Non-woven fabric, and processing equipment and processing method for non-woven fabric production

Abstract

The invention discloses a non-woven fabric and a processing device and method for non-woven fabric production, the non-woven fabric comprises a pricking needle module and a visual processing module which are fixedly arranged on a rack, through cooperative work of the pricking needle module and the visual processing module, real-time automatic detection of the pricking needle state in the non-woven fabric production process is achieved, and the non-woven fabric production efficiency is improved. In the gap of conventional puncture operation, the system triggers detection steps according to an intelligent period inversely proportional to the gram weight of cloth, accurate positioning is carried out through transverse and longitudinal camera movement assemblies, fixed-point flying shooting and dynamic multi-angle flying shooting are carried out on each set of puncture needles in sequence, the deformation and abrasion state of each puncture needle can be comprehensively captured without large-area shutdown, and the work efficiency is improved. The problems of raw material waste and high defective rate caused by difficulty in real-time monitoring due to dense arrangement of needles and lagging in fault discovery of a traditional needling machine are effectively solved, the production cost of replacing the whole needle plate due to local damage of the needles is remarkably reduced, and meanwhile, the product yield of non-woven fabrics and the continuous operation efficiency of a production line are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent production of non-woven fabrics, in particular to a non-woven fabric and a detection device and method thereof. BACKGROUND

[0002] Needle-punched non-woven fabric is a kind of non-woven fabric made by mechanical reinforcement. Its core process is to use barbed needles to repeatedly pierce the loose fiber web. During the piercing process, the barbs on the needles hook some surface fibers and bring them into the web. When the needles are withdrawn, the hooked fibers stay in the web due to the direction of the barbs or friction with surrounding fibers, and they intertwine and entangle the upper and lower layers of fibers like "pins". After a large number of intensive punctures, the originally loose and weak fiber web becomes a cloth-like material with certain strength, thickness, density, and structural stability.

[0003] A needle-punching machine is a device that turns a certain thickness of non-woven fabric into a non-woven fabric by compaction and needle-punching. It uses needles with triangular or other shapes and barbs on the edges to repeatedly pierce the fiber web. The fiber web formed by cross-web or air-laid web is very fluffy when fed into the needle-punching machine, and only has some strength due to the cohesive force between fibers, but the strength is very poor. When multiple needles pierce the web, the barbs on the needles will move the surface and subsurface fibers of the web from the horizontal direction to the vertical direction, causing the fibers to shift up and down, and the fibers that have shifted up and down will compress the web.

[0004] As disclosed in the published patent "Non-woven fabric needle-punching machine" with publication number CN219653253U, a needle-punching machine is a device that uses needles to repeatedly pierce non-woven fabric to improve its performance. The needles of the needle-punching machine are installed on a mounting plate that can be lifted and lowered. During the downward movement of the mounting plate, all the needles will pierce the non-woven fabric.

[0005] The existing needle-punching machine has the following disadvantages: after long-term use, some needles will be worn out, which will reduce the piercing effect and the quality of the produced products.

[0006] As disclosed in the published patent "Non-woven fabric needle-punching machine" with publication number CN210085730U, the existing needle-punching machine forms non-woven fabric with certain strength by needle-punching the web-formed fibers. Currently, when the needles on the needle-punching plate are damaged to a certain extent, they need to be replaced.

[0007] As disclosed in the published patent "Non-woven fabric needling device" with publication number CN114250548A, non-woven fabric, also known as non-woven cloth, is made of oriented or random fibers and is a new generation of environmentally friendly material. It has the characteristics of moisture resistance, air permeability, flexibility, light weight, non-combustion, easy decomposition, non-toxic and non-irritating, rich color, low price, recyclability, etc. It is a non-woven cloth formed by directly using polymer chips, short fibers or filaments, and then forming a fiber web by airflow or mechanical webbing, and then performing water jet, needle punching or hot rolling reinforcement, and finally performing post-finishing. Needle punching is a typical mechanical reinforcement method, which reinforces and holds the fluffy fiber web through the puncture of the needle of the needle machine.

[0008] The existing needle device mainly includes a rack and a needle plate with needles. If a few needles are damaged, the non-woven fabric production equipment needs to be temporarily stopped, and the entire needle plate needs to be replaced manually. Because a few damaged needles cause the entire needle plate to be discarded, the cost is greatly increased, and resources are wasted.

[0009] In summary, in the production process of non-woven fabric, after the needle machine works for a certain period of time, some needles will be worn or damaged. The worn or damaged needles will seriously affect the yield of non-woven fabric products, causing some products to be scrapped, and the production needs to be stopped for maintenance to reduce the impact on production.

[0010] The applicant points out that in the existing production process, the needles on the needle plate are arranged densely and are not convenient for direct real-time monitoring of the state of the needles. Generally, the needle web after needle processing is detected to evaluate whether the state of the needle is qualified. However, in the actual production process, due to the relatively long production line of non-woven fabric processing (as shown in Figure 23 When it is found that the needle web after needle processing has quality problems, a long section of the needle web has already been processed with problems, which will cause great waste of raw materials and labor costs.

[0011] In summary, in the production process of non-woven fabric, the state of the needle cannot be monitored in real time, which affects the production efficiency and also causes waste of raw materials and labor costs. SUMMARY

[0012] To overcome the above-mentioned deficiencies, the present application aims to provide a technical solution to solve the above-mentioned problems.

[0013] To achieve the above-mentioned purpose, the present application provides the following technical solution: A processing method for non-woven fabric production, comprising a needle module fixed on a rack and a visual processing module; The needle module comprises a needle box connected with an auxiliary power unit, a plurality of power units D1, D2, …, Dn are installed on the needle box in sequence along the longitudinal direction, a plurality of needle groups C1, C2, …, Cn composed of a plurality of needle units are fixedly installed on the working shaft of the power units D1, D2, …, Dn, and the needle groups C1, C2, …, Cn each comprise a plurality of needle units Z1, Z2, …, Zn arranged in parallel; The visual processing module comprises a transverse camera movement assembly arranged transversely on the rack and a longitudinal camera movement assembly arranged longitudinally on the rack, the transverse camera movement assembly comprises a transverse linear module fixedly arranged on the rack along the transverse direction, and a transverse camera unit is arranged on the sliding block of the transverse linear module, and the longitudinal camera movement assembly comprises a longitudinal linear module fixedly arranged on the rack along the longitudinal direction, and a longitudinal camera unit is arranged on the sliding block of the longitudinal linear module; The action of the needle module comprises the following steps: S001: The action of the needle module comprises daily operation steps and detection steps; S002: When the needle module performs the daily operation steps, the power units D1, D2, …, Dn drive the needle groups C1, C2, …, Cn to keep rising, and the auxiliary power unit drives the needle box to rise and fall for puncture operation; S003: When the needle module performs the detection steps, the auxiliary power unit drives the needle box to keep rising, and the power units D1, D2, …, Dn drive the needle groups C1, C2, …, Cn to rise and fall regularly in sequence; S004: The detection steps are inserted into the daily operation steps at a detection interval time t0, and the detection interval time t0 is inversely proportional to the grammage of the non-woven fabric; The operation of the visual processing module comprises the following steps: S100: When the needle module performs the detection steps, the power unit D1 drives the needle group C1 to descend and keep the descending state, and the detection of the needle group C1 is performed; the transverse camera unit is shifted to a detection position Ph to perform fixed-point imaging to obtain a first transverse picture Th1, and the detection position Ph corresponds to the transverse center position of the needle group C1; S110: The longitudinal camera unit is shifted to a detection position Pz1 to perform fixed-point imaging to obtain a first longitudinal picture Tz1, and the detection position Pz1 is collinear with the needle group C1; S120: After the detection of the needle group C1 is completed, the power unit D1 drives the needle group C1 to rise, the power unit D2 drives the needle group C2 to descend and keep the descending state, and the detection of the needle group C2 is performed; in sequence, the fixed-point imaging of the needle groups C2 to Cn is completed, a second transverse picture Th2, …, an n-th transverse picture Thn is obtained, and a second longitudinal picture Tz2, …, an n-th longitudinal picture Tzn is obtained. S130: respectively, Th1, Th2, …, Thn and Tz1, Tz2, …, Tzn are subjected to bending correction and passivation correction; S200: in step S100, when the first lateral picture Th1 is acquired; The lateral camera moves at a maximum uniform speed along the lateral direction, and the lateral camera takes multiple dynamic pictures during the movement to obtain a first lateral flash picture group of the needle group C1, and the first lateral flash picture group includes multiple pictures Thf1, Thf2, …, Thfn taken at multiple angles along the front surface of C1. S210: in step S110, when the first longitudinal picture Tz1 is acquired; The longitudinal camera moves at a maximum uniform speed along the longitudinal direction, and the longitudinal camera takes multiple dynamic pictures during the movement to obtain a first longitudinal flash picture group of the needle group C1, and the first longitudinal flash picture group includes multiple pictures Tzf1, Tzf2, …, Tzfn taken at multiple angles along the side surface of C1. S220: in step S120, by analogy, dynamic imaging of needle groups C2 to Cn is completed, and second lateral flash picture groups, …, n-th lateral flash picture groups are obtained, and second longitudinal flash picture groups, …, n-th longitudinal flash picture groups are obtained. S230: eccentricity correction is performed on the second lateral flash picture group, …, the n-th lateral flash picture group and the second longitudinal flash picture group, …, the n-th longitudinal flash picture group, respectively.

[0014] Compared with the prior art, the beneficial effects of the present application are as follows: The present application realizes real-time automatic detection of the state of the needle in the production process of non-woven fabric through the cooperative work of the needle module and the visual processing module. In the gap of conventional puncture operation, the system triggers the detection step according to the intelligent period inversely proportional to the grammage of the cloth, and the lateral and longitudinal camera movement assemblies accurately position and perform point flash and dynamic multi-angle flash on each group of needles. Without large-area shutdown, the deformation and wear state of each needle can be fully captured, effectively solving the problems of raw material waste and high scrap rate caused by the lagging fault discovery of real-time monitoring of densely arranged traditional needle punching machines, significantly reducing the production cost of replacing the whole needle plate due to local needle damage, and ensuring the product yield and production line continuous operation efficiency of non-woven fabric. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a structural perspective view of the present application; Figure 2 is another structural perspective view of the present application; Figure 3is a structure perspective view of the needle module and the visual processing module in the present application; Figure 4 is another structure perspective view of the needle module and the visual processing module in the present application; Figure 5 is a front view of the needle module and the visual processing module in the present application; Figure 6 is a side view of the needle module and the visual processing module in the present application; Figure 7 is a double perspective view detection schematic diagram of the needle group C1 and a top view diagram of the needle group C1 in the present application; Figure 8 is a horizontal structure schematic diagram of the needle group C1 in the detection step of the present application; Figure 9 is a longitudinal structure schematic diagram of the needle group C1 in the detection step of the present application; Figure 10 is a schematic diagram when the horizontal camera member carries out C1 fixed-point imaging in the present application; Figure 11 is a schematic diagram when the longitudinal camera member carries out C1 fixed-point imaging in the present application; Figure 12 is a schematic diagram when the longitudinal camera member carries out C2 fixed-point imaging in the present application; Figure 13 is a schematic diagram of a first horizontal picture point cloud model in the present application; Figure 14 is a schematic diagram of a standard reference model Z 标h in the present application in comparison with the first horizontal picture point cloud model; Figure 15 is a schematic diagram of a standard reference model Z 标z in the present application in comparison with the first longitudinal picture point cloud model; Figure 16 is a schematic diagram of a reference line X 基 and a reference line X 参 in the present application; Figure 17 is a dynamic imaging schematic diagram of the needle group C1 in the present application; Figure 18 is a schematic diagram in comparison of a point cloud model of a first horizontal snapshot picture group in the present application with a standard reference model Z 标f ; Figure 19 is a schematic diagram in comparison of a point cloud model of a first longitudinal snapshot picture group in the present application with a standard reference model Z 标f ; Figure 20 is a structure schematic diagram in the non-woven fabric processing process in the present application; Figure 21is the profile of a single product unit and the fiber density waveform of the skeleton layer in the present application; Figure 22 is a schematic diagram of common defects of the needle member; Figure 23 is a real photo in the production process of the applicant; The reference signs and names in the drawings are as follows: Rack-001, Needle Module-100, Visual Processing Module-200, Auxiliary Power Member-101, Needle Box-102, Transverse Camera Motion Assembly-201, Longitudinal Camera Motion Assembly-202, Transverse Linear Module-203, Transverse Camera Member-204, Longitudinal Linear Module-205, Longitudinal Camera Member-206. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the 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. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0017] Please refer to Figure 1-23 A processing method for non-woven fabric production, comprising a needle module 100 and a visual processing module 200 fixed on a rack 001; The needle module comprises a needle box 102 connected with an auxiliary power member 101, and a plurality of power members D1, D2, …, Dn are installed in sequence and side by side on the needle box along the longitudinal direction. A needle group C1, C2, …, Cn composed of a plurality of needle members is fixedly installed on the working shaft of each power member D1, D2, …, Dn along the transverse direction. The needle group C1, C2, …, Cn comprises a plurality of needle members Z1, Z2, …, Zn arranged side by side; The visual processing module comprises a transverse camera motion assembly 201 arranged transversely on the rack and a longitudinal camera motion assembly 202 arranged longitudinally on the rack. The transverse camera motion assembly comprises a transverse linear module 203 fixedly arranged on the rack along the transverse direction. A transverse camera member 204 is arranged on the sliding block of the transverse linear module. The longitudinal camera motion assembly comprises a longitudinal linear module 205 fixedly arranged on the rack along the longitudinal direction. A longitudinal camera member 206 is arranged on the sliding block of the longitudinal linear module; The action of the needle module comprises the following steps: S001: The action of the needle module comprises daily operation steps and detection steps; S002: When the needle module performs the daily operation step, the power elements D1, D2, …, Dn drive the needle groups C1, C2, …, Cn to keep rising, and the auxiliary power element drives the needle box to rise and fall for puncture operation; S003: When the needle module performs the detection step, the auxiliary power element drives the needle box to keep rising, and the power elements D1, D2, …, Dn drive the needle groups C1, C2, …, Cn to rise and fall regularly; S004: The detection interval time t0 is inserted into the daily operation step, and the detection interval time t0 is inversely proportional to the grammage of the non-woven fabric; The operation of the visual processing module includes the following steps: S100: When the needle module performs the detection step, the power element D1 drives the needle group C1 to descend and keep descending, and detects the needle group C1 (as shown in Figure 7 and Figure 8 ); The lateral camera element is displaced to the detection position P h中 to obtain the first lateral picture T h1 , and the detection position P h中 corresponds to the lateral center position of the needle group C1; S110: The longitudinal camera element is displaced to the detection position P z1 to obtain the first longitudinal picture T z1 , and the detection position P z1 is collinear with the needle group C1; S120: After the detection of the needle group C1 is completed, the power element D1 drives the needle group C1 to rise, the power element D2 drives the needle group C2 to descend and keep descending, and the detection of the needle group C2 is performed; In turn, the fixed-point imaging of the needle groups C2 to Cn is completed, the second lateral picture T h2 , …, the n-th lateral picture T hn , the second longitudinal picture T z2 , …, the n-th longitudinal picture T zn is obtained; S130: T h1 , T h2 , …, T hn and T z1 , T z2 , …, T zn are respectively subjected to bending correction and bluntness correction; S200: In step S100, when the first lateral picture T h1 is obtained; The lateral camera moves along the lateral direction at the maximum uniform speed, and the lateral camera takes multiple dynamic pictures during the movement to obtain a first lateral flash picture group of the needle group C1, and the first lateral flash picture group includes multiple pictures T hf1 , T hf2 ,..., T hfn ; S210: in step S110, when the first longitudinal picture T z1 is obtained; The longitudinal camera moves along the longitudinal direction at the maximum uniform speed, and the longitudinal camera takes multiple dynamic pictures during the movement to obtain a first longitudinal flash picture group of the needle group C1, and the first longitudinal flash picture group includes multiple pictures T zf1 , T zf2 ,..., T zfn ; S220: in step S120, by analogy, dynamic imaging of the needle groups C2 to Cn is completed, and a second lateral flash picture group,..., an n-th lateral flash picture group are obtained, and a second longitudinal flash picture group,..., an n-th longitudinal flash picture group are obtained; S230: eccentricity correction is performed on the second lateral flash picture group,..., the n-th lateral flash picture group and the second longitudinal flash picture group,..., the n-th longitudinal flash picture group, respectively; The application realizes real-time online monitoring of densely arranged needles of a needle punching machine for non-woven fabric by innovatively deeply integrating an automatic visual detection system with a step-by-step and group-controlled needle module and designing a detection triggering mechanism embedded in a production process, as shown in Figure 22 In the existing production and processing process of non-woven fabric, the needle piece is prone to have defects such as bending, wear or eccentricity; By actively triggering the detection step in the production gap (detection interval time t0), and using the structure of the needle module itself (such as Figure 1 , 2 , 3 and 4, the power pieces D1-Dn can independently control the lifting of the needle groups C1-Cn), the state of each needle group (C1-Cn) can be checked one by one and in order without long-term shutdown and manual intervention, the real-time monitoring problem is completely solved, and early fault warning is realized; The accurate displacement control (lateral linear module and longitudinal linear module) of the lateral camera movement assembly and the longitudinal camera movement assembly, combined with the "flash" technology, can realize the working position of the needle group without changing (the working position of the needle group does not change, only the state of the needle group changes from the raised state to the lowered state, as shown in Figure 5As shown, the lance needle member and the lance needle mounting shaft protrude from the lance needle box), quickly and accurately obtain lance needle images (such as Figure 9 , 10 , 11 and 12) from the front (central position Ph) and side (collinear position Pz) two key dimensions, overcoming the inherent obstacles that manual observation is difficult to observe in real time due to the dense arrangement of lance needles; This makes it possible to discover the bending, dulling and other microscopic wear and damage of individual lance needles in a timely manner during production, rather than being passive in response to large-scale quality problems in downstream webs, achieving true preventive maintenance and early fault warning; The detection step not only includes static point shooting (S100-S130) for basic bending and dulling correction, but also introduces dynamic multi-angle shooting (S200-S230). In dynamic shooting, the horizontal camera member moves uniformly in the horizontal direction and takes multiple timed shots (T hf1 -T hfn ) of the front of the lance needle group, and the vertical camera member moves uniformly in the vertical direction and takes multiple timed shots (T zf1 -T zfn ) of the side of the lance needle group. This is equivalent to "scanning" the lance needle from multiple consecutive angles (as shown in Figure 17 ), creating more comprehensive visual information. This multi-angle, continuous image acquisition method greatly improves the comprehensiveness of the detection, especially for finding the eccentricity problem of the lance needle (by correcting multiple angle pictures), and effectively avoids visual obstruction or misjudgment that may be caused by single angle shooting, significantly improving the reliability and accuracy of the detection results; The detection action is intelligently embedded into the normal "daily operation steps" (S001-S004). For different production lines, the detection interval time t0 is designed to be inversely proportional to the grammage of the non-woven fabric, where the grammage is measured in grams per square meter, indicating the weight of the non-woven fabric per unit area. A higher grammage means thicker and higher fiber density, so the puncture resistance is greater, the wear on the lance needle member is greater, and the probability of bending the lance needle member is greater, so the detection interval can be shorter. The grammage of the fabric is smaller, the puncture is easier, the wear and force on the lance needle member are smaller, and the detection interval time t0 can be longer. For example, the non-woven fabric processed by production line one has a large grammage, which shortens the detection interval time t0 and increases the detection frequency. The non-woven fabric processed by production line two has a small grammage, which prolongs the detection interval time t0 and reduces the detection frequency. This adaptive mechanism ensures that the detection frequency meets the quality monitoring requirements and minimizes the interference with the production rhythm, reduces downtime, and improves production efficiency; Since the detection is carried out when the needle box is lifted (the auxiliary power member keeps the box lifted) and only the specific needle group needs to be lowered, the main body of the production line does not need to be stopped for a long time, and the production can be immediately put into operation after the detection is completed, realizing online detection in a true sense and avoiding the huge production capacity loss caused by traditional large-scale shutdown detection. In an embodiment, at least two groups of the needle module and the visual processing module are arranged on the needle punching machine, when one group of the needle module performs the detection step, the other group of the needle module performs the daily operation step, so as to seamlessly connect, and the main body of the production line does not need to be stopped, which greatly improves the production capacity and greatly reduces the waste of raw materials and production cost: By real-time monitoring and early detection of the faulty needle (group), timely intervention can be made when a small number of needles are problematic (rather than waiting until a large area is damaged or the web is seriously defective), which effectively prevents the defective needles from producing a large number of defective webs, fundamentally solves the core pain point of the prior art that a large amount of raw materials is wasted when the problem is found, and in actual production process, when an emergency occurs, large-scale shutdown maintenance is required, all maintenance personnel cannot be on duty at this time, multiple coordination is required, which greatly affects the production efficiency, while the present application detects at a predetermined detection interval time t0, discovers possible hidden dangers in time, arranges maintenance personnel to be on duty in advance, and improves the production efficiency; Precise positioning to the specific damaged needle group (C1-Cn) rather than the entire needle plate enables maintenance to replace or repair only the specific needle group, avoiding the huge resource waste and cost expenditure caused by discarding the entire expensive needle plate due to damage of a small number of needles (such as the problem disclosed in the published patent with publication number CN114250548A); The present application provides continuous and accurate needle state monitoring, which is an important guarantee for maintaining high product yield and quality stability, reduces defects such as fabric flaws and uneven strength caused by poor needle state, and guarantees product yield and quality stability; The present application realizes real-time automatic detection of the needle state in the non-woven fabric production process through the cooperative work of the needle module and the visual processing module, triggers the detection step according to the intelligent period inversely proportional to the fabric weight during the gap of the conventional piercing operation, accurately positions by the horizontal and vertical camera movement assemblies, and performs point flying and dynamic multi-angle flying on each group of needles in turn, so that the deformation and wear state of each needle can be fully captured without large-scale shutdown, effectively solving the problems of raw material waste and high defective rate caused by the lagging fault discovery and the difficulty in real-time monitoring of the densely arranged needles of the traditional needle punching machine, significantly reducing the production cost of replacing the entire needle plate due to local needle damage, and guaranteeing the product yield of the non-woven fabric and the continuous operation efficiency of the production line.

[0018] In the embodiment of the present application, the step S130 includes the following steps: S131: obtaining the first lateral picture T h1 image data of the needle members Z1, Z2, Z3, …, Zn, generating the first lateral picture point cloud model Z1', Z2', Z3', …, Zn' (as shown in Figure 13 S132: obtaining the first longitudinal picture T z1 image data of the needle members Z1, Z2, Z3, …, Zn, generating the first longitudinal picture point cloud model Z 纵 ', setting up a vertical unified reference surface T, placing Z1', Z2', Z3', …, Zn' and Z 纵 ' on the unified reference surface T, and the unified reference surface T including a horizontal axis x-axis and a vertical axis z-axis; S132: setting a standard reference model Z 标h of the first lateral picture point cloud model on the unified reference surface T, wherein the Z 标h includes Z1 标 , Z2 标 , …, Zn 标 , and the Z1', Z2', Z3', …, Zn' correspond to Z1 标 , Z2 标 , …, Z n标 respectively (as shown in Figure 14 ); grabbing the top midpoint coordinates of Z1' and the top midpoint coordinates of Z1 标 , comparing the matching rate of Z1' and Z1, grabbing the top midpoint coordinates of Z2' and the top midpoint coordinates of Z2 标 , comparing the matching rate of Z2' and Z2 标 , …, grabbing the top midpoint coordinates of Zn' and the top midpoint coordinates of Zn 标 , and comparing the matching rate of Zn' and Zn 标 ; S133: when the matching rate of any one of Z1', Z1 标 , Z2', Z2 标 , …, Zn', and Zn 标 is close to the lowest threshold value, triggering an expected maintenance task; S133: when the matching rate of any one of Z1', Z1 标 , Z2', Z2 标 , …, Zn', and Zn 标 is lower than the lowest threshold value, triggering a shutdown maintenance task; After the bending correction of the first lateral picture T h1 is completed, the second lateral picture T h2 , …, the n-th lateral picture Thn bend correction, S134: A standard reference model Z of the first longitudinal picture point cloud model is set on the unified reference plane T 标z (as shown in Figure 15 Z 标z includes a reference line Z 参 parallel to the z-axis, which is matched with the coordinate of the top midpoint of the first longitudinal picture point cloud model Z 参 , and Z 参 is the symmetry axis, and reference line Z 参 and reference line Z 左 are respectively arranged on the two sides of Z 右 , the distance between the reference line Z 左 and the reference line Z 右 is equal to the diameter of the standard needle, and the matching rate of Z 左 and Z 右 is compared as the reference, and the matching rate of Z 纵 and Z 标 z is compared as the reference; S135: When the part of Z 纵 ' exceeding the reference line Z 左 or the reference line Z 右 approaches the highest threshold, an expected maintenance task is triggered; When the part of Z 左 exceeding the reference line Z 右 exceeds the highest threshold, a shutdown maintenance task is triggered; After the bend correction of the first longitudinal picture T z1 is completed, the bend correction of the second longitudinal picture T z2 ,..., and the n-th longitudinal picture T zn is completed in turn. Traditional manual visual inspection or simple image comparison is easily affected by subjective factors such as light, angle, and experience, and cannot uniformly and accurately define “slight bending” or “severe bending”. The present application realizes the objectivity and accurate quantization of the detection standard, greatly improving the judgment accuracy. The present application generates high-precision point cloud models (Z1', Z2' and Z 标h , Z 标z ), converts the visual information of the needle into quantifiable spatial data, and through strict coordinate matching (such as top midpoint coordinate matching) and digital matching rate calculation with the standard reference model (Z 标h , Z 标z ) on the unified reference plane T, the subjective deviation is completely eliminated; the “matching rate” and “exceeding amount” are specific numerical indicators, so that the bending degree of the needle can be accurately measured and recorded, realizing data-driven and standardization of quality control, and the judgment result is reliable. The introduction of a predictive maintenance mechanism shifts from a passive approach to a proactive one. Existing technologies typically only trigger downtime for maintenance after needles have completely failed or defective products have been produced, which is a costly afterthought. The present invention employs a two-tiered triggering mechanism: Predicted maintenance tasks (early warnings): These tasks are triggered when the matching rate approaches a minimum threshold or the excess quantity approaches a maximum threshold. This serves as an early warning system for failures. While not an immediate production halt, they signal to operators that a needle is undergoing minor deformation and is on track to fail in the future. Downtime maintenance tasks: These tasks are triggered only when the matching rate falls below a threshold or the excess quantity exceeds a threshold. This indicates that the needle has been confirmed as defective and must be replaced as soon as possible. However, these replacements are controlled by the system and result in less labor and material losses than unplanned downtime. This hierarchical mechanism enables maintenance teams to plan ahead and intervene at the most appropriate times (e.g., during scheduled maintenance or shift handovers), minimizing unplanned downtime and production disruptions. On "Z1' and Z1 标 ", "Z2' and Z2 标 "... until "Zn' and Zn 标 "A one-by-one, independent comparison is performed, which means that the system can not only detect that "a needle is broken", but also accurately locate which needle (or needles) have a problem, realizing independent monitoring and precise positioning of each needle, improving maintenance efficiency; Maintenance personnel no longer need to conduct extensive inspections of the entire needle board or directly replace the entire, expensive needle board. Instead, they can quickly locate and replace specific problem needles based on system reports. This greatly shortens maintenance time, reduces spare parts consumption, and significantly reduces maintenance costs. Not only using horizontal pictures (T h1 ) to detect the bending of the needle in the XZ plane (front view) (steps S132-S133), and innovatively utilize the longitudinal image (T z1 ) The generated point cloud model is generated by setting the reference line Z 左 and Z 右 To detect the deflection or bending of the needle in the Y-axis direction (side view) (steps S134-S135), the two-dimensional detection is expanded to the three-dimensional space verification, eliminating the detection blind area; Figure 7 As shown in the figure, this dual-view verification ensures that no matter which direction the needle bends, it can be effectively captured, overcoming the defect of a single view that may miss certain directions of bending, making the detection results more comprehensive and reliable; All the data generated in the detection process (the matching rate of each needle each time, the excess amount, the early warning record, the maintenance record) will be saved by the system, and these massive and high-value data can be used for trend analysis, analysis of the average service life and wear curve of a specific type of needle, process optimization, study of the influence of different gram weights of non-woven fabric and different puncture frequencies on needle wear, optimization of production parameters, and provision of core data support for digitalization and intelligentization of the production process. By introducing a digital comparison and hierarchical early warning mechanism based on a point cloud model, the accuracy, automation level and response efficiency of the needle bending detection are significantly improved; by converting the collected horizontal and vertical images into high-precision point cloud models and performing coordinate alignment and matching rate calculation with the standard model in a unified reference plane, quantitative judgment of the bending degree of each needle is realized, thereby overcoming the subjectivity and inefficiency of manual visual inspection; at the same time, the scheme sets a double-trigger mechanism of "approaching threshold early warning" and "threshold exceeding shutdown", which can both prevent problems from occurring and arrange expected maintenance to reduce unplanned shutdowns, and can also intervene decisively when a fault is critical to avoid the production of defective products, ultimately realizing the transition from passive maintenance to predictive intelligent maintenance, greatly ensuring production continuity and product quality stability.

[0019] In the embodiment of the application, the expected maintenance task in steps S133 and S135 includes the following steps: S137: the difference between the matching rates in Z1' and Z1, Z2' and Z2,..., and Zn' and Zn is recorded as the horizontal offset, and the part exceeding the reference line Z left or the reference line Z right in Z vertical' is recorded as the vertical offset; similarly, the offset statistics of the second horizontal picture T 标 , the n-th horizontal picture T 标 , and the second vertical picture T 标 ,..., the n-th vertical picture T h2 are completed respectively; hn z2 zn S138: the horizontal offset and the vertical offset are uniformly marked as the offset Z 偏 , and the expected maintenance time is t 检 , wherein Z 偏 and t 检 satisfy the following corresponding relationship: t 检 =k / λZ 偏 , wherein λ is the gram weight of the non-woven fabric, and k is a processing constant; ​​​Alerts alone (such as the aforementioned "approaching threshold") are insufficient. They only answer the questions "what is broken" and "is likely to break down" but fail to answer the question "when must it be repaired at the latest?" Operators are still faced with the dilemma of whether to shut down the machine immediately (potentially causing production losses) or delay the operation (potentially leading to an accident). By setting a formula for expected maintenance tasks, a qualitative leap from "static alerts" to "dynamic and precise decision-making" has been achieved. Formula t 检 = k / λZ 偏 This decision-making problem is solved. Instead of outputting a vague warning, it outputs a precise, quantified time point (expected maintenance time t检验). This has transformed maintenance scheduling from an experience-based, vague "decision" to a data-driven, precise scientific calculation. The system can clearly inform production planners: "Based on the current needle bending degree and the fabric being produced, it is recommended to schedule maintenance in X hours and X minutes." This has achieved automation, precision, and scientific maintenance decision-making, innovatively linking "equipment health status" with "production conditions" intelligently, and offset Z 偏 It is a quantitative indicator of the health status of the equipment, Z 偏 The larger the value, the more serious the needle bending or deflection is, and the risk of needle breakage or defective product production increases, indicating an urgent situation. The gram weight λ of nonwoven fabric is a quantitative indicator of the current production load and wear intensity. The larger the gram weight λ, the thicker and denser the nonwoven fabric is, the greater the force required for puncture and the faster the wear rate of the needle. Under high gram weight conditions, a tiny defect will worsen at a faster rate. The algorithm cleverly links these two most critical factors. When Z 偏 When it is very large (the needle is in very bad condition), no matter what is produced, the t inspection will become very small, and the system will require immediate or quick arrangement of maintenance, giving priority to safety and quality; When λ is very large (e.g., when producing high-wear products), the algorithm will proactively shorten the t-inspection even if the Z deviation does not appear to be particularly severe at the moment. This is because it predicts that the defect in this needle will rapidly expand under high-intensity wear and must be addressed before the problem occurs. This embodies true "predictive" maintenance, not just "preventive" maintenance. In the traditional mode, for the sake of safety, people may tend to stop the machine for inspection immediately, but this may interrupt a production task that could have been running safely for a long time. This solution allows the equipment to continue running until the optimal time point t under the premise of controllable risks through calculation. 检This fully exploits the equipment's potential for safe operation, avoids unnecessary production interruptions, and maximizes production efficiency and resource utilization. This model provides key input to the entire plant's production planning system, allowing the maintenance team to receive a list of tasks requiring maintenance in the coming hours or days, along with their priorities (sorted by urgency). This allows them to scientifically arrange manpower and prepare spare parts, achieving efficient allocation of maintenance resources and reducing waiting times for both personnel and equipment. The machining constant k is not a fixed value. It can be used as an initial empirical value. By continuously collecting historical data (such as the deviation between actual failure time and maintenance records and predicted t-tests) and performing machine learning optimization, the system will become increasingly "smarter" and the predicted t-tests will become more and more accurate, thus forming a continuously self-improving intelligent maintenance closed loop. By introducing the offset Z 偏 and dynamic algorithm model of nonwoven fabric weight λ (t 检 =k / λZ 偏 ), upgrades the expected maintenance tasks from static early warning to precise dynamic scheduling and decision support; its core benefit is to achieve intelligent linkage between maintenance strategy and actual equipment wear status and production load, offset Z 偏 The severity and urgency of needle bending are quantified, while the nonwoven fabric weight λ reflects the wear intensity of the needles caused by the current production task; the algorithm automatically calculates the optimal expected maintenance time t 检 , ensuring high wear (high λ) or high risk (high Z 偏 ) status, maintenance can be arranged more quickly, thereby maximizing the effective operation time of the equipment while ensuring production safety and product quality, achieving a leap from "early warning" to "precise execution", and greatly improving the scientific nature of production plans and the utilization efficiency of maintenance resources.

[0020] In the embodiment of the present invention, the left side horizontal safety lines Z are respectively provided on both sides of the standard reference model of the first horizontal image point cloud model. 左h and right side horizontal safety line Z 右h The left side of the standard reference model of the first longitudinal image point cloud model is provided with a left longitudinal safety line Z 左z The right side of the standard reference model of the nth longitudinal image point cloud model is provided with a right longitudinal safety line Z 右z , The left side transverse safety line Z 左h The distance from Z1 to Z1' is equal to the distance from the needle piece Z1 to the edge of the needle box. 右h The distance to Zn' is equal to the distance from the needle element Zn to the edge of the needle box; The left longitudinal safety line Z 左zThe distance of the point cloud model Zlong' of the first longitudinal picture to the distance of the needle group C1 to the edge of the needle box is equal to the right longitudinal safety line Z 右z The distance of the point cloud model Zlong' of the nth longitudinal picture to the distance of the needle group Cn to the edge of the needle box is equal to the right longitudinal safety line Z The first transverse picture T h1 , …, the nth transverse picture T hn , the point cloud model of any transverse picture exists beyond the left transverse safety line Z 左h and the right transverse safety line Z 右h , emergency stop is triggered; The first longitudinal picture Tzn, …, the nth transverse picture Tzn, the point cloud model of any longitudinal picture exists beyond the left longitudinal safety line Z 左z and the right longitudinal safety line Z 右z , emergency stop is triggered; The monitoring angle is raised from the microscopic 'needle' to the macroscopic'space relationship between the needle module and the machine structure', and its core goal is to protect the needle machine body and the surrounding production line equipment worth hundreds of thousands or even millions of yuan from serious impact, which is a higher level and more critical protection because the loss and downtime of equipment damage are much higher than the production of some defective products; Traditional mechanical anti-collision measures (such as physical limit switches) usually trigger after or during collision, which is too late and damage has already occurred. The present application creates a virtual 'digital safety fence' to prevent problems before they occur. As shown in Figure 14 and Figure 15 , by accurately setting the left transverse safety line (Z 左h ), the right transverse safety line (Z 右h ), the left longitudinal safety line (Z 左z ), and the right longitudinal safety line (Z 右z ) around the point cloud model, a virtual safety area that cannot be crossed is defined for the movement range of the needle group in the digital world. The boundaries of this area are set according to the real physical distance of the needle to the edge of the box, ensuring that the rules of the digital world and the physical world are consistent. Any needle (or its point cloud model) that crosses this 'electronic fence' in the digital world is determined to have a high risk of collision in the physical world, and the most extreme measure (emergency stop) is taken before physical contact occurs. Due to loose installation bolts, control failure of auxiliary power components (such as cylinders / motors driving the needle box to rise and fall), or transmission mechanism failure, the entire needle box may be misaligned in the transverse or longitudinal direction, at which time all needles will be offset in the same direction. The present scheme monitors the overall point cloud model and Z 左h / Z右h or Z 左z / Z 右z The systematic deviation can be immediately found and the emergency brake can be activated to prevent the whole row of needles from hitting the frame; Individual needle may break or fall off due to material fatigue, in which case the needle will stretch out of the normal range of the array like an out-of-control metal hook, crossing the lateral safety line (Z 左h / Z 右h The design of the lateral safety line (Z 左z / Z 右z ) is to capture such dangerous abnormal state that is far beyond the normal bending correction range; the fixed base of a group of needles (C1, C2...Cn) may be loose, causing the group of needles to shift in the longitudinal direction (the direction of the production line), and the longitudinal safety line (Z 左h / Z 右h ) is specifically designed to detect such position abnormality in the longitudinal direction to prevent interference with adjacent rollers and other equipment; The needles are made of high-strength steel, while the frame, the supporting net plate, the stripping net plate and the conveying rollers of the needle loom are usually thick and heavy precision machined parts; the needles in high-speed motion collide with these parts, which may cause the needles to break, the roller surface to be scratched, the needle plate to be deformed, the transmission system to be damaged, and even the main structure of the frame to be damaged, resulting in a very long downtime and high cost of spare parts and labor for maintenance. The solution of the present application directly reduces the risk to almost zero, and a sudden collision not only damages the equipment, but also may tear or scratch the web being produced, resulting in a large amount of in-process products being scrapped and raw materials being wasted; Unplanned emergency stop will completely disrupt the production plan, cause the order to be delayed, generate liquidated damages and affect the reputation of the enterprise, while the "emergency stop" triggered by the present solution is a controlled and protective stop, which aims to avoid a longer and more loss-causing unplanned stop, and greatly guarantees the stability and continuity of the production plan as a whole; By establishing the lateral and longitudinal three-dimensional safety boundaries (Z 左z / Z 右z), the detection range is improved from the health status of a single lancet to the system-level monitoring of the safety of the entire lancet module and machine structure, realizing real-time detection of abnormal overall deviation or serious deformation of individual lancets in the spatial position of the lancet array as a whole; by accurately setting the position of the safety line (which is equal to the actual physical distance of the lancet or lancet group to the edge of the box), the system can sensitively identify any point cloud model exceeding the virtual safety boundary and immediately trigger an emergency shutdown of the highest priority, thereby intervening in milliseconds before a physical collision occurs, completely avoiding catastrophic equipment damage, long-term production interruption and the resulting significant economic losses that may be caused by lancet impact on the rack, conveying roller or other production line equipment, and constitutes a stable and key intelligent defense line to ensure the physical safety operation of the equipment.

[0021] In the embodiment of the application, the passivation calibration in step S130 comprises the following steps: S1331: when the matching rate of any lancet part is not less than 95% in steps S133 and S134, passivation calibration is performed; S1332: a reference line X is provided on the unified reference plane T 基 and the reference line X 参 , the X 基 is parallel to the x-axis and the X 基 is collinear with the upper vertices of Z1', Z2', Z3',..., and Zn'; the X 参 is parallel to the x-axis, the X 参 is provided below the X 基 , and a standard height is provided between the X 参 and the X 基 , the standard height is equal to the length of the standard lancet part (as shown in Figure 16 ); The lower end points of Z1', Z2', Z3',..., and Zn' are compared with the X 参 respectively, and the differences between the lower end points of Z1', Z2', Z3',..., and Zn' and the X 参 are recorded as wear values respectively; S1333: when any wear value in Z1', Z2', Z3',..., and Zn' approaches the highest threshold value, an expected maintenance task is triggered; After completing the passivation calibration of the first horizontal picture T h1 , the same is true for the second horizontal picture T h2 ,..., and the n-th horizontal picture T hn ; Because only when the bending of the needle is within a reasonable range, it makes sense to consider blunting, and the impact of blunting on production is not as great as bending; At the same time, because blunting is basically impossible to be caused in a short period of time, there is no need to consider the shutdown maintenance task; Step S1331 sets "the matching rate is not lower than 95%" to start blunting correction, which is an extremely engineering-wise optimization step, which first admits a fact: a bent (matching rate lower than 95%) needle itself has precision problems, which must be considered first, and it is unnecessary and uneconomical to spend computing resources to detect whether its tip is blunt at this time, the system will automatically skip the subsequent blunting analysis of these "precision-reduced" needles, and concentrate the limited computing power and detection time on those "healthy body but may have'skin disease' (blunting)" needles, which avoids invalid detection and significantly improves the efficiency of the whole system; The "sharpness" or "blunting" of the needle in the past is a very subjective judgment, which is heavily dependent on the "eyesight" and "hand feeling" of the master, and cannot be standardized. The present scheme creates a new measurement method by introducing the reference line (X 基 ) and the reference line (X 参 ), and using the standard height (equal to the standard needle length) between them as a ruler; By calculating the vertical difference between the lower end point of the needle point cloud model and X 参 , the abstract "blunting" degree is directly converted into a specific, recordable value - wear value, which accurately reflects the actual length loss caused by needle tip wear. Wear 0.1mm and wear 0.5mm represent completely different wear stages, and the system can accurately distinguish and record them, completely eliminating human subjective errors; Only set "trigger expected maintenance task when close to threshold", but not set "shutdown maintenance", which is deeply consistent with the mechanism of blunting failure. Unlike bending, which can occur instantly, needle tip wear is a slow, gradual, and predictable material loss process that does not immediately cause web tearing or equipment risk like broken needles or severe bending, but gradually reduces product quality (such as loose cloth). This design fully fits the predictive maintenance strategy of the wear characteristics, maximizing production continuity. This design allows the system to give the maintenance team a long-term (possibly several days or even weeks) warning, such as: "No. X needle is expected to reach the wear limit after Y time"; This allows the replacement of the needle to be comfortably integrated into the established production plan, shift handover, or regular maintenance window, thereby completely avoiding unplanned production interruptions caused by handling blunting problems. This is a low-cost, less disruptive maintenance mode for production processes; The system continuously records the wear value of each needle and its change rate, which constitutes a valuable "needle full life cycle database". By analyzing these data, a needle life database is constructed, and the enterprise can scientifically judge the wear resistance and service life of needles of different brands and different materials in actual production, provide solid data support for procurement decision-making, study the wear rate of needles for different fiber raw materials and different gram weight products, and then optimize the piercing depth, frequency and other parameters, while ensuring quality, prolong the service life of the needle. Based on accurate needle life prediction, an optimal spare parts inventory strategy can be developed to avoid stockout risk and reduce capital occupation.

[0022] In the embodiment of the application, the eccentricity correction in step S230 includes the following steps: S231: When the matching rate of any needle part is not less than 95% in steps S133 and S134, and the wear value of any needle part is less than 5% in step S1332, eccentricity correction is performed; Obtain image data of multiple pictures Thf1, Thf2, …, Thfn in the first horizontal flyby picture group, respectively generate point cloud models Thf1', Thf2', …, Thfn', obtain image data of multiple pictures Tzf1, Tzf2, …, Tzfn in the first vertical flyby picture group, respectively generate point cloud models Tzf1', Tzf2', …, Tzfn', and place Thf1', Thf2', …, Thfn' and Tzf1', Tzf2', …, Tzfn' on a unified reference surface T respectively; S232: A unified standard reference model Z is provided on the unified reference plane T, which has the first horizontal flyby picture group point cloud model and the first vertical flyby picture group point cloud model 标f The point cloud model of a single needle part in Thf1', Thf2', …, Thfn' and Tzf1', Tzf2', …, Tzfn' is compared and matched with the unified standard reference model Z 标f Compare and match, and record the matching rate; S233: When any matching rate in Thf1', Thf2', …, Thfn' and Tzf1', Tzf2', …, Tzfn' approaches the minimum threshold, trigger the expected maintenance task; After completing the eccentricity correction of the first horizontal flyby picture group and the first vertical flyby picture group, the same is done for the second horizontal flyby picture group, …, the nth horizontal flyby picture group and the second vertical flyby picture group, …, the nth vertical flyby picture group respectively; By setting the two extremely stringent prerequisites of "matching rate ≥ 95%" and "wear value < 5%", it is clearly defined that the target objects of eccentricity correction are needles that are almost completely healthy in terms of static geometry (bend) and basic performance (bluntness). This marks the maturity of the system's detection logic: first solve the serious problems that directly affect production (bend, severe bluntness), and then concentrate superior resources on overcoming more subtle and potential problems (eccentricity); For a bent or blunt needle, the eccentricity test result is invalid and consumes a lot of computing resources. This filter condition ensures that the system only performs this complex analysis on needles that are "worthy of testing", greatly improving the efficiency of high-level testing and the utilization of computing resources, avoiding resource waste. Previous detection was based on one or two “static photos” (T h1 , T z1 ), this solution uses the first horizontal flying picture group and the first vertical flying picture group, which is equivalent to using a high-speed camera to record a "video" of the needle from the front and side, from static snapshots to dynamic images; Figure 18 and 19 As shown, all the images of these consecutive frames are converted into point cloud models (Thf1'...Thfn', Tzf1'...Tzfn') and placed on a unified reference plane T. This is equivalent to reconstructing the motion trajectory and morphological changes of the needle throughout the entire shooting process in the digital world. This analysis no longer focuses on the "appearance" of a single needle, but rather on the "movement" of a single needle. The slight eccentricity of the needle is often difficult to show in a static state due to its rigidity. Such eccentricity defects can only be discovered through multi-angle and consecutive frame comparison. The first horizontal and first vertical flying pictures can quickly reveal eccentric needles by using multiple consecutive pictures in each group. Comparing all dynamically acquired point cloud models with a unified, ideal standard model eliminates comparison errors caused by different angles and lighting, ensuring that all test results are judged based on the same benchmark, making the diagnosis objective and accurate. An independent matching rate calculation is performed for each point cloud model in the dynamic sequence, meaning the system evaluates the state of the needle at every moment throughout its entire motion cycle, rather than an "average state." This allows for precise quantification of the magnitude and frequency of eccentricity, thereby determining its severity. In summary, bending, blunting, and eccentricity together form an impeccable needle health monitoring system. This system covers the three main failure modes of needles: geometric deformation, performance degradation, and precision defects, forming a complete monitoring closed loop covering the failure modes of the needle throughout its life cycle. The eccentricity detection often finds the earliest stage of the fault, and the system can send an early warning at the fault budding stage by triggering the expected maintenance task when the matching rate is close to the minimum threshold, allowing the maintenance personnel to quickly solve the problem at the next planned shutdown; By setting high-standard double-pre-filtering conditions (matching rate >= 95% and wear value < 5%), the eccentricity correction is locked to the geometrically almost perfect needle, ensuring the accuracy and effectiveness of the detection object; the core beneficial effect is that the massive point cloud model (Thf1'-Thfn', Tzf1'-Tzfn') generated by multi-angle flying shots constructs the dynamic motion trajectory atlas of the needle in the three-dimensional space, and by comparing with the unified standard model Z 标f The super-high-precision all-around comparison realizes the sensitive capture and quantitative diagnosis of the micro eccentricity, and finally forms a full-coverage and dead-angle-free monitoring system for the needle state from macro bending to micro passivation to eccentricity, greatly improving the predictability and fine level of quality control.

[0023] In the embodiment of the application, the steps S133, S135, S1333, S233 respectively include a dust suction operation: The dust suction operation includes the following steps: When the matching rate or the wear value is close to the threshold, the current step is repeated after several times of dust suction operation; As disclosed in the published patent CN219450073U, "Needle Sticking Device for Manufacturing Elastic Nonwoven Fabric", when the needle sticking device performs needle sticking processing on nonwoven fabric, fibers will inevitably adhere to the metal needle. The longer the needle sticking time is, the more fibers will adhere to the metal needle. In order to ensure the needle sticking effect and the quality of the produced nonwoven fabric, the worker needs to regularly remove the fibers on the metal needle. Generally, the worker removes the fibers manually, which is not only time-consuming and laborious, but also has the risk of being pricked; After the needle pierces the fiber web, fibers will inevitably adhere to the needle. These entangled fibers will significantly change the apparent geometric shape of the needle in visual detection. For example, a bundle of fibers on the needle tip may be misjudged as a "passivation block". The fiber entanglement on the needle body may be identified as "bending" or "thickening" by the point cloud model, resulting in a decrease in the matching rate and thus a serious false alarm; In the present application, instead of using complex image algorithms to distinguish fibers and metals, the most direct and reliable physical method "cleaning" is adopted. When the system is about to make a key judgment (the matching rate or the wear value is close to the threshold), the dust collection operation is automatically triggered. This operation can strip the non-essential interference attached to the lancet, so that the subsequent repeated detection can capture the real metal surface state of the lancet, which greatly improves the accuracy and reliability of the basic data of all detection algorithms (bending, passivation, eccentricity correction), and eliminates false judgments caused by pollution from the source; This step defines an advanced decision logic: when the system finds an anomaly, its first reaction is not to alarm immediately, but to start a "diagnostic" process - "whether the anomaly is caused by a removable contaminant", first clean up, and then repeat the detection to verify whether the anomaly still exists. This process simulates the reasoning process of a mechanic. If the matching rate returns to normal after cleaning, it means that it is a fiber interference and no maintenance action is needed, effectively reducing unnecessary maintenance orders. If the anomaly still exists after cleaning, it is confirmed as a real hardware damage, at which time the warning is triggered, making the system evolve from a simple "sensor" to an intelligent terminal with "primary diagnostic capability". It should be noted that the dust collection operation is performed by a human or a robot arm, which is a common technical means in the industry and will not be described here. By automatically triggering the dust collection operation when the matching rate or the wear value of the monitored lancet is close to the threshold, the industry problem of fiber attachment interference with detection accuracy is effectively solved. By introducing intelligent cleaning intervention at the key judgment node (near the threshold), the system can automatically distinguish between real mechanical damage of the lancet and temporary detection deviation caused by fiber winding, thereby greatly reducing the false judgment rate and avoiding unnecessary maintenance triggered by fiber shielding. This design not only ensures the accuracy and reliability of the bending, passivation and eccentricity correction results, reduces the frequency and safety risk of manual cleaning, but also improves the intelligence and decision-making science of the entire condition monitoring system as a whole, ensuring that maintenance actions are always based on the real state of the lancet rather than surface stains.

[0024] In the embodiment of the present application, the broken needle detection is included in steps S133 and S134: The broken needle detection includes the following steps: In steps S133 and S134, when the matching rate is below 20%, the subsequent device is simultaneously started to remove the broken needle from the non-woven fabric blank. As described in the published patent CN115772746A, the lancet may be accidentally broken during the process of piercing the non-woven fabric blank. After the non-woven fabric blank is compressed, the broken needle heads embedded therein need to be detected and removed. The existing non-woven fabric needle punching machine only detects the broken needle head mixed in the non-woven fabric embryo through a metal detector during use to show whether the broken needle head is mixed, but cannot circle and select the area mixed with the broken needle head, so that the subsequent needle head removal lacks pertinence and is not conducive to improving the subsequent work efficiency; In the present application, the criterion of "a matching rate below 20%" for broken needles has high engineering rationality; if the matching rate of the point cloud model of a needle and the standard model is as low as this, it is very likely that the physical form of the needle has been fundamentally changed - the needle body is mostly or entirely broken and missing, and this threshold setting can effectively exclude other interference and achieve high accuracy in diagnosing broken needle faults; The detection is directly embedded in the conventional bending correction steps (S133 and S134), which means that the system synchronously screens this fault in each routine inspection, realizes real-time monitoring and immediate discovery of broken needles, and is not only a "diagnosis system" but also an "execution system". After discovering a broken needle, the system does not simply issue an alarm and wait for manual processing, but immediately and automatically issues an instruction to the subsequent equipment to start the targeted removal program. The broken metal needle head will be mixed into the fluffy fiber web, and as the embryo enters the subsequent process, it may scratch or even pierce the surface of expensive hot rolling rollers, fabric guide rollers and other precision equipment, causing permanent damage, high maintenance costs, and the broken metal needle head is rolled into the fabric surface, becoming a hard metal flaw in the product, resulting in downgrading or even scrapping of the entire roll of fabric. After shutdown, workers need to spend a lot of time searching for broken needle fragments in the huge embryo, causing a long production interruption, The present scheme aims to physically remove the broken needle fragments from the production line before they cause greater damage through immediate linkage and removal, thereby avoiding secondary disasters, protecting downstream equipment and ensuring the quality of subsequent products.

[0025] A processing device for non-woven fabric production includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above processing method when executing the computer program.

[0026] A non-woven fabric produced by the above processing method includes a first spunbond layer, a first short fiber PET skeleton layer, a meltblown layer, a second short fiber PET skeleton layer, and a second spunbond layer arranged from top to bottom. The non-woven fabric includes a plurality of product units arranged from left to right. In the single product unit, the fiber density of the first short fiber PET skeleton layer and the second short fiber PET skeleton layer gradually increases from both sides to the middle, reaches a wave crest at 1 / 4, then decreases to a wave trough at 3 / 8, and then uniformly transitions to the middle position. like Figure 20 As shown, in the actual processing process, after being processed by the needle machine, the non-woven fabric is formed by processes such as hot rolling, and then the non-woven fabric is cut into multiple long strip product units based on the single product unit. The long strip product units are then cut as needed to finally make finished products such as masks and protective masks. For example, in masks, existing SMS structure masks generally include a first spunbond layer, a meltblown layer, and a second spunbond layer. Although they can provide virus protection, they are too soft, so when worn, they generally stick to the mouth and nose. The mask will be contaminated with water vapor exhaled by the human body, causing the part in contact with the mouth and nose to be wet and uncomfortable, and at the same time reducing the protective effect. In order to avoid this problem, KN95 masks are formed by arching the entire mask. However, due to the need for molding, the overall material of the KN95 mask is thicker, the production cost is higher, and some structures need to rely on multiple metal or non-metal frames. The assembly process is cumbersome and the labor cost is high. In the present invention, by adding a skeleton layer to a conventional SMS structure mask and designing the fiber density of the skeleton layer, the use effect of the mask is enhanced while controlling the cost and not increasing the production process; First of all, the two sides of a single product unit are relatively soft, which is convenient for fitting with the facial contour to ensure the overall protective sealing of the mask. Figure 21 As shown, the fiber density of the first staple PET skeleton layer and the second staple PET skeleton layer gradually increases from both sides to the middle, reaches a peak at 1 / 4, and then decreases toward the middle to 3 / 8 and reaches a trough, so that a single product unit forms a force-bearing unit that can support force near both sides of the mouth and nose. The force-bearing structure of the force-bearing unit is based on the change in the fiber density of the first staple PET skeleton layer and the second staple PET skeleton layer. Based on this force-bearing structure, the single product unit can be manually lifted to slightly bulge from the left 1 / 4 to the right 1 / 4, thereby preventing the center of the mask from fitting against the mouth and nose. In addition, the structure in the middle of the mask does not change much compared with before, thereby ensuring the basic use effect of the mask while preventing the mask from being too stuffy. The present invention can improve the use effect of the mask by changing the structural level and designing the fiber density without changing the original production process.

[0027] It will be apparent to those skilled in the art that the application is not limited to the details of the above-exemplified embodiments and that the present application can be implemented in other particular forms without departing from the spirit or essential characteristics of the present application. The embodiments should therefore be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the above description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No reference signs in the claims should be considered as limiting the scope of the claims with respect to the figures of the patent document.

Claims

1. A processing method for nonwoven fabric production, characterized in that: It includes a needle module and a visual processing module fixed on the frame; The needle module includes a needle housing connected to an auxiliary power member, on which a plurality of power members D1, D2, ..., Dn are longitudinally and sequentially mounted in parallel, and needle groups C1, C2, ..., Cn consisting of a plurality of parallel needle members are transversely fixedly mounted on the working axes of the power members D1, D2, ..., Dn, respectively, and the needle groups C1, C2, ..., Cn respectively include a plurality of parallel needle members Z1, Z2, ..., Zn; The visual processing module includes a transverse camera motion component disposed transversely on the frame and a longitudinal camera motion component disposed longitudinally on the frame, wherein the transverse camera motion component includes a transverse linear module fixedly disposed transversely on the frame, and a transverse camera component is mounted on a slider of the transverse linear module; and the longitudinal camera motion component includes a longitudinal linear module fixedly disposed longitudinally on the frame, and a longitudinal camera component is mounted on a slider of the longitudinal linear module; The operation of the needle module includes the following steps: S001: The actions of the acupuncture module include daily operation steps and detection steps; S002: When the acupuncture module performs daily operation steps, the power components D1, D2, ..., Dn respectively drive the needle groups C1, C2, ..., Cn to remain in a raised state, and the auxiliary power component drives the needle box to rise and fall to perform the puncture operation; S003: When the acupuncture module performs the detection step, the auxiliary power component drives the needle box to remain in a raised state, and the power components D1, D2, ..., Dn respectively drive the needle groups C1, C2, ..., Cn to rise and fall regularly in sequence; S004: The detection step is interspersed with the daily operation steps with a detection interval time t0, and the detection interval time t0 is inversely proportional to the grammage of the non-woven fabric; The operation of the visual processing module includes the following steps: S100: When the needle module is performing the detection step, the power component D1 drives the needle group C1 to descend and maintain the descending state to detect the needle group C1; the horizontal camera component moves to the detection position P h中 Perform fixed-point photography to obtain the first horizontal image T h1 , the detection position P h中 Corresponding to the horizontal center position of the needle group C1; S110: The longitudinal camera element moves to the detection position P z1 Perform fixed-point photography to obtain the first vertical picture T z1 , the detection position P z1 Collinear with needle group C1; S120: After the needle group C1 is detected, the power member D1 drives the needle group C1 to rise, and the power member D2 drives the needle group C2 to descend and maintain the descending state, and the needle group C2 is detected; and so on, the fixed-point camera of the needle groups C2 to Cn is completed to obtain the second horizontal image T h2 , ..., nth horizontal picture T hn , get the second vertical picture T z2 , ..., the nth vertical picture T zn ; S130: T h1 、T h2 ,……,T hn and T z1 、T z2 ,……,T zn Perform bending and passivation calibration; S200: In step S100, when the first horizontal picture T h1 After the acquisition is completed; The horizontal camera element moves horizontally at a maximum uniform speed. During the movement, the horizontal camera element takes multiple dynamic pictures to obtain a first horizontal flying picture group of the needle group C1. The first horizontal flying picture group includes multiple pictures T taken at multiple angles along the front of C1. hf1 、T hf2 ,……,T hfn ; S210: In step S110, when the first vertical picture T z1 After the acquisition is completed; The longitudinal camera element moves longitudinally at a maximum uniform speed. During the movement, the longitudinal camera element takes multiple dynamic pictures to obtain a first longitudinal flying picture group of the needle group C1. The first longitudinal flying picture group includes multiple pictures T taken at multiple angles along the side of C1. zf1 、T zf2 ,……,T zfn ; S220: In step S120, the dynamic video recording of the needle groups C2 to Cn is completed, and the second horizontal flying picture group, ..., the nth horizontal flying picture group, the second vertical flying picture group, ..., the nth vertical flying picture group are obtained. S230: Perform eccentricity correction on the second horizontal flying picture group, ..., the nth horizontal flying picture group and the second vertical flying picture group, ..., the nth vertical flying picture group respectively.

2. A processing method for nonwoven fabric production according to claim 1, characterized in that: In step S130, the bending calibration includes the following steps: S131: Obtain the first horizontal image T h1 The image data of the needle parts Z1, Z2, Z3, ..., Zn are used to generate the first horizontal image point cloud model Z1', Z2', Z3', ..., Zn', and obtain the first vertical image T z1 The image data of the needle parts Z1, Z2, Z3, ..., Zn are used to generate the first longitudinal image point cloud model Z 纵 ', set up a vertical unified reference plane T, and divide Z1', Z2', Z3', ..., Zn' and Z 纵 ' respectively placed on a uniform reference plane T, the uniform reference plane T including a horizontal axis x-axis and a vertical axis z-axis; S132: A standard reference model Z of the first horizontal image point cloud model is set on the unified reference plane T 标h , the Z 标h Including Z1 标 、Z2 标 、……、Zn 标 , the Z1', Z2', Z3', ..., Zn' are respectively 标 、Z2 标 、……、Z n标 correspond; Grab the top midpoint coordinates of Z1' and Z1 标 The top midpoint coordinates of Z1' and Z1' are aligned, and the top midpoint coordinates of Z2' and Z2 are compared. 标 The coordinates of the top midpoint are aligned, and Z2' is compared with Z2 标 The matching rate, ..., grab the top midpoint coordinates of Zn' and Zn 标 The coordinates of the top midpoint are aligned, and Zn' and Zn are compared. 标 Match rate; S133: Z1' and Z1 标 , Z2' and Z2 标 , ..., Zn' and Zn 标 When any matching rate approaches the minimum threshold, the expected maintenance task is triggered; Z1' and Z1 标 , Z2' and Z2 标 , ..., Zn' and Zn 标 When any matching rate falls below the minimum threshold, a shutdown and maintenance task is triggered; Complete the first horizontal picture T h1 After the bending proofreading, the second horizontal picture T is completed in the same way. h2 , ..., nth horizontal picture T hn Bend proofreading; S134: A standard reference model Z of the first longitudinal image point cloud model is set on the unified reference plane T. 标z , Z 标z Includes a reference line Z parallel to the z-axis 参 , the Z 参 Align with the top midpoint coordinate of the Z vertical 'of the first vertical image point cloud model, with Z 参 is the axis of symmetry, at Z 参 There are reference lines Z on both sides 左 and reference line Z 右 , the reference line Z 左 and reference line Z 右 The spacing is equal to the diameter of the standard needle piece, with reference line Z 左 and reference line Z 右 As a benchmark, compare Z 纵 'With Z 标 The matching rate of z; S135:Z 纵 'Exceeds the reference line Z 左 or reference line Z 右 When the part approaches the highest threshold, the expected maintenance task is triggered; Z' exceeds the reference line Z 左 or reference line Z 右 When the part exceeds the highest threshold, the shutdown maintenance task is triggered; Complete the first vertical picture T z1 After the bending proofreading, the second vertical picture T is completed by analogy. z2 , ..., the nth vertical picture T zn Bend proofreading.

3. A processing method for nonwoven fabric production according to claim 2, characterized in that: The expected maintenance tasks in steps S133 and S135 include the following steps: S137: Statistics Z1' and Z1 标 , Z2' and Z2 标 , ..., Zn' and Zn 标 The difference in the matching rate is recorded as the horizontal offset, and the part of the Z vertical 'that exceeds the reference line Z left or the reference line Z right is recorded as the vertical offset; and so on, the second horizontal picture T is completed respectively. h2 , ..., nth horizontal picture T hn and the second vertical picture T z2 , ..., the nth vertical picture T zn Offset statistics; S138: Mark the horizontal offset and vertical offset as offset Z 偏 , let the expected maintenance time be t 检 , where Z 偏 and t 检 Satisfies the following corresponding relationship: t 检 =k / λZ 偏 , where λ is the gram weight of the nonwoven fabric and k is the processing constant.

4. A processing method for nonwoven fabric production according to claim 3, characterized in that: The left side horizontal safety line Z is set on both sides of the standard reference model of the first horizontal image point cloud model. 左h and right side horizontal safety line Z 右h The left side of the standard reference model of the first longitudinal image point cloud model is provided with a left longitudinal safety line Z 左z The right side of the standard reference model of the nth longitudinal image point cloud model is provided with a right longitudinal safety line Z 右z , The left side transverse safety line Z 左h The distance from Z1 to Z1' is equal to the distance from the needle piece Z1 to the edge of the needle box. 右h The distance to Zn' is equal to the distance from the needle element Zn to the edge of the needle box; The left longitudinal safety line Z 左z The distance from the point cloud model Zlongitudinal' of the first longitudinal picture is equal to the distance from the needle group C1 to the edge of the needle box, and the right longitudinal safety line Z 右z The distance to the point cloud model Z' of the nth longitudinal image is equal to the distance from the needle group Cn to the edge of the needle box; First horizontal picture T h1 , ..., nth horizontal picture T hn In any horizontal image, the point cloud model exceeds the left horizontal safety line Z 左h and right side horizontal safety line Z 右h If the part is too large, an emergency stop is triggered; In the first vertical image Tzn, ..., the nth horizontal image Tzn, the point cloud model of any vertical image exceeds the left longitudinal safety line Z 左z and right longitudinal safety line Z 右z If the part is exceeded, an emergency stop is triggered.

5. A processing method for nonwoven fabric production according to claim 4, characterized in that: In step S130, the passivation check includes the following steps: S1331: in step S133 and step S134, when the matching rate of any needle member is not less than 95%, perform passivation calibration; S1332: There is a reference line X on the unified reference plane T 基 and reference line X 参 , the X 基 parallel to the x-axis and the X 基 Collinear with the upper vertices of Z1', Z2', Z3', ..., Zn'; The X 参 Parallel to the x-axis, the X 参 Located at X 基 Below, X 参 With X 基 There is a standard height between them, which is equal to the length of a standard needle piece; Place the lower endpoints of Z1', Z2', Z3', ..., Zn' at the same level as X 参 Compare and compare the lower endpoints of Z1', Z2', Z3', ..., Zn' with X 参 The difference between and was recorded as the wear value; S1333: When any wear value among Z1', Z2', Z3', ..., Zn' approaches the maximum threshold, the expected maintenance task is triggered; Complete the first horizontal picture T h1 After the passivation proofreading, the second horizontal picture T is completed in the same way. h2 , ..., nth horizontal picture T hn Passivation proofreading.

6. A processing method for nonwoven fabric production according to claim 5, characterized in that: In step S230, the eccentricity calibration includes the following steps: S231: In step S133 and step S134, when the matching rate of any needle member is not less than 95%, and in step S1332, when the wear value of any needle member is less than 5%, eccentricity correction is performed; Obtain image data of multiple images Thf1, Thf2, ..., Thfn in the first horizontal flying image group, and generate point cloud models Thf1', Thf2', ..., Thfn' respectively; obtain image data of multiple images Tzf1, Tzf2, ..., Tzfn in the first vertical flying image group, and generate point cloud models Tzf1', Tzf2', ..., Tzfn' respectively; and place Thf1', Thf2', ..., Thfn' and Tzf1', Tzf2', ..., Tzfn' respectively on a unified reference plane T; S232: A unified standard reference model Z of the first horizontal flying picture group point cloud model and the first vertical flying picture group point cloud model is provided on the unified reference plane T. 标f , the point cloud models of individual needle pieces in Thf1', Thf2', ..., Thfn' and Tzf1', Tzf2', ..., Tzfn' are compared with the unified standard reference model Z 标f Perform comparison and matching, and record the matching rate; S233: When any matching rate among Thf1', Thf2', ..., Thfn' and Tzf1', Tzf2', ..., Tzfn' is close to the minimum threshold, the expected maintenance task is triggered; After completing the eccentricity correction of the first horizontal flying picture group and the first vertical flying picture group, the eccentricity correction of the second horizontal flying picture group, ..., the nth horizontal flying picture group and the second vertical flying picture group, ..., the nth vertical flying picture group is completed respectively.

7. A processing method for nonwoven fabric production according to claim 6, characterized in that: The steps S133, S135, S1333 and S233 respectively include the following dust collection operations: The dust collection operation comprises the following steps: When the matching rate or the wear value approaches the threshold, the current step is repeated after performing several vacuuming operations.

8. The processing method for nonwoven fabric production according to claim 2, characterized in that: Steps S133 and S134 include the broken needle detection The broken needle detection comprises the following steps: In steps S133 and S134, when the matching rate is below 20%, subsequent equipment is synchronously started to remove broken needles from the non-woven fabric blank in a targeted manner.

9. A processing equipment for non-woven fabric production, characterized in that, The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processing method according to any one of claims 1 to 8 is implemented when the processor executes the computer program.

10. A nonwoven fabric, characterized in that: The non-woven fabric is produced by the processing method according to any one of claims 1 to 8, wherein the non-woven fabric comprises a first spunbond layer, a first staple fiber PET skeleton layer, a meltblown layer, a second staple fiber PET skeleton layer, and a second spunbond layer arranged in order from top to bottom; The nonwoven fabric includes a plurality of identical product units from left to right; In the single product unit, the fiber density of the first staple fiber PET skeleton layer and the second staple fiber PET skeleton layer gradually increases from both sides to the middle, reaches a peak at 1 / 4, then decreases toward the middle to a trough at 3 / 8, and then evenly transitions to the middle position.

Images