Intelligent blasting parameter calculation platform

Abstract

The invention discloses a blasting parameter intelligent computing platform, which belongs to the field of blasting engineering, and comprises a distributed computing resource scheduling module, which adopts a distributed computing architecture to decompose a computing task into a plurality of sub-tasks, and distributes the sub-tasks to different computing nodes for parallel computing; through an intelligent scheduling algorithm, according to factors such as the load condition and the computing power of each computing node, computing tasks are dynamically allocated, dispersed computing resources in a network are fully utilized, the utilization rate of the computing resources is improved, and dependence on single high-performance computing equipment is reduced; data are processed in real time through the edge calculation preprocessing module, and the data transmission time is shortened; the lightweight calculation model library can quickly complete blasting parameter calculation; and the real-time feedback and dynamic optimization module realizes real-time adjustment and optimization of blasting parameters, so that accurate blasting parameters can be provided in time according to actual conditions in the tunnel construction process, and the construction progress is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of blasting engineering, and in particular to an intelligent calculation platform for blasting parameters. Background Art

[0002] Among existing blasting parameter calculation technologies, Patent 202410386738 discloses a tunnel blasting parameter acquisition method based on artificial intelligence and simulation. This method uses numerical simulation and machine learning to iteratively optimize surrounding rock parameters. While this improves the accuracy of blasting parameter calculation, it still has significant limitations:

[0003] High dependence on computing resources: This method relies on a single high-performance computing device to perform complex numerical simulations. For multi-section surrounding rock blasting parameter calculations or multi-scheme comparison scenarios, large amounts of data need to be processed serially, resulting in calculations taking several hours or even longer. In addition, the hardware cost is high, making it difficult to apply to small and medium-sized projects with limited resources.

[0004] Lack of real-time performance: Its data processing process requires the complete transmission of original geological data (such as 3D laser scanning point clouds and borehole camera images) to the cloud without edge pre-processing, resulting in high data transmission latency (especially during deep tunnel construction); at the same time, model optimization relies on a single neural network trained offline. Faced with real-time feedback scenarios such as sudden changes in surrounding rock grade and excessive over-excavation after blasting, it is difficult to complete parameter adjustments within construction breaks (usually 2-4 hours), resulting in the blasting plan lagging behind actual engineering needs.

[0005] In addition, traditional technologies generally have the problem of difficulty in balancing model calculation complexity and accuracy: although complex simulation models have high accuracy, their calculation efficiency is low; although simple empirical formulas are fast, they are difficult to adapt to changing geological conditions.

[0006] Therefore, there is a need for an intelligent computing platform for blasting parameters that can effectively reduce computing resource costs, improve computing real-time performance, and realize intelligent dynamic optimization. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides an intelligent calculation platform for blasting parameters to solve the problems of high computing resource cost and difficulty in ensuring real-time performance in the existing technology, realize efficient and real-time blasting parameter calculation, and meet actual engineering needs.

[0008] Technical Solution: To address the above technical issues, according to one aspect of the present invention, more specifically, a blasting parameter intelligent computing platform includes a distributed computing resource scheduling module. This module utilizes a distributed computing architecture to decompose computing tasks into multiple subtasks, which are then assigned to different computing nodes for parallel computing. Through an intelligent scheduling algorithm, computing tasks are dynamically allocated based on factors such as the load and computing power of each computing node, fully utilizing distributed computing resources in the network, improving computing resource utilization, and reducing reliance on a single high-performance computing device, thereby lowering computing costs. Furthermore, for large-scale numerical simulation tasks, distributed computing can significantly shorten computation time and improve computational efficiency.

[0009] Edge computing preprocessing module: A preprocessing unit is installed on edge devices close to the data collection end to perform preliminary processing and analysis of collected raw data, such as tunnel geological data and construction progress data. The edge computing preprocessing module processes collected data in real time, quickly extracting key information, reducing data transmission volume, and alleviating cloud computing pressure. For example, feature extraction of geological data transmits only key geological feature data relevant to blasting parameter calculations to the cloud, significantly reducing data transmission time and cloud computing workload, thus supporting real-time calculation of blasting parameters.

[0010] Lightweight Computational Model Library: This library of lightweight blasting parameter calculation models has been constructed, including a variety of simplified models based on machine learning and deep learning. These models are trained and optimized using extensive historical blasting data and simulation results, reducing computational complexity and the number of parameters while ensuring accuracy. In actual calculations, appropriate lightweight models are selected for blasting parameter calculations based on specific engineering scenarios and data characteristics, enabling rapid results and meeting real-time requirements. Furthermore, lightweight models require fewer computing resources, further reducing computational costs.

[0011] Real-time Feedback and Dynamic Optimization Module: This module monitors the actual conditions during tunnel construction, such as blasting results and surrounding rock changes, in real time, and compares and analyzes these conditions with the calculated blasting parameters. Based on these feedback results, a dynamic optimization algorithm is used to adjust and optimize the calculation model and parameters in real time. For example, if a deviation between the actual blasting results and the predicted results is detected, the module can quickly adjust the calculation model parameters, recalculate the blasting parameters, and promptly provide the optimized parameters to construction personnel, enabling dynamic adjustment of the blasting parameters and ensuring smooth construction.

[0012] Data Management and Sharing Module: This module manages and stores all types of data on the platform, including historical blasting data, geological data, and calculation results. Using data mining and analysis techniques, it deeply processes the data, extracting valuable information to support the training and optimization of computational models. Furthermore, the module supports data sharing across departments and personnel, facilitating collaborative work for construction personnel, technicians, and managers by enabling them to access the data they need.

[0013] The beneficial effects of the intelligent calculation platform for blasting parameters of the present invention are:

[0014] (1) The present invention makes full use of dispersed computing resources through a distributed computing resource scheduling module and a lightweight computing model library, reduces dependence on high-performance computing equipment, and reduces computing costs; the edge computing preprocessing module reduces data transmission volume and cloud computing volume, further reducing costs.

[0015] (2) The present invention processes data in real time through the edge computing preprocessing module, reducing data transmission time; the lightweight computing model library can quickly complete the blasting parameter calculation; the real-time feedback and dynamic optimization module realizes the real-time adjustment and optimization of the blasting parameters, thereby ensuring that accurate blasting parameters can be provided in time according to actual conditions during the tunnel construction process, thereby improving the construction progress. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0017] Figure 1 It is a structural schematic diagram of the present invention;

[0018] Figure 2 It is a schematic diagram of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0020] In order to make the technical solution of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] Reference Figure 1 、 Figure 2 , a blasting parameter intelligent calculation platform, implemented as follows:

[0022] 1. System deployment and data collection

[0023] Distributed data acquisition equipment is deployed at the tunnel construction site, including geological exploration sensors (such as acoustic wave detectors and borehole cameras), construction progress monitoring equipment (such as laser rangefinders and blast volume scanners), and environmental sensors (such as vibration sensors and surrounding rock displacement meters). The collected raw data includes:

[0024] Geological data: rock layer thickness, rock compressive strength, joint distribution density, groundwater content, etc.;

[0025] Construction data: current tunnel excavation progress, adopted blasting parameters (such as blasthole depth, charge, and detonation sequence), and post-blasting surrounding rock stability indicators (such as over-excavation and under-excavation, and surrounding rock deformation);

[0026] Real-time feedback data: blasting vibration waveform, blast pile size distribution, surrounding rock crack expansion, etc.

[0027] 2. Implementation of Edge Computing Preprocessing Module

[0028] Data cleaning and noise reduction: Filter the original geological data and construction data to remove outliers (such as erroneous data generated by transient sensor failures), and use a sliding window algorithm to smooth time series data (such as surrounding rock displacement).

[0029] Feature extraction and dimensionality reduction: For geological data, extract features that are strongly correlated with blasting parameters (such as rock blastability index and rock mass integrity coefficient). Use principal component analysis (PCA) to map high-dimensional geological data into key feature vectors of 10-15 dimensions.

[0030] For the construction progress data, features such as excavation speed and surrounding rock grade change rate are extracted to form structured input data (such as JSON format).

[0031] Data compression and transmission: Only pre-processed key feature data is transmitted to the cloud (the data volume is compressed to 1 / 10 to 1 / 5 of the original data), and transmitted to the distributed computing resource scheduling module via 5G or industrial Ethernet to reduce network latency and cloud load.

[0032] 3. Operation process of distributed computing resource scheduling module

[0033] Task decomposition strategy: According to the calculation task type (such as blasting plan preliminary calculation, parameter optimization inversion), the task is decomposed into subtasks:

[0034] If the calculation is for regional blasting parameters (such as blasthole arrangement in different surrounding rock sections), subtasks are assigned according to tunnel mileage (every 5 to 10 meters as a section);

[0035] If it is a multi-scheme comparison simulation, independent subtasks are generated according to different explosive types, detonation time differences and other parameter combinations.

[0036] Intelligent scheduling algorithm implementation: real-time collection of computing node status (CPU utilization, memory usage, network bandwidth) to establish a node load matrix;

[0037] Adopt dynamic load balancing algorithms (such as the minimum completion time priority algorithm) to prioritize allocating computationally intensive tasks to high-computing power nodes and data transmission tasks to nodes close to edge devices;

[0038] Supports task priority configuration, such as real-time optimization tasks taking priority over historical data training tasks, ensuring that urgent computing needs are processed first.

[0039] 4. Calling and Calculating the Lightweight Computing Model Library

[0040] Model selection mechanism: When the input data is a geological feature vector, the surrounding rock blasting response prediction model based on convolutional neural network (CNN) is automatically called to output parameters such as initial blasthole spacing and charge concentration;

[0041] When the input data includes historical blasting effect feedback, the model switches to a parameter inversion model combined with a genetic algorithm (GA) and iteratively solves for the optimal parameters by optimizing the objective function (e.g., minimizing overexcavation and maximizing blasting energy utilization);

[0042] For scenarios with rapid changes in construction progress (such as sudden changes in surrounding rock grade), the random forest classification model is enabled to quickly screen the optimal blasting parameters of similar historical cases as a reference.

[0043] Model lightweighting: Using model pruning techniques (such as channel pruning) to reduce the number of CNN model parameters by more than 60%, and increase the inference speed to 200 times / second;

[0044] The genetic algorithm is coded and optimized to control the single iteration time within 100ms to meet real-time computing requirements.

[0045] 5. Real-time Feedback and Dynamic Optimization Module Workflow

[0046] Real-time monitoring and deviation analysis:

[0047] Collect actual data within 10 minutes after blasting (e.g., overexcavation exceeding ±5% of the design value, surrounding rock vibration velocity exceeding the limit), compare it with the predicted results of the calculated parameters, and generate a deviation matrix (e.g., charge deviation rate, blasthole angle error);

[0048] A threshold trigger mechanism is used, and when the deviation exceeds the preset project allowable range (such as over-excavation > 10cm), the dynamic optimization process is automatically triggered.

[0049] Online model parameter adjustment: Based on the reinforcement learning algorithm, the bias matrix is ​​used as the reward function input to fine-tune the weight parameters of the lightweight model in real time (such as adjusting the parameters of the fully connected layer of the CNN model);

[0050] If the optimization fails to achieve the expected results after three consecutive times, it will automatically switch to the backup model (such as switching from CNN to support vector machine SVM model) to avoid falling into the local optimal solution.

[0051] VI. Implementation Methods of Data Management and Sharing Module

[0052] Data storage architecture: A distributed database (HBase) is used to store structured data (such as blasting parameter tables and surrounding rock feature libraries), and object storage is used to store unstructured data (geological images and vibration waveform files).

[0053] Supports data version control, generates a unique version number for each blasting parameter calculation result, and associates the corresponding geological data, calculation model and optimization records.

[0054] Data sharing and collaboration: Provides a web-based visualization interface for the construction side, displaying the current optimal blasting parameters and surrounding rock stability prediction results in real time;

[0055] Data access rights are opened to third-party systems through the API interface (RESTful API), allowing technicians to query historical blasting effect data through the mobile app or upload new engineering cases for model training.

[0056] 7. Inter-module interaction and closed-loop processing

[0057] Each module interacts through a standardized data interface (Protobuf format):

[0058] Edge computing preprocessing module → distributed computing scheduling module: transmits preprocessed feature vectors (including engineering scenario labels, such as "hard rock section" and "fault zone");

[0059] Distributed computing scheduling module → lightweight model library: calls the corresponding model according to the task type, inputs feature data and obtains calculation results;

[0060] Lightweight model library → Real-time feedback module: outputs preliminary blasting parameters (such as the number of blastholes and the total amount of charge) and receives optimized model parameters at the same time;

[0061] Real-time feedback module → data management module: stores actual blasting effect data and optimization process records, forming a closed-loop process of "data collection-calculation-optimization-storage".

[0062] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A blasting parameter intelligent computing platform, including a distributed computing resource scheduling module, an edge computing preprocessing module, a lightweight computing model library, a real-time feedback and dynamic optimization module, and a data management and sharing module, characterized by: The distributed computing resource scheduling module is used to adopt a distributed computing architecture to decompose computing tasks into multiple subtasks and assign them to different computing nodes for parallel computing. It uses an intelligent scheduling algorithm to dynamically allocate tasks according to the load and computing power of each computing node to improve computing resource utilization, reduce computing costs and shorten computing time. The edge computing preprocessing module is set up in the edge device close to the data collection end, and is used to perform preliminary processing and analysis on the collected raw data, extract key information, reduce the amount of data transmission, and reduce the computing pressure on the cloud; The lightweight computing model library stores a variety of simplified models based on machine learning and deep learning. The models are trained and optimized through historical blasting data and simulation results to reduce computational complexity and the number of parameters while ensuring computational accuracy. It supports the selection of appropriate models based on engineering scenarios and data characteristics to quickly calculate blasting parameters. The real-time feedback and dynamic optimization module is used to monitor the actual data during the construction process in real time. The actual data includes but is not limited to the data of blasting effects and surrounding rock changes, compares and analyzes them with the calculated blasting parameters, and uses the dynamic optimization algorithm to adjust and optimize the calculation model and parameters in real time; The data management and sharing module is used to uniformly manage and store historical blasting data, geological data, and calculation result data, extract valuable information through data mining and analysis, and support data sharing among different departments and personnel.

2. The blasting parameter intelligent calculation platform according to claim 1, characterized in that: The intelligent scheduling algorithm of the distributed computing resource scheduling module further dynamically adjusts the task allocation strategy according to the real-time load of the computing nodes, historical computing efficiency, and network transmission delay factors to achieve balanced distribution and parallel acceleration of computing tasks.

3. The blasting parameter intelligent calculation platform according to claim 1, characterized in that: The raw data of the edge computing preprocessing module includes tunnel geological data and construction progress data. The preliminary processing includes data cleaning and feature extraction, and only the key feature data related to the blasting parameter calculation is transmitted to the cloud.

4. The blasting parameter intelligent calculation platform according to claim 1, characterized in that: The simplified models in the lightweight computing model library include a surrounding rock characteristic analysis model based on a convolutional neural network, a blasting effect prediction model based on a random forest, and a parameter inversion model optimized in combination with a genetic algorithm.

5. The blasting parameter intelligent calculation platform according to claim 1, characterized in that: The real-time feedback and dynamic optimization module further updates the parameters of the lightweight computing model online through a reinforcement learning algorithm based on the deviation between the actual blasting effect and the predicted result, thereby realizing dynamic correction of the blasting parameters.

6. The blasting parameter intelligent calculation platform according to claim 1, characterized in that: The data management and sharing module supports structured storage, version control and authority management of data, and provides training data and input parameters to the distributed computing resource scheduling module and lightweight computing model library through the API interface.

7. The blasting parameter intelligent calculation platform according to claim 1, characterized in that: Each module interacts through a standardized data interface, forming a closed-loop processing flow from data acquisition, preprocessing, distributed computing, model optimization to result output, meeting the needs of real-time parameter calculation and dynamic adjustment in tunnel blasting projects.

Images