Scraper chain tension dynamic optimization method based on real-time load feedback

Abstract

The present application relates to a method for dynamically optimizing the tension of a scraper chain based on real-time load feedback, belonging to the technical field of scraper chain control. The method comprises: obtaining load data of the scraper chain in multiple operating states; extracting features after standardizing the load data in multiple operating states to obtain load time series features; processing the load time series features and the load data in multiple operating states using a time attention mechanism to obtain importance weights; calculating the tension value based on each operating state load data and its importance weight; calculating the real-time working condition coefficient of the current time node and dynamically adjusting the preset safe tension range to obtain the real-time safe tension range of the current time node; comparing the tension value with the real-time safe tension range and generating a control instruction based on the comparison result; and executing the control instruction. The present application optimizes the tension control method of the scraper chain, improving the control accuracy of the scraper chain tension.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of scraper chain control, and particularly relates to a scraper chain tension dynamic optimization method based on real-time load feedback. BACKGROUND

[0002] As a key equipment for material conveying in coal mines, metallurgy and other industries, the running stability of the scraper chain directly affects the operation efficiency and safety performance of the whole machine. As a core parameter reflecting the running state of the chain, the chain tension often fluctuates dynamically under complex working conditions, leading to excessive tension or relaxation of the chain, causing faults, increasing equipment maintenance costs, and even causing production interruption.

[0003] Currently, when controlling the tension of the scraper chain, a static control method is often used, that is, a fixed tension value is preset according to the design parameters of the scraper, and the hydraulic system is used to maintain the tension. However, the working conditions of the scraper are complex and changeable in actual operation. Controlling the tension of the scraper chain according to the fixed tension value is difficult to adapt to the complex and changing working conditions of the scraper chain, resulting in insufficient control accuracy of the scraper chain. SUMMARY

[0004] To solve the above technical problems, the present application provides a scraper chain tension dynamic optimization method based on real-time load feedback. The technical scheme of the present application is as follows:

[0005] The scraper chain tension dynamic optimization method based on real-time load feedback comprises:

[0006] S1, acquiring time-aligned multiple operating state load data of the scraper chain including multiple time nodes of the current time node;

[0007] S2, extracting features after standardizing all time node multiple operating state load data, to obtain load time sequence features;

[0008] S3, performing time attention mechanism processing on the load time sequence features and the multiple operating state load data of each time node, to obtain the importance weight of each operating state load data of each time node;

[0009] S4, calculating the tension value of the current time node of the scraper chain according to each operating state load data of all time nodes and its importance weight;

[0010] S5, calculating the real-time working condition coefficient of the current time node according to each operating state load data of the current time node and its importance weight, and dynamically adjusting the preset safe tension range according to the real-time working condition coefficient of the current time node, to obtain the real-time safe tension range of the current time node;

[0011] S6, comparing the tension value of the current time node of the scraper chain with the real-time safe tension range thereof, and generating a control instruction according to the comparison result;

[0012] S7, adjusting the scraper chain according to the control instruction, so that the tension value of the future time node of the scraper chain is always within the corresponding real-time safe tension range thereof.

[0013] Preferably, the S2 comprises:

[0014] S21, standardizing a plurality of operating state load data of all time nodes to obtain a plurality of standard operating state load data of all time nodes;

[0015] S22, calculating a statistical characteristic of each standard operating state load data of all time nodes, and generating a first feature of each standard operating state load data of each time node according to the statistical characteristic of each standard operating state load data, wherein the statistical characteristic comprises a mean value, a standard deviation and a maximum value;

[0016] S23, converting each standard operating state load data into a frequency domain operating state load data through Fourier transform, and obtaining an instantaneous frequency feature of each frequency domain operating state load data at each time node, and taking the instantaneous frequency feature of each time node as a second feature;

[0017] S24, taking each time node, the first feature and the second feature as corresponding elements of each time node, and combining the corresponding elements of all time nodes to obtain a load time sequence feature.

[0018] Preferably, the S22 comprises, when generating the first feature of each standard operating state load data of each time node according to the statistical characteristic of each standard operating state load data:

[0019] S221, constructing a matching relationship between the maximum value of each standard operating state load data and the time node corresponding to the maximum value;

[0020] S222, constructing a matching relationship between the mean value and the standard deviation of each standard operating state load data and all time nodes;

[0021] S223, combining the matching relationship between the maximum value of each standard operating state load data and the time node corresponding to the maximum value, the matching relationship between the mean value and the standard deviation of each standard operating state load data and all time nodes, to obtain the first feature of each time node.

[0022] Preferably, the S3 comprises:

[0023] S31, constructing a load matrix according to the multiple operating state load data of each time node, wherein each row of the matrix load represents one operating state load data, and each column represents the multiple operating state load data of one time node;

[0024] S32, inputting the load matrix and the load time sequence feature into a channel attention module, and generating a query vector, a key vector and a value vector of each operating state load data at each time step according to the channel attention module;

[0025] S33, calculating an attention score of each operating state load data at each time step according to the query vector, the key vector and the value vector of each operating state load data at each time step;

[0026] S34, normalizing the attention scores of each operating state load data at all time steps to obtain an attention weight of each operating state load data at each time step as an importance weight of each operating state load data at each time node.

[0027] Preferably, the S33 is realized by formula (1) when calculating the attention score a of the operating state load data at the i th time step according to the query vector, the key vector and the value vector of the operating state load data at each time step:

[0028] (1);

[0029] In formula (1), exp( ) represents an exponential function, Q i represents the query vector of the operating state load data at the i th time step, K j represents the key vector of the operating state load data at the j th time step, K i represents the key vector of the operating state load data at the i th time step, d represents the sum of the dimensions of the query vector and the key vector of the operating state load data, and T represents the number of time steps.

[0030] Preferably, the S4 comprises:

[0031] S41, fitting each operating state load data and its importance weight of all time nodes to obtain a chain tension curve;

[0032] S42, obtaining a predicted chain tension corresponding to the current time node according to the chain tension curve;

[0033] S43, calculating a tension value of the chain of the scraper at the current time node according to the predicted chain tension and the importance weight of the multiple operating state load data of the current time node.

[0034] Preferably, the S43 is implemented by formula (2) when calculating the tension value P of the current time node of the scraper chain according to the importance weight of the plurality of operating state load data of the predicted chain tension and the current time node:

[0035] (2);

[0036] In formula (2), max(a) represents the maximum value of the importance weight of all operating state load data of the current time node, min(a) represents the minimum value of the importance weight of all operating state load data of the current time node, and P0 represents the predicted chain tension.

[0037] Preferably, the S5 comprises:

[0038] S51, weighting and summing all operating state load data of the current time node with the importance weight thereof and accumulating a preset load coefficient to obtain a load value of the current time node;

[0039] S52, determining a preset amplification coefficient corresponding to the load value of the current time node according to a load value-preset amplification coefficient mapping table, and multiplying the load value of the current time node by the corresponding preset amplification coefficient to obtain a real-time working condition coefficient of the current time node;

[0040] S53, multiplying the left end value and the right end value of the preset safe tension range by the real-time working condition coefficient of the current time node to obtain a real-time left end value and a real-time right end value, and constructing a real-time safe tension range of the current time node according to the real-time left end value and the real-time right end value.

[0041] Preferably, the S6 comprises:

[0042] S61, if the tension value of the current time node of the scraper chain is within the real-time safe tension range thereof, it is determined that the tension of the scraper chain of the current time node is normal, and it is determined that there is no early warning level;

[0043] S62, if the tension value of the current time node of the scraper chain is not within the real-time safe tension range thereof, and the minimum value of the absolute value of the difference between the real-time left end value and the real-time right end value of the real-time safe tension range is less than or equal to a critical threshold value, it is determined that the tension of the scraper chain of the current time node is at risk, and it is determined that there is an early warning level and the early warning level is intermediate;

[0044] S63, if the tension value of the current time node of the scraper chain is not within the real-time safe tension range thereof, and the minimum value of the absolute value of the difference between the real-time left end value and the real-time right end value of the real-time safe tension range is greater than the critical threshold value, it is determined that the tension of the scraper chain of the current time node is at greater risk, and it is determined that there is an early warning level and the early warning level is high;

[0045] S64, when it is determined that there is a warning level, a control strategy of the scraper chain is generated according to the minimum value of the absolute value of the difference, and the control instruction is generated by combining the control strategy with the warning level.

[0046] All the optional technical solutions described above can be combined arbitrarily, and the application does not describe the structures after combination in detail.

[0047] Through the above scheme, the beneficial effects of the application are as follows:

[0048] By performing time attention mechanism processing on the load time sequence features and the load data of various operating states, the importance weight of each operating state load data at each time node is obtained, and the tension value of the scraper chain at the current time node is calculated based on this, so that different weights are given to different operating state load data at each time node by using the load time sequence features and the time attention mechanism, the tension can be dynamically determined according to the complex and changeable working conditions of the scraper in actual operation, and the tension value of the scraper chain at the current time node can be more accurately calculated.

[0049] The real-time working condition coefficient of the current time node is calculated according to each operating state load data and its importance weight, the preset safe tension range is dynamically adjusted according to the real-time working condition coefficient, the real-time safe tension range of the current time node is obtained, the tension value is compared with the real-time safe tension range, the control instruction is generated according to the comparison result, and the relationship between the tension adjustment of the scraper chain and the real-time working condition of the scraper chain is established, so that the tension control method of the scraper chain is optimized, and the control precision of the tension of the scraper chain is improved.

[0050] The above description is only a summary of the technical solutions of the application, in order to more clearly understand the technical means of the application, and the content of the specification can be implemented, the following preferred embodiments of the application are described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a flowchart of the scraper chain tension dynamic optimization method based on real-time load feedback provided by the embodiments of the application. DETAILED DESCRIPTION

[0052] The specific embodiments of the application will be further described in detail below with reference to the accompanying drawings and embodiments. The following embodiments are used to illustrate the application, but not to limit the scope of the application.

[0053] As Figure 1 shown, the scraper chain tension dynamic optimization method based on real-time load feedback provided by the embodiments of the application includes steps S1-S7, specifically:

[0054] S1, acquire time-aligned multiple operating state load data of multiple time nodes including a current time node of the scraper chain.

[0055] In the embodiment of the application, the multiple operating state load data is acquired in real time by multiple types of sensors with the same acquisition frequency and time alignment pre-configured on the scraper; wherein the operating state load data includes chain running speed data, chain load data, chain strain data and chain vibration data, and the sensor types include speed sensors, load sensors, strain sensors and vibration sensors. The multiple time nodes including the current time node are generally time nodes within 30 minutes before the current time.

[0056] S2, after standardizing the multiple operating state load data of all time nodes, extracting features to obtain load time sequence features.

[0057] Specifically, the standardization is achieved by performing maximum and minimum standardization on the multiple operating state load data. In the embodiment of the application, the extracting features refers to extracting the mean value, standard deviation, maximum value and instantaneous frequency features of each operating state load data after standardization, and then according to the above features of the multiple operating state load data of each time node, the load time sequence features are constituted.

[0058] In one specific embodiment, the S2 includes S21 to S24:

[0059] S21, standardizing the multiple operating state load data of all time nodes to obtain multiple standard operating state load data of all time nodes.

[0060] Specifically, the multiple standard operating state load data are standardized in [0, 1].

[0061] S22, calculating statistical features of each standard operating state load data of all time nodes, and generating first features of each standard operating state load data of each time node according to the statistical features of each standard operating state load data, wherein the statistical features include mean value, standard deviation and maximum value.

[0062] Specifically, the mean value and the variance are obtained by calculating the mean value and the variance of each standard operating state load data of all time nodes; for any standard operating state load data, the maximum value is obtained by finding the maximum value and the minimum value of the standard operating state load data in all time nodes.

[0063] In one specific embodiment, the S22, when generating the first features of each standard operating state load data of each time node according to the statistical features of each standard operating state load data, includes:

[0064] S221, constructing a matching relationship between the extreme value of each standard running state load data and the time node corresponding to the extreme value.

[0065] Specifically, for any standard running state load data, since its extreme value can only exist in two time nodes, when constructing a matching relationship between the extreme value of any standard running state load data and the time node corresponding to the extreme value, a relationship between the maximum value and the minimum value of the standard running state load data and the corresponding time nodes is constructed.

[0066] S222, constructing a matching relationship between the mean value and the standard deviation of each standard running state load data and all time nodes.

[0067] Specifically, for any standard running state load data, after calculating the mean value and the standard deviation of the standard running state load data, a matching relationship between the mean value and the standard deviation and each time node is constructed, so that the mean value and the standard deviation are used as one of the first features of the standard running state load data of each time node.

[0068] S223, combining the matching relationship between the extreme value of each standard running state load data and the corresponding time node, the matching relationship between the mean value and the standard deviation and all time nodes, to obtain the first feature of each time node.

[0069] Specifically, the mean value, variance and extreme value of each time node of each standard running state load data are combined according to the time node sorting to obtain the first feature of each time node. For example, the first feature of the standard chain running speed data and the standard chain load data at the 5th time node can be represented as [5, (0.6, 0.06, 0.01, 0.08), (0.5, 0.05, 0.01, 0.09)], wherein 5 represents the 5th time node, 0.6, 0.06, 0.01 and 0.08 represent the mean value, variance, minimum value and maximum value of the standard chain running speed data, and the first feature of the standard chain load data is the same. It should be noted that if there is no extreme value of a certain standard running state load data at a certain time node, the element corresponding to the extreme value in the first feature of the standard running state load data at the time node is empty, and the elements after the extreme value are filled in the position.

[0070] S23, converting each standard running state load data into a frequency domain running state load data by Fourier transform, and obtaining the instantaneous frequency feature of each frequency domain running state load data at each time node, and taking the instantaneous frequency feature of each time node as the second feature.

[0071] Specifically, the Fourier transform is performed on the standard running state load data x(y) to obtain frequency domain running state load data X(f), and formula (3) is used to realize the following formula:

[0072] (3);

[0073] In formula (3), f represents a preset frequency, x(y) represents the data of the yth time node of the standard running state load data, m represents the number of time nodes, represents a circular constant, and J represents an imaginary unit.

[0074] The instantaneous frequency feature refers to the instantaneous frequency change of each frequency domain running state load data at each time node, including instantaneous frequency and instantaneous phase and the like.

[0075] S24, each time node, the first feature and the second feature are combined as the corresponding elements of each time node, and the elements corresponding to all time nodes are combined to obtain the load time sequence feature.

[0076] Specifically, still taking the example in S223 as an example, if the instantaneous frequency is 20 and the instantaneous phase is 2 in the second feature corresponding to the standard chain running speed data, the element corresponding to the 5th time node of the standard chain running speed data is [5, (0.6, 0.06, 0.01, 0.08, 20, 2 )].

[0077] Further, the elements corresponding to all time nodes are combined in time sequence to obtain the load time sequence feature.

[0078] S3, the load time sequence feature and the multiple running state load data of each time node are processed by a time attention mechanism to obtain the importance weight of each running state load data of each time node.

[0079] Specifically, the attention mechanism processing is used to assign different importance weights to each running state load data of each time node, so as to improve the accuracy of subsequent tension value calculation.

[0080] In one specific embodiment, the S3 comprises:

[0081] S31, a load matrix is constructed according to the multiple running state load data of each time node, wherein each row of the matrix represents one kind of running state load data, and each column represents the multiple running state load data of one time node.

[0082] Specifically, the number of rows of the load matrix represents the number of types of running state load data, and the number of columns represents the number of time nodes.

[0083] S32, input the load matrix and the load time sequence feature into the channel attention module, and generate a query vector, a key vector and a value vector of each time step and each operating state load data according to the channel attention module.

[0084] Specifically, the channel attention module is a module for assigning different importance weights to different kinds of operating state load data. In the embodiment of the application, the channel attention module can refer to the channel mechanism in the convolutional neural network. The query vector is a vector for comparing the correlation degree between each operating state load data at each time node, which is obtained by multiplying the load time sequence feature by the linear change of the load matrix; each key vector is generated by the channel attention module from each element in the load time sequence feature; and each value vector is generated by the channel attention module from each element in each row and each column of the load matrix.

[0085] S33, calculate the attention score of each time step and each operating state load data according to the query vector, the key vector and the value vector of each time step and each operating state load data.

[0086] Specifically, the attention score refers to the contribution degree of each value vector when calculating the importance weight of each operating state load data at each time node. In the embodiment of the application, one time step corresponds to one time node.

[0087] In one specific embodiment, the S33 calculates the attention score a of the operating state load data at the i-th time step according to the query vector, the key vector and the value vector of any operating state load data at each time step, and is realized by formula (1):

[0088] (1);

[0089] In formula (1), exp( ) represents an exponential function, Qi represents the query vector of the operating state load data at the i-th time step, Kj represents the key vector of the operating state load data at the j-th time step, Ki represents the key vector of the operating state load data at the i-th time step, d represents the sum of the dimensions of the query vector and the key vector of the operating state load data, and T represents the number of time steps.

[0090] S34, normalize the attention scores of each operating state load data at all time steps to obtain the attention weight of each operating state load data at each time step, as the importance weight of each operating state load data at each time node.

[0091] Specifically, the importance weights of each operating state load data at each time node are not necessarily the same, and the sum of the importance weights of all operating state load data at each time node is 1.

[0092] S4, calculate the tension value of the scraper chain at the current time node according to the load data of each operating state of all time nodes and the importance weight thereof.

[0093] Specifically, the tension value refers to the value of the sum of the forces applied to the scraper chain.

[0094] In one specific embodiment, the S4 comprises:

[0095] S41, fit the load data of each operating state of all time nodes and the importance weight thereof to obtain a chain tension curve.

[0096] Specifically, the load data of each operating state of a certain time node and the importance weight thereof are processed by weighted summation to obtain the chain tension value of the time node, and the chain tension values of all time nodes are linearly regressed and fitted to obtain a chain tension curve, wherein the horizontal axis of the chain tension curve represents the time node, and the vertical axis represents the chain tension value.

[0097] S42, obtain the predicted chain tension corresponding to the current time node according to the chain tension curve.

[0098] Specifically, the chain tension value corresponding to the current time node in the chain tension curve is obtained as the predicted chain tension.

[0099] S43, calculate the tension value of the scraper chain at the current time node according to the predicted chain tension and the importance weight of the load data of each operating state of the current time node.

[0100] Specifically, since the chain tension curve obtains predicted data, in order to more accurately calculate the tension value of the scraper chain at the current time node, the predicted chain tension is adjusted by the importance weight of the load data of each operating state of the current time node.

[0101] In one specific embodiment, the S43 is implemented by formula (2) when calculating the tension value P of the scraper chain at the current time node according to the predicted chain tension and the importance weight of the load data of each operating state of the current time node.

[0102] (2);

[0103] In formula (2), max(a) represents the maximum value of the importance weight of all operating state load data of the current time node, min(a) represents the minimum value of the importance weight of all operating state load data of the current time node, and P0 represents the predicted chain tension.

[0104] Specifically, the formula (2) corrects the predicted chain tension by taking the maximum and minimum of the importance weight of all running state load data at the current time node, and obtains a more accurate tension value of the scraper chain at the current time node.

[0105] S5, calculating a real-time working condition coefficient of the current time node according to each running state load data and its importance weight at the current time node, and dynamically adjusting a preset safe tension range according to the real-time working condition coefficient of the current time node to obtain a real-time safe tension range of the current time node.

[0106] Specifically, the real-time working condition coefficient is a coefficient dynamically calculated according to each running state load data and its importance weight at a certain time node, indicating the working state of the scraper chain at the time node. The preset safe tension range is a safe running range of the tension value obtained according to the historical experience data of the scraper chain.

[0107] In one specific embodiment, the S5 comprises:

[0108] S51, weighting and summing all running state load data at the current time node and its importance weight and adding a preset load coefficient to obtain a load value at the current time node.

[0109] Specifically, the preset load coefficient is a coefficient preset according to standard operating conditions, indicating the load capacity of the scraper chain under specific conditions.

[0110] S52, determining a preset amplification coefficient corresponding to the load value at the current time node according to a load value-preset amplification coefficient mapping table, and multiplying the load value at the current time node by the corresponding preset amplification coefficient to obtain a real-time working condition coefficient at the current time node.

[0111] Specifically, the load value-preset amplification coefficient mapping table is a table preset to represent the relationship between the load value and the preset amplification coefficient, which can refer to Table 1. The preset amplification coefficient is a preset coefficient for adjusting the preset safe tension range according to a specific load value.

[0112]

[0113] S53, multiplying the left end value and the right end value of the preset safe tension range by the real-time working condition coefficient at the current time node to obtain a real-time left end value and a real-time right end value, and constructing a real-time safe tension range of the current time node according to the real-time left end value and the real-time right end value.

[0114] Specifically, the numerical range between the real-time left end value and the real-time right end value constitutes the real-time safe tension range.

[0115] It should be noted that in the embodiments of the present application, the real-time working condition coefficient of different time nodes is changed in real time according to the load data of each operating state of different time nodes and the importance weight, so that the real-time working condition coefficient corresponding to different time nodes is dynamically changed, and then the real-time safety tension range corresponding to different time nodes is dynamically changed.

[0116] S6, compare the tension value of the current time node of the scraper chain with the real-time safety tension range thereof, and generate a control instruction according to the comparison result.

[0117] Specifically, the control instruction is a comparison result generated by comparing the tension value of the current time node of the scraper chain with the real-time safety tension range thereof, and is an instruction for adjusting the parameters of the scraper chain, such as intermediate early warning, increasing the chain running speed, high-level early warning, increasing the motor load, etc.

[0118] In a specific embodiment, the S6 comprises:

[0119] S61, if the tension value of the current time node of the scraper chain is within the real-time safety tension range thereof, it is determined that the tension of the scraper chain at the current time node is normal, and it is determined that there is no early warning level.

[0120] S62, if the tension value of the current time node of the scraper chain is not within the real-time safety tension range thereof, and the minimum value of the absolute value of the difference between the tension value of the current time node of the scraper chain and the real-time left end value and the real-time right end value of the real-time safety tension range is less than or equal to a critical threshold value, it is determined that the tension of the scraper chain at the current time node is at risk, and it is determined that there is an early warning level and the early warning level is intermediate.

[0121] Specifically, the critical threshold value is pre-set according to an empirical value, and is used to determine whether the tension value of the current time node of the scraper chain is away from the real-time safety tension range and is in a greater risk. The minimum value of the absolute value of the difference refers to the smaller one of the absolute values of the differences between the tension value of the current time node of the scraper chain and the real-time left end value and the real-time right end value of the real-time safety tension range.

[0122] S63, if the tension value of the current time node of the scraper chain is not within the real-time safety tension range thereof, and the minimum value of the absolute value of the difference between the tension value of the current time node of the scraper chain and the real-time left end value and the real-time right end value of the real-time safety tension range is greater than the critical threshold value, it is determined that the tension of the scraper chain at the current time node is in a greater risk, and it is determined that there is an early warning level and the early warning level is high.

[0123] S64, when it is determined that there is an early warning level, a control strategy of the scraper chain is generated according to the minimum value of the absolute value of the difference, and the control strategy is combined with the early warning level to generate a control instruction.

[0124] Specifically, if the tension value of the scraper chain at the current time node is less than the real-time left end value of the real-time safe tension range and the warning level is medium, the control strategy is to increase the running speed of the scraper chain. If the tension value of the scraper chain at the current time node is greater than the real-time right end value of the real-time safe tension range and the warning level is high, the control strategy is to reduce the running speed of the scraper chain and reduce the motor load. The control instruction is obtained by combining the control strategy with the warning level.

[0125] S7, adjusting the scraper chain according to the control instruction, so that the tension value of the scraper chain at the future time node is always within the corresponding real-time safe tension range.

[0126] Based on all the above embodiments, the scraper chain tension dynamic optimization method based on real-time load feedback provided by the embodiments has the following beneficial effects:

[0127] Firstly, by obtaining various operating state load data of the scraper chain and extracting features, load time sequence features are obtained, which provide accurate data basis for subsequent analysis of the tension value of the scraper chain.

[0128] Then, the load time sequence features and various operating state load data are processed by time attention mechanism to obtain the importance weight of each operating state load data at each time node, and the tension value of the scraper chain at the current time node is calculated according to all operating state load data at each time node and the importance weight thereof, so as to give different weights to different operating state load data at each time node by using load time sequence features and time attention mechanism, dynamically adjust the tension according to the complex and changeable working conditions of the scraper in actual operation, and more accurately calculate the tension value of the scraper chain at the current time node, which provides high-quality data for subsequent optimization control of the scraper chain.

[0129] Finally, the real-time working condition coefficient of the current time node is calculated according to each operating state load data and the importance weight thereof, the preset safe tension range is dynamically adjusted according to the real-time working condition coefficient, the real-time safe tension range of the current time node is obtained, the tension value is compared with the real-time safe tension range, the control instruction is generated according to the comparison result, and the control instruction is executed, thereby optimizing the tension control method of the scraper chain and improving the control precision of the tension of the scraper chain.

[0130] The above only describes the preferred embodiments of the present application and is not used to limit the present application. It should be pointed out that for ordinary skilled persons in the technical field, without departing from the technical principles of the present application, a number of improvements and modifications can be made, and these improvements and modifications should be regarded as the protection scope of the present application.

Claims

1. A method for dynamic optimization of scraper conveyor chain tension based on real-time load feedback, characterized in that, include: S1, acquire various operating state load data of the scraper conveyor chain, including multiple time nodes aligned to the current time node; S2, after standardizing the load data of various operating states at all time points, extract features to obtain load time series features; S3, the time series characteristics of the load and the load data of multiple operating states at each time node are processed by the time attention mechanism to obtain the importance weight of each type of load data at each time node. S3 includes: S31, construct a load matrix based on the load data of multiple operating states at each time point, wherein each row of the matrix load represents a load data of one operating state, and each column represents the load data of multiple operating states at a time point. S32, input the load matrix and load time series features into the channel attention module, and generate query vector, key vector and value vector of load data for each time step and each running state according to the channel attention module; S33, calculate the attention score for each type of runtime state load data at each time step based on the query vector, key vector, and value vector of each type of runtime state load data at each time step; S34, normalize the attention score of each running state load data at all time steps to obtain the attention weight of each running state load data at each time step, which serves as the importance weight of each running state load data at each time node. S4, calculate the tension value of the scraper conveyor chain at the current time node based on the load data of each operating state at all time nodes and their importance weights; S4 includes: S41, fit the load data of each operating state at all time points and its importance weight to obtain the chain tension curve; S42, Obtain the predicted chain tension corresponding to the current time node based on the chain tension curve; S43, calculate the tension value of the scraper conveyor chain at the current time node based on the importance weight of the predicted chain tension and the various operating state load data at the current time node; When S43 calculates the tension value P of the scraper conveyor chain at the current time node based on the importance weights of the predicted chain tension and the various operating state load data at the current time node, it does so through formula (2): （2）； In formula (2), max(a) represents the maximum value of the importance weight of all operating state load data at the current time node, min(a) represents the minimum value of the importance weight of all operating state load data at the current time node, and P0 represents the predicted chain tension. S5. Calculate the real-time operating condition coefficient of the current time node based on the load data of each operating state at the current time node and its importance weight, and dynamically adjust the preset safety tension range based on the real-time operating condition coefficient of the current time node to obtain the real-time safety tension range of the current time node. S6, compare the tension value of the scraper conveyor chain at the current time node with its real-time safe tension range, and generate control commands based on the comparison results; S7 adjusts the scraper conveyor chain according to the control command so that the tension value of the scraper conveyor chain at future time nodes is always within its corresponding real-time safe tension range.

2. The method for dynamic optimization of scraper conveyor chain tension based on real-time load feedback according to claim 1, characterized in that, S2 includes: S21, standardize the load data of various operating states at all time points to obtain various standard operating state load data of all time points; S22, calculate the statistical characteristics of each standard operating state load data at all time points, and generate the first feature of each standard operating state load data at each time point based on the statistical characteristics of each standard operating state load data, wherein the statistical characteristics include the mean, standard deviation and extreme values; S23, convert each standard operating state load data into frequency domain operating state load data through Fourier transform, and obtain the instantaneous frequency characteristics of each frequency domain operating state load data at each time node, and use the instantaneous frequency characteristics of each time node as the second feature. S24, take each time node, the first feature and the second feature as the elements corresponding to each time node, and combine the elements corresponding to all time nodes to obtain the load time series features.

3. The method for dynamic optimization of scraper conveyor chain tension based on real-time load feedback according to claim 2, characterized in that, When S22 generates the first feature of the load data for each standard operating state at each time point based on the statistical characteristics of the load data for each standard operating state, it includes: S221, Construct the matching relationship between the maximum and minimum values ​​of load data for each standard operating state and the time nodes corresponding to the maximum and minimum values; S222, construct the matching relationship between the mean and standard deviation of the load data for each standard operating state and all time points; S223, combine the extreme values ​​of each standard operating state load data and their corresponding time node matching relationships, mean and standard deviation matching relationships with all time nodes to obtain the first feature of each time node.

4. The method for dynamic optimization of scraper conveyor chain tension based on real-time load feedback according to claim 1, characterized in that, When S33 calculates the attention score 'a' of the operating state load data at the i-th time step based on the query vector, key vector, and value vector of any operating state load data at each time step, it does so through formula (1): （1）； In formula (1), exp() represents the exponential function, Qi represents the query vector of the running state load data at the i-th time step, Kj represents the key vector of the running state load data at the j-th time step, Ki represents the key vector of the running state load data at the i-th time step, d represents the sum of the dimensions of the query vector and the key vector of the running state load data, and T represents the number of time steps.

5. The method for dynamic optimization of scraper conveyor chain tension based on real-time load feedback according to claim 1, characterized in that, S5 includes: S51, weight all the running status load data of the current time node and their importance weights, and sum them up by adding the preset load coefficient to obtain the load value of the current time node; S52, determine the preset amplification factor corresponding to the load value at the current time node according to the load value-preset amplification factor mapping table, and multiply the load value at the current time node with the corresponding preset amplification factor to obtain the real-time operating condition factor at the current time node. S53, multiply the left and right values ​​of the preset safe tension range by the real-time operating condition coefficient of the current time node to obtain the real-time left and right values, and construct the real-time safe tension range of the current time node based on the real-time left and right values.

6. The method for dynamic optimization of scraper conveyor chain tension based on real-time load feedback according to claim 5, characterized in that, S6 includes: S61, if the tension value of the scraper conveyor chain at the current time node is within its real-time safe tension range, then the tension of the scraper conveyor chain at the current time node is determined to be normal, and it is determined that there is no warning level. S62, if the tension value of the scraper conveyor chain at the current time node is not within its real-time safe tension range, and the minimum absolute value of the difference between the real-time left end value and the real-time right end value of the real-time safe tension range is less than or equal to the critical threshold, then it is determined that there is a risk in the tension of the scraper conveyor chain at the current time node, and a warning level is determined to be medium. S63, if the tension value of the scraper conveyor chain at the current time node is not within its real-time safe tension range, and the minimum absolute value of the difference between the real-time left end value and the real-time right end value of the real-time safe tension range is greater than the critical threshold, then it is determined that the tension of the scraper conveyor chain at the current time node has a significant risk, and a warning level is determined to exist and the warning level is high. S64, when a warning level is determined, a control strategy for the scraper conveyor chain is generated based on the minimum absolute value of the difference, and the control strategy is combined with the warning level to generate control commands.

Images