Intelligent dust removal system and dust removal method for mechanical rock breaking

Abstract

The invention relates to the technical field of excavator dust removal, in particular to an intelligent dust removal system and method for mechanical rock breaking, laser beams are emitted to an operation area where an excavator execution component works through a laser module, and when flying dust appears in the operation area, a visual sensor can observe the reflected laser beams; the dust raising range and the dust raising concentration are determined according to the length and the brightness of the laser beams, so that a spraying parameter combination of the spraying module is generated, the spraying parameter combination is sent to the spraying module, the spraying module sprays water mist towards the rock breaking working face generating dust raising, and dust falling is achieved through combination of the water mist and dust raising particles. The whole working process does not need manual intervention, the safety of operators is improved while the good dust falling effect is achieved, meanwhile, manual operation is not needed, and the labor cost is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of excavator dust removal, in particular to a mechanical rock breaking intelligent dust removal system and a dust removal method. BACKGROUND

[0002] In the mechanical excavation process in the fields of hydropower engineering, mine engineering and traffic road engineering, more and more mechanical rock breaking methods are used, which have significant advantages in ensuring the excavation efficiency and reducing the disturbance to the surrounding rock compared with the traditional drilling and blasting method. Mechanical rock breaking mainly adopts hydraulic support breaking, or uses an arm of an excavator equipped with a pangolin hook or an excavator equipped with a breaking hammer to break rocks for tunnel or surrounding rock construction.

[0003] However, a large amount of dust is generated during the process of mechanical impact or excavation of rocks. In order to enable the driver to see the working face, the current measure is to have a worker stand near the working face to use a water pipe for dust removal. However, this method has two significant disadvantages: on the one hand, it increases labor costs; on the other hand, it is difficult to avoid flying stones during the process of mechanical rock breaking, which can easily cause personal injury. SUMMARY

[0004] The main purpose of the present application is to provide a mechanical rock breaking intelligent dust removal system and a dust removal method to solve the technical problems of poor safety and high labor cost of manual dust removal of the working face in the prior art.

[0005] To achieve the above purpose, the present application provides a mechanical rock breaking intelligent dust removal system applied to an excavator, which comprises a laser module, a sensor module, a control module and a spraying module; wherein; The laser module is fixed on the execution component of the excavator, and the laser beam emitted by the laser module is directed to the rock breaking working face; The sensor module is arranged on the excavator, and the sensor module comprises a visual sensor, the collection direction of the visual sensor is arranged at an angle with the laser beam; The spraying module is arranged on both sides of the execution component, and the spraying direction of the spraying module is directed to the rock breaking working face; The control module is connected with the laser module, the sensor module and the spraying module respectively.

[0006] Further, the laser module comprises a laser emitter and a detachable connecting seat, the laser emitter is detachably connected with the detachable connecting seat, and the detachable connecting seat is installed on the execution component of the excavator.

[0007] Further preferably, the detachable connecting seat comprises a seat body, an operating member, a magnet, a buffer unit and a sliding frame, the magnet is movably installed in the seat body, the operating member is installed on the seat body, the operating member is connected with the magnet to control the movement of the magnet in the seat body to realize suction locking or unlocking, the sliding frame is in sliding connection with the seat body, and the buffer unit is connected between the sliding frame and the seat body; and the laser emitter is installed on the sliding frame.

[0008] Further preferably, the buffer unit comprises an elastic member, a piston, a rod body and a control valve, the seat body is provided with a sliding groove, and the sliding frame is in sliding connection in the sliding groove; The seat body is provided with a damping cavity, the piston is in sliding connection in the damping cavity, a first end of the rod body is fixedly connected with the piston, and a second end of the rod body is fixedly connected with the sliding frame after penetrating out of the damping cavity; The piston divides the damping cavity into a rod cavity and a rod-free cavity, the seat body is further provided with a first channel and a second channel, the first channel is in communication with the rod-free cavity and the outside, and the second channel is in communication with the rod cavity and the outside; The control valve is in sliding connection with the seat body, the control valve intersects with the first channel and the second channel at the same time to control the sizes of the first channel and the second channel; The elastic member is arranged on a side of the sliding frame away from the rod body, one end of the elastic member is in abutment with the seat body, and the other end of the elastic member is in abutment with the sliding frame.

[0009] Further preferably, the control valve comprises a valve rod and a valve cap, the seat body is provided with a valve hole, the valve hole intersects with the first channel and the second channel at the same time, the valve cap is coaxially and fixedly connected at one end of the valve rod, and the valve rod is in threaded connection with the valve hole; the valve rod comprises a first taper surface and a second taper surface, the first taper surface is arranged at a position corresponding to the first channel to adjust the size of the flow passage cross-sectional area of the first channel, and the second taper surface is arranged at a position corresponding to the second channel to adjust the size of the flow passage cross-sectional area of the second channel.

[0010] The application further provides a mechanical rock breaking intelligent dust removal method applied to the mechanical rock breaking intelligent dust removal system. S1, collecting dust raising characteristic data, environmental parameters and operation intensity data in a mechanical rock breaking operation process, pre-processing the collected data to obtain a real-time data matrix; wherein the dust raising characteristic data comprises dust raising concentration, dust raising range and dust raising average particle size, the environmental parameters comprise environmental wind speed, environmental humidity and operation surface humidity, and the operation intensity data comprises rock breaking output frequency and hydraulic system pressure; S2, obtain a calculation weight matching model, identify a current construction scene, and obtain a calculation weight matrix corresponding to the current construction scene according to the calculation weight matching model; wherein the calculation weight matching model comprises a mapping relationship between a construction scene and a calculation weight of each parameter in a real-time data matrix; S3, obtain a spraying parameter decision model, input the real-time data matrix and the calculation weight matrix into the spraying parameter decision model, and obtain a spraying parameter combination; wherein the spraying parameters include a spraying flow, a droplet size, a spraying angle, and a spraying frequency; S4, control operation of the spraying module according to the spraying parameter combination, and collect a real-time dust concentration after waiting for a preset time length; S5, determine whether the real-time dust concentration is less than or equal to a preset dust concentration threshold; if yes, maintain the operation of the spraying module and return to step S4; if no, obtain a preset correction amount of the spraying parameter combination, correct the spraying parameter combination according to the preset correction amount, and then return to step S4.

[0011] Further preferably, step S2 specifically comprises the following steps: obtain a scene classifier, collect core identification features of a current construction scene, and identify a category of the current construction scene through the scene classifier according to the core identification features; wherein the scene classifier classifies a mechanical rock breaking construction scene into a tunnel scene, an open-air scene, and a foundation pit scene, and the scene classifier takes three types of parameters, including wind speed stability, work sealing, and work surface slope, as core identification features for scene classification; obtain a calculation weight matching model, and obtain a calculation weight matrix corresponding to the current construction scene according to the category of the current construction scene; wherein the calculation weight matrix is a preset initial weight matrix, and the setting of initial weights in the initial weight matrix is based on the influence degree of dust characteristic data, environmental parameters, and work intensity data on the spraying effect in different scenes.

[0012] Further preferably, step S5 further comprises the following steps: obtain a real-time data matrix after spraying, determine a unit time change rate of each parameter in the real-time data matrix, and determine whether the unit time change rate of each parameter is greater than a first preset threshold; If the unit time change rate is greater than the first preset threshold, it is determined that the real-time data is mutation data, and it is determined whether there is any or more than one mutation data in the real-time data matrix after spraying. If there is any or more than one mutation data in the real-time data matrix after spraying, the initial weight corresponding to the mutation data in the initial weight matrix is increased by a preset correction step, and the process returns to step S3.

[0013] Further preferably, step S5 further comprises the following steps: obtaining the real-time data matrix after spraying, determining the deviation of each parameter in the real-time data matrix from the average value of each parameter under the same construction scene, and judging whether the deviation is greater than a second preset threshold value; If the deviation is greater than the second preset threshold value, the real-time data is determined as deviation abnormal data, and it is judged whether there is any one or more than one deviation abnormal data in the real-time data matrix after spraying; If there is any one or more than one deviation abnormal data in the real-time data matrix after spraying, an initial weight matrix of the open-air scene is obtained, an initial weight ratio of the open-air scene is obtained according to the initial weight matrix of the open-air scene, the initial weight of the initial weight matrix of the current construction scene is corrected according to the initial weight ratio with a preset correction step, and the step S3 is returned.

[0014] Further preferably, step S3 specifically comprises the following steps: obtaining a spraying parameter decision model, the spraying parameter decision model comprising a random forest model and a feature attention layer, the feature attention layer being embedded between an input layer and a decision tree training layer of the random forest model; inputting the real-time data matrix and the calculated weight matrix to the input layer, the feature attention layer matching a weight coefficient for each parameter of the real-time data matrix according to the calculated weight matrix to obtain a weighted feature vector; inputting the feature vector to each decision tree of the random forest model for independent training, each decision tree outputting a respective parameter prediction result, and the parameter prediction results of all the decision trees being fused by weighted voting to obtain a final spraying parameter combination.

[0015] Compared with the prior art, the present application has the following beneficial effects: The present application can determine the dust raising range and dust raising concentration according to the length and brightness of the laser beam when the visual sensor can observe the reflected laser beam when dust raising occurs in the working area of the excavator implement component, so as to generate a spraying parameter combination of the spraying module. By sending the spraying parameter combination to the spraying module, the spraying module sprays water mist towards the rock breaking working surface where dust raising occurs, and the water mist and dust particles combine to achieve dust setting. When the laser beam is longer and brighter, the spraying module sprays more dust setting water, so that the dust can be quickly settled. When the laser beam is shorter and darker, the spraying module sprays less dust setting water, so as to achieve dust setting effect while saving water and energy. The whole working process does not need manual intervention, improves the safety of the operator while achieving good dust setting effect, and does not need manual operation, thereby reducing labor cost.

[0016] The dust removal method provided by the application collects multi-dimensional data related to dust raising in the operation, provides comprehensive consideration of multi-dimensional parameter changes for subsequent decision-making, uses a calculation weight matching model with a calculation weight mapping relationship of scenes and parameters to adapt the weight distribution to the current construction scene and improve the decision-making pertinence, uses a spraying parameter decision-making model to fuse a real-time data matrix and a calculation weight matrix to output an adaptive parameter combination, realizes accurate parameter matching to the working condition, and dynamically adjusts the spraying parameter combination according to the real-time dust concentration after spraying and a preset correction amount to ensure the dynamic adaptability of dust removal control. The application integrates scene-based weight matching and dynamic parameter correction into a spraying decision-making process driven by multi-dimensional data, realizes intelligent adaptive dust removal control for different working conditions of mechanical rock breaking, and improves the dust removal effect and control accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from the structures shown in these drawings without creative labor.

[0018] Figure 1 The first perspective view of the overall structure of an embodiment of the present application installed on an excavator; Figure 2 The second perspective view of the overall structure of an embodiment of the present application installed on an excavator; Figure 3 The local enlarged view of the laser module in an embodiment of the present application installed on the excavator; Figure 4 The internal structure of the magnetic connecting seat in an embodiment of the present application; Figure 5 The enlarged view of A in FIG. 6; Figure 4 Figure 6 The overall structure of the magnetic connecting seat in an embodiment of the present application; Figure 7 The flowchart of the mechanical rock breaking intelligent dust removal method in an embodiment of the present application.

[0019] ​1, laser module; 11, laser emitter; 12, magnetic connecting seat; 121, seat body; 1211, sliding groove; 1212, damping cavity; 1213, first channel; 1214, second channel; 122, operating piece; 123, magnet; 124, buffer unit; 1241, elastic piece; 1242, piston; 1243, rod body; 1244, control valve; 1245, valve rod; 1246, valve cap; 1247, first taper surface; 1248, second taper surface; 125, sliding frame; 126, guide rod; 2, visual sensor; 3, spraying module; 31, pumping unit; 311, water tank; 312, water pump; 32, water delivery pipe; 321, fixed hard pipe; 322, connecting hose; 323, flexible positioning pipe; 33, spray head; 34, electromagnetic valve.

[0020] The purposes, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein merely exemplify the present application and are not intended to limit the present application.

[0022] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of protection of the present application.

[0023] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative positional relationship, movement condition, etc. between components in a certain posture (as described in the drawings), and if the certain posture changes, the directional indications also change accordingly.

[0024] In addition, the descriptions of “first”, “second” and the like in the present application are only for descriptive purposes, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implying the number of the indicated technical features. Therefore, the features limited by “first”, “second” can explicitly or implicitly include at least one of the features. In addition, the technical solutions of various embodiments can be combined with each other, but it must be based on the realization of those skilled in the art, and when the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, and is not within the scope of protection claimed by the present application.

[0025] When a tunnel working face is constructed by a breaking hammer mounted on an excavator, dust will be generated at the position of the breaking hammer, which will hinder the working view of the operator of the excavator, so that the operator cannot clearly see the working face, and thus the construction of the tunnel working face by the operator is affected. It is found through research that if laser is used to irradiate the dust, when the laser beam passes through the dust, diffuse reflection will occur at the position of the dust, and thus the diffuse reflection of the laser beam in segments can be clearly observed. It should be noted that when there is no dust in the air, no diffuse reflection will occur in the air through laser irradiation, and thus the laser beam cannot be observed. The range of the dust can be determined by observing the length of the laser beam, the longer the observed laser beam, the greater the range of the dust, and vice versa; the concentration of the dust can be determined by observing the brightness of the laser beam, the brighter the laser beam, the more light reflected by the diffuse reflection, the more dust particles contained in the air, that is, the higher the concentration of the dust. Based on this finding, the inventors have designed the following mechanical rock breaking intelligent dust removal system.

[0026] Referring to Figures 1 to 6 The embodiment provides a mechanical rock breaking intelligent dust removal system applied to an excavator, which comprises a laser module 1, a sensor module, a control module and a spraying module 3; wherein The laser module 1 is fixed on an executing component of the excavator, and the laser beam emitted by the laser module 1 is directed to a rock breaking working face; The sensor module is arranged on the excavator, and the sensor module comprises a visual sensor 2, the collection direction of the visual sensor 2 is arranged at an angle with the laser beam; The spraying module 3 is arranged on both sides of the executing component, and the spraying direction of the spraying module 3 is directed to the rock breaking working face; The control module is connected with the laser module 1, the sensor module and the spraying module 3 respectively.

[0027] The embodiment emits a laser beam towards a working area of the executing component of the excavator through the laser module 1, when dust appears in the working area, the visual sensor 2 can observe the reflected laser beam, the range and concentration of the dust are determined according to the length and brightness of the laser beam, and thus a spraying parameter combination of the spraying module 3 is generated, the spraying parameter combination is sent to the spraying module 3, the spraying module 3 sprays water mist towards the rock breaking working face where the dust is generated, and the water mist and the dust particles are combined to achieve dust fall. When the laser beam is longer and the laser beam is brighter, the spraying module 3 sprays more dust fall water, so that the dust can be quickly settled; when the laser beam is shorter and the laser beam is dimmer, the spraying module 3 sprays less dust fall water, so that the dust fall effect is achieved while water and energy are saved. The whole working process does not need manual intervention, the safety of the operator is improved while a good dust fall effect is achieved, and manual operation is not needed, so that the labor cost is reduced.

[0028] It is worth noting that the direction of the laser beam is at an angle with the direction of the visual sensor 2, and the angle is controlled between 45 degrees and 90 degrees (when the angle is 90 degrees, the length of the laser beam photographed by the visual sensor 2 is equal to the actual length of the laser beam; the smaller the angle, the smaller the length of the laser beam photographed by the visual sensor 2 relative to the actual length of the laser beam). The spraying module 3 adopts an intermittent spraying mode. It can be known that through intermittent spraying, on the one hand, the operator's line of sight can be as little affected as possible during spraying, and on the other hand, continuous spraying can prevent the laser beam from producing diffuse reflection of water, thereby causing the spraying module 3 to spray incorrectly.

[0029] It is found in use that the laser module 1 is a relatively precise part, and the excavator's execution component will produce strong vibration in work, which can easily damage the laser module 1; at the same time, when the excavator is working, the execution component generally adopts a breaking hammer for working on hard rock, and uses a grappling hook or a pangolin hook arm for working on a relatively soft soil and rock layer; therefore, the spraying module 3 and the laser module 1 connected with the execution component need to be frequently disassembled and assembled; based on this, the present embodiment is further improved as follows.

[0030] Further in the present embodiment, the laser module 1 comprises a laser emitter 11 and a detachable connecting seat, the laser emitter 11 is detachably connected with the detachable connecting seat, and the detachable connecting seat is installed on the execution component of the excavator. Specifically, the detachable connecting seat in the present embodiment is a magnetic connecting seat 12, and the laser emitter 11 is an infrared laser emitter, which has higher recognition and is beneficial to image collection by the visual sensor 2.

[0031] It can be known that the laser emitter 11 is used for emitting a laser beam, in the present embodiment, the laser beam is a fixed infrared laser beam, which irradiates the rock breaking working surface, and when dust is generated at the working position, a bright laser beam will appear at the dust position. The magnetic connecting seat 12 has magnetism, the execution component of the excavator is made of steel material, the laser emitter 11 is quickly and detachably connected to the execution component of the excavator through the magnetic connecting seat 12, in the present application, the execution component of the excavator is taken as an example of a breaking hammer, and the laser emitter 11 is quickly and detachably connected to the side wall of the breaking hammer through the magnetic connecting member; thereby realizing quick installation and disassembly of the laser emitter 11. When the breaking hammer needs to be replaced by a pangolin hook arm, the laser emitter 11 can be conveniently disassembled and assembled.

[0032] Further preferably, the detachable connecting seat comprises a seat body 121, an operating member 122, a magnet 123, a buffer unit 124 and a sliding frame 125, the magnet 123 is movably installed in the seat body 121, the operating member 122 is installed on the seat body 121, the operating member 122 is connected with the magnet 123 to control the movement of the magnet 123 in the seat body 121 to achieve locking or unlocking, the sliding frame 125 is in sliding connection with the seat body 121, the buffer unit 124 is connected between the sliding frame 125 and the seat body 121, and the laser emitter 11 is installed on the sliding frame 125. Specifically, the operating member 122 is a knob or a handle, by rotating the knob or the handle, the magnet 123 is driven to move in the seat body 121, which is a conventional technology and is not shown in detail in the drawings, so that the magnet 123 in the seat body 121 moves towards the excavator's executing member to be attracted, thereby being locked on the executing member of the excavator by the magnet 123, when unlocking is needed, the knob or the handle is rotated in the opposite direction to move the magnet 123 away from the executing member, thereby achieving unlocking.

[0033] It can be known that, in use, the executing member of the excavator vibrates, the seat body 121 is fixedly connected to the executing member, so that the seat body 121 vibrates with the executing member, by the arrangement of the buffer assembly, the effect of shock absorption and buffering is achieved, so that less vibration is transmitted to the sliding frame 125 and the laser emitter 11 installed on the sliding frame 125, thereby preventing the laser emitter 11 from being damaged by strong vibration, and achieving the purpose of improving the service life of the laser emitter 11.

[0034] Further preferably in the embodiment, the buffer unit 124 comprises an elastic member 1241, a piston 1242, a rod body 1243 and a control valve 1244, the seat body 121 is provided with a sliding groove 1211, and the sliding frame 125 is in sliding connection in the sliding groove 1211. The seat body 121 is provided with a damping cavity 1212, the piston 1242 is in sliding connection in the damping cavity 1212, the first end of the rod body 1243 is fixedly connected with the piston 1242, and the second end of the rod body 1243 is fixedly connected with the sliding frame 125 after penetrating out of the damping cavity 1212. The piston 1242 divides the damping cavity 1212 into a rod cavity and a rod-free cavity, the seat body 121 is further provided with a first channel 1213 and a second channel 1214, the first channel 1213 communicates the rod-free cavity with the outside, and the second channel 1214 communicates the rod cavity with the outside. The control valve 1244 is in sliding connection with the seat body 121, and the control valve 1244 intersects with the first channel 1213 and the second channel 1214 at the same time, so as to control the size of the first channel 1213 and the second channel 1214. The elastic member 1241 is arranged on the side of the sliding frame 125 away from the rod body 1243, one end of the elastic member 1241 abuts against the seat body 121, and the other end of the elastic member 1241 abuts against the sliding frame 125. Specifically, the elastic member 1241 is a compression coil spring, and in some other embodiments of the present application, the elastic member 1241 can also be a gas spring, a leaf spring, or other components or structures capable of accumulating elastic potential energy.

[0035] It can be known that, in the process of executing the component, the seat body 121 vibrates with the execution component, that is, the seat body 121 reciprocates in a direction, and in the process of reciprocating of the seat body 121, the sliding frame 125 and the laser emitter 11 remain stationary or slightly move relative to the ground under the action of inertia, that is, in the vibration process, the seat body 121 reciprocates relative to the sliding frame 125; in the process of the seat body 121 moving relative to the sliding frame 125, because the piston 1242 and the rod body 1243 are fixedly connected to the sliding frame 125, when the seat body 121 moves relative to the sliding frame 125, the piston 1242 reciprocates in the damping cavity 1212, and when the piston 1242 moves in the damping cavity 1212, the external gas is compressed or drawn in, and when the external gas enters the rodless cavity from the first channel 1213 or the gas in the rodless cavity is discharged to the outside from the first channel 1213, energy is consumed due to the damping effect of air, that is, the vibration energy of the seat body 121 is prevented from being transmitted to the sliding frame 125; at the same time, when the external gas enters the rod cavity from the second channel 1214 or the gas in the rod cavity is discharged to the outside from the second channel 1214, energy is also consumed, further preventing the vibration energy of the seat body 121 from being transmitted to the sliding frame 125, and playing a role of vibration isolation.

[0036] As a further preferred, the control valve 1244 comprises a valve rod 1245 and a valve cap 1246, the seat body 121 is provided with a valve hole, the valve hole intersects the first channel 1213 and the second channel 1214 at the same time, the valve cap 1246 is coaxially fixedly connected to one end of the valve rod 1245, and the valve rod 1245 is threadedly connected with the valve hole; the valve rod 1245 comprises a first taper surface 1247 and a second taper surface 1248, the first taper surface 1247 is arranged at a position corresponding to the first channel 1213 of the valve rod 1245 to adjust the size of the flow passage cross-sectional area of the first channel 1213, and the second taper surface 1248 is arranged at a position corresponding to the second channel 1214 of the valve rod 1245 to adjust the size of the flow passage cross-sectional area of the second channel 1214.

[0037] It can be known that the valve cap 1246 is rotated to drive the valve stem 1245 to move along its own axis direction, so that the first conical surface 1247 is close to or away from the first channel 1213. When the first conical surface 1247 is close to the first channel 1213, the first conical surface 1247 makes the flow area of the first channel 1213 smaller, so that the gas is difficult to pass through the first channel 1213, and the vibration is easily transmitted to the sliding frame 125; when the first conical surface 1247 is away from the first channel 1213, the flow area of the first channel 1213 is larger, so that the gas is easy to pass through the first channel 1213, and the vibration is not easy to be transmitted to the sliding frame 125, but it is easy to cause the sliding frame 125 to move to the end of the sliding groove 1211 and collide with the inner wall of the seat body 121, so it is necessary to adjust the valve and the valve stem 1245, so that the valve stem 1245 is in the appropriate position, so that the buffer unit 124 has better buffering performance, and the sliding frame 125 does not collide with the inside of the seat body 121. The first conical surface 1247 and the second conical surface 1248 are arranged on the valve stem 1245, and the first channel 1213 and the second channel 1214 can be adjusted by adjusting the valve stem 1245.

[0038] Further, the magnetic connecting seat 12 further comprises a guide rod 126, both ends of the guide rod 126 are fixedly connected to the seat body 121, and the guide rod 126 penetrates through the sliding frame 125 and is arranged, and the sliding frame 125 and the guide rod 126 are in sliding fit arrangement. Through the arrangement of the guide rod 126, the sliding frame 125 is guided by the guide rod 126, so that the sliding frame 125 moves along the axis direction of the guide rod 126 when moving relative to the seat body 121, so as to improve the stability of the sliding frame 125, and the buffer unit 124 can play a better buffering effect.

[0039] In one embodiment, the spraying module 3 comprises a pumping unit 31, a water delivery pipe 32, a spray head 33 and a solenoid valve 34. One end of the water delivery pipe 32 is led out from the pumping unit 31, the second end of the water delivery pipe 32 is led out to the excavator operating component, the spray head 33 is installed at the second end of the water delivery pipe 32, and the solenoid valve 34 is installed at one end of the water delivery pipe 32 close to the spray head 33. The water delivery pipe 32 is installed on the operating component of the excavator through the magnetic connecting seat 12, the solenoid valve 34 is electrically connected with the visual sensor 2, and the solenoid valve 34 is used for receiving and responding to the electrical signal of the visual sensor 2.

[0040] It can be known that the pumping unit 31 pumps water outward, the water pumped by the pumping unit 31 is transported to the spray head 33 through the water delivery pipe 32, the electromagnetic valve 34 controls the water delivery of the water delivery pipe 32, and then controls the water spraying time and the water spraying time length of the spray head 33. Specifically, when the visual sensor 2 detects the laser beam, an opening instruction is sent to the electromagnetic valve 34, and at the same time the electromagnetic valve 34 is opened, water is sprayed from the spray head 33 to form a water mist, and dust falling in the working area is realized. At the same time, through the arrangement of the magnetic connection seat 12, the vibration of the water delivery pipe 32 and the electromagnetic valve 34 is reduced, so as to improve the service life of the water delivery pipe 32 and the electromagnetic valve 34.

[0041] Further, the pumping unit 31 comprises a water tank 311 and a water pump 312, the input end of the water pump 312 extends into the water tank 311, and the output end of the water pump 312 is communicated with the water delivery pipe 32; the water delivery pipe 32 comprises a fixed hard pipe 321, a connecting hose 322 and a flexible positioning pipe 323, the flexible positioning pipe 323 is arranged at the downstream end, and the tail end of the flexible positioning pipe 323 is connected with the spray head 33, the fixed hard pipe 321 is used for fixedly connecting on the excavator, the upstream end of the fixed hard pipe 321 is communicated with the water pump 312 through the connecting hose 322, and the downstream end of the fixed hard pipe 321 is communicated with the flexible positioning pipe 323 through the connecting hose 322. Specifically, the fixed hard pipe 321 is fixedly connected on the excavating arm of the excavator through a pipe clamp, and the fixed hard pipe 321 is arranged in two, one of which is fixedly connected on the large arm of the excavator through a pipe clamp, and the other is fixedly connected on the small arm of the excavator through a pipe clamp; the two fixed hard pipes 321 are connected through the connecting hose 322.

[0042] It can be known that the water tank 311 is used for storing water, the water in the water tank 311 is pumped out by the water pump 312, the water is delivered to the fixed hard pipe 321 through the connecting hose 322, and then delivered to the flexible positioning pipe 323, and finally sprayed out from the spray head 33 at the downstream end of the flexible positioning pipe 323, so as to realize dust falling. It should be noted that the flexible positioning pipe 323 is a goose neck pipe which can be plastically deformed or a bamboo joint pipe which can change direction and position, and a goose neck pipe is adopted in the application, the position of the downstream end of the goose neck pipe is adjusted, so that the spray head 33 sprays the working area of the execution component to realize dust falling.

[0043] In one embodiment, the spray head 33 preferably adopts a piezoelectric ceramic driven spray head, the droplet size is 5-50 μm; the spray head 33 is also connected with a micro rudder to control the pitch and yaw of the spray head 33; the water pump 312 adopts a 12V direct current frequency conversion water pump, and the frequency conversion pressure is 1.0 MPa.

[0044] Further in the embodiment, the spraying module 3 is arranged in two groups, and the two groups of spraying module 3 are arranged on two sides of the excavator implement respectively. By arranging the two groups of spraying module 3 on the two sides of the implement respectively, and spraying the working area from the two sides simultaneously by the two groups of spraying module 3, the spraying area is increased, and the dust can settle more quickly.

[0045] In one embodiment, the sensor module further comprises a wind speed sensor, a humidity sensor and a laser particle size analyzer, the wind speed sensor, the humidity sensor and the laser particle size analyzer are connected with the control module respectively, the wind speed sensor is used for measuring the environmental wind speed, the humidity sensor is used for detecting the working surface humidity and the environmental humidity, and the laser particle size analyzer is used for detecting the dust particle size.

[0046] Please refer to Figure 7 The embodiment also provides a mechanical rock breaking intelligent dust removal method, which is applied to the mechanical rock breaking intelligent dust removal system and comprises the following steps: S1, collecting dust characteristic data, environmental parameters and working intensity data in a mechanical rock breaking operation process, pre-processing the collected data to obtain a real-time data matrix; wherein the dust characteristic data comprises dust concentration, dust range and dust average particle size, the environmental parameters comprise environmental wind speed, environmental humidity and working surface humidity, and the working intensity data comprises rock breaking output frequency and hydraulic system pressure; S2, obtaining a calculation weight matching model, identifying a current construction scene, and obtaining a calculation weight matrix corresponding to the current construction scene according to the calculation weight matching model; wherein the calculation weight matching model contains a mapping relationship between the construction scene and the calculation weights of each parameter in the real-time data matrix; S3, obtaining a spraying parameter decision model, inputting the real-time data matrix and the calculation weight matrix into the spraying parameter decision model, and obtaining a spraying parameter combination; wherein the spraying parameters comprise spraying flow, droplet particle size, spraying angle and spraying frequency; S4, controlling the operation of the spraying module 3 according to the spraying parameter combination, and collecting real-time dust concentration after waiting for a preset time length; S5, judging whether the real-time dust concentration is less than or equal to a preset dust concentration threshold value; if yes, maintaining the operation of the spraying module 3, and returning to step S4; if no, obtaining a preset correction amount of the spraying parameter combination, correcting the spraying parameter combination according to the preset correction amount, and then returning to step S4.

[0047] The dust removal method provided by the embodiment collects multi-dimensional data related to dust raising in the operation, provides comprehensive consideration of multi-dimensional parameter changes for subsequent decision-making, uses a calculation weight matching model with a scene and parameter calculation weight mapping relationship to adapt the weight distribution to the current construction scene and improve the decision-making pertinence, uses a spraying parameter decision-making model to fuse a real-time data matrix and a calculation weight matrix to output an adaptive parameter combination, realizes accurate parameter matching to the working condition, and dynamically adjusts the spraying parameter combination according to the real-time dust concentration after spraying and a preset correction amount to ensure the dynamic adaptability of dust removal control. The scene weight matching and dynamic parameter correction are integrated into the spraying decision-making process driven by multi-dimensional data to realize intelligent adaptive dust removal control for different working conditions of mechanical rock breaking, and improve the dust removal effect and control accuracy.

[0048] Specifically, the dust concentration and dust range in the dust characteristic data in step S1 are collected by a laser emitting assembly and a visual sensor 2. The laser emitting assembly emits a laser beam to pass through the dust area to generate diffuse reflection. The visual sensor 2 determines the dust concentration and dust range by identifying the brightness and length of the laser beam in the diffuse reflection. The collection frequency is preferably 10 Hz. The dust particle size is collected by a laser particle size analyzer. The laser particle size analyzer reverses the particle size distribution of the dust particles based on the Mie scattering principle, thereby dividing the dust particle size into three categories: fine particles (<10 μm), medium particles (10-50 μm), and coarse particles (>50 μm). The collection frequency is 5 Hz, and the particle size detection accuracy is ±0.1 μm.

[0049] In the embodiment, the environmental wind speed in the environmental parameters is measured by a miniature ultrasonic wind speed sensor. The measurement range is 0-10 m / s, and the measurement accuracy is ±0.1 m / s. The miniature ultrasonic wind speed sensor is installed on the top of the excavator cab and faces the working area. The environmental humidity and the working surface humidity are measured by a double-path humidity sensor. One path of the double-path humidity sensor is installed on the top of the cab to collect the environmental humidity. The measurement range is 0-100% RH, and the accuracy is ±2% RH. The other path is installed near the execution component by a magnetic base to collect the working surface humidity to avoid the interference of the environmental humidity. The collection frequency is 5 Hz.

[0050] The operation intensity data is collected by the CAN bus and the excavator control system. The breaking hammer striking frequency (i.e., the rock breaking output frequency) is collected in real time. The measurement range is 500-1500 times / min, and the hydraulic system pressure is measured. The measurement range is 0-35 MPa, thereby indirectly representing the operation intensity. The higher the rock breaking output frequency and the greater the hydraulic pressure, the greater the dust generation amount. The collection frequency is 10 Hz.

[0051] In this embodiment, a sliding window filtering algorithm is used to denoise all collected data, and the window size is set to 50 ms to eliminate sensor data fluctuations caused by construction vibration. At the same time, the data is standardized and then mapped to the 0-1 interval to avoid the dimensional differences of different dimensions affecting the subsequent algorithm analysis. It should be understood that the sliding window filtering algorithm in this embodiment is prior art, which will not be described in detail here.

[0052] In one embodiment, as a further preferred, step S2 specifically comprises the following steps: obtaining a scene classifier, collecting core identification features of the current construction scene, and identifying the category of the current construction scene through the scene classifier according to the core identification features; wherein the scene classifier classifies the mechanical rock breaking construction scene into a tunnel scene, an open-air scene, and a foundation pit scene, and the scene classifier takes wind speed stability, work sealing, and work surface slope as the core identification features of scene classification; obtaining a calculation weight matching model, and obtaining a calculation weight matrix corresponding to the current construction scene according to the category of the current construction scene; wherein the calculation weight matrix is a preset initial weight matrix, and the setting of the initial weight in the initial weight matrix is based on the influence degree of dust characteristic data, environmental parameters, and work intensity data on the spraying effect in different scenes.

[0053] Specifically, the data of wind speed stability are as follows: tunnel scene: wind speed < 1 m / s and fluctuation ≤ 0.2 m / s; open-air scene: wind speed > 3 m / s and fluctuation ≥ 0.5 m / s; foundation pit scene: wind speed 1-3 m / s and fluctuation 0.2-0.5 m / s. The visual sensor 2 collects images of the work surface environment, identifies the proportion of the closed area through image segmentation, and the data of work sealing are as follows: tunnel scene > 80%, open-air scene < 20%, and foundation pit scene 20-80%. The work surface slope is obtained through the attitude sensor of the excavator, and the work surface slope in the foundation pit scene is > 10°, and the work surface slope in the tunnel scene or open-air scene is ≤ 10°.

[0054] The initial weight matrix of each type of construction scene is as follows: tunnel scene: dust concentration 30%, dust particle size 20%, work intensity 20%, work surface humidity 15%, wind speed 10%, and environmental humidity 5%; open-air scene: wind speed 35%, dust concentration 20%, dust particle size 20%, work intensity 10%, environmental humidity 10%, and work surface humidity 5%; foundation pit scene: dust particle size 25%, work surface humidity 20%, dust concentration 15%, work intensity 15%, wind speed 15%, and environmental humidity 10%.

[0055] The embodiment extracts three core identification features of wind speed stability, operation closure and operation surface slope through a scene classifier, realizes accurate identification of the construction scene category, and provides a scene basis for weight distribution; with the aid of a calculation weight matching model, the preset initial weight matrix is called according to the influence degree of each parameter on the spraying effect under different scenes, so that the weight distribution is accurately adapted to the scene characteristics. The application clearly defines the core feature dimension of scene classification, presets the initial weight according to the influence degree of the parameter on the spraying effect, realizes the accurate association of scene identification and weight distribution, improves the pertinence and rationality of feature weight distribution, and lays a foundation for accurate decision of subsequent spraying parameter combination.

[0056] In one embodiment, as a further preferred, step S5 further comprises the following steps: The real-time data matrix after spraying is obtained, the unit time change rate of each parameter in the real-time data matrix is determined, and it is judged whether the unit time change rate of each parameter is greater than a first preset threshold; in the embodiment, the first preset threshold is 30%; If the unit time change rate is greater than the first preset threshold, it is determined that the real-time data is mutation data, and it is judged whether there is any or more than one mutation data in the real-time data matrix after spraying; If there is any or more than one mutation data in the real-time data matrix after spraying, the initial weight corresponding to the mutation data in the initial weight matrix is increased by a preset correction step, and step S3 is returned, and the preset correction step is 2%.

[0057] By real-time monitoring of the dynamic change of data after spraying, the working condition mutation is identified, the corresponding parameter weight is adjusted, the feature weight distribution is more suitable for real-time working condition fluctuation, more accurate weight support is provided for subsequent spraying parameter decision, a weight dynamic optimization mechanism based on feedback of data change after spraying is constructed, and the adaptability and response accuracy of spraying parameter decision to working condition mutation are further improved.

[0058] As a further preferred, step S5 further comprises the following steps: The real-time data matrix after spraying is obtained, the deviation of each parameter in the real-time data matrix from the average value of each parameter under the same construction scene is determined, and it is judged whether the deviation is greater than a second preset threshold; in the embodiment, the second preset threshold is 50%; If the deviation is greater than the second preset threshold, it is determined that the real-time data is deviation abnormal data, and it is judged whether there is any or more than one deviation abnormal data in the real-time data matrix after spraying; If there is any one or more than one deviation abnormal data in the real-time data matrix after the spraying, an initial weight matrix of the open-air scene is obtained, an initial weight ratio of the open-air scene is obtained according to the initial weight matrix of the open-air scene, the initial weight of the initial weight matrix of the current construction scene is corrected according to the initial weight ratio with a preset correction step, and the step S3 is returned.

[0059] The embodiment constructs a weight adaptation mechanism based on scene parameter deviation feedback, identifies abnormal working conditions by monitoring the deviation of the parameters from the scene reference value, realizes targeted correction of the weight by means of the weight ratio of the open-air scene, makes the weight distribution more suitable for the characteristics of the abnormal working conditions, and further improves the adaptation accuracy of the spraying parameter decision to the fluctuations of complex working conditions.

[0060] As a further optimization, step S3 specifically includes the following steps: A spraying parameter decision model is obtained, the spraying parameter decision model including a random forest model and a feature attention layer, the feature attention layer being embedded between an input layer and a decision tree training layer of the random forest model; The real-time data matrix and the calculated weight matrix are input to the input layer, the feature attention layer matching a weight coefficient for each parameter of the real-time data matrix according to the calculated weight matrix to obtain a weighted feature vector; The feature vector is input to each decision tree of the random forest model for independent training, each decision tree outputting a respective parameter prediction result, the parameter prediction results of all the decision trees being fused by weighted voting to obtain a final spraying parameter combination.

[0061] Specifically, the spraying parameter decision model in the embodiment adopts a multi-fusion architecture of a random forest model and an attention mechanism, taking into account the strong generalization ability of the random forest model and the key feature focusing ability of the attention mechanism, and the specific architecture is as follows: The random forest model adopts 100 CART decision trees (classification and regression trees) to form a forest main body, 100 sub-datasets are extracted from the original training sample library by Bootstrap sampling with replacement, and the sample amount of each sub-dataset is 70% of the original dataset, so as to ensure that the training data of each decision tree has diversity. When training each decision tree, a feature random selection strategy is adopted to randomly select parameters from the 6 types of collected data, i.e., dust concentration, dust particle size, environmental wind speed, environmental humidity, working surface humidity and working intensity, as the splitting basis of the current decision tree, and the splitting criterion adopts the principle of minimum mean square error, so as to be suitable for the spraying parameter regression prediction task. When the depth of the decision tree reaches 15 layers or the number of leaf nodes is ≤5, the growth is stopped to avoid overfitting.

[0062] A feature attention layer is embedded between the input layer and the decision tree training layer of the random forest model to form a closed-loop architecture of feature weighting, forest training, and result fusion. The feature attention layer adopts a lightweight structure of multiple perceptrons stacked with softmax normalization processing, and the sum of the weight coefficients output by the feature attention layer is 1.

[0063] The feature attention layer matches weight coefficients to each parameter of the real-time data matrix according to the calculated weight matrix to obtain a weighted feature vector. The weighted feature vector is input to each decision tree of the random forest model for independent training. Each decision tree outputs a respective parameter prediction result. The parameter prediction result includes flow rate, droplet size, angle, and frequency. The results of all decision trees are finally fused by a weighted voting method. Decision trees that are sensitive to features with high attention weights are given higher fusion weights, with a weight proportion of 20-30%. The final spray parameter combination is obtained.

[0064] More specifically, the model training and updating of the spray parameter decision model in the present embodiment adopts a two-stage training strategy of offline pre-training and online fine-tuning to ensure that the model has both full-condition adaptation capability and rapid response to new scenarios. The specific process is as follows: Offline pre-training stage: Training sample library construction: a three-dimensional labeled sample library of multi-dimensional data-spray parameter-dedusting effect or water-saving effect is constructed. The samples are derived from three core construction scenarios, including tunnel scenarios, open-air scenarios, and foundation pit scenarios. The data is obtained by a combination of field construction collection and simulation experiments. The samples cover the full-condition range: wind speed 0-10 m / s, humidity 20-100% RH, work intensity 500-1500 times / min, particle size 0.1-100 μm. 100 groups of effective data are collected under each condition, and the total sample size is ≥100,000 groups.

[0065] Sample labeling and preprocessing: each group of samples is labeled in three dimensions, i.e., input feature labeling, i.e., 6 types of core collected data, spray parameter labeling (including the optimal combination of flow rate, droplet size, angle, and frequency determined by orthogonal experiment), and effect indicator labeling (including dust removal efficiency and water-saving rate). The labeled samples are subjected to outlier rejection and standardization processing according to the 3σ criterion, and are divided into training set, validation set, and test set in a ratio of 7:2:1.

[0066] Model pre-training: the training set is input to the spray parameter decision model, and the batch gradient descent algorithm is used to optimize the model parameters. The learning rate is set to 0.01, the number of iterations is 500, the model performance is verified every 50 iterations using the validation set, and the spray parameter prediction error ≤5% and the dust removal efficiency prediction error ≤3% are set as the qualified standards. When the validation set error does not decrease continuously for 10 iterations, the training is stopped, and the pre-trained model parameters are saved.

[0067] Model performance verification: the pre-trained model is evaluated for performance using the test set, with a requirement that the prediction accuracy of the pre-trained model for the spray parameter combination is ≥95%, the prediction accuracy for the dust removal efficiency is ≥97%, and the prediction accuracy for the water saving rate is ≥96%. If the requirements are not met, the number of decision trees of the random forest model is adjusted or the number of MLP layers is increased / decreased, and the model is retrained until the requirements are met.

[0068] In this embodiment, the parameter ranges of the spray parameter combination are as follows: Flow rate: 0.3-1.0 MPa (corresponding to flow rate 5-50 L / min); Droplet size: 5-50 μm (continuously adjustable); Angle: pitch angle -10°~30°, deflection angle -45°~45°; Frequency: 1-5 times per second, intermittent spraying is used, high-frequency continuous spraying is used for high-concentration dust, and low-frequency intermittent spraying is used for low-concentration dust.

[0069] The above are only preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent flow transformation using the content of the specification and drawings, or direct or indirect application in other related technical fields, is also included in the patent protection scope of the present application.

Claims

1. A mechanical rock-breaking intelligent dust removal system, applied to excavators, characterized in that, Includes a laser module, a sensor module, a control module, and a spray module; in; The laser module is fixed to the actuating component of the excavator, and the laser beam emitted by the laser module is directed towards the rock-breaking working face; The sensor module is mounted on the excavator. The sensor module includes a vision sensor, and the acquisition direction of the vision sensor is set at an angle to the laser beam. The spraying module is located on both sides of the actuating component, and the spraying direction of the spraying module is towards the rock breaking working face; The control module is connected to the laser module, the sensor module, and the spray module, respectively.

2. The intelligent dust removal system for mechanical rock breaking according to claim 1, characterized in that, The laser module includes a laser emitter and a detachable connector. The laser emitter is detachably connected to the detachable connector, which is mounted on the actuating component of the excavator.

3. The intelligent dust removal system for mechanical rock breaking according to claim 2, characterized in that, The detachable connector includes a base, an operating component, a magnet, a buffer unit, and a sliding frame. The magnet is movably installed in the base. The operating component is installed on the base and connected to the magnet to control the movement of the magnet within the base to achieve attraction, locking, or unlocking. The sliding frame is slidably connected to the base. The buffer unit is connected between the sliding frame and the base. The laser emitter is installed on the sliding frame.

4. The intelligent dust removal system for mechanical rock breaking according to claim 3, characterized in that, The buffer unit includes an elastic element, a piston, a rod, and a control valve. The seat is provided with a sliding groove, and the sliding frame is slidably connected in the sliding groove. The seat body is provided with a damping cavity, the piston is slidably connected in the damping cavity, the first end of the rod is fixedly connected to the piston, and the second end of the rod passes through the damping cavity and is fixedly connected to the sliding frame. The piston divides the damping chamber into a rod chamber and a rodless chamber. The seat is also provided with a first channel and a second channel. The first channel connects the rodless chamber to the outside, and the second channel connects the rod chamber to the outside. The control valve is slidably connected to the seat body, and the control valve intersects with both the first channel and the second channel to control the size of the first channel and the second channel; The elastic element is arranged on the side of the sliding frame away from the rod, with one end of the elastic element abutting against the seat and the other end of the elastic element abutting against the sliding frame.

5. The intelligent dust removal system for mechanical rock breaking according to claim 4, characterized in that, The control valve includes a valve stem and a valve cap. The seat has a valve hole that intersects with both the first channel and the second channel. The valve cap is coaxially fixed to one end of the valve stem. The valve stem is threadedly connected to the valve hole. The valve stem includes a first conical surface and a second conical surface. The first conical surface is positioned on the valve stem corresponding to the first channel to adjust the cross-sectional area of ​​the first channel. The second conical surface is positioned on the valve stem corresponding to the second channel to adjust the cross-sectional area of ​​the second channel.

6. A method for intelligent dust removal in mechanical rock breaking, applied to the intelligent dust removal system for mechanical rock breaking as described in any one of claims 1-5, characterized in that, Includes the following steps: S1. Collect dust characteristics data, environmental parameters, and work intensity data during the mechanical rock breaking operation. Preprocess the collected data to obtain a real-time data matrix. The dust characteristics data include dust concentration, dust range, and average dust particle size. The environmental parameters include ambient wind speed, ambient humidity, and work surface humidity. The work intensity data includes rock breaking output frequency and hydraulic system pressure. S2. Obtain the calculation weight matching model, identify the current construction scenario, and obtain the calculation weight matrix corresponding to the current construction scenario based on the calculation weight matching model; wherein, the calculation weight matching model contains the mapping relationship between the construction scenario and the calculation weights of each parameter in the real-time data matrix; S3. Obtain the spray parameter decision model, input the real-time data matrix and the calculated weight matrix into the spray parameter decision model to obtain the spray parameter combination; wherein, the spray parameters include spray flow rate, droplet size, spray angle and spray frequency; S4. Control the operation of the spray module according to the spray parameter combination, and collect the real-time dust concentration after waiting for a preset time. S5. Determine whether the real-time dust concentration is less than or equal to the preset dust concentration threshold; if yes, maintain the operation of the spray module and return to step S4; if no, obtain the preset correction amount of the spray parameter combination, correct the spray parameter combination according to the preset correction amount, and then return to step S4.

7. The intelligent dust removal method for mechanical rock breaking according to claim 6, characterized in that, Step S2 specifically includes the following steps: A scene classifier is obtained to collect the core identification features of the current construction scene. Based on the core identification features, the scene classifier identifies the category of the current construction scene. The scene classifier classifies mechanical rock breaking construction scenes into tunnel scenes, open-pit scenes, and foundation pit scenes. The scene classifier uses three parameters—wind speed stability, operation enclosure, and working face slope—as the core identification features for scene classification. Obtain the weight matching model and obtain the weight matrix corresponding to the current construction scenario based on the category of the current construction scenario; wherein, the weight matrix is ​​a preset initial weight matrix, and the initial weights in the initial weight matrix are set according to the degree of influence of dust characteristics data, environmental parameters and work intensity data on the spraying effect under different scenarios.

8. The intelligent dust removal method for mechanical rock breaking according to claim 7, characterized in that, Step S5 is followed by the following steps: Obtain the real-time data matrix after spraying, determine the rate of change of each parameter in the real-time data matrix per unit time, and determine whether the rate of change of each parameter per unit time is greater than a first preset threshold. If the rate of change per unit time is greater than the first preset threshold, the real-time data is determined to be a mutation data, and it is determined whether there is any one or more mutation data in the real-time data matrix after spraying. If there is one or more abrupt changes in the real-time data matrix after spraying, the initial weights in the initial weight matrix corresponding to the abrupt changes are increased by a preset correction step size, and the process returns to step S3.

9. The intelligent dust removal method for mechanical rock breaking according to claim 7, characterized in that, Step S5 is followed by the following steps: Obtain the real-time data matrix after spraying, determine the deviation of each parameter in the real-time data matrix from the average value of each parameter under the same construction scenario, and determine whether the deviation is greater than a second preset threshold. If the deviation is greater than the second preset threshold, the real-time data is determined to be abnormal deviation data, and it is determined whether there is any one or more abnormal deviation data in the real-time data matrix after spraying. If any one or more deviation abnormal data exist in the real-time data matrix after spraying, the initial weight matrix of the open-air scene is obtained, the initial weight ratio of the open-air scene is obtained based on the initial weight matrix of the open-air scene, the initial weight of the initial weight matrix of the current construction scene is corrected according to the initial weight ratio with a preset correction step size, and the process returns to step S3.

10. The intelligent dust removal method for mechanical rock breaking according to claim 6, characterized in that, Step S3 specifically includes the following steps: A spray parameter decision model is obtained, which includes a random forest model and a feature attention layer, wherein the feature attention layer is embedded between the input layer and the decision tree training layer of the random forest model; The real-time data matrix and the calculated weight matrix are input to the input layer. The feature attention layer matches weight coefficients according to the parameters of the calculated weight matrix and the real-time data matrix to obtain a weighted feature vector. The feature vector is input into each decision tree of the random forest model for independent training. Each decision tree outputs its own parameter prediction results. The parameter prediction results of all decision trees are merged by weighted voting to obtain the final spray parameter combination.

Images