A high-precision positioning automatic control method and system

Abstract

This invention discloses a high-precision positioning automatic control method and system, comprising: receiving a work task sent by a user terminal; acquiring historical data of a grader and generating a machine profile of the grader based on the historical data; acquiring environmental data perceived by the grader's sensing module; generating a digital three-dimensional design benchmark model of the work task based on the machine profile and environmental data; acquiring the current mechanical blade position and motion posture of the grader when performing the work task; comparing the current mechanical blade position and motion posture with the digital three-dimensional design benchmark model, obtaining the comparison result and sending it to the grader; and automatically controlling the mechanical blade of the grader based on the comparison result. This achieves automatic and precise digital grader operation, while simultaneously monitoring and adjusting the mechanical blade position and motion posture in real time during automated operation, greatly improving construction accuracy and ensuring construction quality.

Description

Technical Field

[0001] This invention relates to the field of control technology, and in particular to a high-precision positioning automatic control method and system. Background Technology

[0002] Graders are the main machinery used for shaping and leveling operations in earthwork engineering. Traditional fine leveling of earthwork requires surveyors to measure the original ground surface in advance, calculate the loose paving thickness based on the original ground elevation, and manually mark and lay out lines. The construction process requires multiple rounds of compaction, measurement, and fine leveling, resulting in high costs for both manual and mechanical inputs and poor results, making it extremely difficult to meet the elevation requirements of +10mm and -20mm in the affected area of ​​airport pavements. Existing graders cannot operate automatically according to user instructions, and during automated operation, they cannot monitor the position and movement of the mechanical blades in real time and make timely adjustments, resulting in the inability to achieve fine leveling operations and low construction accuracy and quality. Summary of the Invention

[0003] This invention aims to at least partially solve one of the technical problems in the aforementioned technologies. Therefore, the first objective of this invention is to propose a high-precision positioning automatic control method to achieve automated precision leveling digital construction operations using a grader. Simultaneously, during automated operation, the position and movement posture of the mechanical blade are monitored in real time and adjusted promptly. Employing automatic blade control technology, the grader blade automatically performs precision leveling operations according to a design benchmark model. Only 1-2 passes are needed to achieve an accuracy within ±10mm, greatly improving construction precision and ensuring construction quality.

[0004] The second objective of this invention is to provide a high-precision positioning automatic control system.

[0005] To achieve the above objectives, a first aspect of the present invention provides a high-precision positioning automatic control method, comprising:

[0006] Receive work tasks sent by the user client;

[0007] Acquire historical data of the grader and generate a machine profile of the grader based on the historical data;

[0008] Acquire environmental data sensed by the sensor module based on the grader;

[0009] A digital 3D design baseline model of the work task is generated based on the equipment profile and environmental data;

[0010] Obtain the current position and movement posture of the mechanical blades of the grader when it is performing a work task;

[0011] The current position and motion posture of the mechanical blade are compared with the digital 3D design benchmark model, the comparison results are obtained and sent to the grader;

[0012] The grader automatically controls the current mechanical blade based on the comparison results.

[0013] According to some embodiments of the present invention, acquiring historical data of a motor grader and generating a device profile of the motor grader based on the historical data includes:

[0014] Data analysis is performed on historical data to identify the relationships between the data.

[0015] Based on the relationships between the data, a profile of the grader is generated.

[0016] According to some embodiments of the present invention, data analysis is performed on historical data to obtain the correlation between data, including:

[0017] Historical data is preprocessed and transformed into document data; wherein the document data contains multiple sentences;

[0018] The document data is context-encoded based on a pre-trained language coding model to determine the position information of each sentence in the document data;

[0019] Based on a pre-trained entity extraction model, entities are extracted from each sentence to determine their location information within the sentence;

[0020] An initial association graph is constructed based on the position information of each sentence in the document data and the position information of entities in the sentences; the initial association graph includes the initial representation of entity nodes;

[0021] Based on a pre-trained semantic recognition model, semantic recognition is performed on the entity nodes in the initial association graph. The semantic recognition results are then labeled on the entity nodes to obtain semantic entity nodes. The semantic entity nodes are then divided into a first part of semantic entity nodes and a second part of semantic entity nodes.

[0022] The first part of semantic entity nodes is predicted based on the feedforward neural network model, and the predicted association relationship labels between the first part of semantic entity nodes are determined.

[0023] Calculate the loss value between the predicted association relationship label between the semantic entity nodes in the first part and the association relationship label in the first part. Based on the loss value and the backpropagation algorithm, iteratively optimize the model parameters of the feedforward neural network model and determine the corrected feedforward neural network model.

[0024] The semantic entity nodes of the second part are identified based on the modified feedforward neural network model to obtain the association labels of the second part.

[0025] Based on the first part of the relationship labels and the second part of the relationship labels, the relationships between the data are obtained.

[0026] According to some embodiments of the present invention, acquiring environmental data sensed by the sensing module based on a grader includes:

[0027] The three-dimensional data of the environment in which the grader is located is perceived by the sensing module; the three-dimensional data includes location information, obstacle information and height information.

[0028] According to some embodiments of the present invention, a digital three-dimensional design baseline model for a work task is generated based on equipment profiles and environmental data, including:

[0029] A 3D model is generated based on the 3D data of the environment in which the grader is located, and feasible paths are planned for the work tasks on the 3D model to obtain a set of feasible paths.

[0030] Based on the equipment profile, the feasible path set is filtered and optimized to determine the target execution path. Based on the target execution path, a digital 3D design benchmark model of the work task is generated.

[0031] According to some embodiments of the present invention, feasible paths are planned for work tasks on a three-dimensional model diagram to obtain a set of feasible paths, including:

[0032] Information about obstacles is output based on a convolutional neural network;

[0033] Mark the start and end positions of the work tasks on the 3D model diagram; construct a path mesh based on the obstacle information, start and end positions to obtain a set of feasible paths.

[0034] According to some embodiments of the present invention, after the grader automatically controls the current mechanical blade based on the comparison result, the method further includes:

[0035] Acquire construction data, monitor whether construction quality indicators meet design and specification requirements based on the construction data, and issue an alarm when it is determined that the construction quality indicators do not meet the design and specification requirements.

[0036] According to some embodiments of the present invention, the grader automatically controls the current mechanical blade based on comparison results, including:

[0037] The grader automatically performs corresponding hydraulic closed-loop control and automatic control to adjust the blade posture based on the comparison results.

[0038] According to some embodiments of the present invention, it further includes:

[0039] When the grader is performing its work tasks, acquire the grader's operating status data;

[0040] The system analyzes operational status data based on the operational status database and issues an alarm when an anomaly is detected, according to the analysis results.

[0041] The operational status data is analyzed based on the operational status database, including:

[0042] The operational status database contains m distinct data entries, and each data entry includes the values ​​of n indicators used to evaluate the operational status data, generating matrix B; at the same time, the operational status database contains information on whether each data entry is abnormal, generating matrix C;

[0043] The matrix B is balanced to obtain the corrected matrix;

[0044]

[0045] Where, bip is the value in the i-th row and p-th column of matrix B, which is the value of the p-th indicator of the i-th data; bb ip The value of matrix B after equalization processing of the value in the i-th row and p-th column; i = 1, 2, 3...m; p = 1, 2, 3...n; Based on the processing of each value in matrix B, the corrected matrix W is obtained;

[0046] Calculate the weighting coefficient for each indicator in the data:

[0047]

[0048] Where bbip is the value in the i-th row and p-th column of the correction matrix W; bb tp To correct the value in the t-th row and p-th column of matrix W; wp is the weight coefficient of the p-th indicator in the data, i = 1, 2, 3...m; t = 1, 2, 3...m; p = 1, 2, 3...n, ln() is the logarithm of the logarithm to the base e;

[0049] Calculate the feature value for each data point:

[0050]

[0051] Where Fi is the feature value of the i-th data; i = 1, 2, 3...m;

[0052] Based on the values ​​of n indicators used to evaluate the operational status data and the weight coefficient of each indicator in the data, the characteristic value of the operational status data is calculated.

[0053] Calculate the absolute value of the difference between the feature value of the running status data and the feature value of each data point in the running status database. Select the data in the running status database with the smallest absolute value of the difference as the matching data. Determine the value in matrix C corresponding to the matching data as the analysis result.

[0054] To achieve the above objectives, a second aspect of the present invention provides a high-precision positioning automatic control system, comprising:

[0055] The receiving module is used to receive work tasks sent by the user terminal.

[0056] The first acquisition module is used to acquire historical data of the grader and generate a device profile of the grader based on the historical data.

[0057] The second acquisition module is used to acquire environmental data perceived by the sensor module based on the grader;

[0058] The generation module is used to generate a digital 3D design baseline model of the work task based on the equipment profile and environmental data;

[0059] The third acquisition module is used to acquire the current position and movement posture of the mechanical blade when the grader is performing a work task;

[0060] The comparison module is used to compare the current position and motion posture of the mechanical blade with the digital 3D design baseline model, obtain the comparison results, and send them to the grader.

[0061] The grader automatically controls the current mechanical blade based on the comparison results.

[0062] This invention proposes a high-precision positioning automatic control method and system to realize automated precision leveling digital construction operations using a grader. Simultaneously, during automated operation, the system monitors and adjusts the position and movement of the mechanical blade in real time. Employing automatic blade control technology, the grader blade automatically performs precision leveling according to the design benchmark model. Only 1-2 passes are needed to achieve an accuracy within ±10mm, significantly improving construction precision and ensuring construction quality. Furthermore, adjustments are made based on the equipment image when generating the digital 3D design benchmark model, further enhancing the accuracy of the resulting model.

[0063] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0064] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0065] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0066] Figure 1 This is a flowchart of a high-precision positioning automatic control method according to an embodiment of the present invention;

[0067] Figure 2 This is a schematic diagram of a grader according to an embodiment of the present invention;

[0068] Figure 3 This is a block diagram of a high-precision positioning automatic control system according to an embodiment of the present invention. Detailed Implementation

[0069] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0070] like Figure 1 As shown, the first aspect of the present invention proposes a high-precision positioning automatic control method, including steps S1-S7:

[0071] S1. Receive work tasks sent by the user terminal;

[0072] S2. Obtain historical data of the grader and generate a device profile of the grader based on the historical data;

[0073] S3. Obtain environmental data perceived by the sensor module based on the grader;

[0074] S4. Generate a digital 3D design baseline model of the work task based on the equipment profile and environmental data;

[0075] S5. Obtain the current position and movement posture of the mechanical blade when the grader is performing a work task;

[0076] S6. Compare the current position and motion posture of the mechanical blade with the digital three-dimensional design benchmark model, obtain the comparison results, and send them to the grader;

[0077] S7. The grader automatically controls the current mechanical blade based on the comparison results.

[0078] The working principle of the above technical solution: In this embodiment, the user terminal can be a mobile phone, computer, etc., which makes it convenient for users to assign work tasks to the grader and monitor it in real time.

[0079] In this embodiment, based on the historical data of the grader, a machine profile of the grader is generated; this facilitates the data integration of the overall equipment status, such as the working accuracy of the equipment, the various adjustment accuracy and adjustment methods of the mechanical blades, the responsiveness of the equipment, and other basic information about the equipment.

[0080] like Figure 2 As shown, in this embodiment, the grader includes an antenna column and a GNSS antenna for signal and data transmission. An attitude sensor detects the current position and movement attitude of the mechanical blade; it also includes a display terminal, a main control unit, etc. The grader's sensing module includes a radar module and a camera module for sensing environmental data.

[0081] In this embodiment, existing technologies can generate a digital 3D design baseline model for a work task using environmental data. However, they cannot adjust and optimize this model based on the equipment's own information to better suit the equipment's operation. Adjusting the model based on the equipment profile improves the accuracy of the resulting digital 3D design baseline model and, consequently, the accuracy of the equipment's operation.

[0082] In this embodiment, the current position and motion posture of the mechanical blade of the grader are acquired when it performs a work task; the current position and motion posture of the mechanical blade are compared with the digital three-dimensional design reference model, and the comparison result is sent to the grader; the grader automatically controls the current mechanical blade according to the comparison result. This achieves automatic control of the current mechanical blade, facilitating precision leveling operations and improving work efficiency and leveling quality.

[0083] The beneficial effects of the above technical solution are as follows: It enables automated precision leveling digital construction operations using graders, while simultaneously monitoring and adjusting the position and movement of the mechanical blades in real time during automated operation. Employing automatic blade control technology, the grader blades automatically perform precision leveling according to the design benchmark model, requiring only 1-2 passes to achieve an accuracy within ±10mm, significantly improving construction precision and ensuring construction quality. Furthermore, adjustments are made based on the equipment image when generating the digital 3D design benchmark model, improving the accuracy of the resulting model.

[0084] According to some embodiments of the present invention, acquiring historical data of a motor grader and generating a device profile of the motor grader based on the historical data includes:

[0085] Data analysis is performed on historical data to identify the relationships between the data.

[0086] Based on the relationships between the data, a profile of the grader is generated.

[0087] The working principle and beneficial effects of the above technical solution are as follows: Data analysis is performed on historical data to obtain the correlations between data points. Based on these correlations, the data is organized and analyzed to create a profile of the grader. Equipment profiling refers to the in-depth analysis and description of the equipment using various data, information, and data relationships, thereby generating a comprehensive image of the equipment. Equipment profiling can include basic equipment information (such as model, production date, power, etc.), working accuracy, adjustment accuracy, execution logic, network response, and other aspects. Through equipment profiling, the equipment can be operated more effectively, providing a responsive execution model for the equipment.

[0088] According to some embodiments of the present invention, data analysis is performed on historical data to obtain the correlation between data, including:

[0089] Historical data is preprocessed and transformed into document data; wherein the document data contains multiple sentences;

[0090] The document data is context-encoded based on a pre-trained language coding model to determine the position information of each sentence in the document data;

[0091] Based on a pre-trained entity extraction model, entities are extracted from each sentence to determine their location information within the sentence;

[0092] An initial association graph is constructed based on the position information of each sentence in the document data and the position information of entities in the sentences; the initial association graph includes the initial representation of entity nodes;

[0093] Based on a pre-trained semantic recognition model, semantic recognition is performed on the entity nodes in the initial association graph. The semantic recognition results are then labeled on the entity nodes to obtain semantic entity nodes. The semantic entity nodes are then divided into a first part of semantic entity nodes and a second part of semantic entity nodes.

[0094] The first part of semantic entity nodes is predicted based on the feedforward neural network model, and the predicted association relationship labels between the first part of semantic entity nodes are determined.

[0095] Calculate the loss value between the predicted association relationship label between the semantic entity nodes in the first part and the association relationship label in the first part. Based on the loss value and the backpropagation algorithm, iteratively optimize the model parameters of the feedforward neural network model and determine the corrected feedforward neural network model.

[0096] The semantic entity nodes of the second part are identified based on the modified feedforward neural network model to obtain the association labels of the second part.

[0097] Based on the first part of the relationship labels and the second part of the relationship labels, the relationships between the data are obtained.

[0098] The working principle of the above technical solution is as follows: In this embodiment, context encoding is performed on document data based on a pre-trained language encoding model to determine the position information of each sentence in the document data. This includes using BERT (Bidirectional Encoder Representations from Transformers) or other pre-trained Transformer models. These models have already learned sentence-level and document-level semantic representations during training and can handle context information well. Determining the position information of each sentence in the document data includes: Sentence segmentation: The document data is segmented into sentences according to sentence termination symbols (such as periods, question marks, exclamation marks, etc.). Serialization: Each sentence is converted into an input format acceptable to the model, usually by converting the sentence into a sequence of word vectors and adding special symbols (such as [CLS] and [SEP]) for labeling. Context encoding: The serialized sentences are input into the pre-trained language encoding model to obtain the semantic representation of the sentences in the document data. Obtaining positional information: By using the hidden states or other information output by the model, the positional information of a sentence within the document data can be obtained, such as the sentence's start position, end position, and relative position. This allows for the determination of the positional information of each sentence in the document data.

[0099] In this embodiment, entities in each sentence are extracted based on a pre-trained entity extraction model to determine the entity's position within the sentence. The entity extraction model includes rule-based models and machine learning-based models (such as CRF). After entity extraction is complete, the entity's position within the sentence can be determined using the model's output. The entity's position can be represented by its start and end positions within the sentence, for example, by specifying the start and end character indices of the entity.

[0100] In this embodiment, an initial association graph is constructed based on the position information of each sentence in the document data and the position information of the entity in the sentence; the initial association graph contains the initial representation of the entity node; the entity node in the initial association graph is semantically recognized based on a pre-trained semantic recognition model to realize the semantic update of the entity node and obtain semantic entity node.

[0101] In one embodiment, the number of the first part of semantic entity nodes and the number of the second part of semantic entity nodes are the same.

[0102] In this embodiment, the first part of semantic entity nodes is predicted based on a feedforward neural network model to determine the predicted association labels between the first part of semantic entity nodes; the loss value between the predicted association labels between the first part of semantic entity nodes and the first part of the association labels is calculated; the model parameters of the feedforward neural network model are iteratively optimized based on the loss value and the backpropagation algorithm to determine the corrected feedforward neural network model; the second part of semantic entity nodes is identified based on the corrected feedforward neural network model to obtain the second part of the association labels; and the association relationship between the data is obtained based on the first part of the association labels and the second part of the association labels.

[0103] The beneficial effects of the above technical solution are as follows: the data is processed into nodes, and an initial association graph is constructed based on the position information of each sentence in the document data and the position information of entities in the sentences; the entity nodes in the initial association graph are semantically recognized based on a pre-trained semantic recognition model to obtain semantic entity nodes; the feedforward neural network model is updated based on the first part of semantic entity nodes to improve the accuracy of the obtained modified feedforward neural network model, thereby accurately determining the second part of the association label, and thus accurately determining the association relationship between the data.

[0104] According to some embodiments of the present invention, acquiring environmental data sensed by the sensing module based on a grader includes:

[0105] The three-dimensional data of the environment in which the grader is located is perceived by the sensing module; the three-dimensional data includes location information, obstacle information and height information.

[0106] According to some embodiments of the present invention, a digital three-dimensional design baseline model for a work task is generated based on equipment profiles and environmental data, including:

[0107] A 3D model is generated based on the 3D data of the environment in which the grader is located, and feasible paths are planned for the work tasks on the 3D model to obtain a set of feasible paths.

[0108] Based on the equipment profile, the feasible path set is filtered and optimized to determine the target execution path. Based on the target execution path, a digital 3D design benchmark model of the work task is generated.

[0109] The working principle and beneficial effects of the above technical solution are as follows: A 3D model is generated based on the 3D data of the grader's environment, and feasible paths are planned for the work tasks on the 3D model to obtain a set of feasible paths. The set of feasible paths is then filtered and optimized based on the equipment profile, including selecting suitable paths based on equipment characteristics and fine-tuning these suitable paths based on equipment characteristics to obtain the target execution path. Based on the target execution path, a digital 3D design benchmark model for the work tasks is generated. This improves the accuracy of obtaining the digital 3D design benchmark model.

[0110] According to some embodiments of the present invention, feasible paths are planned for work tasks on a three-dimensional model diagram to obtain a set of feasible paths, including:

[0111] Information about obstacles is output based on a convolutional neural network;

[0112] Mark the start and end positions of the work tasks on the 3D model diagram; construct a path mesh based on the obstacle information, start and end positions to obtain a set of feasible paths.

[0113] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, obstacle information is output based on a convolutional neural network. The convolutional neural network structure includes multiple convolutional layers, pooling layers, and fully connected layers. A 3D model graph is used as input. Features are extracted and mapped through the filters of the convolutional layers to generate new feature maps. Spatial dimensionality reduction is then performed through the pooling layers to retain key information. The fully connected layers integrate the information and output obstacle information. The start and end positions of the task are marked on the 3D model graph. Based on the obstacle information, start and end positions, a path grid is constructed to accurately obtain a set of feasible paths.

[0114] According to some embodiments of the present invention, after the grader automatically controls the current mechanical blade based on the comparison result, the method further includes:

[0115] Acquire construction data, monitor whether construction quality indicators meet design and specification requirements based on the construction data, and issue an alarm when it is determined that the construction quality indicators do not meet the design and specification requirements.

[0116] The beneficial effects of the above technical solution are: it facilitates timely adjustments to the construction operation and ensures construction quality.

[0117] According to some embodiments of the present invention, the grader automatically controls the current mechanical blade based on comparison results, including:

[0118] The grader automatically performs corresponding hydraulic closed-loop control and automatic control to adjust the blade posture based on the comparison results.

[0119] The working principle and beneficial effects of the above technical solution: The grader automatically performs corresponding hydraulic closed-loop control and automatic control adjustment of the blade posture based on the comparison results, so as to adjust the current position and posture of the mechanical blade to be consistent with the preset parameters in the digital three-dimensional design benchmark model, thereby realizing automated control and operation.

[0120] According to some embodiments of the present invention, it further includes:

[0121] When the grader is performing its work tasks, acquire the grader's operating status data;

[0122] The system analyzes operational status data based on the operational status database and issues an alarm when an anomaly is detected, according to the analysis results.

[0123] The operational status data is analyzed based on the operational status database, including:

[0124] The operational status database contains m distinct data entries, and each data entry includes the values ​​of n indicators used to evaluate the operational status data, generating matrix B; at the same time, the operational status database contains information on whether each data entry is abnormal, generating matrix C;

[0125] The matrix B is balanced to obtain the corrected matrix;

[0126]

[0127] Where, bip is the value in the i-th row and p-th column of matrix B, which is the value of the p-th indicator of the i-th data; bb ip The value of matrix B after equalization processing of the value in the i-th row and p-th column; i = 1, 2, 3...m; p = 1, 2, 3...n; Based on the processing of each value in matrix B, the corrected matrix W is obtained;

[0128] Calculate the weighting coefficient for each indicator in the data:

[0129]

[0130] Where bbip is the value in the i-th row and p-th column of the correction matrix W; bb tp To correct the value in the t-th row and p-th column of matrix W; wp is the weight coefficient of the p-th indicator in the data, i = 1, 2, 3...m; t = 1, 2, 3...m; p = 1, 2, 3...n, ln() is the logarithm of the logarithm to the base e;

[0131] Calculate the feature value for each data point:

[0132]

[0133] Where Fi is the feature value of the i-th data; i = 1, 2, 3...m;

[0134] Based on the values ​​of n indicators used to evaluate the operational status data and the weight coefficient of each indicator in the data, the characteristic value of the operational status data is calculated.

[0135] Calculate the absolute value of the difference between the feature value of the running status data and the feature value of each data point in the running status database. Select the data in the running status database with the smallest absolute value of the difference as the matching data. Determine the value in matrix C corresponding to the matching data as the analysis result.

[0136] The working principle and beneficial effects of the above technical solution are as follows: When the grader performs its work tasks, it acquires the grader's operating status data; the operating status data is analyzed based on the operating status database; and an alarm is issued when an anomaly is detected, based on the analysis results. This facilitates monitoring of the grader's operating status, determining the presence of anomalies, enabling timely handling, and improving the control reliability of automated operations. The analysis of the operating status data based on the operating status database includes: performing matrix processing on m different data points in the operating status database; simultaneously, performing equalization processing on matrix B to obtain a correction matrix, which reduces computational complexity; reducing differences; calculating the weight coefficient of each indicator in the data; and accurately calculating the characteristic value of each data point. The absolute value of the difference between the characteristic value of the operating status data and the characteristic value of each data point in the operating status database is calculated; the data in the operating status database with the smallest absolute difference is used as the matching data; the value in matrix C corresponding to the matching data is determined as the analysis result. This facilitates accurate analysis of the operating status data, accurately determines the presence of anomalies, and improves the reliability of system control.

[0137] like Figure 3 As shown, a second aspect of the present invention provides a high-precision positioning automatic control system, comprising:

[0138] The receiving module is used to receive work tasks sent by the user terminal.

[0139] The first acquisition module is used to acquire historical data of the grader and generate a device profile of the grader based on the historical data.

[0140] The second acquisition module is used to acquire environmental data perceived by the sensor module based on the grader;

[0141] The generation module is used to generate a digital 3D design baseline model of the work task based on the equipment profile and environmental data;

[0142] The third acquisition module is used to acquire the current position and movement posture of the mechanical blade when the grader is performing a work task;

[0143] The comparison module is used to compare the current position and motion posture of the mechanical blade with the digital 3D design baseline model, obtain the comparison results, and send them to the grader.

[0144] The grader automatically controls the current mechanical blade based on the comparison results.

[0145] The beneficial effects of the above technical solution are as follows: It enables automated precision leveling digital construction operations using graders, while simultaneously monitoring and adjusting the position and movement of the mechanical blades in real time during automated operation. Employing automatic blade control technology, the grader blades automatically perform precision leveling according to the design benchmark model, requiring only 1-2 passes to achieve an accuracy within ±10mm, significantly improving construction precision and ensuring construction quality. Furthermore, adjustments are made based on the equipment image when generating the digital 3D design benchmark model, improving the accuracy of the resulting model.

[0146] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A high-precision positioning automatic control method, characterized in that, include: Receive work tasks sent by the user client; Acquire historical data of the grader and generate a machine profile of the grader based on the historical data; Acquire environmental data sensed by the sensor module based on the grader; A digital 3D design baseline model of the work task is generated based on the equipment profile and environmental data; Obtain the current position and movement posture of the mechanical blades of the grader when it is performing a work task; The current position and motion posture of the mechanical blade are compared with the digital 3D design benchmark model, the comparison results are obtained and sent to the grader; The grader automatically controls the current mechanical blade based on the comparison results; Obtain historical data of the grader and generate a machine profile of the grader based on the historical data, including: Data analysis is performed on historical data to identify the relationships between the data. Based on the relationships between the data, generate a profile of the grader. Data analysis of historical data reveals the relationships between data points, including: Historical data is preprocessed and transformed into document data; wherein the document data contains multiple sentences; The document data is context-encoded based on a pre-trained language coding model to determine the position information of each sentence in the document data; Based on a pre-trained entity extraction model, entities are extracted from each sentence to determine their location information within the sentence; An initial association graph is constructed based on the position information of each sentence in the document data and the position information of entities in the sentences; the initial association graph includes the initial representation of entity nodes; Based on a pre-trained semantic recognition model, semantic recognition is performed on the entity nodes in the initial association graph. The semantic recognition results are then labeled on the entity nodes to obtain semantic entity nodes. The semantic entity nodes are then divided into a first part of semantic entity nodes and a second part of semantic entity nodes. The first part of semantic entity nodes is predicted based on the feedforward neural network model, and the predicted association relationship labels between the first part of semantic entity nodes are determined. Calculate the loss value between the predicted association relationship label between the semantic entity nodes in the first part and the association relationship label in the first part. Based on the loss value and the backpropagation algorithm, iteratively optimize the model parameters of the feedforward neural network model and determine the corrected feedforward neural network model. The semantic entity nodes of the second part are identified based on the modified feedforward neural network model to obtain the association labels of the second part. Based on the first part of the relationship labels and the second part of the relationship labels, the relationships between the data are obtained.

2. The high-precision positioning automatic control method as described in claim 1, characterized in that, Acquire environmental data sensed by the grader's sensing module, including: The three-dimensional data of the environment in which the grader is located is perceived by the sensing module; the three-dimensional data includes location information, obstacle information and height information.

3. The high-precision positioning automatic control method as described in claim 1, characterized in that, Based on equipment profiles and environmental data, a digital 3D design baseline model for the work task is generated, including: A 3D model is generated based on the 3D data of the environment in which the grader is located, and feasible paths are planned for the work tasks on the 3D model to obtain a set of feasible paths. Based on the equipment profile, the feasible path set is filtered and optimized to determine the target execution path. Based on the target execution path, a digital 3D design benchmark model of the work task is generated.

4. The high-precision positioning automatic control method as described in claim 3, characterized in that, On the 3D model diagram, feasible paths are planned for the work tasks, resulting in a set of feasible paths, including: Information about obstacles is output based on a convolutional neural network; Mark the start and end positions of the work tasks on the 3D model diagram; construct a path mesh based on the obstacle information, start and end positions to obtain a set of feasible paths.

5. The high-precision positioning automatic control method as described in claim 1, characterized in that, After the grader automatically controls the current mechanical blade based on the comparison results, it also includes: Acquire construction data, monitor whether construction quality indicators meet design and specification requirements based on the construction data, and issue an alarm when it is determined that the construction quality indicators do not meet the design and specification requirements.

6. The high-precision positioning automatic control method as described in claim 1, characterized in that, The grader automatically controls the current mechanical blade based on the comparison results, including: The grader automatically performs corresponding hydraulic closed-loop control and automatic control to adjust the blade posture based on the comparison results.

7. The high-precision positioning automatic control method as described in claim 1, characterized in that, Also includes: When the grader is performing its work tasks, acquire the grader's operating status data; The system analyzes operational status data based on the operational status database and issues an alarm when an anomaly is detected, according to the analysis results. The operational status data is analyzed based on the operational status database, including: The operational status database contains m distinct data entries, and each data entry includes the values ​​of n indicators used to evaluate the operational status data, generating matrix B; at the same time, the operational status database contains information on whether each data entry is abnormal, generating matrix C; The matrix B is balanced to obtain the corrected matrix; ； in, This is the value in the i-th row and p-th column of matrix B, which is the value of the p-th indicator of the i-th data point. The value of matrix B after equalization processing of the value in the i-th row and p-th column; i = 1, 2, 3...m; p = 1, 2, 3...n; Based on processing each value in matrix B, the corrected matrix W is obtained; Calculate the weighting coefficient for each indicator in the data: ； in, To correct the value in the i-th row and p-th column of matrix W; To correct the value in the t-th row and p-th column of matrix W; Let t = 1, 2, 3...m; t = 1, 2, 3...m; p = 1, 2, 3...n; ln() is the logarithm of logarithm base e; Calculate the feature value for each data point: ； in, Let be the feature value of the i-th data point; i = 1, 2, 3, ..., m; Based on the values ​​of n indicators used to evaluate the operational status data and the weight coefficient of each indicator in the data, the characteristic value of the operational status data is calculated. Calculate the absolute value of the difference between the feature value of the running status data and the feature value of each data point in the running status database. Select the data in the running status database with the smallest absolute value of the difference as the matching data. Determine the value in matrix C corresponding to the matching data as the analysis result.

8. A high-precision positioning automatic control system, characterized in that, include: The receiving module is used to receive work tasks sent by the user terminal. The first acquisition module is used to acquire historical data of the grader and generate a device profile of the grader based on the historical data. The second acquisition module is used to acquire environmental data perceived by the sensor module based on the grader; The generation module is used to generate a digital 3D design baseline model of the work task based on the equipment profile and environmental data; The third acquisition module is used to acquire the current position and movement posture of the mechanical blade when the grader is performing a work task; The comparison module is used to compare the current position and motion posture of the mechanical blade with the digital 3D design baseline model, obtain the comparison results, and send them to the grader. The grader automatically controls the current mechanical blade based on the comparison results; Obtain historical data of the grader and generate a machine profile of the grader based on the historical data, including: Data analysis is performed on historical data to identify the relationships between the data. Based on the relationships between the data, generate a profile of the grader. Data analysis of historical data reveals the relationships between data points, including: Historical data is preprocessed and transformed into document data; wherein the document data contains multiple sentences; The document data is context-encoded based on a pre-trained language coding model to determine the position information of each sentence in the document data; Based on a pre-trained entity extraction model, entities are extracted from each sentence to determine their location information within the sentence; An initial association graph is constructed based on the position information of each sentence in the document data and the position information of entities in the sentences; the initial association graph includes the initial representation of entity nodes; Based on a pre-trained semantic recognition model, semantic recognition is performed on the entity nodes in the initial association graph. The semantic recognition results are then labeled on the entity nodes to obtain semantic entity nodes. The semantic entity nodes are then divided into a first part of semantic entity nodes and a second part of semantic entity nodes. The first part of semantic entity nodes is predicted based on the feedforward neural network model, and the predicted association relationship labels between the first part of semantic entity nodes are determined. Calculate the loss value between the predicted association relationship label between the semantic entity nodes in the first part and the association relationship label in the first part. Based on the loss value and the backpropagation algorithm, iteratively optimize the model parameters of the feedforward neural network model and determine the corrected feedforward neural network model. The semantic entity nodes of the second part are identified based on the modified feedforward neural network model to obtain the association labels of the second part. Based on the first part of the relationship labels and the second part of the relationship labels, the relationships between the data are obtained.

Images