Excavation resistance prediction method and system of digital-analog hybrid drive excavator

By establishing the mechanism model of the interaction between materials and buckets of excavators and combining machine learning algorithms, a mining resistance prediction model is constructed, which solves the problem of rigidity of bucket materials model in the existing technology, and realizes accurate prediction of mining resistance and generalization of model The improvement of capacity has laid the foundation for the energy conservation, emission reduction and automated operations of excavators.

CN119004936APending Publication Date: 2024-11-22XIAMEN UNIV
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410339338.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-03-25
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing bucket material model based on mechanism analysis is too rigid, making it difficult to effectively deal with complex systems and unknown mechanisms in excavator excavation process. The data-driven model is sensitive to data quality and has poor interpretation, making it difficult to accurately predict mining resistance in small samples. .

Method used

By obtaining the actual mining data of the excavator, a mechanism model for interaction between materials and buckets is established, and a mining resistance prediction model is constructed in combination with machine learning algorithms. The loss function is used to integrate mechanism model information to improve the generalization ability and prediction accuracy of the model.

Benefits of technology

Accurate prediction of excavation resistance is achieved, the generalization ability and prediction accuracy of the model are improved, and the foundation is provided for the excavator's energy conservation, emission reduction and automated operations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004936A_ABST
    Figure CN119004936A_ABST
Patent Text Reader

Abstract

The invention discloses an excavation resistance prediction method and system for a digital-analog hybrid drive excavator. The method comprises the steps that actual excavation data of the excavator in the material excavation process are obtained; preprocessing the actual mining data to obtain mining parameters and actual mining resistance; establishing a material and bucket interaction mechanism model, and calculating theoretical excavation resistance; a machine learning algorithm is utilized to construct an excavation resistance prediction model, the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters (excavation depth, excavation angle and the like) as input, and predicted resistance is output; constructing a loss function according to the theoretical excavation resistance and the actual excavation resistance, and training an excavation resistance prediction model; and realizing excavation resistance prediction by using the trained excavation resistance prediction model. According to the method, the excavation resistance of the excavator can be accurately predicted by analyzing the interaction state of the bucket and the material in the material excavation process of the excavator, establishing the material and bucket interaction mechanism model and combining the machine learning algorithm.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of engineering machinery, and in particular to a method and system for predicting excavation resistance of an excavator. Background Art

[0002] As a kind of engineering machinery widely used in infrastructure and mining, the energy saving and consumption reduction and automation research of excavators have always attracted people's attention. Accurately predicting the excavation resistance on the excavation trajectory can provide a basis for optimizing the excavation trajectory and achieving energy conservation and emission reduction. By accurately predicting the excavation resistance, the energy consumption of the excavator can be effectively managed, thereby achieving the goal of energy conservation and emission reduction. By optimizing the working mode and excavation parameters of the excavator, the energy consumption caused by the excavation resistance can be minimized, and energy-saving control of the excavation process can be achieved, which helps to reduce carbon emissions and environmental impact.

[0003] However, the interaction between materials and buckets during excavation is a very complex process, which not only involves many influencing parameters but is also full of uncertainty. Existing bucket material models based on mechanism analysis provide strong interpretability and strong adaptability to scarce data, but they are challenging in dealing with system complexity and unknown mechanisms, and may be too rigid. In addition, data-driven models excel in complex nonlinear relationships with their flexibility and generalization ability, but are sensitive to data quality, have relatively poor interpretability, and are prone to overfitting in small sample situations. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that the bucket material model based on mechanism analysis in the prior art is too rigid. It analyzes the interaction state between the bucket and the material during the excavation process of the excavator and establishes a mechanism model of the interaction between the material and the bucket. Then, combined with the machine learning algorithm, an accurate excavation resistance prediction method for the excavator is proposed.

[0005] The technical solution adopted by the present invention to solve the technical problem is: to provide a method for predicting the excavation resistance of a digital-analog hybrid drive excavator, comprising the following steps:

[0006] Acquire actual excavation data of the excavator during the excavation of materials, wherein the actual excavation data includes material parameters, bucket parameters and excavation trajectory;

[0007] Preprocess the actual mining data to obtain mining parameters and actual mining resistance;

[0008] Establish a material-bucket interaction model and calculate the theoretical excavation resistance based on the input material parameters, bucket parameters and excavation parameters;

[0009] Constructing an excavation resistance prediction model, wherein the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters as inputs and outputs predicted resistance;

[0010] The loss function is constructed based on the theoretical mining resistance and the actual mining resistance, and the mining resistance prediction model is trained;

[0011] The trained excavation resistance prediction model is used to realize excavation resistance prediction.

[0012] Preferably, the actual excavation data of the excavator during the material excavation process is obtained as follows:

[0013] Use empirical values ​​as material parameters, including the internal friction angle between the material and the bucket, material density, material internal friction, and material cohesion coefficient;

[0014] Obtain the excavator bucket parameters through actual measurement, including bucket width, bucket bottom plate length, and the angle ε between the bucket bottom plate and the side edge;

[0015] The wire-pull displacement sensor is used to obtain the excavation trajectory, including the boom cylinder displacement, dipper arm cylinder displacement and bucket displacement during the excavation process.

[0016] Preferably, the actual mining data also includes:

[0017] The surface morphology point cloud of the material to be excavated is obtained by using a binocular camera;

[0018] The pin force is obtained using the pin force sensor.

[0019] Preferably, the preprocessing of the actual mining data includes:

[0020] The pin force obtained by the pin force sensor is combined with the bucket dynamics model to obtain the actual digging resistance;

[0021] A kinematic model of the excavator working device is established, and a base coordinate system is established with the excavator's rotation center and the ground contact position. The surface morphology point cloud of the material to be excavated based on the binocular camera coordinate system is converted to the base coordinate system, and the fitting function of the material surface is obtained by polynomial fitting. The excavation parameters are obtained according to the fitting function of the material surface.

[0022] Preferably, the excavation parameters are obtained according to the fitting function of the material surface, and the excavation parameters include material accumulation angle, excavation depth, and excavation front angle. The acquisition process is specifically as follows:

[0023] The position of the bucket tip and the bucket posture in the base coordinate system are solved according to the displacement of the boom cylinder, the arm cylinder and the bucket;

[0024] Differentiate the fitting function of the material surface to obtain the material accumulation angle at different positions;

[0025] The excavation depth and excavation front angle are solved according to the fitting function of the material surface, the position of the bucket tip and the bucket posture.

[0026] Preferably, the establishment of a material-bucket interaction model that takes into account different interaction states between the material and the bucket is specifically as follows:

[0027] The interaction between the bucket and the material during the excavation process is analyzed in stages, and the relationship between the input parameters and the theoretical excavation resistance is established; the excavation process is divided into four stages, including: the first stage, the bucket is not inserted into the material, and the excavation resistance is zero; the second stage, the bucket bottom plate is inserted into the material, and the bucket rear plate does not contact the material; the third stage, the bucket bottom plate is completely inserted into the material, and the material is in contact with the bucket rear plate; the fourth stage, the bucket tip leaves the material and the excavation is completed.

[0028] Preferably, the training mining resistance prediction model is as follows:

[0029] The data set is constructed by taking material parameters, bucket parameters and excavation parameters as input and actual excavation resistance as label;

[0030] Design the loss function of the excavation resistance prediction model. The loss function consists of two parts. The first part is the variance between the output value of the machine learning model and the actual excavation resistance. The second part is the variance between the output value of the machine learning model and the excavation resistance calculated by the mechanism model.

[0031] The model is trained using the data set. During the training process, the model parameters are continuously adjusted to minimize the loss function, thereby obtaining the optimal mining resistance prediction model.

[0032] Preferably, the mining resistance prediction is realized by using the trained mining resistance prediction model, specifically:

[0033] Given excavation conditions, including excavation trajectory, material information of the material to be excavated, and bucket parameters, the material information includes the material surface morphology point cloud and material parameters;

[0034] Pre-process the excavation trajectory and material information to obtain the excavation parameters;

[0035] The material parameters, bucket parameters and excavation parameters are input into the trained excavation resistance prediction model to predict the excavation resistance for given excavation conditions.

[0036] The present invention also provides a digital-analog hybrid drive excavator excavation resistance prediction system, comprising:

[0037] A data acquisition module, which acquires actual excavation data of the excavator during the excavation of materials, wherein the actual excavation data includes material parameters, bucket parameters and excavation trajectory;

[0038] Data preprocessing module, which preprocesses the actual mining data to obtain mining parameters and actual mining resistance;

[0039] Theoretical value calculation module, which establishes the interaction model between materials and buckets, and calculates the theoretical excavation resistance according to the input material parameters, bucket parameters and excavation parameters;

[0040] A prediction model building module is used to build an excavation resistance prediction model, wherein the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters as inputs and outputs predicted resistance;

[0041] The prediction model training module constructs a loss function based on theoretical mining resistance and actual mining resistance to train the mining resistance prediction model;

[0042] The prediction model application module uses the trained excavation resistance prediction model to realize excavation resistance prediction.

[0043] The present invention has the following beneficial effects: by analyzing the differences in the physical interaction states between the bucket and the excavator during the excavation process, a model of the interaction mechanism between the material and the bucket is constructed, and the excavation process data of different types of materials are collected to construct a data set. In the machine learning algorithm, the information of the mechanism model is integrated into the machine learning algorithm by designing a loss function, so as to improve the generalization ability of the model prediction while ensuring the accuracy of the model. The constructed excavation resistance prediction model can pre-calculate the excavation resistance on the given excavation trajectory according to the trajectory, laying the foundation for trajectory planning during the excavation process of the excavator, thereby helping to achieve energy conservation and emission reduction as well as automated and unmanned operation of the excavator.

[0044] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments, but the present invention is not limited to the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A method step diagram of an embodiment of the present invention;

[0046] Figure 2 is a detailed flow chart of an embodiment of the present invention;

[0047] Figure 3 Schematic diagram of the data collection process for mining;

[0048] Figure 4 It is a schematic diagram of different states of interaction between bucket and material;

[0049] Figure 5 4 is a system structure diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0050] See also Figure 1 and Figure 2The method step diagram and detailed flow chart of the embodiment of the present invention are shown, which include the following steps:

[0051] S101, obtaining actual excavation data of an excavator in a process of excavating materials, wherein the actual excavation data includes material parameters, bucket parameters, and excavation trajectory;

[0052] S102, preprocessing the actual mining data to obtain mining parameters and actual mining resistance;

[0053] S103, establishing a material and bucket interaction model, and calculating theoretical excavation resistance according to input material parameters, bucket parameters, and excavation parameters;

[0054] S104, constructing an excavation resistance prediction model, wherein the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters as inputs and outputs predicted resistance;

[0055] S105, constructing a loss function based on the theoretical mining resistance and the actual mining resistance, and training a mining resistance prediction model;

[0056] S106, using the trained excavation resistance prediction model to implement excavation resistance prediction.

[0057] Specifically, S101 uses on-board sensors and data acquisition systems to obtain data such as the surface morphology of materials, the digging trajectory of the bucket tip, and the actual digging resistance during the excavation process.

[0058] For details, see Excavation Device Figure 2 As shown, it includes an excavator 9, a binocular camera 6, displacement sensors 4, 5, 7 and pin force sensors 2, 3. The displacement data of each cylinder of the working device during the excavation process of the excavator are obtained by using the displacement sensors 4, 5, 7; the pin at the bucket is replaced with the pin sensors 2, 3, and the actual excavation resistance data during the excavation process is obtained by using the bucket dynamics model; the empirical data is sorted out, the fuzzy relationship between the material type and the material attribute is constructed, and the material attribute parameter P is obtained. m , including the friction angle δ between the bucket and the material, the internal friction angle φ of the material, and the cohesion coefficient c of the material, that is, P m =(δ,φ,c); obtain the excavator bucket parameter P through actual measurement b , including bucket width w, bucket bottom length l b , angle between bucket bottom plate and side ε, P b =(w,l b ,ε); During the data collection process, mining operations are performed on different materials 1 along different mining trajectories 8 to obtain rich mining process data.

[0059] Specifically, S102 preprocesses the collected data to obtain the mining process parameter Pe and actual excavation resistance, specifically:

[0060] The coordinate system of the excavator is constructed based on the DH coordinate system, and the base coordinate system is established at the rotation center of the excavator and the ground contact position; the position of the bucket tip in the base coordinate system and the bucket posture are solved based on the collected cylinder displacement data and the kinematic model of the excavator working device; the material surface morphology point cloud is converted to the base coordinate system, and the surface morphology function of the material is fitted using a polynomial function; the surface morphology function is differentiated to obtain the material accumulation angle α at different positions, and the excavation depth d and the excavation front angle β are solved based on the material surface morphology function, the bucket tip position and the bucket posture; the accumulated soil volume V during the excavation process is obtained by integrating the material area swept by the bucket g ; Therefore, the mining process parameter is P e =(d,β,V g ,α);

[0061] According to the bucket dynamic model, the actual operating resistance F during the excavation process is calculated by the pin sensor force r .

[0062] Specifically, S103 establishes a material-bucket interaction mechanism model by analyzing the interaction state between the bucket and the material during the excavation process. According to the different interaction postures between the bucket and the material, the excavation process is divided into four stages, such as Figure 4 As shown in the figure, they are: the first stage is when the bucket is not inserted with materials, such as Figure 4 As shown in ①; the second stage is when the bucket bottom plate is inserted into the material, and the bucket rear plate does not contact the material, such as Figure 4 As shown in ②; the third stage is when the bucket bottom plate is completely inserted into the material and the material contacts the bucket rear plate, as shown in Figure 4 As shown in ③; the fourth stage is when the bucket tip leaves the material and the excavation is completed, as shown in Figure 4 As shown in ④; and analyze the interaction between the material and the bucket in the four stages to establish the parameters (P e ,P m ,P b ) and digging resistance O m The relationship between materials and buckets is constructed. m =f m (P e ,P m ,P b ).

[0063] Specifically, in S104 and S105, a digital-analog hybrid excavator excavation resistance prediction model is constructed and trained, and the mechanism model is combined with the machine learning algorithm to ensure the prediction accuracy while improving its generalization ability. Specifically:

[0064] The input value of the mechanism model (P e ,P m ,P b ) as input value, the actual excavation resistance F r As the output value, construct the data set D = (P e ,P m ,P b ; F r );

[0065] Using machine learning algorithms and combining mechanism models to build a mining resistance prediction model d =f d (P e ,P m ,P b ), the loss function of the designed model is Loss = L1(O d ,F r )+L2(O d ,O m ), the loss function consists of two parts, the first part L1(O d ,F r ) is the variance between the output value of the machine learning model and the actual mining resistance. The second part L2(O d ,O m ) is the variance of the output value of the machine learning model and the mining resistance calculated by the mechanism model;

[0066] During the model training process, the model parameters are adjusted through continuous training so that the loss function reaches the minimum value, thereby obtaining the optimal mining resistance prediction model.

[0067] Specifically, in S106, given the excavation trajectory, material attribute information and bucket parameters, the input parameter input of the excavation resistance prediction model is calculated according to the kinematic model of the excavator working device. e ,P m ,P b ), and input it into the prediction model f d (input), you can get the excavation resistance O on the specified excavation trajectory p .

[0068] See also Figure 5 As shown, it is a system structure diagram of an embodiment of the present invention, including:

[0069] The data acquisition module 501 acquires actual excavation data of the excavator during the excavation of materials, wherein the actual excavation data includes material parameters, bucket parameters and excavation trajectory;

[0070] The data preprocessing module 502 preprocesses the actual mining data to obtain mining parameters and actual mining resistance;

[0071] Theoretical value calculation module 503 establishes a material and bucket interaction model and calculates theoretical excavation resistance according to input material parameters, bucket parameters and excavation parameters;

[0072] A prediction model building module 504 builds an excavation resistance prediction model, wherein the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters as inputs and outputs predicted resistance;

[0073] The prediction model training module 505 constructs a loss function based on the theoretical mining resistance and the actual mining resistance to train the mining resistance prediction model;

[0074] The prediction model application module 506 implements excavation resistance prediction using the trained excavation resistance prediction model.

[0075] It can be seen that the present invention constructs an excavation resistance prediction model that can accurately predict the excavation resistance and ensure the generalization ability of the model. The excavation resistance on the trajectory can be calculated in advance according to the given excavation conditions, laying the foundation for trajectory planning during the excavation process of the excavator.

[0076] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting excavation resistance of a digital-analog hybrid excavator, characterized in that: The following steps are involved: Acquire actual excavation data of the excavator during the excavation of materials, wherein the actual excavation data includes material parameters, bucket parameters and excavation trajectory; Preprocess the actual mining data to obtain mining parameters and actual mining resistance; Establish a material-bucket interaction model and calculate the theoretical excavation resistance based on the input material parameters, bucket parameters and excavation parameters; Constructing an excavation resistance prediction model, wherein the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters as inputs and outputs predicted resistance; The loss function is constructed based on the theoretical mining resistance and the actual mining resistance, and the mining resistance prediction model is trained; The trained excavation resistance prediction model is used to realize excavation resistance prediction.

2. The method for predicting excavation resistance of a digital-analog hybrid excavator according to claim 1, characterized in that: The actual excavation data of the excavator during the material excavation process is obtained as follows: Use empirical values ​​as material parameters, including the internal friction angle between the material and the bucket, material density, material internal friction, and material cohesion coefficient; Obtain the excavator bucket parameters through actual measurement, including bucket width, bucket bottom plate length, and the angle ε between the bucket bottom plate and the side edge; The wire-pull displacement sensor is used to obtain the excavation trajectory, including the boom cylinder displacement, dipper arm cylinder displacement and bucket displacement during the excavation process.

3. The method for predicting excavation resistance of a digital-analog hybrid excavator according to claim 2, characterized in that: The actual mining data also includes: The surface morphology point cloud of the material to be excavated is obtained by using a binocular camera; The pin force is obtained using the pin force sensor.

4. The method for predicting excavation resistance of a digital-analog hybrid excavator according to claim 3 is characterized in that: The preprocessing of the actual mining data includes: The pin force obtained by the pin force sensor is combined with the bucket dynamics model to obtain the actual digging resistance; A kinematic model of the excavator working device is established, and a base coordinate system is established with the excavator's rotation center and the ground contact position. The surface morphology point cloud of the material to be excavated based on the binocular camera coordinate system is converted to the base coordinate system, and the fitting function of the material surface is obtained by polynomial fitting. The excavation parameters are obtained according to the fitting function of the material surface.

5. The method for predicting excavation resistance of a digital-analog hybrid excavator according to claim 4 is characterized in that: The excavation parameters are obtained according to the fitting function of the material surface, and the excavation parameters include material accumulation angle, excavation depth, and excavation front angle. The acquisition process is specifically as follows: The position of the bucket tip and the bucket posture in the base coordinate system are solved according to the displacement of the boom cylinder, the arm cylinder and the bucket; Differentiate the fitting function of the material surface to obtain the material accumulation angle at different positions; The excavation depth and excavation front angle are solved according to the fitting function of the material surface, the position of the bucket tip and the bucket posture.

6. The method for predicting excavation resistance of a digital-analog hybrid excavator according to claim 1, characterized in that: The material-bucket interaction model considering different interaction states between the material and the bucket is established as follows: The interaction between the bucket and the material during the excavation process is analyzed in stages to establish the relationship between the input parameters and the theoretical excavation resistance; The excavation process is divided into four stages, including: the first stage, when the bucket is not inserted into the material, the excavation resistance is zero; In the second stage, the bucket bottom plate is inserted into the material, and the bucket rear plate does not contact the material; in the third stage, the bucket bottom plate is completely inserted into the material, and the material is in contact with the bucket rear plate; in the fourth stage, the bucket tip leaves the material and the excavation is completed.

7. The method for predicting excavation resistance of a digital-analog hybrid excavator according to claim 4, characterized in that: The training mining resistance prediction model is as follows: The data set is constructed by taking material parameters, bucket parameters and excavation parameters as input and actual excavation resistance as label; Design the loss function of the excavation resistance prediction model. The loss function consists of two parts. The first part is the variance between the output value of the machine learning model and the actual excavation resistance. The second part is the variance between the output value of the machine learning model and the excavation resistance calculated by the mechanism model. The model is trained using the data set. During the training process, the model parameters are continuously adjusted to minimize the loss function, thereby obtaining the optimal mining resistance prediction model.

8. The method for predicting excavation resistance of an excavator with digital-analog hybrid drive according to claim 1, characterized in that: The mining resistance prediction is realized by using the trained mining resistance prediction model, specifically: Given excavation conditions, including excavation trajectory, material information of the material to be excavated, and bucket parameters, the material information includes the material surface morphology point cloud and material parameters; Pre-process the excavation trajectory and material information to obtain the excavation parameters; The material parameters, bucket parameters and excavation parameters are input into the trained excavation resistance prediction model to predict the excavation resistance for given excavation conditions.

9. A digital-analog hybrid excavator excavation resistance prediction system, characterized in that: include: A data acquisition module, which acquires actual excavation data of the excavator during the excavation of materials, wherein the actual excavation data includes material parameters, bucket parameters and excavation trajectory; Data preprocessing module, which preprocesses the actual mining data to obtain mining parameters and actual mining resistance; Theoretical value calculation module, which establishes the interaction model between materials and buckets, and calculates the theoretical excavation resistance according to the input material parameters, bucket parameters and excavation parameters; A prediction model building module is used to build an excavation resistance prediction model, wherein the excavation resistance prediction model takes material parameters, bucket parameters and excavation parameters as inputs and outputs predicted resistance; The prediction model training module constructs a loss function based on theoretical mining resistance and actual mining resistance to train the mining resistance prediction model; The prediction model application module uses the trained excavation resistance prediction model to realize excavation resistance prediction.

Citation Information

Cited By

  • Bucket tooth tip trajectory planning method based on intelligent detection of electric shovel bucket material volume

    CN121381711A