A forming monitoring system and method for a galvanized steel strand

Abstract

This invention discloses a forming monitoring system and method for galvanized steel strand, relating to the field of processing monitoring technology. Based on a process baseline map, the system identifies and binds process response characteristics with the current production cycle time, delineates the quality manifold boundary, and uses the quality manifold boundary contour to separate process fluctuations within the acceptable tolerance zone from abnormal pull-out damping abrupt changes. Abnormal abrupt change signals are extracted, and for these signals, the abrupt change rates of the transient yaw tension of a single filament, the integral of acoustic emission energy in the die zone, and the spindle drag torque are mathematically normalized to construct an asymmetric stranding distortion index. This index is used to assess the degradation of electromechanical-acoustic combined strand forming scenarios. If an out-of-limit trend leading to wire breakage and shutdown is detected, an emergency stop command is issued to the PLC. This monitoring method effectively improves the quality stability and intelligent level of galvanized steel strand production.

Description

Technical Field

[0001] This invention relates to the field of processing monitoring technology, and specifically to a forming monitoring system and method for galvanized steel strand. Background Technology

[0002] Galvanized steel strand, used as a load-bearing component in power plants, bridges, and geotechnical anchoring applications, requires processes such as high-carbon steel wire rod drawing, hot-dip galvanizing, and tube stranding. Existing monitoring methods cannot correlate multi-dimensional data such as stranding trajectory, acoustic emission, and drag torque, making it difficult to distinguish between normal fluctuations and genuine anomalies, resulting in high false negative and missed detection rates and a lack of proactive early warning. While machine vision solutions can identify surface defects, they lack in-depth correlation with the internal mechanical evolution and process parameter matching during the stranding process, making it impossible to deduce the forming quality from a mechanistic perspective and intervene before wire breakage. Therefore, there is an urgent need for a monitoring solution that covers the entire stranding process, integrates multi-source parameters, and has predictive interruption capabilities to address the pain points of traditional monitoring methods, such as lag, missed detections, and high safety risks. Summary of the Invention

[0003] The purpose of this invention is to provide a forming monitoring system and method for galvanized steel strand to address the shortcomings in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring the forming of galvanized steel strand, the monitoring method comprising the following steps: S1: During the start-up and steady-state operation phases, the sensing layer continuously captures multimodal operation data. Combined with the improved flexible body spatial spiral twisting algorithm and trajectory compensation mechanism, the spatial twisting evolution curve during the galvanized steel wire convergence and forming process is deduced to generate a process baseline map. S2: Based on the process baseline map, the process response characteristics are bound to the current production cycle time for identification, and the quality manifold boundary is delineated; S3: Using the mass manifold boundary profile, process fluctuations within the acceptable tolerance zone are separated from abnormal pull-out damping abrupt changes, and abnormal abrupt change signals are extracted; S4: For abnormal abrupt change signals, the abrupt change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the pressing zone, and the main shaft driving torque are mathematically normalized to construct an asymmetric twisting distortion index. S5: Degradation judgment is performed on the electromechanical and acoustic combined yarn forming scenario. If an over-limit trend that would cause yarn breakage and shutdown is detected, an emergency stop interruption command is sent to the electrical control PLC.

[0005] Preferably, the step of improving the flexible body spatial helical twisting algorithm in step S1 includes: Based on the differential geometric modeling logic of the spatial helical structure of steel wire rope, multiple vectors of the center line of the steel wire are defined with the Frenet frame as the reference. The multiple parameters of the steel wire are mapped to the rotation angle and pitch variables of the spatial curve. A parameterized description of the spatial curve of a single galvanized steel wire in the stranding process is established, and the three-dimensional spatial pose of the steel wire in the multi-layer stranding process is deduced. Based on the microstructure evolution law of the entire process of high carbon galvanized steel wire drawing-hot galvanizing, the parameters of sorbite lamellar spacing, cementite fragmentation degree, and dislocation density are mapped to the equivalent elastic modulus and yield limit of the steel wire. When the drawing stress is close to the dynamic yield threshold of the material, the curvature calculation value of the space curve is positively compensated to correct the trajectory deviation of the steel wire caused by plastic deformation. Based on the test results of friction characteristics between galvanized steel wires, the zinc layer thickness, surface roughness, and contact stress are mapped to dynamic friction coefficients. For the contact slip scenarios between the steel wire and the die, and between the steel wires, negative compensation is made to the calculated tangential displacement value of the space curve to counteract the trajectory lag caused by friction resistance. The two types of compensation quantities are weighted and fused together and then superimposed onto the original spatial curve calculation results to deduce the entire process spatial twisting evolution curve of a single galvanized steel wire from laying, entering the mold to twisting and forming.

[0006] Preferably, the step of generating the process reference base map in step S1 includes: Based on the spatial twisting evolution curve obtained by deduction, the cycle time parameters of the current production process are synchronously associated, the process parameters are bound to the time nodes of the evolution curve, and the process response characteristic range covering the qualified tolerance zone is defined. Integrated anti-fragmentation and wire breakage safety prediction logic: When the interlayer relative displacement of adjacent wires in the evolution curve exceeds the critical threshold of delamination slip, it is marked as a strand risk area; When the equivalent torsional stress of a single filament exceeds the torsional fatigue threshold under the current dislocation density, it is marked as a filament breakage risk area.

[0007] Preferably, in step S1, the sensing layer captures multimodal operating data, including mechanical oscillation signals, wire pulling stress, and transient slip acoustic signatures.

[0008] Preferably, in step S2, the process response characteristics are bound to the current production cycle time based on the process baseline map, and the quality manifold boundary is delineated by the following steps: For the adaptive adjustment of the entire stranding process of galvanized steel strand, benchmark ranges of process parameters are established for the corresponding cycle time: In the acceleration and start-up phase, the spindle speed ramp rate and the wire tension gradient rise curve are used as binding parameters. In the constant speed twisting phase, the average single filament tension under stable twist pitch and the friction temperature rise rate in the die zone are used as binding benchmarks. In the transition phase, the focus is on binding the drawing resistance change threshold corresponding to the die hole wear amount. In the tension adaptive adjustment phase, the output response delay and steady-state error range of the tension closed-loop controller are associated. The time nodes of the spatial twisting evolution curve derived from the process baseline map are matched with the real-time acquired process cycle state signals. When the cycle state changes, the corresponding process response feature subset is automatically called. After dividing the K-nearest neighbor region based on manifold distance, the global feature matrix and the local feature matrix are calculated separately. The fusion adjustment parameter of the global and local features is determined by the feature matrix spectral radius analysis method. The two types of features are fused to obtain the dimension reduction matrix, which is then projected onto the boundary of the mass manifold.

[0009] Preferably, in step S3, by using the mass manifold boundary profile, process fluctuations within the acceptable tolerance zone are separated from abnormal pull-out damping abrupt changes, and abnormal abrupt change signals are extracted, including the following steps: By comparing the characteristic fingerprint of the mutation signal with the patterns of the three types of anomalies, the process fluctuations and abnormal pull-out damping mutations are completely separated, and the abnormal mutation signals corresponding to the defect cause characteristics are extracted.

[0010] Preferably, in step S4, for abnormal abrupt change signals, the abrupt change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the molding zone, and the abrupt change rate of the main shaft driving torque is mathematically normalized to construct the asymmetric twisting distortion index. The theoretical tensile reference value is calculated based on the ratio of stranding pitch to nominal wire diameter. The deviation between the real-time collected yaw tensile force and the theoretical reference value is used as the original input. The deviation is compressed to the range of 0 to 1 through piecewise linear mapping logic. When the deviation is lower than the lower limit of the qualified tolerance zone, it is mapped to 0. When it is higher than the upper limit of the abnormal threshold, it is mapped to 1. The intermediate range is linearly scaled according to the deviation ratio. The acoustic emission energy integral of the molding area is bandpass filtered according to the abnormal characteristic frequency band, and the energy accumulation value within the time window corresponding to the sudden change signal is extracted. Based on the energy baseline under the mold wear state, the energy is normalized and the logarithm of the ratio of the current energy integral value to the baseline is mapped to the interval of 0 to 1. The spindle drive torque mutation rate is calculated by the slope of torque change within adjacent sampling periods, and then the ratio is calculated with the rated torque change rate threshold at the current production speed. The ratio is then mapped to the 0 to 1 interval through an S-shaped saturation function. The weighted normalized values ​​of the three categories are summed to obtain the asymmetric twisting distortion index.

[0011] Preferably, the degradation judgment step in step S5 includes: Establish a dynamic threshold system for exceeding limits, with the threshold dynamically adjusted according to the production cycle time; The degradation judgment is performed using a sliding time window logic. The distortion index is collected continuously for 5 sampling periods. If the average index value within the window exceeds the dynamic over-limit threshold corresponding to the current process, and the index rise slope exceeds the preset wire breakage precursor slope threshold, it is determined that there is an over-limit trend that will cause wire breakage and shutdown.

[0012] Preferably, the emergency stop and blocking steps in step S5 include: After confirming a tendency for the system to stop beyond its limits due to wire breakage, a priority judgment is performed. If the current system is in an emergency stop protection state, the command will not be issued again. If it is in a normal operating state, the three-level blocking logic is triggered. The first level is torque limiting, which reduces the upper limit of the spindle drive torque output to 30% of the rated value. The second stage is tension release, where the tension controller of the wire feeding frame gradually reduces the tension of the single filament to zero. The third stage is the main spindle braking, which sends a brake trigger signal to the PLC. The main spindle braking deceleration is set at 80% of the yield strength of the steel wire.

[0013] A forming monitoring system for galvanized steel strand, comprising: The simulation module continuously captures multimodal operation data during the start-up and steady-state operation phases. Combining the improved flexible body spatial spiral twisting algorithm and trajectory compensation mechanism, it simulates the spatial twisting evolution curve during the galvanized steel wire convergence and forming process, and generates a process baseline map. Signal extraction module: Based on the process baseline map, the process response features are bound and identified with the current production cycle time, the quality manifold boundary is defined, and the process fluctuations within the acceptable tolerance zone and abnormal pull-out damping abrupt changes are separated by the quality manifold boundary contour, and abnormal abrupt change signals are extracted. Command issuance module: For abnormal sudden change signals, the sudden change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the pressing zone and the main shaft driving torque is mathematically normalized to construct an asymmetric stranding distortion index, and the degradation judgment is made for the electromechanical-acoustic combined yarn forming scenario. If an over-limit trend that causes yarn breakage and shutdown is detected, an emergency stop interruption command is issued to the electrical control PLC.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: By continuously capturing multimodal operation data during the start-up and steady-state phases, and combining the improved flexible body spatial spiral twisting algorithm and trajectory compensation mechanism to deduce the spatial twisting evolution curve and generate the process baseline map, and then binding the process response characteristics with the production process cycle time to delineate the quality manifold boundary, normal process fluctuations within the qualified tolerance zone can be accurately separated, and abnormal pull-out damping mutation signals can be extracted separately, avoiding the problem of traditional monitoring misjudging normal fluctuations as abnormalities.

[0015] Asymmetric stranding distortion index was constructed by mathematically normalizing the transient yaw tension of a single filament, the integral of acoustic emission energy in the die pressing area, and the mutation rate of the main shaft driving torque. This enabled the quantitative fusion evaluation of multi-dimensional electromechanical and acoustic parameters, solving the problem of high missed detection rate in single parameter monitoring. Ultimately, it was used to make deterioration judgments in the electromechanical and acoustic combined stranding scenario. When the trend of wire breakage and shutdown exceeding the limit was detected, an emergency stop command was directly issued to the PLC, replacing the traditional post-processing mode that relied on human experience. Production was stopped in time before the defect expanded, effectively improving the quality stability and intelligent level of galvanized steel strand production. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the monitoring method of the present invention. Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example: This example provides a forming monitoring system and method for galvanized steel strand. Please refer to [link / reference]. Figure 1 As shown, the monitoring method includes the following steps: By importing the equipment's digital model into the on-site 3D contour mapping, the geometric topological relationships of stations such as the wire feeding frame, wire distribution plate, deformation pressing die, and traction winch are restored, and the basic force transmission equations between each level of guide wheel and steel wire are established. Along the steel wire routing path, composite sensing nodes (covering dynamic tension, acoustic emission, and high-frequency torque sensors) are distributed and implanted in the tension adjustment swing arm, the strand confluence die, and the track traction section.

[0020] During the start-up and steady-state operation phases, the sensing layer continuously captures multimodal mechanical oscillations, wire pull-out stress, and transient slippage acoustic signatures, among other underlying operational data. Combined with an improved flexible body spatial helical twisting algorithm, a trajectory compensation mechanism based on material yield characteristics and zinc layer friction coefficient is introduced to deduce the spatial twisting evolution curve of a single galvanized steel wire during the convergence and forming process, generating a predictive baseline map for preventing strand breakage and wire detachment in the safety process.

[0021] The sensing layer needs to continuously capture multimodal underlying operational data, covering three dimensions: mechanical oscillation signals, wire drawing stress time-series data, and transient slippage acoustic signature characteristics. Among them, mechanical oscillation signals need to be collected by high-frequency accelerometers arranged on the main shaft of the wire stranding machine, the die pressing area, and the pay-off frame, with a sampling frequency of not less than 10kHz to cover the high-frequency vibration components during the wire stranding process; wire drawing stress needs to be collected in real time by strain gauge force sensors embedded in the pay-off tension wheel and traction roller, synchronously correlated with the theoretical tension threshold at the current production speed; transient slippage acoustic signatures need to be collected by piezoelectric acoustic emission sensors to collect high-frequency acoustic signals of friction between the wire and the die in the die pressing area and slippage between single wire layers.

[0022] The improved steps of the flexible body spatial helical twisting algorithm are as follows: Based on the differential geometric modeling logic of the spatial helical structure of steel wire rope, a parameterized description of the spatial curve of a single galvanized steel wire during the stranding process is established. Using the Frenet frame as a reference, the tangent vector, normal vector and binormal vector of the steel wire centerline are defined. The twist pitch, helical radius and twist direction parameters of the steel wire are mapped to the rotation angle and pitch variables of the spatial curve to describe the three-dimensional spatial pose of the steel wire during the multi-layer stranding process.

[0023] Based on this, a trajectory compensation mechanism is introduced, and the compensation logic consists of two parts: First, material yield characteristics compensation: based on the microstructure evolution law of the entire process of high carbon galvanized steel wire drawing-hot galvanizing, parameters such as sorbite lamellar spacing, cementite fragmentation degree, dislocation density, etc. are mapped to the equivalent elastic modulus and yield limit of the steel wire. When the drawing stress is close to the dynamic yield threshold of the material, the curvature calculation value of the space curve is positively compensated to correct the trajectory deviation of the steel wire caused by plastic deformation. Secondly, zinc layer friction coefficient compensation is implemented. Based on the test results of friction characteristics between galvanized steel wires, the zinc layer thickness, surface roughness, and contact stress are mapped to a dynamic friction coefficient. For contact slippage scenarios between the steel wire and the die, and between interlayer steel wires, negative compensation is applied to the calculated tangential displacement value of the spatial curve to offset the trajectory lag caused by frictional resistance. The two types of compensation are weighted and fused together and then superimposed onto the original spatial curve calculation results. Finally, the entire spatial twisting evolution curve of a single galvanized steel wire from unwinding, die insertion to twisting is derived.

[0024] When generating a predictive process baseline map based on the aforementioned spatial twisting evolution curve, the cycle time parameters of the current production process are simultaneously associated. Process parameters such as spindle speed, traction speed, and wire tension setting are bound to the time nodes of the evolution curve. A process response characteristic interval covering the qualified tolerance zone is defined. This interval needs to be set in conjunction with the influence law of the wire rope lay multiple on the contact angle and curvature, and the curvature fluctuation threshold and the allowable range of contact angle under different structures of co-twist and interlocking twists are clarified.

[0025] The base chart integrates safety prediction logic for both preventing strand breakage and preventing thread breakage: The anti-stripping prediction is based on the layered slip theory of the joint working mode between the steel wires of the cable stay. When the interlayer relative displacement of adjacent steel wires in the evolution curve exceeds the critical threshold of layered slip, it is marked as a stripping risk area. The wire breakage prediction is based on the correspondence between the torsional properties and microstructure of high carbon galvanized steel wire. When the equivalent torsional stress of a single wire exceeds the torsional fatigue threshold under the current dislocation density, it is marked as a wire breakage risk area. The final process reference map includes both the spatial trajectory reference of steel wire forming and the prediction boundary of multi-dimensional safety risks.

[0026] In a preferred embodiment, the parameterized description of the space curve uses the arc length parameter s of the wire centerline as the independent variable, the x-coordinate = helix radius r × cosine (2πs ÷ twist p), the y-coordinate = helix radius r × sine (2πs ÷ twist p), and the z-coordinate = pitch h × s ÷ twist p. A 1×19 galvanized steel strand with a nominal diameter of 5mm is used. The helix radius r is 2.3mm, and the twist pitch p is 12 times the single wire diameter, i.e., 60mm. The twist direction parameter ω is +1 under the same twist structure and -1 under the alternating twist structure. The relationship between the pitch h and the twist direction parameter is h=ω×twist p×tangent twist angle β. When the twist angle β is 6°, h is calculated to be 6.31mm. Substituting the arc length parameter s=100mm (corresponding to a 100mm long steel wire segment after laying out), the coordinates of the point are calculated to be (2.3×cos(10.47°), 2.3×sin(10.47°), 6.31×100÷60), i.e. (2.26, 0.42, 10.52)mm, thus completing the quantitative description of the single wire spatial pose.

[0027] The expression for calculating the equivalent elastic modulus E_eq is: E_eq = Elastic modulus of annealed high carbon steel E_0 × (1 - 0.12 × (Sorbite lamellar spacing d - 0.15μm) ÷ 0.1) × (1 + 0.08 × Cementite fragmentation degree η) × (1 - 0.05 × Natural logarithm (Dislocation density ρ ÷ 10^10 m)) -2 )).

[0028] The expression for calculating the yield strength σ_s is: σ_s = annealed yield strength σ_s0 × (1 + 0.15 × (0.2μm - d) ÷ 0.1) × (1 + 0.1 × η). Taking E_0 as 206 GPa, σ_s0 as 420 MPa, and after hot-dip galvanizing, the measured sorbite lamellar spacing d = 0.18 μm, cementite fragmentation degree η = 82%, and dislocation density ρ = 3 × 10^10 m. -2The calculation yields E_eq = 206 × (1 - 0.12 × 0.3) × 1.0656 × (1 - 0.05 × ln3) ≈ 199 GPa, and σ_s = 420 × (1 + 0.15 × 0.2) × 1.082 ≈ 468 MPa.

[0029] When the pull-out stress σ approaches the material's dynamic yield threshold of 0.8σ_s, i.e., 374 MPa, positive compensation is applied to the calculated curvature value κ of the space curve. The expression for the compensation amount Δκ_1 is as follows: Δκ_1 = (σ - 0.8σ_s) ÷ (E_eq × 10³ × helix radius r) × twist parameter ω. Substituting σ = 380MPa, we get Δκ_1 = (380 - 374) / (199 × 10³ × 2.3) × 1 ≈ 1.31 × 10⁻⁵ mm. -1 The original curvature κ_0 = 1 / r = 0.435 mm -1 After correction, the curvature κ = κ_0 + Δκ_1 = 0.4350131 mm-1, thus completing the trajectory offset correction caused by plastic deformation.

[0030] The expression for calculating the dynamic friction coefficient μ_d is: μ_d = smooth steel-steel reference friction coefficient μ_0 × (1 + 0.2 × (zinc layer thickness t - 15μm) ÷ 10) × (1 + 0.1 × natural logarithm (surface roughness Ra ÷ 0.8μm)) × (1 + 0.05 × (contact stress σ_c - 50MPa) ÷ 20). Taking μ_0 as 0.15, the measured zinc layer thickness t = 20μm, surface roughness Ra = 1.2μm, and contact stress σ_c in the molding area = 65MPa, we calculate μ_d = 0.15 × (1 + 0.2 × 0.5) × (1 + 0.1 × ln1.5) × (1 + 0.05 × 0.75) ≈ 0.173.

[0031] For scenarios involving contact slippage between the steel wire and the die, and between interlayer steel wires, negative compensation is applied to the calculated tangential displacement Δs of the spatial curve. The compensation amount Δs_c is calculated as follows: Δs_c = (μ_d × normal contact force in the die area F_N × contact length between the steel wire and the die L) ÷ (E_eq × 103 × cross-sectional area of ​​a single wire A). Substituting F_N = 120N, L = 15mm, and A = 19.6mm² (corresponding to a single wire with a diameter of 5mm), we calculate Δs_c = (0.173 × 120 × 15) / (199 × 103 × 19.6) ≈ 7.98 × 10⁻⁵ mm. The original tangential displacement Δs_0 = 0.12mm, and the corrected Δs = Δs_0 - Δs_c = 0.11992mm, thus offsetting the trajectory lag caused by frictional resistance.

[0032] The two types of compensation quantities are weighted and fused together and then superimposed onto the original spatial curve calculation results. Under the same twist structure, the material yield compensation weight is 0.6 and the zinc layer friction compensation weight is 0.4; under the alternating twist structure, the material yield compensation weight is 0.4 and the zinc layer friction compensation weight is 0.6. The calculation expression of the fused coordinate r_comp(s) is: r_comp(s) = original coordinate r(s) + weight 1 × Δκ_1 × normal vector n(s) × s squared + weight 2 × Δs_c × tangent vector τ(s).

[0033] Substituting the structural parameters of the same-direction twist structure, w1=0.6, w2=0.4, s=100mm, the normal vector n(s)=(-sin(10.47°), cos(10.47°), 0)=(-0.182, 0.983, 0), and the tangent vector τ(s)=(-2.3×2π / p sin(10.47°), 2.3×2π / p cos(10.47°), h / p)=(-0.0024, 0.013, 0.105), we calculate the x-component: 2.26 + 0.6×1.31×10⁻⁵×(-0.182)×10000 ≈2.246mm, and the y-component: 0.42 + 0.6×1.31×10⁻⁵×0.983×10000 ≈0.497mm, z-component: 10.52 + 0.6×1.31×10⁻⁵×0×10000 + 0.4×7.98×10⁻⁵×0.105×100 ≈10.523mm. Finally, the spatial twisting evolution curve of a single galvanized steel wire from unwinding, entering the mold to twisting and forming is derived.

[0034] When generating a predictive process baseline map based on the aforementioned spatial twisting evolution curve, the cycle time parameters of the current production process are synchronously associated. Process parameters such as spindle speed, traction speed, and wire tension setpoints are bound to the timing nodes of the evolution curve. The calculation expression for the timing node t_node is as follows: t_node = arc length node s_node ÷ traction speed v + corner node θ_node ÷ (2π × spindle speed n). Taking the steady-state condition of spindle speed n = 120 r / min and traction speed v = 60 m / min, s_node = 100 mm = 0.1 m, θ_node = 10.47° = 0.1827 rad, we calculate t_node = 0.1 ÷ 60 + 0.1827 ÷ (2π × 120) ≈ 0.00191 min = 0.1146 s, thus completing the time sequence binding of the process cycle and evolution curve. The process response characteristic range covering the qualified tolerance zone is defined. This range is set in combination with the influence law of wire rope lay multiple on contact angle and curvature. Under the same-direction lay structure, the curvature fluctuation threshold is ±0.02 mm-1, and the allowable range of contact angle is 5.8°~6.2°; under the alternating lay structure, the curvature fluctuation threshold is ±0.015 mm-1, and the allowable range of contact angle is 5.9°~6.1°.

[0035] The critical threshold Δd_c is calculated as follows: Δd_c = 0.1 × helix radius r × (1 - dynamic friction coefficient μ_d ÷ 0.2). Substituting r = 2.3 mm and μ_d = 0.173, we get Δd_c = 0.1 × 2.3 × (1 - 0.865) = 0.031 mm. When the measured interlayer relative displacement Δd = 0.035 mm, it exceeds the critical threshold and is marked as a strand breakage risk area. Wire breakage prediction is based on the correspondence between the torsional properties and microstructure of high-carbon galvanized steel wire. When the equivalent torsional stress of a single wire exceeds the torsional fatigue threshold under the current dislocation density, it is marked as a wire breakage risk area. The torsional fatigue threshold τ_f is calculated as follows: τ_f = σ_s ÷ √3 × (1 - 0.1 × natural logarithm (ρ ÷ 10^10)). Substituting σ_s=468MPa and ρ=3×10^10m-2, we calculate τ_f=468÷1.732×(1-0.1×ln3)≈242MPa. When the equivalent torsional stress of a single filament τ_eq=250MPa, it exceeds the fatigue threshold and is marked as a filament breakage risk area.

[0036] Based on the process baseline map, the process response characteristics are linked to the current production cycle time (such as acceleration start-up, constant speed stranding, die transition, and adaptive tension adjustment) to identify and adaptively define the mass manifold boundary under the stranding parameters. Using the mass manifold boundary profile, inherent mechanical vibrations of the equipment and elastic pulsation allowances of the wire are filtered out. Normal process fluctuations within the acceptable tolerance zone are separated from abnormal pull-out damping abrupt changes caused by die wear, single-wire tension imbalance, or excessive zinc layer extrusion and scraping, thus extracting abnormal abrupt change signals.

[0037] For the various operating conditions in the entire process of galvanized steel strand stranding, including accelerated start-up, constant speed stranding, die transition, and adaptive tension adjustment, benchmark ranges of process parameters under corresponding cycle times are established: In the acceleration and start-up phase, the spindle speed ramp rate and the wire tension gradient rise curve are used as binding parameters. In the constant speed stranding phase, the average tension of the single filament under stable twist pitch and the friction temperature rise rate in the die zone are used as binding benchmarks. In the die transition phase, the focus is on binding the threshold of the drawing resistance change corresponding to the die hole wear amount. In the tension adaptive adjustment phase, the output response delay and steady-state error range of the tension closed-loop controller are associated.

[0038] The binding process adopts a timestamp alignment mechanism, which matches each time node of the spatial twisting evolution curve derived from the process baseline map with the real-time acquired process cycle status signal. When the cycle status changes, the corresponding process response feature subset is automatically called to avoid parameter confusion under different working conditions and ensure the adaptability of working conditions in subsequent boundary delineation.

[0039] A computational framework based on manifold learning is constructed to adaptively define the mass manifold boundary under the twisting parameters: Using multi-dimensional parameters of the process response characteristics (including single-filament yaw tension, acoustic emission energy, spindle torque, stranding curvature, etc.) as input, a high-dimensional feature space set of data points is constructed. The manifold distance between data points is calculated by preserving the local neighborhood structure, replacing the traditional Euclidean distance to adapt to the distribution characteristics of the nonlinear parameters in the stranding process. The expression for calculating the manifold distance is: Data points are mapped to vertices in a graph structure. Adjacent vertices are connected to form an edge set. All paths connecting two data points are traversed. The Euclidean distances between adjacent vertices on the path are accumulated, and a scaling factor is introduced to correct nonlinear deviations. The minimum accumulated value among all paths is taken as the final manifold distance.

[0040] After dividing the K-nearest neighbor regions based on manifold distance, the global feature matrix and local feature matrix are calculated separately: the global feature matrix is ​​obtained by seeking the low-dimensional representation with the largest data variance, corresponding to the principal component analysis logic, to maximize the variance consistency between the dimensionality-reduced dataset and the original dataset; the local feature matrix is ​​obtained by minimizing the similarity deviation of data points within the neighborhood. When constructing the similarity matrix, the manifold distance is used as the independent variable, and the Gaussian kernel function is used to map the similarity coefficient. The fusion adjustment parameters of global and local features are determined by the feature matrix spectral radius analysis method. After fusing the two types of features, the dimensionality-reduced matrix is ​​obtained and projected onto the quality manifold boundary of the high-dimensional feature space. This boundary is the threshold range for distinguishing between qualified and abnormal.

[0041] By utilizing the quality manifold boundary profile to filter out inherent disturbances and elastic pulsations in the equipment, process fluctuations within the acceptable tolerance band are identified. The inherent mechanical vibration of the equipment is determined by vibration spectrum analysis of the spindle, die area, and pay-off frame. The steady-state vibration component in the 10Hz to 200Hz frequency band is extracted as the baseline vibration threshold. The elastic pulsation margin of the steel wire is calculated based on the stress-strain hysteresis characteristics of high-carbon galvanized steel wire. The pulsation amplitude range caused by elastic recovery is mapped by drawing speed and tension setpoint. The two types of interference signals are separated from the original signal by band-stop filtering and pulsation compensation algorithm, and only the residual signal that exceeds the mass manifold boundary contour is retained.

[0042] When identifying anomalous changes in the residual signal, it is necessary to distinguish the characteristic differences caused by different factors: Abnormal pull-out damping abrupt changes caused by die wear manifest as a stepwise continuous increase in damping value, accompanied by a slow increase in the energy integral of acoustic emission energy in the 100kHz~300kHz frequency band; abrupt changes caused by single-wire tension imbalance manifest as pulse-like jumps in the transient yaw tension of the single wire, with the tension difference between adjacent wires exceeding the 5% process allowable threshold; abrupt changes caused by excessive extrusion and scraping of the zinc layer are accompanied by high-frequency spikes in the acoustic emission signal of the die area, while the spindle drive torque shows a millisecond-level abrupt increase in abrupt change rate.

[0043] By comparing the characteristic fingerprints of the mutation signals with the patterns of the three types of anomalies, the conventional process fluctuations and abnormal pull-out damping mutations are completely separated. The extracted abnormal mutation signals have been freed from all interference components, retaining only the characteristics of the corresponding defect causes.

[0044] In a preferred embodiment, the galvanized steel wire used has a diameter of 4.57 mm and a lay length of 415 mm, which meets the lay length requirements of 1×37 structural steel strand. When binding and identifying the process response characteristics with the current production cycle based on the process baseline map, for the twisting scenario of galvanized steel strand with a nominal diameter of 32 mm, the time taken for the spindle speed to climb from 0 r / min to 120 r / min in the acceleration start-up phase is 5 s, and the climbing rate calculation expression is Δn / Δt=120 / 5=24 r / min / s. The time taken for the wire tension to gradually increase from 0 N to 500 N is 5 s, and the slope of the gradient increase curve is 500 / 5=100 N / s. These two parameters serve as the binding baseline for the acceleration start-up phase.

[0045] In the constant-speed twisting stage, the twisting speed was 45 m / min and the stable twist pitch was 415 mm. The average tension of the single filament was 500 N. The frictional temperature rise rate in the die pressing area was measured to be 0.8 ℃ / s using an infrared thermal imager, which served as the binding benchmark for this stage. In the die transition stage, for tungsten steel dies with a hole wear of 0.05 mm, the corresponding threshold for the change in pull-out resistance was measured to be an increase of 18 N through a pull-out test, which served as the core binding parameter for this stage.

[0046] During the tension adaptive adjustment phase, the output response delay of the associated tension closed-loop controller is 12ms, and the steady-state error range is ±8N, completing the binding parameter configuration for the corresponding operating condition. The binding process adopts a timestamp alignment mechanism, matching each timing node of the spatial twisting evolution curve derived from the process baseline map with the real-time acquired process cycle state signal. The time deviation compensation value is 2ms. When the cycle state changes, the corresponding process response feature subset is automatically called.

[0047] The computational framework based on manifold learning adaptively delineates the mass manifold boundary under stranding parameters. It constructs a high-dimensional feature space set of data points using multi-dimensional parameters of the process response characteristics as input. The four-dimensional feature vector of the i-th data point is taken as X_i=(F_i, E_i, T_i, κ_i), where F_i is the yaw tension of the single filament (N); E_i is the acoustic emission energy integral (mV·s); T_i is the principal shaft torque (N·m); and κ_i is the stranding curvature (mm). -1 .

[0048] The manifold distance between data points is calculated by preserving the local neighborhood structure, replacing the traditional Euclidean distance to adapt to the distribution characteristics of nonlinear parameters in the twisting process. The calculation expression of the manifold distance is as follows: map the data points to vertices in the graph structure, connect adjacent vertices to form an edge set, traverse all paths connecting two data points, accumulate the Euclidean distances of adjacent vertices on the path, introduce a scaling factor to correct the nonlinear deviation, and take the minimum accumulated value among all paths as the final manifold distance.

[0049] The Euclidean distance is calculated as d_euc(X_i, X_j)=√[(F_i-F_j)²+(E_i-E_j)²+(T_i-T_j)²+(κ_i-κ_j)²], with a scaling factor of α=1.12, corresponding to the nonlinear deviation coefficient of the twisting process. The manifold distance is calculated as d_man(X_i, X_j)=min(∑{k=1} n α·d_euc(X_k,X{k+1})), where n is the number of vertices on the path. Taking two adjacent data points actually collected, X_1=(520,1200,85,0.22) and X_2=(535,1350,88,0.23), we can calculate the Euclidean distance d_euc=√[(520-535)²+(1200-1350)²+(85-88)²+(0.22-0.23)²]=√(225+22500+9+0.0001)=√22734.0001≈150.78, and the manifold distance d_man=1.12×150.78≈169.07.

[0050] Based on the manifold distance, K-nearest neighbor regions are divided, and K=5 is selected. The global feature matrix and local feature matrix are calculated respectively. The global feature matrix is ​​obtained by seeking the low-dimensional representation with the largest data variance. Corresponding to the principal component analysis logic, the variance consistency between the dimensionality-reduced dataset and the original dataset is maximized. The contribution rate of the first principal component is 82%, which covers most of the variance information.

[0051] The local feature matrix is ​​obtained by minimizing the similarity deviation of data points in the neighborhood. When constructing the similarity matrix, the manifold distance is used as the independent variable, and the Gaussian kernel function is used to map the similarity coefficient. The similarity calculation expression is S_ij=exp(-d_man(X_i,X_j)² / (2σ²)), taking σ=150, and substituting the manifold distance between X_1 and X_2, we get S_12=exp(-169.07² / (2×150²))=exp(-28584.66 / 45000)=exp(-0.6352)≈0.529.

[0052] The fusion adjustment parameter λ=0.65 for global and local features was determined using the characteristic matrix spectral radius analysis method. After fusing the two types of features, a dimension-reduced matrix was obtained and projected onto the mass manifold boundary of the high-dimensional feature space. The threshold range of this boundary was determined as follows: single-filament yaw tension 480–520 N, acoustic emission energy integral 1000–1400 mV·s, spindle torque 80–90 N·m, and twisting curvature 0.20–0.24 mm. -1 .

[0053] When filtering out the inherent mechanical vibration of the equipment and the elastic pulsation margin of the steel wire using the mass manifold boundary profile, the inherent mechanical vibration of the equipment is determined by the vibration spectrum analysis of the spindle, the die area, and the pay-off frame. The steady-state vibration component in the frequency band of 10Hz to 200Hz is extracted as the baseline vibration threshold, which is set to 3.0mm / s. The effective value of the measured spindle vibration velocity is 2.7mm / s, which is within the baseline threshold range and is determined to be inherent mechanical vibration. It is then removed from the original signal by a band-stop filtering algorithm.

[0054] The elastic pulsation margin of the steel wire is calculated based on the stress-strain hysteresis characteristics of high-carbon galvanized steel wire. The calculation formula is ΔT=0.024×v×T0, where v is the drawing speed, taken as 45m / min (0.75m / s), and T0 is the tension setpoint, taken as 500N. Substituting these values, we get ΔT=0.024×0.75×500=9N. The elastic pulsation amplitude range is ±9N, and the measured tension fluctuation is ±7N, which is within the allowable range. The pulsation compensation algorithm is used to extract the residual signal from the original signal, retaining only the residual signal that exceeds the mass manifold boundary profile.

[0055] When identifying abnormal changes in the residual signal, it is necessary to distinguish the characteristic differences of different causes. Abnormal pull-out damping changes caused by die wear are manifested as a step-like continuous increase in damping value, increasing by 5 N·s / m for every 100m of twisting. Accompanying this is the slow increase in the energy integral of acoustic emission energy in the 100kHz~300kHz frequency band from 800mV·s to 1500mV·s, exceeding the upper limit of 1400mV·s of the mass manifold boundary.

[0056] The sudden change caused by the tension imbalance of a single wire manifests as a pulse-like jump in the transient yaw tension of the single wire. The tensions of adjacent wires are 510N and 480N, respectively, with a tension difference of 30N, accounting for 6% of the tension set value, exceeding the 5% process allowable threshold. The sudden change caused by excessive extrusion and scraping of the zinc layer is accompanied by a high-frequency spike in the acoustic emission signal of the die area in the 300kHz frequency band, with an amplitude of 850mV. At the same time, the spindle driving torque increases from 85N·m to 120N·m within 2ms. The sudden change rate is calculated as ΔT / Δt=(120-85) / 0.002=17500N·m / s, which exceeds the 10000N·m / s threshold of the mass manifold boundary.

[0057] By comparing the characteristic fingerprints of the mutation signals with the patterns of the three types of anomalies, the normal process fluctuations and abnormal pull-out damping mutations within the acceptable tolerance zone are completely separated. The extracted abnormal mutation signals have been freed from all interference components, retaining only the characteristics of the corresponding defect causes.

[0058] For the extracted abnormal mutation signals, the mutation rate of the transient yaw tension of the monofilament, the integral of the acoustic emission energy in the pressing zone, and the main shaft drive torque is mathematically normalized to construct an asymmetric stranding distortion index, and a degradation judgment is made for the electromechanical-acoustic combined yarn forming scenario. If an out-of-limit trend that causes yarn breakage and shutdown is detected, an emergency stop interruption command is issued to the electronic control PLC.

[0059] For the extracted anomalous mutation signals, mathematical normalization was performed on three types of parameters: transient yaw tension of the single filament, integral of acoustic emission energy in the molding zone, and mutation rate of spindle drive torque. Dimension elimination and distribution adaptation were then performed on these three types of parameters. The transient yaw tension of a single wire needs to be calculated based on the ratio of the stranding pitch to the nominal diameter of the wire. The deviation between the real-time collected yaw tension and the theoretical reference value is used as the original input. The deviation is compressed to the range of 0 to 1 through piecewise linear mapping logic. When the deviation is lower than the lower limit of the qualified tolerance zone, it is mapped to 0. When it is higher than the upper limit of the abnormal threshold, it is mapped to 1. The intermediate range is linearly scaled according to the deviation ratio. The acoustic emission energy integration in the molding area needs to be bandpass filtered according to the abnormal characteristic frequency band of 100kHz~300kHz first, and the energy accumulation value in the corresponding time window of the sudden signal is extracted. Then, the energy baseline under the mold wear state is normalized. The logarithm of the ratio of the current energy integral value to the baseline is mapped to the 0 to 1 interval. Logarithmic processing can suppress the dimension jump caused by occasional high frequency spikes such as zinc layer scratch. The sudden change rate of spindle drive torque needs to be calculated by first calculating the slope of torque change within adjacent sampling periods, and then calculating the ratio with the rated torque change rate threshold at the current production speed. The ratio is mapped to the 0 to 1 range through an S-shaped saturation function to avoid overflow of the normalized value caused by instantaneous torque impact. After the three types of parameters are normalized, they need to be weighted according to the twisting structure. The yaw tension has a higher weight in the same twist structure, and the acoustic emission energy integral has a higher weight in the alternating twist structure. The weight coefficient is determined by statistical analysis of the characteristic contribution of historical wire breakage samples. Finally, the weighted normalized values ​​of the three types are summed to obtain the asymmetric twisting distortion index.

[0060] When assessing degradation in electromechanical-acoustic combined spinning scenarios based on the asymmetric twisting distortion index, a dynamic threshold system is established. This threshold is not a fixed value but is dynamically adjusted according to the production cycle time. The threshold for the acceleration start-up phase is set to 1.3 times that of the steady-state phase to accommodate the inherent parameter fluctuations during the start-up process; the threshold for the mold transition phase is set to 1.2 times that of the steady-state phase to accommodate the transient damping changes caused by mold switching; the steady-state threshold is used for the constant speed stranding and tension adaptive adjustment phases, and the threshold reference is obtained by mapping the mass manifold boundary in the process reference base map.

[0061] The degradation judgment adopts the logic of sliding time window, continuously collecting the distortion index for 5 sampling periods. If the average index within the window exceeds the dynamic threshold corresponding to the current process, and the index rise slope exceeds the preset wire breakage precursor slope threshold, it is judged that there is an over-limit trend that will cause wire breakage and shutdown. If, at the same time the distortion index exceeds the limit, there is a continuous energy spike of more than 150kHz in the acoustic emission signal of the molding zone, a pulse jump of more than 10% in the yaw tension of the single wire, and the spindle torque mutation rate is positive for three consecutive cycles, then the validity of the exceeding trend is further confirmed, and misjudgment caused by occasional sensor interference is ruled out.

[0062] The theoretical tensile reference value of the transient yaw tension of a single wire is calculated based on the ratio of the stranding pitch to the nominal diameter of the wire. The stranding pitch is taken as 15 times the nominal diameter of the wire. This value conforms to the conventional parameter range of stranding process in the cable industry. The ratio of pitch to nominal diameter is the pitch ratio. Corresponding to the definition of pitch ratio in the stranding process, taking a 1×19 structural galvanized steel stranded wire with a nominal diameter of 5mm, the stranding pitch is 75mm, and the pitch ratio is 75÷5=15. The reference tensile force under the unit pitch ratio is taken as 40N. The theoretical tensile reference value = pitch ratio × reference tensile force per unit pitch ratio, which is calculated to be 15×40=600N.

[0063] The real-time yaw thrust is 636N, and the deviation is 636-600=36N. The acceptable tolerance zone is ±5% of the theoretical reference value, that is, the lower limit of the deviation is -30N and the upper limit is +30N. The upper limit of the abnormal threshold is +10% of the theoretical reference value, that is, the deviation is +60N. Below -30N is mapped to 0, and above +60N is mapped to 1. The intermediate range is linearly scaled according to the deviation ratio. The normalization value is calculated by subtracting the lower limit of the deviation from the deviation, and dividing by subtracting the lower limit of the deviation from the upper limit of the abnormal threshold. Substituting the values, we get 36-(-30)=66, 60-(-30)=90, and 66÷90≈0.733.

[0064] The acoustic emission energy integral of the molding zone is first bandpass filtered according to the abnormal characteristic frequency band of 100kHz~300kHz. The energy accumulation value within a 50ms time window corresponding to the sudden change signal is extracted. The measured current energy integral value is 2100mV·s. The energy baseline based on the mold wear state is 1200mV·s, and the upper limit of the abnormal threshold is 3000mV·s. The normalization logic is to take the logarithm of the ratio of the current energy integral value to the baseline to the base 10, and then divide it by the abnormal threshold. The ratio of the limit to the baseline is taken as the logarithm to the base 10 and mapped to the interval between 0 and 1. The calculated ratio of the current energy integral to the baseline is 2100 ÷ 1200 = 1.75, and the logarithm to the base 10 is approximately 0.243. The ratio of the upper limit of the abnormal threshold to the baseline is 3000 ÷ 1200 = 2.5, and the logarithm to the base 10 is approximately 0.398. The normalized value is 0.243 ÷ 0.398 ≈ 0.611. Logarithmic processing can suppress the dimensional jumps caused by occasional high-frequency spikes such as zinc layer scratches.

[0065] The sudden change in spindle drive torque is first calculated by the slope of torque change within adjacent 10ms sampling periods. The measured torque increases from 85 N·m to 102 N·m, and the slope of change is (102-85)÷0.01s=1700 N·m / s. The rated torque change rate threshold at the current production speed of 60m / min is 1500 N·m / s. The ratio is calculated as 1700÷1500≈1.133. The ratio is mapped to the interval between 0 and 1 through the S-shaped saturation function. The logic of the S-shaped saturation function is 1÷(1+e^(-10×(ratio-0.5))). Substituting the values, we get 1÷(1+e^(-10×(1.133-0.5))=1÷(1+e^(-6.33))≈1÷(1+0.0018)≈0.998, which avoids overflow of the normalization value caused by instantaneous torque impact.

[0066] After normalizing the three types of parameters, weights are allocated according to the twisting structure. The yaw tension has a higher weight in the same-direction twist structure, while the acoustic emission energy integral has a higher weight in the alternating twist structure. The weight coefficients are determined by statistical analysis of the characteristic contribution of historical wire breakage samples. Taking the same-direction twist structure scenario, the yaw tension weight is 0.5, the acoustic emission energy integral weight is 0.2, and the spindle drive torque mutation rate weight is 0.3. The asymmetric twisting distortion index is calculated by multiplying the three normalized values ​​by their corresponding weights and then summing them. Substituting the values, we get 0.733×0.5+0.611×0.2+0.998×0.3≈0.3665+0.1222+0.2994≈0.788.

[0067] When assessing degradation in electromechanical-acoustic combined yarn-making scenarios based on the asymmetric stranding distortion index, a dynamic threshold system is established. This threshold is not a fixed value but is dynamically adjusted according to the production cycle. The steady-state threshold is obtained by mapping the mass manifold boundary in the process baseline diagram and is set to 0.6. The threshold for the acceleration start-up stage is set to 1.3 times that of the steady-state stage, calculated as 0.6 × 1.3 = 0.78, to accommodate the inherent parameter fluctuations during start-up. The threshold for the mold transition stage is set to 1.2 times that of the steady-state stage, calculated as 0.6 × 1.2 = 0.72, to accommodate the transient damping changes caused by mold switching. The steady-state threshold of 0.6 is used in the constant-speed stranding and tension adaptive adjustment stages. Currently, we are in the constant-speed stranding stage, so the applicable threshold is 0.6. The calculated distortion index 0.788 > 0.6, which meets the limit-breaking condition.

[0068] The degradation judgment adopts the logic of sliding time window, continuously collecting the distortion index for 5 sampling periods, with a window length of 50ms. The distortion indices for the 5 periods are 0.68, 0.71, 0.74, 0.76, and 0.788, respectively. The average index within the window is calculated as (0.68+0.71+0.74+0.76+0.788)÷5≈0.736, which exceeds the dynamic threshold of 0.6 corresponding to the current process. The index rise slope is calculated as (0.788-0.68)÷0.04=2.7 / s, which exceeds the preset pre-breakage slope threshold of 2 / s, and is judged to be an out-of-limit trend that will cause wire breakage and shutdown.

[0069] While the distortion index exceeded the limit, a continuous energy spike above 150kHz appeared in the acoustic emission signal of the pressure mold area, corresponding to an energy integral of 252mV·s; the yaw tension of the single filament was 666N, which was 11% higher than the theoretical reference value of 600N, exceeding the 10% pulse jump threshold; the spindle torque mutation rate was +150N·m / s, +170N·m / s, and +190N·m / s for three consecutive cycles, all of which were positive, further confirming the validity of the exceeding trend and eliminating misjudgment caused by occasional sensor interference.

[0070] After confirming a tendency for the system to stop beyond its limits due to wire breakage, an emergency stop command is sent to the PLC. Before issuing the command, an execution priority check is performed. If the system is currently in normal operation, the three-level blocking logic is triggered. The first stage is torque limiting, which reduces the upper limit of the spindle drive torque output to 30% of the rated value. The rated torque is 200 N·m, and the upper limit after limiting is 60 N·m, which suppresses the continued force during the twisting process. The second stage is tension release, which controls the tension controller of the wire feeding frame to gradually reduce the tension of the single wire from 600N to zero. The release time is 120ms to prevent the wire from being pulled apart during the emergency stop. The third stage is the main spindle braking, which sends a brake trigger signal to the PLC. The main spindle braking deceleration is set at 80% of the steel wire yield strength. With a yield strength of 1200MPa, the deceleration is set to the braking parameters corresponding to 960MPa, ensuring rapid stopping while preventing new plastic deformation of the steel wire due to excessive braking. After the emergency stop command is issued, the storage of all current multimodal operating data is locked simultaneously. The time node of the abnormality occurrence is marked as 127.4s after startup, with a corresponding distortion index value of 0.788. This provides data support for subsequent defect tracing and process parameter optimization. The response time of the entire blocking process from trend confirmation to completion of main spindle braking is 185ms, which does not exceed the requirement of 200ms, allowing intervention to be completed before wire breakage occurs.

[0071] After confirming a tendency for the system to stop beyond its limit due to wire breakage, an emergency stop interruption command is sent to the PLC. Before sending the command, an execution priority judgment must be performed. If the current system is already in emergency stop protection mode, the command will not be sent again. If the system is in normal operation mode, the three-level interruption logic will be triggered. The first stage is torque limiting, which reduces the upper limit of the spindle drive torque output to 30% of the rated value, suppressing the continued force during the twisting process; The second stage is tension relief, which controls the tension controller of the wire feeding frame to gradually reduce the tension of the single wire to zero, so as to prevent the wire from being pulled apart during the emergency stop. The third stage is the main spindle braking, which sends a brake trigger signal to the PLC. The main spindle braking deceleration is set at 80% of the yield strength of the steel wire, which ensures rapid stopping while avoiding new plastic deformation of the steel wire due to excessive braking.

[0072] After the emergency stop command is issued, the storage of all current multimodal operating data needs to be locked simultaneously, and the time node of the anomaly and the corresponding distortion index value should be marked to provide data support for subsequent defect tracing and process parameter optimization. The response time of the entire blocking process from trend confirmation to completion of spindle braking does not exceed 200ms, which can effectively complete the intervention before the wire breakage occurs and reduce the loss caused by the scrapping of the entire batch of products.

[0073] This embodiment also provides a forming monitoring system for galvanized steel strand, including: The simulation module continuously captures multimodal operation data during the start-up and steady-state operation phases. Combining the improved flexible body spatial spiral twisting algorithm and trajectory compensation mechanism, it simulates the spatial twisting evolution curve during the galvanized steel wire convergence and forming process, and generates a process baseline map. Signal extraction module: Based on the process baseline map, the process response features are bound and identified with the current production cycle time, the quality manifold boundary is defined, and the process fluctuations within the acceptable tolerance zone and abnormal pull-out damping abrupt changes are separated by the quality manifold boundary contour, and abnormal abrupt change signals are extracted. Command issuance module: For abnormal sudden change signals, the sudden change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the pressing zone and the main shaft driving torque is mathematically normalized to construct an asymmetric stranding distortion index, and the degradation judgment is made for the electromechanical-acoustic combined yarn forming scenario. If an over-limit trend that causes yarn breakage and shutdown is detected, an emergency stop interruption command is issued to the electrical control PLC.

[0074] In the description of this specification, references to the terms "an embodiment," "example," "example," etc., indicate that a feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the described features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0075] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to only certain implementations. Clearly, many modifications and variations can be made based on the content of this specification. These embodiments have been selected and described in this specification to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for monitoring the forming of galvanized steel strand, characterized in that: The monitoring method includes the following steps: S1: During the start-up and steady-state operation phases, the sensing layer continuously captures multimodal operation data. Combined with the improved flexible body spatial spiral twisting algorithm and trajectory compensation mechanism, the spatial twisting evolution curve during the galvanized steel wire convergence and forming process is deduced to generate a process baseline map. S2: Based on the process baseline map, the process response characteristics are bound to the current production cycle time for identification, and the quality manifold boundary is delineated; S3: Using the mass manifold boundary profile, process fluctuations within the acceptable tolerance zone are separated from abnormal pull-out damping abrupt changes, and abnormal abrupt change signals are extracted; S4: For abnormal abrupt change signals, the abrupt change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the pressing zone, and the main shaft driving torque are mathematically normalized to construct an asymmetric twisting distortion index. S5: Degradation judgment is performed on the electromechanical and acoustic combined yarn forming scenario. If an over-limit trend that would cause yarn breakage and shutdown is detected, an emergency stop interruption command is sent to the electrical control PLC.

2. The method for monitoring the forming of galvanized steel strand according to claim 1, characterized in that: The steps in step S1 to improve the flexible body spatial helical twisting algorithm include: Based on the differential geometric modeling logic of the spatial helical structure of steel wire rope, multiple vectors of the center line of the steel wire are defined with the Frenet frame as the reference. The multiple parameters of the steel wire are mapped to the rotation angle and pitch variables of the spatial curve. A parameterized description of the spatial curve of a single galvanized steel wire in the stranding process is established, and the three-dimensional spatial pose of the steel wire in the multi-layer stranding process is deduced. Based on the microstructure evolution law of the entire process of high carbon galvanized steel wire drawing-hot galvanizing, the parameters of sorbite lamellar spacing, cementite fragmentation degree, and dislocation density are mapped to the equivalent elastic modulus and yield limit of the steel wire. When the drawing stress is close to the dynamic yield threshold of the material, the curvature calculation value of the space curve is positively compensated to correct the trajectory deviation of the steel wire caused by plastic deformation. Based on the test results of friction characteristics between galvanized steel wires, the zinc layer thickness, surface roughness, and contact stress are mapped to dynamic friction coefficients. For the contact slip scenarios between the steel wire and the die, and between the steel wires, negative compensation is made to the calculated tangential displacement value of the space curve to counteract the trajectory lag caused by friction resistance. The two types of compensation quantities are weighted and fused together and then superimposed onto the original spatial curve calculation results to deduce the entire process spatial twisting evolution curve of a single galvanized steel wire from laying, entering the mold to twisting and forming.

3. The method for monitoring the forming of galvanized steel strand according to claim 2, characterized in that: The steps in step S1 to generate the process baseline map include: Based on the spatial twisting evolution curve obtained by deduction, the cycle time parameters of the current production process are synchronously associated, the process parameters are bound to the time nodes of the evolution curve, and the process response characteristic range covering the qualified tolerance zone is defined. Integrated anti-fragmentation and wire breakage safety prediction logic: When the interlayer relative displacement of adjacent wires in the evolution curve exceeds the critical threshold of delamination slip, it is marked as a strand risk area; When the equivalent torsional stress of a single filament exceeds the torsional fatigue threshold under the current dislocation density, it is marked as a filament breakage risk area.

4. The method for monitoring the forming of galvanized steel strand according to claim 1, characterized in that: In step S1, the sensing layer captures multimodal operating data, including mechanical oscillation signals, wire pulling stress, and transient slip acoustic signatures.

5. The method for monitoring the forming of galvanized steel strand according to claim 1, characterized in that: In step S2, the process response characteristics are bound to the current production cycle time based on the process baseline map, and the quality manifold boundary is delineated, including the following steps: For the adaptive adjustment of the entire stranding process of galvanized steel strand, benchmark ranges of process parameters are established for the corresponding cycle time: In the acceleration and start-up phase, the spindle speed ramp rate and the wire tension gradient rise curve are used as binding parameters. In the constant speed twisting phase, the average single filament tension under stable twist pitch and the friction temperature rise rate in the die zone are used as binding benchmarks. In the transition phase, the focus is on binding the drawing resistance change threshold corresponding to the die hole wear amount. In the tension adaptive adjustment phase, the output response delay and steady-state error range of the tension closed-loop controller are associated. The time nodes of the spatial twisting evolution curve derived from the process baseline map are matched with the real-time acquired process cycle state signals. When the cycle state changes, the corresponding process response feature subset is automatically called. After dividing the K-nearest neighbor region based on manifold distance, the global feature matrix and the local feature matrix are calculated separately. The fusion adjustment parameter of the global and local features is determined by the feature matrix spectral radius analysis method. The two types of features are fused to obtain the dimension reduction matrix, which is then projected onto the boundary of the mass manifold.

6. The method for monitoring the forming of galvanized steel strand according to claim 1, characterized in that: In step S3, using the mass manifold boundary profile, process fluctuations within the acceptable tolerance zone are separated from abnormal pull-out damping abrupt changes, and abnormal abrupt change signals are extracted, including the following steps: By comparing the characteristic fingerprint of the mutation signal with the patterns of the three types of anomalies, the process fluctuations and abnormal pull-out damping mutations are completely separated, and the abnormal mutation signals corresponding to the defect cause characteristics are extracted.

7. The method for monitoring the forming of galvanized steel strand according to claim 1, characterized in that: In step S4, for abnormal abrupt change signals, the abrupt change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the molding zone, and the spindle drive torque is mathematically normalized to construct the asymmetric twisting distortion index. The steps are as follows: The theoretical tensile reference value is calculated based on the ratio of stranding pitch to nominal wire diameter. The deviation between the real-time collected yaw tensile force and the theoretical reference value is used as the original input. The deviation is compressed to the range of 0 to 1 through piecewise linear mapping logic. When the deviation is lower than the lower limit of the qualified tolerance zone, it is mapped to 0. When it is higher than the upper limit of the abnormal threshold, it is mapped to 1. The intermediate range is linearly scaled according to the deviation ratio. The acoustic emission energy integral of the molding area is bandpass filtered according to the abnormal characteristic frequency band, and the energy accumulation value within the time window corresponding to the sudden change signal is extracted. Based on the energy baseline under the mold wear state, the energy is normalized and the logarithm of the ratio of the current energy integral value to the baseline is mapped to the interval of 0 to 1. The spindle drive torque mutation rate is calculated by the slope of torque change within adjacent sampling periods, and then the ratio is calculated with the rated torque change rate threshold at the current production speed. The ratio is then mapped to the 0 to 1 interval through an S-shaped saturation function. The asymmetric twisting distortion index is obtained by summing the weighted normalized values ​​of the three categories.

8. The method for monitoring the forming of galvanized steel strand according to claim 1, characterized in that: The degradation judgment steps in step S5 include: Establish a dynamic threshold system for exceeding limits, with the threshold dynamically adjusted according to the production cycle time; The degradation judgment is performed using a sliding time window logic. The distortion index is collected continuously for 5 sampling periods. If the average index value within the window exceeds the dynamic over-limit threshold corresponding to the current process, and the index rise slope exceeds the preset wire breakage precursor slope threshold, it is determined that there is an over-limit trend that will cause wire breakage and shutdown.

9. The method for monitoring the forming of galvanized steel strand according to claim 8, characterized in that: The emergency stop and intervention steps in step S5 include: After confirming a tendency for the system to stop beyond its limits due to wire breakage, a priority judgment is performed. If the current system is in an emergency stop protection state, the command will not be issued again. If it is in a normal operating state, the three-level blocking logic is triggered. The first level is torque limiting, which reduces the upper limit of the spindle drive torque output to 30% of the rated value. The second stage is tension release, where the tension controller of the wire feeding frame gradually reduces the tension of the single filament to zero. The third stage is the main spindle braking, which sends a brake trigger signal to the PLC. The main spindle braking deceleration is set at 80% of the yield strength of the steel wire.

10. A forming monitoring system for galvanized steel strand, used to implement the monitoring method according to any one of claims 1-9, characterized in that: include: The simulation module continuously captures multimodal operation data during the start-up and steady-state operation phases. Combining the improved flexible body spatial spiral twisting algorithm and trajectory compensation mechanism, it simulates the spatial twisting evolution curve during the galvanized steel wire convergence and forming process, and generates a process baseline map. Signal extraction module: Based on the process baseline map, the process response features are bound and identified with the current production cycle time, the quality manifold boundary is defined, and the process fluctuations within the acceptable tolerance zone and abnormal pull-out damping abrupt changes are separated by the quality manifold boundary contour, and abnormal abrupt change signals are extracted. Command issuance module: For abnormal sudden change signals, the sudden change rate of the transient yaw tension of the single filament, the integral of the acoustic emission energy in the pressing zone and the main shaft driving torque is mathematically normalized to construct an asymmetric stranding distortion index, and the degradation judgment is made for the electromechanical-acoustic combined yarn forming scenario. If an over-limit trend that causes yarn breakage and shutdown is detected, an emergency stop interruption command is issued to the electrical control PLC.

Images