Crane positioning anti-swing closed-loop control method and system

Abstract

The invention discloses a crane positioning anti-swing closed-loop control method and system, and relates to the field of crane control. The method comprises the following steps of: firstly, respectively calculating difference values between real-time positions and target positions of the crane at a plurality of preset time points to obtain a target position trend time sequence input vector, and then arranging motor driving power values at the plurality of preset time points into a motor driving power time sequence input vector according to a time dimension; then, local time sequence feature extraction and cause-fruit time sequence interaction correlation analysis are carried out on the target position trend time sequence input vector and the motor driving power time sequence input vector to obtain motor driving-position change time sequence interaction features, and finally, based on the motor driving-position change time sequence interaction features, the motor driving-position change time sequence interaction features are obtained. And determining that the motor driving power value at the current time point should be increased, decreased or kept, so that accurate positioning and anti-swing control can be realized, and the working efficiency and safety of the crane are improved.

Description

Technical Field

[0001] The present application relates to the field of cranes, and more specifically, to a crane positioning and anti-sway closed-loop control method and system. Background Art

[0002] Crane anti-sway control is an important technology to ensure the safe and efficient operation of the crane. The anti-sway control system can automatically eliminate the sway of the load during operation, reduce mechanical wear, shorten the load transfer time, and reduce the risk of safety accidents.

[0003] Traditional crane positioning control methods usually use sensors to measure the position of the crane, and control the movement speed of the crane through a motor driver to achieve position adjustment. For example, the Chinese patent with application number 202111015635.5 discloses a bridge crane positioning anti-sway control method, which obtains the swing amplitude value Ab of the load-bearing hook through a swing measurement module combined with a data processing module; the data acquisition module obtains the current load-bearing weight of the bridge crane; if the current load-bearing weight is greater than the maximum load-bearing weight Gcm; then an alarm signal is sent to the controller for overload alarm; if the current load-bearing weight is within the range of the maximum load-bearing weight Gcm, the data processing module compares the swing amplitude value Ab with the swing amplitude threshold; the data acquisition module obtains the current bridge crane operating speed, if the bridge crane operating speed is greater than the operating speed threshold corresponding to the swing amplitude threshold, the data processing module sends a signal to the controller, and the controller connects to the swing adjustment module to control the operating speed, completing the anti-sway control of the bridge crane.

[0004] However, due to the accuracy deviation of sensor measurement data and the lag effect of motor drive control, the traditional method cannot meet the requirements of control accuracy and real-time performance, resulting in poor positioning accuracy and anti-sway effect of the crane.

[0005] Therefore, a closed-loop control solution for crane positioning and anti-sway is needed. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a closed-loop control method and system for positioning and anti-swaying of a crane, which can achieve accurate positioning and anti-swaying control and improve the working efficiency and safety of the crane.

[0007] According to one aspect of the present application, a crane positioning anti-sway closed-loop control method is provided, which includes:

[0008] Acquire the real-time position of the crane at a plurality of predetermined time points within a predetermined time period collected by the sensor group, and the motor drive power values ​​at the plurality of predetermined time points;

[0009] Calculating the difference between the real-time position and the target position of the crane at the plurality of predetermined time points respectively to obtain a target position trend time series input vector;

[0010] Arranging the motor drive power values ​​at the plurality of predetermined time points into a motor drive power timing input vector according to the time dimension;

[0011] Performing local time series feature extraction on the target position tendency time series input vector and the motor drive power time series input vector respectively to obtain a sequence of target position tendency local time series feature vectors and a sequence of motor drive power local time series feature vectors;

[0012] Processing the sequence of the target position tendency local time series feature vectors and the sequence of the motor drive power local time series feature vectors to obtain a motor drive-position change time series interaction feature; and

[0013] Based on the motor drive-position change timing interaction characteristics, it is determined whether the motor drive power value at the current time point should be increased, decreased or maintained.

[0014] Specifically, local time series feature extraction is performed on the target position tendency timing input vector and the motor drive power timing input vector respectively to obtain a sequence of target position tendency local time series feature vectors and a sequence of motor drive power local time series feature vectors, including:

[0015] Performing vector segmentation on the target position approach timing input vector and the motor drive power timing input vector to obtain a sequence of target position approach local timing input vectors and a sequence of motor drive power local timing input vectors; and

[0016] The sequence of the target position tendency local timing input vector and the sequence of the motor drive power local timing input vector are respectively passed through a timing feature extractor based on a deep neural network model to obtain a sequence of the target position tendency local timing feature vector and a sequence of the motor drive power local timing feature vector.

[0017] Specifically, the temporal feature extractor based on the deep neural network model is a temporal feature extractor based on a one-dimensional convolutional layer.

[0018] Specifically, the sequence of the target position tendency local time series feature vectors and the sequence of the motor drive power local time series feature vectors are processed to obtain the motor drive-position change time series interaction feature, including:

[0019] Reshaping the sequence of the target position trend local time series feature vectors to obtain a target position trend full time domain aggregated feature map;

[0020] Reshaping the sequence of local time series feature vectors of the motor drive power to obtain a full-time domain aggregate feature map of the motor drive power;

[0021] Performing multimodal fine-grained semantic compensation encoding on the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain a motor drive-position change time series interaction feature map;

[0022] The motor drive-position change timing interaction feature map is feature flattened to obtain the motor drive-position change timing interaction feature vector.

[0023] Specifically, multimodal fine-grained semantic compensation encoding is performed on the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain a motor drive-position change time series interaction feature map, including:

[0024] Performing feature dissociation on the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain a set of target position tendency full-time domain aggregated feature matrices and a set of motor drive power full-time domain aggregated feature matrices;

[0025] Fine-grained semantic feature compensation and integration are performed on the set of the target position tendency full-time domain aggregated feature matrices and the set of the motor drive power full-time domain aggregated feature matrices to obtain the motor drive-position change timing interaction feature map.

[0026] Specifically, feature dissociation is performed on the target position tendency full-time domain aggregate feature map and the motor drive power full-time domain aggregate feature map to obtain a set of target position tendency full-time domain aggregate feature matrices and a set of motor drive power full-time domain aggregate feature matrices, including:

[0027] The target position tendency full time domain aggregated feature map and the motor drive power full time domain aggregated feature map are feature dissociated along the channel dimension of the target position tendency full time domain aggregated feature map and the motor drive power full time domain aggregated feature map to obtain a set of target position tendency full time domain aggregated feature matrices and a set of motor drive power full time domain aggregated feature matrices.

[0028] Specifically, a set of the target position tendency full-time domain aggregate feature matrices and a set of the motor drive power full-time domain aggregate feature matrices are integrated with fine-grained semantic feature compensation to obtain the motor drive-position change time series interaction feature map, including:

[0029] Input the target position tendency full-time domain aggregate feature matrix and the motor drive power full-time domain aggregate feature matrix of each corresponding channel dimension in the set of the target position tendency full-time domain aggregate feature matrix and the set of the motor drive power full-time domain aggregate feature matrix into the cross-domain semantic information extraction module guided by the twin network to obtain a set of motor drive-position change shared semantic feature matrices;

[0030] Performing semantic feature compensation on the set of motor drive-position change shared semantic feature matrices to obtain a set of motor drive-position change semantic compensation feature matrices;

[0031] Based on the set of the motor drive-position change semantic compensation feature matrices, the set of the motor drive-position change shared semantic feature matrices is semantically enhanced to obtain a motor drive-position change timing interaction feature graph as the motor drive-position change timing interaction feature.

[0032] Specifically, based on the motor drive-position change timing interaction characteristics, determining whether the motor drive power value at the current time point should be increased, decreased, or maintained includes:

[0033] The motor drive-position change time series interaction feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the motor drive power value at the current time point should be increased, decreased or maintained.

[0034] According to another aspect of the present application, a crane positioning anti-sway closed-loop control system is provided, comprising:

[0035] A data acquisition module, used to acquire the real-time position of the crane at a plurality of predetermined time points within a predetermined time period acquired by the sensor group, and the motor drive power values ​​at the plurality of predetermined time points;

[0036] A position difference calculation module, used for respectively calculating the difference between the real-time position of the crane and the target position at the plurality of predetermined time points to obtain a target position trend time series input vector;

[0037] A power value vectorization module, used for arranging the motor drive power values ​​at the plurality of predetermined time points into a motor drive power timing input vector according to a time dimension;

[0038] A local timing feature extraction module, used to extract local timing features from the target position tendency timing input vector and the motor drive power timing input vector respectively to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors;

[0039] A motor drive-position change timing interaction analysis module, used for processing the sequence of the target position tendency local timing feature vector and the sequence of the motor drive power local timing feature vector to obtain the motor drive-position change timing interaction feature;

[0040] The motor drive control module is used to determine whether the motor drive power value at the current time point should be increased, decreased or maintained based on the motor drive-position change timing interaction characteristics.

[0041] This application has at least the following technical effects:

[0042] Compared with the prior art, the crane positioning and anti-sway closed-loop control method and system provided in the present application first calculates the difference between the real-time position and the target position of the crane at multiple predetermined time points to obtain the target position trend timing input vector, then arranges the motor drive power values ​​at multiple predetermined time points according to the time dimension as a motor drive power timing input vector, and then performs local timing feature extraction and cause-effect timing interaction correlation analysis on the target position trend timing input vector and the motor drive power timing input vector to obtain the motor drive-position change timing interaction feature. Finally, based on the motor drive-position change timing interaction feature, it is determined whether the motor drive power value at the current time point should be increased, decreased or maintained, so as to achieve accurate positioning and anti-sway control and improve the working efficiency and safety of the crane. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present application.

[0044] Figure 1 The figure is a flow chart of a closed-loop control method for crane positioning and anti-swaying according to an embodiment of the present application.

[0045] Figure 2 Schematic diagram of the architecture of a crane positioning anti-sway closed-loop control method according to an embodiment of the present application.

[0046] Figure 3 It is a flowchart of sub-step S140 of the crane positioning anti-sway closed-loop control method according to an embodiment of the present application.

[0047] Figure 4 It is a block diagram of a crane positioning anti-sway closed-loop control system according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0049] As shown in this application and claims, 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.

[0050] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0051] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0052] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0053] In view of the above technical problems, in the technical solution of this application, a crane positioning anti-sway closed-loop control method is proposed. Figure 1 The figure is a flow chart of a closed-loop control method for crane positioning and anti-swaying according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the structure of the crane positioning anti-sway closed-loop control method according to an embodiment of the present application. Figure 1 and Figure 2As shown, the closed-loop control method for crane positioning and anti-sway according to the embodiment of the present application includes the following steps: S110, acquiring the real-time position of the crane at multiple predetermined time points within a predetermined time period collected by the sensor group, and the motor drive power values ​​at the multiple predetermined time points; S120, respectively calculating the difference between the real-time position of the crane and the target position at the multiple predetermined time points to obtain a target position trend time series input vector; S130, arranging the motor drive power values ​​at the multiple predetermined time points according to the time dimension as a motor drive power time series input vector; S140, respectively calculating the difference between the real-time position of the crane and the target position at the multiple predetermined time points; S150, arranging the motor drive power values ​​at the multiple predetermined time points according to the time dimension as a motor drive power time series input vector; S160, respectively calculating the difference between the real-time position of the crane and the target position at the multiple predetermined time points; S170, arranging the motor drive power values ​​at the multiple predetermined time points according to the time dimension as a motor drive power time series input vector; S180, respectively calculating the difference between the real-time position of the crane and the target position at the multiple predetermined time points; S190, arranging the motor drive power values ​​at the multiple predetermined time points according to the time dimension as a motor drive power time series input vector; S191, arranging the real-time position of the crane and the target position at the multiple predetermined time points according to the time dimension; S192, arranging the real-time position of the crane and the target position at the multiple predetermined time points according to the time dimension; S193, arranging the real-time position of the crane and the target position at the multiple predetermined time points according to the time dimension; S194, arranging the real-time position of the crane and the target position at the multiple predetermined time points according to the time dimension; S195, arranging the real-time position of the crane and the target position at the multiple predetermined time points according to the time dimension; S196, ar The target position tendency timing input vector and the motor drive power timing input vector are subjected to local timing feature extraction to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors; S150, the sequence of target position tendency local timing feature vectors and the sequence of motor drive power local timing feature vectors are processed to obtain a motor drive-position change timing interaction feature; S160, based on the motor drive-position change timing interaction feature, it is determined whether the motor drive power value at the current time point should be increased, decreased or maintained.

[0054] Specifically, in the technical solution of the present application, first, the real-time position of the crane at multiple predetermined time points within a predetermined time period collected by the sensor group, and the motor drive power values ​​at the multiple predetermined time points are obtained. Next, considering that in the positioning control process of the crane, the deviation between the current position of the crane and the target position and the change trend of the deviation are the key to the positioning control of the crane. Therefore, in order to obtain the trend information of the position change of the crane so as to perform more precise control, in the technical solution of the present application, the difference between the real-time position of the crane and the target position at the multiple predetermined time points is calculated respectively to obtain the target position trend time series input vector. By calculating the difference between the real-time position of the crane and the target position, the deviation of the current position of the crane relative to the target position can be understood. In particular, this difference can also be used to indicate in which direction the crane needs to move to approach the target position, that is, it provides information about the change direction and change trend of the current position of the crane relative to the target position. In this way, it can help to judge whether the current position of the crane is tending to the target position, and make corresponding adjustments according to the trend information.

[0055] Furthermore, considering that the motor drive power value has a dynamic change rule in time series in the time dimension, that is, the motor drive power values ​​at the multiple predetermined time points have a correlation in time series. Therefore, it is necessary to arrange the motor drive power values ​​at the multiple predetermined time points into a motor drive power time series input vector according to the time dimension, so as to integrate the distribution information of the motor drive power value in time series.

[0056] In the control process of the crane, there may be differences in the target position trend changes and motor drive power changes of the crane in different time periods. Therefore, the positioning and anti-sway control of the crane needs to take into account the target position trend changes and motor drive power changes in different time periods. Therefore, in order to capture the local timing characteristics of the target position trend and motor drive power of the crane in different time periods, it is necessary to vectorize the target position trend timing input vector and the motor drive power timing input vector, and decompose them into a sequence of multiple local timing input vectors to obtain a sequence of target position trend local timing input vectors and a sequence of motor drive power local timing input vectors. In this way, each local timing input vector represents the target position trend timing changes and motor drive power timing changes within a period of time, which is conducive to more fine analysis and processing of the local timing characteristics of these data to perform real-time control of the motor drive power.

[0057] Then, the sequence of the local time series input vectors of the target position tendency and the sequence of the local time series input vectors of the motor drive power are respectively subjected to feature mining in a time series feature extractor based on a one-dimensional convolutional layer to respectively extract the local time series feature distribution information of the target position tendency of the crane and the motor drive power value in the time dimension, thereby obtaining a sequence of the local time series feature vectors of the target position tendency and a sequence of the local time series feature vectors of the motor drive power.

[0058] Accordingly, if Figure 3 As shown, in step S140, local timing feature extraction is performed on the target position tendency timing input vector and the motor drive power timing input vector respectively to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors, including: S141, vector segmentation is performed on the target position tendency timing input vector and the motor drive power timing input vector to obtain a sequence of target position tendency local timing input vectors and a sequence of motor drive power local timing input vectors; and S142, the sequence of target position tendency local timing input vectors and the sequence of motor drive power local timing input vectors are respectively passed through a timing feature extractor based on a deep neural network model to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors.

[0059] Among them, in step S142, the time series feature extractor based on the deep neural network model is a time series feature extractor based on a one-dimensional convolutional layer.

[0060] It is worth mentioning that the one-dimensional convolution layer is a common layer in deep neural networks, which is used to process time series data or one-dimensional sequence data. It is one of the key components of the Convolutional Neural Network (CNN). In the one-dimensional convolution layer, the input data is a one-dimensional sequence, such as time series data or text data. This layer extracts local features in the input sequence by convolving the input data using a set of learnable convolution kernels (also called filters). Each convolution kernel is a small window, and the convolution operation is performed on the input sequence through a sliding window. The result calculated is called a feature map. The main function of the one-dimensional convolution layer is to capture local patterns and features in the input sequence through convolution operations. The size and number of convolution kernels can be adjusted as needed to adapt to different data characteristics. Through the combination of multiple convolution kernels, the one-dimensional convolution layer can extract features of different sizes and abstraction levels, thereby better representing the information of the input sequence. One-dimensional convolution layers are often used in combination with other types of layers (such as pooling layers and fully connected layers) to build more complex deep neural network models. In time series data processing tasks, one-dimensional convolutional layers can effectively capture the time series features of the input sequence, such as temporal patterns, trends, and periodicity.

[0061] In step S142, a one-dimensional convolutional layer-based temporal feature extractor is used to process the sequence of local temporal input vectors of target position tendency and the sequence of local temporal input vectors of motor drive power to extract their temporal features. In this way, the input sequence can be converted into a more expressive feature vector sequence, providing better input for subsequent tasks (such as target position prediction or motor drive power prediction).

[0062] In the positioning and anti-sway control of the crane, there is a mutual influence and interactive relationship between the motor drive power and the target position trend change of the crane. In addition, it is also considered that the mutual relationship between the motor drive power and the target position trend of the crane exists in each local time period, and the degree of mutual influence and correlation in each local time period is different. Therefore, in order to be able to more fully and accurately interact with the motor drive power and the target position trend information of the crane, so as to achieve more accurate motor drive power control, in the technical solution of the present application, a cause-effect timing interactor is further used to process the sequence of the local timing feature vector of the target position trend and the sequence of the local timing feature vector of the motor drive power to obtain the motor drive-position change timing interaction feature vector. It should be understood that the cause-effect timing interactor is a method for processing time series data, which can interact with two time series sequences to obtain the interaction characteristics between them. Therefore, in the technical solution of the present application, by using the cause-effect timing interactor, the timing characteristic information between the motor drive power and the target position trend information of the crane in each local time period can be interactively fused, thereby providing dynamic interaction information between the motor drive power and the target position trend information of the crane, which helps to more accurately position and prevent the crane from swaying.

[0063] Correspondingly, in step S150, the sequence of the target position tendency local timing feature vectors and the sequence of the motor drive power local timing feature vectors are processed to obtain the motor drive-position change timing interaction feature, including: using a cause-effect timing interactor to process the sequence of the target position tendency local timing feature vectors and the sequence of the motor drive power local timing feature vectors to obtain the motor drive-position change timing interaction feature vector as the motor drive-position change timing interaction feature.

[0064] More specifically, a cause-effect timing interactor is used to process the sequence of the target position tendency local timing feature vectors and the sequence of the motor drive power local timing feature vectors to obtain the motor drive-position change timing interaction feature vector as the motor drive-position change timing interaction feature, including: based on the correlation between the sequence of the target position tendency local timing feature vectors and the sequence of the motor drive power local timing feature vectors, attention enhancement is performed to obtain a sequence of attention-enhanced target position tendency local timing feature vectors and a sequence of attention-enhanced motor drive power local timing feature vectors; the feature vectors of corresponding positions in the sequence of the target position tendency local timing feature vectors and the sequence of the attention-enhanced target position tendency local timing feature vectors are fused to obtain the target position The method comprises the steps of: performing maximum pooling processing on the sequence of target position tendency local fusion feature vectors to obtain a target position tendency local fusion maximum pooling feature vector, and performing maximum pooling processing on the sequence of motor drive power local fusion feature vectors to obtain a motor drive power local fusion maximum pooling feature vector; and fusing the target position tendency local fusion maximum pooling feature vector and the motor drive power local fusion maximum pooling feature vector to obtain the motor drive-position change timing interaction feature vector.

[0065] Here, the sequence of the local time series feature vectors of the target position tendency and the sequence of the local time series feature vectors of the motor drive power respectively express the local time domain temporal correlation characteristics of the target position tendency values ​​and the motor drive power values ​​of the multiple predetermined time points in the local time domain determined by vector segmentation in the global time domain. Therefore, when the cause-effect temporal interactor is used to process the sequence of the local time series feature vectors of the target position tendency and the sequence of the local time series feature vectors of the motor drive power, the local time domain temporal correlation characteristics in the local time domain can be fused based on the cause-effect temporal interaction of the local time domain sequences in the global time domain. At this time, it is necessary to focus on the temporal feature sharing of the temporal correlation features of the sequence of the local time series feature vectors of the target position tendency and the sequence of the local time series feature vectors of the motor drive power based on the local time domain sequence distribution under the global time domain. In order to improve the temporal feature sharing of the temporal correlation features of the sequence of the local time series feature vectors of the target position tendency and the sequence of the local time series feature vectors of the motor drive power based on the local time domain sequence distribution under the global time domain distribution,

[0066] In a preferred embodiment of the present application, the process of processing the sequence of the target position tendency local timing feature vectors and the sequence of the motor drive power local timing feature vectors to obtain the motor drive-position change timing interaction feature vector includes: reshaping the feature shape of the sequence of the target position tendency local timing feature vectors to obtain the target position tendency full time domain aggregated feature map; reshaping the feature shape of the sequence of the motor drive power local timing feature vectors to obtain the motor drive power full time domain aggregated feature map; performing multimodal fine-grained semantic compensation encoding on the target position tendency full time domain aggregated feature map and the motor drive power full time domain aggregated feature map to obtain the motor drive-position change timing interaction feature map; and flattening the motor drive-position change timing interaction feature map to obtain the motor drive-position change timing interaction feature vector.

[0067] Specifically, multimodal fine-grained semantic compensation encoding is performed on the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain the motor drive-position change timing interaction feature map, including: feature dissociation of the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain a set of target position tendency full-time domain aggregated feature matrices and a set of motor drive power full-time domain aggregated feature matrices; fine-grained semantic feature compensation integration is performed on the set of target position tendency full-time domain aggregated feature matrices and the set of motor drive power full-time domain aggregated feature matrices to obtain the motor drive-position change timing interaction feature map.

[0068] Specifically, the target position tendency full-time domain aggregate feature map and the motor drive power full-time domain aggregate feature map are feature-dissociated to obtain a set of target position tendency full-time domain aggregate feature matrices and a set of motor drive power full-time domain aggregate feature matrices, including: along the channel dimension of the target position tendency full-time domain aggregate feature map and the motor drive power full-time domain aggregate feature map, the target position tendency full-time domain aggregate feature map and the motor drive power full-time domain aggregate feature map are feature-dissociated to obtain a set of target position tendency full-time domain aggregate feature matrices and a set of motor drive power full-time domain aggregate feature matrices. Specifically, the process can be expressed by the formula:

[0069] Decouple(F1)={F 11 ,F 12 ,...,F 1i ,...,F 1n}

[0070] Decouple(F2)={F 21 ,F 22 ,...,F 2i,...,F 2n}

[0071] Among them, F1 and F2 are the target position trend full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map, respectively. Decouple(·) is to perform feature dissociation operation on the feature map. 11 , F 12 , F 1i and F 1n are the first, second, i-th and n-th target position tending full-time domain aggregate feature matrices in the set of target position tending full-time domain aggregate feature matrices, F 21 , F 22 , F 2i and F 2n They are respectively the 1st, 2nd, i-th and n-th motor drive power full-time domain aggregated feature matrices in the set of motor drive power full-time domain aggregated feature matrices.

[0072] It should be understood that dissociating the feature graph along the channel dimension means separating different types of information that were originally integrated together (such as data from different sensors, different physical quantities, etc.) to form a series of independent but related feature matrices. Doing so helps to highlight the unique information carried by each channel and enables subsequent analysis to focus on this specific information. The target position trend full-time domain aggregated feature matrix set and the motor drive power full-time domain aggregated feature matrix set obtained by feature dissociation actually represent the position change trend of the crane and the change of the motor drive power at different times. Each matrix contains time series data on a specific aspect (i.e., a channel), which makes it possible to examine the behavior patterns of these aspects separately.

[0073] Specifically, fine-grained semantic feature compensation and integration are performed on the set of target position tendency full-time domain aggregated feature matrices and the set of motor drive power full-time domain aggregated feature matrices to obtain the motor drive-position change timing interaction feature map, including: inputting the target position tendency full-time domain aggregated feature matrix and the motor drive power full-time domain aggregated feature matrix of each corresponding channel dimension in the set of target position tendency full-time domain aggregated feature matrices and the set of motor drive power full-time domain aggregated feature matrices into a twin network-guided cross-domain semantic information extraction module to obtain a set of motor drive-position change shared semantic feature matrices; performing semantic feature compensation on the set of motor drive-position change shared semantic feature matrices to obtain a set of motor drive-position change semantic compensation feature matrices; based on the set of motor drive-position change semantic compensation feature matrices, semantically enhancing the set of motor drive-position change shared semantic feature matrices to obtain a motor drive-position change timing interaction feature map as the motor drive-position change timing interaction feature.

[0074] Specifically, the target position tendency full-time domain aggregate feature matrix and the motor drive power full-time domain aggregate feature matrix of each corresponding channel dimension in the set of the target position tendency full-time domain aggregate feature matrix and the set of the motor drive power full-time domain aggregate feature matrix are input into the cross-domain semantic information extraction module guided by the twin network to obtain a set of motor drive-position change shared semantic feature matrices. The process can be expressed as follows:

[0075]

[0076] Among them, ⊕, ⊙ and They are point-by-point addition, point-by-point multiplication, and point-by-point subtraction, Concat(·;·;· ) For cascade processing, Cov 1×1 (·) is the point convolutional coding, sigmoid(·) is the sigmoid function, S i Yes F 1i and F 2i The motor drive-position changes between the shared semantic feature matrix.

[0077] Specifically, when these fine-grained feature matrices are input into the shared semantic information extraction module based on the twin network structure, the model is actually learning the common patterns or common features between different features (such as position change trends and motor drive power). The twin network is particularly good at capturing this type of association, which can help identify potential connections between the two, such as how a specific position change trend is often accompanied by a motor power adjustment. Even under different operating conditions, such as different loads or environmental factors, the system is able to identify similar operating patterns and apply the knowledge learned from one situation to another, thereby promoting the transfer of cross-domain knowledge. In addition, since each feature matrix represents time series data of a specific aspect, the relationship between these data can be analyzed at a very detailed level. This not only deepens the understanding of the laws of crane movement, but also helps to discover subtle changes that are difficult to detect based on macro-level data alone. Through in-depth analysis of these fine-grained features, future behavioral trends can be predicted more accurately, providing a basis for more precise and smooth control.

[0078] Specifically, the set of motor drive-position change shared semantic feature matrices is subjected to semantic feature compensation to obtain a set of motor drive-position change semantic compensation feature matrices, including: inputting each motor drive-position change shared semantic feature matrix in the set of motor drive-position change shared semantic feature matrices into a semantic enhancement decoding module guided by a large language model to obtain a set of motor drive-position change semantic compensation text descriptions; inputting each motor drive-position change semantic compensation text description in the set of motor drive-position change semantic compensation text descriptions into a semantic encoder based on a text convolutional neural network model to obtain a set of motor drive-position change semantic compensation feature matrices. Specifically, the process can be expressed as follows:

[0079] T i =LLM{S i}

[0080] M i =TextCNN{T i}

[0081] Among them, T i For S i The corresponding motor drive-position change semantic compensation text description, LLM is the semantic compensation decoding operation, TextCNN is the text convolutional coding, M i For S i The corresponding motor drive-position change semantic compensation feature matrix.

[0082] That is, first, through the processing of the semantic enhancement decoding module guided by the large language model, the originally abstract motor drive-position change shared semantic feature matrix is ​​converted into a text description with clear meaning. These text descriptions not only contain the information carried by the original feature matrix, but also integrate the prior knowledge and context understanding ability of the large language model. This means that for each specific time point or operation state, the system can obtain a richer and more explanatory semantic representation, which helps to capture those subtle but important operation details, such as the subtle adjustment of the crane when approaching the target position or the immediate response to emergencies. Next, when these semantically compensated text descriptions are fed into the semantic encoder based on the text convolutional neural network model, Text CNN can efficiently extract and encode the text description. It constructs a new feature representation form, namely the semantic compensation feature matrix, based on the vocabulary, grammatical structure and logical relationship between sentences in the text. Compared with the original shared semantic feature matrix, the feature matrix after this processing emphasizes the semantic association between different operation states, and because of the intuitiveness and concreteness brought by the integration of text descriptions, these features are easier to be used in subsequent analysis and decision-making processes. Ultimately, through this process, the system can further enhance its understanding of the complex interaction between motor drive and position change. This enhancement is not only reflected in the improvement of data accuracy, but more importantly, it gives the system greater adaptability and flexibility.

[0083] Specifically, based on the set of the motor drive-position change semantic compensation feature matrices, the set of the motor drive-position change shared semantic feature matrices is semantically enhanced to obtain a motor drive-position change timing interaction feature graph as the motor drive-position change timing interaction feature. Specifically, the process can be expressed as follows:

[0084]

[0085] F f =couple{S b1 ,S b2 ,...,S bi ,...,S bn}

[0086] in, M i The transposed matrix of is matrix multiplication, D is M i The scale is the matrix width multiplied by the matrix height, softmax(·) is the softmax function, α and β are weighting coefficients, S b1 , S b2 , S bi and Sbn are the first, second, i-th and n-th motor drive-position change interaction semantic reinforcement feature matrices in the set of motor drive-position change interaction semantic reinforcement feature matrices, n is the number of feature matrices in the set of motor drive-position change interaction semantic reinforcement feature matrices, couple{·,·,...,·} is to aggregate the set of feature matrices along the channel dimension, F f It is the motor drive-position change timing interaction characteristic diagram.

[0087] Specifically, when the semantic compensation feature matrix processed by the semantic enhancement decoding module guided by the large language model is combined with the original shared semantic feature matrix, an additional semantic layer is actually added to the original one. These semantic compensation feature matrices not only retain the original time series features, but also introduce the intuitiveness and specificity brought by the text description, as well as the extensive background knowledge provided by the large language model. This combination enables the system to more accurately capture the complex relationship between motor drive and position change, especially in those subtle but critical operating moments, such as the deceleration adjustment when the crane approaches the target position or the immediate response to an emergency. By fusing the two feature matrix sets, a new feature representation that integrates multiple aspects of information is obtained - the motor drive-position change time series interaction feature map. This feature map is not just a static data snapshot, it reflects the operation dynamics that evolve over time, and because it incorporates semantic compensation information, it can better express the intrinsic connection between different operating states. For example, at a specific point in time, if the change in a motor power is always accompanied by a certain position trend, then such an association will be clearly reflected in the feature map. This provides a strong basis for subsequent analysis, allowing the system to make more intelligent and timely decisions based on these rich features. In this way, the output of the motor driver can be adjusted in real time through closed-loop control according to the difference between the actual position and the target position of the crane, thereby improving control accuracy and real-time control to achieve precise positioning and anti-sway control, and improve the working efficiency and safety of the crane.

[0088] Then, the motor drive-position change time series interaction feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the motor drive power value at the current time point should be increased, decreased, or maintained. In other words, the local time series interaction correlation feature information between the local time series change feature of the motor drive power and the local time series change feature of the target position trend of the crane is used for classification processing to adjust the output of the motor driver according to the difference between the actual position and the target position of the crane. In this way, the accuracy of the sensor measurement data can be improved, and the control hysteresis effect of the motor driver can be avoided, thereby achieving more accurate positioning and anti-sway control.

[0089] Accordingly, in step S160, based on the motor drive-position change timing interaction characteristics, it is determined whether the motor drive power value at the current time point should be increased, decreased or maintained, including: passing the motor drive-position change timing interaction feature vector through a classifier to obtain a classification result, and the classification result is used to indicate that the motor drive power value at the current time point should be increased, decreased or maintained.

[0090] Specifically, the motor drive-position change timing interaction feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the motor drive power value at the current time point should be increased, decreased or maintained, including: using the fully connected layer of the classifier to fully connect encode the motor drive-position change timing interaction feature vector to obtain an encoded classification feature vector; and inputting the encoded classification feature vector into the Softmax classification function of the classifier to obtain the classification result.

[0091] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression and SVM are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification. However, this is prone to errors and is not efficient. Commonly used multi-classification methods include the Softmax classification function.

[0092] In summary, the closed-loop control method for crane positioning and anti-swaying based on the embodiment of the present application is explained, which can achieve precise positioning and anti-swaying control and improve the working efficiency and safety of the crane.

[0093] Figure 4 FIG. 1 is a block diagram of a crane positioning anti-sway closed-loop control system 100 according to an embodiment of the present application. Figure 4As shown, the crane positioning anti-sway closed-loop control system 100 according to the embodiment of the present application includes: a data acquisition module 110, which is used to obtain the real-time position of the crane at multiple predetermined time points within a predetermined time period collected by the sensor group, and the motor drive power values ​​at the multiple predetermined time points; a position difference calculation module 120, which is used to respectively calculate the difference between the real-time position of the crane at the multiple predetermined time points and the target position to obtain the target position trend time series input vector; a power value vectorization module 130, which is used to arrange the motor drive power values ​​at the multiple predetermined time points into a motor drive power time series input vector according to the time dimension; a local time series feature extraction module 140, which is used to Local timing feature extraction is performed on the target position tendency timing input vector and the motor drive power timing input vector respectively to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors; a motor drive-position change timing interaction analysis module 150 is used to process the sequence of target position tendency local timing feature vectors and the sequence of motor drive power local timing feature vectors to obtain motor drive-position change timing interaction characteristics; and a motor drive control module 160 is used to determine whether the motor drive power value at the current time point should be increased, decreased or maintained based on the motor drive-position change timing interaction characteristics.

[0094] In one example, in the above-mentioned crane positioning and anti-sway closed-loop control system 100, the local timing feature extraction module 140 includes: a vector segmentation unit, which is used to perform vector segmentation on the target position tendency timing input vector and the motor drive power timing input vector to obtain a sequence of target position tendency local timing input vectors and a sequence of motor drive power local timing input vectors; and a timing feature extraction unit, which is used to pass the sequence of target position tendency local timing input vectors and the sequence of motor drive power local timing input vectors through a timing feature extractor based on a deep neural network model respectively to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors.

[0095] The specific functions and operations of each module in the above-mentioned crane positioning anti-sway closed-loop control system 100 have been referred to above. Figures 1 to 3 The description of the crane positioning anti-sway closed-loop control method has been introduced in detail, and therefore, its repeated description will be omitted.

[0096] As described above, the crane positioning anti-sway closed-loop control system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a crane positioning anti-sway closed-loop control algorithm. In one example, the crane positioning anti-sway closed-loop control system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the crane positioning anti-sway closed-loop control system 100 can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the crane positioning anti-sway closed-loop control system 100 can also be one of the many hardware modules of the wireless terminal.

[0097] Alternatively, in another example, the crane positioning anti-sway closed-loop control system 100 and the wireless terminal may also be separate devices, and the crane positioning anti-sway closed-loop control system 100 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.

[0098] Various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0099] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0100] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application are described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. A closed-loop control method for crane positioning and anti-swaying, characterized in that: include: Acquire the real-time position of the crane at a plurality of predetermined time points within a predetermined time period collected by the sensor group, and the motor drive power values ​​at the plurality of predetermined time points; Calculating the difference between the real-time position and the target position of the crane at the plurality of predetermined time points respectively to obtain a target position trend time series input vector; Arranging the motor drive power values ​​at the plurality of predetermined time points into a motor drive power timing input vector according to the time dimension; Performing local time series feature extraction on the target position tendency time series input vector and the motor drive power time series input vector respectively to obtain a sequence of target position tendency local time series feature vectors and a sequence of motor drive power local time series feature vectors; Processing the sequence of the target position tendency local time series feature vectors and the sequence of the motor drive power local time series feature vectors to obtain a motor drive-position change time series interaction feature; Based on the motor drive-position change timing interaction characteristics, it is determined whether the motor drive power value at the current time point should be increased, decreased or maintained.

2. The crane positioning anti-sway closed-loop control method according to claim 1 is characterized in that: The local time series feature extraction is performed on the target position tendency time series input vector and the motor drive power time series input vector respectively to obtain a sequence of the target position tendency local time series feature vector and a sequence of the motor drive power local time series feature vector, including: Performing vector segmentation on the target position approach timing input vector and the motor drive power timing input vector to obtain a sequence of target position approach local timing input vectors and a sequence of motor drive power local timing input vectors; and The sequence of the target position tendency local timing input vector and the sequence of the motor drive power local timing input vector are respectively passed through a timing feature extractor based on a deep neural network model to obtain a sequence of the target position tendency local timing feature vector and a sequence of the motor drive power local timing feature vector.

3. The crane positioning anti-sway closed-loop control method according to claim 2 is characterized in that: The temporal feature extractor based on the deep neural network model is a temporal feature extractor based on a one-dimensional convolutional layer.

4. The crane positioning anti-sway closed-loop control method according to claim 3 is characterized in that: The sequence of the target position tendency local time series feature vectors and the sequence of the motor drive power local time series feature vectors are processed to obtain the motor drive-position change time series interaction feature, including: Reshaping the sequence of the target position trend local time series feature vectors to obtain a target position trend full time domain aggregated feature map; Reshaping the sequence of local time series feature vectors of the motor drive power to obtain a full-time domain aggregate feature map of the motor drive power; Performing multimodal fine-grained semantic compensation encoding on the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain a motor drive-position change time series interaction feature map; The motor drive-position change timing interaction feature map is feature flattened to obtain the motor drive-position change timing interaction feature vector.

5. The crane positioning anti-sway closed-loop control method according to claim 4 is characterized in that: The target position trend full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map are subjected to multimodal fine-grained semantic compensation encoding to obtain a motor drive-position change time series interaction feature map, including: Performing feature dissociation on the target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map to obtain a set of target position tendency full-time domain aggregated feature matrices and a set of motor drive power full-time domain aggregated feature matrices; Fine-grained semantic feature compensation and integration are performed on the set of the target position tendency full-time domain aggregated feature matrices and the set of the motor drive power full-time domain aggregated feature matrices to obtain the motor drive-position change timing interaction feature map.

6. The crane positioning anti-sway closed-loop control method according to claim 5, characterized in that: The target position tendency full-time domain aggregated feature map and the motor drive power full-time domain aggregated feature map are subjected to feature dissociation to obtain a set of target position tendency full-time domain aggregated feature matrices and a set of motor drive power full-time domain aggregated feature matrices, including: The target position tendency full time domain aggregated feature map and the motor drive power full time domain aggregated feature map are feature dissociated along the channel dimension of the target position tendency full time domain aggregated feature map and the motor drive power full time domain aggregated feature map to obtain a set of target position tendency full time domain aggregated feature matrices and a set of motor drive power full time domain aggregated feature matrices.

7. The crane positioning anti-sway closed-loop control method according to claim 6, characterized in that: The set of the target position tendency full-time domain aggregate feature matrix and the set of the motor drive power full-time domain aggregate feature matrix are integrated with fine-grained semantic feature compensation to obtain the motor drive-position change time series interaction feature map, including: Input the target position tendency full-time domain aggregate feature matrix and the motor drive power full-time domain aggregate feature matrix of each corresponding channel dimension in the set of the target position tendency full-time domain aggregate feature matrix and the set of the motor drive power full-time domain aggregate feature matrix into the cross-domain semantic information extraction module guided by the twin network to obtain a set of motor drive-position change shared semantic feature matrices; Performing semantic feature compensation on the set of motor drive-position change shared semantic feature matrices to obtain a set of motor drive-position change semantic compensation feature matrices; Based on the set of the motor drive-position change semantic compensation feature matrices, the set of the motor drive-position change shared semantic feature matrices is semantically enhanced to obtain a motor drive-position change timing interaction feature graph as the motor drive-position change timing interaction feature.

8. The crane positioning anti-sway closed-loop control method according to claim 7, characterized in that: Based on the motor drive-position change timing interaction characteristics, determining whether the motor drive power value at the current time point should be increased, decreased, or maintained includes: The motor drive-position change time series interaction feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the motor drive power value at the current time point should be increased, decreased or maintained.

9. Crane positioning anti-sway closed-loop control system, characterized in that: include: A data acquisition module, used to acquire the real-time position of the crane at a plurality of predetermined time points within a predetermined time period acquired by the sensor group, and the motor drive power values ​​at the plurality of predetermined time points; A position difference calculation module, used for respectively calculating the difference between the real-time position of the crane and the target position at the plurality of predetermined time points to obtain a target position trend time series input vector; A power value vectorization module, used for arranging the motor drive power values ​​at the plurality of predetermined time points into a motor drive power timing input vector according to a time dimension; A local timing feature extraction module, used to extract local timing features from the target position tendency timing input vector and the motor drive power timing input vector respectively to obtain a sequence of target position tendency local timing feature vectors and a sequence of motor drive power local timing feature vectors; A motor drive-position change timing interaction analysis module, used for processing the sequence of the target position tendency local timing feature vector and the sequence of the motor drive power local timing feature vector to obtain the motor drive-position change timing interaction feature; The motor drive control module is used to determine whether the motor drive power value at the current time point should be increased, decreased or maintained based on the motor drive-position change timing interaction characteristics.

10. The crane positioning anti-sway closed-loop control system according to claim 9, characterized in that: The local temporal feature extraction module comprises: A vector segmentation unit, used for performing vector segmentation on the target position tendency timing input vector and the motor drive power timing input vector to obtain a sequence of target position tendency local timing input vectors and a sequence of motor drive power local timing input vectors; A timing feature extraction unit is used to pass the sequence of the target position tendency local timing input vector and the sequence of the motor drive power local timing input vector through a timing feature extractor based on a deep neural network model to obtain the sequence of the target position tendency local timing feature vector and the sequence of the motor drive power local timing feature vector.

Images