Transient bright and full flow prediction method for large rainstorm drainage tunnel system

Abstract

The invention relates to the technical field of drainage risk prediction, in particular to a transient bright and full flow prediction method for a large rainstorm drainage tunnel system. According to the method, firstly, all changing curves are corrected, then the risk probability of transient clear full flow of each monitoring point is obtained at each monitoring moment, and then all risk sections and corresponding risk intensity in a tunnel are determined; matching the risk segments at different monitoring moments, and obtaining impurity blocking parameters of each risk segment at the current moment according to changes of the matched risk segments; and then, at the current moment, obtaining expansion influence degrees among different risk segments, further obtaining a danger prediction index of each risk segment, and finally determining all transient bright full flow segments. According to the method, the transient bright and full flow risk sections are preliminarily determined according to the corrected acquired information, then the risk index of each risk section is accurately analyzed and predicted in combination with the expansion change and impurity deposition influence of the transient bright and full flow risk sections, and the transient bright and full flow prediction effect of the large rainstorm drainage tunnel system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of drainage risk prediction, and in particular to a method for predicting transient open-full flow in a large-scale storm drainage tunnel system. Background Art

[0002] Normally, water in large stormwater drainage tunnel systems flows smoothly as open flow (where the water level is below the tunnel top, resulting in a free surface). However, during extreme weather events such as heavy rain, the water flows as full flow (where the water level contacts the tunnel top, forming a pressurized flow), and even transient alternations between open and full flow may occur. Transient open and full flow refers to an unstable state in which the water flow alternates between open and full flow, which can cause local damage to the drainage tunnel structure, poor drainage, and even lead to waterlogging disasters. Therefore, predicting transient open and full flow is crucial.

[0003] Existing technologies typically use models to simulate water flow changes in drainage tunnels to assess and predict the risk of transient full flow. However, during heavy rain, precipitation can carry a large amount of impurities into the tunnel. Once deposited, these impurities can dynamically change the local water flow area and inner wall roughness of the tunnel, thereby affecting water flow behavior and increasing the model's prediction complexity. At the same time, model predictions also need to incorporate information such as the water level and flow rate in the tunnel. However, heavy rain can cause sensors to be washed and soaked, interfering with information collection and affecting the accuracy of model predictions, which in turn affects the accuracy of transient full flow predictions. Summary of the Invention

[0004] In order to solve the technical problem of poor prediction effect of transient clear-fill flow in large-scale storm drainage tunnel systems, the present invention aims to provide a method for predicting transient clear-fill flow in large-scale storm drainage tunnel systems. The technical solution adopted is as follows:

[0005] During the preset historical monitoring period at the current moment, obtain the change curve of each monitoring point at each monitoring section of the tunnel under each hydraulic index, and the change curve of the flow area at each monitoring section;

[0006] Based on the local fluctuation changes of each data point in each change curve, the corresponding change curve is corrected and the predicted data of each hydraulic index at the future monitoring time of each monitoring time is obtained; at each monitoring point, based on the fluctuation changes of the corrected change curve under different hydraulic indexes and the predicted data of each hydraulic index at the future monitoring time of each monitoring time, the risk probability of transient full flow at each monitoring time is obtained; at each monitoring time, all risk sections in the tunnel are determined based on the risk probability, and the risk intensity of each risk section is obtained based on the flow area of ​​all monitoring sections in each risk section in the corresponding corrected change curve, as well as the number of monitoring points and risk probability therein;

[0007] According to the overlap of different risk segments, the risk segments at different monitoring moments are matched, and based on the changes in the matched risk segments, the impurity blockage parameters of each risk segment at the current moment are obtained; at the current moment, based on the risk intensity of each risk segment and the risk probability of each monitoring point within it, combined with the spatial distance between different risk segments, the degree of expansion impact between different risk segments is obtained; at the current moment, based on the relative change in the risk intensity of each risk segment and the impurity blockage parameters, combined with the degree of expansion impact between different risk segments, the hazard prediction index of each risk segment is obtained;

[0008] Determine all transient full flow sections of the tunnel based on the hazard prediction index.

[0009] Furthermore, the method for correcting the change curve and obtaining the prediction data includes:

[0010] In each of the change curves, the curve segment between adjacent extreme values ​​is regarded as a monotonic subsegment, and the abnormal parameters of each data point are obtained according to the fluctuation deviation of the data point at each monitoring moment and the local change trend of the monotonic subsegment to which it belongs, and the data point whose abnormal parameter is greater than a preset threshold is regarded as an abnormal data point;

[0011] In each monotonic subsegment, the difference between each data point and the previous adjacent data point is used as the variation parameter of each data point; the difference between the variation parameter of each abnormal data point and the average level is mapped to the range [-1, 1], and the mapping result is subtracted from the constant 1 as the correction weight; the amplitude of the corresponding abnormal data point is weighted using the correction weight, and the weighted result is used as the correction result of the abnormal data point to obtain the corresponding corrected change curve;

[0012] Under each hydraulic index, the monitoring data of the data point at each monitoring moment in the modified change curve is added with the change parameter to obtain the predicted data at the future monitoring moment of each monitoring moment.

[0013] Furthermore, the method for obtaining the abnormal parameters includes:

[0014] A data point at any monitoring moment in the change curve is taken as a target data point, a monotonic subsegment to which the target data point belongs is taken as a target subsegment, and a monotonic subsegment with the same monotonicity as the target subsegment is taken as a reference subsegment;

[0015] The sum of the absolute values ​​of the differences between the corresponding ranges of the target subsegment and each reference subsegment is used as the trend deviation index of the target subsegment; the amplitude deviation index of the target data point is obtained based on the deviation of the target data point from the average level of the data points in the change curve; and the negative correlation mapping result of the length of the target subsegment is used as the change rate of the target subsegment;

[0016] The trend deviation index, amplitude deviation index and change rate are integrated to obtain the abnormal parameters of the target data point.

[0017] Furthermore, the method for obtaining the risk probability includes:

[0018] At each monitoring point, any hydraulic index is used as the target index, and the remaining hydraulic indexes are used as reference indexes. The attention weight under the target index is obtained based on the fluctuation similarity between the target index and the corresponding correction change curves of each reference index.

[0019] The sum of the attention weights of all hydraulic indicators is 1; at each monitoring point, the attention weight under each hydraulic indicator is used to weight the predicted data of the corresponding hydraulic indicator at each future monitoring moment, and the weighted sum value of all hydraulic indicators is used as the risk probability of transient full flow at the corresponding monitoring point at the corresponding monitoring moment.

[0020] Furthermore, the method for obtaining the risk segment and the risk intensity includes:

[0021] At each monitoring moment, the monitoring point where the risk probability is greater than the preset probability threshold is regarded as a risk point, the monitoring section where the risk point exists is regarded as a risk section, and the tunnel section corresponding to the continuous risk sections in the tunnel is regarded as a risk section;

[0022] In each of the risk segments, the proportion of the number of risk points in all monitoring points is taken as the first risk parameter, the mean of the risk probabilities of all the risk points is taken as the second risk parameter, and the negative correlation mapping result of the mean of the flow areas of all the risk sections in the corresponding corrected change curve is taken as the third risk parameter; the first risk parameter, the second risk parameter and the third risk parameter are integrated to obtain the risk intensity.

[0023] Furthermore, the method of matching risk segments at different monitoring moments includes:

[0024] Taking any of the risk segments at the current moment as the target segment, obtain the number of overlapping sections between the target segment and the monitoring sections of each risk segment at each monitoring moment except the current moment; at each monitoring moment except the current moment, take the risk segment corresponding to the maximum number of overlapping sections as the matching risk segment of the target segment; wherein, when the number of overlapping sections is 0, it is determined that the target segment has no matching risk segment at the corresponding monitoring moment.

[0025] Furthermore, the method for obtaining the impurity blocking parameter includes:

[0026] Taking any of the risk segments at the current moment as the target segment, the risk intensities corresponding to the target segment and all its matching risk segments are sorted in chronological order to construct an intensity change sequence; the total number of monitoring sections corresponding to the target segment and all its matching risk segments are sorted in chronological order to construct a length change sequence;

[0027] A first blocking parameter is obtained based on the difference between the total number of the first and last monitoring sections in the length change sequence and the sequence length of the length change sequence; a second blocking parameter is obtained based on the increase in risk intensity in the intensity change sequence; and the first blocking parameter and the second blocking parameter are fused to obtain the impurity blocking parameter of the target section.

[0028] Furthermore, the method for obtaining the expansion impact degree includes:

[0029] In each risk segment at the current moment, the monitoring point with the smallest risk probability is directed in the direction of the monitoring point with the largest risk probability as the expansion direction of the corresponding risk segment;

[0030] Any of the risk segments at the current moment is used as a target segment, and each of the remaining risk segments is used as a reference segment; between the target segment and each reference segment, the negative correlation normalized value of the cosine value of the angle between the corresponding expansion directions is used as a first influence weight, and the negative correlation normalized value of the corresponding spatial distance is used as a second influence weight;

[0031] The first impact weight and the second impact weight are integrated to obtain an expansion impact weight; the risk intensity of the corresponding reference segment is weighted using the expansion impact weight, and the weighted result is used as the expansion impact degree of each reference segment on the target segment.

[0032] Furthermore, the method for obtaining the risk prediction index includes:

[0033] Obtaining a first risk parameter of the target segment based on the degree of influence of all reference segments on the expansion of the target segment; obtaining a second risk parameter of the target segment based on the difference between the risk intensity of the target segment and the risk intensity of the risk segment matching the previous monitoring moment;

[0034] The first risk parameter, the second risk parameter and the impurity blockage parameter of the target segment are fused, and a normalized result of the fusion result is used as a risk prediction index of the target segment.

[0035] Furthermore, the method for obtaining the transient full flow section includes:

[0036] Among all risk segments at the current moment, all risk segments whose danger prediction index is greater than a preset prediction threshold are regarded as transient full flow segments.

[0037] The present invention has the following beneficial effects:

[0038] The present invention first obtains the change curve of each monitoring point at each monitoring section of the tunnel under each hydraulic index and the change curve of the flow area at each monitoring section, and corrects the corresponding change curve and obtains the predicted data of each hydraulic index at the future monitoring moment of each monitoring moment, providing a data basis for the subsequent accurate prediction of the risk of transient bright full flow in the tunnel; then, at each monitoring point, the risk probability of transient bright full flow at each monitoring moment is obtained to prepare for determining all risk sections in the tunnel; at each monitoring moment, according to the corrected flow area of ​​the monitoring section reflecting the impurity deposition situation in each risk section, the number of monitoring points reflecting the length of the risk section, and each The risk probability of the monitoring point is used to comprehensively evaluate the risk intensity of each risk section; the risk sections at different monitoring moments are further matched, and according to the changes in the matched risk sections, the impact of the dynamic changes of impurities on the transient clear full flow is evaluated to obtain the impurity blocking parameters of each risk section at the current moment; then, at the current moment, according to the risk intensity of each risk section and the risk probability of each monitoring point within it, combined with the spatial distance between different risk sections, the expansion influence degree between different risk sections is obtained, and the relative change of the risk intensity of each risk section and the impurity blocking parameters can be further combined to obtain the danger prediction index of each risk section, thereby determining all transient clear full flow sections of the tunnel. The present invention preliminarily determines the transient clear full flow risk section based on the corrected collected information, and then combines its expansion change and the influence of impurity deposition to accurately analyze and predict the danger index of each risk section, thereby improving the prediction effect of transient clear full flow of large-scale storm drainage tunnel systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 A flow chart of a method for predicting transient full flow in a large-scale storm drainage tunnel system provided by one embodiment of the present invention;

[0041] Figure 2 A schematic diagram of the distribution of monitoring points at a monitoring section provided by one embodiment of the present invention;

[0042] Figure 3 A comparison chart of water level predictions at a certain monitoring point provided by one embodiment of the present invention;

[0043] Figure 4A schematic diagram of the risk probability distribution of transient full flow in a certain monitoring section at different monitoring times provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0044] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of a method for predicting transient full flow in a large-scale stormwater drainage tunnel system proposed in accordance with the present invention. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0045] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0046] The following describes in detail a specific scheme of a method for predicting transient full flow in a large-scale storm drainage tunnel system provided by the present invention with reference to the accompanying drawings.

[0047] See also Figure 1 , which shows a flow chart of a method for predicting transient full flow in a large-scale storm drainage tunnel system provided by one embodiment of the present invention, specifically comprising:

[0048] Step S1, within a preset historical monitoring period at the current moment, obtain the change curve of each monitoring point at each monitoring section of the tunnel under each hydraulic index, and the change curve of the flow area at each monitoring section.

[0049] In one embodiment of the present invention, all monitoring sections are first determined in a large-scale storm drainage tunnel system, specifically at preset intervals such as 10 meters, and a number of monitoring points are evenly set on the inner wall of each monitoring section corresponding to the tunnel; see Figure 2 , which shows a distribution diagram of monitoring points at a monitoring section provided by an embodiment of the present invention, Figure 2 The circle in the figure represents a monitoring section of the tunnel, and the center of each sector corresponds to a monitoring point. Various sensors are then set up at each monitoring point to collect information on each hydraulic indicator and evaluate the flow area of ​​the monitoring section, providing a data basis for subsequent analysis.

[0050] Among them, the hydraulic indicators include at least the water level in the tunnel, the water flow velocity and the pressure on the inner wall of the tunnel; specifically, a water level sensor, a flow velocity sensor and a pressure sensor are installed at each monitoring point, and the sensors collect relevant indicator information at a preset frequency, such as once per second, and then construct a change curve; then, within the preset historical monitoring period at the current moment, specifically the preset historical monitoring period is set to the historical 24 hours, and the indicator information collected by each sensor at each monitoring moment within the historical 24 hours is used as a data point and mapped to a timestamp to fit the change curve of the corresponding hydraulic indicator at the corresponding monitoring point;

[0051] At the same time, detection devices such as ultrasonic sensors or geological radars are set at each monitoring point. The detection results of the detection devices can be used to measure the distance information from the detection point to the inner wall of the tunnel, and then all the detection results at the monitoring section at each monitoring moment can be integrated to evaluate the flow area. The flow area of ​​each monitoring section at each monitoring moment is used as a data point and mapped to the timestamp to fit the change curve of each monitoring flow area.

[0052] It should be noted that in other embodiments of the present invention, the implementer may also define preset intervals to determine the monitoring section, may also define the monitoring point layout plan and preset historical monitoring period at each monitoring section, and may also define the types and quantities of hydraulic indicators.

[0053] It should be noted that the flow area is the area through which water can flow at the corresponding monitoring section. It changes dynamically due to the influence of sediments in the tunnel and is not the cross-sectional area of ​​the tunnel. The fitting of the change curve and the acquisition of the flow area are both existing technologies and will not be repeated here.

[0054] Step S2, according to the local fluctuation changes of each data point in each change curve, correct the corresponding change curve and predict the predicted data of each hydraulic indicator at the future monitoring time of each monitoring moment; at each monitoring point, according to the fluctuation changes of the corrected change curve under different hydraulic indicators and the predicted data of each hydraulic indicator at the future monitoring time of each monitoring moment, obtain the risk probability of transient full flow at each monitoring moment; at each monitoring moment, determine all risk sections in the tunnel according to the risk probability, and obtain the risk intensity of each risk section according to the flow area of ​​all monitoring sections in each risk section in the corresponding corrected change curve, and the number and risk probability of the monitoring points therein.

[0055] Taking into account that after heavy rainwater flows into the tunnel, the water flow may cause certain scouring interference to some sensors, which may affect the accuracy of the collection of the change curve, and further affect the accuracy of the subsequent risk prediction of transient clear full flow in the tunnel; and considering that the more the fluctuation of the data points in the change curve deviates relatively, the more likely it is to be abnormal, and the local fluctuation information of the data points can provide a certain correction reference for the abnormally deviated data points; therefore, the embodiment of the present invention will correct the corresponding change curve according to the local fluctuation changes of each data point in each change curve, so as to prepare for the subsequent accurate prediction of the transient clear full flow risk in the tunnel.

[0056] Preferably, in one embodiment of the present invention, considering that the data output by the sensor changes relatively smoothly and continuously under a stable water flow state, while under the impact of the water flow, the data output by the sensor will experience violent abnormal fluctuations in a short period of time, and there is a large difference compared with the data changes under a stable water flow state, it is first possible to determine the monotonically changing curve sub-segments in each change curve, and then determine the abnormal data points based on the change differences between the curve sub-segments with the same monotonicity and the abnormal offset of each data point; and considering that the change information in each monotonically changing curve sub-segment can provide a certain correction reference for the deviation abnormality of some abnormal data points; based on this, the correction method of the change curve and the method for obtaining the predicted data include:

[0057] In each variation curve, the curve segment between adjacent extreme values ​​is regarded as a monotonic subsegment. Based on the fluctuation deviation of the data point at each monitoring moment and the local variation trend of the monotonic subsegment, the abnormal parameters of each data point are obtained, and the data points with abnormal parameters greater than the preset threshold are regarded as abnormal data points.

[0058] In each monotonic subsegment, the difference between each data point and the previous adjacent data point is used as the change parameter of each data point; the difference between the change parameter of each abnormal data point and the average level is mapped to [-1, 1], and the constant 1 minus the mapping result is used as the correction weight; the amplitude of the corresponding abnormal data point is weighted using the correction weight, and the weighted result is used as the correction result of the abnormal data point to obtain the corresponding correction change curve;

[0059] Under each hydraulic index, the monitoring data of the data point at each monitoring moment in the modified change curve is added with the change parameter to obtain the predicted data at the future monitoring moment of each monitoring moment.

[0060] As an example, taking any change curve as an example, first calculate the abnormal parameters of each data point, and then set the preset threshold to 0.6 to filter out all abnormal data points; then, in the monotonic subsegment to which each abnormal data point belongs, obtain the mean of the change parameters of all data points, and map the difference between the change parameter of each abnormal data point and the mean to [-1,1] through the premnmx() function, and then obtain the correction weight of the corresponding abnormal data point; multiply the amplitude of each abnormal data point by the corresponding correction weight to obtain the correction result; finally, replace all abnormal data points in the change curve with the corresponding correction result to obtain the corrected change curve; then at each monitoring moment, add the data point in the change curve under each hydraulic index to its difference relative to the previous adjacent data point to obtain the predicted data at the next adjacent future monitoring moment of each monitoring moment.

[0061] See also Figure 3 , which shows a comparison diagram of water level prediction at a certain monitoring point provided by an embodiment of the present invention; Figure 3 A total of five water depth prediction results are shown. Water depth refers to water level. The curve corresponding to the measured water depth data is the corrected change curve of the water level, and the curve corresponding to the current prediction data is the curve fitted by the prediction data at each future monitoring moment calculated according to the above steps. The water depth predicted by the current scheme is closer to the corrected change curve than other schemes, and the prediction effect is more accurate, which prepares for the subsequent assessment of risk probability.

[0062] Among them, when the change parameter of each abnormal data point is larger relative to the mean, it means that the abnormal data point is abnormally large, and the mapping result is closer to 1. Then, after subtracting the mapping result from the constant 1, the correction weight is smaller, and the amplitude of the abnormal data point can be reduced; similarly, when the change parameter of the abnormal data point is smaller relative to the mean, it means that the abnormal data point is abnormally small, and the mapping result tends to -1, and the correction weight is larger, which can increase the amplitude of the abnormal data point.

[0063] It should be noted that, considering the special situation that transient full flow may cause negative pressure on the inner wall of the tunnel, that is, the data points in some change curves may be negative, the correction weight is set to a constant of 1 plus the mapping result to obtain the correction result.

[0064] In other examples, implementers may also customize preset thresholds or use other mapping functions, which will not be described in detail.

[0065] In a preferred embodiment of the present invention, considering that monotonic subsegments of the same monotonicity can provide a reference basis for evaluating abnormal deviations of data points, any data point in the change curve is first analyzed as a target data point to determine the target subsegment to which it belongs and the reference subsegment that can provide an analysis reference; considering that the range in the monotonic subsegment can reflect its change trend, the deviation of the change trend of the target subsegment from the change trend of the reference subsegment can be compared; the deviation of the target data point from the average level can reflect its abnormal deviation degree, and the length of the target subsegment to which it belongs can reflect its change rate and change intensity to a certain extent. The shorter the length, the faster and more intense the change; all three can reflect the abnormal possibility of the target data point; the method for obtaining abnormal parameters includes:

[0066] The data point at any monitoring moment in the change curve is taken as the target data point, the monotonic subsegment to which the target data point belongs is taken as the target subsegment, and the monotonic subsegment with the same monotonicity as the target subsegment is taken as the reference subsegment;

[0067] The sum of the absolute values ​​of the differences between the target subsegment and the corresponding ranges of each reference subsegment is used as the trend deviation index of the target subsegment. The amplitude deviation index of the target data point is obtained based on the deviation of the target data point from the average level of the data points in the change curve. The negative correlation mapping result of the length of the target subsegment is used as the change rate of the target subsegment.

[0068] The trend deviation index, amplitude deviation index and change rate are integrated to obtain the abnormal parameters of the target data point.

[0069] As an example, first obtain the trend deviation index of the target subsegment, then use the absolute value of the difference between the target data point and the mean of the data points in the change curve as the amplitude deviation index of the target data point, further perform the inverse operation on the length of the target subsegment, and use the inverse as the change rate of the target subsegment; finally, multiply and fuse the trend deviation index, amplitude deviation index and change rate, and then normalize the product through the sigmoid function to obtain the abnormal parameters of the target data point; by changing the target data point, the abnormal parameters of each data point in each change curve can be obtained.

[0070] In other examples, the implementer may also combine the three through other basic mathematical operations such as addition or weighted summation, and may also normalize the combined result through other mapping functions or normalization methods, which will not be repeated here.

[0071] Taking into account that in the case of transient full flow, the water level is close to the top of the tunnel, the water flow velocity is relatively high, and the pressure on the inner wall caused by the water filling the tunnel is also greater, the possibility of transient full flow can be evaluated at each monitoring point by integrating the indicator information under different hydraulic indicators. The embodiment of the present invention will evaluate the risk probability of transient full flow at each monitoring moment based on the corrected change curves under different hydraulic indicators; the risk probability reflects the possibility of transient full flow occurring at the monitoring point at the monitoring moment, preparing for the subsequent determination of the risk section of transient full flow.

[0072] Preferably, in one embodiment of the present invention, considering that heavy rain causes a surge in water flow in the tunnel, the water level rises rapidly, and the pressure and flow rate should also increase rapidly, there should be a change correlation between different hydraulic indicators at the same monitoring point; and considering that the fluctuation similarity between the change curves under different hydraulic indicators can reflect the change correlation therebetween, the change correlation reflects to a certain extent the degree of influence of each hydraulic indicator on other hydraulic indicators, and indirectly reflects the influence weight of its change on transient full flow. Therefore, based on the change correlation, the attention weight of each hydraulic indicator is determined, and then the indicator information under different hydraulic indicators is weightedly integrated based on the attention weight to evaluate the risk probability; the method for obtaining the risk probability includes:

[0073] At each monitoring point, any hydraulic index is used as the target index, and the remaining hydraulic indexes are used as reference indexes. The attention weight under the target index is obtained based on the fluctuation similarity between the target index and the corresponding correction change curves of each reference index.

[0074] The sum of the attention weights of all hydraulic indicators is 1. At each monitoring point, the attention weight of each hydraulic indicator is used to weight the predicted data of the corresponding hydraulic indicator at each future monitoring moment, and the weighted sum of all hydraulic indicators is used as the risk probability of transient full flow at the corresponding monitoring point at the corresponding monitoring moment.

[0075] As an example, first obtain the Pearson correlation coefficient between the target indicator and the corresponding corrected change curve of each reference indicator, and use the sum of the Pearson correlation coefficients as the focus parameter under the target indicator; by changing the target indicator, the focus parameter of each hydraulic indicator can be obtained; then divide the focus parameter of each hydraulic indicator by the sum of the focus parameters of all hydraulic indicators to obtain the focus weight of each hydraulic indicator, normalize the focus parameter of each hydraulic indicator, and make the sum of the focus weights of all hydraulic indicators 1; the larger the focus weight, the greater the impact of the change of the corresponding hydraulic indicator on the transient full flow, and then fuse the data points of the hydraulic indicator of each monitoring point at each monitoring time by weighted summation, so as to obtain the risk probability of transient full flow at the corresponding monitoring point at the corresponding monitoring time.

[0076] See also Figure 4, which shows a schematic diagram of the risk probability distribution of transient full flow in a certain monitoring section at different monitoring times provided by one embodiment of the present invention; Figure 4 The color of each sector grid corresponding to the monitoring point in each monitoring section reflects the risk probability. The darker the color, the greater the risk probability.

[0077] It should be noted that in other examples, implementers may also use DTW similarity and other means to measure the fluctuation similarity between different change curves, or adopt other normalization means. These, together with the Pearson correlation coefficient and weighted summation, are already existing technologies well known to those skilled in the art and will not be described in detail.

[0078] After obtaining the risk probability of each monitoring point at each monitoring moment, all risk sections in the tunnel can be further determined based on the risk probability at each monitoring moment.

[0079] In a preferred embodiment of the present invention, at each monitoring moment, a monitoring point with a risk probability greater than a preset probability threshold is regarded as a risk point, a monitoring section with a risk point is regarded as a risk section, and a tunnel section corresponding to continuous risk sections in the tunnel is regarded as a risk section; wherein, the preset probability threshold is 0.6, which can also be defined by the implementer.

[0080] Considering that under normal circumstances, the tunnel cross-section is relatively uniform, the occurrence location of transient bright full flow is relatively fixed and can quickly restore stable flow, that is, the frequency of transient bright full flow is low and the duration is short; under the influence of impurity deposition, the flow area of ​​the tunnel cross-section is reduced and the instability of the flow pattern is aggravated, resulting in the frequent occurrence of bright full flow transients and the possibility of their range expanding, and the risk intensity of transient bright full flow is relatively high; considering that the longer the risk section, the greater the risk probability of the monitoring points within it, which also indirectly indicates that the risk intensity of transient bright full flow is relatively high, and the total number of risk points within the risk section can, to a certain extent, reflect the length of the risk section;

[0081] Based on this, the embodiment of the present invention will obtain the risk intensity of each risk segment according to the flow area of ​​all monitoring sections in each risk segment in the corresponding corrected change curve, and the number and risk probability of the monitoring points therein, to prepare for the subsequent analysis of the expansion changes of the risk segment at different monitoring times and the assessment of the risk level of the risk segment.

[0082] Preferably, in one embodiment of the present invention, the method for obtaining risk intensity includes:

[0083] In each risk segment, the proportion of risk points in all monitoring points is taken as the first risk parameter, the mean of the risk probabilities of all risk points is taken as the second risk parameter, and the negative correlation mapping result of the mean of the flow areas of all risk sections in the corresponding corrected change curve is taken as the third risk parameter; the first risk parameter, the second risk parameter and the third risk parameter are integrated to obtain the risk intensity.

[0084] As an example, the mean of the flow areas of all risk sections in the corresponding corrected change curve is subjected to an inverse operation to perform a negative correlation mapping, so that the smaller the mean, the smaller the average flow area, the higher the flow complexity, and the larger the third risk parameter; then the first risk parameter, the second risk parameter and the third risk parameter are multiplied and combined, and the product is used as the risk intensity of the corresponding risk segment.

[0085] In other examples, the implementer may also use the mean as the x in the exponential function exp(-x) with the natural constant e as the base to perform negative correlation mapping, or may use other means such as addition or weighted summation to fuse the three, which will not be repeated here.

[0086] Step S3, according to the overlap of different risk segments, match the risk segments at different monitoring moments, and according to the changes of the matched risk segments, obtain the impurity blockage parameters of each risk segment at the current moment; at the current moment, according to the risk intensity of each risk segment and the risk probability of each monitoring point therein, combined with the spatial distance between different risk segments, obtain the expansion influence degree between different risk segments; at the current moment, according to the relative change of the risk intensity of each risk segment and the impurity blockage parameters, combined with the expansion influence degree between different risk segments, obtain the hazard prediction index of each risk segment.

[0087] Considering that the water flow in the tunnel is affected by many factors, resulting in more complex flow characteristics, the transient full flow or risk segment may change at different times, such as expansion or contraction, but its position will not change significantly in the short term. Therefore, in order to analyze the changes in the risk segment, the embodiment of the present invention will match the risk segments at different monitoring times according to the overlap of different risk segments.

[0088] Preferably, in one embodiment of the present invention, the method for matching risk segments at different monitoring moments includes:

[0089] Taking any risk segment at the current moment as the target segment, obtain the number of overlapping sections between the target segment and the monitoring sections of each risk segment at each monitoring moment except the current moment; at each monitoring moment except the current moment, take the risk segment corresponding to the maximum number of overlapping sections as the matching risk segment of the target segment; among them, when the number of overlapping sections is 0, it is determined that the target segment has no matching risk segment at the corresponding monitoring moment.

[0090] By changing the target segment, we can obtain the matching risk segments of each risk segment at the current moment and at each monitoring moment.

[0091] Considering that the flow of impurity sediments in the tunnel causes the flow area of ​​the tunnel section to change, which is one of the reasons for the frequent occurrence or change of transient full flow phenomenon, by analyzing the changes in the matching risk section, it is possible to evaluate its blockage effect by impurity sediments, that is, the impurity blockage parameter, which can provide an analytical basis for the subsequent assessment of the danger level of the risk section.

[0092] Preferably, in one embodiment of the present invention, the change in risk intensity of the matched risk segment reflects, to a certain extent, the degree of blocking influence of impurity deposits; and the change in length and duration of the matched risk segment can, to a certain extent, reflect the expansion of the risk segment. The greater the expansion trend and the persistence of the expansion, the greater the blocking influence of impurity deposits. Therefore, the method for obtaining impurity blocking parameters includes:

[0093] Taking any risk segment at the current moment as the target segment, the risk intensities corresponding to the target segment and all its matching risk segments are sorted in chronological order to construct an intensity change sequence; the total number of monitoring sections corresponding to the target segment and all its matching risk segments are sorted in chronological order to construct a length change sequence;

[0094] The first blocking parameter is obtained based on the difference between the total number of the first and last monitoring sections in the length change sequence and the sequence length of the length change sequence; the second blocking parameter is obtained based on the growth of risk intensity in the intensity change sequence; and the impurity blocking parameter of the target section is obtained by fusing the first blocking parameter and the second blocking parameter.

[0095] As an example, a target segment is first identified, and an intensity change sequence and a length change sequence are constructed for subsequent analysis. The total number of monitored sections reflects the length of the corresponding risk segment, and the sequence length of the length change sequence indirectly reflects the duration of the risk segment. The two sequences respectively reflect the time-series risk intensity change and length change of the risk segment matching the target segment. The ratio between the last element and the first element in the length change sequence is then used as the length change parameter. The length change parameter is multiplied by the sequence length of the length change sequence to obtain the first blocking parameter. When the ratio is greater than 1 and the length change parameter is larger, it indicates that the risk segment is expanding. At the same time, the longer the sequence length, the longer the risk segment lasts and the greater the impact of impurity blocking. Positive values ​​in the first-order difference sequence of the risk intensity change sequence are further obtained. Positive values ​​reflect the relative growth of risk intensity. The ratio of the total number of positive values ​​to the total number of difference values ​​is multiplied by the mean of all positive values ​​to obtain the second blocking parameter. Finally, the first blocking parameter and the second blocking parameter are multiplied and combined to obtain the impurity blocking parameter of the target segment. The target segment is changed to obtain the impurity blocking parameters of all risk segments at the current moment.

[0096] In other examples, the implementer may also use the normalized value of the difference between the last element and the first element in the length change sequence as the length change parameter; or directly use the normalized value of the slope of the fitting line of the risk intensity change sequence as the second blocking parameter. The slope is positive and the larger it is, the larger the normalized value is, the greater the trend of increasing risk intensity is, and the larger the second blocking parameter is; other means such as addition or weighted summation may also be used to fuse the first blocking parameter and the second blocking parameter, which will not be repeated here.

[0097] Taking into account that at a certain monitoring moment, there is a certain amount of energy flow or transfer between different risk segments. This energy transfer will affect the flow state of the water, thereby making the water flow more unstable, which may lead to the expansion of the risk segment and an increase in the risk intensity. That is, there is a certain expansion effect between different risk segments, and the expansion will also be affected by the spatial distribution and changes of the risk segments. Based on this, the embodiment of the present invention will, at the current moment, according to the risk intensity of each risk segment and the risk probability of each monitoring point therein, combined with the spatial distance between different risk segments, obtain the degree of expansion influence between different risk segments, so as to prepare for the subsequent evaluation of the danger prediction of each risk segment at the current moment.

[0098] Preferably, in one embodiment of the present invention, if a risk segment is close to the other risk segments and their expansion directions are relatively consistent, flow energy may be transferred to the risk segment. Meanwhile, if the risk intensity of the other risk segments is also relatively high, the flow pattern of the water flow in the segment will be more unstable. The transient bright full flow intensity of the risk segment may continue to increase, that is, each risk segment is more affected by the other risk segments. The direction of change of the risk probability of the monitoring points in the risk segment can reflect the expansion tendency of the risk segment to a certain extent. Therefore, the method for obtaining the expansion influence degree includes:

[0099] In each risk segment at the current moment, the monitoring point with the lowest risk probability is pointed in the direction of the monitoring point with the highest risk probability as the expansion direction of the corresponding risk segment;

[0100] Any risk segment at the current moment is taken as the target segment, and each of the remaining risk segments is taken as the reference segment. Between the target segment and each reference segment, the negative correlation normalized value of the cosine value of the corresponding expansion direction is used as the first influence weight, and the negative correlation normalized value of the corresponding spatial distance is used as the second influence weight.

[0101] The first impact weight and the second impact weight are integrated to obtain the expansion impact weight; the risk intensity of the corresponding reference segment is weighted using the expansion impact weight, and the weighted result is used as the expansion impact degree of each reference segment on the target segment.

[0102] As an example, after determining the target segment and all reference segments, the cosine value of the angle between the target segment and the corresponding expansion direction of each reference segment is specifically reciprocally calculated to perform negative correlation normalization, so that the smaller the angle, the smaller the cosine value, the larger the corresponding negative correlation normalization value, and the larger the first influence weight; then the Euclidean distance between the corresponding center points of the target segment and the reference segment is reciprocally calculated to perform negative correlation normalization, so that the smaller the Euclidean distance, the larger the second influence weight; then the first influence weight is multiplied and combined with the second influence weight, and the product is used as the expansion influence weight of each reference segment on the target segment, and then the expansion influence weight is weighted with the risk intensity of the corresponding reference segment, so that the reference segment with a closer expansion direction, a closer distance, and a greater risk intensity has a greater influence on the expansion of the target segment; by changing the target segment, the degree to which each risk segment is affected by the expansion of the remaining risk segments at the current moment can be obtained.

[0103] In other examples, the implementer may also evaluate the expansion direction based on the difference between the target segment and the matching risk segment at the previous adjacent monitoring moment; the sum of the risk probabilities of all monitoring points at each monitoring section in the risk segment may be used as the section risk probability, and the direction of the section corresponding to the minimum section risk probability pointing to the section corresponding to the maximum section risk probability may be used as the expansion direction; other negative correlation normalization methods such as negative exponential functions may be used, or the first influence weight and the second influence weight may be fused by addition or weighted fusion, which will not be elaborated here.

[0104] At the current moment, after obtaining the degree to which each risk segment is affected by the expansion of other risk segments, we can further combine the relative changes in the risk intensity of each risk segment and the impurity blockage parameters to obtain the danger prediction index of each risk segment. The danger prediction index is the predicted danger of transient full flow occurring in the risk segment at the current moment.

[0105] Preferably, in one embodiment of the present invention, considering that the greater the influence of other risk segments on the expansion of a certain risk segment, the greater the possibility of the risk segment expanding and increasing the risk intensity, and the greater the degree of danger; considering that if the risk intensity of a certain risk segment increases relative to adjacent monitoring moments and the greater the influence of impurity blockage it suffers, it has a certain tendency to expand and the degree of danger is also greater; therefore, the method for obtaining the risk prediction index includes:

[0106] The first hazard parameter of the target segment is obtained based on the degree of influence of all reference segments on the expansion of the target segment. The second hazard parameter of the target segment is obtained based on the difference between the risk intensity of the target segment and the risk intensity of the risk segment matching the previous monitoring moment. The first hazard parameter, second hazard parameter and impurity blockage parameter of the target segment are fused, and the normalized result of the fusion result is used as the hazard prediction index of the target segment.

[0107] As an example, the influence of all reference segments on the expansion of the target segment is averaged to obtain the first risk parameter; then the risk intensity of the target segment is used as the numerator, the risk intensity of the risk segment matching the previous monitoring moment is used as the denominator, and the fractional ratio is used as the second risk parameter. A ratio greater than 1 indicates that the risk intensity of the target segment has increased relative to the historical monitoring moment, the degree of danger has increased, and the second risk parameter is also larger; then the first risk parameter, the second risk parameter and the impurity blockage parameter of the target segment are multiplied and combined, the product is normalized by the sigmoid function, and the normalized value is used as the risk prediction index of the target segment.

[0108] In other examples, the implementer may also use the normalized value of the difference between the risk intensity of the target segment and the risk intensity of the risk segment matching the previous monitoring moment as the second risk parameter; or may fuse the three by addition or weighted fusion, and then use other normalization methods to obtain the risk prediction index, which will not be repeated here.

[0109] Step S4: determining all transient full flow sections of the tunnel according to the hazard prediction index.

[0110] At the current moment, after obtaining the hazard prediction index of each risk section in the tunnel, all transient full flow sections can be further determined.

[0111] Preferably, in one embodiment of the present invention, among all risk segments at the current moment, all risk segments whose danger prediction index is greater than a preset prediction threshold are regarded as transient bright full flow segments; wherein, the preset prediction threshold is 0.6, which can also be customized by the implementer; the transient bright full flow segment is obtained based on the analysis and prediction of sensor collected data.

[0112] In summary, the present invention first corrects all the change curves, then obtains the risk probability of transient bright full flow at each monitoring point at each monitoring moment, and then determines all risk sections in the tunnel, and further obtains the risk intensity of each risk section; matches the risk sections at different monitoring moments, and according to the changes of the matched risk sections, obtains the impurity blocking parameters of each risk section at the current moment; then, at the current moment, obtains the expansion influence degree between different risk sections, and then obtains the danger prediction index of each risk section, and finally determines all transient bright full flow sections of the tunnel. The present invention preliminarily determines the transient bright full flow risk section based on the corrected collected information, and then combines its expansion changes and impurity deposition effects during the monitoring period to accurately analyze and predict the danger index of each risk section, thereby improving the prediction effect of transient bright full flow of large-scale storm drainage tunnel systems.

[0113] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0114] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for predicting transient full flow in a large-scale storm drainage tunnel system, characterized in that: The method comprises: During the preset historical monitoring period at the current moment, obtain the change curve of each monitoring point at each monitoring section of the tunnel under each hydraulic index, and the change curve of the flow area at each monitoring section; Based on the local fluctuation of each data point in each variation curve, the corresponding variation curve is corrected and the predicted data of each hydraulic index at each future monitoring moment is predicted; at each monitoring point, the risk probability of transient full flow at each monitoring moment is obtained based on the fluctuation of the corrected variation curve under different hydraulic indexes and the predicted data of each hydraulic index at each future monitoring moment; at each monitoring moment, all risk sections in the tunnel are determined based on the risk probability, and the risk intensity of each risk section is obtained based on the flow area of ​​all monitoring sections in each risk section in the corresponding corrected variation curve, as well as the number of monitoring points and risk probability therein; According to the overlap of different risk segments, the risk segments at different monitoring moments are matched, and based on the changes in the matched risk segments, the impurity blockage parameters of each risk segment at the current moment are obtained; at the current moment, based on the risk intensity of each risk segment and the risk probability of each monitoring point within it, combined with the spatial distance between different risk segments, the degree of expansion impact between different risk segments is obtained; at the current moment, based on the relative change in the risk intensity of each risk segment and the impurity blockage parameters, combined with the degree of expansion impact between different risk segments, the hazard prediction index of each risk segment is obtained; Determine all transient full flow sections of the tunnel based on the hazard prediction index.

2. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1 is characterized in that: The method for correcting the change curve and the method for obtaining the prediction data include: In each of the change curves, the curve segment between adjacent extreme values ​​is regarded as a monotonic subsegment, and the abnormal parameters of each data point are obtained according to the fluctuation deviation of the data point at each monitoring moment and the local change trend of the monotonic subsegment to which it belongs, and the data point whose abnormal parameter is greater than a preset threshold is regarded as an abnormal data point; In each monotonic subsegment, the difference between each data point and the previous adjacent data point is used as the variation parameter of each data point; the difference between the variation parameter of each abnormal data point and the average level is mapped to the range [-1, 1], and the mapping result is subtracted from the constant 1 as the correction weight; the amplitude of the corresponding abnormal data point is weighted using the correction weight, and the weighted result is used as the correction result of the abnormal data point to obtain the corresponding corrected change curve; Under each hydraulic index, the monitoring data of the data point at each monitoring moment in the modified change curve is added with the change parameter to obtain the predicted data at the future monitoring moment of each monitoring moment.

3. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 2, characterized in that: The method for obtaining the abnormal parameters includes: A data point at any monitoring moment in the change curve is taken as a target data point, a monotonic subsegment to which the target data point belongs is taken as a target subsegment, and a monotonic subsegment with the same monotonicity as the target subsegment is taken as a reference subsegment; The sum of the absolute values ​​of the differences between the corresponding ranges of the target subsegment and each reference subsegment is used as the trend deviation index of the target subsegment; the amplitude deviation index of the target data point is obtained based on the deviation of the target data point from the average level of the data points in the change curve; and the negative correlation mapping result of the length of the target subsegment is used as the change rate of the target subsegment; The trend deviation index, amplitude deviation index and change rate are integrated to obtain the abnormal parameters of the target data point.

4. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1, characterized in that: The method for obtaining the risk probability includes: At each monitoring point, any hydraulic index is used as the target index, and the remaining hydraulic indexes are used as reference indexes. The attention weight under the target index is obtained based on the fluctuation similarity between the target index and the corresponding correction change curves of each reference index. The sum of the attention weights of all hydraulic indicators is 1; at each monitoring point, the attention weight under each hydraulic indicator is used to weight the predicted data of the corresponding hydraulic indicator at each future monitoring moment, and the weighted sum value of all hydraulic indicators is used as the risk probability of transient full flow at the corresponding monitoring point at the corresponding monitoring moment.

5. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1, characterized in that: The method for obtaining the risk segment and the risk intensity includes: At each monitoring moment, the monitoring point where the risk probability is greater than the preset probability threshold is regarded as a risk point, the monitoring section where the risk point exists is regarded as a risk section, and the tunnel section corresponding to the continuous risk sections in the tunnel is regarded as a risk section; In each of the risk segments, the proportion of the number of risk points in all monitoring points is taken as the first risk parameter, the mean of the risk probabilities of all the risk points is taken as the second risk parameter, and the negative correlation mapping result of the mean of the flow areas of all the risk sections in the corresponding corrected change curve is taken as the third risk parameter; the first risk parameter, the second risk parameter and the third risk parameter are integrated to obtain the risk intensity.

6. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1, characterized in that: The method for matching risk segments at different monitoring moments includes: Taking any of the risk segments at the current moment as the target segment, obtain the number of overlapping sections between the target segment and the monitoring sections of each risk segment at each monitoring moment except the current moment; at each monitoring moment except the current moment, take the risk segment corresponding to the maximum number of overlapping sections as the matching risk segment of the target segment; wherein, when the number of overlapping sections is 0, it is determined that the target segment has no matching risk segment at the corresponding monitoring moment.

7. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1, characterized in that: The method for obtaining the impurity blocking parameter includes: Taking any of the risk segments at the current moment as the target segment, the risk intensities corresponding to the target segment and all its matching risk segments are sorted in chronological order to construct an intensity change sequence; the total number of monitoring sections corresponding to the target segment and all its matching risk segments are sorted in chronological order to construct a length change sequence; A first blocking parameter is obtained based on the difference between the total number of the first and last monitoring sections in the length change sequence and the sequence length of the length change sequence; a second blocking parameter is obtained based on the increase in risk intensity in the intensity change sequence; and the first blocking parameter and the second blocking parameter are fused to obtain the impurity blocking parameter of the target section.

8. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1, characterized in that: The method for obtaining the expansion impact degree includes: In each risk segment at the current moment, the monitoring point with the smallest risk probability is directed in the direction of the monitoring point with the largest risk probability as the expansion direction of the corresponding risk segment; Any of the risk segments at the current moment is used as a target segment, and each of the remaining risk segments is used as a reference segment; between the target segment and each reference segment, the negative correlation normalized value of the cosine value of the angle between the corresponding expansion directions is used as a first influence weight, and the negative correlation normalized value of the corresponding spatial distance is used as a second influence weight; The first impact weight and the second impact weight are integrated to obtain an expansion impact weight; the risk intensity of the corresponding reference segment is weighted using the expansion impact weight, and the weighted result is used as the expansion impact degree of each reference segment on the target segment.

9. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 8, characterized in that: The method for obtaining the risk prediction index includes: Obtaining a first risk parameter of the target segment based on the degree of influence of all reference segments on the expansion of the target segment; obtaining a second risk parameter of the target segment based on the difference between the risk intensity of the target segment and the risk intensity of the risk segment matching the previous monitoring moment; The first risk parameter, the second risk parameter and the impurity blockage parameter of the target segment are fused, and a normalized result of the fusion result is used as a risk prediction index of the target segment.

10. The method for predicting transient full flow in a large-scale storm drainage tunnel system according to claim 1, characterized in that: The method for obtaining the transient bright full flow section includes: Among all risk segments at the current moment, all risk segments whose danger prediction index is greater than a preset prediction threshold are regarded as transient full flow segments.

Images