calculating the DTW distance between the first sequence and the second sequence to obtain the DTW distance between the associated segments.

[0044] It should be noted that when calculating the DTW distance between associated segments, the purpose of linearly scaling the sequence with a smaller number of data is to make the two sequences have the same length on the time axis so as to enable effective comparison and distance calculation.

[0045] Step S5: Obtain precipitation data at a set time and a target lag period before the set time, and input it into the trained ARIMA model to predict the water level at the set time.

[0046] Among them, a training data of the ARIMA model includes t The water level at time 1 and t The precipitation at 2 o'clock, t 1- t 2= G. In one embodiment, constructing the ARIMA model includes: obtaining a training set, the training set is a time series data set, which includes multiple training data, wherein the collection time corresponding to the water level or precipitation in each training data is different, and the difference between the collection time of the water level and the collection time of the precipitation in the training data is equal to the lag period; obtaining an ARIMA initial model, wherein the parameters of the ARIMA initial model are all preset values; training the ARIMA initial model through the training set to obtain the ARIMA model. In this embodiment, predicting the water level at a predetermined time includes: obtaining precipitation data at a target time, wherein the difference between the target time and the predetermined time is equal to the target lag period; after inputting the precipitation data at the target time into the ARIMA model, the output of the ARIMA model is determined as the water level at the predetermined time.

[0047] In another embodiment, the training set includes a plurality of training subsets, the training subsets include a first time series data set and a second time series data set, the first time series data set includes water level data collected at a frequency f of a predetermined size in a first time period, and the second time series data set includes precipitation data collected at a frequency f of a predetermined size in a second time period. The time lengths of the first time period and the second time period are equal and both are T, and the difference between the initial moment of the first time period and the initial moment of the second time period is equal to the target lag period. In this embodiment, the ARIMA model obtained by training the training set receives a time series precipitation data and outputs a time series water level data, the time series precipitation data is historical data (i.e., precipitation data collected before the current moment), and the output time series water level data includes water level data at multiple moments, and the water level data at the expected moment (the expected moment is after the current moment) can be obtained according to the time series water level data. For example, a time series of precipitation data is input to the ARIMA model in this embodiment, and the time series of precipitation data includes precipitation data collected by the precipitation sensor at a frequency of f from time T1 to time T2, and the precipitation data is arranged in a time series order, where T2-T1=T. At this time, the ARIMA model inputs a time series of water level data, and the time series of water level data includes Q water level data from time T3 to time T4 (the water level data is the predicted water level data), and the water level data is arranged in a time series order, where T4-T3=T, Q=T×f, and the qth data in the time series of water level data represents the water level data at time (T3+q×f), q is a positive integer and q≤Q. Based on this, the water level data at the expected time (i.e., the water level prediction data) can be obtained.

[0048] Furthermore, the precipitation data at any moment includes the precipitation at that moment, for example, a The precipitation value at time isA Alternatively, the water level data at any moment includes the average value of the water level within a certain period of time before (including) that moment. a The water level data A collected at all times meets the following requirements: ,in T represents a positive integer, A i Representative i The precipitation value at a certain time point, represent A 1 to A T The sum of precipitation at each point in time during a period of time, A 1 represents the precipitation value at the earliest time point, A 1 to A T is the water level data at continuous moments, A T Represents the current time t of the collection a precipitation data.

[0049] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0050] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A water situation safety monitoring method based on digital twins, characterized in that: include: Collect historical water regime data and draw precipitation curves and water level curves based on the historical water regime data; Taking the previous data point of an inflection point between two adjacent peaks in a precipitation curve or a water level curve as a segmentation point, the corresponding curve is divided into multiple segments; when the absolute value of the difference between the moment of the peak in any segment of the water level curve and the moment of the peak in a segment of the precipitation curve is less than a threshold, the corresponding segments of the precipitation curve and the water level curve are determined to be mutually related segments; for all related segments of the segment, the related segment with the largest peak value is determined as the associated segment of the segment; Calculate the hysteresis period of the associated segment G m , all positive lag periods are weighted summed to obtain the target lag period; , F m is the peak moment in the precipitation curve, E m is the moment of the inflection point closest to the peak in the precipitation curve, f m is the peak moment of the water level curve, e m is the moment of the inflection point closest to the peak in the water level curve, and E m < F m , e m < f m ; The precipitation data at the set time and the target lag period before the set time are obtained, and input into the trained ARIMA model to predict the water level at the set time.

2. A water situation safety monitoring method based on digital twins according to claim 1, characterized in that: The weighted sum of all positive lags to obtain the target lag includes: The association segments whose hysteresis period is positive are taken as target association segments, and the weights of the hysteresis period of each pair of target association segments are calculated, wherein the weights of the hysteresis period of a pair of target association segments are proportional to the DTW distance between the two segments; Weighted summation of the lag periods for each pair of target-associated segments: ,in G is the target lag period, For the i The hysteresis period for target-related segments, η i For the i The weight of the lag period for the target association segment, N The total number of pairs of segments associated with the target.

3. A water situation safety monitoring method based on digital twin according to claim 2, characterized in that: Calculate the n The weights for the lag period of the target association segment include: Calculate the DTW distance between two segments in each pair of target associated segments; Calculate the n Weights for the lag period of the target-related segment: ,in For the n The weight of the lag period for the target association segment, H n For the n For the DTW distance between associated segments, the exp() function is an exponential function with the natural constant e as the base. n Is a positive integer.

4. A water situation safety monitoring method based on digital twins according to claim 3, characterized in that: Calculating the DTW distance between two segments in a pair of target associated segments includes: Obtain precipitation data at multiple moments included in a segment corresponding to a precipitation curve in a pair of target associated segments, recorded as a first sequence; obtain water level data at multiple moments included in a segment corresponding to a water level curve in a pair of target associated segments, recorded as a second sequence, wherein the data in the first sequence and the second sequence are arranged in time sequence; Linearly scaling the sequence with less data in the first sequence and the second sequence so that the first sequence and the second sequence have the same number of data; After linear scaling is performed on the first sequence or the second sequence, the DTW distance between the first sequence and the second sequence is calculated to obtain the DTW distance between two segments in a pair of target associated segments.

5. The water situation safety monitoring method based on digital twin according to claim 1 is characterized in that: Drawing precipitation curves includes: After preprocessing the historical water regime data, the precipitation data in the historical water regime data is obtained, wherein the preprocessing of the historical water regime data includes denoising the historical water regime data, removing outliers, and filling in missing values; Fitting the precipitation data into a first curve through a fitting algorithm; The first curve is smoothed to obtain the precipitation curve.

6. A water situation safety monitoring method based on digital twins according to claim 1, characterized in that: Drawing a water level curve includes: After preprocessing the historical water regime data, water level data in the historical water regime data is obtained, wherein the preprocessing of the historical water regime data includes denoising the water level data, removing abnormal values, and filling in missing values; Fitting the water level data into a second curve through a fitting algorithm; The second curve is smoothed to obtain the water level curve.

7. The water situation safety monitoring method based on digital twin according to claim 1 is characterized in that: Building the ARIMA model includes: Obtaining a training set, the training set comprising a plurality of training data, wherein the collection time corresponding to the water level or precipitation in each training data is different, and the difference between the collection time of the water level and the collection time of the precipitation in the training data is equal to the lag period; Obtain the ARIMA initial model, where the parameters of the ARIMA initial model are all preset values; The ARIMA initial model is trained using a training set to obtain the ARIMA model.

8. The water situation safety monitoring method based on digital twin according to claim 1 is characterized in that: Obtaining precipitation data at the set time and the target lag period before the set time, and inputting it into the trained ARIMA model to predict the water level at the set time includes: Determine the set time; After the precipitation data of the target lag period before the set time is input into the ARIMA model, the output of the ARIMA model is determined to be the result of predicting the water level at the set time.

Images