Strip mine electric shovel excavation operation control system and method

Abstract

The embodiment of the invention provides a strip mine electric shovel excavation operation control system and method, and the method comprises the steps: collecting dynamic mechanical parameters of a plurality of key parts of a strip mine electric shovel in real time through a distributed torque sensor network, arranging the dynamic mechanical parameters into a real-time torque distribution matrix, constructing a spatial distribution matrix of the plurality of key parts, and carrying out the real-time torque distribution matrix of the plurality of key parts; a neural network model based on deep learning is further adopted to conduct deep interaction analysis on the torque distribution characteristics and the spatial relation characteristics, and accurate sensing and dynamic evaluation of the overall load state of the strip mine electric shovel are achieved based on the analysis result. Through the mode, the local overload risk of the equipment can be quickly identified, the comprehensiveness and real-time performance of equipment load monitoring are improved, equipment damage caused by local overload is effectively prevented, meanwhile, the risk of manual operation errors is reduced through intelligent control, and the safety and efficiency of strip mine electric shovel operation are guaranteed.

Description

Technical Field

[0001] Embodiments of the present disclosure belong to the technical field of intelligent control, and particularly relate to an open-pit electric shovel excavation operation control system and method. Background Art

[0002] In open-pit electric shovel excavation operations, the safety and efficiency of equipment are of utmost importance. During the loading process, if the electric shovel malfunctions or operates overloaded, it may lead to equipment damage, tipping over, or even more serious accidents. To prevent such situations, modern open-pit mines are gradually adopting intelligent technologies to improve the safety performance of equipment. For example, by installing an intelligent monitoring system, the working status of the electric shovel can be monitored in real time, and the operating parameters can be automatically adjusted according to the actual situation to avoid overload.

[0003] Traditional load monitoring systems usually rely on a single sensor or the torque values at a few key positions to evaluate the load status of the equipment. This method has obvious defects: since only limited data points are collected, it is impossible to comprehensively reflect the actual stress conditions of each key part of the equipment, making it difficult to accurately judge the overall load distribution status of the equipment. For example,

[0004] During the excavation process, different components such as the dipper arm, boom, and bucket will bear different degrees of pressure and tension, and the distribution of these forces may be uneven. If only the torque values at some of these positions are monitored, the potential overload risks at other parts may be overlooked, resulting in the equipment not being adjusted in a timely manner when some parts are already close to the limit load, increasing the likelihood of equipment damage.

[0005] In addition, insufficient real-time performance is also a major shortcoming of traditional load monitoring systems. In a dynamically changing mining environment, the working conditions of the electric shovel may change at any time, which requires the monitoring system to respond quickly and make corresponding adjustments. However, traditional systems often cannot provide sufficient real-time feedback, resulting in a lag in the operator's response when facing emergencies and being unable to take timely measures to avoid accidents, leading to the inability to give timely warnings when the equipment is approaching the overload threshold and increasing the risks of equipment damage and safety accidents. Summary of the Invention

[0006] Embodiments of the present disclosure aim to solve at least one of the technical problems existing in the prior art, and provide an open-pit electric shovel excavation operation control system and method. The system and method collect dynamic mechanical parameters of multiple key parts of the open-pit electric shovel in real time through a distributed torque sensor network and arrange them into a real-time torque distribution matrix. At the same time, a spatial distribution matrix of the multiple key parts is constructed. Further, a neural network model based on deep learning is used to perform in-depth interaction analysis on the torque distribution characteristics and spatial relationship characteristics, and based on the analysis results, the overall load state of the open-pit electric shovel is accurately perceived and dynamically evaluated. In this way, the risk of local overload of the equipment can be quickly identified, the comprehensiveness and real-time performance of equipment load monitoring are improved, equipment damage caused by local overload is effectively prevented, and at the same time, the risk of human operation errors is reduced through intelligent control, ensuring the safety and efficiency of the open-pit electric shovel operation.

[0007] On the one hand, embodiments of the present disclosure provide an open-pit electric shovel excavation operation control system, including:

[0008] A real-time torque value acquisition module, configured to acquire real-time torque values of multiple key parts of the open-pit electric shovel by using a distributed torque sensor network to obtain a set of real-time torque values;

[0009] An arrangement module, configured to arrange the set of real-time torque values into a real-time torque distribution matrix according to the position distribution of the multiple key parts of the open-pit electric shovel;

[0010] A spatial distribution matrix construction module, configured to construct a spatial distribution matrix of the multiple key parts;

[0011] A feature extraction module, configured to extract features from the real-time torque distribution matrix and the spatial distribution matrix to obtain real-time torque distribution features and key part spatial distribution features;

[0012] A feature interaction analysis module, configured to perform torque spatial distribution feature interaction analysis guided by core information on the real-time torque distribution features and the key part spatial distribution features to obtain real-time torque spatial distribution response interaction fusion coding features;

[0013] A shutdown instruction control module, configured to determine whether to generate an automatic shutdown instruction based on the real-time torque spatial distribution response interaction fusion coding features.

[0014] Optionally, the multiple key parts of the open-pit electric shovel include a dipper arm, a robotic arm, a bucket, a slewing platform, a chassis, a support frame, a winch, a steel cable, a connecting pin shaft, and a motor.

[0015] Optionally, the values of each position at non-diagonal positions in the spatial distribution matrix are Euclidean distance values of corresponding two key parts.

[0016] Optionally, the feature extraction module is further configured to:

[0017] Perform dilated convolution encoding on the real-time torque distribution matrix and the spatial distribution matrix to obtain a real-time torque distribution feature vector and a key part spatial distribution feature vector as the real-time torque distribution feature and the key part spatial distribution feature.

[0018] Optionally, the feature interaction analysis module includes:

[0019] A core information anchoring unit for constructing a semantic self-correlation associated distribution feature representation of the real-time torque distribution feature vector and the key part spatial distribution feature vector, and the semantic self-correlation associated distribution feature representations of the real-time torque distribution feature vector and the key part spatial distribution feature vector are respectively used to anchor the core features of the real-time torque distribution feature vector and the key part spatial distribution feature vector;

[0020] A response interaction encoding unit for performing real-time torque spatial distribution response interaction encoding on the real-time torque distribution feature vector and the key part spatial distribution feature vector based on the core features of the real-time torque distribution feature vector and the key part spatial distribution feature vector to obtain a real-time torque spatial distribution response interaction fusion encoding vector as the real-time torque spatial distribution response interaction fusion encoding feature.

[0021] Optionally, the core information anchoring unit is further configured to:

[0022] Construct a semantic self-correlation associated matrix of the real-time torque distribution feature vector and the key part spatial distribution feature vector to obtain a real-time torque distribution feature semantic self-correlation associated matrix and a key part spatial distribution feature semantic self-correlation associated matrix;

[0023] Input the real-time torque distribution feature semantic self-correlation associated matrix and the key part spatial distribution feature semantic self-correlation associated matrix into a core information anchoring network based on self-correlation decoupling to obtain a real-time torque distribution feature core information anchoring encoding vector and a key part spatial distribution feature core information anchoring encoding vector respectively as the core features of the real-time torque distribution feature vector and the key part spatial distribution feature vector.

[0024] Optionally, the response interaction encoding unit is further configured to:

[0025] Input the real-time torque distribution feature core information anchoring encoding vector and the key part spatial distribution feature core information anchoring encoding vector into a feature granularity response interaction encoder to obtain a feature granularity real-time torque spatial distribution response interaction encoding vector;

[0026] Input the real-time torque distribution feature core information anchoring coding vector and the key part spatial distribution feature core information anchoring coding vector into the eigenvalue granularity response interaction encoder to obtain the eigenvalue granularity real-time torque spatial distribution response interaction coding vector;

[0027] Fuse the feature granularity real-time torque spatial distribution response interaction coding vector and the eigenvalue granularity real-time torque spatial distribution response interaction coding vector to obtain the real-time torque spatial distribution response interaction fusion coding vector.

[0028] Optionally, the shutdown instruction control module is further configured to:

[0029] Input the real-time torque spatial distribution response interaction fusion coding vector into the shutdown instruction control module based on a classifier to obtain a shutdown instruction control result, and the shutdown instruction control result is used to indicate whether to generate an automatic shutdown instruction.

[0030] On the other hand, an embodiment of the present disclosure further provides an open-pit electric shovel excavation operation control method, including:

[0031] Use a distributed torque sensor network to collect the real-time torque values of multiple key parts of the open-pit electric shovel to obtain a set of real-time torque values;

[0032] Arrange the set of real-time torque values into a real-time torque distribution matrix according to the position distribution of multiple key parts of the open-pit electric shovel;

[0033] Construct a spatial distribution matrix of the multiple key parts;

[0034] Perform feature extraction on the real-time torque distribution matrix and the spatial distribution matrix to obtain real-time torque distribution features and key part spatial distribution features;

[0035] Input the real-time torque distribution features and the key part spatial distribution features into the torque spatial distribution feature interaction analysis based on core information decoupling to obtain real-time torque spatial distribution response interaction fusion coding features;

[0036] Based on the real-time torque spatial distribution response interaction fusion coding features, determine whether to generate an automatic shutdown instruction.

[0037] Compared with the related art, an open-pit electric shovel excavation operation control system and method provided by an embodiment of the present disclosure collect dynamic mechanical parameters of multiple key parts of an open-pit electric shovel in real time through a distributed torque sensor network and arrange them into a real-time torque distribution matrix. At the same time, a spatial distribution matrix of the multiple key parts is constructed, and a neural network model based on deep learning is further used to perform in-depth interaction analysis on the torque distribution characteristics and spatial relationship characteristics, and based on the analysis results, to achieve accurate perception and dynamic evaluation of the overall load state of the open-pit electric shovel. In this way, the risk of local overload of the equipment can be quickly identified, the comprehensiveness and real-time performance of equipment load monitoring are improved, equipment damage caused by local overload is effectively prevented, and at the same time, the risk of human operation errors is reduced through intelligent control, ensuring the safety and efficiency of open-pit electric shovel operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a block diagram of an open-pit electric shovel excavation operation control system according to an embodiment of the present disclosure;

[0039] Figure 2 is a schematic diagram of data flow of an open-pit electric shovel excavation operation control system according to an embodiment of the present disclosure;

[0040] Figure 3 is a block diagram of a feature interaction analysis module in an open-pit electric shovel excavation operation control system according to an embodiment of the present disclosure;

[0041] Figure 4 is a flowchart of an open-pit electric shovel excavation operation control method according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] Hereinafter, exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments of the present disclosure. It should be understood that the present disclosure is not limited by the exemplary embodiments described herein.

[0043] As shown in the present disclosure and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "including" and "comprising" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0044] Although the present disclosure makes various references to certain modules in the system according to the embodiments of the present disclosure, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0045] In the present disclosure, flowcharts are used to illustrate the operations performed by the systems according to the embodiments of the present disclosure. It should be understood that the operations above or below do not necessarily have to be performed precisely in sequence. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more steps can be removed from these processes.

[0046] Next, example embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments of the present disclosure. It should be understood that the present disclosure is not limited by the example embodiments described herein.

[0047] In the technical solution of the present disclosure, an open-pit electric shovel excavation operation control system is proposed. Figure 1 FIG. is a block diagram of an open-pit electric shovel excavation operation control system according to an embodiment of the present disclosure. Figure 2 FIG. is a schematic diagram of data flow of an open-pit electric shovel excavation operation control system according to an embodiment of the present disclosure. As Figure 1 and Figure 2 shown, the open-pit electric shovel excavation operation control system 300 according to the embodiment of the present disclosure includes: a real-time torque value acquisition module 310, configured to acquire real-time torque values of multiple key parts of the open-pit electric shovel by using a distributed torque sensor network to obtain a set of real-time torque values. An arrangement module 320, configured to arrange the set of real-time torque values into a real-time torque distribution matrix according to the position distribution of the multiple key parts of the open-pit electric shovel. A spatial distribution matrix construction module 330, configured to construct a spatial distribution matrix of the multiple key parts. A feature extraction module 340, configured to perform feature extraction on the real-time torque distribution matrix and the spatial distribution matrix to obtain a real-time torque distribution feature and a key part spatial distribution feature. A feature interaction analysis module 350, configured to perform a torque spatial distribution feature interaction analysis guided by core information on the real-time torque distribution feature and the key part spatial distribution feature to obtain a real-time torque spatial distribution response interaction fusion coding feature. A shutdown instruction control module 360, configured to determine whether to generate an automatic shutdown instruction based on the real-time torque spatial distribution response interaction fusion coding feature.

[0048] Exemplarily, the real-time torque value acquisition module 310 is configured to acquire the real-time torque values of multiple key parts of the surface mine electric shovel by using a distributed torque sensor network to obtain a set of real-time torque values. Among them, the multiple key parts of the surface mine electric shovel include the dipper arm, the robotic arm, the bucket, the slewing platform, the chassis, the support frame, the winch, the steel cable, the connecting pin shaft, and the motor. It should be understood that as a large and complex construction machinery, the working environment of the electric shovel is harsh, the load changes frequently, and each key part bears complex mechanical forces during the operation. Through the distributed sensor network, all key stress parts of the equipment can be comprehensively covered, and multi-source mechanical data can be collected in real time. In this way, not only can the local overload risk be quickly identified, but also the operation parameters can be automatically adjusted according to the actual situation to avoid overload. The deployment of the distributed sensor network can achieve a comprehensive perception of the overall load state of the equipment, avoiding misjudgment or missed judgment caused by local data loss. At the same time, the real-time acquisition of multi-source data also provides a solid foundation for subsequent intelligent analysis and dynamic control.

[0049] In one example, during the operation of the equipment, various components such as the dipper arm, boom, bucket, slewing platform, chassis, support frame, winch, steel cable, connecting pin shaft, and motor all bear different mechanical forces, and their stress states directly reflect the load conditions of the equipment. First, as one of the main load-bearing components of the electric shovel, the dipper arm bears huge pushing and pulling forces during the excavation process. The moment value of the dipper arm can directly reflect the load conditions of the excavation operation. If the moment value of the dipper arm exceeds the safety threshold, it may mean that the excavation resistance is too large and there is a risk of equipment overload. Second, the boom is an important component connecting the dipper arm and the bucket, and its stress state directly affects the stability and operation efficiency of the equipment. The boom bears complex bending and torsional moments during the operation. If the moment value of the boom is abnormal, it may mean that the equipment load is unevenly distributed or there are structural problems. The bucket, as the component directly in contact with the material, its stress state directly reflects the load conditions of the excavation operation. The moment value of the bucket can reflect the hardness of the material and the excavation depth. If the moment value of the bucket is too large, it may mean that the material is too hard or the excavation is too deep, and there is a risk of equipment overload. The slewing platform is a key component for the electric shovel to achieve turning and positioning, and its stress state directly affects the movement flexibility and stability of the equipment. The slewing platform bears complex torsional moments during the operation. If the moment value of the slewing platform is abnormal, it may mean that the turning resistance of the equipment is too large or there are structural problems. The chassis, as the support foundation of the electric shovel, its stress state directly affects the overall stability of the equipment. The chassis bears huge pressure and torsional moments during the operation. If the moment value of the chassis is abnormal, it may mean that the equipment load is unevenly distributed or there are structural problems. The support frame is an important component connecting the chassis and the upper structure, and its stress state directly affects the overall rigidity and stability of the equipment. The support frame bears complex bending and torsional moments during the operation. If the moment value of the support frame is abnormal, it may mean that the equipment load is unevenly distributed or there are structural problems. The winch and the steel cable are key components for the electric shovel to achieve lifting and lowering, and their stress states directly affect the operation efficiency and safety of the equipment. The winch and the steel cable bear huge tensile and torsional moments during the operation. If the moment value of the winch and the steel cable is abnormal, it may mean that the lifting resistance is too large or there are structural problems. The connecting pin shaft is the connection point between various components of the electric shovel, and its stress state directly affects the overall stability and safety of the equipment. The connecting pin shaft bears complex shear and torsional moments during the operation. If the moment value of the connecting pin shaft is abnormal, it may mean that the equipment load is unevenly distributed or there are structural problems. Finally, the motor, as the power source of the electric shovel, its stress state directly affects the operation efficiency and safety of the equipment. The motor bears huge torque and load during the operation. If the moment value of the motor is abnormal, it may mean that the equipment load is too large or there are problems with the power system. Therefore, by real-time monitoring and analysis of the moment values of each key component, the system can comprehensively and accurately evaluate the load state of the equipment.

[0050] Exemplarily, the arrangement module 320 is configured to arrange the set of real-time torque values into a real-time torque distribution matrix according to the position distribution of multiple key parts of the surface mine electric shovel. Considering that traditional load monitoring systems usually only focus on the torque values of a single or a few key parts, while ignoring the spatial distribution characteristics of these torque values, resulting in an inability to accurately evaluate the overall load status of the equipment. In the technical solution of the present disclosure, the set of real-time torque values is arranged into a real-time torque distribution matrix according to the position distribution of multiple key parts of the surface mine electric shovel. By arranging the real-time collected torque values according to the spatial positions of the key parts of the equipment, a matrix structure reflecting the overall load distribution of the equipment can be constructed, which can not only intuitively reflect the torque values of each key part, but also reflect the mutual relationship of these torque values in space, thereby providing more comprehensive and accurate data support for the dynamic evaluation and risk warning of the equipment load status. Specifically, the construction of the real-time torque distribution matrix can help the system quickly identify the local overload risk of the equipment and achieve dynamic adjustment and optimization of the equipment through intelligent control, thereby effectively preventing the occurrence of equipment damage and safety accidents.

[0051] Exemplarily, the spatial distribution matrix construction module 330 is configured to construct the spatial distribution matrix of the multiple key parts. It should be understood that traditional monitoring methods only rely on limited data points to evaluate the working state of the electric shovel, which cannot comprehensively reflect the actual stress conditions of each key part of the equipment and is difficult to accurately judge the overall load distribution state of the equipment. For example, during the excavation process, different components such as the dipper arm, the boom, and the bucket will bear different degrees of pressure and tension, and the distribution of these forces may be uneven. Therefore, in the technical solution of the present disclosure, the spatial distribution matrix of the multiple key parts is constructed, wherein the values at each position in the non-diagonal positions of the spatial distribution matrix are the Euclidean distance values of the corresponding two key parts. By constructing the spatial distribution matrix of multiple key parts and setting the values in the non-diagonal positions as the Euclidean distance between the corresponding two key parts, the physical structure characteristics of the entire system can be depicted more precisely. In this way, not only can the position information of each key part be reflected, but also the relative position relationship between each key part can be revealed, which is crucial for the training and feature extraction of subsequent deep learning models.

[0052] Exemplarily, the feature extraction module 340 is used to extract features from the real-time moment distribution matrix and the spatial distribution matrix to obtain real-time moment distribution features and key parts spatial distribution features. In the technical solution disclosed in the present invention, the real-time moment distribution matrix and the spatial distribution matrix are subjected to hole convolution coding to obtain real-time moment distribution feature vectors and key parts spatial distribution feature vectors as the real-time moment distribution features and the key parts spatial distribution features. It should be understood that the real-time moment distribution matrix provides the force conditions of each key part at a specific time point, while the spatial distribution matrix reflects the spatial relationship between these parts. As a special convolution method, hole convolution coding can expand the receptive field without increasing the computational complexity, so that the model can capture a wider range of contextual information. Through hole convolution coding, the system can extract high-level moment distribution features and spatial relationship features from the real-time moment distribution matrix and the spatial distribution matrix, thereby providing more comprehensive and accurate data support for subsequent intelligent analysis and dynamic control. In this way, the system can not only comprehensively capture the torque distribution characteristics and spatial relationship characteristics of each key part of the equipment, but also understand the propagation path and impact range of these risks in the entire equipment, thereby making more accurate early warning and control decisions.

[0053] Exemplarily, the feature interaction analysis module 350 is used to perform a torque spatial distribution feature interaction analysis based on core information guidance on the real-time torque distribution feature and the key part spatial distribution feature to obtain a real-time torque spatial distribution response interaction fusion coding feature. It should be understood that the interaction between different key parts of the electric shovel (such as the dipper arm, the mechanical arm, and the bucket) is very complex. This complexity is not only reflected in the torque change of a single part, but also includes how the various parts work together to cope with different excavation conditions. For example, when excavating hard rocks, the bucket and the dipper arm may be subjected to greater pressure than usual. When carrying a large amount of loose materials, the stability of the mechanical arm and the rotary platform is more tested. These different operating conditions require the monitoring system to not only capture the state changes of a single key part, but also understand and predict the complex interaction patterns between the various parts. The traditional shallow feature interaction method cannot capture such a complex interaction relationship. Therefore, in the technical solution of the present disclosure, the real-time torque distribution feature and the key part spatial distribution feature are subjected to a torque spatial distribution feature interaction analysis based on core information guidance to obtain a real-time torque spatial distribution response interaction fusion coding feature. This step aims to highlight the parts that are highly relevant to the core semantics and remove redundant information, thereby extracting a refined core anchor point representation. For an electric shovel, this means identifying which areas or components are most likely to have overload risks, thereby quickly identifying local overload risks and automatically adjusting equipment operating parameters when necessary to avoid equipment damage or accidents. In a specific example of the present disclosure,Figure 3 As shown, the feature interaction analysis module 350 includes: a core information anchoring unit 351, configured to construct a semantic self - correlation associated distribution feature representation of the real - time torque distribution feature vector and the key - part spatial distribution feature vector, where the semantic self - correlation associated distribution feature representations of the real - time torque distribution feature vector and the key - part spatial distribution feature vector are respectively used to anchor the core features of the real - time torque distribution feature vector and the key - part spatial distribution feature vector; and a response interaction encoding unit 352, configured to perform real - time torque spatial distribution response interaction encoding on the real - time torque distribution feature vector and the key - part spatial distribution feature vector based on the core features of the real - time torque distribution feature vector and the key - part spatial distribution feature vector, so as to obtain a real - time torque spatial distribution response interaction fusion encoding vector as the real - time torque spatial distribution response interaction fusion encoding feature.

[0054] Specifically, the core information anchoring unit 351 is configured to construct a semantic self - correlation associated distribution feature representation of the real - time torque distribution feature vector and the key - part spatial distribution feature vector, where the semantic self - correlation associated distribution feature representations of the real - time torque distribution feature vector and the key - part spatial distribution feature vector are respectively used to anchor the core features of the real - time torque distribution feature vector and the key - part spatial distribution feature vector. In an embodiment of the present disclosure, first, a semantic self - correlation associated matrix of the real - time torque distribution feature vector and the key - part spatial distribution feature vector is constructed to obtain a real - time torque distribution feature semantic self - correlation associated matrix and a key - part spatial distribution feature semantic self - correlation associated matrix. In this process, through dot - product operations or other similarity - measurement methods, the internal relationship of torque changes between different key parts is revealed. In an example, it should be understood that since the rock is extremely hard, the bucket is subjected to a great resistance, and this resistance not only directly affects the bucket itself but also is transmitted along the robotic arm to the arm, thus affecting the stability of the entire system. By calculating the self - correlation of the torque values between these components, the system can identify potential chain - reaction patterns, thereby providing basic data for subsequent risk assessment. In a specific example of the present disclosure, the following semantic self - correlation associated matrix construction formula is used to construct the semantic self - correlation associated matrix of the real - time torque distribution feature vector and the key - part spatial distribution feature vector to obtain a real - time torque distribution feature semantic self - correlation associated matrix and a key - part spatial distribution feature semantic self - correlation associated matrix. Wherein, the semantic self - correlation associated matrix construction formula is:

[0055]

[0056] where v1 represents the real - time torque distribution feature vector, and v2 represents the key - part spatial distribution feature vector. Let φ(·) be a feature mapping function, such as a linear mapping or a non-linear kernel function. M1 represents the semantic autocorrelation association matrix of the real-time torque distribution features, and M2 represents the semantic autocorrelation association matrix of the spatial distribution features of the key parts.

[0057] Furthermore, the semantic autocorrelation association matrix of the real-time torque distribution features and the semantic autocorrelation association matrix of the spatial distribution features of the key parts are respectively input into the core information anchoring network based on autocorrelation decoupling to obtain the core information anchoring coding vectors of the real-time torque distribution features and the core information anchoring coding vectors of the spatial distribution features of the key parts, which are respectively used as the core features of the real-time torque distribution feature vector and the spatial distribution feature vector of the key parts. That is, after generating the semantic autocorrelation association matrix of the real-time torque distribution features and the semantic autocorrelation association matrix of the spatial distribution features of the key parts, the core information anchoring network based on autocorrelation decoupling is used to refine and purify the internal information of the features. Specifically, through the method of feature distillation, the part highly correlated with the core semantics in the feature vector is highlighted, the redundant information is denoised, and the core anchor point representation with refined structure is extracted therefrom. During this process, the system can accurately locate which regions or components are most likely to have overload risks. In a specific example of the present disclosure, the semantic autocorrelation association matrix of the real-time torque distribution features and the semantic autocorrelation association matrix of the spatial distribution features of the key parts are respectively input into the core information anchoring network based on autocorrelation decoupling according to the following formula to obtain the core information anchoring coding vectors of the real-time torque distribution features and the core information anchoring coding vectors of the spatial distribution features of the key parts. Wherein, the formula is:

[0058]

[0059] Wherein, decouple(·) represents the feature decoupling operation, x 11 , x 12 , x 1i , x 1n respectively represent the row vectors of the semantic autocorrelation association matrix of the spatial distribution features of the key parts, W 1i and b 1i are the weight matrix and the bias vector respectively, is the matrix multiplication, is the node position modulation vector corresponding to the node position of x 1i in the row vectors of the semantic autocorrelation association matrix of the spatial distribution features of the key parts, e 1i represents the core information anchoring factor of the real-time torque distribution features, sigmoid(·) is the sigmoid function, a 1iDenote the real-time torque distribution feature core information anchoring modulation vector, n is the number of row vectors of the semantic autocorrelation correlation matrix of the spatial distribution feature of the key part, c1 represents the real-time torque distribution feature core information anchoring coding vector, x 21 , x 22 , x 2i , x 2n respectively represent the respective row vectors of the semantic autocorrelation correlation matrix of the spatial distribution feature of the key part, the node position modulation vector corresponding to the node position in each row vector of the semantic autocorrelation correlation matrix of the spatial distribution feature of the key part, W 2i and b 2i are the weight matrix and the bias vector respectively, e 2i represents the key part spatial distribution feature core information anchoring factor, a 2i represents the key part spatial distribution feature core information anchoring modulation vector, m is the number of row vectors of the semantic autocorrelation correlation matrix of the spatial distribution feature of the key part, c2 represents the key part spatial distribution feature core information anchoring coding vector.

[0060] Specifically, the response interaction encoding unit 352 is configured to perform real-time torque spatial distribution response interaction encoding on the real-time torque distribution feature vector and the key part spatial distribution feature vector based on the core features of the real-time torque distribution feature vector and the key part spatial distribution feature vector, so as to obtain a real-time torque spatial distribution response interaction fusion encoding vector as the real-time torque spatial distribution response interaction fusion encoding feature. In an embodiment of the present disclosure, first, the real-time torque distribution feature core information anchored encoding vector and the key part spatial distribution feature core information anchored encoding vector are input into a feature granularity response interaction encoder to obtain a feature granularity real-time torque spatial distribution response interaction encoding vector. Then, the real-time torque distribution feature core information anchored encoding vector and the key part spatial distribution feature core information anchored encoding vector are input into a feature value granularity response interaction encoder to obtain a feature value granularity real-time torque spatial distribution response interaction encoding vector. That is, the feature granularity response interaction encoder and the feature value granularity response interaction encoder are respectively used to perform response interaction encoding on the real-time torque distribution feature core information anchored encoding vector and the key part spatial distribution feature core information anchored encoding vector, so as to explore the complex dependence relationships between torque distribution features from different dimensions. Among them, the interaction modeling at the feature granularity level captures the semantic coupling at the microscopic level, while the feature value granularity regresses to more underlying numerical operations to characterize the mutual dependence relationships between feature values. Thus, the system can not only identify which areas have overload risks, but also understand how these risks spread among different components, providing a basis for formulating precise countermeasures. Subsequently, the feature granularity real-time torque spatial distribution response interaction encoding vector and the feature value granularity real-time torque spatial distribution response interaction encoding vector are fused to obtain the real-time torque spatial distribution response interaction fusion encoding vector. That is, by fusing the feature granularity real-time torque spatial distribution response interaction encoding vector and the feature value granularity real-time torque spatial distribution response interaction encoding vector, a real-time torque spatial distribution response interaction fusion encoding vector that can comprehensively reflect the real-time torque state and its spatial distribution of each key part of the electric shovel is obtained. Based on this, the system can make more accurate and timely responses when facing a complex mining environment, preventing possible equipment damage and safety accidents. In a specific example of the present disclosure, based on the core features of the real-time torque distribution feature vector and the key part spatial distribution feature vector, the real-time torque distribution feature vector and the key part spatial distribution feature vector are subjected to real-time torque spatial distribution response interaction encoding according to the following response interaction encoding formula to obtain a real-time torque spatial distribution response interaction fusion encoding vector. Wherein, the response interaction encoding formula is:

[0061]

[0062] Vf = concat[E granular ; E value

[0063] Among them, is addition by position, W VT and b VT are the weight matrix and the bias vector respectively, tanh(·) is the tanh function, and E granular represents the real-time torque space distribution response interaction coding vector of the feature granularity, and E value represents the real-time torque space distribution response interaction coding vector of the eigenvalue granularity. Concat{·; ·} is the concatenation operation, and V f represents the real-time torque space distribution response interaction fusion coding vector.

[0064] In particular, the shutdown instruction control module 360 is used to determine whether to generate an automatic shutdown instruction based on the real-time torque space distribution response interaction fusion coding feature. In a specific example of the present disclosure, the real-time torque space distribution response interaction fusion coding vector is input into the shutdown instruction control module based on the classifier to obtain the shutdown instruction control result, and the shutdown instruction control result is used to indicate whether to generate an automatic shutdown instruction. That is, by extracting high-level torque distribution features and spatial relationship features from the real-time torque space distribution response interaction fusion coding vector, the dynamic evaluation and risk warning of the equipment load state are realized. In this way, more accurate data support can be provided for the intelligent control of the equipment, and through real-time monitoring and feedback, the automatic adjustment and optimization of the equipment operation can be realized, thereby improving the operation safety and efficiency.

[0065] As described above, the open-pit electric shovel excavation operation control system 300 according to the embodiments of the present disclosure can be implemented in various wireless terminals, such as a server with an open-pit electric shovel excavation operation control algorithm. In a possible implementation manner, the open-pit electric shovel excavation operation control system 300 according to the embodiments of the present disclosure can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the open-pit electric shovel excavation operation control system 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal. Of course, the open-pit electric shovel excavation operation control system 300 can also be one of the many hardware modules of the wireless terminal.

[0066] Alternatively, in another example, the open-pit electric shovel excavation operation control system 300 and the wireless terminal can also be separate devices, and the open-pit electric shovel excavation operation control system 300 can be connected to the wireless terminal through a wired and / or wireless network and transmit interaction information according to a predefined data format. ​

[0067] Furthermore, an embodiment of the present disclosure also provides a method for controlling the excavation operation of an open-pit electric shovel.

[0068] Figure 4 It is a flowchart of the method for controlling the excavation operation of an open-pit electric shovel according to an embodiment of the present disclosure. As Figure 4 shown, according to the method for controlling the excavation operation of an open-pit electric shovel according to an embodiment of the present disclosure, it includes the steps of: S1, using a distributed torque sensor network to collect real-time torque values of multiple key parts of the open-pit electric shovel to obtain a set of real-time torque values. S2, arranging the set of real-time torque values into a real-time torque distribution matrix according to the position distribution of multiple key parts of the open-pit electric shovel. S3, constructing a spatial distribution matrix of the multiple key parts. S4, performing feature extraction on the real-time torque distribution matrix and the spatial distribution matrix to obtain a real-time torque distribution feature and a key part spatial distribution feature. S5, inputting the real-time torque distribution feature and the key part spatial distribution feature into an interactive analysis of torque spatial distribution features based on core information decoupling to obtain a real-time torque spatial distribution response interactive fusion coding feature. S6, determining whether to generate an automatic shutdown instruction based on the real-time torque spatial distribution response interactive fusion coding feature.

[0069] In summary, the method for controlling the excavation operation of an open-pit electric shovel according to an embodiment of the present disclosure is elucidated. It uses a distributed torque sensor network to collect dynamic mechanical parameters of multiple key parts of the open-pit electric shovel in real time and arranges them into a real-time torque distribution matrix. At the same time, it constructs a spatial distribution matrix of the multiple key parts, further uses a neural network model based on deep learning to perform in-depth interactive analysis on the torque distribution feature and the spatial relationship feature, and realizes precise perception and dynamic evaluation of the overall load state of the open-pit electric shovel based on the analysis results. In this way, it can quickly identify the local overload risk of the equipment, improve the comprehensiveness and real-time performance of equipment load monitoring, effectively prevent equipment damage caused by local overload, and at the same time reduce the risk of human operation errors through intelligent control, ensuring the safety and efficiency of the open-pit electric shovel operation.

[0070] The various embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technologies in the market, or to enable other ordinary technical personnel in the technical field to understand the disclosed embodiments. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present disclosure, and these modifications and improvements are also regarded as the protection scope of the present disclosure.

Claims

1. A control system for open-pit mine electric shovel excavation operation, characterized in that: include: A real-time torque value collection module is used to collect real-time torque values ​​of multiple key parts of the open-pit mine shovel using a distributed torque sensor network to obtain a set of real-time torque values; An arrangement module, used for arranging the set of real-time torque values ​​into a real-time torque distribution matrix according to the position distribution of multiple key parts of the open-pit mining electric shovel; A spatial distribution matrix construction module, used to construct the spatial distribution matrix of the plurality of key parts; A feature extraction module, used for extracting features from the real-time moment distribution matrix and the spatial distribution matrix to obtain real-time moment distribution features and key parts spatial distribution features; A feature interaction analysis module, used for performing a torque spatial distribution feature interaction analysis on the real-time torque distribution feature and the key part spatial distribution feature based on core information guidance to obtain a real-time torque spatial distribution response interaction fusion coding feature; A stop instruction control module is used to determine whether to generate an automatic stop instruction based on the real-time torque spatial distribution response interactive fusion coding characteristics.

2. The open-pit mine electric shovel excavation control system according to claim 1, characterized in that: The multiple key parts of the open-pit mine electric shovel include a dipper arm, a mechanical arm, a bucket, a rotary platform, a chassis, a support frame, a winch, a steel cable, a connecting pin shaft and a motor.

3. The open-pit mine electric shovel excavation operation control system according to claim 1, characterized in that: The value of each non-diagonal position in the spatial distribution matrix is ​​the Euclidean distance value between the corresponding two key parts.

4. The open-pit mine electric shovel excavation operation control system according to claim 1, characterized in that: The feature extraction module is further used for: The real-time moment distribution matrix and the spatial distribution matrix are subjected to dilated convolution coding to obtain a real-time moment distribution feature vector and a key-part spatial distribution feature vector as the real-time moment distribution feature and the key-part spatial distribution feature.

5. The open-pit mine electric shovel excavation control system according to claim 4, characterized in that: The feature interaction analysis module includes: A core information anchoring unit, used to construct a semantic autocorrelation associated distribution feature representation of the real-time moment distribution feature vector and the key part spatial distribution feature vector, wherein the semantic autocorrelation associated distribution feature representation of the real-time moment distribution feature vector and the key part spatial distribution feature vector are used to anchor the core features of the real-time moment distribution feature vector and the key part spatial distribution feature vector respectively; A response interaction coding unit is used to perform real-time moment space distribution response interaction coding on the real-time moment distribution feature vector and the key part space distribution feature vector based on the core features of the real-time moment distribution feature vector and the key part space distribution feature vector to obtain a real-time moment space distribution response interaction fusion coding vector as the real-time moment space distribution response interaction fusion coding feature.

6. The open-pit mine electric shovel excavation control system according to claim 5, characterized in that: The core information anchoring unit is further used for: Constructing a semantic autocorrelation association matrix of the real-time moment distribution feature vector and the key part spatial distribution feature vector to obtain a real-time moment distribution feature semantic autocorrelation association matrix and a key part spatial distribution feature semantic autocorrelation association matrix; The real-time moment distribution feature semantic autocorrelation association matrix and the key part spatial distribution feature semantic autocorrelation association matrix are respectively input into the core information anchoring network based on autocorrelation decoupling to obtain the real-time moment distribution feature core information anchoring coding vector and the key part spatial distribution feature core information anchoring coding vector as the core features of the real-time moment distribution feature vector and the key part spatial distribution feature vector respectively.

7. The open-pit mine electric shovel excavation control system according to claim 6, characterized in that: The response interaction encoding unit is further used for: Input the real-time moment distribution feature core information anchor coding vector and the key part spatial distribution feature core information anchor coding vector into a feature granularity response interactive encoder to obtain a feature granularity real-time moment spatial distribution response interactive coding vector; Input the real-time moment distribution feature core information anchor coding vector and the key part spatial distribution feature core information anchor coding vector into an eigenvalue granularity response interactive encoder to obtain an eigenvalue granularity real-time moment spatial distribution response interactive coding vector; The feature granularity real-time moment space distribution response interactive coding vector and the eigenvalue granularity real-time moment space distribution response interactive coding vector are fused to obtain the real-time moment space distribution response interactive fusion coding vector.

8. The open-pit mine electric shovel excavation control system according to claim 7, characterized in that: The shutdown instruction control module is also used for: The real-time torque spatial distribution response interactive fusion coding vector is input into a classifier-based stop instruction control module to obtain a stop instruction control result, and the stop instruction control result is used to indicate whether an automatic stop instruction is generated.

9. A method for controlling excavation operations of an open-pit mine shovel, characterized in that: include: Using a distributed torque sensor network to collect real-time torque values ​​of multiple key parts of the open-pit mine shovel to obtain a set of real-time torque values; Arranging the set of real-time torque values ​​into a real-time torque distribution matrix according to the position distribution of multiple key parts of the open-pit mining shovel; Constructing a spatial distribution matrix of the plurality of key parts; Performing feature extraction on the real-time moment distribution matrix and the spatial distribution matrix to obtain real-time moment distribution features and key parts spatial distribution features; Interactive analysis of the moment spatial distribution characteristics based on core information decoupling is performed on the real-time moment distribution characteristics and the key parts spatial distribution characteristics to obtain the real-time moment spatial distribution response interactive fusion coding characteristics; Based on the real-time torque spatial distribution response interactive fusion coding features, it is determined whether to generate an automatic shutdown instruction.

Images