Intelligent coordination control subsystem of coal mining machine

Abstract

The present application relates to a kind of intelligent coordination control subsystem of coal mining machine, comprising: coordination control module, for obtaining planned cutting path, real-time acquisition coal mining machine operating parameter, when the operating parameter of coal mining machine meets preset demand, first cutting motor control instruction and first traction motor control instruction are generated based on planned cutting path, motor control module is based on first cutting motor control instruction and first traction motor control instruction according to the planned cutting path Execution coal mining operation;Otherwise, second cutting motor control instruction and second traction motor control instruction are generated based on current coal mining machine operating parameter, and motor control module is used to control the traction motor and cutting motor of coal mining machine based on second cutting motor control instruction and second traction motor control instruction operation, based on current coal mining machine operating parameter Execution coal mining operation.The present application improves the accuracy and reliability of coal mining machine coordination control.

Description

Technical Field

[0001] This invention relates to the field of intelligent coal mining technology, and in particular to an intelligent coordination control subsystem for a coal mining machine. Background Technology

[0002] my country is the world's largest coal producer, accounting for over 35% of global coal production. Manual mining accounts for over 90% of my country's total coal output, the highest proportion in the world. Improving the technological level of the coal mining industry and developing intelligent mining technologies can not only effectively increase mining efficiency but also reduce the number of workers in hazardous environments, thereby enhancing mine safety. Therefore, research into unmanned mining technology is of great significance for promoting a revolution in my country's energy production and consumption, and for building a safe and efficient energy system.

[0003] However, existing technologies face at least the following technical problems in realizing intelligent coal mining technology: (1) Low level of intelligence and low mining efficiency; (2) Due to the irregular surface of the coal face, laser focusing is difficult, which makes it impossible to guarantee the quality of spectral acquisition. The quality of spectral acquisition is low and the coal and rock identification effect is poor. (3) The low accuracy of coal and rock classification leads to inaccurate depiction of the coal and rock interface, low accuracy of cutting path planning, and poor reliability.

[0004] (4) When the coal mining machine travels based on the cutting path, it is necessary to coordinate and control the cutting motor and the traction motor in combination with the cutting parameters. However, the coordination control in the existing technology has low accuracy and poor reliability. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent coordination and control subsystem for a coal mining machine, so as to solve, or at least partially solve, the technical problems mentioned in the background art.

[0006] On one hand, this invention provides an intelligent coordinated control subsystem for a coal mining machine, including a coordinated control module and a motor control module, wherein, The coordination and control module is used to obtain the planned cutting path and also to obtain the operating parameters of the coal mining machine in real time. When the operating parameters of the coal mining machine meet the preset requirements, it generates a first cutting motor control command and a first traction motor control command based on the planned cutting path and sends the first cutting motor control command and the first traction motor control command to the motor control module. When the operating parameters of the coal mining machine do not meet the preset requirements, a second cutting motor control command and a second traction motor control command are generated based on the current operating parameters of the coal mining machine. The second cutting motor control command and the second traction motor control command are sent to the motor control module. The second cutting motor control command and the second traction motor control command have higher priority than the first cutting motor control command and the first traction motor control command. The motor control module is used to control the operation of the traction motor and the cutting motor of the coal mining machine based on the first cutting motor control command and the first traction motor control command when no second cutting motor control command and the second traction motor control command are received, and to perform coal mining operations according to the planned cutting path; when the second cutting motor control command and the second traction motor control command are received, the module controls the operation of the traction motor and the cutting motor of the coal mining machine based on the second cutting motor control command and the second traction motor control command, and to perform coal mining operations based on the current operating parameters of the coal mining machine.

[0007] The above technical solution coordinates and controls the cutting motor and traction motor of the coal mining machine to perform coal mining operations based on the planned cutting path and the operating parameters of the coal mining machine, thereby improving the accuracy and reliability of the coordinated control of the coal mining machine. Attached Figure Description

[0008] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0009] Figure 1 This is a schematic diagram illustrating the basic principle of the intelligent coal mining system based on LIBS technology provided in Example 1. Figure 2 This is a schematic diagram of different focusing methods of the intelligent sampling device provided in Embodiment 2; Figure 3 The flowchart of the coal and rock identification method based on LIBS single spectral lines provided in Example 3; Figure 4 This is a schematic diagram illustrating the dimensionality reduction effect of PCA on LIBS data provided in Example 4; Figure 5 This is a schematic diagram of coal and rock identification with a fixed identification resolution in the PCA coal and rock identification model of Example 5; Figure 6 This is a schematic diagram of the coal and rock identification results at the dynamic identification resolution of the PLS coal and rock identification model in Example 5; Figure 7 This is a diagram illustrating the effect of selecting the target simplified spectral line wavelength based on VIP in Example 6. Figure 8 This is a schematic diagram of the intelligent coordination control subsystem for the coal mining machine in Example 7. Detailed Implementation To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] Laser-induced breakdown spectroscopy (LIBS) is an atomic spectroscopy technique used for material analysis and elemental detection in samples. The principle is that a high-energy-density laser pulse irradiates the sample, generating a high electric field at the focal spot on the sample's surface, creating a near-neutral "laser plasma." This plasma emits a spectrum, known as the laser-induced breakdown spectrum (LIBS spectrum). Each spectral line in the LIBS spectrum corresponds to a unique transition in a specific atom, ion, or molecule. Because each element emits light at its own specific spectral wavelength, the LIBS spectral information, effectively collected by an optical detection system, can be used as a "fingerprint" to identify the elemental composition of the sample, enabling qualitative and quantitative analysis of the elements contained within. LIBS technology can effectively detect low atomic number elements such as carbon (C), magnesium (Mg), aluminum (Al), and silicon (Si), which are crucial for coal and rock identification but difficult to detect using other analytical techniques. Therefore, LIBS technology is an ideal material analysis technology for coal and rock identification. This invention proposes several embodiments based on LIBS technology to promote intelligent and unmanned coal mining operations, which not only improves mining efficiency but also keeps miners away from dangerous and harsh working environments, achieving the goal of safe and efficient mining.

[0011] To improve the intelligence level and mining efficiency of coal mining machines, Example 1 is proposed.

[0012] Example 1 Example 1 provides an intelligent coal mining system based on LIBS technology, including a LIBS spectral intelligent acquisition subsystem, a LIBS spectral intelligent analysis subsystem, and an intelligent coordination control subsystem.

[0013] The LIBS spectral intelligent acquisition subsystem is used to acquire LIBS spectral data of samples within a preset wavelength range and send it to the LIBS spectral intelligent analysis subsystem. As an example, the preset wavelength range is [196.08nm, 507.78nm]. Selecting [196.08nm, 507.78nm] includes reference spectral lines of metallic and non-metallic elements that play a decisive role in coal and petrology analysis, thus improving the accuracy of coal and petrology identification based on LIBS data.

[0014] The LIBS spectral intelligent analysis subsystem is used to train and generate a coal and rock identification model based on the LIBS spectral data corresponding to the samples. The LIBS spectral intelligent acquisition subsystem is also used to set sampling points on the target working face, acquire the LIBS spectral data corresponding to the sampling points, and send them to the coal and rock identification model.

[0015] The coal and rock identification model is used to generate coal and rock identification results for the received sampling points based on the LIBS spectral data corresponding to the sampling points. These results are then fed back to the LIBS spectral intelligent acquisition subsystem. Furthermore, a planned truncation path is generated based on the coal and rock identification results, and this planned truncation path is sent to the intelligent coordination and control subsystem. It should be noted that the coal and rock identification result represents the coal and rock type, which can be a binary classification result or a multi-class classification result. A binary classification result includes both coal and rock, while a multi-class classification result can further subdivide the coal and rock types. For example, coal types include raw coal, refined coal, and middlings; rock types include sandy mudstone and gangue.

[0016] The intelligent coordination and control subsystem is used to control the coal mining machine to cut coal based on the planned cutting path, and at the same time, to acquire the operating parameters of the coal mining machine and adjust the cutting path in real time based on the operating parameters of the coal mining machine.

[0017] It should be noted that the intelligent coordination and control subsystem can effectively control the coal mining machine to operate according to the planned cutting path. In addition, in the event of an anomaly, it can use information such as coal and rock distribution and coal mining machine operating parameters as an "alarm signal" for cutting rock, automatically or with manual intervention to terminate or resume the operation of the coal mining machine.

[0018] The intelligent coal mining system, through the collaborative cooperation of its three subsystems, can effectively perceive coal and rock information and effectively control the coal mining machine based on this perception, laying the foundation for the realization of a remote intelligent control platform for unmanned coal mining faces.

[0019] like Figure 1 As shown, the LIBS intelligent spectral acquisition subsystem ( Figure 1The intelligent acquisition system (hereinafter referred to as the intelligent acquisition system) acquires LIBS spectral data of sampling points based on the actual distribution of coal and rock at the working face, and then processes the LIBS spectral data through the LIBS spectral intelligent analysis subsystem ( Figure 1 The intelligent analysis system (referred to as the intelligent analysis system in Chinese) generates coal and rock identification results for each sampling point, describes and generates a coal and rock distribution interface based on the coal and rock distribution interface, and realizes cutting path planning based on the coal and rock distribution interface. Figure 1 The intelligent coordination control system (hereinafter referred to as the intelligent coordination control system) controls the coal mining machine to mine coal based on the cutting path planning. At the same time, it also obtains the operating parameters of the coal mining machine according to the actual distribution of coal and rock in the working face, and adjusts the cutting path in real time based on the operating parameters of the coal mining machine.

[0020] As an example, the LIBS spectral intelligent acquisition subsystem is also used to dynamically adjust the sampling interval based on the coal and rock identification results of the sampling points. For instance, it determines sensitive and non-sensitive areas based on the coal and rock identification results. Sensitive areas are the regions near the coal-rock boundary, while non-sensitive areas are regions containing only coal or only rock. For non-sensitive areas, a larger sampling interval is set to maintain a lower identification resolution, saving the number of sampling points to meet the speed requirements of coal and rock identification at the coal mining face. In sensitive areas, the identification resolution should be increased, the sampling interval reduced, and the number of sampling points increased to meet the accuracy requirements of the coal mining machine's cutting trajectory for coal and rock identification. This intelligent dynamic coal and rock identification resolution improves the effective utilization of sampling points.

[0021] The control device is used to determine whether the object distance between the sampled substance and the LIBS spectral acquisition device is equal to the preset fixed object distance. If it is not equal, the control device moves along the object distance direction so that the object distance between the sampled substance and the LIBS spectral acquisition device is equal to the preset fixed object distance. This enables the LIBS spectral intelligent acquisition subsystem to overcome the difficulty of laser focusing caused by the irregular surface of the coal wall, realize intelligent focusing of the LIBS analyzer on the sampling point, and ensure the quality of LIBS spectral acquisition.

[0022] The LIBS spectral acquisition device generates a laser pulse and focuses it onto the surface of the sampled material when the object distance between the sampled material and the LIBS spectral acquisition device is equal to a preset fixed object distance. The surface of the sampled material is ablated to form plasma and emit a spectrum, thus acquiring the corresponding LIBS spectral data. The sampled material can be a sample or a sampling point on a target working surface. To improve the accuracy and reliability of the acquired LIBS spectral data, the LIBS spectral acquisition device can perform multiple acquisitions on the surface of the sampled material to obtain multiple acquisition data points. The average of these multiple acquisition data points is then used to generate the corresponding LIBS spectral data. As a preferred example, the single-pulse energy of the LIBS spectral acquisition device does not exceed 100 mJ.

[0023] As an example, each sample includes N LIBS spectral data (A n B n A n A is the spectral wavelength corresponding to the nth LIBS spectral data of the sample. n Located within the preset band interval, B n For the sample in A n The corresponding LIBS spectral intensity, where N ranges from 1 to N, is determined based on the sampling accuracy of the LIBS spectral intelligent acquisition subsystem. As an example, the preset band interval is [196.08nm, 507.78nm], and the value of N is 6074.

[0024] Since the coal samples provided are usually in the form of coal powder, necessary simple preparation can be performed to ensure uniform material distribution and compact structure, avoiding agitation of the sample surface by the pulsed laser, which could affect the accuracy and stability of LIBS spectral acquisition. The sample preparation process is as follows: all coal powder is sieved, shaken, and then pressed into tablets using a mold and tablet press to form compact, flat coal cake samples; coal gangue remains in its original form, with only simple surface cleaning. It should be noted that samples involved in subsequent embodiments can also be prepared using this method.

[0025] It should be noted that all existing methods for training and generating coal and rock recognition models fall within the scope of protection of this invention, and the intelligent coal mining system can use any existing method to train and generate the coal and rock recognition model.

[0026] As an example, the LIBS spectral intelligent analysis subsystem is used to determine the target spectral wavelength corresponding to the target spectral line based on the LIBS spectral data corresponding to all samples. If the difference in LIBS spectral intensity between coal and rock at the target spectral line is greater than a preset difference threshold, a confidence interval [BI1, BI2] for the target LIBS spectral intensity is determined based on the target spectral wavelength, and a single-spectral-line coal-rock identification model is generated based on [BI1, BI2]. The LIBS spectral intelligent acquisition subsystem is used to send the LIBS spectral intensity corresponding to the target spectral wavelength of the sampling point to the single-spectral-line coal-rock identification model. If the LIBS spectral intensity corresponding to the target spectral wavelength of the sampling point is within [BI1, BI2], the identification result output by the single-spectral-line coal-rock identification model is coal; otherwise, the identification result is rock. The specific method for generating the single-spectral-line coal-rock identification model is described in Example 3.

[0027] As an example, the LIBS spectral intelligent analysis subsystem is used to perform principal component analysis based on the LIBS spectral data corresponding to all samples, and to determine M principal component extraction modules {W1, W2, ..., W...}. M}, W m The m-th principal component extraction module is used, where m ranges from 1 to M. The LIBS spectral intelligent analysis subsystem uses these M principal component extraction modules to convert the N LIBS spectral data of each sample into M principal grouping component variables (C1, C2, ..., C...). M ), and the M principal grouping component variables (C1, C2, ..., C) of each sample. M Using M principal component extraction modules as input and the corresponding type of the sample as output, a neural network model is trained to generate a neural network classification model; the M principal component extraction modules and the neural network classification model are combined to generate a coal and rock identification model. The LIBS spectral intelligent analysis subsystem determines M principal component extraction modules based on the PCA algorithm, and combines these modules with the neural network classification model to generate a Principal Components Analysis (PCA) coal and rock identification model. The LIBS spectral intelligent analysis subsystem also determines M principal component extraction modules based on the PLS algorithm, and combines these modules with the neural network classification model to generate a partial least-squares (PLS) coal and rock identification model. Specific technical details of principal component analysis-based coal and rock identification are described in Example 4. Specific technical details of partial least-squares-based coal and rock identification are described in Example 5.

[0028] As an example, the intelligent coordinated control subsystem includes a coordinated control module and a motor control module. The coordinated control module is used to acquire the planned cutting path and the operating parameters of the coal mining machine, generate coordinated control commands based on the planned cutting path and the operating parameters of the coal mining machine, and control the motor control module based on the coordinated control commands to realize the control of the cutting motor and traction motor of the coal mining machine.

[0029] Example 1 achieves intelligent acquisition of LIBS spectral data by setting up a LIBS spectral intelligent acquisition subsystem. Based on the LIBS spectral intelligent analysis subsystem, a coal and rock identification model is constructed, and a planned cutting path is generated based on the coal and rock identification model. The intelligent coordination and control subsystem controls the operation of the cutting motor and traction motor based on the planned cutting path and the operating parameters of the coal mining machine. Through the coordinated control of the LIBS spectral intelligent acquisition subsystem, the LIBS spectral intelligent analysis subsystem, and the intelligent coordination and control subsystem, intelligent coal mining is achieved, improving the level of intelligence and mining efficiency.

[0030] Coal mining faces typically consist of irregularly undulating coal walls. When inspecting the coal walls of a coal mining face, the parameters of the laser equipment must meet the safety requirements of underground coal mines, and the difficulties in laser focusing caused by variations in the height of coal and rock on the irregular surface of the coal wall must be overcome to ensure the stability of LIBS spectral acquisition. For in-situ sampling and analysis of the coal wall itself, the most convenient and efficient method is to use an online analyzer to directly emit a laser onto the area to be measured on the coal mining face, and to achieve coverage of the area to be identified on the coal wall by moving the sampling point along a plane formed by the X and Y axes parallel to the coal mining face. However, this method usually suffers from significant measurement errors and high uncertainties. The main reason is that the coal face is not a regular plane. Due to the varying sizes of coal blocks or rock particles, and the cutting action of the cutting drum on the coal face, there are significant irregularities on the surface of the coal face. This causes significant fluctuations in the object distance (Z-axis direction) between the LIBS spectral acquisition device and the sampling point, resulting in poor focusing of the LIBS analyzer on the coal and rock identification sampling points. Consequently, the LIBS spectral data at the sampling points is unstable, and the measurement has large errors, severely impacting the coal and rock identification effect of the coal face. Therefore, when acquiring spectra, the LIBS spectral acquisition device must not only move in the X and Y axes to cover the coal face with the sampling range, but also adjust its position in real time in the Z-axis direction to achieve precise focusing on the sampling points. Based on this, the present invention further proposes Embodiment Two.

[0031] Example 2 Example 2 provides a LIBS spectral intelligent acquisition subsystem, including a LIBS spectral acquisition device and a control device. The LIBS spectral acquisition device includes a laser, a LIBS spectral acquisition unit, an acquisition track, and a moving device. The laser and the LIBS spectral acquisition unit are mounted on the moving device, which is mounted on the acquisition track. The moving device can move along the acquisition track at preset sampling intervals. The acquisition track is located parallel to the coal mining face. The moving device can also move in a direction perpendicular to the coal mining face.

[0032] When the LIBS spectral intelligent acquisition subsystem samples the sampling points on the coal mining face, the control device is used to determine whether the object distance between the current sampling point and the LIBS spectral acquisition device is equal to the preset fixed object distance. If they are not equal, the control device moves the LIBS spectral acquisition device in a direction perpendicular to the coal mining face so that the object distance between the current sampling point and the LIBS spectral acquisition device is equal to the preset fixed object distance, thereby maintaining a stable object distance.

[0033] The laser is used to generate a laser pulse and focus it on the surface of the sampled material when the object distance between the sampled material and the LIBS spectral acquisition device is equal to a preset fixed object distance. The surface of the sampled material is ablated to form plasma and emit a spectrum. As an example, based on the safety requirements of coal mine environments and to ensure a certain degree of safety redundancy, this embodiment of the invention uses a compact nanosecond solid-state laser with spectral parameters of a spectral wavelength of 532 nm and a laser energy of 25 mJ / pulse. It is small in size, compact in structure, robust and durable, and highly engineered.

[0034] The LIBS spectrometer is used to acquire LIBS spectral data corresponding to the sampling point when the object distance between the sampled substance and the LIBS spectrometer is equal to a preset fixed object distance. As an example, the LIBS spectrometer has high resolution and a wide spectral coverage range; the resolution can be specifically set to 0.07 nm, and the spectral coverage range can be specifically set to 0-1500 nm. The LIBS spectrometer has a calibration system to enable the LIBS system to achieve a high signal-to-noise ratio; equipped with a high-efficiency optical laser focusing and signal collection system, it can achieve a lower detection limit and a shorter integration time.

[0035] This embodiment utilizes a moving device to maintain a stable object distance, enabling the LIBS spectral acquisition device to automatically, quickly, and accurately focus on irregular coal surface. That is, without changing the original focal length of the LIBS spectral acquisition device, the LIBS spectral acquisition device is mounted on the moving device, and the movement of the moving device on the track offsets changes in object distance, ensuring that the laser emitted by the laser is always focused at the ideal position of the sampling point, thus achieving dynamic and precise focusing on irregular surfaces.

[0036] Figure 2 Two different focusing methods are shown; the area within the dashed box is the LIBS spectral acquisition device. Figure 2 This is a simplified diagram of the intelligent sampling device (hereinafter referred to as the intelligent sampling device). States 1 to 3 represent focusing achieved through optical path adjustment of the LIBS spectral acquisition unit (hereinafter referred to as the LIBS instrument in the diagram). States 1 to 2 represent using the intelligent sampling device to stabilize the fixed object distance between the LIBS analyzer and the coal wall sampling point. Figure 2 The object distance (D) is used to ensure that the object distance is always matched with the focal length of the LIBS analyzer for focusing. The intelligent sampling device uses a moving device (i.e., Figure 2 The focusing process can be achieved by using a high-precision position servo control on a mobile platform. The device has a simple structure and is easy to implement. Its response speed and accuracy are better than optical path zoom. Moreover, it does not require adjustment of the laser energy value of the laser source. Therefore, the method of using an intelligent sampling device for assisted focusing is more feasible.

[0037] As an example, the control device is also used to obtain the coal and rock type of the sampling point based on the LIBS spectral data corresponding to the sampling point, and dynamically adjust the sampling interval according to the change of coal and rock type of V consecutive sampling points of different types, where V≥2. The coal and rock types include clean coal, coal slime, raw coal, middlings, sandy mudstone, and gangue, etc. Different coal and rock types have different gangue contents, and the sampling interval is inversely proportional to the amount of gangue.

[0038] Achieving precise movement of the mobile device is crucial for maintaining a fixed distance from the object. However, this mobile device involves numerous uncertainties and complexities, such as unstable motor torque, variations in track friction torque, dead zones of components, and the accuracy of position sensors. Establishing an accurate mathematical model presents significant challenges. When using traditional PID control, these complexities are often not adequately considered or addressed, resulting in suboptimal control performance. PID control refers to proportional-integral-derivative control.

[0039] To achieve precise movement of the mobile device, as an example, the mobile device includes a rangefinder, a BP (BackPropagation) neural network model, a PID controller, and a moving motor. Precise control of the moving motor is key to achieving a fixed object distance. The rangefinder is used to monitor the current distance y(k) between the sampling point and the mobile device in real time. The PID controller determines the current object distance difference e(k) based on the current distance y(k) and a preset fixed object distance u(k), and inputs the current object distance difference e(k) into the BP neural network. The BP neural network outputs the proportional control parameter K of the moving motor based on the current object distance difference e(k). p Integral control parameter K of the moving motor i and the differential control parameter K of the moving motor d The PID controller is also used to obtain the K output of the BP neural network. p K i K d Based on K p K i K d Output the target distance u(k), and set K p K i K dThe system operates on the mobile motor to control the movement of the mobile device, acquiring the actual distance v(k). The actual distance v(k) and the target distance u(k) are then input into the BP neural network. The BP neural network model is also used to update its parameters based on the actual distance v(k), the target distance u(k), and the object distance difference e(k). By combining the BP neural network model with a PID controller, the intelligent algorithm of the BP neural network model can adjust various parameters of the motor servo control system in real time according to the operational requirements of the LIBS spectral acquisition device, eliminating negative interference from various nonlinear factors, ensuring the entire LIBS spectral acquisition device always operates at a good level, and greatly improving the control accuracy of the mobile motor. The combination of PID control and intelligent algorithms enhances the dynamic and precise focusing effect of the LIBS spectral acquisition device.

[0040] As an example, the BP neural network model is as follows: in, Let u(k) be the difference between the target distance and the actual distance v(k), e(k-1) be the difference between the previous object distance and e(k-2) be the difference between the previous two object distances.

[0041] As an example, the moving motor is a servo motor, AC motor, synchronous motor, or asynchronous motor, etc.

[0042] Example 2 involves setting up a LIBS spectral acquisition device and a control device. The control device dynamically controls the position of the LIBS spectral acquisition device based on changes in the coal mining face, ensuring that LIBS spectral acquisition is always achieved when the object distance between the current sampling point and the LIBS spectral acquisition device is equal to the preset fixed object distance. This guarantees the accurate focusing of the LIBS spectral acquisition device and improves the spectral acquisition quality and coal and rock identification effect.

[0043] The core technology of the LIBS spectral intelligent analysis subsystem lies in how to achieve coal petrification identification. The spectral lines contained in LIBS spectra correspond to unique transitions of specific atoms, ions, or molecules. After the spectral information is effectively collected by the LIBS spectral intelligent acquisition unit, it can be used as a "fingerprint" to identify the elemental composition of a sample. Therefore, elemental analysis can be used to determine the coal petrification classification of the sample. However, the LIBS spectra of coal exhibit significant matrix effects and self-absorption effects, resulting in computationally complex and unsatisfactory results when using LIBS spectroscopy for elemental analysis of coal. Therefore, it is necessary to establish a reasonable and effective LIBS coal petrification identification model to achieve rapid and accurate identification of coal petrification. Based on this, Example 3 is further proposed.

[0044] Example 3 Example 3 provides a coal and rock identification method based on LIBS single spectral lines, such as... Figure 3 As shown, it includes: Step S1: Obtain coal sample set and rock sample set. Each sample includes N LIBS spectral data. The LIBS spectral data includes spectral line wavelength and corresponding LIBS spectral intensity.

[0045] The coal and rock sample sets can come from the same mining face or from multiple different mining faces. If the coal and rock sample sets come from the same mining face, the generated single-spectral coal-rock identification model is only applicable to that mining face and lacks universality. If the coal and rock sample sets come from multiple different mining faces, the generated single-spectral coal-rock identification model has universality. It is understandable that the larger the sample size, the higher the accuracy of the single-spectral coal-rock identification model.

[0046] Step S2: Based on the LIBS spectral intensity of all samples in the coal and rock sample sets, determine the target spectral wavelength that can significantly distinguish between coal and rock.

[0047] To achieve the goal of coal and rock identification of unknown samples, it is necessary to find the target spectral wavelength that is significantly different between samples in the "coal" group and the "rock" group, and to construct a coal and rock identification model based on the target spectral wavelength.

[0048] Step S3: Construct a single-spectral-line coal and rock identification model based on the LIBS spectral intensity data corresponding to the target spectral line wavelengths of all samples.

[0049] Step S4: Obtain the LIBS spectral intensity of the target spectral line wavelength corresponding to the sampling point on the target working surface, input it into the single-spectral-line coal and rock identification model, and generate the coal and rock identification result of the sampling point.

[0050] As an example, the coal sample set is {X1, X2, ..., X...} r ,…,X R The rock sample set is {Y1, Y2, ..., Y}. s ,…Y S}, where X r Let Y be the r-th sample of type coal, where r ranges from 1 to R, and R is the total number of coal samples; s Let X be the s-th sample of type rock, where s ranges from 1 to S, and S is the total number of rock samples; r =[(A1 Xr B1 Xr ),(A2 Xr B2 Xr ),…,(A n Xr B n Xr ),…, (AN Xr B N Xr )], Y s =[(A1 Ys B1 Ys ), (A2 Ys B2 Ys ),…,(A n Ys B n Ys ),…, (A N Ys B N Ys )] ;(A n Xr B n Xr ) is X r The nth LIBS spectral data, A n Xr For X r The corresponding wavelength of the nth spectral line, B n Xr For X r In A n Xr The corresponding LIBS spectral intensity; (A n Ys B n Ys ) is Y s The nth LIBS spectral data, A n Ys For Y s The corresponding wavelength of the nth spectral line, B n Ys For Y s In A n Ys The corresponding LIBS spectral intensity, where n ranges from 1 to N, and N is the total number of LIBS spectral data corresponding to each sample. N is determined based on the sampling accuracy of the LIBS spectral intelligent acquisition subsystem. A n Xr A n Ys All of them belong to the preset band range, which is [196.08nm, 507.78nm].

[0051] As an example, step S2 includes: Step S21, put all A n Xr Corresponding B n Xr And all A nYs Corresponding B n Ys Mix them and generate each B in order of increasing LIBS spectral intensity. n Xr The corresponding rank T i Xrn And each B n Ys The corresponding rank T j Ysn The value of i ranges from 1 to R, and the value of j ranges from 1 to S. Step S22: Obtain B at the wavelength of the nth spectral line. n Ys Greater than B n Xr The number W XY Based on T i Xrn Determine the rank-sum statistic W for coal samples X Based on T j Ysn Determine the rank-sum statistic W of the rock sample Y : W X = ; W Y = Step S23: Based on W corresponding to the wavelength of the nth spectral line XY W X W Y A rank-sum test is performed to generate a test P-value corresponding to the wavelength of the nth spectral line. The test P-value is used to represent the probability that there is no difference in the LIBS spectral intensity of the coal sample and the rock sample at the wavelength of the nth spectral line.

[0052] It should be noted that the test P-value for each spectral line wavelength can be obtained based on the Wilcoxon rank-sum test, which will not be elaborated here.

[0053] Step S24: Determine the spectral line wavelengths whose test P value is less than the preset test P value threshold as candidate spectral line wavelengths, and construct a candidate spectral line wavelength set.

[0054] Step S25: Select a target spectral line wavelength from the candidate spectral line wavelength set.

[0055] It should be noted that when the sample size is greater than or equal to the preset sample size threshold, the accuracy of the data can be guaranteed. At this time, any spectral line wavelength randomly selected from the candidate spectral line wavelength set can be used as the target spectral line wavelength. As a preferred example, when the sample size is greater than or equal to the preset sample size threshold, in said step S25, the spectral line wavelength with the smallest test P-value in the candidate spectral line wavelength set is determined as the target spectral line wavelength. When the accuracy of the test P-value is guaranteed, the smaller the test P-value is, the greater the difference in LIBS spectral intensity between the corresponding coal sample and rock sample is, and the easier it is to distinguish coal from rock.

[0056] When the sample size is smaller than the preset sample size threshold, it may be affected by other factors, which makes it impossible to guarantee the accuracy of the test P-value corresponding to each spectral line wavelength. Therefore, directly randomly selecting or selecting the spectral line wavelength with the smallest test P-value may fail to select a spectral line wavelength that can significantly distinguish coal samples and rock samples. As an example, the step S25 comprises: Step S251: acquiring a preset element spectral line mapping table, wherein the preset element spectral line mapping table includes H preset element spectral line wavelengths (AZ1, AZ2, …, AZ h , …, AZ H ), AZ h is the h-th preset element spectral line wavelength, and the significance of AZ1, AZ2, …, AZ h , …, AZ H that can distinguish between target spectral wavelengths of coal and rock decreases in sequence.

[0057] Wherein, the preset element spectral line wavelengths (AZ1, AZ2, …, AZ h , …, AZ H ) can be directly specified by the user.

[0058] Step S252: setting h=1.

[0059] Step S253: determining whether AZ h belongs to said candidate spectral line wavelength set. If it belongs, AZ h is determined as the target spectral line wavelength, and the process is ended; if it does not belong, step S254 is performed.

[0060] Step S254: if h<H, setting h=h+1, returning to perform step S253; if h=H, determining the spectral line wavelength with the smallest test P-value in the candidate spectral line wavelength set as the target spectral line wavelength.

[0061] As an example, the final target spectral wavelength was determined to be 247.95 nm, corresponding to element C. It should be noted that when the sample size is less than a preset sample size threshold, other factors may influence the selection, potentially compromising the accuracy of the test P-value for each spectral wavelength. The test P-value for the spectral wavelength corresponding to element C might be too high, preventing its selection as the target spectral wavelength. However, through steps S251-S254, AZ1 can be set as the spectral wavelength corresponding to element C. As long as the spectral wavelength corresponding to element C is also in the candidate spectral wavelength set, it can be determined as the target spectral wavelength. It is understood that the above is merely an example, and the spectral wavelength corresponding to element C is not necessarily set to AZ1.

[0062] As an example, step S3 includes: Step S31: Obtain the spectral light average value of the LIBS spectral intensity corresponding to the target spectral line wavelength of all coal samples in the coal sample set. and spectral intensity standard deviation . Step S32, based on and Determine the confidence interval for the sample type as coal. The confidence level of the confidence interval is b, where a is a preset coefficient and R is the total number of coal samples in the coal sample set.

[0063] As an example, the value of a is 1.96, and the value of b is 95%.

[0064] As an example, step S4 includes: Step S41: Obtain the LIBS spectral intensity of the target spectral line wavelength corresponding to the sampling point on the target working surface, and input it into the single-spectral-line coal and rock identification model.

[0065] Specifically, the LIBS spectral intelligent acquisition subsystem in Embodiment 2 can be used to obtain the LIBS spectral intensity of the target spectral line wavelength corresponding to the sampling point. It should be noted that after the single-spectral-line coal and rock identification model is established, it is no longer necessary to collect LIBS spectral data based on the preset band interval. The band interval can be narrowed down to include only the target spectral line wavelength, thereby reducing the amount of data collected.

[0066] Step S42: If the LIBS spectral intensity of the target spectral line wavelength corresponding to the sampling point is within the confidence interval, then the type of the sampling point is determined to be coal; otherwise, the type of the sampling point is determined to be rock.

[0067] Example 3: By determining the target spectral wavelength that can significantly distinguish between coal and rock, a single-spectral-line coal and rock identification model is constructed based on the LIBS spectral intensity data corresponding to the target spectral wavelength of all samples. Based on the single spectral line, the coal and rock type identification of the working face sampling points is realized. With sufficient sample size, the accuracy of coal and rock classification is improved, thereby improving the accuracy and reliability of generating cutting path planning, and reducing the computational load and cost of coal and rock identification.

[0068] Single-line coal and rock identification models utilize only one spectral line from the original LIBS spectrum of coal and rock samples for identification. The data information contained in single-line models is very limited, and the analysis is often isolated. Furthermore, due to matrix effects, self-absorption effects, inter-elemental interactions, and plasma parameters, spectral line overlap and inter-elemental interference are common in LIBS spectra. The information contained in a single spectral line may come not only from a specific element and its abundance but also from other factors present in the plasma or even noise interference. With limited sample size, single-line coal and rock identification models have low resolution of the original LIBS spectral data, resulting in less than ideal identification performance. The only way to improve the model's accuracy is to increase the model's sample size. However, most coal mining faces require coal and rock identification models with small sample size modeling capabilities. To achieve good identification accuracy with small sample size modeling, it is necessary to find coal and rock classification patterns from as much spectral information as possible and to handle the complex relationships between spectral lines caused by various influencing factors, thereby improving the accuracy of the identification model. Large datasets composed of multiple variables undoubtedly provide rich information for coal and rock classification; however, many variables may be correlated, potentially increasing data redundancy. Furthermore, performing classification and regression on multiple input variables is computationally intensive, and the computation time and hardware costs far exceed the permissible limits for coal and rock identification in actual coal mining operations. Based on this, the present invention further proposes Embodiment Four.

[0069] Example 4 Example 4 provides a coal and rock identification method based on principal component analysis, including: Step C1: Obtain coal sample sets and rock sample sets. Each sample includes N LIBS spectral data, which include spectral line wavelengths and corresponding LIBS spectral intensities.

[0070] The coal and rock sample sets can come from the same mining face or from multiple different mining faces. If the coal and rock sample sets come from the same mining face, the generated binary classification model is only applicable to that mining face and lacks universality. If the coal and rock sample sets come from multiple different mining faces, the generated binary classification model has universality.

[0071] Step C2: Determine the number of principal components D based on all samples in the coal and rock sample sets, and the D principal component extraction framework {z(x)}. 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D}, where z(x) d Let z(x) be the principal component extraction architecture for the d-th principal component, where d ranges from 1 to D, N >> D. 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D They are orthogonal to each other, z(x) 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D The sum of the explanatory power of the overall sample data is greater than the preset explanatory power threshold.

[0072] Wherein, through z(x) 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D To interpret as much raw LIBS data as possible, for each z(x)... d All are linear transformations of the original variables, and the components are orthogonal. Each principal component retains the maximum information from the original data along its corresponding coordinate axis. In the subsequent construction of the binary classification model, z(x) will be the primary consideration. 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D The data structure is greatly simplified within the allowable range of information loss.

[0073] Step C3, based on {z(x)} 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D Extract D principal components corresponding to each sample from the coal sample set and the rock sample set.

[0074] Step C4: Use the D principal components corresponding to each sample in the coal sample set and the rock sample set as input to the preset classification model, and use the coal-rock binary classification result as output to train the binary classification model.

[0075] It is understood that the binary classification model is the principal component analysis coal and rock identification model mentioned in Example 1.

[0076] Step C5: Identify coal and rock based on the binary classification model.

[0077] As an example, step C5 includes: Step C51: Obtain N LIBS spectral data corresponding to the sampling points on the target working surface, and input them into D principal component extraction architectures respectively to generate D principal component variables corresponding to the sampling points; Step C52: Input the D principal component variables corresponding to the sampling points into the binary classification model to generate the coal and rock identification results of the sampling points.

[0078] As an example, the coal sample set is {X1, X2, ..., X...} r ,…,X R The rock sample set is {Y1, Y2, ..., Y}. s ,…Y S}, where X r Let Y be the r-th sample of type coal, where r ranges from 1 to R, and R is the total number of coal samples; s Let X be the s-th sample of type rock, where s ranges from 1 to S, and S is the total number of rock samples; r =[(A1 Xr B1 Xr ),(A2 Xr B2 Xr ),…,(A n Xr B n Xr ),…, (A N Xr B N Xr )], Y s =[(A1 Ys B1 Ys ), (A2 Ys B2 Ys ),…,(A n Ys B n Ys ),…, (A N Ys B N Ys )] ;(A n Xr B n Xr ) is X r The nth LIBS spectral data, A n Xr For X r The corresponding wavelength of the nth spectral line, B n XrFor X r In A n Xr The corresponding LIBS spectral intensity; (A n Ys B n Ys ) is Y s The nth LIBS spectral data, A n Ys For Y s The corresponding wavelength of the nth spectral line, B n Ys For Y s In A n Ys The corresponding LIBS spectral intensity, where n ranges from 1 to N, and N is the total number of LIBS spectral data points for each sample. A n Xr A n Ys All belong to the preset band range. As an example, the preset band range is [196.08nm, 507.78nm]. N is determined based on the sampling accuracy of the LIBS spectral intelligent acquisition subsystem. In this example, the value of N is 6074.

[0079] This embodiment utilizes principal component analysis (PCA) to effectively reduce the dimensionality of LIBS spectral data. PCA is a statistical analysis method that reduces dimensionality by decreasing the correlation between variables through orthogonal transformation. While minimizing the loss of original data information, PCA can effectively simplify the dataset, transforming potentially correlated multidimensional variables into a smaller number of uncorrelated principal component variables. In this embodiment, D is set to 3, and z(x)... 1 The explanatory power of z(x) for the overall sample data is 0.616. 2 The explanatory power of z(x) for the overall sample data is 0.316. 3 The explanatory power of z(x) for the overall sample data is 0.0286. 1 z(x) 2 z(x) 3 The sum is 0.9606, and the first three principal components collectively retain 96% of the spectral information of the original data. Increasing to four principal components does not significantly improve the overall data interpretability; therefore, considering the effect of simplifying the data structure, the first three principal components are used.

[0080] Figure 4 A schematic diagram illustrating the dimensionality reduction effect of PCA is shown. Through PCA analysis, the data structure of the LIBS spectrum was reduced from 6074 dimensions to 3 dimensions. Figure 4The PCA scatter plot shows that the spatial distribution of the coal and rock sample sets clearly reflects the differences between the groups, indicating that there are significant differences between coal and rock in LIBS spectral statistical characteristics. This achieves the expected effect of classifying coal and rock based on LIBS spectral statistical characteristics, and is fully reflected by the PCA analysis algorithm.

[0081] As an example, step C2 includes: Step C21: Construct the initial principal component extraction architecture: z(x) = l1x1 + l2x2 + ... + l n x n …+L N x N , where l n x is the coefficient corresponding to the wavelength of the nth spectral line. n This represents the LIBS spectral intensity variable corresponding to the wavelength of the nth spectral line. Step C22: Input all samples from the coal sample set and the rock sample set into z(x) = l1x1 + l2x2 + ... + l n x n …+L N x N Adjust l1, l2, ... l n …、L N Generate group D l1 d l2 d ...l n d …、L N d Simultaneously satisfying: Condition 1, each group l1 d l2 d ...l n d …、L N d satisfy .

[0082] Condition 2, for each group l1 d l2 d ...l n d …、L N d z(x) satisfies the condition for all samples d The difference between the variance of the sample and the original variance of all samples in the coal and rock sample sets is less than the preset variance threshold.

[0083] Condition 3, Group D, l1 d l2 d ...l n d …、L N dSatisfying z(x) 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D The sum of the explanatory power of the corresponding total sample data is greater than the preset explanatory power threshold. Step C23, based on group D l1 d l2 d ...l n d …、L N d Generate {z(x)} 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D}

[0084] As an example, the preset classification model is a binary classification neural network model. Specifically, the neural network model may include D inputs, 2 outputs, and 2 hidden layers, wherein the first hidden layer includes 20 neurons and the second hidden layer includes 10 neurons.

[0085] Analysis of the PCA coal and rock identification model using Matlab shows that its goodness of fit for binary classification of coal and rock samples is approximately 1, indicating a high degree of fit. Matlab randomly selected samples from the modeling sample library as the training set to train the coal and rock identification model, and verification experiments showed that all samples were correctly identified. Furthermore, the self-learning effect of the neural network classifier changes each time the samples selected in the training set change. Due to the limited number of modeling samples, multiple modeling and recognition verifications are necessary by continuously changing the sample data in the training and test sets to find the optimal recognition accuracy of the established coal and rock identification model and observe the stability of its recognition performance. Through multiple recognition verification experiments with continuously changing the sample composition of the training and test sets, the PCA coal and rock identification model can stably maintain a 100% recognition accuracy for binary classification of coal and rock.

[0086] The PCA coal and rock identification model, constructed based on statistical analysis algorithms and combined with neural network algorithms, not only transforms a "single" variable analysis of spectral line variables into a "multi" variable analysis of three principal component variables, but also significantly improves the quality of the variables. This is because each principal component variable is actually derived from a linear transformation of all variables in the original spectrum, containing a large amount of original spectral information. Furthermore, a neural network classifier is incorporated when performing classification and regression using multiple principal component variables. The most prominent advantage of neural networks is that they do not require establishing a specific mathematical model of the object; they only need to know the input and output information of the object to establish a mapping relationship from the input space to the output space. This model fully utilizes the characteristics of artificial intelligence algorithms, achieving not only excellent recognition results but also endowing the PCA coal and rock identification model with self-learning and adaptive capabilities, making it intelligent and capable of coping with complex geological conditions. Compared with methods that directly use 6074 variables from the original spectrum for classification, the PCA coal and rock identification model's neural network classifier input signal only has three principal component variables, greatly simplifying the network structure and computational complexity of the neural network, reducing the algorithmic pressure on the classifier, and significantly reducing the time and space overhead of the algorithm implementation. This lowers the hardware and software costs of implementing the intelligent coal and rock sensing system at the model level.

[0087] In Example 4, principal component analysis was first performed on the coal and rock sample sets to extract multiple principal components. The sum of the explanatory power of the corresponding components for the overall sample data was greater than the preset explanatory power threshold. The principal component corresponding to each sample was used as the input of the preset classification model, and the coal-rock binary classification result was used as the output of the preset classification model. The binary classification model was trained, which greatly simplified the complexity of the sample input data and improved the accuracy of coal-rock classification, thereby improving the accuracy and reliability of the cutting path planning.

[0088] The LIBS-based coal and rock identification method is essentially a "point identification" method. It uses the coal and rock identification results from sampling points scattered throughout the coal face to describe the coal and rock distribution, ultimately aiming to characterize the coal-rock interface and plan the cutting trajectory of the mining machine. For a coal and rock area to be measured in a coal face, the more sampling points, the higher the identification resolution, and the more accurate the characterization of the coal-rock interface. However, the number of sampling points also affects the time required to complete coal and rock identification in that area; identification resolution and speed are mutually restrictive. To characterize the coal-rock interface as quickly and accurately as possible under certain conditions, the LIBS coal and rock identification model needs further optimization to improve the effective utilization rate of sampling points and maximize identification accuracy while limiting the number of sampling points.

[0089] The identification resolution near the coal-rock interface directly determines whether the coal mining machine will cause "missed mining" or "over-mining" of coal due to excessive sampling intervals. Clearly, the higher the identification resolution, the better the effect of avoiding "missed mining" and "over-mining." The identification resolution in other areas is less important for coal-rock identification. The area near the coal-rock interface is defined as the sensitive area, and other areas as non-sensitive areas. Effective identification of the sensitive area becomes crucial for achieving dynamic sampling. The closer to the coal-rock interface, the higher the content of impurities (coal gangue, roof and floor mudstone, etc.) in the coal, forming a "coal-rock transition zone" with varying coal quality. This characteristic of coal quality change in the coal-rock transition zone is used as the basis for determining sensitive areas. Therefore, the coal-rock identification model must be able to accurately identify coal quality changes to achieve dynamic identification resolution coal-rock characterization of the area to be measured in the working face. In summary, the LIBS coal-rock identification model, in addition to achieving binary coal-rock classification, must also have the ability to identify coal quality changes; therefore, a multi-classification LIBS coal-rock identification model is needed.

[0090] When performing coal-rock binary classification based on PCA, coal and rock can be distinguished. However, when coal and rock types are further subdivided, the spatial distinguishability between different coal types obtained by PCA is limited, and samples of different coal types may be mixed together. This indicates that the data differences within coal group samples are not ideal under PCA analysis. Therefore, the accuracy of PCA coal-rock identification models based on PCA principal component analysis is also affected when performing multi-class identification, meaning that PCA coal-rock identification models are not suitable for multi-class coal-rock identification. During principal component extraction in PCA analysis, dimensionality reduction is mainly achieved by reducing the correlation between the wavelengths of independent variables' spectral lines, ignoring the interpretative role of independent variable spectral line wavelengths in the dependent variable coal-rock identification result. This may lead to highly variable noise being treated as key information and calculated into the principal components, while variables that are truly related to the dependent variable coal-rock identification result may be ignored. This is the reason why the PCA-based coal-rock identification model has a low recognition accuracy in multi-class identification. Based on this, the present invention further proposes Example 5.

[0091] Example 5 This invention provides a coal and rock identification method based on partial least squares, comprising: Step E1: Obtain P1 coal sample sets and Q1 rock sample sets, where P1≥2 and Q1≥1. Different sample sets correspond to different coal and rock identification types, and the samples in different sample sets correspond to different gangue contents. Each sample includes N LIBS spectral data, which include spectral line wavelengths and corresponding LIBS spectral intensities.

[0092] The samples can be taken from the same coal face or from different coal faces. The P1 coal sample sets include clean coal, middlings coal, and raw coal samples; the Q1 rock sample sets include sandy mudstone and gangue samples, with corresponding coal-rock classification results including clean coal, middlings coal, raw coal, sandy mudstone, and gangue. Raw coal refers to coal excluding visible gangue, without any other treatment. Clean coal refers to high-quality coal after washing. Middlings coal has a gangue content between raw coal and clean coal. The quality of clean coal, middlings coal, and raw coal decreases progressively, while the gangue content increases sequentially. These three samples can simulate the coal quality changes in the coal-rock transition zone of a coal face. Furthermore, the P1 coal sample sets also include a coal slime sample set, with corresponding coal-rock classification results including coal slime. Coal slime refers to a semi-solid substance formed from water-containing coal powder. Coal slime samples are used to simulate coal wetted by water mist used to cool coal mining machines. Because coal slime contains water, its composition and LIBS spectrum differ from normal coal. Therefore, accurately identifying coal slime helps reduce the interference of cooling water mist on coal and rock identification, ensuring the effectiveness of coal and rock identification.

[0093] Step E2: Determine the number of principal components F based on all samples in the P1 coal sample sets and Q1 rock sample sets, and define the F principal component extraction framework {V(x)}. 1 V(x) 2 V(x) f V(x) F}, where V(x) f Let f be the principal component extraction architecture, where f ranges from 1 to F, N >> F, and V(x). 1 V(x) 2 V(x) f V(x) F The sum of the explanatory power of the corresponding total sample data is greater than the preset first explanatory power threshold; V(x) 1 V(x) 2 V(x) f V(x) F The sum of the explanatory power of the corresponding independent variable spectral line wavelengths for the dependent variable coal and rock classification results is greater than the preset second explanatory power threshold.

[0094] The partial least squares (PLS) method was used to determine the number of principal components F and the extraction framework for F principal components. PLS is a multivariate analysis method that simplifies the data structure while considering the correlation between independent and dependent variables. It projects the high-dimensional data space of the independent variable (spectral wavelength) and the dependent variable (coal and rock identification results) into a corresponding low-dimensional space, obtaining the mutually orthogonal principal components of the independent variable (spectral wavelength) and the dependent variable (coal and rock identification results). Then, a linear regression relationship is established between the principal components of the independent variable (spectral wavelength) and the dependent variable (coal and rock identification results). The detailed algorithm of PLS ​​is not elaborated here. As an example, F is set to 3, and V(x)... 1 The explanatory power of V(x) for the overall sample data is 0.615. 2 The explanatory power of V(x) for the overall sample data is 0.316. 3 The explanatory power of V(x) for the overall sample data is 0.028. 1 V(x) 2 V(x) 3 The sum is 0.959, and the first three principal components collectively retain 95.9% of the spectral information of the original data. Increasing to four principal components does not significantly improve the overall data interpretability; therefore, considering the effect of simplifying the data structure, the first three principal components are used.

[0095] Step E3, based on {V(x)} 1 V(x) 2 V(x) f V(x) F Extract F principal components corresponding to each sample in the coal sample set and the rock sample set.

[0096] It should be noted that, based on {V(x)} 1 V(x) 2 V(x) f V(x) F Extracting principal components can emphasize the explanatory role of the independent variable spectral line wavelength on the dependent variable coal and rock identification results, and can also reduce the influence of noise, which is beneficial to coal and rock identification and classification.

[0097] Step E4: Take the F principal components corresponding to each sample in the coal sample set and the rock sample set as the input of the preset classification model, and take the coal-rock W classification result as the output of the preset classification model to train the W classification model, where W>2.

[0098] Step E5: Identify coal and rock based on the W classification model.

[0099] When performing coal and petrography multi-classification based on partial least squares (PLS), the identification results are significantly improved compared to those based on PCA, both between coal and petrography samples and within each coal and petrography group. As an example, step E5 includes: Step E51: Obtain N LIBS spectral data corresponding to the sampling points on the target working surface, and input them into F principal component extraction architectures to generate F principal component variables corresponding to the sampling points.

[0100] Step E52: Input the F principal component variables corresponding to the sampling points into the W classification model to generate the coal and rock identification results of the sampling points.

[0101] As an example, step E5 further includes: Step E53: Obtain the coal and rock identification results of V consecutive sampling points on the target working face, and dynamically adjust the sampling interval of the sampling points according to the changes in the coal and rock type of the V consecutive sampling points. The sampling interval is inversely proportional to the amount of gangue.

[0102] Based on step E53, intelligent distribution of sampling points can be achieved, dynamic recognition resolution can be realized for different areas, the effective utilization rate of sampling points can be improved, and the coal-rock interface can be more accurately depicted with the same number of sampling points.

[0103] Taking a simulated coal mining face with randomly distributed coal-rock interfaces as an example, the model is 1.7 meters long and 1.2 meters wide. The coal-rock transition zone near the coal-rock interface of the physical model is simulated by coal of different qualities. Figure 5 The results of coal and rock identification using the PCA coal and rock identification model at a fixed resolution are shown. Because the PCA model cannot effectively identify changes in coal quality, it cannot determine sensitive areas. Therefore, the sampling point interval cannot be dynamically adjusted during the identification process; the entire identification process is completed at a fixed resolution with a sampling point interval of 0.1 meters. The identification results of each sampling point are used as the identification results for the corresponding grid. A total of 204 sampling points are evenly distributed in the area to be identified. The coal and rock distribution of the simulated coal mining face will be characterized by these 204 grids.

[0104] Figure 6The results of coal and rock identification using the PLS coal and rock identification model at dynamic resolution are shown. Because the PLS model can effectively identify changes in coal quality, the sampling point interval can be dynamically adjusted during the identification process after identifying sensitive areas, thus completing the entire identification process with dynamic resolution. Sampling points are identified on the simulated coal mining face model using a left-to-right line-by-line scanning method. During the identification process, the horizontal sampling interval is adjusted in real time according to changes in coal quality. When a large amount of gangue is detected in the coal, the sampling interval is shortened; conversely, the interval is increased. The final coal and rock distribution of the simulated coal mining face is characterized by 184 squares, a reduction of approximately 10% compared to the 204 sampling points used with a fixed resolution. The identification results show that the PLS coal and rock identification model has a more intelligent sampling point distribution and a more accurate depiction of the coal-rock interface. In actual planning of the coal mining machine's cutting trajectory, this can effectively reduce the possibility of "over-mining" and "under-mining." As an example, the spectral wavelength corresponding to each sample belongs to the preset band range, which is [196.08nm, 507.78nm].

[0105] As an example, the preset classification model is a W-classification neural network model.

[0106] In the identification and verification of the modeling sample library, the PLS coal and rock identification model can maintain a 6-class classification accuracy rate of over 94.4%, reaching a maximum of 100%. The PLS coal and rock identification model outperforms the PCA coal and rock identification model in 6-class classification accuracy, which is consistent with PLS's advanced algorithm and powerful spectral analysis capabilities. The PLS coal and rock identification model can achieve multi-class classification of coal and rock, possessing the ability to identify coal quality changes in the mining face, providing theoretical basis and data support for determining sensitive areas, and achieving intelligent dynamic coal and rock identification resolution. The PLS coal and rock identification model is suitable for mining faces with a large range of coal quality variations and high requirements for the accuracy of coal-rock interface characterization.

[0107] Example 5 uses partial least squares to perform principal component analysis on coal and rock sample sets, extracting multiple principal components. This not only reduces the dimensionality of the input sample data but also incorporates the interpretative role of spectral wavelengths in coal and rock identification results. The principal component corresponding to each sample is used as the input to a pre-defined classification model, and the binary classification result of coal and rock is used as the output of the pre-defined classification model. This trains a multi-classification model, greatly simplifying the complexity of the input sample data and improving the accuracy of coal and rock classification, thereby enhancing the accuracy and reliability of cut-off path planning.

[0108] While both PCA and PLS algorithms can reduce LIBS spectral data from 6074 spectral line variables to 3 principal component variables, greatly simplifying the spectral data structure while retaining sufficient spectral information, their analysis is based on a 6074×60 dataset. Furthermore, each principal component is calculated from all variable data of the original spectrum, and its expression includes all spectral line data. The statistical analysis and principal component calculation process remains relatively complex, placing high demands on data processing chips and data storage space for engineering applications. Therefore, this invention further proposes Embodiment Six.

[0109] Example 6 Example 6 provides a LIBS spectral data processing method based on variable projection importance, including: Step F1: Obtain P1 coal sample sets and Q1 rock sample sets, where P1≥2 and Q1≥1. Different sample sets correspond to different coal and rock identification types, and the samples in different sample sets correspond to different gangue contents. Each sample includes N LIBS spectral data, which include spectral line wavelengths and corresponding LIBS spectral intensities.

[0110] Step F2: Generate a weight value VIP for the influence of each spectral wavelength on the final coal and rock identification result based on the LIBS spectral data of all samples in the P1 coal sample set and the Q1 rock sample set. Step F3: Sort all VIP values ​​in descending order, and select the spectral wavelengths corresponding to the top L VIP values ​​as the candidate spectral wavelength set {A1}. VIP A2 VIP ,…,A l VIP ,…,A L VIP}, A l VI Let A1 be the wavelength of the l-th candidate spectral line, where l ranges from 1 to L. VIP A2 VIP ,…,A l VIP ,…,A L VIP The corresponding VIP values ​​decrease sequentially. Step F4: Obtain each A l VIP The corresponding element, for any two candidate spectral line wavelengths A l1 VIP and A l2 VIP If A l1 VIP and A l2 VIPThe corresponding elements are the same, and A l1 VIP VIP value greater than A l2 VIP If the VIP value is A, then A l2 VIP From the candidate spectral line wavelength set, a target simplified spectral line wavelength set is generated. The number of spectral line wavelengths in the target simplified spectral line wavelength set is much smaller than N. The spectral line wavelengths in the target simplified spectral line wavelength set are the spectral line wavelengths that play a dominant role in coal and rock identification, wherein A l1 VIP and A l2 VIP All belong to the candidate spectral wavelength set, with l1 ranging from 1 to L and l2 ranging from 2 to L.

[0111] This embodiment can analyze and optimize the original spectrum before implementing Embodiments 4 and 5, simplifying it into a "dominant factor simplified spectrum" consisting of only a small number of spectral lines that play a dominant role in the coal and rock identification results. This reduces the amount of data and computation when calculating principal components, thereby reducing the software and hardware costs of coal and rock identification and improving the identification speed.

[0112] Partial Least Squares Discriminant Analysis (PLS-DA) is a further application of PLS ​​and a regression analysis method specifically designed for classification problems. When the dependent variable is categorical data, PLS-DA analysis can not only find the principal components obtained from linear combinations of variables, but also analyze the VIP (Variable Importance in Projection) value of the influence of each variable in the original data on the final discriminant classification. That is, in LIBS coal rock identification spectral data analysis, PLS-DA analysis can obtain the weight of each spectral line's influence on the final coal rock identification result. PLS-DA analysis and VIP value calculation can be performed using data analysis software such as SIMCA-P. As an example, step F2 includes: Step F21: Import all samples from P1 coal sample sets and Q1 rock sample sets into SIMCA-P software to establish a database. The database includes PQ1 sample information records, where PQ1 is the total number of samples in P1 coal sample sets and Q1 rock sample sets. The sample information records include the sample coal mining machine intelligent coordination control subsystem field, the sample type field, and the LIBS spectral intensity field corresponding to N spectral line wavelengths.

[0113] Step F22: Scale the sample data in the database.

[0114] Specifically, UV data scaling can be used to scale the sample data in the database. UV data scaling is a scaling method provided by the SIMCA-P software. The UV algorithm best takes care of small peaks with small peak areas. If there are many small peaks in the data, the UV algorithm can be used. Therefore, this embodiment chooses the UV data scaling method.

[0115] Step F23: Obtain the VIP value corresponding to each spectral line wavelength based on SIMCA-P software and scaled sample data.

[0116] Since the VIP value of a spectral line indicates the extent to which the spectral line contributes to the final coal and rock identification, the VIP value is used as a reference condition for screening the dominant factors. Spectral lines with large VIP values ​​are selected, while spectral lines with small VIP values ​​that are not helpful for coal and rock identification are deleted.

[0117] To further improve the accuracy and reliability of the generated target simplified spectral wavelength set, as an example, the following is included between step F2 and step F3: Step F30: Generate a test P value corresponding to each spectral wavelength based on the LIBS spectral data of all samples in the P1 coal sample set and Q1 rock sample set. The test P value is used to represent the probability that there is no difference in the LIBS spectral intensity of the coal sample and the delayed sample at the nth spectral wavelength. Spectral wavelengths with test P values ​​greater than or equal to the preset test P value threshold are removed, and then step F3 is executed.

[0118] As an example, in step F30, the Wilcoxon rank-sum test algorithm is used to obtain the test P value corresponding to each spectral line wavelength. The specific Wilcoxon rank-sum test algorithm will not be described in detail here.

[0119] The p-value is related to its within-group variance and between-group variation, and it is only related to the change of the variable itself, not to the change of the overall data. The VIP value, however, examines both the change of the variable itself and its proportion in the overall data; the greater the change and the greater the proportion, the larger the VIP value. In actual calculations, VIP places more emphasis on the proportion of the variable in the overall data, and only secondarily on the change of the substance itself. In this example, there are more than 4500 spectral wavelengths with VIP < 1, while the VIP value of the wavelength with the highest VIP ranking exceeds 6. This indicates that the original LIBS spectral data of coal and rock contains, on the one hand, a large number of spectral lines that are useless for coal and rock identification, resulting in high data redundancy and significant room for optimization; on the other hand, the original spectrum also contains spectral lines that are extremely important for coal and rock identification, with a large range of VIP values. This distribution characteristic of VIP values ​​is very conducive to the selection of dominant factors, fully achieving the expected results. Taking a VIP value of 1 and a p-value threshold of 0.05 as an example, some variables may have high VIP values, resulting in a VIP greater than 1, but their p-values ​​may still be greater than 0.05 (without significant difference). Therefore, in addition to using VIP > 1 as a limiting condition for selecting variables for coal and rock identification, we also use the condition p < 0.05 to select dominant factors.

[0120] When constructing a simplified spectrum by screening dominant factors, in addition to considering the screening conditions, the actual physical meaning of the spectral line should also be considered. For similar spectral lines belonging to the same element's reference spectral range, only the one with the highest VIP value should be selected to ensure quality. For example, although the spectral line corresponding to carbon (C) is relatively difficult to excite at low laser intensities, resulting in a lower VIP value ranking in the PLS-DA analysis of this study's sample library, considering the actual physical meaning of C as a typical element contained in coal, as an example, step F4 also includes: Step F41: Determine whether the VIP value corresponding to each preset spectral wavelength in the preset spectral wavelength set is greater than the preset VIP threshold. If it is greater, proceed to step F42. The preset spectral wavelength is the spectral wavelength corresponding to a preset element that can distinguish coal and rock in a physical or chemical sense and is difficult to excite.

[0121] The preset elements include C elements and N elements, etc. As an example, the preset VIP threshold is equal to 1.

[0122] Step F42: Determine whether the preset spectral line wavelength belongs to the candidate spectral line wavelength set. If it does not belong, add the preset spectral line wavelength to the target simplified spectral line wavelength set.

[0123] Steps F41-F42 not only expand the band range of the simplified spectrum but also describe the spectral changes between bands in more detail, thus preserving more of the useful data information for coal and rock identification in the original LIBS spectrum of the sample.

[0124] As an example, step F4 is followed by: Step F5: Input the generated coal and rock identification model based on the LIBS spectral intensity corresponding to the target simplified spectral wavelength in the sample target simplified spectral wavelength set, and train the coal and rock identification model.

[0125] Step F5 can be specifically trained to generate a PCA coal and rock identification model in the manner described in Example 4 or a PLS coal and rock identification model in the manner described in Example 5. It does not require directly obtaining the corresponding principal component extraction architecture based on the N LIBS spectral data for each sample. Instead, a target simplified spectral wavelength set is first generated from the N LIBS spectral data. Based on the LIBS spectral intensity corresponding to the target simplified spectral wavelength in the sample's target simplified spectral wavelength set, the corresponding principal component extraction architecture is obtained. Then, based on the corresponding principal component extraction architecture and the LIBS spectral intensity corresponding to the target simplified spectral wavelength in the sample's target simplified spectral wavelength set, the principal component corresponding to each sample is generated as input to the preset classification model. The coal and rock binary classification result is used as the output of the preset classification model, training to obtain a binary classification model or a W-classification model, i.e., a PCA coal and rock identification model or a PLS coal and rock identification model. Other related technical details in Examples 4 and 5 will not be repeated here. The PCA or PLS coal rock identification model built using the target simplified spectral wavelength still achieves the best identification accuracy comparable to the PCA or PLS coal rock identification model built using the original spectral wavelength while significantly reducing the computational load. The dominant factor simplified spectral model based on the VIP value can reasonably and effectively optimize the original spectrum.

[0126] Figure 7The diagram shows the effect of selecting target simplified spectral wavelengths based on VIP. Each initial sample includes 6074 LIBS data points. The element corresponding to the fourth spectral line with a VIP value is Mg, and the element corresponding to the second spectral line with a VIP value is also Mg. Therefore, the fourth spectral line is removed. Similarly, the seventh, eighth, and ninth spectral lines are also removed. Then, based on steps F41-F42, the spectral lines corresponding to element C are added. As can be seen from the table, the VIP value of element C is ranked 1176th. This is because carbon (C) spectral lines are difficult to excite at low laser intensities, resulting in a relatively low VIP value ranking for the C spectral lines in the PLS-DA analysis of this study's sample library. However, element C is a typical element contained in coal and has practical physical significance; therefore, the spectral lines corresponding to element C are also selected. Finally, seven spectral wavelengths are chosen as the target simplified wavelengths.

[0127] Example 6 obtains the weight value VIP of the influence of each spectral wavelength on the final coal and rock identification result in P1 coal sample sets and Q1 rock sample sets. Based on the VIP, a target simplified spectral wavelength set is generated. The spectral wavelengths in the target simplified spectral wavelength set are the spectral wavelengths that play a dominant role in coal and rock identification. This greatly simplifies the LIBS spectral data structure and can further reduce the overhead of coal and rock identification based on LIBS spectral data while maintaining the accuracy of coal and rock identification.

[0128] Developing comprehensive supporting equipment for high-yield and high-efficiency coal mines is the mainstream of coal technology development. High-power, deep-cutting electric traction drum coal mining machines have been widely used. The motor is one of the main power sources and actuators of the electric traction coal mining machine, mainly consisting of a cutting motor and a traction motor. The cutting motor drives the coal mining machine drum to cut coal, while the traction motor drives the entire machine to move along the traction direction of the working face. The reliability and effectiveness of the motor drive control are crucial to the actual working effect of the coal mining machine and ensure the completion of the cutting path planning by the coal and rock intelligent sensing and other information sensing systems. After sampling the coal and rock at the above embodiments and generating the planned cutting path, errors may exist, or other situations may arise during coal mining. Therefore, how to achieve intelligent coordinated control of the coal mining machine is also particularly critical. Based on this, this invention further proposes Embodiment Seven.

[0129] Example 7 Example 7 provides an intelligent coordination control subsystem for a coal mining machine, such as... Figure 8As shown, it includes a coordination control module and a motor control module. The coordination control module is used to obtain the planned cutting path and to obtain the operating parameters of the coal mining machine in real time. When the operating parameters of the coal mining machine meet the preset requirements, it generates a first cutting motor control command and a first traction motor control command based on the planned cutting path and sends the first cutting motor control command and the first traction motor control command to the motor control module.

[0130] The planned cutting path can be generated based on the coal and rock identification results of Examples 3, 4, and 5. It can also be achieved by acquiring LIBS data from sampling points through the LIBS spectral intelligent acquisition subsystem described in Example 2. Specific technical details have been described in the preceding examples and will not be repeated here. As an example, the operating parameters of the coal mining machine may specifically include current, voltage, and vibration values. It is understood that as the gangue content increases, the current, voltage, and vibration values ​​of the acquisition machine will also increase.

[0131] When the operating parameters of the coal mining machine do not meet the preset requirements, a second cutting motor control command and a second traction motor control command are generated based on the current operating parameters of the coal mining machine. The second cutting motor control command and the second traction motor control command are sent to the motor control module. The second cutting motor control command and the second traction motor control command have higher priority than the first cutting motor control command and the first traction motor control command.

[0132] The motor control module is used to control the operation of the traction motor and cutting motor of the coal mining machine based on the first cutting motor control command and the first traction motor control command when no second cutting motor control command and second traction motor control command are received, and to perform coal mining operations according to the planned cutting path. When the second cutting motor control command and second traction motor control command are received, the module controls the operation of the traction motor and cutting motor of the coal mining machine based on the second cutting motor control command and the second traction motor control command, and to perform coal mining operations based on the current operating parameters of the coal mining machine. Understandably, if the second cutting motor control command and the second traction motor control command are not received, it means that the coal mining machine can operate normally based on the planned cutting path. At this time, the coal mining operation can be performed directly according to the planned cutting path. When the second cutting motor control command and the second traction motor control command are received, it means that an abnormality has occurred. It may be that the planned cutting path is inaccurate, causing the mining machine to cut into or about to cut into the rock, or the coal mining machine itself is abnormal. At this time, it is necessary to generate the second cutting motor control command and the second traction motor control command based on the current operating parameters of the coal mining machine, so that the coal mining machine can perform the coal mining operation based on the second cutting motor control command and the second traction motor control command.

[0133] The intelligent coordination and control subsystem for the coal mining machine integrates coal and rock sensing information based on the planned cutting path and the machine's operating parameters. It coordinates the control of the cutting motor and traction motor, enabling timely deceleration or even stopping of cutting and movement when the cutting teeth are about to cut rock or have already cut rock, and restarting and adjusting speed when normal operation resumes. In case of other emergency abnormalities, remote monitoring personnel can also manually intervene through the coordination and control system.

[0134] As an example, the coordination control module is used to determine that the operating parameters of the current coal mining machine meet the preset requirements when the current current is less than or equal to a preset current threshold, the current voltage is less than or equal to a preset voltage threshold, and the current vibration value is less than or equal to a preset vibration threshold, and to generate a first cutting motor control command and a first traction motor control command based on the planned cutting path.

[0135] As an example, the coordination control module is used to determine that the operating parameters of the current coal mining machine do not meet the preset requirements if at least one of the following conditions occurs: the current current is greater than a preset current threshold, the current voltage is greater than a preset voltage threshold, or the current vibration value is greater than a preset vibration threshold. Based on at least one of the differences between the current current and the preset current threshold, the current voltage and the preset voltage threshold, or the current vibration value and the preset vibration threshold, the module generates a second cutting motor control command and a second traction motor control command. It should be noted that as the amount of gangue in the area mined by the coal mining machine increases, the current, voltage, and vibration value may all increase. Increases in current, voltage, and vibration values ​​may also occur when the coal mining machine itself malfunctions or when environmental factors such as temperature are abnormal.

[0136] The operation of the traction motor and cutting motor of the coal mining machine is controlled based on the first cutting motor control command and the first traction motor control command, or based on the second cutting motor control command and the second traction motor control command. Specifically, this includes controlling the start and stop of the traction motor and / or cutting motor; controlling the frequency of the traction motor and / or cutting motor; controlling the forward and reverse rotation of the traction motor and / or cutting motor; and controlling the speed of the traction motor and / or cutting motor. It is understood that the specific controlled parameters are determined based on the planned cutting path and the operating parameters of the coal mining machine.

[0137] As an example, if the current, voltage, and vibration values ​​all return to their corresponding normal ranges, the coordination control module generates a first cutting motor control command and a first traction motor control command based on the planned cutting path. It is understood that when the current, voltage, and vibration values ​​all return to their corresponding normal ranges, it indicates that the abnormal situation has disappeared, or that the planned cutting path with a large error has been bypassed, and therefore, travel can continue based on the planned cutting path.

[0138] To meet the needs of high production, high efficiency, and rapid cutting under different geological conditions in fully mechanized mining faces, coal mining machines used for various coal seam thicknesses are constantly increasing their installed power and power supply voltage. As an example, the motor control module is a three-level converter, which is a back-to-back NPC type three-level control system that can be used for the drive control of a 1000kW high-power asynchronous motor with a voltage level of 1140V in underground coal mines.

[0139] As an example, the three-level converter includes a core control board and an interface board, with the core control board connected to the interface board. The core control board includes a DSP module and an FPGA module. The DSP module is used to implement the main program for controlling the traction motor or cutting motor and the user interface communication. The FPGA module is used for control logic coupling and pulse signal control, and completes AD chip selection, IO control and PWM signal output based on the instructions sent by the DSP module.

[0140] As an example, the DSP module includes a system control unit, an operating status monitoring unit, and a communication unit. The system control unit implements system control via external interrupts, performing AD sampling, system protection, external I / O input / output sampling, coordinate transformation, space vector pulse width modulation (SVPWM) modulation, and proportional and integral regulation (PI regulation). The operating status monitoring unit monitors user-input control words and switches the inverter system state machine based on these control words. The communication unit receives commands from the host computer and human-machine interface to set the system control words.

[0141] Example 7 uses the above technical solution to coordinate and control the cutting motor and traction motor of the coal mining machine to perform coal mining operations based on the planned cutting path and the operating parameters of the coal mining machine, thereby improving the accuracy and reliability of the coordinated control of the coal mining machine.

[0142] It should be noted that the same or related technical features among the examples of this invention can be used in combination, and will not be described in detail here. Furthermore, the steps described in Embodiments Three, Four, Five, and Six of this invention can be performed in different orders and / or in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of this invention is not limited in this respect.

[0143] This invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being configured to perform the methods described in embodiments three, four, five, and six of this invention.

[0144] This invention also provides a computer-readable storage medium, wherein the computer instructions are used to execute the methods described in embodiments three, four, five, and six of this invention.

[0145] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

Claims

1. A coal mining machine intelligent coordination control subsystem, characterized in that, It includes a coordination control module and a motor control module, among which, The coordination and control module is used to obtain the planned cutting path and also to obtain the operating parameters of the coal mining machine in real time. When the operating parameters of the coal mining machine meet the preset requirements, it generates a first cutting motor control command and a first traction motor control command based on the planned cutting path and sends the first cutting motor control command and the first traction motor control command to the motor control module. The planned cutting path is generated based on the coal and rock identification results obtained in steps C1 to C5: Step C1: Obtain coal sample set and rock sample set. Each sample includes N LIBS spectral data. The LIBS spectral data includes spectral line wavelength and corresponding LIBS spectral intensity. Step C2: Determine the number of principal components D based on all samples in the coal and rock sample sets, and the D principal component extraction framework {z(x)}. 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D }, where z(x) d Let z(x) be the principal component extraction architecture for the d-th principal component, where d ranges from 1 to D, N >> D. 2 ,…,z(x) d ,…,z(x) D They are orthogonal to each other, z(x) 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D The sum of the explanatory power of the overall sample data is greater than the preset explanatory power threshold; Step C3, based on {z(x)} 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D Extract D principal components from each sample in the coal and rock sample sets; Step C4: Use the D principal components corresponding to each sample in the coal sample set and the rock sample set as the input of the preset classification model, and use the coal-rock binary classification result as the output of the preset classification model to train the binary classification model. Step C5: Perform coal and rock identification based on the binary classification model; Step C2 includes: Step C21: Construct the initial principal component extraction architecture: z(x) = l1x1 + l2x2 + ... + l n x n …+L N x N , where l n x is the coefficient corresponding to the wavelength of the nth spectral line. n The LIBS spectral intensity variable corresponding to the wavelength of the nth spectral line; Step C22: Input all samples from the coal sample set and the rock sample set into z(x) = l1x1 + l2x2 + ... + l n x n …+L N x N Adjust l1, l2, ... l n …、L N Generate group D l1 d l2 d ...l n d …、L N d Simultaneously satisfying: Condition 1, each group l1 d l2 d ...l n d …、L N d satisfy ; Condition 2, for each group l1 d l2 d ...l n d …、L N d z(x) satisfies the condition for all samples d The difference between the variance of the sample and the original variance of all samples in the coal and rock sample sets is less than the preset variance threshold. Condition 3, Group D, l1 d l2 d ...l n d …、L N d Satisfying z(x) 1 ,z(x) 2 ,…,z(x) d ,…,z(x) D The sum of the explanatory power of the overall sample data is greater than the preset explanatory power threshold; When the operating parameters of the coal mining machine do not meet the preset requirements, a second cutting motor control command and a second traction motor control command are generated based on the current operating parameters of the coal mining machine. The second cutting motor control command and the second traction motor control command are sent to the motor control module. The second cutting motor control command and the second traction motor control command have higher priority than the first cutting motor control command and the first traction motor control command. The motor control module is used to control the operation of the traction motor and the cutting motor of the coal mining machine based on the first cutting motor control command and the first traction motor control command when no second cutting motor control command and the second traction motor control command are received, and to perform coal mining operations according to the planned cutting path; when the second cutting motor control command and the second traction motor control command are received, the module controls the operation of the traction motor and the cutting motor of the coal mining machine based on the second cutting motor control command and the second traction motor control command, and to perform coal mining operations based on the current operating parameters of the coal mining machine.

2. The system according to claim 1, characterized in that, The operating parameters of the coal mining machine include current, voltage, and vibration values.

3. The system according to claim 2, characterized in that, The coordination control module is used to determine that the operating parameters of the current coal mining machine meet the preset requirements when the current current is less than or equal to a preset current threshold, the current voltage is less than or equal to a preset voltage threshold, and the current vibration value is less than or equal to a preset vibration threshold. Based on the planned cutting path, it generates the first cutting motor control command and the first traction motor control command.

4. The system according to claim 2, characterized in that, The coordination control module is used to determine that the operating parameters of the current coal mining machine do not meet the preset requirements if at least one of the following conditions occurs: the current current is greater than a preset current threshold, the current voltage is greater than a preset voltage threshold, or the current vibration value is greater than a preset vibration threshold. Based on at least one of the differences between the current current and the preset current threshold, the current voltage and the preset voltage threshold, or the current vibration value and the preset vibration threshold, the module generates a second cutting motor control command and a second traction motor control command.

5. The system according to claim 4, characterized in that, If the current, voltage, and vibration values ​​all return to their corresponding normal ranges, the coordination control module generates a first cutting motor control command and a first traction motor control command based on the planned cutting path.

6. The system according to claim 1, characterized in that, The motor control module is a three-level converter, which includes a core control board and an interface board, and the core control board is connected to the interface board. The core control board includes a DSP module and an FPGA module. The DSP module is used to implement the main program for controlling the traction motor or cutting motor and the communication with the user interface. The FPGA module is used for control logic coupling and pulse signal control, and completes AD chip selection, IO control and PWM signal output based on the instructions sent by the DSP module.

7. The system according to claim 6, characterized in that, The DSP module includes a system control unit, a working status monitoring unit, and a communication unit. The system control unit is used to implement system control in the DSP external interrupt. The system control is used to implement AD sampling, system protection, external I / O input / output sampling, coordinate transformation, SVPWM modulation, and PI regulation. The working status monitoring unit is used to monitor the user input control words and complete the switching of the inverter system state machine according to the relevant control words; The communication unit is used to receive instructions from the host computer and the human-machine interface to complete the setting of system control words.

8. The system according to claim 6, characterized in that, The three-level converter is a back-to-back NPC-type three-level control system.

9. The system according to claim 1, characterized in that, The planned cutting path is generated based on the coal and rock identification results of the sampling points on the target working face generated by the preset coal and rock identification model.

Images