Scraper conveyer attitude prediction method

Abstract

The invention discloses a scraper conveyor attitude prediction method, and belongs to the technical field of equipment monitoring, and the method comprises the steps: S1, collecting multi-source data, including equipment operation state data, support attitude data and equipment physical structure parameters; s2, performing time synchronization on the acquired equipment operation state data and the bracket attitude data, and further calculating the bottom cutting depth of a bracket chute and an advanced state vector of a pushing and sliding bracket; s3, according to the calculated bottom cutting depth and the advanced state vector, extracting time sequence characteristics of attitude data and bracket attitude data during bottom cutting of the coal mining machine through the first sub-network; extracting structural features of the physical structural parameters of the equipment through the second sub-network; and S4, splicing the extracted time sequence features and structural features, and outputting an attitude prediction value of the scraper conveyor through an attitude prediction model. According to the method, the deep learning prediction model is constructed by fusing multi-source heterogeneous data, so that the problems of low prediction precision, poor real-time performance and incapability of dynamic optimization in the prior art are solved.

Description

Technical Field

[0001] The invention belongs to the technical field of equipment monitoring, and in particular relates to a method for predicting the posture of a scraper conveyor. Background Art

[0002] In the process of comprehensive mechanized mining, the scraper conveyor is a key equipment for coal transportation. Its posture directly affects the transportation efficiency and equipment safety. Currently, the posture monitoring methods of the scraper conveyor include: (1) Based on the spline interpolation method and Kalman filter algorithm, the real-time position and posture of the scraper conveyor are accurately simulated to improve the reliability of unmanned mining.

[0003] (2) By obtaining the shape of the scraper conveyor and the structural parameters of the shearer, the pitch angle obtained during the actual operation of the shearer through the strapdown inertial navigation system and the dual-axis tilt sensor is reversely mapped to the distance the shearer travels on the scraper conveyor at a certain moment, thereby positioning the shearer. It is necessary to install a tilt sensor on the scraper conveyor chute. This method is currently rarely used in actual applications. The tilt sensor is mainly installed on the bracket, which also has limitations.

[0004] (3) By calculating the difference between the hydraulic support stroke value and the slope of the curved section, the position of the scraper conveyor can be accurately identified and the positioning accuracy can be improved. This method is simple and easy to use, but its accuracy will be greatly affected by terrain fluctuations.

[0005] The above-mentioned existing methods have shortcomings such as poor adaptability and difficulty in implementation. At the same time, they lack verification and feedback mechanisms for prediction results, and are unable to dynamically adjust the prediction model according to actual working conditions, resulting in continuous accumulation of prediction errors after long-term operation. Summary of the Invention

[0006] In response to the above-mentioned deficiencies in the prior art, the scraper conveyor posture prediction method provided by the present invention solves the problems of poor adaptability and implementation difficulties of the existing methods, lack of verification and feedback mechanism for prediction results, and inability to dynamically adjust the prediction process according to actual working conditions, resulting in the continuous accumulation of prediction errors after long-term operation.

[0007] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is: a method for predicting the posture of a scraper conveyor, comprising the following steps: S1. Collect multi-source data, including equipment operation status data, bracket posture data, and equipment physical structure parameters; S2. Time synchronization of the collected equipment operation status data and support posture data, and then calculation of the bottom cutting depth of the support chute and the leading state vector of the push-slide support; S3. Based on the calculated cutting depth and the advance state vector, the first sub-network is used to extract the time series features of the shearer's cutting posture data and the support posture data, including the posture angle change characteristics when the support is pushed, the correlation characteristics between the push rod length and the cutting depth, and the influence characteristics of the adjacent supports on the current support. extracting structural features of physical structural parameters of the device through the second sub-network; S4. The extracted time series features and structural features are spliced ​​together, and the attitude prediction value of the scraper conveyor is output through the attitude prediction model.

[0008] Furthermore, in step S2, the time synchronization processing of the equipment operation status data and the bracket posture data is performed, including: S201, performing clock synchronization on a data acquisition device for collecting device operation status data based on the NTP protocol; S202: Based on clock synchronization, the trigger position of the support pushing and sliding action is used as a reference to align the timestamps of the posture data of each support, and align the spatial positions of the shearer cutting data corresponding to the support; S203 : Based on the alignment of the timestamp and the spatial position, unified resampling is performed to obtain time-synchronized device operation status data and bracket posture data.

[0009] Furthermore, the step S202 is specifically as follows: When the bracket completes the pulling action and triggers the completion flag, the current global timestamp is recorded and bracket position , and when the push-slide action flag is ready to be triggered, the time for preparing the push-slide work is recorded as ; Data points of any bracket attitude sensor , take the time lie in and The data values ​​between them are used, and the posture data of the bracket is obtained through median filtering to complete the timestamp alignment of the posture data of each bracket and the spatial position alignment of the coal cutting machine bottom data; wherein, Indicates the corrected timestamp, Indicates the data value collected by the attitude sensor.

[0010] Furthermore, in step S2, the cutting depth of the support chute is for: Where, Indicates the vertical height of the rocker arm end, , Indicates the cutting radius of the shearer drum, Indicates the height of the fuselage connection point from the bottom plate. Indicates the rocker arm length, Indicates the pitch angle of the coal mining machine body, Indicates the inclination angle of the rocker arm relative to the fuselage; The method for determining the leading state vector of the push-and-slide support is: Setting the advanced state threshold ; when When , the leading state is that the left frame leads; when When , the leading state is that the right frame is ahead; when When , the leading state is no offset; The determined advance state is uniquely encoded to obtain the advance state vector .

[0011] Furthermore, in step S3, the first sub-network is an LSTM model, which extracts the temporal features of the input data through a set number of neurons and outputs the hidden state of the last time step; The input data is , represents the number of samples, represents the time step, The features of each time step include the attitude angle of the current bracket, the cutting depth, the push rod length, the attitude angles of the left and right brackets, the three dimensions of the leading state vector, and the position; the attitude angles include the pitch angle, the roll angle, and the heading angle.

[0012] Furthermore, in step S3, the second sub-network is a PSFC network, and the PSFC network includes a dimension expansion layer, a first fully connected layer, a ReLU activation function, a second fully connected layer and an output layer connected in sequence; The dimension expansion layer expands the dimension of the input data by constructing implicit relationships that derive explicit encoding physical parameters; The input data of the PSFC network include support width, slide length, maximum horizontal bending angle, maximum vertical full angle and shearer sliding shoe spacing; After the dimension expansion of the dimension expansion layer, the data input to the first fully connected layer also includes the maximum offset in the horizontal direction, the maximum fluctuation in the vertical direction, and the number of sliding shoe coverage nodes.

[0013] Furthermore, in step S4, the posture prediction model includes a third fully connected layer and a fourth fully connected layer connected in sequence.

[0014] Furthermore, the loss function of the posture prediction model is for: Where, represents the forecast error term, represents the physical constraint term, represents the total constraint weight, represents the position constraint, Represents an angular constraint.

[0015] Furthermore, the angle constraint term for: Where, Represents the horizontal constraint, Indicates the vertical constraint. and Respectively and The constraint weight coefficient, Indicates the maximum horizontal bending angle of the scraper conveyor. Indicates the maximum vertical bending angle of the scraper conveyor. Indicates the vertical angle difference of the scraper conveyor, Indicates the vertical angle difference of the scraper conveyor, subscript Indicates the chute sequence number of the scraper conveyor. Indicates the total number of chutes on the scraper conveyor.

[0016] Furthermore, the position constraint for: Where, Represents the horizontal position constraint, Represents a vertical position constraint, and Respectively and The constraint weight coefficient, and Respectively represent the horizontal offset and vertical offset of the scraper conveyor, and Respectively represent the maximum horizontal and vertical offset of the scraper conveyor, the subscript Indicates the chute sequence number of the scraper conveyor. Indicates the total number of chutes on the scraper conveyor.

[0017] The beneficial effects of the present invention are: (1) The morphology of the scraper conveyor is affected by nonlinear factors such as the pushing sequence, the coordinated action of the supports, and the resistance of the coal seam. Traditional physical models are difficult to accurately describe (such as the friction at the hinge of the middle trough and the coupling effect of multiple pushers). The present invention uses the LSTM model to automatically learn the implicit mapping between multiple variables through data-driven learning, which is particularly suitable for processing the dynamic relationship between the support pushing sequence and the scraper morphology.

[0018] (2) When the scraper conveyor is pushed, it will be affected by the left and right brackets, and will deviate in the direction of the pushed slide, thereby affecting the posture change of the scraper conveyor. In the method of the present invention, the relative position of the pushing bracket and the adjacent bracket is quantified through three states: left bracket ahead, right bracket ahead, and no deviation. By quantifying the physical characteristics, the accuracy of obtaining the scraper conveyor posture change data is improved.

[0019] (3) In addition to extracting the temporal features of posture data through the LSTM model, the present invention also designs a PSFC network to extract the structural features of the physical structure parameters of the device, and splices them with the temporal features as the basis for subsequent posture prediction. Compared with the traditional method of only extracting temporal features to predict posture data, the physical structure features of the device are extracted through a separate PSFC network, which improves the accuracy of the posture prediction results while reducing the network complexity.

[0020] (4) The present invention trains the posture prediction model by combining the loss function of physical constraints, so that the model can automatically adjust the parameters according to the actual error, and the limitation provides the accuracy of the long-term operation of the method, so that the posture prediction results conform to the laws of mechanical motion, and enhance the reliability of the method of the present invention in actual production. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flow chart of the scraper conveyor posture prediction method provided by the present invention. DETAILED DESCRIPTION

[0022] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0023] The embodiment of the present invention provides a method for predicting the posture of a scraper conveyor. Figure 1 As shown, the following steps are included: S1. Collect multi-source data, including equipment operation status data, bracket posture data, and equipment physical structure parameters; S2. Time synchronization of the collected equipment operation status data and support posture data, and then calculation of the bottom cutting depth of the support chute and the leading state vector of the push-slide support; S3. Based on the calculated cutting depth and the advance state vector, the first sub-network is used to extract the time series features of the shearer's cutting posture data and the support posture data, including the posture angle change characteristics when the support is pushed, the correlation characteristics between the push rod length and the cutting depth, and the influence characteristics of the adjacent supports on the current support. extracting structural features of physical structural parameters of the device through the second sub-network; S4. The extracted time series features and structural features are spliced ​​together, and the attitude prediction value of the scraper conveyor is output through the attitude prediction model.

[0024] In step S1 of an embodiment of the present invention, the collected equipment operation status data include coal mining machine body inertial navigation data, inertial measurement unit data, coal mining machine attitude data, support attitude data, support push rod stroke data and the position of the side support of the push-slide support; wherein, the coal mining machine body inertial navigation data includes acceleration data and angular velocity data obtained by the inertial measurement unit, the coal mining machine attitude data includes the body attitude and left and right roller lying status data obtained in real time by a high-precision angle sensor, the support attitude data includes the bottom plate pitch angle, roll angle and heading angle data measured by the inclinometer, the support push rod stroke data is collected by a magnetostrictive displacement sensor, and the leading state of the left and right frames of the push-slide support is judged by recording the frame pulling and pushing action.

[0025] In step S1 of the embodiment of the present invention, the collected support posture data is the support posture data of the previous N push and slide times, which is used to analyze the posture change trend, where N = support length / single push and slide length. In step S1 of the embodiment of the present invention, the collected physical structural parameters include coal mining machine size data (arm length and fuselage size), support size data (width and bottom plate size) and scraper conveyor size data (fuselage height, chute width and chute gap).

[0026] In step S2 of the embodiment of the present invention, a dual-dimensional alignment method based on the bracket position and timestamp is adopted, with the bracket push-slide trigger position being prioritized as a benchmark, and the NTP protocol is combined to achieve multi-source data synchronization.

[0027] Specifically, in step S2 of this embodiment, the time synchronization processing of the device operation status data and the bracket posture data is performed, including: S201, performing clock synchronization based on the NTP protocol on a data acquisition device that collects device operating status data and bracket posture data; S202: Based on clock synchronization, the trigger position of the support pushing and sliding action is used as a reference to align the timestamps of the posture data of each support, and align the spatial positions of the shearer cutting data corresponding to the support; S203 : Based on the alignment of the timestamp and the spatial position, unified resampling is performed to obtain time-synchronized device operation status data and bracket posture data.

[0028] In step S201 of this embodiment, each data acquisition device synchronizes with the ground clock server through the NTP protocol to correct the local clock deviation; wherein the timestamp correction formula is: Where, represents the corrected global unified timestamp, Indicates the local original timestamp of the device. Indicates the clock offset calculated by the NTP protocol ,in The time when the client sends the request. The server receives the time. is the server response time, The time it takes for the client to receive a response.

[0029] Step S202 in this embodiment is specifically as follows: When the bracket completes the pulling action and triggers the completion flag, the current global timestamp is recorded and bracket position , and when the push-slide action flag is ready to be triggered, the time for preparing the push-slide work is recorded as ; Among them, the bracket position Identified by bracket number or absolute coordinates; Data points of any bracket attitude sensor , take the time lie in and The data values ​​between them are used, and the posture data of the bracket is obtained through median filtering to complete the timestamp alignment of the posture data of each bracket and the spatial position alignment of the coal cutting machine bottom data; wherein, Indicates the corrected timestamp, Indicates the data value collected by the attitude sensor.

[0030] In this embodiment, the bottom cutting data is calculated from the posture data of the coal shearer. When the front drum of the coal shearer is cutting the bottom of the N bracket, the N-20 bracket may be preparing to push and slide.

[0031] In this embodiment, spatial alignment of each support's posture data primarily aligns the support's posture data with the shearer drum's bottom-cutting data. This is because the shearer's position refers to the center point of the shearer, which corresponds to a support number. The drums, however, are located on either side of the shearer, and their corresponding support numbers are not the shearer's. The support numbers corresponding to the left and right drums can be calculated from the shearer's supports, thereby achieving spatial alignment of the support's posture data. Regarding temporal alignment, generally speaking, the support's bottom-cutting time is earlier than the support's pull-out time, which is earlier than the support's push-out time. Since push-out calculations are required, temporal data alignment is necessary.

[0032] In step S2 of the embodiment of the present invention, the cutting depth of the corresponding support chute is calculated by a geometric algorithm based on the shearer arm inclination angle and the body state, combined with the shearer size data; specifically, the cutting depth is defined as the vertical distance between the cutting point at the end of the shearer arm and the bottom surface of the scraper conveyor chute. for: Where, Indicates the vertical height of the rocker arm end, , Indicates the cutting radius of the shearer drum, Indicates the height of the fuselage connection point from the bottom plate. Indicates the rocker arm length, Indicates the pitch angle of the coal mining machine body, Indicates the inclination angle of the rocker arm relative to the fuselage.

[0033] In this embodiment, the relative position of the push-and-slide support and the adjacent support is quantified by three states: "left support leading / right support leading / no offset"; specifically, the method for determining the leading state vector of the push-and-slide support is: Setting the advanced state threshold ; when When , the leading state is that the left frame leads; when When , the leading state is that the right frame is ahead; when When , the leading state is no offset; The determined advance state is uniquely encoded to obtain the advance state vector .

[0034] In step S3 of the embodiment of the present invention, the first sub-network is an LSTM model, which extracts the temporal features of the input data through a set number of neurons and outputs the hidden state of the last time step; Specifically, the LSTM model in this embodiment sets one hidden layer, each layer has 64 neurons, and the input data is , represents the number of samples, represents the time step, The features of each time step include the attitude angle of the current bracket, the cutting depth, the push rod length, the attitude angles of the left and right brackets, the three dimensions of the leading state vector, and the position; the attitude angles include the pitch angle, the roll angle, and the heading angle.

[0035] In this embodiment, taking the time step of 3 as an example, the LSTM model outputs the hidden state of the last time step (shape (N, 64)), and extracts the time series features of the three pushing and sliding data, including the posture angle change characteristics of the bracket when pushing and sliding, the correlation characteristics between the push rod length and the bottom cutting depth, and the influence characteristics of the adjacent brackets on the current bracket.

[0036] In this embodiment, the LSTM model solves the gradient vanishing and gradient exploding problems of traditional recurrent neural networks (RNNs) through a gating mechanism. The forget gate determines which information to discard, the input gate controls the addition of new information, and the output gate determines the output content. The cell state is responsible for long-term information transmission. It can effectively process the temporal characteristics of the coal mining machine bottom cutting data and the historical posture data of the support, and learn the changes in data over time.

[0037] In step S3 of the embodiment of the present invention, the second sub-network is a PSFC network, wherein the PSFC network includes a dimension expansion layer, a first fully connected layer, a ReLU activation function, a second fully connected layer, and an output layer connected in sequence; Among them, the dimension expansion layer expands the dimension of the input data by constructing implicit relationships that explicitly encode physical parameters, allowing the PSFC network to learn richer structural information.

[0038] The input data of the PSFC network include support width, slide length, maximum horizontal bending angle, maximum vertical full angle and shearer skid shoe spacing; After the dimension expansion of the dimension expansion layer, the data input to the first fully connected layer also includes the maximum horizontal offset, the maximum vertical fluctuation, and the number of sliding shoe covered sliding joints. The maximum horizontal offset is the maximum allowable misalignment of adjacent sliding joints in the horizontal direction, which is expressed as The maximum vertical fluctuation is the maximum allowable vertical fluctuation between adjacent chute sections, expressed as The number of sliding shoe sections covered is the number of middle slots affected by the shearer. .

[0039] In this embodiment, in the PSFC network, the first fully connected layer maps the 8-dimensional input to 16 dimensions to preliminarily extract the linear / nonlinear relationship between parameters; the ReLU activation function introduces nonlinearity to avoid the inability of pure linear transformation to model complex physical constraints; the second fully connected layer maps the 16-dimensional features to 32 dimensions and outputs the final structural features.

[0040] In this embodiment, the PSFC network performs linear transformation and activation operations on the physical structure parameters, converting them into a representation consistent with the dimensions of other feature vectors, and then splicing them with the time series features output by the LSTM network, so that the subsequent posture prediction model can comprehensively consider the impact of the physical characteristics of the equipment on the scraper conveyor posture when predicting the posture data.

[0041] In step S4 of an embodiment of the present invention, the posture prediction model includes a third fully connected layer and a fourth fully connected layer connected in sequence, mapping the feature vector of the spliced ​​temporal features and structural features into a posture prediction value of the scraper conveyor, including pitch angle, roll angle, heading angle and offset.

[0042] In this embodiment of the present invention, a transfer learning algorithm is used to update the parameters of the posture prediction model based on the prediction error. The Adam optimizer (learning rate 0.001, batch size 64) is used for training for 100 epochs, and an early stopping strategy is applied to prevent overfitting.

[0043] In this embodiment, the physical constraint term is introduced to optimize the loss function, and the loss function of the posture prediction model is obtained. for: Where, represents the forecast error term, represents the physical constraint term, represents the total constraint weight, represents the position constraint, Represents an angular constraint.

[0044] In this embodiment, when calculating the angle constraint term, it is assumed that the scraper conveyor consists of n sections, and the predicted posture parameters of each section include the horizontal angle (heading angle) , the vertical angle (pitch angle) is , the angle difference between adjacent sliding sections includes the horizontal angle difference and vertical angle difference ; Based on this, at the maximum horizontal bending angle and maximum vertical bending angle Based on the angle constraint for: Where, Represents the horizontal constraint, Represents the vertical constraint item, and Respectively and The constraint weight coefficient, Indicates the maximum horizontal bending angle of the scraper conveyor. Indicates the maximum vertical bending angle of the scraper conveyor. Indicates the vertical angle difference of the scraper conveyor, Indicates the vertical angle difference of the scraper conveyor, subscript Indicates the chute sequence number of the scraper conveyor. Indicates the total number of chutes on the scraper conveyor.

[0045] In this embodiment, when calculating the position constraint, the three-dimensional coordinates of each chute of the scraper conveyor are , define the horizontal position offset as the difference in the projection distance of adjacent chute sections on the horizontal plane, and the vertical displacement offset as the difference in the projection distance of adjacent chute sections on the vertical plane, and then calculate the displacement offset of adjacent chute sections including the horizontal offset and vertical offset ; According to the length L of the slide and the maximum bending angle ( , ), calculate the maximum allowable offset of the mechanical structure including the maximum horizontal offset and vertical maximum offset ; Based on this, the position constraint for: Where, Represents the horizontal position constraint, Represents a vertical position constraint, and Respectively and The constraint weight coefficient, and Respectively represent the horizontal offset and vertical offset of the scraper conveyor, and Respectively represent the maximum horizontal and vertical offset of the scraper conveyor, the subscript Indicates the chute number of the scraper conveyor. Indicates the total number of chutes on the scraper conveyor.

[0046] In the above-mentioned loss function provided by an embodiment of the present invention, the horizontal and vertical angle differences and position differences between each pair of adjacent slide joints are calculated by designing physical constraint items. For the angle differences and positions that exceed the physical limits, the max function is used to calculate the penalty value; on this basis, the penalty values ​​of all adjacent slide joint pairs are summed to obtain the total physical constraint item, and the strength of the physical constraint is controlled by the weight coefficient.

[0047] The loss function provided by this invention ensures that when the predicted angle and position differences between adjacent joints are within the physically permitted range, the constraint term is zero, unaffecting training. When the differences exceed the physical limits, the constraint term incurs a penalty proportional to the amount exceeded. Furthermore, the weight coefficients can be adjusted to balance the importance of prediction accuracy and physical constraints.

[0048] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

[0049] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A method for predicting the posture of a scraper conveyor, characterized in that: The following steps are involved: S1. Collect multi-source data, including equipment operation status data, bracket posture data, and equipment physical structure parameters; S2. Time synchronization of the collected equipment operation status data and support posture data, and then calculation of the bottom cutting depth of the support chute and the leading state vector of the push-slide support; S3. Based on the calculated cutting depth and the advance state vector, the first sub-network is used to extract the time series features of the shearer's cutting posture data and the support posture data, including the posture angle change characteristics when the support is pushed, the correlation characteristics between the push rod length and the cutting depth, and the influence characteristics of the adjacent supports on the current support. extracting structural features of physical structural parameters of the device through the second sub-network; S4. The extracted time series features and structural features are spliced ​​together, and the attitude prediction value of the scraper conveyor is output through the attitude prediction model.

2. The method for predicting the posture of a scraper conveyor according to claim 1, characterized in that: In step S2, time synchronization processing of the equipment operation status data and the bracket posture data is performed, including: S201, performing clock synchronization on a data acquisition device for collecting device operation status data based on the NTP protocol; S202: Based on clock synchronization, the trigger position of the support pushing and sliding action is used as a reference to align the timestamps of the posture data of each support, and align the spatial positions of the shearer cutting data corresponding to the support; S203 : Based on the alignment of the timestamp and the spatial position, unified resampling is performed to obtain time-synchronized device operation status data and bracket posture data.

3. The method for predicting the posture of a scraper conveyor according to claim 2, characterized in that: The step S202 is specifically as follows: When the bracket completes the pulling action and triggers the completion flag, the current global timestamp is recorded and bracket position , and when the push-slide action flag is ready to be triggered, the time for preparing the push-slide work is recorded as ; Data points of any bracket attitude sensor , take the time lie in and The data values ​​between them are used, and the posture data of the bracket is obtained through median filtering to complete the timestamp alignment of the posture data of each bracket and the spatial position alignment of the coal cutting machine bottom data; wherein, Indicates the corrected timestamp, Indicates the data value collected by the attitude sensor.

4. The method for predicting the posture of a scraper conveyor according to claim 1, wherein: In step S2, the bottom cutting depth of the bracket chute is for: Where, Indicates the vertical height of the rocker arm end, , Indicates the cutting radius of the shearer drum, Indicates the height of the fuselage connection point from the bottom plate. Indicates the rocker arm length, Indicates the pitch angle of the coal mining machine body, Indicates the inclination angle of the rocker arm relative to the fuselage; The method for determining the leading state vector of the push-and-slide support is: Setting the advanced state threshold ; when When , the leading state is that the left frame leads; when When , the leading state is that the right frame is ahead; when When , the leading state is no offset; The determined advance state is uniquely encoded to obtain the advance state vector .

5. The method for predicting the posture of a scraper conveyor according to claim 4, characterized in that: In step S3, the first sub-network is an LSTM model, which extracts the temporal features of the input data through a set number of neurons and outputs the hidden state of the last time step; The input data is , represents the number of samples, represents the time step, The features of each time step include the attitude angle of the current bracket, the cutting depth, the push rod length, the attitude angles of the left and right brackets, the three dimensions of the leading state vector, and the position; the attitude angles include the pitch angle, the roll angle, and the heading angle.

6. The method for predicting the posture of a scraper conveyor according to claim 1, characterized in that: In step S3, the second sub-network is a PSFC network, and the PSFC network includes a dimension expansion layer, a first fully connected layer, a ReLU activation function, a second fully connected layer and an output layer connected in sequence; The dimension expansion layer expands the dimension of the input data by constructing implicit relationships that derive explicit encoding physical parameters; The input data of the PSFC network include support width, slide length, maximum horizontal bending angle, maximum vertical full angle and shearer sliding shoe spacing; After the dimension expansion of the dimension expansion layer, the data input to the first fully connected layer also includes the maximum offset in the horizontal direction, the maximum fluctuation in the vertical direction, and the number of sliding shoe coverage nodes.

7. The method for predicting the posture of a scraper conveyor according to claim 1, wherein: In step S4, the posture prediction model includes a third fully connected layer and a fourth fully connected layer connected in sequence.

8. The method for predicting the posture of a scraper conveyor according to claim 1, wherein: The loss function of the posture prediction model is for: Where, represents the forecast error term, represents the physical constraint term, represents the total constraint weight, represents the position constraint, Represents an angular constraint.

9. The method for predicting the posture of a scraper conveyor according to claim 8, characterized in that: The angle constraint for: Where, Represents the horizontal constraint, Represents the vertical constraint item, and Respectively and The constraint weight coefficient, Indicates the maximum horizontal bending angle of the scraper conveyor. Indicates the maximum vertical bending angle of the scraper conveyor. Indicates the vertical angle difference of the scraper conveyor, Indicates the vertical angle difference of the scraper conveyor, subscript Indicates the chute number of the scraper conveyor. Indicates the total number of chutes on the scraper conveyor.

10. The method for predicting the posture of a scraper conveyor according to claim 8, wherein: The position constraint for: Where, Represents the horizontal position constraint, Represents a vertical position constraint, and Respectively and The constraint weight coefficient, and Respectively represent the horizontal offset and vertical offset of the scraper conveyor, and Respectively represent the maximum horizontal and vertical offset of the scraper conveyor, the subscript Indicates the chute number of the scraper conveyor. Indicates the total number of chutes on the scraper conveyor.

Images