Rainfall type loess slope instability early warning method and system based on multi-source data fusion

Abstract

The invention relates to the technical field of rainfall type loess side slope instability early warning methods, and discloses a rainfall type loess side slope instability early warning method and system based on multi-source data fusion, and the method comprises the steps: S1, arranging a sensor network on a side slope, and collecting rainfall capacity, surface displacement and soil pressure data in real time; s2, preprocessing the collected multi-source data, including data cleaning, missing value filling and data standardization; and S3, extracting characteristic indexes capable of reflecting the stability state of the slope from the preprocessed data. Environmental data of the loess slope, such as rainfall, evaporation capacity, temperature and humidity, are acquired by using a sensor network; the deformation data comprises earth surface displacement, deep displacement and crack width; the mechanical data comprises soil pressure, pore water pressure and soil moisture content; image data: a slope surface image and a crack development image are collected, an early warning model is constructed based on a machine learning algorithm, and data features can be automatically learned.

Description

Technical Field

[0001] The present invention relates to the technical field of early warning methods for rainfall-induced instability of loess slopes, and specifically to an early warning method and system for rainfall-induced instability of loess slopes based on multi-source data fusion. Background Art

[0002] Landslides and collapses caused by unstable slopes account for 84.5% of the total number of geological disasters. Due to the extensive development of loess slopes, the thickness of loess in some areas is relatively large, up to more than 400 meters at most. Unsaturated loess also has relatively high strength, so loess slopes of hundreds of meters can often be formed, which are major hidden dangers of geological disasters. Loess slopes are extremely prone to instability and failure under rainfall conditions, seriously threatening people's lives and property safety.

[0003] Traditional slope monitoring methods mostly use single indicators, which are difficult to comprehensively reflect the stability state of slopes. Moreover, the early warning thresholds are mostly determined based on experience and lack theoretical basis, resulting in low early warning accuracy. Therefore, it is of great significance to invent an early warning method and system for rainfall-induced instability of loess slopes based on multi-source data fusion. Summary of the Invention

[0004] The present invention provides an early warning method and system for rainfall-induced instability of loess slopes based on multi-source data fusion, which has the advantages of multi-data fusion and solves the problems raised in the above background art.

[0005] The present invention provides the following technical solutions: An early warning method and system for rainfall-induced instability of loess slopes based on multi-source data fusion, including the following steps;

[0006] Step S1: Arrange a sensor network on the slope;

[0007] Step S2: Preprocess the multi-source data collected;

[0008] Step S3: Extract characteristic indicators from the preprocessed data that can reflect the stability state of the slope.

[0009] Preferably, step S1 specifically includes:

[0010] Collect the environmental data of the slope through the sensor network arranged on the slope: rainfall, evaporation, temperature, humidity; deformation data: surface displacement, deep displacement, crack width; mechanical data: earth pressure, pore water pressure, soil moisture content; image data: slope surface image, crack development image.

[0011] Preferably, step S3 specifically includes:

[0012] Extract the rainfall characteristic indexes on the slope: cumulative rainfall, rainfall intensity, rainfall duration; deformation characteristic indexes: displacement rate, displacement acceleration, crack propagation rate; mechanical characteristic indexes: soil pressure change rate, pore water pressure change rate, soil moisture content change rate; and image characteristic indexes: crack length, width, and trend.

[0013] Preferably, the warning model construction unit: constructs a rainfall-induced loess slope instability warning model based on a machine learning algorithm, uses the extracted characteristic indexes as inputs, and the slope stability state as outputs, trains and optimizes the model, and the machine learning algorithm uses GRU to process data information;

[0014] Data prediction unit: determines the thresholds corresponding to different warning levels using statistical methods based on historical data and model prediction results;

[0015] Data training unit: inputs the real-time collected data into the trained warning model.

[0016] Preferably, the warning model construction unit includes;

[0017] Update gate formula: z t =g(W (x) x t +U (x) h t-1 )

[0018] Reset gate formula: r t =g(W (N) x t +U (r) h t-1 )

[0019] Current memory content: h′ i =tanh(Wx i +r i )U i-1 )

[0020] Current time final memory: h t =z t ⊙h t-1 +(1-z t )⊙h。

[0021] Model training, and its training steps include:

[0022] a. Initialization of parameters: Parameters such as weights and learning rates need to be set in order to be correctly applied to the model;

[0023] b. Establish the loss function and set the stopping condition: The loss function is an important learning criterion that can be used to measure the performance of a network and is often related to the optimization quantity. The mean squared error loss function is used to calculate the mean squared error between them, and its calculation formula is as follows:

[0024]

[0025] where N is the length of the input sequence, is the predicted value of the model, and is the true value;

[0026] c. By applying the Adam optimizer, the weight parameters and bias parameters in the model can be effectively optimized;

[0027] d. Through careful adjustment of the parameters and functions, an effective network training process can be achieved, that is, according to the preset number of training times, continuously adjust the parameters of the model until a certain specific threshold is reached. In this way, even if the expected number of training times is not reached, the model can stop running when the threshold is met. After 3 epochs, if the loss function has not improved significantly, the deep learning model will be terminated.

[0028] Preferably, the data prediction unit includes;

[0029] Blue warning: The slope is in a stable state, but there are potential risks;

[0030] Yellow warning: The slope is in a sub-stable state and needs to be monitored intensively;

[0031] Orange warning: The slope is in an unstable state and engineering measures need to be taken;

[0032] Red warning: The slope is in an extremely unstable state and may undergo instability failure at any time.

[0033] The present invention has the following beneficial effects:

[0034] 1. The rainfall-induced loess slope instability warning method and system based on multi-source data fusion collect the environmental data of the loess slope: rainfall, evaporation, temperature, humidity; deformation data: surface displacement, deep displacement, crack width; mechanical data: earth pressure, pore water pressure, soil moisture content; image data: slope surface image, crack development image through the sensor network, and at the same time construct a warning model based on machine learning algorithms, which can automatically learn data features and improve the warning accuracy.

[0035] 2. The rainfall-induced loess slope instability warning method and system based on multi-source data fusion dynamically adjusts the warning threshold based on historical data and model prediction results. Blue warning: The slope is in a stable state but there are potential risks; Yellow warning: The slope is in a sub-stable state and monitoring needs to be strengthened; Orange warning: The slope is in an unstable state and engineering measures need to be taken; Red warning: The slope is in an extremely unstable state and may undergo instability failure at any time. At the same time, the stability state of the slope can be monitored in real time, and warning information can be released in a timely manner, providing a scientific basis for disaster prevention and mitigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic flowchart of the present invention;

[0037] Figure 2 It is a schematic structural block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] Please refer to Figure 1 - Figure 2 , a rainfall-induced loess slope instability warning method and system based on multi-source data fusion, comprising the following steps:

[0040] Step S1: Arrange a sensor network on the slope;

[0041] Step S2: Preprocess the collected multi-source data;

[0042] Step S3: Extract characteristic indicators from the preprocessed data that can reflect the stability state of the slope;

[0043] In a preferred embodiment, step S1 specifically includes;

[0044] Collect the environmental data of the slope through the sensor network arranged on the slope: rainfall, evaporation, temperature, humidity; deformation data: surface displacement, deep displacement, crack width; mechanical data: earth pressure, pore water pressure, soil moisture content; image data: slope surface image, crack development image.

[0045] In a preferred embodiment, step S3 specifically includes;

[0046] Extract the rainfall characteristic indexes on the slope: cumulative rainfall, rainfall intensity, rainfall duration; deformation characteristic indexes: displacement rate, displacement acceleration, crack propagation rate; mechanical characteristic indexes: earth pressure change rate, pore water pressure change rate, soil moisture content change rate; and image characteristic indexes: crack length, width, and trend.

[0047] In a preferred embodiment, the early warning model construction unit: constructs a rainfall-induced loess slope instability early warning model based on a machine learning algorithm, uses the extracted characteristic indexes as inputs, and the slope stability state as outputs, and trains and optimizes the model. The machine learning algorithm uses GRU to process data information;

[0048] The data prediction unit: determines the thresholds corresponding to different early warning levels using statistical methods based on historical data and model prediction results;

[0049] In a preferred embodiment, the data training unit: inputs the real-time collected data into the trained early warning model.

[0050] The early warning model construction unit includes:

[0051] Update gate formula: z t =g(W (x) x t +U (x) h t-1 )

[0052] Reset gate formula: r t =g(W (N) x t +U (r) h t-1 )

[0053] Current memory content: h′ i =tanh(Wx i +r i )U i-1 )

[0054] Current time final memory: h t =z t ⊙h t-1 +(1-z t )⊙h。

[0055] Model training, and its training steps include:

[0056] a. Initialization of parameters: Parameters such as weights and learning rates need to be set in order to be correctly applied to the model;

[0057] b. Establish the loss function and set the stopping condition: The loss function is an important learning criterion that can be used to measure the performance of a network. It is often related to the optimization quantity. The mean squared error loss function is used to calculate the mean squared error between them. The calculation formula is as follows:

[0058]

[0059] where N is the length of the input sequence, is the predicted value of the model, and is the true value;

[0060] c. By applying the Adam optimizer, the weight parameters and bias parameters in the model can be effectively optimized;

[0061] d. Through careful adjustment of the parameters and functions, an effective network training process can be achieved, that is, according to the preset number of training times, continuously adjust the parameters of the model until a certain specific threshold is reached. In this way, even if the expected number of training times is not reached, the model can stop running when the threshold is met. After 3 epochs, if the loss function has not improved significantly, the deep learning model will be terminated.

[0062] In a preferred embodiment, step S5 specifically includes:

[0063] Blue warning: The slope is in a stable state but there are potential risks;

[0064] Yellow warning: The slope is in a sub-stable state and needs to be monitored intensively;

[0065] Orange warning: The slope is in an unstable state and engineering measures need to be taken;

[0066] Red warning: The slope is in an extremely unstable state and may undergo instability failure at any time.

[0067] Working principle: Through the sensor network arranged on the slope, rainfall, surface displacement, and earth pressure data are collected in real time;

[0068] Environmental data: rainfall, evaporation, temperature, humidity;

[0069] Deformation data: surface displacement, deep displacement, crack width;

[0070] Mechanical data: earth pressure, pore water pressure, soil moisture content;

[0071] Image data: slope surface image, crack development image.

[0072] Preprocess the collected multi-source data, including data cleaning, missing value filling, and data standardization;

[0073] Extract characteristic indicators that can reflect the stability state of the slope from the preprocessed data;

[0074] This indicator includes:

[0075] Rainfall characteristic indicators: cumulative rainfall, rainfall intensity, rainfall duration;

[0076] Deformation characteristic indicators: displacement rate, displacement acceleration, crack propagation rate;

[0077] Mechanical characteristic indicators: earth pressure change rate, pore water pressure change rate, soil moisture content change rate;

[0078] Image characteristic indicators: crack length, width, orientation

[0079] Construct a rainfall-induced loess slope instability warning model based on machine learning algorithms. Use the extracted characteristic indicators as input and the slope stability state as output to train and optimize the model. The machine learning algorithm uses GRU to process data information. The basic architecture is as follows:

[0080] Update gate formula: z t = g(W (x) x t + U (x) h t-1 )

[0081] Reset gate formula: r t = g(W (N) x t + U (r) h t-1 )

[0082] Current memory content: h′ i = tanh(Wx i + r i )U i-1 )

[0083] Current time final memory: h t = z t ⊙ h t-1 + (1 - z t ) ⊙ h.

[0084] Model training, and its training steps include:

[0085] a. Initialization of parameters: Parameters such as weights and learning rates need to be set in order to be correctly applied to the model;

[0086] b. Establish the loss function and set the stopping condition: The loss function is an important learning criterion that can be used to measure the performance of a network and is often related to the optimization quantity. The mean squared error loss function is used to calculate the mean squared error between them, and its calculation formula is as follows:

[0087]

[0088] where N is the length of the input sequence, is the predicted value of the model, and y i is the true value;

[0089] c. By applying the Adam optimizer, the weight parameters and bias parameters in the model can be effectively optimized;

[0090] d. Through careful adjustment of the parameters and functions, an effective network training process can be achieved, that is, according to the preset number of training times, continuously adjust the parameters of the model until a certain specific threshold is reached. In this way, even if the expected number of training times is not reached, the model can stop running when the threshold is met. After 3 epochs, if the loss function has not improved significantly, the deep learning model will be terminated

[0091] At this time, based on historical data and model prediction results, statistical methods are used to determine the thresholds corresponding to different warning levels. For example:

[0092] Blue warning: The slope is in a stable state but there are potential risks;

[0093] Yellow warning: The slope is in a sub - stable state and needs to be monitored intensively;

[0094] Orange warning: The slope is in an unstable state and engineering measures need to be taken;

[0095] Red warning: The slope is in an extremely unstable state and may undergo instability failure at any time.

[0096] Input the real - time collected data into the trained warning model, calculate the slope stability state, and issue corresponding warning information according to the warning threshold.

[0097] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "including", "comprising" or any other variant thereof is intended to cover non - exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or elements inherent to such process, method, article or device.

[0098] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A rainfall-induced loess slope instability early warning method based on multi-source data fusion, characterized in that: The following steps are involved: Step S1: Arrange a sensor network on the slope; Step S2: preprocessing the collected multi-source data; Step S3: extracting characteristic indicators that can reflect the stability state of the slope from the preprocessed data.

2. The rainfall-induced loess slope instability early warning method based on multi-source data fusion according to claim 1 is characterized by: Step S1 specifically includes: Through the sensor network arranged on the slope, the environmental data of the slope: rainfall, evaporation, temperature, humidity; deformation data: surface displacement, deep displacement, crack width; mechanical data: soil pressure, pore water pressure, soil moisture content; image data: slope surface image, crack development image are collected.

3. The rainfall-induced loess slope instability early warning method based on multi-source data fusion according to claim 1 is characterized by: Step S3 specifically includes: Characteristics of rainfall on the slope: cumulative rainfall, rainfall intensity, and rainfall duration; Deformation characteristic indicators: displacement rate, displacement acceleration, crack extension rate; Mechanical characteristic indicators: soil pressure change rate, pore water pressure change rate, soil moisture content change rate; Characteristic indicators such as crack length, width and direction are extracted.

4. A rainfall-induced loess slope instability early warning system based on multi-source data fusion according to claims 1-3, characterized in that: Early warning model construction unit: construct a rainfall-induced loess slope instability early warning model based on machine learning algorithm, take the extracted characteristic indicators as input, and the slope stability state as output, train and optimize the model, and the machine learning algorithm uses GRU to process data information; Data prediction unit: Based on historical data and model prediction results, statistical methods are used to determine the thresholds corresponding to different warning levels; Data training unit: input the real-time collected data into the trained early warning model.

5. The rainfall-induced loess slope instability early warning system based on multi-source data fusion according to claim 4 is characterized by: The early warning model building unit includes: Update gate formula: z t =g(W (x) x t +U (x) h t-1 ) Reset gate formula: r t =g(W (N) x t +U (r) h t-1 ) Current memory content: h′ i =tanh(Wx i +r i ) i-1 ) Current time final memory: h t =z t ⊙h t-1 +(1-z t )⊙h. Model training, the training steps include: a. Parameter initialization: setting the relevant parameters of weight and learning rate; b. Establish the loss function and set the stopping condition: Use the mean square error loss function to calculate the mean square error between them. The calculation formula is as follows: Among them, N is the length of the input sequence, is the predicted value of the model, and is the true value; c. Optimize the weight and bias parameters in the model by applying the Adam optimizer; d. According to the pre-set number of training times, the model parameters are continuously adjusted until a certain threshold is reached. If the expected number of training times is not reached, the model can stop running when the threshold is met. After 3 epochs, if the loss function has not been significantly improved, the deep learning model will be terminated.

6. The rainfall-induced loess slope instability early warning system based on multi-source data fusion according to claim 4 is characterized by: The data prediction unit includes: Blue warning: The slope is stable, but there are potential risks; Yellow warning: The slope is in an unstable state and requires enhanced monitoring; Orange warning: The slope is in an unstable state and engineering measures are required; Red warning: The slope is in an extremely unstable state and may become unstable and damaged at any time.

Images