CN111860973B - Debris flow intelligent early warning method based on multi-objective optimization - Google Patents
Debris flow intelligent early warning method based on multi-objective optimization Download PDFInfo
- Publication number
- CN111860973B CN111860973B CN202010610972.8A CN202010610972A CN111860973B CN 111860973 B CN111860973 B CN 111860973B CN 202010610972 A CN202010610972 A CN 202010610972A CN 111860973 B CN111860973 B CN 111860973B
- Authority
- CN
- China
- Prior art keywords
- debris flow
- early warning
- dominant
- predicted
- pareto
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 72
- 238000005457 optimization Methods 0.000 title claims abstract description 42
- 239000011435 rock Substances 0.000 claims description 38
- 239000003795 chemical substances by application Substances 0.000 claims description 23
- 230000005284 excitation Effects 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 10
- 239000002245 particle Substances 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 8
- 230000006698 induction Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000000638 stimulation Effects 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 abstract description 11
- 230000007547 defect Effects 0.000 abstract description 2
- 239000000243 solution Substances 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 6
- 230000002265 prevention Effects 0.000 description 6
- 238000002360 preparation method Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 239000007787 solid Substances 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 3
- 239000004575 stone Substances 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000007123 defense Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000000611 regression analysis Methods 0.000 description 2
- 229920006395 saturated elastomer Polymers 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000037007 arousal Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000004927 clay Substances 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000010432 diamond Substances 0.000 description 1
- 238000010790 dilution Methods 0.000 description 1
- 239000012895 dilution Substances 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 239000002002 slurry Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/10—Alarms for ensuring the safety of persons responsive to calamitous events, e.g. tornados or earthquakes
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
- G08B21/182—Level alarms, e.g. alarms responsive to variables exceeding a threshold
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- General Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Quality & Reliability (AREA)
- Geology (AREA)
- Environmental & Geological Engineering (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Computer Security & Cryptography (AREA)
- Educational Administration (AREA)
- Primary Health Care (AREA)
Abstract
The invention discloses a debris flow intelligent early warning method based on multi-objective optimization, which belongs to the field of geological disaster monitoring and early warning, particularly relates to the field of debris flow early warning, and aims to overcome the defects of larger error, inaccuracy and poor reliability of the existing debris flow prediction method; constructing a proxy model for evaluating and exciting the debris flow; solving the pareto threshold front edge of the debris flow proxy model; inputting the front edge of a debris flow pareto threshold of a debris flow sample point to be predicted and a debris flow proxy model into a debris flow early warning discriminator to perform dominant discrimination comparison, and outputting dominant discrimination results, wherein dominant discrimination results are that early warning is not performed, non-dominant discrimination results are that yellow early warning is performed, and dominant discrimination results are that red early warning is performed; evaluating the support degree of red early warning; and carrying out debris flow early warning according to the dominance judgment result and the support degree. And a more accurate proxy model is constructed, and the reliability and operability of debris flow monitoring and early warning are greatly improved.
Description
Technical Field
The invention discloses a debris flow intelligent early warning method based on multi-objective optimization, belongs to the field of geological disaster monitoring and early warning, and particularly relates to the technical field of debris flow early warning.
Background
Debris flow is the flood flow formed by saturated dilution of a sandy and soft soil mountain body containing sand and stones through rainstorm and flood, and is large in area, volume and flow rate, and landslide is a small-area of the diluted soil mountain body, and typical debris flow is composed of viscous slurry which is suspended with coarse solid debris and is rich in silt and clay. Under proper topographic conditions, a large amount of water soaks solid accumulated substances in a flowing water hillside or a ditch bed, so that the stability of the solid accumulated substances is reduced, and the solid accumulated substances saturated with water move under the action of self gravity to form a debris flow. Debris flow is a disastrous geological phenomenon. Usually the debris flow is sudden, violent and can carry huge stones. It is extremely destructive because it has a strong energy due to its high speed of travel.
With the continuous development of computer technology, intelligent means is more and more applied to early warning of debris flow.
The conventional early warning technology obtains an approximate debris flow early warning curve by regression analysis of historical data. The method generally adopts a relatively universal fitting model of a linear function or a logistic function, and different models cannot be used for analysis according to different characteristics of each region. Other debris flow early warning methods based on machine learning are to consider whether debris flow occurs or not as a two-classification problem to train an optimal classifier, such as a neural network or a support vector machine.
The above methods using regression analysis, neural networks and support vector machines require training the classifier using sample data for which debris flow has occurred and has not occurred. The sample data of the non-occurrence debris flow can not deduce the deviation of the sample data of the non-occurrence debris flow, so that the error of the classifier is large, and only two classifications can be carried out on the monitoring data, namely, whether the debris flow occurs or not is predicted.
Disclosure of Invention
The invention aims to: the method is used for solving the defects that the existing debris flow early warning method has larger error of a classifier because sample data which does not generate debris flow is adopted, the deviation of exciting the debris flow can not be deduced, and in addition, only two classifications can be carried out on monitoring data, namely whether the debris flow is generated or not is predicted, so that the method is inaccurate and not high in reliability.
The technical scheme adopted by the invention is as follows:
an intelligent debris flow early warning method based on multi-objective optimization comprises the following steps
Step 1, collecting a data sample of the generated debris flow;
step 2, constructing a proxy model for evaluating and exciting the debris flow;
step 3, solving the pareto threshold front edge of the debris flow proxy model;
step 4, inputting the front edge of a debris flow pareto threshold value of a debris flow sample point to be predicted and a debris flow proxy model into a debris flow early warning discriminator to perform dominant discrimination comparison, and outputting a dominant discrimination result, wherein dominant discrimination result is red early warning;
and 6, carrying out debris flow early warning according to the dominance judgment result and the support degree.
Preferably, historical sample data of the debris flow which has occurred in the monitoring early warning area is collected.
Preferably, the debris flow sample point to be predicted is also in the monitoring early warning area.
Preferably, the debris flow proxy model in the step 2 is a multi-target proxy model, and the multi-target proxy model has a plurality of input variables, namely a plurality of debris flow influence factors, and a plurality of output variables, namely evaluation indexes for exciting the debris flow.
Preferably, the multi-target agent model is one or a group of display functions (regression equation, kernel function, etc.) or implicit models (such as neural network expressed by connecting weights and threshold values) obtained by learning and training the mud stone flow data samples, and the learning and training method of the multi-target agent model can be one or more integration of machine learning methods.
Preferably, the input of the multi-target agent model is arousal rain intensity I max And the accumulated rainfall E induced by the excitation process is output as the average hourly rainfall intensity I of the excitation process and the rainfall induction duration D of the debris flow excitation process, the optimized two targets are the average hourly rainfall intensity I of the excitation process and the rainfall induction duration D of the debris flow excitation process, the optimized algorithm adopts a multi-target particle swarm algorithm, and the obtained two targets are the mostThe small pareto front is a collection of data and does not represent a particular curve.
Preferably, the front edge of the pareto threshold of the debris flow is composed of a group of debris flow excitation index values which are not dominant, and the front edge is obtained by carrying out optimization solution on a multi-objective proxy model through a multi-objective optimization method, wherein the multi-objective optimization method comprises a data optimization method or a multi-objective evolutionary optimization method, such as a multi-objective genetic algorithm and a multi-objective particle swarm optimization algorithm. The multi-objective minimized Pareto front on the data samples of the occurring debris flow can represent a minimum threshold of an evaluation index of the occurring debris flow.
Preferably, the multi-target agent model is constructed as follows: firstly, a full-connection network is used for fitting historical data of the occurring debris flow, and the following steps are carried out according to 4:1, dividing training data and test data in proportion, then adopting a loss function and a regularization term as two targets in a double-target optimization problem in a training process, then adopting a multi-target particle swarm optimization method to obtain a corresponding debris flow pareto threshold front edge, then adopting a two-stage optimal decision method to select 5 candidate solution sets in the debris flow pareto threshold front edge, and determining parameters of a proxy model according to the final performance of the 5 solution sets on prediction data. And finally, selecting a group of parameters with the best performance according to the performances of the agent models corresponding to different solution sets on the test data, and completing the construction of the multi-target agent model. At the moment, the multi-target agent model is not a classifier, so that two target minimum pareto curves of the multi-target agent model need to be found through multi-target optimization, the multi-target particle swarm optimization is used for carrying out two target minimum optimization on average hourly rainfall intensity I of two output excitation processes of the neural network and rainfall induction duration D of a debris flow excitation process, and the corresponding pareto curves are found and input into a debris flow early warning discriminator. And then, acquiring debris flow observation data, namely debris flow sample points needing to be predicted, through a debris flow online monitor, and inputting the data into a debris flow early warning discriminator to perform dominant discrimination and comparison.
Preferably, the debris flow sample point to be predicted and the debris flow pareto threshold front have the same physical index.
Preferably, the dominance is compared by discriminating and comparing, the magnitude of two multidimensional variables is compared, and a Pareto dominance comparison method, a relaxed Pareto dominance comparison method (such as epsilon-Pareto dominance) or any comparison method for distinguishing multidimensional variables is adopted.
Preferably, the debris flow sample points to be predicted are obtained by acquiring data on an online debris flow monitor and performing data preprocessing, such as filtering and denoising.
Preferably, the mud-rock flow sample point to be predicted is dominant at least at one point on the pareto curve, and the mud-rock flow sample point to be predicted and the point on the pareto curve are not dominant if the dominant relationship cannot be compared.
Preferably, the debris flow sample points to be predicted are dominant, the possibility of debris flow occurrence is considered to be very low, early warning is not needed, the debris flow sample points to be predicted are dominant, the debris flow is considered to be excited to send out red early warning, and the debris flow sample points to be predicted are non-dominant, the continuous close observation is needed, and the yellow early warning is sent out.
Preferably, the debris flow sample point to be predicted is dominated by a point N on a pareto curve where the debris flow sample point to be predicted is dominated according to the point N i To point N on the total pareto curve all Calculating a support S from the ratio of degree If the support is S degree =N i /N all For the dominant red warning point, when S degree When the mud-rock flow rate is less than 10%, considering that the mud-rock flow occurring at the mud-rock flow sample point needing to be predicted is slight, and making mud-rock flow prevention preparation work by governments and related departments according to duties; when S is degree When the current is 10-20%, considering that debris flow generated by debris flow sample points needing to be predicted is in the middle of the debris flow, and the like, making debris flow prevention preparation work by governments and related departments according to duties, and simultaneously cutting off dangerous outdoor power supplies to transfer personnel in danger zones to safe places; when S is degree At 20-40%, the mud-rock flow is considered to need predictionThe debris flow generated at the sample point is serious, the government and related departments make debris flow emergency work according to duties, and the traffic management department adopts traffic control on part of road sections according to conditions; when S is degree When the number of the mud-rock flow sample points is more than 40%, the mud-rock flow which needs to be predicted is considered to be very serious, the government and related departments carry out emergency work of the mud-rock flow according to responsibilities, the defense and emergency work of disasters such as the mud-rock flow and the like should be carried out, and meanwhile, the related meetings and the courseware are stopped. Therefore, the index can reflect the severity of the occurrence of debris flow.
Preferably, in the step 4, a dominance judgment result is output, and the dominance is not early-warned, so that early warning is not needed; the early warning is yellow when the early warning is not dominant, continuous close observation is needed, and the support degree is not required to be calculated when the early warning is dominant or not dominant.
In the technical scheme of the application, pareto is Pareto.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. according to the invention, a more accurate proxy model can be constructed without too much prior knowledge, and the reliability and operability of debris flow monitoring and early warning can be greatly improved.
2. In the invention, only the samples of the debris flow are used when the proxy model is constructed, so that on one hand, the deviation caused by the samples of the debris flow can be avoided, and on the other hand, the required sample amount can be reduced.
3. According to the invention, through multi-objective intelligent optimization, the lower boundary of the proxy model can be quickly found, and the operability of debris flow monitoring and early warning is improved.
4. In the invention, by introducing a method for comparing the support degree and the dominance, not only can the result of three classifications be obtained, but also the severity can be judged according to the support degree, thereby more scientifically and efficiently deploying the pre-alarm work.
5. The invention provides a debris flow only early warning method based on multi-objective optimization, which can construct a proxy model without relying on prior knowledge and does not depend on samples without debris flow; the early warning point set can be found more scientifically and efficiently by adopting a multi-objective optimization method when the front edge of the Pareto threshold of the debris flow is found; and in the last early warning judgment part, the early warning work of the debris flow can be guided to be better carried out by a method of comparing the support degree with the dominance.
Drawings
FIG. 1 is a flow chart of an intelligent debris flow early warning method based on multi-objective optimization according to the invention;
fig. 2 is a diagram of a case of early warning debris flow in embodiment 2 of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1
As shown in FIG. 1, the debris flow intelligent early warning method based on multi-objective optimization comprises the following steps
Step 1, collecting data samples of debris flow in a monitoring early warning area;
step 2, constructing a proxy model for evaluating and exciting the debris flow, wherein the debris flow proxy model is a multi-target proxy model, and the multi-target proxy model is provided with a plurality of input variables, namely a plurality of debris flow influence factors, and a plurality of output variables, namely evaluation indexes for exciting the debris flow; the multi-target agent model is one or a group of display functions (regression equation, kernel function and the like) or implicit models (such as a neural network expressed by connecting weight and threshold value) obtained by learning and training the existing debris flow data samples, and the learning and training method of the multi-target agent model can be one or more integration of machine learning methods; input of the multi-target agent model is aroused rain intensity I max And the cumulative rainfall E induced by the excitation process is output as the average hourly rainfall intensity I of the excitation process and the rainfall induction duration D of the debris flow excitation process, the optimized two targets are the average hourly rainfall intensity I of the excitation process and the rainfall induction duration D of the debris flow excitation process, the optimized algorithm adopts a multi-target particle swarm algorithm, and the obtained minimum pareto frontier of the two targets is dataDoes not represent a particular curve; constructing a multi-target agent model: first, a fully connected network is used to fit historical data of the occurring debris flow, and the data is obtained according to the following steps of 4:1, dividing training data and test data in proportion, then adopting a loss function and a regularization term as two targets in a double-target optimization problem in a training process, then adopting a multi-target particle swarm optimization method to obtain a corresponding debris flow pareto threshold front edge, then adopting a two-stage optimal decision method to select 5 candidate solution sets in the debris flow pareto threshold front edge, and determining parameters of a proxy model according to the final performance of the 5 solution sets on prediction data. According to the performance of the agent models corresponding to different solution sets on test data, a group of parameters with the best performance (the parameters with the minimum Mean Square Error (MSE) on the test data) is finally selected, and the construction of the multi-target agent model is completed. At the moment, the multi-target agent model is not a classifier, so that two target minimum pareto curves of the multi-target agent model need to be found through multi-target optimization, the multi-target particle swarm optimization is used for carrying out two target minimum optimization on average hourly rainfall intensity I of two output excitation processes of the neural network and rainfall induction duration D of the debris flow excitation process, and the corresponding pareto curves are found and input into the debris flow early warning discriminator. Then, acquiring debris flow observation data, namely debris flow sample points needing to be predicted, through a debris flow online monitor, and inputting the data into a debris flow early warning discriminator to perform dominant discrimination and comparison;
and 3, solving the pareto threshold front edge of the debris flow proxy model, wherein the pareto threshold front edge of the debris flow is composed of a group of debris flow excitation index values which are not dominant, and the pareto threshold front edge is obtained by carrying out optimization solution on the multi-target proxy model through a multi-target optimization method, wherein the multi-target optimization method comprises a data optimization method or a multi-target evolution optimization method, such as a multi-target genetic algorithm and a multi-target particle swarm optimization algorithm. The method comprises the steps that a multi-target minimized Pareto front edge on a data sample of the existing debris flow can represent a minimum threshold value of an evaluation index of the existing debris flow; the debris flow sample point to be predicted and the front edge of the debris flow pareto threshold have the same physical index;
step 4, inputting the debris flow sample points to be predicted in the monitoring early-warning area and the front edge of the debris flow pareto threshold of the debris flow proxy model into a debris flow early-warning discriminator to perform dominant discrimination comparison, and outputting a dominant discrimination result, wherein dominant discrimination means no early warning, non-dominant discrimination means yellow early warning, and dominant discrimination means red early warning; judging and comparing the dominance, comparing the sizes of the two multidimensional variables, and adopting a Pareto dominance comparison method, a loose Pareto dominance comparison method (such as epsilon-Pareto dominance) or any comparison method for distinguishing the multidimensional variables; obtaining data of a debris flow sample point to be predicted by a debris flow on-line monitor and carrying out data preprocessing, such as filtering and denoising; the mud-rock flow sample point to be predicted is dominant at least at one point on the pareto curve, and the mud-rock flow sample point to be predicted and the point on the pareto curve are not dominant if the dominant relation cannot be compared; if the mud-rock flow sample points to be predicted are dominant, the possibility of the occurrence of the mud-rock flow is considered to be very low, early warning is not needed, if the mud-rock flow sample points to be predicted are dominant, the mud-rock flow is considered to be excited to send out red early warning, if the mud-rock flow sample points to be predicted are not dominant, the mud-rock flow sample points to be predicted need to be continuously and closely observed to send out yellow early warning;
step 6, carrying out debris flow early warning according to the dominance judgment result and the support degree; the mudslide sample point to be predicted is dominant, and the point N on the pareto curve where the mudslide sample point to be predicted is dominant i To point N on the total pareto curve all Calculating the support S from the ratio of degree If the support is S degree =N i /N all For a dominant red warning point, when S degree When the mud-rock flow is less than 10%, considering that the mud-rock flow generated at the mud-rock flow sample point needing to be predicted is slight, and making mud-rock flow prevention preparation work by governments and related departments according to responsibilities; when S is degree 10-20%, the debris flow sample point needing prediction is considered to beIn the middle of the raw debris flow, government and related departments prepare for debris flow prevention according to duties, and should cut off dangerous outdoor power supply to transfer personnel in dangerous areas to safe places; when S is degree When the mud-rock flow is 20-40%, the mud-rock flow occurring at the mud-rock flow sample points needing to be predicted is considered to be serious, the government and related departments make mud-rock flow emergency work according to duties, and the traffic management department needs to adopt traffic control on part of road sections according to conditions; when S is degree When the number of the mud-rock flow sample points is more than 40%, the mud-rock flow which needs to be predicted is considered to be very serious, the government and related departments carry out emergency work of the mud-rock flow according to responsibilities, the defense and emergency work of disasters such as the mud-rock flow and the like should be carried out, and meanwhile, the related meetings and the courseware are stopped. Therefore, the index can reflect the severity of the occurrence of debris flow.
Example 2
On the basis of the embodiment 1, 9 points of data of the pu-er ditch and the wenchuan in 2019 are selected, wherein the mud-rock flow occurs for 5 times and does not occur for 4 times, and the discriminator can output the early warning condition and the support degree at the moment. If the point is dominant, the possibility of debris flow is considered to be very low, and early warning is not needed; if the point is dominant, the possibility of debris flow is considered to be high, and early warning is needed; if the points are not dominant, the decision is made by continuously and closely observing the points. As shown in fig. 2, 3 of the 5 points where debris flow has occurred are classified as red, i.e., early warning is required, and 2 are classified as yellow, i.e., close observation is continuously required for decision making. And 4 points without debris flow are divided into blue, namely early warning is not needed.
While the point is dominant or dominant at point N on the pareto curve i To point N on the total pareto curve all Calculating the support S from the ratio of degree . As shown in fig. 2 (ordinate in fig. 2 indicates average rainfall intensity per hour in the excitation process), the support degrees of the 3 black regular triangle points from left to right are respectively 20.58% (considering that the generated debris flow is serious, the government and related departments have emergency work of the debris flow according to duties, and traffic management departments should adopt traffic control on partial road sections according to conditions), 6.53% (the generated debris flow is slight,the government and the related departments make the mud-rock flow prevention preparation work according to the duty), 4.24% (the generated mud-rock flow is slight, and the government and the related departments make the mud-rock flow prevention preparation work according to the duty), because the support degree of the black triangular point at the leftmost side is the highest, the possibility of the mud-rock flow is the highest. The 2 gray regular triangles and the discrimination curve are in a state of not dominance (needing continuous close observation), so that the support degree is not calculated. 4 black diamond points are non-early warning points (no measures are taken), so that the support degree is not calculated.
The above description is intended to be illustrative of the preferred embodiment of the present invention and should not be taken as limiting the invention, but rather, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.
Claims (7)
1. The debris flow intelligent early warning method based on multi-objective optimization is characterized by comprising the following steps
Step 1, collecting a data sample of the generated debris flow;
step 2, constructing an agent model for evaluating and exciting the debris flow, wherein the agent model for evaluating and exciting the debris flow is a multi-target agent model, the multi-target agent model is provided with a plurality of input variables and a plurality of output variables, and the input of the multi-target agent model is the strength of the excited rainI max And triggering process to induce accumulated rainfallEThe output is the average hourly rainfall intensity of the excitation processIAnd duration of rainfall induction in mud-rock flow excitation processDWhile two goals for optimization are average hourly rainfall intensity for the stimulation processIAnd duration of rainfall induction in mud-rock flow excitation processDThe optimized algorithm adopts a multi-target particle swarm algorithm;
step 3, solving the pareto threshold front edge of the debris flow proxy model;
step 4, inputting the front edge of a debris flow pareto threshold value of a debris flow sample point to be predicted and a debris flow proxy model into a debris flow early warning discriminator to perform dominant discrimination comparison, and outputting a dominant discrimination result, wherein dominant discrimination result is red early warning;
step 5, evaluating the support degree of the red early warning;
and 6, carrying out debris flow early warning according to the dominance judgment result and the support degree, wherein the debris flow sample point to be predicted is dominated, and the debris flow sample point to be predicted is dominated according to the point on the pareto curveN i To a point on the total pareto curveN all Calculating the support degree from the ratio of (A) to (B)S degree Then the support degree isFor a dominant red warning point, whenS degree When the mass flow rate is less than 10%, considering that the debris flow occurring at the debris flow sample point needing to be predicted is slight; when the temperature is higher than the set temperatureS degree When the mass flow rate is 10-20%, determining that the debris flow occurring at the debris flow sample point needs to be predicted is medium; when in useS degree When the mass flow rate is 20-40%, considering that the debris flow occurring at the debris flow sample point needing to be predicted is serious; when in useS degree Above 40%, the debris flow occurring at the debris flow sample point that needs to be predicted is considered to be very severe.
2. The intelligent debris flow early warning method based on multi-objective optimization as claimed in claim 1, wherein the multi-objective agent model is one or a group of display functions or implicit models obtained by learning and training samples of generated debris flow data.
3. The intelligent debris flow early warning method based on multi-objective optimization as claimed in claim 2, wherein the front edge of the pareto threshold of the debris flow is composed of a group of debris flow excitation index values which are not dominant, and the debris flow is obtained by performing optimization solution on the multi-objective agent model through a multi-objective optimization method.
4. The multi-objective optimization-based intelligent debris flow early warning method as claimed in claim 1, wherein the dominance discrimination and comparison adopts a Pareto dominance comparison method, a loose Pareto dominance comparison method or any comparison method for distinguishing multidimensional variables.
5. The intelligent debris flow early warning method based on multi-objective optimization as claimed in claim 1, wherein the debris flow sample points to be predicted are obtained by acquiring data on a debris flow online monitor and preprocessing the data.
6. The multi-objective optimization-based intelligent debris flow early warning method as claimed in claim 1, wherein at least one dominant point on a pareto curve of a debris flow sample point to be predicted is dominant, at least one dominant point on the pareto curve of the debris flow sample point to be predicted is dominant, and if the dominant relationship between the debris flow sample point to be predicted and the pareto curve cannot be compared, the dominant relationship is not dominant.
7. The debris flow intelligent early warning method based on multi-objective optimization as claimed in claim 1, wherein step 4 outputs a dominance judgment result, and the dominance is not early warning, and no measure is taken; the early warning of yellow color is conducted when the color is not dominant, and the color needs to be continuously and closely observed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010610972.8A CN111860973B (en) | 2020-06-30 | 2020-06-30 | Debris flow intelligent early warning method based on multi-objective optimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010610972.8A CN111860973B (en) | 2020-06-30 | 2020-06-30 | Debris flow intelligent early warning method based on multi-objective optimization |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111860973A CN111860973A (en) | 2020-10-30 |
CN111860973B true CN111860973B (en) | 2023-04-18 |
Family
ID=72989292
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010610972.8A Active CN111860973B (en) | 2020-06-30 | 2020-06-30 | Debris flow intelligent early warning method based on multi-objective optimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111860973B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104318717A (en) * | 2014-10-21 | 2015-01-28 | 四川大学 | Rainstorm debris flow early warning method under shortage conditions of historical data |
CN104866692A (en) * | 2015-06-18 | 2015-08-26 | 北京理工大学 | Aircraft multi-objective optimization method based on self-adaptive agent model |
CN105590141A (en) * | 2015-12-15 | 2016-05-18 | 东北大学 | Genetic algorithm initial population construction method applied to optimized design of complex products |
CN109118718A (en) * | 2018-07-09 | 2019-01-01 | 中国科学院、水利部成都山地灾害与环境研究所 | Rainfall I-D curve threshold value construction method, basin debris flow early-warning method occur for mud-rock flow |
CN109615118A (en) * | 2018-11-23 | 2019-04-12 | 泉州装备制造研究所 | Based on big data hazards control Informatization Service integrated control system and method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7664622B2 (en) * | 2006-07-05 | 2010-02-16 | Sun Microsystems, Inc. | Using interval techniques to solve a parametric multi-objective optimization problem |
-
2020
- 2020-06-30 CN CN202010610972.8A patent/CN111860973B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104318717A (en) * | 2014-10-21 | 2015-01-28 | 四川大学 | Rainstorm debris flow early warning method under shortage conditions of historical data |
CN104866692A (en) * | 2015-06-18 | 2015-08-26 | 北京理工大学 | Aircraft multi-objective optimization method based on self-adaptive agent model |
CN105590141A (en) * | 2015-12-15 | 2016-05-18 | 东北大学 | Genetic algorithm initial population construction method applied to optimized design of complex products |
CN109118718A (en) * | 2018-07-09 | 2019-01-01 | 中国科学院、水利部成都山地灾害与环境研究所 | Rainfall I-D curve threshold value construction method, basin debris flow early-warning method occur for mud-rock flow |
CN109615118A (en) * | 2018-11-23 | 2019-04-12 | 泉州装备制造研究所 | Based on big data hazards control Informatization Service integrated control system and method |
Non-Patent Citations (3)
Title |
---|
Antonio C. Briza 等.Stock trading system based on the multi-objective particle swarm optimization of technical indicators on end-of-day market data.Applied Soft Computing.2011,第11卷(第1期),1191-1201. * |
吕志明.基于多代理模型的群智能优化算法研究.中国博士学位论文全文数据库信息科技辑.2020,(第01期),I140-42. * |
张猛.降雨引发泥石流人工神经网络预警模型剖析.黑龙江水利科技.2015,第43卷(第12期),55-56,90. * |
Also Published As
Publication number | Publication date |
---|---|
CN111860973A (en) | 2020-10-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113837477B (en) | Method, device and equipment for predicting power grid faults under typhoon disasters driven by data | |
CN110807562B (en) | Regional bridge risk prediction method and system | |
CN106021875B (en) | Multi-scale debris flow risk assessment method for seismic disturbance zone | |
CN111047099A (en) | Regional torrential flood risk prediction method and system | |
CN111861274A (en) | Water environment risk prediction and early warning method | |
CN104766129A (en) | Subway shield construction surface deformation warning method based on temporal and spatial information fusion | |
CN103093400A (en) | Adjacent building safety quantitative evaluation method in tunnel construction | |
Yang et al. | A CNN-LSTM model for tailings dam risk prediction | |
CN105260599A (en) | Rockburst dynamic prediction method based on BP neural network modeling | |
CN115966074B (en) | Machine learning-based shield cutter disc mud cake early warning method and system | |
CN114971301B (en) | Ecological interference risk identification and evaluation method based on automatic parameter adjustment optimization model | |
CN103512751A (en) | Bearing health state identification method based on probabilistic neural network | |
Cheng et al. | A Swarm-Optimized Fuzzy Instance-based Learning approach for predicting slope collapses in mountain roads | |
CN106780266B (en) | Principal component contribution degree parameter-based accident hotspot internal characteristic analysis and driving guidance method | |
CN110110425A (en) | A kind of slope sliding force prediction method based on reliability rule-based reasoning | |
CN115983093A (en) | Power transmission line tower footing geological disaster prediction method and device | |
CN113673119A (en) | Dynamic and static coupling evaluation method for coal mine rock burst danger based on Bayes method | |
CN116415481A (en) | Regional landslide hazard risk prediction method and device, computer equipment and storage medium | |
CN115359629A (en) | Regional landslide disaster early warning method based on sequential evaluation method | |
CN111860973B (en) | Debris flow intelligent early warning method based on multi-objective optimization | |
CN117238126A (en) | Traffic accident risk assessment method under continuous flow road scene | |
CN104462787A (en) | RST/IAHP-based aeronautical maintenance human factor reliability assessment method | |
CN103577700B (en) | Boat firefighting system interlock failure prediction method | |
CN116090696A (en) | Landslide geological disaster risk classification prediction method suitable for mountain railway line | |
Wengang et al. | Application of machine learning in slope stability assessment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |