WO2022100609A1 - 用于工程机械的地质硬度识别方法及其系统 - Google Patents

用于工程机械的地质硬度识别方法及其系统 Download PDF

Info

Publication number
WO2022100609A1
WO2022100609A1 PCT/CN2021/129762 CN2021129762W WO2022100609A1 WO 2022100609 A1 WO2022100609 A1 WO 2022100609A1 CN 2021129762 W CN2021129762 W CN 2021129762W WO 2022100609 A1 WO2022100609 A1 WO 2022100609A1
Authority
WO
WIPO (PCT)
Prior art keywords
geological
equipment
hardness
real
sample
Prior art date
Application number
PCT/CN2021/129762
Other languages
English (en)
French (fr)
Inventor
黎伟福
邢柳
黄忠睿
何欢
张华�
Original Assignee
上海中联重科桩工机械有限公司
中联重科股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 上海中联重科桩工机械有限公司, 中联重科股份有限公司 filed Critical 上海中联重科桩工机械有限公司
Publication of WO2022100609A1 publication Critical patent/WO2022100609A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Definitions

  • the invention relates to the technical field of foundation construction, in particular to a geological hardness identification method and a system thereof for construction machinery.
  • Rotary drilling rig is a kind of foundation construction machinery, which is mainly used for pile foundation construction, especially in the construction of bridges, viaducts and house foundations.
  • the drill bit is under the ground for construction, and the operator cannot observe the geological conditions of the construction in real time.
  • an existing construction method is that the operator operates according to the operating experience and the parameter information fed back by the equipment during the construction process, so the operation of the rotary drilling rig requires higher experience and technical level of the operator; In addition to the operating experience, the operator also needs to be familiar with the drilling tools, formation geology and construction technology.
  • the corresponding drilling tools need to be selected according to the geological conditions. Excellent operators can select the corresponding drilling tools and construction methods to operate according to the geological conditions, which can not only improve the construction efficiency, but also save energy. Since the operation of the rotary drilling rig has high requirements on the operator, the labor cost is correspondingly high.
  • Another method for obtaining geological conditions is to use theoretical formulas to calculate, that is, to use real-time operating parameters fed back by equipment to calculate the current geological hardness through theoretical formulas.
  • the update and calculation of the above data is a continuous process, but the calculated theoretical value is a quantitative value at a certain moment; and the drilling process will change due to the operator's construction method and instantaneous data value changes. Affecting the calculation results, the calculation results vary greatly and can only be used as a reference, but cannot be used to continuously and accurately confirm the construction geology.
  • the purpose of the present invention is to provide a geological hardness identification method and a system thereof for construction machinery.
  • the method improves the accuracy of geological hardness identification through a machine learning algorithm, and through the cloud model training-edge model prediction layout, so that the model Continuous iteration in the cloud improves the robustness and accuracy of the model.
  • the operator can select the corresponding drilling tool and construction method according to the geological hardness label predicted by the model to operate to improve construction efficiency. .
  • one aspect of the present invention provides a geological hardness identification method for construction machinery, which is executed by edge equipment, including: acquiring real-time operating parameters of the equipment and a geological hardness identification model trained by cloud equipment; running the equipment in real time The parameters are input into the geological hardness identification model, and the drilling process label and the geological hardness label are obtained.
  • the real-time operating parameters of the equipment into the geological hardness identification model to obtain the drilling process label and the geological hardness label, including: cleaning the null values, invalid values and abnormal values in the real-time operating parameters of the equipment; judging the cleaned Whether the real-time operating parameters of the equipment are real-time data of the excavation process; if the real-time operating parameters of the cleaned equipment are real-time data of the excavation process, determine that the label of the drilling process is in progress; determine the label of geological hardness according to the real-time data of the excavation process.
  • the geological hardness identification method for construction machinery further includes: in the case that the real-time operating parameters of the cleaned equipment are not real-time data of the driving process, determining that the drilling process is marked as not driving.
  • the geological hardness identification method for construction machinery further includes: uploading real-time operating parameters of the device and its corresponding drilling process labels and geological hardness labels to the cloud device for iterative training of the geological hardness identification model.
  • the geological hardness identification method for construction machinery further includes: sending the drilling process label and the geological hardness label to the display screen for display to the operator.
  • the present invention also provides a geological hardness identification method for construction machinery, which is executed by a cloud device, including: acquiring an offline test report, where the offline test report includes the geological report and the equipment operating parameters corresponding to the geological report; The report determines the sample features and sample labels for training the geological hardness recognition model; uses the sample features and sample labels to train the geological hardness recognition model; sends the trained geological hardness recognition model to the edge device for calculating and outputting the geological recognition results.
  • determining sample features and sample labels for training the geological hardness identification model according to the offline test report including: performing data processing on equipment operating parameters to obtain running data of the excavation process; obtaining each geological category from the geological report as a sample label; The sample data corresponding to the geological category is selected from the running data of the excavation process, and the sample characteristics of the corresponding sample label are determined according to the sample data.
  • performing data processing on the operating parameters of the equipment to obtain the operating data of the excavation process including: cleaning null values, invalid values and abnormal values in the operating parameters of the equipment; splicing the operating parameters of the equipment after cleaning according to the requirements of the operating parameters of the equipment and interpolation; select the running data of the tunneling process under the tunneling conditions from the equipment operating parameters after splicing and interpolation.
  • the sample data corresponding to the geological category is selected from the running data of the excavation process, and the sample characteristics of the corresponding sample label are determined according to the sample data, including: selecting the sample data corresponding to the geological category from the running data of the excavation process; Perform feature calculation to obtain sample features.
  • the machine learning algorithm for training the geological hardness identification model includes: a gradient boosting classification tree algorithm.
  • the equipment operating parameters include at least one of the following: power head torque, main hoisting wire rope tension, main hoisting load, motor speed, hole depth, drill pipe lowering distance, and power head lowering distance.
  • the present invention also provides a geological hardness identification system for construction machinery, including: a construction machinery controller, used to obtain real-time operation parameters of equipment feedback from construction parts and equipment operation parameters in an offline test report; cloud-based equipment , used to train the geological hardness identification model according to the offline test report; edge equipment, used to input the real-time operating parameters of the equipment into the geological hardness identification model to obtain the drilling process label and geological hardness label; display screen, used to display the drilling process label and geological hardness labels.
  • a construction machinery controller used to obtain real-time operation parameters of equipment feedback from construction parts and equipment operation parameters in an offline test report
  • cloud-based equipment used to train the geological hardness identification model according to the offline test report
  • edge equipment used to input the real-time operating parameters of the equipment into the geological hardness identification model to obtain the drilling process label and geological hardness label
  • display screen used to display the drilling process label and geological hardness labels.
  • the edge equipment uses the geological hardness recognition model trained by the cloud equipment, and by inputting the real-time operating parameters of the equipment fed back by the construction components into the geological hardness recognition model, the drilling process label and the geological hardness label are obtained, and the operator can According to the predicted accurate geological hardness label, select the corresponding drilling tool and construction method for operation to improve the construction efficiency.
  • FIG. 1 is a flowchart of a method for identifying geological hardness for construction machinery performed by edge end equipment provided by an embodiment
  • Fig. 2 is a flow chart of a method for identifying geological hardness for Cheng machinery and performed by a cloud device provided by an embodiment
  • FIG. 3 is a structural block diagram of a geological hardness identification system for Cheng machinery provided by an embodiment.
  • geological hardness There are two main methods for identifying geological hardness: one is the experience of the operator on the construction site: when the drilling tool is drilled into formations with different hardness, the drilling speed, power head torque, power head rotation speed and equipment vibration, etc. The parameters will change. The operator preliminarily judges the geology of the formation based on the geological report and operating experience of the construction area. At the same time, the drill bit will be pressurized, and then the construction method and drilling tool type will be determined based on the feedback parameters after the pressurization. If it cannot be judged accurately, the operator also needs to use the drilling tool to grind continuously, and fish out the ground and broken geology to judge the actual geological situation. The disadvantage of this method is that the operator's judgment depends on the operation experience. If there is a problem with the selected drill bit through subjective judgment, it will not only reduce the construction efficiency because the drill bit needs to be replaced, but also reduce the service life of the drill bit.
  • Another method is to use the theoretical formula to calculate: using the torque of the power head, the area of the pile hole, the friction coefficient, the downward pressure of the drill bit, the drilling speed, the rotation speed of the power head and other data, the current geological hardness is calculated by the formula.
  • the instantaneous changes of parameters will affect the calculation results.
  • the calculation results vary greatly and can only be used as a reference, but cannot be used to continuously and accurately confirm the construction geology, so it has great limitations.
  • an embodiment of the present invention provides a geological hardness identification method for construction machinery, which is performed by edge end equipment, as shown in FIG. 1 , including steps S102-S104:
  • the real-time operation parameters of the equipment are input into the geological hardness identification model, and the drilling process label and the geological hardness label are obtained.
  • the real-time operation parameters of the equipment are the real-time operation parameters of the equipment fed back by the construction components obtained by the construction machinery controller, including drill pipe information, drilling tool information, power head data, main winch data, pressurized oil cylinder data, mast data, etc.
  • the geological hardness identification model is obtained through cloud-based equipment training, and the edge device inputs the collected real-time operating parameters of the construction site equipment into the trained geological hardness identification model, which outputs the drilling process label and the geological hardness label.
  • the labels of the drilling process include the "Driving" label and the "Not driving label". The operator can judge whether the drilling tool is in the driving state according to these two labels, so as to make correct operation judgments.
  • the geological hardness label reflects the geological conditions of the drilling tool in contact with the construction geology.
  • the label can be geological categories such as “clay”, “marble”, and “granite”, or it can be “extremely soft”, “soft”, “softer” , “hard”, “hard” and other geological hardness categories, the operator can accurately obtain the geological conditions according to the geological hardness label to make correct operation judgments.
  • the above method improves the accuracy of geological hardness identification through machine learning algorithms; Model prediction cannot guarantee the real-time performance of calculation, so this method adopts the layout of cloud model training and edge model prediction, and the edge device performs the model prediction process, so that the operator can quickly and accurately obtain the driving state and geological category. Make correct operating judgments.
  • the label can be sent to the display screen for display to the operator, and the display effect can be "drilling process: tunneling; geological hardness: granite", the operator Based on this information, you can select the appropriate drilling tools and construction techniques by hand.
  • the present invention provides another embodiment to illustrate the process of processing the real-time operating parameters of the equipment by the geological hardness identification model, including steps S1042-S1046:
  • the real-time operating parameters of the cleaned equipment are the real-time data of the excavation process, determine that the label of the drilling process is in progress, and determine the label of geological hardness according to the real-time data of the excavation process; the real-time operating parameters of the cleaned equipment are not the excavation process In the case of real-time data of the process, it is determined that the drilling process is marked as not driving.
  • the real-time data of the excavation process refers to the parameters of the real-time operation parameters of the equipment when the equipment is in the state of excavation.
  • the training of the model also includes the training of the judgment of the drilling process; only when the real-time operation parameters of the equipment are real-time data of the excavation process, further The geological hardness label is predicted according to the real-time data of the excavation process, otherwise only the label of the drilling process that is not in the excavation is output.
  • the edge device can also upload the data involved in the model prediction to the cloud device for training the model for iterative training to improve the robustness and accuracy of the model , which is to upload the real-time operating parameters of the device and its corresponding output drilling process labels and geological hardness labels to the cloud device.
  • the present invention also provides a geological hardness identification method for construction machinery executed by a cloud device, specifically including a training method for a geological hardness identification model, as shown in FIG. 2 , including steps S202-S208:
  • an offline test report is obtained, where the offline test report includes a geological report and equipment operation parameters corresponding to the geological report.
  • the geological report is obtained by the exploration company drilling the ground of the construction site.
  • the information included in the geological report is: the geological category, weathering degree, and geological uniaxial anti-saturation pressure of the construction site corresponding to the different depths of the underground test hole and other hardness parameters distribution.
  • the equipment operating parameters are the real-time parameters fed back by the construction components during the construction process, which are the same as the above-mentioned real-time operating parameters of the equipment, including multi-dimensional data, such as: power head torque, main hoisting wire rope tension, main hoisting load, motor speed, hole depth , Drill pipe lowering distance, power head lowering distance and so on.
  • the corresponding relationship between the geological category and the operating parameters of the equipment may be determined according to the drilling depth and/or the drilling time. For example, when the drilling depth is one meter, the geological category corresponding to the drilling depth in the geological report is "marble", and other dimensional parameters such as the pressurization pressure and drill bit speed corresponding to the drilling depth in the equipment operating parameters are corresponding. value.
  • the samples used to train the model are labeled samples, the sample features are the eigenvalues of the samples, and the sample labels are the label values of the samples.
  • S204 may preferably include steps S2042-S2046:
  • the data processing process preferably includes steps (1)-(3): (1) Null values, invalid values and abnormal values in the equipment operating parameters (2) Splicing and interpolating the operating parameters of the cleaned equipment according to the requirements of the operating parameters of the equipment; (3) Selecting the operating data of the driving process under the driving conditions from the operating parameters of the equipment after the splicing and interpolation, and the driving process running
  • the data are the parameters of the equipment operating parameters when the equipment is in the excavation state.
  • each geological category is obtained from the geological report as a sample label.
  • S2046 Screen out the sample data corresponding to the geological category from the running data of the excavation process, and determine the sample feature corresponding to the sample label according to the sample data.
  • the data in the running data of the excavation process is multi-dimensional and can contain hundreds of data. Therefore, in order to reduce the amount of calculation, the data with high correlation with reflecting the geological category can be selected as sample data;
  • the sample data can be directly used as sample features for model training, and can also be used for feature calculation on sample data as sample features: for example, the torque mean and torque variance can be obtained by calculating the torque value and the variance as sample features, which can avoid the instantaneous The effect of unstable value.
  • the data used for training the model in the above method is obtained from the offline test report, and the model training process is performed by the cloud device, which reduces the computing pressure of the edge device.
  • the machine learning algorithms used in model training include but are not limited to decision trees, logistic regression, Xgboost, support vector machines, CNN (Convolutional Neural Networks, Convolutional Neural Networks), RNN (Recurrent Neural Networks, Recurrent Neural Networks). After testing, Gradient boosting classification tree algorithm can achieve higher computational accuracy.
  • the present invention also provides a geological hardness identification system for construction machinery, as shown in FIG. 3 , comprising: a construction machinery controller, used for acquiring real-time operating parameters of equipment fed back by construction components and operating parameters of equipment in off-line test reports;
  • the cloud device is used to train the geological hardness identification model according to the offline test report;
  • the edge device is used to input the real-time operating parameters of the device into the geological hardness identification model to obtain the drilling process label and the geological hardness label;
  • the display screen is used to display the Drilling process label and geological hardness label.
  • the edge device is also used to upload the real-time operating parameters of the device and its corresponding drilling process labels and geological hardness labels to the cloud device for iterative training of the geological hardness identification model.
  • the edge device and the construction machinery controller realize stable communication through the CAN (Controller Area Network) bus, which can not only obtain the real-time operating parameters of the equipment from the construction machinery controller, but also return the model prediction results to the construction machinery control.
  • the device is delivered to the display screen.
  • the above system establishes a cloud model training-edge model prediction layout, which improves the real-time performance of model prediction, and at the same time enables the model to iterate continuously in the cloud, improving the robustness and accuracy of the model.
  • the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
  • computer-usable storage media including, but not limited to, disk storage, CD-ROM, optical storage, etc.
  • These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions
  • the apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.
  • a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • Memory may include non-persistent memory in computer readable media, random access memory (RAM) and/or non-volatile memory in the form of, for example, read only memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
  • RAM random access memory
  • ROM read only memory
  • flash RAM flash memory
  • Computer-readable media includes both persistent and non-permanent, removable and non-removable media, and storage of information may be implemented by any method or technology.
  • Information may be computer readable instructions, data structures, modules of programs, or other data.
  • Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Flash Memory or other memory technology, Compact Disc Read Only Memory (CD-ROM), Digital Versatile Disc (DVD) or other optical storage, Magnetic tape cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.
  • computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Investigation Of Foundation Soil And Reinforcement Of Foundation Soil By Compacting Or Drainage (AREA)

Abstract

本发明提供一种用于工程机械的地质硬度识别方法及其系统。该方法包括：由边缘端设备执行：获取设备实时运行参数以及经过云端设备训练的地质硬度识别模型；将设备实时运行参数输入到地质硬度识别模型，获得下钻过程标签和地质硬度标签。还包括：由云端设备执行：获取离线测试报告，该离线测试报告包括地质报告以及对应地质报告的设备运行参数；根据离线测试报告确定用于训练地质硬度识别模型的样本特征和样本标签；使用样本特征和样本标签训练地质硬度识别模型；将经过训练的地质硬度识别模型下发给边缘端设备用于计算输出地质识别结果。通过云端模型训练-边缘端模型的布局，使得模型不断在云端迭代，提高模型的鲁棒性和准确率。

Description

用于工程机械的地质硬度识别方法及其系统
相关申请的交叉引用
本申请要求2020年11月11日提交的中国专利申请202011255138.8的权益，该申请的内容通过引用被合并于本文。
技术领域
本发明涉及基础施工技术领域，具体地涉及一种用于工程机械的地质硬度识别方法及其系统。
背景技术
旋挖钻机是基础施工机械的一种，主要用来进行桩基施工，特别是在桥梁、高架和房屋地基的施工中应用尤其广泛。旋挖钻机在旋挖过程中钻头在地面以下进行施工，操作机手无法实时观察到施工的地质情况。鉴于此，现有的一种施工方法是，在施工过程中操作机手根据操作经验和设备反馈的参数信息进行操作，故操作旋挖钻机对于操作机手的经验和技术水平要求较高；除了操作经验外，操作机手还需要对钻具、地层地质和施工工艺都比较熟悉，在钻进到对应地质的时候需要根据地质情况选用对应的钻具。优秀的操作机手能够根据地质情况选取对应的钻具和施工工法进行操作，不仅能够提高施工效率，还能节约能耗。鉴于旋挖钻机的操作对操作机手的要求较高，因此，人力成本也相应较高。
另一种获取地质情况的方法为利用理论公式计算，即利用设备反馈的实时运行参数，通过理论公式计算当前地质硬度。然而在实际施工过程中，上述数据的更新和计算是一个持续不断的过程，但计算出的理论值为某个时刻的定量值；且下钻过程因为操作机手施工工法、瞬时数据值变化都会影响到计算结果，计算结果变化幅度大，仅可以作为参考，而并不能用于连续准确地确认施工地质。
为了降低对操作机手技能经验的要求，同时获取更加准确的地质硬度识别结果，亟需一种能够自动识别地质硬度的方法，以提高施工效率，实现施工智能化和便利化。
发明内容
本发明的目的是提供一种用于工程机械的地质硬度识别方法及其系统，该方法通过机器学习算法提高地质硬度识别的准确性，并通过云端模型训练-边缘端模型预测的布局，使得模型不断在云端迭代，提高模型的鲁棒性和准确率，工程机械在进行旋挖作业时，操作机手能够根据模型预测的地质硬度标签选取对应的钻具和施工工法进行操作，以提高施工效率。
为了实现上述目的，本发明一方面提供一种用于工程机械的地质硬度识别方法，由边缘端设备执行，包括：获取设备实时运行参数以及经过云端设备训练的地质硬度识别模型；将设备实时运行参数输入到地质硬度识别模型，获得下钻过程标签和地质硬度标签。
可选地，将设备实时运行参数输入到地质硬度识别模型，获得下钻过程标签和地质硬度标签，包括：对设备实时运行参数中的空值、无效值和异常值进行清洗；判断清洗后的设备实时运行参数是否为掘进过程实时数据；在清洗后的设备实时运行参数为掘进过程实时数据的情况下，确定下钻过程标签为正在掘进；根据掘进过程实时数据确定地质硬度标签。
可选地，用于工程机械的地质硬度识别方法还包括：在清洗后的设备实时运行参数不为掘进过程实时数据的情况下，确定下钻过程标签为不在掘进。
可选地，用于工程机械的地质硬度识别方法还包括：将设备实时运行参数及其对应的下钻过程标签和地质硬度标签上传至云端设备，用于迭代训练地质硬度识别模型。
可选地，用于工程机械的地质硬度识别方法还包括：将下钻过程标签和地质硬度标签发送给显示屏用于向操作机手展示。
另一方面，本发明还提供一种用于工程机械的地质硬度识别方法，由云端设备执行，包括：获取离线测试报告，离线测试报告包括地质报告以及对应地质报告的设备运行参数；根据离线测试报告确定用于训练地质硬度识别模型的样本特征和样本标签；使用样本特征和样本标签训练地质硬度识别模型；将经过训练的地质硬度识别模型下发给边缘端设备用于计算输出地质识别结果。
可选地，根据离线测试报告确定用于训练地质硬度识别模型的样本特征和样本标签，包括：对设备运行参数进行数据处理获得掘进过程运行数据；从地质报告中获取各地质类别作为样本标签；从掘进过程运行数据中筛选出对应地质类 别的样本数据，根据样本数据确定对应样本标签的样本特征。
可选地，对设备运行参数进行数据处理获得掘进过程运行数据，包括：对设备运行参数中的空值、无效值和异常值进行清洗；根据设备运行参数需求对清洗后的设备运行参数进行拼接和插值；从拼接和插值后的设备运行参数中选取掘进工况下的掘进过程运行数据。
可选地，从掘进过程运行数据中筛选出对应地质类别的样本数据，根据样本数据确定对应样本标签的样本特征，包括：从掘进过程运行数据中筛选出对应地质类别的样本数据；对样本数据进行特征计算获得样本特征。
可选地，训练地质硬度识别模型的机器学习算法包括：梯度提升分类树算法。
可选地，设备运行参数包括以下中的至少一者：动力头扭矩、主卷扬钢丝绳张力、主卷扬载荷、马达转速、孔深、钻杆下放距离、动力头下放距离。
另一方面，本发明还提供一种用于工程机械的地质硬度识别系统，包括：工程机械控制器，用于获取施工部件反馈的设备实时运行参数及离线测试报告中的设备运行参数；云端设备，用于根据离线测试报告训练地质硬度识别模型；边缘端设备，用于将设备实时运行参数输入到地质硬度识别模型获得下钻过程标签和地质硬度标签；显示屏，用于显示下钻过程标签和地质硬度标签。
通过上述技术方案，边缘端设备利用云端设备训练好的地质硬度识别模型，通过将施工部件反馈的设备实时运行参数输入到地质硬度识别模型，获取下钻过程标签和地质硬度标签，操作机手能够根据预测准确的地质硬度标签选取对应的钻具和施工工法进行操作，以提高施工效率。
本发明的其它特征和优点将在随后的具体实施方式部分予以详细说明。
附图说明
附图是用来提供对本发明的进一步理解，并且构成说明书的一部分，与下面的具体实施方式一起用于解释本发明，但并不构成对本发明的限制。在附图中：
图1是一实施例提供的由边缘端设备执行的用于工程机械的地质硬度识别方法流程图；
图2是一实施例提供的由云端设备执行的用于程机械的地质硬度识别方法 流程图；
图3是一实施例提供的用于程机械的地质硬度识别系统结构框图。
具体实施方式
以下结合附图对本发明的具体实施方式进行详细说明。应当理解的是，此处所描述的具体实施方式仅用于说明和解释本发明，并不用于限制本发明。
现有的地质硬度识别方法主要有两种：一种是施工现场操作机手的经验：当钻具下钻到硬度不同的地层时，下钻速度、动力头扭矩、动力头转速和设备振动等参数会出现变化，操作机手根据施工区域的地质报告与操作经验初步判断地层的地质，同时会对钻头进行加压，然后结合加压后的反馈参数决定施工工法和钻具类型。如果不能准确判断，操作机手还需要利用钻具不断打磨，将打磨破碎的地质捞出判断实际的地质情况。该方法的缺点在于：操作机手的判断依靠操作经验，如果通过主观判断选择的钻头有问题，不仅会因为需要更换钻头而降低施工效率，甚至会减少钻头的使用寿命。
另外一种方法是利用理论公式计算：利用动力头扭矩、桩孔面积、摩擦系数、钻头下压力、下钻速度、动力头转速度等数据，通过公式计算当前地质硬度。然而在实际施工过程中，参数的瞬时变化都会影响到计算结果，计算结果变化幅度大，仅可以作为参考，而并不能用于连续准确地确认施工地质，因此具有很大的局限性。
上述两种方法都存在地质硬度识别结果准确性不足的缺陷，从而影响施工效率。有鉴于此，本发明一实施例提供一种用于工程机械的地质硬度识别方法，由边缘端设备执行，如图1所示，包括步骤S102-S104：
S102，获取设备实时运行参数以及经过云端设备训练的地质硬度识别模型。
S104，将设备实时运行参数输入到地质硬度识别模型，获得下钻过程标签和地质硬度标签。
设备实时运行参数是工程机械控制器获取的施工部件反馈的设备实时运行参数，包括钻杆信息、钻具信息、动力头数据、主卷扬数据、加压油缸数据、桅杆数据等。地质硬度识别模型是在云端设备训练获得，边缘端设备将采集的施工现场的设备实时运行参数输入到经过训练的地质硬度识别模型，该模型输出下钻 过程标签和地质硬度标签。下钻过程标签包括“正在掘进”标签和“不在掘进标签”，操作机手可根据这两个标签判断钻具是否处于掘进状态，从而做出正确的操作判断。地质硬度标签反应钻具接触施工地质的地质情况，该标签可以是“黏土”、“大理石”、“花岗岩”此类的地质类别，也可以是“极软”、“软”、“较软”、“较硬”、“坚硬”此类的地质硬度类别，操作机手可根据地质硬度标签准确获取地质情况从而做出正确的操作判断。
一方面考虑到传统理论公式的计算结果准确性有限，故上述方法通过机器学习算法提高了地质硬度识别的准确性；另一方面考虑到工程机械的施工环境可能信号条件欠佳，在云端设备进行模型预测不能保证计算的实时性，故本方法采用云端模型训练-边缘端模型预测的布局，由边缘端设备执行模型预测过程，从而使得操作机手能快速、准确地获取掘进状态和地质类别，做出正确的操作判断。
优选地，获得下钻过程标签和地质硬度标签后，便可将标签发送给显示屏用于向操作机手展示，展示效果可以为“下钻过程：正在掘进；地质硬度：花岗岩”，操作机手便可根据该信息选择合适的钻具和施工工艺。
对于上述S104，优选地，本发明提供另一实施例以说明地质硬度识别模型对设备实时运行参数的处理过程，包括步骤S1042-S1046：
S1042，对设备实时运行参数中的空值、无效值和异常值进行清洗。
S1044，判断清洗后的设备实时运行参数是否为掘进过程实时数据。
S1046，在清洗后的设备实时运行参数为掘进过程实时数据的情况下，确定下钻过程标签为正在掘进，以及根据掘进过程实时数据确定地质硬度标签；在清洗后的设备实时运行参数不为掘进过程实时数据的情况下，确定下钻过程标签为不在掘进。
掘进过程实时数据是指设备实时运行参数中当设备处于正在掘进状态时的参数，对模型的训练也包括对下钻过程判断的训练；只有当设备实时运行参数为掘进过程实时数据时，才进一步根据掘进过程实时数据来预测地质硬度标签，否则只输出不在掘进的下钻过程标签。
优选地，当边缘端设备进行模型预测获得标签后，边缘端设备还可将模型预测所涉及的数据上传至用于训练模型的云端设备，用于迭代训练以提高模型的鲁棒性和准确率，具体为将设备实时运行参数及其对应输出的下钻过程标签和地 质硬度标签上传至云端设备。
本发明还提供一种由云端设备执行的用于工程机械的地质硬度识别方法，具体包括地质硬度识别模型的训练方法，如图2所示，包括步骤S202-S208：
S202，获取离线测试报告，离线测试报告包括地质报告以及对应地质报告的设备运行参数。
地质报告由勘探公司钻取施工工地地面获得，地质报告包括的信息为：施工工地对应试钻孔地下不同深度的地质类别、风化程度和地质单轴抗饱和压力等硬度参数分布情况。设备运行参数是施工过程中施工部件反馈的实时参数，和上述设备实时运行参数相同，包括多维度数据，例如包括：动力头扭矩、主卷扬钢丝绳张力、主卷扬载荷、马达转速、孔深、钻杆下放距离、动力头下放距离等等。而地质类别与设备运行参数的对应关系可根据钻孔深度和/或下钻时间确定。举例来说，当钻孔深度为一米时，地质报告中对应该钻孔深度的地质类别为“大理石”，设备运行参数中对应该钻孔深度的加压力、钻头转速等其他维度参数为相应值。
S204，根据离线测试报告确定用于训练地质硬度识别模型的样本特征和样本标签。
训练模型所使用的样本为有标签样本，样本特征为样本的特征值，样本标签为样本的标签值。
S204可优选包括步骤S2042-S2046：
S2042，对设备运行参数进行数据处理获得掘进过程运行数据。
通过数据处理流程获得优质离线数据以提高所训练的模型的预测准确率，数据处理过程优选包括步骤(1)-(3)：(1)对设备运行参数中的空值、无效值和异常值进行清洗；(2)根据设备运行参数需求对清洗后的设备运行参数进行拼接和插值；(3)从拼接和插值后的设备运行参数中选取掘进工况下的掘进过程运行数据，掘进过程运行数据为设备运行参数中当设备处于掘进状态时的参数。
S2044，从地质报告中获取各地质类别作为样本标签。
例如，将“黏土”、“大理石”、“花岗岩”此类的地质类别作为样本标签。
S2046，从掘进过程运行数据中筛选出对应地质类别的样本数据，根据样本数据确定对应样本标签的样本特征。
容易理解，掘进过程运行数据中的数据是多维度的，可包含上百个数据，因此为减少计算量，可从其中筛选出与反映地质类别相关度较高的数据作为样本数据；筛选出的样本数据可直接作为样本特征用于模型训练，也可对样本数据进行特征计算作为样本特征：例如对扭矩值进行求均值和求方差的计算获得扭矩均值和扭矩方差作为样本特征，可避免由于瞬时值不稳定而带来的影响。
S206，使用样本特征和样本标签训练地质硬度识别模型。
S208，将经过训练的地质硬度识别模型下发给边缘端设备用于计算输出地质识别结果。
上述方法中用于训练模型的数据从离线测试报告获取，并且由云端设备进行模型训练过程，减小了边缘端设备的计算压力。
模型训练所采用的机器学习算法包括但不限于决策树、逻辑回归、Xgboost、支持向量机、CNN(Convolutional Neural Networks，卷积神经网络)、RNN(Recurrent Neural Network，循环神经网络)，经过测试，梯度提升分类树算法可获得较高的计算精度。
本发明还提供一种用于工程机械的地质硬度识别系统，如图3所示，包括：工程机械控制器，用于获取施工部件反馈的设备实时运行参数及离线测试报告中的设备运行参数；云端设备，用于根据离线测试报告训练地质硬度识别模型；边缘端设备，用于将设备实时运行参数输入到地质硬度识别模型获得下钻过程标签和地质硬度标签；以及显示屏，用于显示下钻过程标签和地质硬度标签。
边缘端设备还用于：将设备实时运行参数及其对应的下钻过程标签、地质硬度标签上传至云端设备，用于迭代训练地质硬度识别模型。
边缘端设备与工程机械控制器通过CAN(Controller Area Network，控制器局域网络)总线实现稳定通信，既能从工程机械控制器获取设备实时运行参数，同时也能将模型预测结果返回至工程机械控制器投放至显示屏显示。
上述系统建立起一个云端模型训练-边缘端模型预测的布局，提高了模型预测的实时性，同时使得模型不断在云端迭代，提高了模型的鲁棒性和准确率。
本领域内的技术人员应明白，本申请的实施例可提供为方法、系统、或计 算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本申请是参照根据本申请实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
存储器可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。存储器是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器 (EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括要素的过程、方法、商品或者设备中还存在另外的相同要素。
以上仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。

Claims (12)

  1. 一种用于工程机械的地质硬度识别方法，其特征在于，由边缘端设备执行，包括：
    获取设备实时运行参数以及经过云端设备训练的地质硬度识别模型；
    将所述设备实时运行参数输入到所述地质硬度识别模型，获得下钻过程标签和地质硬度标签。
  2. 根据权利要求1所述的用于工程机械的地质硬度识别方法，其特征在于，所述将所述设备实时运行参数输入到所述地质硬度识别模型，获得下钻过程标签和地质硬度标签，包括：
    对所述设备实时运行参数中的空值、无效值和异常值进行清洗；
    判断清洗后的设备实时运行参数是否为掘进过程实时数据；
    在所述清洗后的设备实时运行参数为所述掘进过程实时数据的情况下，确定所述下钻过程标签为正在掘进；
    根据所述掘进过程实时数据确定所述地质硬度标签。
  3. 根据权利要求2所述的用于工程机械的地质硬度识别方法，其特征在于，还包括：
    在所述清洗后的设备实时运行参数不为所述掘进过程实时数据的情况下，确定所述下钻过程标签为不在掘进。
  4. 根据权利要求2所述的用于工程机械的地质硬度识别方法，其特征在于，还包括：
    将所述设备实时运行参数及其对应的所述下钻过程标签和所述地质硬度标签上传至所述云端设备，用于迭代训练所述地质硬度识别模型。
  5. 根据权利要求1所述的用于工程机械的地质硬度识别方法，其特征在于，还包括：
    将所述下钻过程标签和地质硬度标签发送给显示屏用于向操作机手展示。
  6. 一种用于工程机械的地质硬度识别方法，其特征在于，由云端设备执行，包括：
    获取离线测试报告，所述离线测试报告包括地质报告以及对应所述地质报告的设备运行参数；
    根据所述离线测试报告确定用于训练地质硬度识别模型的样本特征和样本标签；
    使用所述样本特征和所述样本标签训练所述地质硬度识别模型；
    将经过训练的地质硬度识别模型下发给边缘端设备用于计算输出地质识别结果。
  7. 根据权利要求6所述的用于工程机械的地质硬度识别方法，其特征在于，所述根据所述离线测试报告确定用于训练地质硬度识别模型的样本特征和样本标签，包括：
    对所述设备运行参数进行数据处理获得掘进过程运行数据；
    从所述地质报告中获取各地质类别作为所述样本标签；
    从所述掘进过程运行数据中筛选出对应所述地质类别的样本数据，根据所述样本数据确定对应所述样本标签的所述样本特征。
  8. 根据权利要求7所述的用于工程机械的地质硬度识别方法，其特征在于，所述对所述设备运行参数进行数据处理获得掘进过程运行数据，包括：
    对所述设备运行参数中的空值、无效值和异常值进行清洗；
    根据设备运行参数需求对清洗后的设备运行参数进行拼接和插值；
    从拼接和插值后的设备运行参数中选取掘进工况下的所述掘进过程运行数据。
  9. 根据权利要求7所述的用于工程机械的地质硬度识别方法，其特征在于，所述从所述掘进过程运行数据中筛选出对应所述地质类别的样本数据，根据所述样本数据确定对应所述样本标签的所述样本特征，包括：
    从所述掘进过程运行数据中筛选出对应所述地质类别的样本数据，对所述 样本数据进行特征计算获得所述样本特征。
  10. 根据权利要求6所述的用于工程机械的地质硬度识别方法，其特征在于，训练所述地质硬度识别模型的机器学习算法包括：梯度提升分类树算法。
  11. 根据权利要求6所述的用于工程机械的地质硬度识别方法，其特征在于，所述设备运行参数包括以下中的至少一者：
    动力头扭矩、主卷扬钢丝绳张力、主卷扬载荷、马达转速、孔深、钻杆下放距离、动力头下放距离。
  12. 一种用于工程机械的地质硬度识别系统，其特征在于，包括：
    工程机械控制器，用于获取施工部件反馈的设备实时运行参数及离线测试报告中的设备运行参数；
    云端设备，用于根据所述离线测试报告训练地质硬度识别模型；
    边缘端设备，用于将所述设备实时运行参数输入到所述地质硬度识别模型获得下钻过程标签和地质硬度标签；
    显示屏，用于显示所述下钻过程标签和地质硬度标签。
PCT/CN2021/129762 2020-11-11 2021-11-10 用于工程机械的地质硬度识别方法及其系统 WO2022100609A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011255138.8 2020-11-11
CN202011255138.8A CN112446172A (zh) 2020-11-11 2020-11-11 用于工程机械的地质硬度识别方法以及系统

Publications (1)

Publication Number Publication Date
WO2022100609A1 true WO2022100609A1 (zh) 2022-05-19

Family

ID=74736823

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/129762 WO2022100609A1 (zh) 2020-11-11 2021-11-10 用于工程机械的地质硬度识别方法及其系统

Country Status (2)

Country Link
CN (1) CN112446172A (zh)
WO (1) WO2022100609A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115661133A (zh) * 2022-12-07 2023-01-31 山东省地质矿产勘查开发局第一地质大队（山东省第一地质矿产勘查院） 一种地质类型识别方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112446172A (zh) * 2020-11-11 2021-03-05 上海中联重科桩工机械有限公司 用于工程机械的地质硬度识别方法以及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101672184A (zh) * 2009-09-30 2010-03-17 大连理工大学 全断面岩石掘进机滚刀米字型结构布置方法
US20180240021A1 (en) * 2017-02-20 2018-08-23 Saudi Arabian Oil Company Well performance classification using artificial intelligence and pattern recognition
CN111879610A (zh) * 2020-07-10 2020-11-03 武汉大学 一种掘进过程中岩渣力学参数实时测量系统及测量方法
CN112446172A (zh) * 2020-11-11 2021-03-05 上海中联重科桩工机械有限公司 用于工程机械的地质硬度识别方法以及系统

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118071129A (zh) * 2016-12-12 2024-05-24 吉奥奎斯特系统公司 增强的地质服务表征
CN109766335A (zh) * 2019-01-16 2019-05-17 天津大学 基于分类回归决策树算法的盾构施工地质识别方法及系统
GB2583906B (en) * 2019-04-29 2022-01-19 Equinor Energy As Method of estimating a mineral content of a geological structure
CN111160571B (zh) * 2019-12-31 2023-04-07 盾构及掘进技术国家重点实验室 一种基于土压平衡盾构机参数数据驱动反演地质的方法
CN111291500B (zh) * 2020-03-06 2023-03-31 盾构及掘进技术国家重点实验室 一种基于土压平衡盾构机参数数据的反演建模方法
CN111594201B (zh) * 2020-05-28 2021-08-13 中铁工程装备集团有限公司 一种tbm关键参数智能控制系统及方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101672184A (zh) * 2009-09-30 2010-03-17 大连理工大学 全断面岩石掘进机滚刀米字型结构布置方法
US20180240021A1 (en) * 2017-02-20 2018-08-23 Saudi Arabian Oil Company Well performance classification using artificial intelligence and pattern recognition
CN111879610A (zh) * 2020-07-10 2020-11-03 武汉大学 一种掘进过程中岩渣力学参数实时测量系统及测量方法
CN112446172A (zh) * 2020-11-11 2021-03-05 上海中联重科桩工机械有限公司 用于工程机械的地质硬度识别方法以及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WANG, JIAN: "Research on Tbm Performance Prediction and Rock Mass Parameters Characterization Method Based on Machine Learning", CHINESE MASTER’S THESES FULL-TEXT DATABASE (ENGINEERING SCIENCE AND TECHNOLOGY II), 30 May 2017 (2017-05-30), pages 1 - 133, XP055930518, [retrieved on 20220613] *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115661133A (zh) * 2022-12-07 2023-01-31 山东省地质矿产勘查开发局第一地质大队（山东省第一地质矿产勘查院） 一种地质类型识别方法
CN115661133B (zh) * 2022-12-07 2023-05-23 山东省地质矿产勘查开发局第一地质大队（山东省第一地质矿产勘查院） 一种地质类型识别方法

Also Published As

Publication number Publication date
CN112446172A (zh) 2021-03-05

Similar Documents

Publication Publication Date Title
WO2022100609A1 (zh) 用于工程机械的地质硬度识别方法及其系统
RU2740883C2 (ru) Система, основанная на анализе изображений, для буровых работ
US8271246B2 (en) System and method for minimizing lost circulation
CN104806226B (zh) 智能钻井专家系统
EP3899193A1 (en) Image based inspection and automated inventory and supply chain management of well equipment
CA2817515C (en) Apparatus and method for automated drilling of a borehole in a subsurface formation
Singh* et al. Cloud-based ROP prediction and optimization in real-time using supervised machine learning
AU2014348913B2 (en) Wellbore pipe trip guidance and statistical information processing method
KR102294384B1 (ko) 머신러닝을 이용한 굴진율 예측 시추운전가이드 모델 구축방법 및 이를 이용한 굴진율 예측시스템
Garcia et al. Decision support system for an intelligent operator of utility tunnel boring machines
US20210017846A1 (en) Control scheme for surface steerable drilling system
CN111126735A (zh) 一种钻井数字孪生系统
Elmgerbi et al. Machine learning techniques application for real-time drilling hydraulic optimization
Said et al. Theoretical development of a digital-twin based automation system for oil well drilling rigs
CN109736771B (zh) 基于凿岩台车随钻参数的围岩分析方法及系统
US11514382B2 (en) Utilizing micro-services for optimization workflows of borehole operations
US11898441B2 (en) Method and system for optimizing rig energy efficiency using machine learning
US20220316328A1 (en) Real time dull bit grading modeling and process technique
CN113431557B (zh) 一种基于人工智能的井下井眼轨迹跟踪方法
CN114370284B (zh) 盾构开挖面地层性质实时感知与掘进参数调整方法及系统
Løken et al. Automated drilling algorithms implementation on a laboratory drilling rig
Wang et al. Data Integration Enabling Advanced Machine Learning ROP Predictions and its Applications
FR3083259A1 (fr) Sous-système de trépan pour mettre à jour automatiquement une trajectoire de forage
US11859485B1 (en) Trip map for adjusting a tripping operation in a wellbore
US11788400B2 (en) Method for real-time pad force estimation in rotary steerable system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21891129

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21891129

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 21891129

Country of ref document: EP

Kind code of ref document: A1