Bucket wheel machine intelligent control system and method based on space positioning

Abstract

The embodiment of the invention relates to the technical field of bucket wheel machine control, and provides a bucket wheel machine intelligent control system and method based on space positioning, a time sequence of positioning data of a fixed antenna installed on a bucket wheel machine is obtained through a GPS positioning system, the positioning data comprise positioning coordinates and course angles, and the positioning coordinates are used for positioning the bucket wheel machine. And carrying out time sequence characteristic analysis and interaction on the positioning data by adopting a data processing and analysis algorithm based on artificial intelligence, so as to intelligently obtain predicted values of a rotation angle and a pitch angle of the cantilever according to semantic reasoning characteristics of positioning coordinates and a course angle on a full time domain, and controlling the cantilever of the bucket wheel machine to work according to the predicted values. Therefore, the position of the bucket wheel machine is accurately monitored, and the operation accuracy and the operation efficiency of the bucket wheel machine are improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of bucket wheel machine control, and in particular to an intelligent control system and method for bucket wheel machine based on spatial positioning. Background Art

[0002] Bucket wheel excavator is a continuous conveying and loading and unloading machine used in large dry bulk cargo yards. It consists of a belt conveyor arm that can pitch and swing horizontally, a bucket wheel at its front end, a frame, and a running mechanism. In modern industry, bucket wheel excavators are widely used in the stacking and unloading operations of bulk material storage yards such as steel mills, coking plants, coal storage plants, and power plants. They are particularly suitable for handling bulk bulk materials such as coal, ore, and chemical raw materials.

[0003] The boom control of bucket wheel machine is crucial to ensure the safety and efficiency of loading and unloading operations. The control of the boom affects the stability of the bucket wheel machine, especially when the material on the conveyor belt increases the weight of the boom. If the tilt angle of the boom is not properly controlled, it may lead to reduced efficiency or even overturning accidents. Therefore, accurate boom control can prevent equipment damage, extend its service life, and reduce maintenance costs.

[0004] At present, the working environment of bucket wheel machine is usually complex, with various obstacles and uncertain factors. However, the traditional bucket wheel machine control method usually uses rule-based or experience-based control methods to control the cantilever of bucket wheel machine. When dealing with complex and changeable working conditions, the control accuracy is limited and it is difficult to achieve precise control. Summary of the invention

[0005] The present disclosure aims to solve at least one of the problems existing in the prior art and to provide an intelligent control system and method for a bucket wheel excavator based on spatial positioning.

[0006] In one aspect of the present disclosure, a bucket wheel machine intelligent control system based on spatial positioning is provided, and the bucket wheel machine intelligent control system comprises:

[0007] A fixed antenna positioning data acquisition module is used to acquire a time series of positioning data of a fixed antenna installed on a bucket wheel machine through a GPS positioning system; wherein the positioning data includes positioning coordinates and heading angles;

[0008] A data division and regularization module, used to sequentially divide and regularize the time series of the positioning data of the fixed antenna to obtain a sequence of local time series input vectors of the corresponding fixed antenna positioning coordinates and a sequence of local time series input vectors of the fixed antenna heading angle;

[0009] A local timing pattern extraction module is used to extract antenna positioning local timing pattern features from the sequence of local timing input vectors of the fixed antenna positioning coordinates and the sequence of local timing input vectors of the fixed antenna heading angles, respectively, to obtain a corresponding sequence of local timing pattern feature vectors of the fixed antenna positioning coordinates and a sequence of local timing pattern feature vectors of the fixed antenna heading angles;

[0010] A feature interaction fusion module is used to interactively fuse each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors in the sequence of the fixed antenna positioning coordinate local time sequence pattern feature vectors and the sequence of the fixed antenna heading angle local time sequence pattern feature vectors, so as to obtain a sequence of fixed antenna positioning coordinate-heading angle local time sequence element interaction feature vectors;

[0011] A positioning semantic reasoning module, used for inputting the sequence of the fixed antenna positioning coordinate-heading angle local time series element interaction feature vectors into a decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors;

[0012] A full-time-domain contextual semantic understanding module is used to perform full-time-domain contextual semantic understanding of the fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vector sequence to obtain a fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature vector as a fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature;

[0013] The boom angle prediction module is used to determine the predicted values ​​of the boom rotation angle and the boom pitch angle based on the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding characteristics, and control the bucket wheel machine boom to work according to the predicted values ​​of the boom rotation angle and the boom pitch angle.

[0014] Optionally, the data division and regularization module is specifically used to: divide the time series of the positioning data of the fixed antenna according to a predetermined time scale, and regularize the data according to the positioning data parameter sample dimension, so as to obtain the corresponding sequence of the local time series input vector of the fixed antenna positioning coordinates and the sequence of the local time series input vector of the fixed antenna heading angle.

[0015] Optionally, the local timing pattern extraction module is specifically used to: input the sequence of local timing input vectors of the fixed antenna positioning coordinates and the sequence of local timing input vectors of the fixed antenna heading angles into an antenna positioning local timing pattern feature extractor based on a time convolutional network to obtain the corresponding sequence of local timing pattern feature vectors of the fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of the fixed antenna heading angles.

[0016] Optionally, the feature interaction fusion module is specifically used to use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local timing pattern feature vectors and fixed antenna heading angle local timing pattern feature vectors in the sequence of fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of fixed antenna heading angle local timing pattern feature vectors to obtain a sequence of fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors.

[0017] Optionally, the feature interaction fusion module includes:

[0018] A feature cascading unit, used for cascading each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors to obtain a fixed antenna cascade feature vector;

[0019] A threshold calculation unit, used for first multiplying the fixed antenna cascade feature vector by the transformation matrix, then adding the feature vector obtained by the multiplication to the bias vector by position, and then inputting the feature vector obtained by the position addition into a sigmoid activation function to obtain a threshold vector;

[0020] A feature fusion unit is used to calculate the weighted sum of each group of corresponding fixed antenna positioning coordinate local time series pattern feature vectors and fixed antenna heading angle local time series pattern feature vectors based on the threshold values ​​of each position in the threshold vector, so as to obtain the fixed antenna positioning coordinate-heading angle local time series element interaction feature vector;

[0021] Wherein, the feature fusion unit is used to:

[0022] Using the characteristic value of each position in the local time series pattern characteristic vector of the fixed antenna positioning coordinates as the exponent of the natural constant to calculate the product between the exponential function value with the natural constant as the base and one minus the corresponding threshold value, so as to obtain a local time series coding characteristic vector of the fixed antenna positioning coordinates composed of multiple local time series characteristic values ​​of the fixed antenna positioning coordinates;

[0023] Taking the characteristic value of each position in the local timing pattern characteristic vector of the fixed antenna heading angle as the exponent of the natural constant to calculate the product between the exponential function value with the natural constant as the base and the corresponding threshold value, so as to obtain a local timing coding characteristic vector of the fixed antenna heading angle composed of multiple local timing characteristic values ​​of the fixed antenna heading angle;

[0024] The local time sequence coding feature vector of the fixed antenna positioning coordinates and the local time sequence coding feature vector of the fixed antenna heading angle are added by position to obtain the local time sequence element interaction feature vector of the fixed antenna positioning coordinates-heading angle.

[0025] Optionally, the full-time domain context semantic understanding module is specifically used to: input the sequence of the fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vectors into the Transformer-based positioning full-time domain context semantic understander to obtain the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector.

[0026] Optionally, the cantilever angle prediction module includes:

[0027] A cantilever rotation angle prediction unit, used for inputting the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature into a cantilever rotation angle predictor based on a decoder to obtain a recommended cantilever rotation angle prediction value;

[0028] A cantilever pitch angle prediction unit, used for inputting the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature into a cantilever pitch angle predictor based on a decoder to obtain a recommended cantilever pitch angle prediction value;

[0029] The bucket wheel machine boom control unit is used to input the recommended boom rotation angle prediction value and the recommended boom pitch angle prediction value into the bucket wheel machine controller, so that the bucket wheel machine controller controls the bucket wheel machine boom to work according to the recommended boom rotation angle prediction value and the recommended boom pitch angle prediction value.

[0030] Optionally, the bucket wheel machine intelligent control system further includes a training module;

[0031] The training module is used to:

[0032] Training the decoder-based fixed antenna positioning semantic reasoner;

[0033] When the local time series pattern extraction module is specifically used to input the sequence of the local time series input vectors of the fixed antenna positioning coordinates and the sequence of the local time series input vectors of the fixed antenna heading angle into the antenna positioning local time series pattern feature extractor based on the time convolution network, the antenna positioning local time series pattern feature extractor based on the time convolution network is trained;

[0034] When the feature interaction fusion module is specifically used to use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors in the sequence of the fixed antenna positioning coordinate local time sequence pattern feature vectors and the sequence of the fixed antenna heading angle local time sequence pattern feature vectors, the antenna positioning feature local element dynamic interaction module is trained;

[0035] When the full-time-domain context semantic understanding module is specifically used to input the sequence of the fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vector into the Transformer-based positioning full-time-domain context semantic understander, the Transformer-based positioning full-time-domain context semantic understander is trained;

[0036] When the boom angle prediction module includes the boom rotation angle prediction unit and the boom pitch angle prediction unit, the decoder-based boom rotation angle predictor and the decoder-based boom pitch angle predictor are trained.

[0037] Optionally, the training module includes:

[0038] A training data acquisition unit, used to acquire training data and a true value of a boom rotation angle and a true value of a boom pitch angle of a bucket wheel machine, wherein the training data includes a time series of training positioning data of a fixed antenna installed on the bucket wheel machine acquired through a GPS positioning system, and the training positioning data includes training positioning coordinates and a training heading angle;

[0039] A training data regularization unit, used to divide the time series of the training positioning data according to a predetermined time scale, and regularize the data according to the training positioning data parameter sample dimension, so as to obtain a sequence of corresponding training fixed antenna positioning coordinate local time series input vectors and a sequence of training fixed antenna heading angle local time series input vectors;

[0040] A training local temporal pattern feature extraction unit, used for inputting the sequence of the training fixed antenna positioning coordinate local temporal input vectors and the sequence of the training fixed antenna heading angle local temporal input vectors into the antenna positioning local temporal pattern feature extractor based on the time convolution network, so as to obtain a corresponding sequence of training fixed antenna positioning coordinate local temporal pattern feature vectors and a sequence of training fixed antenna heading angle local temporal pattern feature vectors;

[0041] A training local timing element interaction unit is used to interactively fuse each group of corresponding training fixed antenna positioning coordinate local timing pattern feature vectors and training fixed antenna heading angle local timing pattern feature vectors in the sequence of the training fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of the training fixed antenna heading angle local timing pattern feature vectors using the antenna positioning feature local element dynamic interaction module to obtain a sequence of training fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors;

[0042] A training local temporal semantic reasoning unit is used to input the sequence of the training fixed antenna positioning coordinate-heading angle local temporal element interaction feature vectors into the decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of training fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vectors;

[0043] A training full-time domain semantic understanding unit is used to input the sequence of the training fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vector into the Transformer-based positioning full-time domain context semantic understander to obtain the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector;

[0044] a first decoding loss unit, configured to input the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever rotation angle predictor to obtain a predicted value of the cantilever rotation angle, and obtain a first decoding loss function value according to the predicted value of the cantilever rotation angle and the true value of the cantilever rotation angle;

[0045] a second decoding loss unit, configured to input the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever pitch angle predictor to obtain a predicted value of the cantilever pitch angle, and obtain a second decoding loss function value according to the predicted value of the cantilever pitch angle and a true value of the cantilever pitch angle;

[0046] A training unit is used to use the weighted sum of the first decoding loss function value and the second decoding loss function value as the loss function value, and to train the antenna positioning local temporal pattern feature extractor based on the time convolutional network, the antenna positioning feature local element dynamic interaction module, the decoder-based fixed antenna positioning semantic reasoner, the Transformer-based positioning full-time domain context semantic understander, the decoder-based cantilever rotation angle predictor and the decoder-based cantilever pitch angle predictor through gradient descent back propagation.

[0047] Another aspect of the present disclosure provides a bucket wheel machine intelligent control method based on spatial positioning, the bucket wheel machine intelligent control method comprising:

[0048] Acquiring a time series of positioning data of a fixed antenna installed on a bucket wheel machine through a GPS positioning system; wherein the positioning data includes positioning coordinates and heading angles;

[0049] Sequentially performing data division and data regularization on the time series of the positioning data of the fixed antenna to obtain a sequence of corresponding local time series input vectors of the fixed antenna positioning coordinates and a sequence of local time series input vectors of the fixed antenna heading angle;

[0050] Perform antenna positioning local time sequence pattern feature extraction on the sequence of local time sequence input vectors of the fixed antenna positioning coordinates and the sequence of local time sequence input vectors of the fixed antenna heading angles, respectively, to obtain a corresponding sequence of local time sequence pattern feature vectors of the fixed antenna positioning coordinates and a sequence of local time sequence pattern feature vectors of the fixed antenna heading angles;

[0051] Interactively fuse each group of corresponding local timing pattern feature vectors of fixed antenna positioning coordinates and local timing pattern feature vectors of fixed antenna heading angle in the sequence of local timing pattern feature vectors of fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of fixed antenna heading angles to obtain a sequence of interactive feature vectors of local timing elements of fixed antenna positioning coordinates and heading angles;

[0052] Inputting the sequence of the fixed antenna positioning coordinate-heading angle local time series element interaction feature vectors into a decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors;

[0053] Performing positioning full-time-domain context semantic understanding on the sequence of the fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors to obtain the fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature vector as the fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature;

[0054] Based on the full-time domain semantic understanding features of the fixed antenna positioning coordinates-heading angle, the predicted values ​​of the boom rotation angle and the boom pitch angle are determined, and the bucket wheel excavator boom is controlled to work according to the predicted values ​​of the boom rotation angle and the boom pitch angle.

[0055] Compared with the prior art, the present invention obtains the time series of positioning data of a fixed antenna installed on a bucket wheel machine through a GPS positioning system, wherein the positioning data includes positioning coordinates and heading angles, and adopts data processing and analysis algorithms based on artificial intelligence to perform time series feature analysis and interaction on the positioning data, so as to intelligently obtain the predicted values ​​of the cantilever rotation angle and pitch angle according to the semantic reasoning characteristics of the positioning coordinates and heading angles in the full time domain, and control the bucket wheel machine cantilever to work according to the predicted values, thereby realizing accurate monitoring of the bucket wheel machine's position and improving the accuracy and efficiency of the bucket wheel machine's operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0057] Figure 1A schematic diagram of the structure of a bucket wheel machine intelligent control system based on spatial positioning provided in one embodiment of the present disclosure;

[0058] Figure 2 A schematic diagram of data flow of a bucket wheel machine intelligent control system based on spatial positioning provided by another embodiment of the present disclosure;

[0059] Figure 3 A schematic diagram of the structure of a feature interaction fusion module provided in another embodiment of the present disclosure;

[0060] Figure 4 A schematic diagram of the structure of a cantilever angle prediction module provided in another embodiment of the present disclosure;

[0061] Figure 5 A schematic diagram of the structure of a training module provided in another embodiment of the present disclosure;

[0062] Figure 6 A flowchart of an intelligent control method for a bucket wheel excavator based on spatial positioning is provided in accordance with another embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings. However, it can be understood by those skilled in the art that in each embodiment of the present disclosure, many technical details are proposed in order to enable readers to better understand the present disclosure. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed for protection in the present disclosure can also be implemented. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific implementation of the present disclosure. The various embodiments can be combined and referenced with each other without contradiction.

[0064] Unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" 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.

[0065] Although the present disclosure makes various references to certain modules in the system according to embodiments of the present disclosure, any number of different modules can be used and run on a user terminal and / or server. The modules are illustrative only, and different aspects of the systems and methods can use different modules.

[0066] The present disclosure uses a flow chart to illustrate the operations performed according to the method of the embodiments of the present disclosure. It should be understood that the preceding or following operations are not necessarily performed precisely in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0067] In view of the problems existing in the traditional bucket wheel control method, the present invention provides an intelligent control system and method for bucket wheel based on spatial positioning, and its technical concept is as follows: it obtains the time series of positioning data of the fixed antenna installed on the bucket wheel through the GPS positioning system, wherein the positioning data includes positioning coordinates and heading angles, and uses artificial intelligence-based data processing and analysis algorithms to perform time series feature analysis and interaction on the positioning data, thereby intelligently obtaining the predicted values ​​of the cantilever rotation angle and pitch angle according to the semantic reasoning characteristics of the positioning coordinates and heading angles in the full time domain, and controlling the bucket wheel cantilever to work according to the predicted values. Through the above method, accurate monitoring of the bucket wheel position is achieved, and the accuracy and efficiency of bucket wheel operation are improved.

[0068] One embodiment of the present disclosure relates to an intelligent control system 100 for a bucket wheel machine based on spatial positioning, and its structure is as follows: Figure 1 As shown, the data flow is as follows Figure 2 shown.

[0069] Combined Figure 1 and Figure 2 The bucket wheel machine intelligent control system 100 based on spatial positioning provided in this embodiment includes a fixed antenna positioning data acquisition module 110, a data division and regularization module 120, a local time series pattern extraction module 130, a feature interaction fusion module 140, a positioning semantic reasoning module 150, a full time domain context semantic understanding module 160, and a cantilever angle prediction module 170.

[0070] The fixed antenna positioning data acquisition module 110 is used to acquire the time series of positioning data of the fixed antenna installed on the bucket wheel machine through the GPS positioning system. The positioning data includes positioning coordinates and heading angles.

[0071] Specifically, in order to accurately control the movement and operating direction of the bucket wheel machine, the fixed antenna positioning data acquisition module 110 uses the GPS positioning system, i.e., the global positioning system, to obtain the time series of positioning data of the fixed antenna installed on the bucket wheel machine, wherein the positioning data includes positioning coordinates and heading angles. Positioning coordinates refer to the specific position of the bucket wheel machine in the geographic space, which is composed of three values: longitude, latitude, and altitude. They are used to determine the precise position of the bucket wheel machine and are the basis for spatial positioning and monitoring thereof. The heading angle is also called the azimuth, which indicates the orientation of the bucket wheel machine relative to the geographic north or a reference direction. It is an important parameter for describing the direction of the bucket wheel machine and is essential for monitoring the operating status of the bucket wheel machine and performing precise control.

[0072] The data division and regularization module 120 is used to perform data division and data regularization on the time series of the fixed antenna positioning data in sequence to obtain a sequence of corresponding fixed antenna positioning coordinate local time series input vectors and a sequence of fixed antenna heading angle local time series input vectors.

[0073] Exemplarily, the data division and regularization module 120 is specifically used to: divide the time series of the fixed antenna positioning data according to a predetermined time scale, and regularize the data according to the positioning data parameter sample dimension to obtain a corresponding sequence of local time series input vectors of the fixed antenna positioning coordinates and a sequence of local time series input vectors of the fixed antenna heading angle.

[0074] Specifically, considering that the positioning data of the fixed antenna will also change with the change of time, and the characteristic information of the positioning data of the fixed antenna in different time periods is also different, therefore, in order to more carefully analyze the timing characteristic information of the positioning data of the fixed antenna in each different time window, the data division and regularization module 120 divides the time series of the positioning data of the fixed antenna according to a predetermined time scale, wherein data division according to the predetermined time scale can divide the time series of the continuous fixed antenna positioning data into multiple time windows, each time window contains positioning data within a time range, which is conducive to capturing the timing dynamic information and change trends in the positioning data of the fixed antenna in each time period. Since the two types of data, positioning coordinates and heading angles, belong to different positioning data parameter sample dimensions and have different changing trends and fluctuations in time series, the data division and regularization module 120 regularizes the divided data according to the positioning data parameter sample dimensions to obtain the corresponding sequence of local time series input vectors of fixed antenna positioning coordinates and the sequence of local time series input vectors of fixed antenna heading angles. In this way, the positioning coordinates and heading angles can be analyzed and processed separately, so as to help the system better utilize the timing information of the positioning data and realize accurate prediction and intelligent control of the cantilever rotation angle and pitch angle.

[0075] The local timing pattern extraction module 130 is used to extract antenna positioning local timing pattern features from the sequence of local timing input vectors of fixed antenna positioning coordinates and the sequence of local timing input vectors of fixed antenna heading angles, respectively, to obtain the corresponding sequence of local timing pattern feature vectors of fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of fixed antenna heading angles.

[0076] Exemplarily, the local timing pattern extraction module 130 is specifically used to: input the sequence of local timing input vectors of fixed antenna positioning coordinates and the sequence of local timing input vectors of fixed antenna heading angles into an antenna positioning local timing pattern feature extractor based on a time convolutional network to obtain the corresponding sequence of local timing pattern feature vectors of fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of fixed antenna heading angles.

[0077] Specifically, considering that each local time series input vector of the fixed antenna positioning coordinates and the local time series input vector of the fixed antenna heading angle in the sequence of local time series input vectors of the fixed antenna positioning coordinates and the local time series input vector of the fixed antenna heading angle reflects the time series feature information about the positioning coordinates and heading angle of the fixed antenna within a local time period, and the time convolution network is a deep learning network structure suitable for processing time series data, which can effectively capture the local pattern dependencies in the time series data, therefore, the local time series pattern extraction module 130 inputs the sequence of local time series input vectors of the fixed antenna positioning coordinates and the sequence of local time series input vectors of the fixed antenna heading angle into the antenna positioning local time series pattern feature extractor based on the time convolution network, so as to respectively capture and mine the time series pattern characteristics and change trends of the positioning coordinates and heading angles at different local time scales, thereby obtaining the corresponding sequence of local time series pattern feature vectors of the fixed antenna positioning coordinates and the sequence of local time series pattern feature vectors of the fixed antenna heading angle.

[0078] The feature interaction fusion module 140 is used to interactively fuse each group of corresponding fixed antenna positioning coordinate local timing pattern feature vectors and fixed antenna heading angle local timing pattern feature vectors in the sequence of fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of fixed antenna heading angle local timing pattern feature vectors to obtain a sequence of fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors.

[0079] Exemplarily, the feature interaction fusion module 140 is specifically used to use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local timing pattern feature vectors and fixed antenna heading angle local timing pattern feature vectors in the sequence of fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of fixed antenna heading angle local timing pattern feature vectors to obtain a sequence of fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors.

[0080] Specifically, considering that the fixed antenna positioning coordinates and heading angle are two key positioning parameters, the relationship between them is crucial to the bucket wheel boom control, and there is an interactive relationship between the local timing pattern characteristics of the fixed antenna positioning coordinates and the local timing pattern characteristics of the fixed antenna heading angle. Specifically, there is a position and direction relationship between the positioning coordinates of the fixed antenna and the heading angle, that is, the position information of the bucket wheel will affect the change of the heading angle, and the change of the heading angle will also be fed back to the position information. Therefore, in order to obtain a more comprehensive and comprehensive position and direction feature representation, the feature interaction fusion module 140 uses the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local timing pattern feature vectors and fixed antenna heading angle local timing pattern feature vectors in the sequence of fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of fixed antenna heading angle local timing pattern feature vectors to obtain a sequence of fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors. By utilizing the dynamic interaction module of the local elements of the antenna positioning features to interactively fuse the local time series pattern feature vectors of the fixed antenna positioning coordinates and the local time series pattern feature vectors of the fixed antenna heading angle, the fixed antenna position information and direction information can influence and interact with each other in the local time series, thereby better capturing the motion state and direction changes of the bucket wheel machine in each local time period, thereby further optimizing the performance and stability of the bucket wheel machine's intelligent control system.

[0081] For example, Figure 3 As shown, the feature interaction fusion module 140 includes a feature cascade unit 141 , a threshold calculation unit 142 , and a feature fusion unit 143 .

[0082] The feature cascading unit 141 is used to cascade each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors to obtain a fixed antenna cascade feature vector.

[0083] The threshold calculation unit 142 is used to first multiply the fixed antenna cascade eigenvector with the transformation matrix, then add the eigenvector obtained by the multiplication and the bias vector by position, and then input the eigenvector obtained by the position addition into the sigmoid activation function to obtain the threshold vector.

[0084] The feature fusion unit 143 is used to calculate the weighted sum of each group of corresponding fixed antenna positioning coordinate local time series pattern feature vectors and fixed antenna heading angle local time series pattern feature vectors based on the threshold values ​​of each position in the threshold vector, so as to obtain the fixed antenna positioning coordinate-heading angle local time series element interaction feature vector;

[0085] Among them, the feature fusion unit 143 is used to: use the eigenvalue of each position in the local timing pattern eigenvector of the fixed antenna positioning coordinates as the exponent of the natural constant to calculate the product between the exponential function value with the natural constant as the base and one minus the corresponding threshold value, so as to obtain a fixed antenna positioning coordinate local timing coding feature vector composed of multiple fixed antenna positioning coordinate local timing eigenvalues; use the eigenvalue of each position in the local timing pattern eigenvector of the fixed antenna heading angle as the exponent of the natural constant to calculate the product between the exponential function value with the natural constant as the base and the corresponding threshold value, so as to obtain a fixed antenna heading angle local timing coding feature vector composed of multiple fixed antenna heading angle local timing eigenvalues; add the fixed antenna positioning coordinate local timing coding feature vector and the fixed antenna heading angle local timing coding feature vector by position, so as to obtain the fixed antenna positioning coordinate-heading angle local timing element interaction feature vector.

[0086] In other words, the feature interaction fusion module 140 can be used to: use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors with the following feature interaction fusion formula to obtain the fixed antenna positioning coordinate-heading angle local time sequence element interaction feature vector, wherein the feature interaction fusion formula includes:

[0087] V=[V 1 ; V 2 ]

[0088] T=Sigmoid(WV+b)

[0089]

[0090] Among them, V 1 and V 2 denote the local temporal pattern feature vector of the fixed antenna positioning coordinates and the local temporal pattern feature vector of the fixed antenna heading angle corresponding to each group, V is the fixed antenna cascade feature vector, [·; ·] denotes cascade processing, T is the threshold vector, t is the threshold value of each position in the threshold vector, t∈[0,1], W is the transformation matrix, b is the bias vector, sigmoid(·) denotes the sigmoid activation function, v 1 and v 2are the eigenvalues ​​of each position in the local time sequence pattern eigenvector of the fixed antenna positioning coordinates and the local time sequence pattern eigenvector of the fixed antenna heading angle corresponding to each group, v c It is the eigenvalue of each position in the interaction eigenvector of the local time series element of the fixed antenna positioning coordinate-heading angle.

[0091] The positioning semantic reasoning module 150 is used to input the sequence of fixed antenna positioning coordinate-heading angle local time series element interaction feature vectors into the decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors.

[0092] Specifically, considering that in the boom control of the bucket wheel machine, the semantic information of the positioning coordinates and heading angles of the fixed antenna is crucial to the understanding and decision-making of the system, and each fixed antenna positioning coordinate-heading angle local time series element interaction feature vector contains the semantic association relationship between the positioning coordinates and the heading angle in time series, therefore, the positioning semantic reasoning module 150 inputs the sequence of the fixed antenna positioning coordinate-heading angle local time series element interaction feature vectors into the fixed antenna positioning semantic reasoner based on the decoder to obtain the sequence of the fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors. Among them, the fixed antenna positioning semantic reasoner based on the decoder can capture the semantic information and correlation in the fixed antenna positioning coordinate-heading angle local time series element interaction feature, so as to infer the fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature containing richer temporal interdependence between the positioning coordinates and the heading angle, thereby improving the system's reasoning and decision-making capabilities for the bucket wheel machine state.

[0093] The full-time-domain contextual semantic understanding module 160 is used to perform full-time-domain contextual semantic understanding on the sequence of fixed antenna positioning coordinates-heading angle local temporal semantic reasoning feature vectors to obtain the fixed antenna positioning coordinates-heading angle full-time-domain semantic understanding feature vector as the fixed antenna positioning coordinates-heading angle full-time-domain semantic understanding feature.

[0094] Exemplarily, the full-time-domain contextual semantic understanding module 160 is specifically used to: input the sequence of fixed antenna positioning coordinates-heading angle local temporal semantic reasoning feature vectors into the Transformer-based positioning full-time-domain contextual semantic understander to obtain the fixed antenna positioning coordinates-heading angle full-time-domain semantic understanding feature vector.

[0095] Specifically, considering that there are different degrees of contextual semantic associations between different time spans in the sequence of fixed antenna positioning coordinates-heading angle local time series semantic reasoning feature vectors, and the Transformer model has excellent performance in processing sequence data, and can effectively capture long-distance dependencies and semantic information in the sequence, therefore, in order to integrate the semantic interdependence information between the sequence of fixed antenna positioning coordinates-heading angle local time series semantic reasoning feature vectors on the entire time scale, the full-time domain context semantic understanding module 160 inputs the sequence of fixed antenna positioning coordinates-heading angle local time series semantic reasoning feature vectors into the Transformer-based positioning full-time domain context semantic understander to obtain the fixed antenna positioning coordinates-heading angle full-time domain semantic understanding feature vector. Among them, the Transformer-based positioning full-time domain context semantic understander can more comprehensively capture the semantic information and contextual relationships under different time spans in the local time series of fixed antenna positioning coordinates-heading angles, so as to better understand the correlation between positioning coordinates and heading angles in the entire time range, and obtain a fixed antenna positioning coordinates-heading angle full-time domain semantic understanding feature vector with more full-time domain semantics.

[0096] The boom angle prediction module 170 is used to determine the predicted values ​​of the boom rotation angle and the boom pitch angle based on the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding features, and control the bucket wheel machine boom to work according to the predicted values ​​of the boom rotation angle and the boom pitch angle.

[0097] For example, Figure 4 As shown, the boom angle prediction module 170 includes a boom rotation angle prediction unit 171 , a boom pitch angle prediction unit 172 , and a bucket wheel machine boom control unit 173 .

[0098] The boom rotation angle prediction unit 171 is used to input the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding features into the boom rotation angle predictor based on the decoder to obtain the recommended boom rotation angle prediction value.

[0099] The boom pitch angle prediction unit 172 is used to input the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding features into the decoder-based boom pitch angle predictor to obtain a recommended boom pitch angle prediction value.

[0100] The bucket wheel machine boom control unit 173 is used to input the recommended boom rotation angle prediction value and the recommended boom pitch angle prediction value into the bucket wheel machine controller, so that the bucket wheel machine controller controls the bucket wheel machine boom to work according to the recommended boom rotation angle prediction value and the recommended boom pitch angle prediction value.

[0101] Specifically, by utilizing a decoder-based cantilever rotation angle predictor and a decoder-based cantilever pitch angle predictor, the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding features are decoded and processed, so as to intelligently obtain the predicted values ​​of the cantilever rotation angle and pitch angle, and the bucket wheel crane cantilever is controlled to work according to the predicted values, effectively realizing the precise monitoring of the bucket wheel crane position and improving the accuracy and efficiency of the bucket wheel crane operation.

[0102] For example, a person skilled in the art should know that before applying a deep neural network model for inference, the deep neural network model needs to be trained so that the deep neural network can realize a specific function. To this end, the bucket wheel machine intelligent control system 100 based on spatial positioning also includes a training module 200.

[0103] The training module 200 is used to train the decoder-based fixed antenna positioning semantic reasoner, the temporal convolutional network-based antenna positioning local temporal pattern feature extractor, the antenna positioning feature local element dynamic interaction module, the Transformer-based positioning full-time domain context semantic understander, the decoder-based cantilever rotation angle predictor and the decoder-based cantilever pitch angle predictor in the above-mentioned implementation.

[0104] For example, Figure 5 As shown, the training module 200 includes a training data acquisition unit 210, a training data regularization unit 220, a training local temporal pattern feature extraction unit 230, a training local temporal element interaction unit 240, a training local temporal semantic reasoning unit 250, a training full-time domain semantic understanding unit 260, a first decoding loss unit 270, a second decoding loss unit 280, and a training unit 290.

[0105] The training data acquisition unit 210 is used to acquire training data and the true value of the boom rotation angle and the true value of the boom pitch angle of the bucket wheel machine, wherein the training data includes a time series of training positioning data of a fixed antenna installed on the bucket wheel machine obtained through a GPS positioning system, and the training positioning data includes training positioning coordinates and training heading angles.

[0106] The training data regularization unit 220 is used to divide the time series of the training positioning data according to a predetermined time scale, and to regularize the data according to the training positioning data parameter sample dimension, so as to obtain a corresponding sequence of local time series input vectors of the training fixed antenna positioning coordinates and a sequence of local time series input vectors of the training fixed antenna heading angles.

[0107] The training local timing pattern feature extraction unit 230 is used to input the sequence of training fixed antenna positioning coordinate local timing input vectors and the sequence of training fixed antenna heading angle local timing input vectors into the antenna positioning local timing pattern feature extractor based on the time convolution network to obtain the corresponding sequence of training fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of training fixed antenna heading angle local timing pattern feature vectors.

[0108] The training local timing element interaction unit 240 is used to use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding training fixed antenna positioning coordinate local timing pattern feature vectors and training fixed antenna heading angle local timing pattern feature vectors in the sequence of training fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of training fixed antenna heading angle local timing pattern feature vectors to obtain a sequence of training fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors.

[0109] The training local temporal semantic reasoning unit 250 is used to input the sequence of training fixed antenna positioning coordinates-heading angle local temporal element interaction feature vectors into the decoder-based fixed antenna positioning semantic reasoner to obtain the sequence of training fixed antenna positioning coordinates-heading angle local temporal semantic reasoning feature vectors.

[0110] The training full-time domain semantic understanding unit 260 is used to input the sequence of training fixed antenna positioning coordinates-heading angle local temporal semantic reasoning feature vectors into the Transformer-based positioning full-time domain context semantic understander to obtain the training fixed antenna positioning coordinates-heading angle full-time domain semantic understanding feature vector.

[0111] The first decoding loss unit 270 is used to input the full-time domain semantic understanding feature vector of the training fixed antenna positioning coordinate-heading angle into the decoder-based cantilever rotation angle predictor to obtain the predicted value of the cantilever rotation angle, and obtain the first decoding loss function value based on the predicted value of the cantilever rotation angle and the true value of the cantilever rotation angle.

[0112] The second decoding loss unit 280 is used to input the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever pitch angle predictor to obtain the predicted value of the cantilever pitch angle, and obtain the second decoding loss function value based on the predicted value of the cantilever pitch angle and the true value of the cantilever pitch angle.

[0113] The training unit 290 is used to use the weighted sum of the first decoding loss function value and the second decoding loss function value as the loss function value, and to train the antenna positioning local temporal pattern feature extractor based on the time convolutional network, the antenna positioning feature local element dynamic interaction module, the decoder-based fixed antenna positioning semantic reasoner, the Transformer-based positioning full-time domain context semantic understander, the decoder-based cantilever rotation angle predictor and the decoder-based cantilever pitch angle predictor through gradient descent back propagation.

[0114] Specifically, the sequence of local temporal pattern feature vectors of training fixed antenna positioning coordinates and the sequence of local temporal pattern feature vectors of training fixed antenna heading angles respectively represent the temporal distribution of temporal implicit correlation features of training positioning coordinates and the temporal distribution of temporal implicit correlation features of training heading angles. At the same time, there is a correlation between the training heading angle and the training positioning coordinates (that is, the heading angle can be determined based on the spatial coordinates of the fixed station antenna). In this way, when the antenna positioning feature local element dynamic interaction module is used to interactively fuse each group of corresponding training fixed antenna positioning coordinate local time series pattern feature vectors and training fixed antenna heading angle local time series pattern feature vectors in the sequence of training fixed antenna positioning coordinate local time series pattern feature vectors and the sequence of training fixed antenna heading angle local time series pattern feature vectors, the obtained training fixed antenna positioning coordinate-heading angle local time series element interaction feature vector sequence has a certain degree of feature redundancy in each training fixed antenna positioning coordinate-heading angle local time series element interaction feature vector, which makes the training fixed antenna positioning coordinate-heading angle full time domain semantic understanding feature vector finally obtained by Transformer encoding insufficiently constrained by the overall distribution regression of the decoding regression domain, affecting the accuracy of the recommended cantilever rotation angle prediction value and the cantilever pitch angle prediction value. To this end, this embodiment uses the first decoding loss unit 270 and the second decoding loss unit 280 to obtain the first decoding loss function value and the second decoding loss function value respectively, and uses the training unit 290 to train the antenna positioning local temporal pattern feature extractor based on the time convolutional network, the antenna positioning feature local element dynamic interaction module, the decoder-based fixed antenna positioning semantic reasoner, the Transformer-based positioning full-time domain context semantic understander, the decoder-based cantilever rotation angle predictor and the decoder-based cantilever pitch angle predictor based on the first decoding loss function value and the second decoding loss function value and the back propagation of gradient descent.

[0115] Preferably, the first decoding loss unit 270 is specifically used to: perform a dot addition operation on the training fixed antenna positioning coordinate-heading angle full time domain semantic understanding feature vector and the square root of the length of the training fixed antenna positioning coordinate-heading angle full time domain semantic understanding feature vector and the inverse of the square root of the second norm of the training fixed antenna positioning coordinate-heading angle full time domain semantic understanding feature vector to obtain the training fixed antenna positioning coordinate-heading angle full time domain semantic offset feature vector; calculate the exponential function of the training fixed antenna positioning coordinate-heading angle full time domain semantic offset feature vector with the natural constant as the base to obtain the training fixed antenna positioning coordinate-heading angle full time domain semantic class offset feature vector; and add the training fixed antenna positioning coordinate-heading angle The first decoding loss function is to input the optimized training fixed antenna positioning coordinates-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever rotation angle predictor to obtain the first decoding loss function value. The second decoding loss unit 280 is specifically used to: input the optimized training fixed antenna positioning coordinates-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever pitch angle predictor to obtain the second decoding loss function value.

[0116] Specifically, this embodiment uses the structured norm representation of the full-time domain semantic understanding feature vector of the training fixed antenna positioning coordinates-heading angle as the local canonical coordinates for each eigenvalue of the full-time domain semantic understanding feature vector of the training fixed antenna positioning coordinates-heading angle to determine the rotational offset of the vector overall distribution representation of the full-time domain semantic understanding feature vector of the training fixed antenna positioning coordinates-heading angle relative to the eigenvalue based on the scale modulus and the feature distribution modulus, so as to set the offset prediction direction for each eigenvalue of the full-time domain semantic understanding feature vector of the training fixed antenna positioning coordinates-heading angle as the center, and use The vector eigenvalues ​​of the training fixed antenna positioning coordinates-heading angle full time domain semantic understanding feature vector are constrained for the overall distribution bounding box representation, so as to improve the constraint of the training fixed antenna positioning coordinates-heading angle full time domain semantic understanding feature vector under the overall decoding regression distribution, thereby improving the training speed of the model and the accuracy of the recommended cantilever rotation angle prediction value and the recommended cantilever pitch angle prediction value obtained by the training fixed antenna positioning coordinates-heading angle full time domain semantic understanding feature vector respectively through the decoder-based cantilever rotation angle predictor and the decoder-based cantilever rotation angle predictor. In this way, the accurate monitoring of the bucket wheel position is achieved, and the accuracy and efficiency of the bucket wheel operation are improved.

[0117] The bucket wheel machine intelligent control system 100 based on spatial positioning provided in the above embodiment can be implemented in various wireless terminals, such as a server with a bucket wheel machine intelligent control algorithm based on spatial positioning. In a possible implementation, the bucket wheel machine intelligent control system 100 based on spatial positioning can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the bucket wheel machine intelligent control system 100 based on spatial positioning can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the bucket wheel machine intelligent control system 100 based on spatial positioning can also be one of the many hardware modules of the wireless terminal.

[0118] Alternatively, in another example, the bucket wheel machine intelligent control system 100 based on spatial positioning and the wireless terminal may also be separate devices, and the bucket wheel machine intelligent control system 100 based on spatial positioning may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0119] An intelligent control system for a bucket wheel machine based on spatial positioning provided in an embodiment of the present disclosure, compared with the prior art, obtains the time series of positioning data of a fixed antenna installed on the bucket wheel machine through a GPS positioning system, wherein the positioning data includes positioning coordinates and a heading angle, and adopts a data processing and analysis algorithm based on artificial intelligence to perform time series feature analysis and interaction on the positioning data, thereby intelligently obtaining the predicted values ​​of the cantilever rotation angle and the pitch angle according to the semantic reasoning characteristics of the positioning coordinates and the heading angle in the full time domain, and controlling the bucket wheel machine cantilever to work according to the predicted values, thereby realizing accurate monitoring of the bucket wheel machine's position and improving the accuracy and efficiency of the bucket wheel machine's operation.

[0120] Another embodiment of the present disclosure relates to an intelligent control method for a bucket wheel machine based on spatial positioning, such as Figure 6 As shown, including:

[0121] Step S110, obtaining a time series of positioning data of a fixed antenna installed on a bucket wheel machine through a GPS positioning system; wherein the positioning data includes positioning coordinates and heading angles.

[0122] Step S120, sequentially performing data division and data regularization on the time series of the fixed antenna positioning data to obtain a corresponding sequence of local time series input vectors of the fixed antenna positioning coordinates and a sequence of local time series input vectors of the fixed antenna heading angles.

[0123] Step S130, performing antenna positioning local timing pattern feature extraction on the sequence of local timing input vectors of fixed antenna positioning coordinates and the sequence of local timing input vectors of fixed antenna heading angles, respectively, to obtain a corresponding sequence of local timing pattern feature vectors of fixed antenna positioning coordinates and a sequence of local timing pattern feature vectors of fixed antenna heading angles.

[0124] Step S140, interactively fuse each group of corresponding local timing pattern feature vectors of fixed antenna positioning coordinates and local timing pattern feature vectors of fixed antenna heading angle in the sequence of local timing pattern feature vectors of fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of fixed antenna heading angle to obtain a sequence of interactive feature vectors of local timing elements of fixed antenna positioning coordinates and heading angle.

[0125] Step S150, inputting the sequence of fixed antenna positioning coordinates-heading angle local time series element interaction feature vectors into a decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of fixed antenna positioning coordinates-heading angle local time series semantic reasoning feature vectors.

[0126] Step S160, performing full-time-domain context semantic understanding on the sequence of fixed antenna positioning coordinates-heading angle local temporal sequence semantic reasoning feature vectors, and obtaining a full-time-domain semantic understanding feature vector of fixed antenna positioning coordinates-heading angle as a full-time-domain semantic understanding feature of fixed antenna positioning coordinates-heading angle.

[0127] Step S170, based on the full-time domain semantic understanding features of the fixed antenna positioning coordinates-heading angle, the predicted values ​​of the boom rotation angle and the boom pitch angle are determined, and the bucket wheel excavator boom is controlled to work according to the predicted values ​​of the boom rotation angle and the boom pitch angle.

[0128] The specific implementation steps of the bucket wheel excavator intelligent control method based on spatial positioning provided in the embodiment of the present disclosure can be referred to the specific implementation methods of each module in the bucket wheel excavator intelligent control system based on spatial positioning provided in the embodiment of the present disclosure, and will not be repeated here.

[0129] Compared with the prior art, the intelligent control method for bucket wheel machine based on spatial positioning provided in the embodiment of the present disclosure obtains the time series of positioning data of the fixed antenna installed on the bucket wheel machine through the GPS positioning system, wherein the positioning data includes positioning coordinates and heading angle, and adopts the data processing and analysis algorithm based on artificial intelligence to perform time series feature analysis and interaction on the positioning data, so as to intelligently obtain the predicted values ​​of the cantilever rotation angle and pitch angle according to the semantic reasoning characteristics of the positioning coordinates and heading angle in the whole time domain, and control the bucket wheel machine cantilever to work according to the predicted values, thereby realizing the accurate monitoring of the position of the bucket wheel machine and improving the accuracy and efficiency of the bucket wheel machine operation.

[0130] Those skilled in the art will appreciate that the above-mentioned embodiments are specific embodiments for implementing the present disclosure, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present disclosure.

Claims

1. A bucket wheel excavator intelligent control system based on spatial positioning, characterized in that: The bucket wheel machine intelligent control system comprises: A fixed antenna positioning data acquisition module is used to acquire a time series of positioning data of a fixed antenna installed on a bucket wheel machine through a GPS positioning system; wherein the positioning data includes positioning coordinates and heading angles; A data division and regularization module, used to sequentially divide and regularize the time series of the positioning data of the fixed antenna to obtain a sequence of local time series input vectors of the corresponding fixed antenna positioning coordinates and a sequence of local time series input vectors of the fixed antenna heading angle; A local timing pattern extraction module is used to extract antenna positioning local timing pattern features from the sequence of local timing input vectors of the fixed antenna positioning coordinates and the sequence of local timing input vectors of the fixed antenna heading angles, respectively, to obtain a corresponding sequence of local timing pattern feature vectors of the fixed antenna positioning coordinates and a sequence of local timing pattern feature vectors of the fixed antenna heading angles; A feature interaction fusion module is used to interactively fuse each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors in the sequence of the fixed antenna positioning coordinate local time sequence pattern feature vectors and the sequence of the fixed antenna heading angle local time sequence pattern feature vectors, so as to obtain a sequence of fixed antenna positioning coordinate-heading angle local time sequence element interaction feature vectors; A positioning semantic reasoning module, used for inputting the sequence of the fixed antenna positioning coordinate-heading angle local time series element interaction feature vectors into a decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors; A full-time-domain contextual semantic understanding module is used to perform full-time-domain contextual semantic understanding of the fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vector sequence to obtain a fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature vector as a fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature; The boom angle prediction module is used to determine the predicted values ​​of the boom rotation angle and the boom pitch angle based on the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding characteristics, and control the bucket wheel machine boom to work according to the predicted values ​​of the boom rotation angle and the boom pitch angle.

2. The intelligent control system of bucket wheel excavator according to claim 1 is characterized in that: The data division and regularization module is specifically used to: divide the time series of the positioning data of the fixed antenna according to a predetermined time scale, and regularize the data according to the positioning data parameter sample dimension, so as to obtain the corresponding sequence of the local time series input vector of the fixed antenna positioning coordinates and the sequence of the local time series input vector of the fixed antenna heading angle.

3. The intelligent control system of bucket wheel excavator according to claim 1 is characterized in that: The local timing pattern extraction module is specifically used to: input the sequence of local timing input vectors of the fixed antenna positioning coordinates and the sequence of local timing input vectors of the fixed antenna heading angles into the antenna positioning local timing pattern feature extractor based on the time convolution network, so as to obtain the corresponding sequence of local timing pattern feature vectors of the fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of the fixed antenna heading angles.

4. The intelligent control system of bucket wheel excavator according to claim 1, characterized in that: The feature interaction fusion module is specifically used to use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local timing pattern feature vectors and fixed antenna heading angle local timing pattern feature vectors in the sequence of fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of fixed antenna heading angle local timing pattern feature vectors to obtain the sequence of fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors.

5. The intelligent control system of bucket wheel excavator according to claim 4 is characterized in that: The feature interaction fusion module includes: A feature cascading unit, used for cascading each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors to obtain a fixed antenna cascade feature vector; A threshold calculation unit, used for first multiplying the fixed antenna cascade feature vector by the transformation matrix, then adding the feature vector obtained by the multiplication to the bias vector by position, and then inputting the feature vector obtained by the position addition into a sigmoid activation function to obtain a threshold vector; A feature fusion unit is used to calculate the weighted sum of each group of corresponding fixed antenna positioning coordinate local time series pattern feature vectors and fixed antenna heading angle local time series pattern feature vectors based on the threshold values ​​of each position in the threshold vector, so as to obtain the fixed antenna positioning coordinate-heading angle local time series element interaction feature vector; Wherein, the feature fusion unit is used to: Using the characteristic value of each position in the local time series pattern characteristic vector of the fixed antenna positioning coordinates as the exponent of the natural constant to calculate the product between the exponential function value with the natural constant as the base and one minus the corresponding threshold value, so as to obtain a local time series coding characteristic vector of the fixed antenna positioning coordinates composed of multiple local time series characteristic values ​​of the fixed antenna positioning coordinates; Taking the characteristic value of each position in the local timing pattern characteristic vector of the fixed antenna heading angle as the exponent of the natural constant to calculate the product between the exponential function value with the natural constant as the base and the corresponding threshold value, so as to obtain a local timing coding characteristic vector of the fixed antenna heading angle composed of multiple local timing characteristic values ​​of the fixed antenna heading angle; The local time sequence coding feature vector of the fixed antenna positioning coordinates and the local time sequence coding feature vector of the fixed antenna heading angle are added by position to obtain the local time sequence element interaction feature vector of the fixed antenna positioning coordinates-heading angle.

6. The intelligent control system of bucket wheel excavator according to claim 1, characterized in that: The full-time-domain contextual semantic understanding module is specifically used to: input the sequence of the fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vectors into the Transformer-based positioning full-time-domain contextual semantic understander to obtain the fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature vector.

7. The intelligent control system of bucket wheel excavator according to claim 1 is characterized in that: The cantilever angle prediction module comprises: A cantilever rotation angle prediction unit, used for inputting the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature into a cantilever rotation angle predictor based on a decoder to obtain a recommended cantilever rotation angle prediction value; A cantilever pitch angle prediction unit, used for inputting the fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature into a cantilever pitch angle predictor based on a decoder to obtain a recommended cantilever pitch angle prediction value; The bucket wheel machine boom control unit is used to input the recommended boom rotation angle prediction value and the recommended boom pitch angle prediction value into the bucket wheel machine controller, so that the bucket wheel machine controller controls the bucket wheel machine boom to work according to the recommended boom rotation angle prediction value and the recommended boom pitch angle prediction value.

8. The bucket wheel machine intelligent control system according to any one of claims 3, 4, 6 and 7, characterized in that: The bucket wheel machine intelligent control system also includes a training module; The training module is used to: Training the decoder-based fixed antenna positioning semantic reasoner; When the local time series pattern extraction module is specifically used to input the sequence of the local time series input vectors of the fixed antenna positioning coordinates and the sequence of the local time series input vectors of the fixed antenna heading angle into the antenna positioning local time series pattern feature extractor based on the time convolution network, the antenna positioning local time series pattern feature extractor based on the time convolution network is trained; When the feature interaction fusion module is specifically used to use the antenna positioning feature local element dynamic interaction module to interactively fuse each group of corresponding fixed antenna positioning coordinate local time sequence pattern feature vectors and fixed antenna heading angle local time sequence pattern feature vectors in the sequence of the fixed antenna positioning coordinate local time sequence pattern feature vectors and the sequence of the fixed antenna heading angle local time sequence pattern feature vectors, the antenna positioning feature local element dynamic interaction module is trained; When the full-time-domain context semantic understanding module is specifically used to input the sequence of the fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vector into the Transformer-based positioning full-time-domain context semantic understander, the Transformer-based positioning full-time-domain context semantic understander is trained; When the boom angle prediction module includes the boom rotation angle prediction unit and the boom pitch angle prediction unit, the decoder-based boom rotation angle predictor and the decoder-based boom pitch angle predictor are trained.

9. The intelligent control system of bucket wheel excavator according to claim 8, characterized in that: The training module comprises: A training data acquisition unit, used to acquire training data and a true value of a boom rotation angle and a true value of a boom pitch angle of a bucket wheel machine, wherein the training data includes a time series of training positioning data of a fixed antenna installed on the bucket wheel machine acquired through a GPS positioning system, and the training positioning data includes training positioning coordinates and a training heading angle; A training data regularization unit, used to divide the time series of the training positioning data according to a predetermined time scale, and regularize the data according to the training positioning data parameter sample dimension, so as to obtain a sequence of corresponding training fixed antenna positioning coordinate local time series input vectors and a sequence of training fixed antenna heading angle local time series input vectors; A training local temporal pattern feature extraction unit, used for inputting the sequence of the training fixed antenna positioning coordinate local temporal input vectors and the sequence of the training fixed antenna heading angle local temporal input vectors into the antenna positioning local temporal pattern feature extractor based on the time convolution network, so as to obtain a corresponding sequence of training fixed antenna positioning coordinate local temporal pattern feature vectors and a sequence of training fixed antenna heading angle local temporal pattern feature vectors; A training local timing element interaction unit is used to interactively fuse each group of corresponding training fixed antenna positioning coordinate local timing pattern feature vectors and training fixed antenna heading angle local timing pattern feature vectors in the sequence of the training fixed antenna positioning coordinate local timing pattern feature vectors and the sequence of the training fixed antenna heading angle local timing pattern feature vectors using the antenna positioning feature local element dynamic interaction module to obtain a sequence of training fixed antenna positioning coordinate-heading angle local timing element interaction feature vectors; A training local temporal semantic reasoning unit is used to input the sequence of the training fixed antenna positioning coordinate-heading angle local temporal element interaction feature vectors into the decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of training fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vectors; A training full-time domain semantic understanding unit is used to input the sequence of the training fixed antenna positioning coordinate-heading angle local temporal semantic reasoning feature vector into the Transformer-based positioning full-time domain context semantic understander to obtain the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector; a first decoding loss unit, configured to input the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever rotation angle predictor to obtain a predicted value of the cantilever rotation angle, and obtain a first decoding loss function value according to the predicted value of the cantilever rotation angle and the true value of the cantilever rotation angle; a second decoding loss unit, configured to input the training fixed antenna positioning coordinate-heading angle full-time domain semantic understanding feature vector into the decoder-based cantilever pitch angle predictor to obtain a predicted value of the cantilever pitch angle, and obtain a second decoding loss function value according to the predicted value of the cantilever pitch angle and a true value of the cantilever pitch angle; A training unit is used to use the weighted sum of the first decoding loss function value and the second decoding loss function value as the loss function value, and to train the antenna positioning local temporal pattern feature extractor based on the time convolutional network, the antenna positioning feature local element dynamic interaction module, the decoder-based fixed antenna positioning semantic reasoner, the Transformer-based positioning full-time domain context semantic understander, the decoder-based cantilever rotation angle predictor and the decoder-based cantilever pitch angle predictor through gradient descent back propagation.

10. A bucket wheel machine intelligent control method based on spatial positioning, characterized in that: The bucket wheel machine intelligent control method comprises: Acquiring a time series of positioning data of a fixed antenna installed on a bucket wheel machine through a GPS positioning system; wherein the positioning data includes positioning coordinates and heading angles; Sequentially performing data division and data regularization on the time series of the positioning data of the fixed antenna to obtain a sequence of corresponding local time series input vectors of the fixed antenna positioning coordinates and a sequence of local time series input vectors of the fixed antenna heading angle; Perform antenna positioning local time sequence pattern feature extraction on the sequence of local time sequence input vectors of the fixed antenna positioning coordinates and the sequence of local time sequence input vectors of the fixed antenna heading angles, respectively, to obtain a corresponding sequence of local time sequence pattern feature vectors of the fixed antenna positioning coordinates and a sequence of local time sequence pattern feature vectors of the fixed antenna heading angles; Interactively fuse each group of corresponding local timing pattern feature vectors of fixed antenna positioning coordinates and local timing pattern feature vectors of fixed antenna heading angle in the sequence of local timing pattern feature vectors of fixed antenna positioning coordinates and the sequence of local timing pattern feature vectors of fixed antenna heading angles to obtain a sequence of interactive feature vectors of local timing elements of fixed antenna positioning coordinates and heading angles; Inputting the sequence of the fixed antenna positioning coordinate-heading angle local time series element interaction feature vectors into a decoder-based fixed antenna positioning semantic reasoner to obtain a sequence of fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors; Performing positioning full-time-domain context semantic understanding on the sequence of the fixed antenna positioning coordinate-heading angle local time series semantic reasoning feature vectors to obtain the fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature vector as the fixed antenna positioning coordinate-heading angle full-time-domain semantic understanding feature; Based on the full-time domain semantic understanding features of the fixed antenna positioning coordinates-heading angle, the predicted values ​​of the boom rotation angle and the boom pitch angle are determined, and the bucket wheel excavator boom is controlled to work according to the predicted values ​​of the boom rotation angle and the boom pitch angle.

Images