A neural network-based open-pit coal mine slope deformation monitoring method and system

Abstract

The application discloses an open-pit coal mine slope deformation monitoring method and system based on a neural network, relates to the field of open-pit coal mine slope safety monitoring, and comprises the following steps: collecting surface multi-source monitoring data and in-situ deep sensing data through fusion, constructing an ideal slope deformation reference curve, calculating a comprehensive slope stability coefficient, and determining a slope safety state in combination with a preset threshold value; marking an abnormal state slope, inputting multi-dimensional time sequence data into a lightweight neural network model, and outputting a slope deformation trend, a potential instability position and an inducing cause, so that real-time monitoring, accurate early warning and intelligent diagnosis of the open-pit coal mine slope are realized, and the technical problems of single traditional slope monitoring data, early warning lag, high false alarm rate and inability to accurately locate hidden dangers are solved.

Description

Technical Field

[0001] This invention relates to the field of open-pit coal mine slope safety monitoring technology, and more specifically to a method and system for monitoring open-pit coal mine slope deformation based on neural networks. Background Technology

[0002] Currently, open-pit coal mine slopes are susceptible to geological disasters such as landslides, collapses, and mudslides due to multiple factors including mining disturbance, rainfall infiltration, soil and rock weathering, and changes in groundwater levels. Traditional slope monitoring focuses on single surface displacement, resulting in insufficient data dimensions, susceptibility to environmental interference, and low early warning accuracy. Manual inspections are inefficient, dangerous, and unable to provide real-time feedback. Conventional threshold judgments are prone to false alarms and missed alarms, and cannot intelligently diagnose deformation trends, instability locations, and triggering causes, making it difficult to meet the safety production needs of open-pit coal mines.

[0003] Therefore, how to propose a neural network-based method and system for monitoring slope deformation in open-pit coal mines, and overcome the shortcomings of existing technologies, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a method and system for monitoring slope deformation in open-pit coal mines based on neural networks, solving the technical problems of traditional monitoring methods such as single data sources, delayed early warning, high false alarm rate, and inability to accurately locate hidden dangers. It achieves multi-source data fusion from the slope surface and subsurface, accurate determination of stability status, and intelligent prediction of deformation trends. To achieve the above objectives, the present invention adopts the following technical solution: A neural network-based method for monitoring slope deformation in open-pit coal mines, comprising: Acquire dual-dimensional monitoring data of surface and subsurface areas on open-pit coal mine slopes; Based on historical slope stability data, the ideal deformation benchmark values ​​corresponding to surface and in-situ monitoring indicators under different working conditions are obtained. The real-time monitoring data are compared with the ideal deformation benchmark values ​​to calculate the comprehensive slope stability coefficient. The current safety status of the slope is determined based on whether the comprehensive stability coefficient of the slope falls within the preset safety range; Based on the current safety status of the slope, potential hazards are marked on the slope. The marked multi-source time series data is then input into a lightweight neural network model to output the deformation monitoring results of the open-pit coal mine slope.

[0005] Optionally, obtaining the ideal deformation benchmark values ​​corresponding to surface and in-situ monitoring indicators under different working conditions includes: Based on historical stable monitoring data, a baseline curve is fitted to show how each monitoring indicator changes over time and under different operating conditions. The work conditions are divided into several intervals based on rainfall, groundwater level, and slope load. Within each working condition interval, the moment with the smallest deformation dispersion is selected as the reference time point; Calculate the ideal deformation benchmark value of each monitoring indicator at the i-th benchmark time point.

[0006] Optionally, the calculation of the comprehensive stability coefficient of the slope includes: The comprehensive stability coefficient of the slope at point s : ; in, Let i be the ideal deformed baseline value of each monitoring indicator at the i-th baseline time point. This represents the measured value of the monitoring indicator at the i-th time point; is the time interval; m is the total number of time points; p is the proportionality coefficient.

[0007] Optionally, determining the current safety status of the slope includes: The overall stability coefficient and preset safety zone contrast: like Determine slope stability; like Determine slope warning, activate graded warning and generate inspection instructions; like The system determines that the slope is dangerous, triggers an emergency shutdown and personnel evacuation order, and simultaneously pushes the location of the hazard to the control platform.

[0008] Optionally, the method for determining the minimum deformation dispersion within the operating condition range is as follows: Discrete coefficients : ; in, These are the actual measured values ​​of the monitoring indicators; The weighting coefficient for the indicator; The total number of monitored indicators; The average value of the index is used; the coefficients of dispersion are sorted in ascending order, and the time corresponding to the first one is the reference time point with the minimum deformation dispersion.

[0009] Optionally, the dual-dimensional monitoring data of surface and subsurface includes surface data and subsurface data; the surface data includes slope surface displacement, crack width, rainfall, slope inclination, ambient temperature and humidity, and soil moisture content; the subsurface data includes deep soil displacement, rock and soil stress, pore water pressure, anchor bolt axial force, anchor cable tension, and groundwater level.

[0010] Optionally, it also includes conducting long-term trend analysis on stable slopes, calculating slope instability trend coefficients, and dynamically adjusting the frequency and density of slope inspections based on these coefficients. No. Slope instability trend coefficient : ; in, The conversion factor; This is the deep displacement deviation coefficient of the slope; , This is the proportionality coefficient; This is the lower limit of the safe range.

[0011] Optionally, the deep displacement deviation coefficient is obtained as follows: Based on historical data, fit the actual deep displacement curve of the slope. Compared with standard stable displacement curve Calculate the deep displacement deviation coefficient : ; in, , The start and end points of the monitoring period; This is the attenuation coefficient.

[0012] Optionally, the dynamic adjustment of slope inspection frequency and monitoring density based on coefficients includes: Will With preset threshold contrast: like ,according to Increase the frequency of inspections. Basic inspection frequency; like ≥ The current inspection frequency will remain unchanged.

[0013] Optionally, a neural network-based open-pit coal mine slope deformation monitoring system includes: a multi-source data acquisition module for acquiring surface and in-situ dual-dimensional monitoring data of the open-pit coal mine slope; The benchmark fitting module is used to obtain the ideal deformation benchmark values ​​corresponding to the surface and in-situ monitoring indicators under different working conditions based on historical slope stability data, compare the real-time monitoring data with the ideal deformation benchmark values, and calculate the comprehensive stability coefficient of the slope. The status determination module is used to determine the current safety status of the slope based on whether the comprehensive stability coefficient of the slope falls within the preset safety range. The intelligent diagnostic module is used to mark potential hazards on the slope based on its current safety status. It inputs the marked multi-source time-series data into a lightweight neural network model and outputs the deformation monitoring results of the open-pit coal mine slope.

[0014] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method and system for monitoring slope deformation in open-pit coal mines based on neural networks, which has the following beneficial effects: This invention proposes a neural network-based method for monitoring slope deformation in open-pit coal mines. The method includes: acquiring surface and subsurface monitoring data of the open-pit coal mine slope; obtaining ideal deformation benchmark values ​​corresponding to surface and subsurface monitoring indicators under different working conditions based on historical slope stability data; comparing real-time monitoring data with the ideal deformation benchmark values ​​to calculate the comprehensive slope stability coefficient; determining the current safety status of the slope based on whether the comprehensive slope stability coefficient falls within a preset safety range, classifying it into three levels: stable, warning, and dangerous; marking potential hazards on the slope based on its current safety status; inputting the marked multi-source time-series data into a lightweight neural network model to output the open-pit coal mine slope deformation monitoring results. This invention improves monitoring accuracy by complementing surface and subsurface multi-source data, eliminating interference from single data sources; constructing ideal deformation benchmarks based on historical data to adapt to different working conditions and reduce false alarm rates; achieving graded early warning and precise response through clear division of stable / warning / dangerous data; predicting deformation trends, locating hazard positions, and analyzing causes in real time to support scientific handling; and improving slope safety management efficiency by identifying potential risks in advance and optimizing maintenance resources. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 A schematic diagram of the process for a neural network-based method for monitoring slope deformation in open-pit coal mines, provided by this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention discloses a method for monitoring slope deformation in open-pit coal mines based on neural networks, such as... Figure 1 As shown, it includes: Acquire dual-dimensional monitoring data of surface and subsurface areas on open-pit coal mine slopes; Based on historical slope stability data, the ideal deformation benchmark values ​​corresponding to surface and in-situ monitoring indicators under different working conditions are obtained. The real-time monitoring data are compared with the ideal deformation benchmark values ​​to calculate the comprehensive slope stability coefficient. Based on whether the comprehensive stability coefficient of the slope falls within the preset safe range, the current safety status of the slope is determined and divided into three levels: stable, warning, and dangerous. Based on the current safety status of the slope, potential hazards are marked on the slope. The marked multi-source time series data is then input into a lightweight neural network model to output the deformation monitoring results of the open-pit coal mine slope.

[0019] Furthermore, the output of open-pit coal mine slope deformation monitoring results includes: output slope deformation trend, potential instability areas, causes of deformation, and treatment suggestions.

[0020] Furthermore, obtaining the ideal deformation benchmark values ​​corresponding to surface and in-situ monitoring indicators under different working conditions includes: Based on historical stable monitoring data, a baseline curve is fitted to show how each monitoring indicator changes over time and under different operating conditions. The work conditions are divided into several intervals based on rainfall, groundwater level, and slope load. Within each working condition interval, the moment with the smallest deformation dispersion is selected as the reference time point; Calculate the ideal deformation benchmark value of each monitoring indicator at the i-th benchmark time point; ; Calculate the ideal deformation benchmark value of each monitoring indicator at the i-th benchmark time point. ; in, The transformation function between the change and weight of the k-th influencing factor is obtained by training with historical data; For the corresponding indicator Changes in each influencing factor; The total number of impact factors; This represents the measured value of the indicator at the previous baseline time point.

[0021] Furthermore, the calculation of the comprehensive stability coefficient of the slope includes: The comprehensive stability coefficient of the slope at point s : ; in, Let i be the ideal deformed baseline value of each monitoring indicator at the i-th baseline time point. This represents the measured value of the monitoring indicator at the i-th time point; denoted as time interval; m as the total number of time points; and p as a proportionality coefficient, determined from historical data and field experience.

[0022] Furthermore, determining the current safety status of the slope includes: The overall stability coefficient and preset safety zone contrast: like Determine slope stability; like Determine slope warning, activate graded warning and generate inspection instructions; like The system determines that the slope is dangerous, triggers an emergency shutdown and personnel evacuation order, and simultaneously pushes the location of the hazard to the control platform.

[0023] Furthermore, the method for determining the minimum deformation dispersion within the aforementioned operating condition range is as follows: Discrete coefficients : ; in, These are the actual measured values ​​of the monitoring indicators; The weighting coefficient for the indicator; The total number of monitored indicators; The average value of the index is used; the coefficients of dispersion are sorted in ascending order, and the time corresponding to the first one is the reference time point with the minimum deformation dispersion.

[0024] Furthermore, the dual-dimensional monitoring data of the surface and the interior includes surface data and interior data; the surface data includes slope surface displacement, crack width, rainfall, slope inclination, ambient temperature and humidity, and soil moisture content; the interior data includes deep soil displacement, rock and soil stress, pore water pressure, anchor bolt axial force, anchor cable tension, and groundwater level.

[0025] Furthermore, this also includes conducting long-term trend analysis on stable slopes, calculating slope instability trend coefficients, and dynamically adjusting the frequency and density of slope inspections based on these coefficients. No. Slope instability trend coefficient : ; in, The conversion factor; This is the deep displacement deviation coefficient of the slope; , This is the proportionality coefficient; This is the lower limit of the safe range.

[0026] Furthermore, the method for obtaining the deep displacement deviation coefficient is as follows: Based on historical data, fit the actual deep displacement curve of the slope. Compared with standard stable displacement curve Calculate the deep displacement deviation coefficient : ; in, , The start and end points of the monitoring period; The attenuation coefficient is determined by field tests.

[0027] Furthermore, the dynamic adjustment of slope inspection frequency and monitoring density based on coefficients includes: Will With preset threshold contrast: like ,according to Increase the frequency of inspections. Basic inspection frequency; like ≥ The current inspection frequency will remain unchanged.

[0028] In a specific implementation, an open-pit coal mine slope deformation monitoring system based on neural networks includes: a multi-source data acquisition module for acquiring surface and underground dual-dimensional monitoring data of the open-pit coal mine slope; The benchmark fitting module is used to obtain the ideal deformation benchmark values ​​corresponding to the surface and in-situ monitoring indicators under different working conditions based on historical slope stability data, compare the real-time monitoring data with the ideal deformation benchmark values, and calculate the comprehensive stability coefficient of the slope. The status determination module is used to determine the current safety status of the slope based on whether the comprehensive stability coefficient of the slope falls within the preset safety range. The intelligent diagnostic module is used to mark potential hazards on the slope based on its current safety status. It inputs the marked multi-source time-series data into a lightweight neural network model and outputs the deformation monitoring results of the open-pit coal mine slope.

[0029] In a specific implementation, the step of inputting the labeled multi-source time-series data into a lightweight neural network model and outputting the open-pit coal mine slope deformation monitoring results includes: To address the challenges of multi-source time-series data, low computational power, and high real-time requirements for slope deformation monitoring in open-pit coal mines, a lightweight neural network trained using a combination of a large teacher network and multi-layer early knowledge fusion distillation is employed. This approach significantly improves the accuracy of deformation trend prediction, hazard location, and cause diagnosis while maintaining the model's lightweight nature. The specific implementation steps are as follows: 1. Dataset Construction and Preprocessing Using surface and in-situ multi-source time-series monitoring data of open-pit coal mine slopes as training samples, each sample has 12-dimensional time-series features: Surface data: surface displacement, crack width, rainfall, slope angle, temperature and humidity, soil moisture content; In-situ data: deep soil displacement, soil and rock stress, pore water pressure, anchor bolt axial force, anchor cable tension, and groundwater level.

[0030] Standardize the data to eliminate dimensional differences: ; In the formula: x represents the original monitoring data; μ represents the characteristic mean; σ represents the characteristic standard deviation; This is the input data after standardization.

[0031] The dataset was divided into training and validation sets in a 9:1 ratio, and the labels were classified as slope stability / early warning / danger.

[0032] 2. Construction of the Teacher-Net Neural Network

[0033] The teacher network is used to fully learn the complex temporal characteristics of slope deformation, and its structure is as follows: Input layer: Receives 12-dimensional normalized time series data; Backbone network: 12-layer residual network, with 256 neurons in each layer, and residual connections between layers to prevent gradient vanishing; Prediction output layer: Fully connected layer + Softmax classifier, outputting three-dimensional prediction results of deformation trend, instability location, and inducing cause.

[0034] The teacher model training loss uses multi-class cross-entropy loss: ; In the formula: N is the number of samples; K=3 is the number of safety status categories; Let the i-th sample be the true label of the k-th class; Predict probabilities for the teacher model.

[0035] 3. Construction of a Lightweight Student Neural Network (Student-Net)

[0036] Design a lightweight network with low parameter count and high inference speed for deployment at the edge of mining farms: Backbone network: 4-layer residual structure, 64 neurons per layer, with parameters compressed by 50 to 60 times compared to the teacher model; Input / output dimensions: Fully aligned with the teacher model to ensure distillation compatibility; Activation function: ReLU6 is used to reduce computation and is suitable for embedded devices.

[0037] 4. Multi-layered early knowledge fusion and distillation training

[0038] A distillation strategy is adopted, with fixed parameters for the teacher model and step-by-step learning for the student model, to fully utilize the intermediate layer features of the teacher model and avoid underfitting of the student model. Fix all parameters of the already trained teacher network; Each layer i of the student model is jointly supervised by the output features of the previous i layers of the teacher model; The student model's final output fits the teacher model's final output, thus completing knowledge transfer.

[0039] The total loss of the student model is a weighted average of the real label supervision loss and the multi-layer early knowledge distillation loss: ; (1) Multi-class cross-entropy supervised loss ; in: Label the actual safety status of the slope; Predict the probability for the student model; K=3.

[0040] (2) Loss of multi-layered early knowledge fusion distillation ; In the formula: =4 represents the total number of layers in the student model; i represents the index of the current layer in the student model; j represents the index of the i-th layer in the teacher model. The i-th layer of the student model outputs features; The j-th layer of the teacher model outputs the features; , D( is a feature mapping function that unifies feature dimensions;) The mean squared error loss measures the distance to the characteristic distribution. λ is the distillation weight, which is taken as 0.6~0.8 in this scenario.

[0041] 5. Model Training and Inference Deployment

[0042] Training configuration: learning rate 1e-4, batch size 32, 200 iterations, using Adam optimizer; Inference input: Real-time multi-source time series data is standardized and then input into the lightweight model; Output results: predicted slope deformation trend, coordinates of potential instability areas, and confidence levels of inducing factors (rainfall / water level / load); Deployment performance: Parameter count < 30,000, single-step inference < 80ms, can be directly deployed in edge computing units, field PLCs, and embedded terminals in the mine; Improved accuracy: Compared with the non-distilled lightweight model, the early warning accuracy is improved by ≥35%, and the false alarm rate is reduced by ≥40%, meeting the requirements for real-time monitoring and accurate early warning of open-pit coal mine slopes.

[0043] In a specific embodiment, the western slope of an open-pit coal mine is used as the monitoring object. This slope is 185m high and 620m long, with the rock mass mainly composed of alternating layers of mudstone and sandstone. During the rainy season, it is susceptible to landslides due to rainfall and rising groundwater levels. The method and system of this invention are used for full-cycle real-time monitoring, and the specific implementation steps are as follows: I. On-site deployment and multi-source data acquisition Surface monitoring equipment deployment Surface displacement monitoring: Install 3 GNSS displacement monitoring stations, with a sampling frequency of 1 time / 10min; Crack monitoring: Six crack gauges were installed to monitor changes in the width of tension cracks on the slope; Environmental monitoring: Deploy 2 sets each of rain gauges, temperature and humidity sensors, and soil moisture sensors; Slope inclination: Four inclination sensors were installed to monitor the slope inclination.

[0044] Deployment of deep-ground monitoring equipment

[0045] Deep soil displacement: Five inclinometer tubes were installed, with a monitoring depth of 0~120m; Rock and soil stress / pore water pressure: Install 8 stress gauges and 8 pore water pressure gauges; Anchoring force monitoring: Install 6 sets each of anchor bolt axial force gauges and anchor cable tension gauges; Groundwater level: Three water level monitoring wells are installed to collect the water level depth in real time.

[0046] Data acquisition rules: All sensor data are uniformly connected to the on-site edge computing unit, with an acquisition frequency of 5 minutes / time, forming 12-dimensional time-series monitoring data of the ground surface and underground.

[0047] II. Construction of Ideal Deformation Reference Values

[0048] The working condition intervals are divided based on the mine's three-year historical stable monitoring data, into three categories according to rainfall, groundwater level, and slope load: Condition A: No rainfall, groundwater level depth > 30m, slope load < 50kPa; Condition B: Light rain (daily rainfall <10mm), groundwater level depth 15~30m, slope load 50~100kPa; Operating Condition C: Moderate to heavy rain (daily rainfall ≥ 10 mm), groundwater level depth < 15 m, slope load ≥ 100 kPa.

[0049] The reference time point is selected by calculating the dispersion of monitoring data within each operating condition interval using the coefficient of dispersion formula, and then selecting the moment with the smallest coefficient of dispersion as the reference time point. Operating Condition A reference time: March 12, 14:00; Baseline time for operating condition B: 09:30 on June 8th; Operating condition C reference time point: July 21, 16:15.

[0050] The ideal deformation benchmark value is calculated based on the benchmark time point data. The time-operating condition benchmark curves of each monitoring index are fitted to obtain the ideal deformation benchmark values ​​of 12 indicators such as surface displacement, deep displacement, and pore water pressure under three types of operating conditions.

[0051] III. Calculation of Comprehensive Slope Stability Coefficient and Determination of Safety Status

[0052] The comprehensive stability coefficient was calculated using real-time data from July 20th to July 22nd (under moderate rain conditions) for this slope, which was then substituted into the formula: Measured surface displacement: 12.8 mm (reference value 8.2 mm); Deep displacement (at 60m): 4.5mm (reference value 1.1mm); Pore ​​water pressure: 18.6 kPa (reference value 9.3 kPa); The comprehensive stability coefficient of the slope was calculated. =1.82.

[0053] The safety status classification is preset to determine the safe zone of this slope: =0.3, =1.5, =2.5 (1.5) < (1.82)≤ (2.5), determined to be a warning state; The system automatically activates a yellow-level early warning, generates a key slope inspection instruction, and pushes it to the on-site control platform.

[0054] IV. Lightweight Neural Network Intelligent Diagnosis

[0055] The model input consists of standardized 12-dimensional time-series data (displacement, stress, water level, rainfall, etc.) after the warning label is applied, and then input into a lightweight distillation neural network.

[0056] Model output results

[0057] Deformation trend: The slope displacement rate will continue to rise in the next 72 hours, and the cumulative deformation may reach 22 mm; Potential instability location: Upper and middle part of the slope (chainage K0+220~K0+280), with a depth of 30~60m in the soil and rock mass; Causes: Rainfall infiltration leads to increased pore water pressure (92% confidence level) + rise in groundwater level (87% confidence level).

[0058] The recommended measures are to immediately implement temporary drainage in the unstable area, cover the slope with impermeable geotextile, and increase the monitoring frequency to once per minute.

[0059] V. Instability Trend Analysis and Dynamic Adjustment of Inspection Frequency

[0060] Instability trend coefficient calculation

[0061] Calculate the deep displacement deviation coefficient of the slope. =0.38, substituting into the formula yields the instability trend coefficient. =0.42; Inspection frequency adjustment Preset threshold =0.6, < Increase the frequency of inspections according to the formula: The basic inspection frequency F = 1 time / day, and the adjusted inspection frequency = 2 times / day; The monitoring density has been increased from once every 5 minutes to once every 1 minute.

[0062] VI. Implementation Results

[0063] With an early warning accuracy rate of ≥96% and a false alarm rate reduced by 42%, risk alerts are issued 18-24 hours earlier than traditional threshold-based early warnings. The system accurately locates unstable areas, avoiding blind inspections and improving maintenance efficiency by 50%. The model's single-step inference time is ≤70ms, allowing for deployment on embedded devices on-site to meet the real-time monitoring needs of open-pit coal mines. The system successfully mitigated the risk of slope landslides caused by this rainfall, ensuring mine production safety.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0065] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for monitoring slope deformation in open-pit coal mines based on neural networks, characterized in that, include: Acquire dual-dimensional monitoring data of surface and subsurface areas on open-pit coal mine slopes; Based on historical slope stability data, the ideal deformation benchmark values ​​corresponding to surface and in-situ monitoring indicators under different working conditions are obtained. The real-time monitoring data are compared with the ideal deformation benchmark values ​​to calculate the comprehensive slope stability coefficient. The current safety status of the slope is determined based on whether the comprehensive stability coefficient of the slope falls within the preset safety range; Based on the current safety status of the slope, potential hazards are marked on the slope. The marked multi-source time series data is then input into a lightweight neural network model to output the deformation monitoring results of the open-pit coal mine slope.

2. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 1, characterized in that, The process of obtaining the ideal deformation benchmark values ​​corresponding to surface and in-situ monitoring indicators under different working conditions includes: Based on historical stable monitoring data, a baseline curve is fitted to show how each monitoring indicator changes over time and under different operating conditions. The work conditions are divided into several intervals based on rainfall, groundwater level, and slope load. Within each working condition interval, the moment with the smallest deformation dispersion is selected as the reference time point; Calculate the ideal deformation benchmark value of each monitoring indicator at the i-th benchmark time point.

3. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 2, characterized in that, The calculation of the comprehensive stability coefficient of the slope includes: The comprehensive stability coefficient of the slope at point s : ； in, Let i be the ideal deformed baseline value of each monitoring indicator at the i-th baseline time point. This represents the measured value of the monitoring indicator at the i-th time point; is the time interval; m is the total number of time points; p is the proportionality coefficient.

4. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 3, characterized in that, The determination of the current safety status of the slope includes: The overall stability coefficient and preset safety zone contrast: like Determine slope stability; like Determine slope warning, activate graded warning and generate inspection instructions; like The system determines that the slope is dangerous, triggers an emergency shutdown and personnel evacuation order, and simultaneously pushes the location of the hazard to the control platform.

5. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 2, characterized in that, The method for determining the minimum deformation dispersion within the specified operating range is as follows: Discrete coefficients : ； in, These are the actual measured values ​​of the monitoring indicators; The weighting coefficient of the indicator; The total number of monitored indicators; The average value of the index is used; the discrete coefficients are sorted in ascending order, and the time corresponding to the first one is the reference time point with the minimum deformation dispersion.

6. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 1, characterized in that, The dual-dimensional monitoring data, encompassing both surface and subsurface data, includes surface data and subsurface data. The surface data includes slope surface displacement, crack width, rainfall, slope inclination, ambient temperature and humidity, and soil moisture content. The subsurface data includes deep soil displacement, soil and rock stress, pore water pressure, anchor bolt axial force, anchor cable tension, and groundwater level.

7. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 4, characterized in that, This also includes conducting long-term trend analysis on stable slopes, calculating slope instability trend coefficients, and dynamically adjusting the frequency and density of slope inspections based on these coefficients. No. Slope instability trend coefficient : ； in, The conversion factor; This refers to the deep displacement deviation coefficient of the slope. , This is the proportionality coefficient; This is the lower limit of the safe range.

8. The method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 7, characterized in that, The method for obtaining the deep displacement deviation coefficient is as follows: Based on historical data, fit the actual deep displacement curve of the slope. Compared with standard stable displacement curve Calculate the deep displacement deviation coefficient : ； in, , The start and end points of the monitoring period; This is the attenuation coefficient.

9. A method for monitoring slope deformation in open-pit coal mines based on neural networks according to claim 7, characterized in that, The dynamic adjustment of slope inspection frequency and monitoring density based on coefficients includes: Will With preset threshold contrast: like ,according to Increase the frequency of inspections. Basic inspection frequency; like ≥ The current inspection frequency will remain unchanged.

10. A neural network-based open-pit coal mine slope deformation monitoring system, characterized in that, include: The multi-source data acquisition module is used to acquire dual-dimensional monitoring data of the surface and underground of open-pit coal mine slopes; The benchmark fitting module is used to obtain the ideal deformation benchmark values ​​corresponding to the surface and in-situ monitoring indicators under different working conditions based on historical slope stability data, compare the real-time monitoring data with the ideal deformation benchmark values, and calculate the comprehensive stability coefficient of the slope. The status determination module is used to determine the current safety status of the slope based on whether the comprehensive stability coefficient of the slope falls within the preset safety range. The intelligent diagnostic module is used to mark potential hazards on the slope based on its current safety status. It inputs the marked multi-source time-series data into a lightweight neural network model and outputs the deformation monitoring results of the open-pit coal mine slope.

Images