A smart quality control system for the entire process of concrete utility poles
The intelligent quality control system throughout the entire process enables cross-domain transmission and collaborative response of status information between processes during the production of concrete poles, solving the problem of information silos between processes in existing technologies and improving the consistency and adaptability of production quality.
Patent Information
- Application Number
- CN202611338879.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-31
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies, there is a lack of effective status information transmission mechanisms between the centrifugal molding, curing and demolding processes of concrete poles, resulting in a lack of coordination in quality control, the inability to achieve real-time feedback and adjustment, and affecting production quality.
A full-process intelligent quality control system for concrete poles is constructed. Through multi-source signal fusion sensing and multi-dimensional material state characterization, the system realizes quantitative real-time sensing of the material state within the closed mold. Through the linkage control center, the system realizes cross-domain transmission and collaborative response of state information between processes, including closed-loop control of centrifugal molding, curing and demolding.
It has achieved a consistent quality improvement throughout the entire production process of concrete poles. Through an adaptive closed-loop control mode, it dynamically adjusts the parameters of each process, improves the adaptability to material batch fluctuations and environmental changes, and reduces the risk of demolding damage.
Smart Images

Figure CN123103841A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent manufacturing and concrete product production equipment control technology, specifically to an intelligent quality control system for the entire process of concrete pole production. Background Technology
[0002] The centrifugal molding manufacturing process of concrete poles typically includes key steps such as material placement and mold closing, centrifugal molding, steam curing, and demolding and cleaning. The control level of process parameters in each step directly determines the core quality indicators of the final product, such as density, concentricity, and structural strength. Currently, some technological accumulation has been made in the intelligent production of concrete poles: In the centrifugal molding stage, existing technologies have attempted to monitor the equipment's operating status during centrifugation by installing sensors on the centrifuge to collect electrical signal parameters (such as current and torque), and to perform open-loop control based on a preset fixed time-speed curve; in the curing stage, existing solutions mostly use preset time-temperature curves for programmed curing, while some solutions use temperature sensors placed in the curing kiln to achieve real-time monitoring and feedback adjustment of the ambient temperature; in the demolding stage, existing operations are usually performed based on experience or fixed time nodes, and the demolding parameters remain unchanged within batches. In addition, with the development of industrial internet technology, some production lines have achieved data acquisition and local display within a single process, providing basic data support for local optimization within the process.
[0003] However, the existing technology still has the following significant drawbacks: First, during the centrifugal molding process, the concrete material is in a closed mold, and its true rheological properties, distribution uniformity, and internal structural evolution cannot be directly observed. Furthermore, the existing electrical signal monitoring schemes only collect data from a single type of sensor, which is difficult to effectively eliminate equipment noise interference under strong mechanical disturbance conditions. It is also impossible to accurately extract effective characteristic signals that truly reflect the state of the concrete material. Moreover, the control strategy is still mainly based on preset programs and lacks the adaptive adjustment capability based on real-time feedback of the material's internal state. Secondly, and more critically, the three processes of centrifugal molding, curing, and demolding are currently isolated and operate independently in existing technologies, lacking an effective mechanism for transmitting state information between them. The actual hardening state and temperature distribution of the concrete at the end of centrifugal molding directly determine the reasonable starting point of the curing regime. However, in existing solutions, the curing process cannot obtain the true material state information at the final state after centrifugation, and can only execute a fixed regime under conditions of information deficiency, leading to both insufficient and excessive curing. Similarly, the actual hardening degree of the pole at the final curing state directly determines the reasonable value of the demolding parameters, but the demolding process cannot obtain the true state data at the final curing state, and the demolding parameters cannot be pre-calibrated based on the actual hardening degree of the pole. Furthermore, there is a lack of dynamic adjustment capability based on real-time feedback during the demolding process. These information silos between processes result in a lack of coordination in the overall quality control process, becoming a bottleneck restricting further improvement in the production quality of concrete poles. Summary of the Invention
[0004] To address the problems of existing technologies, this invention provides an intelligent quality control system for the entire process of concrete utility poles.
[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: Firstly, a full-process intelligent quality control system for concrete poles, including a centrifugal molding subsystem, a curing subsystem and a demolding subsystem, and also including a linkage control center connecting the various subsystems; The centrifugal forming subsystem includes: The reference parameter identification unit is used to perform a stepped speed-up test under the no-load state of the centrifuge, collect the real-time operating data of the drive motor, and calculate the empty mold impedance parameters that characterize the inherent mechanical resistance of the mold through regression analysis. The empty mold impedance parameters include the empty mold rotational inertia J, the mold viscous friction coefficient B, and the Coulomb friction torque T_c. The mixed signal acquisition unit is used to simultaneously acquire the electromagnetic drive signal of the drive motor, the vibration signal of the mold, and the sound signal inside the cavity during the operation of the centrifuge under load, forming a mixed signal. The signal separation and feature extraction unit is used to perform blind source separation processing on the mixed signal, extract the pure material state signal component, and perform multi-scale permutation entropy analysis on the component to extract feature parameters characterizing the ordered evolution of the internal structure of concrete. The feature parameters include the entropy change rate ΔH / Δs, the entropy inflection point position s_inflection point, and the entropy stability interval. The load torque observation unit is used to acquire the electromagnetic driving torque T_e in real time, and to calculate the mechanical consumption torque T_m of the mold itself at the current speed using the empty mold impedance parameters. The electromagnetic driving torque is subtracted from the mechanical consumption torque to extract the net load torque T_net that only acts on the concrete material. The rheological and distribution state analysis unit is used to perform micro-perturbation response analysis on the net load torque T_net, calculate the equivalent dynamic viscosity η characterizing the flow state of concrete, and extract the frequency domain features of the net load torque T_net to calculate the dimensionless eccentricity factor ε characterizing the uniformity of material distribution. The hardening state assessment unit is used to comprehensively assess the internal density and hardening process of concrete based on the real-time change trend of the characteristic parameters, and generate the centrifugal final state hardening state level H_final and the centrifugal final state temperature distribution field T_dist. An adaptive closed-loop control unit is used to dynamically adjust the centrifuge speed or stage switching timing based on the preset value range of the dimensionless eccentricity factor ε, the equivalent dynamic viscosity η, and the hardening state level. The linkage control center includes a positive linkage control module, which is used to send the centrifugation final hardening state level H_final, centrifugation final temperature distribution field T_dist, and centrifugation final eccentricity factor ε_final as linkage data packets to the maintenance subsystem. The maintenance subsystem includes: A maintenance status receiving unit is used to receive the linkage data packet; The initial maintenance regime generation unit is used to match an initial maintenance scheme from a preset maintenance rule base according to the centrifugation final hardening state level H_final and the centrifugation final temperature distribution field T_dist. The initial maintenance scheme includes an initial heating rate, an initial isothermal temperature, an initial isothermal duration, and an initial cooling rate. The maintenance process monitoring unit is used to continuously collect data on the internal temperature of the pole, the surface temperature of the pole, and the temperature and humidity of the maintenance environment during the maintenance process, and to calculate the real-time hardening degree S(t) and real-time temperature gradient G(t) inside the pole based on these data. The curing parameter dynamic adjustment unit is used to dynamically adjust the curing parameters according to the real-time hardening degree S(t) and the real-time temperature gradient G(t), including: when the real-time hardening degree S(t) deviates from the expected hardening curve S_target(t), adjusting the heating rate or the constant temperature duration for compensation; when the real-time temperature gradient G(t) exceeds the preset safety threshold G_max, reducing the heating rate or extending the constant temperature duration to reduce the temperature gradient. The final state evaluation unit is used to evaluate the final hardening degree S_final_cure and final temperature T_final_cure of the pole at the end of the curing process, and generate the demolding readiness index R_demold accordingly. The positive linkage control module of the linkage control center is also used to send the final hardening degree S_final_cure and the final temperature T_final_cure as demolding pre-parameters to the demolding subsystem; The demolding subsystem includes: The demolding parameter pre-calibration unit is used to receive the demolding pre-parameters, and calculate the demolding parameter calibration coefficient λ in combination with the centrifugal final state eccentricity factor ε_final, to perform pre-calibration on the initial demolding parameters, and generate calibrated demolding parameters. The calibrated demolding parameters include calibrated demolding time t_demold, calibrated demolding speed v_demold, and calibrated demolding force F_demold. The demolding process monitoring unit is used to collect demolding force F(t), demolding displacement D(t) and pole surface strain data in real time during the demolding process, and calculate the real-time bonding force B(t) and demolding resistance change rate ΔR(t) based on these data. The demolding dynamic adjustment unit is used to dynamically adjust the demolding speed and demolding force according to the real-time adhesive force B(t) and the demolding resistance change rate ΔR(t): when the real-time adhesive force B(t) exceeds the preset safety range, the demolding speed is reduced or demolding is paused and auxiliary vibration is started; when the demolding resistance change rate ΔR(t) increases abnormally, an early warning is triggered and the demolding parameters are adjusted. The demolding anomaly diagnosis unit is used to diagnose the cause of the anomaly and generate an anomaly feedback report when abnormal demolding resistance or excessive strain on the pole surface is detected. The linkage control center also includes a reverse linkage control module, which is used to send the abnormal feedback report to the maintenance parameter dynamic adjustment unit of the maintenance subsystem to calibrate the maintenance regime parameters of subsequent batches of poles, and to send it to the adaptive closed-loop control unit of the centrifugal forming subsystem to adjust the centrifugal control strategy of subsequent batches of poles.
[0006] In one specific embodiment of the first aspect, the specific process by which the reference parameter identification unit calculates the empty-mode impedance parameter is as follows: Data on the mechanical angular velocity ω, angular acceleration α, and stator torque current component I_q of the centrifuge during the no-load acceleration process were collected. Construct a system of linear equations based on rigid body dynamics: T_e=J·α+B·ω+T_c Where T_e is the electromagnetic drive torque, T_e=K_t·I_q, and K_t is the motor torque constant; Using the stator torque current component I_q as the input variable and the mechanical angular velocity ω and angular acceleration α as the state variables, the least squares method is used to fit three constant terms J, B, and T_c, which are defined as the moment of inertia of the empty mold, the viscous friction coefficient of the mold, and the Coulomb friction torque, respectively. The combination of J, B, and T_c is set as the empty-mode impedance parameter and stored in the reference parameter identification unit.
[0007] In one specific implementation of the first aspect, the specific process of blind source separation and multi-scale permutation entropy analysis performed by the signal separation and feature extraction unit is as follows: The mixed signal is preprocessed by filtering, denoising, and normalization; The independent component analysis algorithm was used to perform blind source separation on the preprocessed mixed signal, separating the material state signal component y_1(t), the mechanical vibration signal component y_2(t), and the environmental noise signal component y_3(t); The material state signal component y_1(t) is extracted as an effective signal characterizing the internal state of the concrete; The material state signal component y_1(t) is subjected to multi-scale coarsening processing. For the scale factor s, a coarsened sequence is constructed as follows:
[0008] Where N is the signal length, and s takes the value of a positive integer from 1 to 20; For each coarse-grained sequence at scale s, calculate its permutation entropy value H_s with embedding dimension m=4 and delay time τ=1, and construct a multi-scale permutation entropy spectrum {H_s|s=1,2,...,20}; Feature parameters are extracted from the multi-scale entropy spectrum: entropy change rate ΔH / Δs, entropy inflection point position s_inflection point, and entropy stability interval.
[0009] In one specific implementation of the first aspect, the specific process by which the load torque observation unit obtains the net load torque T_net is as follows: The real-time acquired stator current q-axis component I_q is converted into electromagnetic drive torque T_e=K_t·I_q; Based on the current mechanical angular velocity ω and angular acceleration α, and combined with the empty mold impedance parameters J, B, and T_c, the theoretical mechanical torque required for the mold to overcome inertia, viscous friction, and static friction is calculated as follows: T_m = J·α + B·ω + T_c; Calculate the difference between the electromagnetic drive torque and the theoretical mechanical torque, ΔT = T_e - T_m; The difference is superimposed with the compensation value ΔT_comp generated based on the speed observation error, and the final value after superposition is determined as the net load torque T_net = ΔT + ΔT_comp.
[0010] In a specific embodiment of the first aspect, the rheological and distribution state analysis unit performs micro-perturbation response analysis and calculates the equivalent dynamic viscosity η as follows: A sinusoidal wave signal with a preset frequency f and amplitude A is superimposed on the basic speed command of the centrifuge. The frequency f ranges from 1 to 5 Hz, and the amplitude A ranges from 1% to 5% of the basic speed command value. The collected net load torque data and mechanical angular velocity data are subjected to high-pass filtering to remove the DC component and extract the torque fluctuation component ΔT_f and velocity fluctuation component Δω_f, which are consistent with the frequency of the sinusoidal wave signal. The root mean square value (RMS) of the torque fluctuation component (ΔT_f) and the root mean square value (RMS) of the speed fluctuation component (Δω_f) are calculated respectively within a preset time window, wherein the preset time window ranges from 3 to 10 seconds. Calculate the ratio R = RMS(ΔT_f) / RMS(Δω_f); Obtain the inner radius r and effective length L of the mold, and calculate the geometric constant G = 2π·r. 2·L, divide the ratio by the geometric constant to obtain the equivalent dynamic viscosity η=R / G, where the unit of the equivalent dynamic viscosity η is Pascal-second.
[0011] In one specific implementation of the first aspect, the rheological and distribution state analysis unit performs frequency domain feature extraction and eccentricity factor ε calculation as follows: Perform a Fast Fourier Transform on the net load torque T_net to obtain its frequency domain spectrum; Extract the amplitudes A_1, A_2, A_3, and A_4 from the spectrum corresponding to the centrifuge's fundamental frequency f_0 and its second, third, and fourth harmonics; Calculate the dimensionless eccentricity factor ε:
[0012] Wherein, w_2, w_3, and w_4 are preset weight coefficients, and w_2+w_3+w_4=1. The value range of the preset weight coefficients is: w_2 is 0.4~0.6, w_3 is 0.2~0.4, and w_4 is 0.1~0.3.
[0013] In one specific implementation of the first aspect, the adaptive closed-loop control unit generates control commands based on the evaluation parameters using the following specific strategy: When the dimensionless eccentricity factor ε exceeds the preset safety threshold ε_max, an emergency braking command is output, wherein the value of ε_max is in the range of 0.12 to 0.18. When the equivalent dynamic viscosity η is higher than the preset rheological threshold η_max, it is determined that the rheological state of the concrete has not met the standard, and a constant speed shear command is output to maintain the current speed for a preset time. The value range of η_max is 400~600 Pascals per second, and the value range of the preset time is 30~90 seconds. When the equivalent dynamic viscosity η is lower than or equal to the preset rheological threshold η_max and the dimensionless eccentricity factor ε is less than or equal to the preset safety threshold ε_max, it is determined that the rheological state meets the standard and the eccentricity is within the safe range. An adaptive acceleration command is output to accelerate to the target speed of the next stage with a preset acceleration. The value range of the preset acceleration is 0.3~0.8 rpm / s. When the hardening state level reaches the preset density level, a centrifugation termination command is output.
[0014] In one specific embodiment of the first aspect, the specific strategy for the maintenance parameter dynamic adjustment unit to dynamically adjust the maintenance parameters is as follows: Set the expected hardening curve S_target(t), which is pre-calibrated and stored in the curing subsystem according to the pole specifications and concrete mix ratio; During the maintenance process, the deviation between the current actual hardening degree S(t) and the expected hardening degree S_target(t) is calculated at preset time intervals Δt, where the value of the preset time interval Δt ranges from 5 to 20 minutes. When ΔS(t) is greater than the preset positive deviation threshold, it is determined that the hardening speed is too slow, and a heating rate compensation operation is performed: the current heating rate is increased by a first preset percentage, or the current constant temperature is increased by a first preset temperature value. The preset positive deviation threshold ranges from 0.03 to 0.08, the first preset percentage ranges from 5% to 15%, and the first preset temperature value ranges from 1 to 3℃. When ΔS(t) is less than the preset negative deviation threshold, it is determined that the hardening speed is too fast, and a cooling rate compensation operation is performed: the current heating rate is reduced by a second preset percentage, or the current constant temperature is reduced by a second preset temperature value. The preset negative deviation threshold ranges from -0.08 to -0.03, the second preset percentage ranges from 5% to 15%, and the second preset temperature value ranges from 1 to 3℃. When the real-time temperature gradient G(t) exceeds the preset safety threshold G_max, a gradient suppression operation is performed: the heating is paused and the isothermal duration of the current temperature stage is extended until the temperature gradient falls back to the safe range. The value of G_max is 2.0~3.0℃ / cm.
[0015] In one specific embodiment of the first aspect, the specific process by which the demolding parameter pre-calibration unit calculates the demolding parameter calibration coefficient λ and calibrates the initial demolding parameters is as follows: Obtain the final curing hardening degree S_final_cure, the final temperature T_final_cure, and the centrifugal final eccentricity factor ε_final, where the value of S_final_cure ranges from 0 to 1; Obtain the target curing hardening degree S_target_cure and the median value of the optimal demolding temperature range T_target_demold; Calculate the hardening degree deviation ΔS = S_target_cure - S_final_cure; Calculate the temperature deviation ΔT = T_final_cure - T_target_demold; Calculate the overall calibration coefficient λ:
[0016] where α, β, and γ are preset weight coefficients, α+β+γ=1, the value range of α is 0.4 to 0.6, the value range of β is 0.15 to 0.35, and the value range of γ is 0.2 to 0.4; calibrate the initial demolding parameters by using the comprehensive calibration coefficient λ: calibrate the demolding time: t_demold=t_0×(1+k_t·λ), where t_0 is the initial demolding time, and k_t is a preset time calibration coefficient; calibrate the demolding speed: v_demold=v_0×(1-k_v·λ), where v_0 is the initial demolding speed, and k_v is a preset speed calibration coefficient; calibrate the demolding force: F_demold=F_0×(1-k_F·λ), where F_0 is the initial demolding force, and k_F is a preset force calibration coefficient; the value range of k_t is 0.2 to 0.5, the value range of k_v is 0.3 to 0.6, and the value range of k_F is 0.2 to 0.5.
[0017] In a specific embodiment of the first aspect, the specific process for the demolding abnormality diagnosis unit to diagnose the abnormality cause and generate an abnormality feedback report is as follows: When an abnormality of demolding resistance is detected, the demolding force F_abnormal, the demolding displacement D_abnormal and the pole surface strain data ε_strain at the moment when the abnormality occurs are collected; Compare the demolding force F_abnormal with the preset upper limit of bonding force B_max and the preset lower limit of bonding force B_min: If F_abnormal>B_max, it is determined that the abnormality type is over-tight bonding abnormality, and the value range of B_max is 20 to 30kN; If F_abnormal<B_min and the demolding displacement D_abnormal is greater than the preset normal demolding displacement D_normal, it is determined that the abnormality type is loose bonding abnormality, and the value range of B_min is 8 to 15kN; The abnormality cause is diagnosed according to the abnormality type and the pole surface strain data ε_strain, and the abnormality cause includes at least one of local damage on the inner wall of the mold, residual substances on the inner wall of the mold, insufficient local hardening of the electric pole, and excessive local curing of the electric pole; An abnormality feedback report is generated, and the abnormality feedback report includes the abnormality type, abnormality position, abnormality degree and preliminary diagnosis cause; The reverse linkage control module sends the abnormal feedback report to the maintenance parameter dynamic adjustment unit of the maintenance subsystem. The maintenance parameter dynamic adjustment unit adjusts the maintenance regime parameters of subsequent batches of poles according to the abnormality type in the abnormal feedback report: when the abnormality type is excessive adhesion, the constant temperature of subsequent batches of poles is reduced or the constant temperature time is shortened; when the abnormality type is insufficient local hardening of the pole, the constant temperature time of subsequent batches of poles is extended or the constant temperature is increased. The reverse linkage control module sends the abnormal feedback report to the adaptive closed-loop control unit of the centrifugal forming subsystem. The adaptive closed-loop control unit adjusts the centrifugal control strategy of subsequent batches of poles according to the abnormality type in the abnormal feedback report: when the abnormality type is insufficient local hardening of the pole, the running time of the low-speed centrifugal stage is extended or the rotation speed of the low-speed centrifugal stage is increased.
[0018] The beneficial effects of this invention are as follows: 1. This invention achieves quantitative real-time sensing of the physical state of concrete materials within a closed mold by constructing a multi-source signal fusion sensing and multi-dimensional material state characterization system. Specifically, this invention simultaneously collects electromagnetic drive signals, mold vibration signals, and acoustic signals from within the cavity during centrifuge operation to form a ternary mixed signal matrix. Blind source separation technology is then used to extract pure material state signal components, effectively eliminating equipment noise interference under strong mechanical disturbance conditions and solving the technical problems of signal mixing and feature extraction in traditional electrical signal monitoring schemes. Based on this, this invention performs multi-scale permutation entropy analysis on the material state signal components to extract characteristic parameters such as entropy change rate, entropy inflection point position, and entropy stability interval, achieving a quantitative characterization of the ordered evolution process of the internal structure of concrete. Furthermore, by performing micro-perturbation response analysis on the net load torque to calculate the equivalent dynamic viscosity, a quantitative description of the rheological properties of concrete is achieved. Finally, by performing frequency domain harmonic analysis on the net load torque to calculate the dimensionless eccentricity factor, a quantitative evaluation of the uniformity of material distribution is achieved. Therefore, this invention upgrades the traditional open-loop control mode, which relies on manual experience or fixed programs, to an adaptive closed-loop control mode based on real-time feedback of multi-dimensional material state parameters. This allows the timing of speed adjustment and stage switching in the centrifugal forming process to respond precisely to the dynamic changes in the material's internal state, significantly improving the centrifugal forming process's adaptability to uncertainties such as batch fluctuations in materials and changes in environmental conditions. 2. This invention constructs a multi-process linkage control mechanism for the entire process of centrifugal molding, curing, and demolding, which realizes cross-domain transmission and collaborative response of status information between processes, forming a closed-loop control network that contains at least multiple positive and negative correlations. Specifically, this invention transmits the material state information (including the final hardening state level, temperature distribution field, and eccentricity factor) of the centrifugal molding final state to the curing subsystem through a linkage control center. This allows the initial curing scheme to be personalized according to the actual material state at the centrifugal final state, solving the technical defects of starting the curing process in an information blind spot and being unable to adapt to material conditions. During the curing process, this invention continuously monitors the real-time hardening degree and temperature gradient inside the pole, and dynamically adjusts the heating rate, constant temperature, and constant temperature duration accordingly. This allows the curing parameters to adaptively compensate for the actual hardening process of the concrete, avoiding the problems of insufficient or excessive curing caused by a fixed curing scheme. After the final curing state assessment is completed, this invention transmits the final hardening degree and final temperature as pre-demolding parameters to the demolding subsystem. Through a comprehensive calibration coefficient, the demolding time, demolding speed, and demolding force are pre-calibrated, allowing the demolding operation to be adaptively adjusted according to the actual material state at the final curing state, reducing the risk of demolding damage caused by deviations in hardening degree. Therefore, this invention integrates the three originally separate processes of centrifugal molding, curing, and demolding into an organic whole through a linkage control center. The output status of the upstream process determines the input parameters of the downstream process, and the execution effect of the downstream process is fed back to the parameter optimization of the upstream process. This achieves integrated and coordinated quality control throughout the entire process and effectively improves the quality consistency of the entire concrete pole production process. Attached Figure Description
[0019] Figure 1 This is a schematic diagram of the overall system architecture of the present invention.
[0020] Figure 2 This is a schematic diagram of the centrifugal molding process control flow of the present invention.
[0021] Figure 3 This is a schematic diagram of the dynamic adjustment control process for maintenance parameters according to the present invention.
[0022] Figure 4 This is a schematic diagram of the demolding parameter calibration and dynamic adjustment process of the present invention. Detailed Implementation
[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0024] like Figures 1 to 4 The diagram shows an intelligent quality control system for the entire process of concrete utility poles.
[0025] System Initialization Before the system is put into use for the first time or after a major change to the equipment, the following initialization procedure shall be performed: The first step is hardware self-test. After the system is powered on, the linkage control center sends self-test commands to the centrifugal molding subsystem, curing subsystem, and demolding subsystem respectively. Each subsystem executes the following in sequence: sensor channel verification (current / voltage sensors, acceleration sensors, acoustic sensors, temperature sensors, tension sensors, displacement sensors, strain gauges, etc., confirming that each channel signal is normal and there are no open circuits or short circuits), actuator status verification (centrifuge frequency converter, curing kiln temperature control device, demolding traction device, auxiliary vibration device, etc., confirming that communication is normal and there are no fault alarms), and data acquisition verification (each sensor acquires known standard signals, confirming that the acquisition accuracy meets the requirements). After the self-test passes, each subsystem returns a "ready" signal to the linkage control center, and the system enters standby mode; if the self-test fails, the linkage control center issues an audible and visual alarm and displays the specific fault location on the human-machine interface. After the fault is resolved, the self-test is executed again.
[0026] The second step is communication link confirmation. A communication connection is established between the central control unit and each subsystem via industrial Ethernet (based on TCP / IP or Profinet protocol). The central control unit sequentially sends heartbeat packets (ping commands) to the three subsystems, and each subsystem replies with an acknowledgment frame upon receiving the heartbeat packet. A communication delay of no more than 10ms and a packet loss rate of no more than 0.1% are considered acceptable. After confirming that all communication links are normal, the system enters the operational ready state.
[0027] The third step is to check the baseline parameter status. The linkage control center checks whether the baseline parameter identification unit of the centrifugal molding subsystem has stored the empty mold impedance parameters of the current mold. If it is empty (first use), or the current mold model does not match the stored parameters, or the time since the last calibration exceeds the preset validity period (e.g., 90 days), or the mold has undergone mechanical maintenance (replacing bearings, support rollers, and other transmission components), the baseline parameter identification process is triggered.
[0028] I. Specific Implementation Methods of the Centrifugal Molding Subsystem (a) Implementation of the reference parameter identification unit The reference parameter identification unit is calibrated under no-load conditions after the centrifuge is first put into use or after the mold is replaced. The specific operation is as follows: The centrifuge is controlled to start from a standstill and gradually accelerate to the target speed of 500 rpm at a preset acceleration of 0.3~0.5 rpm / s. During the acceleration process, the stator torque current component I_q, mechanical angular velocity ω and angular acceleration α of the drive motor are continuously collected at a sampling frequency of not less than 100 Hz.
[0029] After collecting the data, substitute the data from each sampling moment into the rigid body dynamics equations: T_e = J·α + B·ω + T_c Where T_e = K_t·I_q, K_t is the motor torque constant (provided by the motor manufacturer, with a typical value of 0.8~1.5 N·m / A).
[0030] Using ω and α as independent variables and T_e as the dependent variable, a binary linear regression was performed using the least squares method to obtain three constant terms J, B, and T_c. To ensure fitting accuracy, at least 200 data points were collected, covering the entire speed range from start-up to full speed.
[0031] The fitted J, B, and T_c combinations are set as the unloaded mold impedance parameters and stored in the memory of the reference parameter identification unit for subsequent load operation. The unloaded mold impedance parameters of the same mold can be used continuously as long as the mold structure remains unchanged; however, if the mold has been repaired or key transmission components have been replaced, the unloaded calibration must be performed again.
[0032] (II) Implementation of the Hybrid Signal Acquisition Unit After the concrete is laid and the formwork is closed, the centrifuge is started simultaneously with the start of the mixed signal acquisition unit.
[0033] The current / voltage sensor is a Hall effect type sensor, installed on the three-phase lines of the drive motor power supply line, with a sampling frequency of 1kHz and a range of 1.5 times the rated current of the motor, to collect three-phase current and voltage signals.
[0034] The accelerometer is a piezoelectric accelerometer (model PCB 352C33 or equivalent) which is fixedly mounted on the centrifuge bearing housing by a magnetic base or bolts. The sampling frequency is set to 5kHz, the range is ±50g, and the resolution is 0.001g to collect vibration signals from the mold and centrifuge support structure.
[0035] The acoustic sensor uses a prepolarized capacitor microphone (model GRAS 46AE or equivalent) and is installed on the outside of the centrifuge base at a distance of about 0.3~0.8m from the outer wall of the mold. The sampling frequency is set to 10kHz and the range is 20~20kHz to collect the sound signals generated by the movement and collision of concrete materials in the mold cavity during centrifugation.
[0036] The three sensors collect data synchronously, covering the entire centrifugation process. The collected data are then aligned with timestamps to form a mixed signal matrix X(t) = [x1(t), x2(t), x3(t)]ᵀ, where x1 is the electromagnetic signal, x2 is the vibration signal, and x3 is the sound signal.
[0037] (III) Implementation of the Signal Separation and Feature Extraction Unit After the mixed signal acquisition is completed, the signal separation and feature extraction unit processes the mixed signal matrix X(t) as follows: The first step is preprocessing. Bandpass filtering is applied to each of the three signals, with a passband frequency range of 10Hz to 5kHz, to remove low-frequency drift (<10Hz) and high-frequency noise (>5kHz). Next, mean reduction is performed, subtracting the time average value from each signal to make the signal mean zero. Finally, variance normalization is performed to ensure that the amplitudes of all signals are on the same order of magnitude, facilitating the convergence of the subsequent blind source separation algorithm.
[0038] The second step is blind source separation. Independent component analysis (ICA) is used to perform blind source separation on the preprocessed mixed signal. Specifically, let the separated signal be Y(t) = [y1(t), y2(t), y3(t)]ᵀ, and solve for the unmixing matrix W (a 3×3 matrix) such that the components of Y(t) = W·X(t) are statistically independent. The unmixing matrix W is solved using the FastICA algorithm, with no more than 1000 iterations and a convergence accuracy of 1×10⁻⁶. -6 .
[0039] After separation, the physical source of each component is identified by its spectral characteristics: y1(t) is a signal component whose main frequency component is related to the centrifuge's rotation fundamental frequency and its harmonics, and is identified as a material state signal component; y2(t) is a signal component whose main frequency component is concentrated near the natural frequency of the mechanical equipment (usually 50~200Hz), and is identified as a mechanical vibration signal component; y3(t) is a signal component with a lower amplitude and a relatively flat spectrum, and is identified as an environmental noise signal component.
[0040] y1(t) is extracted as the effective signal for subsequent analysis, while the other two components are archived for future reference but are not included in subsequent calculations.
[0041] The third step is multi-scale entropy analysis. The material state signal component y1(t) (signal length N is typically 10,000–50,000 sampling points, depending on the total centrifugation time) undergoes multi-scale coarse-graining. For the scale factor s (s takes positive integers from 1 to 20), a coarse-grained sequence is constructed:
[0042] For each coarse-grained sequence at scale s, phase space reconstruction is performed with embedding dimension m=4 and delay time τ=1. The permutation pattern probability of each reconstructed vector is calculated, and then the permutation entropy value H_s at that scale is calculated:
[0043] Where p_i is the probability of the i-th permutation pattern occurring. The permutation entropy value H_s ranges from 0 to ln(m!), that is, 0 to ln(24)≈3.178. The closer the H_s value is to 0, the more regular (orderly) the signal is; the closer the H_s value is to 3.178, the more random (disorderly) the signal is.
[0044] For rapid-hardening concrete (such as early-strength cement), the value range of the scale factor s can be reduced to 1~12; for ordinary silicate concrete, the value range of the scale factor s remains 1~20.
[0045] The permutation entropy values at each scale are plotted as a multi-scale permutation entropy spectrum {H_s | s = 1, 2, ..., 20}. The following feature parameters are extracted from this spectrum: Entropy change rate ΔH / Δs: In the initial segment of the entropy spectrum (s=1~5), the slope of the linear fit is calculated, reflecting the evolution rate of the ordering of the internal structure of concrete, and the unit is bit / scale; Entropy inflection point location s_inflection point: The scale value corresponding to the turning point where H_s in the entropy spectrum changes from rapid decrease to gradual flattening, reflecting the transformation point of the internal structure of concrete from disorder to order; Entropy stability range: The continuous scale range in the entropy spectrum where H_s tends to be stable (entropy value change between adjacent scales < 0.05), reflecting the range in which the concrete compaction state meets the standard.
[0046] (iv) Implementation of the load torque observation unit The load torque observation unit operates continuously throughout the entire process of the centrifuge under load, calculating the net load torque in real time at an update frequency of no less than 100Hz.
[0047] The first step is to obtain the electromagnetic drive torque. The q-axis component I_q (unit: A) of the stator current is read in real-time from the drive motor controller and multiplied by the motor torque constant K_t (unit: N·m / A) to obtain the electromagnetic drive torque. T_e = K_t · I_q The second step is to calculate the theoretical mechanical torque. The current mechanical angular velocity ω (unit: rad / s) is obtained in real time from the speed sensor. The angular acceleration α (unit: rad / s) is obtained by numerically differentiating ω (using the three-point central difference method). 2 Based on the empty mode impedance parameters J, B, and T_c stored in the reference parameter identification unit, the theoretical mechanical torque required for the mold to overcome inertia, viscous friction, and Coulomb friction is calculated: T_m = J·α + B·ω + T_c The third step is to calculate the difference and apply compensation. Calculate the original difference between the electromagnetic drive torque and the theoretical mechanical torque: ΔT_raw = T_e - T_m To eliminate the influence of speed measurement noise and model errors, a compensation mechanism based on speed observation error is constructed. The measured value ω of the mechanical angular velocity is compared with the observed value ω_obs estimated based on the torque balance equation to obtain the speed observation error e_ω = ω - ω_obs. e_ω is input into a PI controller (proportional coefficient K_p = 10~50, integral coefficient Ki = 1~10) to generate the compensation value ΔT_comp.
[0048] The final net load torque is: T_net = ΔT_raw + ΔT_comp T_net is the effective torque (unit: N·m) acting solely on the concrete material after removing the mechanical consumption of the mold itself. Its magnitude and trend directly reflect the rheological properties and distribution state of the concrete material during the centrifugation process.
[0049] (v) Implementation of the Rheological and Distribution State Analysis Unit 1. Calculation of equivalent dynamic viscosity η The rheological and distributed state analysis unit performs micro-perturbation response analysis at a frequency of once every 30 to 120 seconds.
[0050] First, a sinusoidal wave signal is superimposed on the current base speed command ω_base of the centrifuge: ω_cmd(t) = ω_base + A·sin(2πf·t) The frequency f is 1~5Hz (preferably 2Hz), and the amplitude A is 1%~5% (preferably 3%) of the base speed command value, i.e. A = 0.01~0.05 × ω_base.
[0051] The duration of the sinusoidal oscillation signal is 5 to 15 seconds (preferably 10 seconds). During this period, the net load torque T_net(t) and mechanical angular velocity ω(t) are synchronously acquired at a sampling frequency of not less than 200Hz.
[0052] After the data acquisition is completed, high-pass filtering is performed on T_net(t) and ω(t) respectively, with the cutoff frequency set to 0.5f (i.e. 0.5~2.5Hz) to filter out DC components and low-frequency trend terms, and to extract the torque fluctuation component ΔT_f(t) and speed fluctuation component Δω_f(t) that are consistent with the frequency of the sinusoidal wave signal.
[0053] Within a preset time window (3~10 seconds, preferably 5 seconds), calculate the root mean square values of the two fluctuation components respectively:
[0054]
[0055] Where M is the number of sampling points within the time window.
[0056] Calculate the ratio R = RMS(ΔT_f) / RMS(Δω_f), with units of N·m·s / rad.
[0057] Obtain the inner radius r (in meters) and effective length L (in meters) of the current mold, and calculate the geometric constants: G = 2π·r 2 ·L This geometric constant is derived from the application of Newton's law of internal friction in annular flow channels. For a given mold, G is a fixed value.
[0058] Dividing the ratio R by the geometric constant G yields the equivalent dynamic viscosity: η = R / G The unit of η is Pascal-second (Pa·s). The higher the value of η, the greater the flow resistance of the concrete material (i.e., the more viscous and the worse the fluidity); the lower the value of η, the thinner the material and the better the fluidity.
[0059] The suitable range of equivalent dynamic viscosity η was determined through the following calibration test: C50 concrete mixtures with different water-cement ratios (0.32, 0.35, 0.38, 0.41, 0.44) were used for centrifugal molding tests under standard centrifugation conditions. Net load torque was collected and the η value was calculated. The test results showed that when η > 600 Pa·s (corresponding to a water-cement ratio ≤ 0.32), the material flowability was insufficient, and the proportion of honeycomb and void defects in the poles after centrifugation reached 15.2%; when η < 400 Pa·s (corresponding to a water-cement ratio ≥ 0.44), the material was too thin, and the proportion of aggregate segregation and delamination defects in the poles after centrifugation reached 12.8%; when η was in the range of 400~600 Pa·s (corresponding to a water-cement ratio of 0.35~0.41), the appearance and internal quality of the poles were both qualified, and the defect rate was less than 2.0%. Based on the above test results, the suitable range of η was determined to be 400~600 Pa·s, with an optimal value of 500 Pa·s.
[0060] 2. Calculation of the dimensionless eccentricity factor ε The rheological and distribution state analysis unit performs frequency domain feature extraction every 10 to 30 seconds.
[0061] Take the net load torque T_net(t) data within the time window of 30-120 seconds before the current time (sampling frequency not less than 100Hz), and perform a Fast Fourier Transform:
[0062] Where N is the number of sampling points within the time window.
[0063] Extract the amplitudes A1, A2, A3, and A4 corresponding to the centrifuge's fundamental rotation frequency f0 (f0 = ω / (2π), unit: Hz) and its 2nd, 3rd, and 4th harmonics from the spectrum.
[0064] The fundamental frequency amplitude A1 reflects the basic load level of the centrifuge rotation; the harmonic amplitudes A2, A3, and A4 reflect the periodic load fluctuations caused by mass eccentricity due to uneven material distribution. The larger the harmonic amplitude relative to the fundamental frequency amplitude, the more severe the eccentricity.
[0065] Calculate the dimensionless eccentricity factor:
[0066] Where w2, w3, and w4 are preset weighting coefficients, satisfying w2 + w3 + w4 = 1. Preferably, w2 is 0.4~0.6 (preferably 0.5), w3 is 0.2~0.4 (preferably 0.3), and w4 is 0.1~0.3 (preferably 0.2), because the second harmonic has the most direct and significant impact on eccentricity, so it is given a higher weight.
[0067] ε is a dimensionless quantity. The closer its value is to 0, the more uniform the material distribution; the larger the value of ε, the more severe the eccentricity.
[0068] The determination of ε_max is based on the following: Centrifugal forming models of electric poles with different degrees of eccentricity were established through finite element simulation, and the eccentricity coefficient of the dynamic load on the mold support points corresponding to different ε values was calculated. Simulation results show that: when ε=0.12, the eccentricity coefficient is 1.38 (i.e., the load on one support point is 38% higher than the other); when ε=0.15, the eccentricity coefficient is 1.52; and when ε=0.18, the eccentricity coefficient is 1.68. Simultaneously, statistical analysis of actual production data shows that when ε>0.15, the centrifuge bearing failure rate is 3.2 times higher than when ε≤0.15, and the proportion of electric pole wall thickness deviations exceeding the standard allowable value reaches 9.5%. Considering both equipment safety and product quality, the range of ε_max is determined to be 0.12~0.18, with 0.15 being the preferred value.
[0069] (vi) Implementation of the Hardening Status Assessment Unit The hardening state assessment unit evaluates the internal density of concrete at a frequency of once every 30 to 120 seconds, based on the multi-scale entropy spectrum characteristic parameters output by the signal separation and feature extraction unit.
[0070] The evaluation rules are as follows: When the rate of change of entropy |ΔH / Δs| > 0.15 bit / scale, the internal structure of the concrete is determined to be in a rapid ordering stage (dense process). When the rate of change of entropy |ΔH / Δs| ≤ 0.05 bit / scale and the stable entropy range continuously covers at least 5 scales (i.e., the change of H_s in the range of s=16~20 is <0.05), it is determined that the internal structure of concrete has become stable (compaction is basically completed). When the rate of change of entropy is between 0.05 and 0.15 bit / scale, the concrete is considered to be in a moderately dense state.
[0071] Based on the above judgment results, a hardening state level H is generated, with the value of H ranging from 0 to 1. H < 0.3: Not compacted (material is still in a fluid state); 0.3 ≤ H < 0.6: Partially compacted (material begins to form a skeletal structure); 0.6 ≤ H < 0.85: Moderately dense (structure is forming but not yet stable); H ≥ 0.85: Compaction meets the standard (the structure is stable and can proceed to the next stage).
[0072] Meanwhile, the temperature of each measuring point is collected in real time by distributed temperature sensors (no less than 6 measuring points, evenly distributed along the pole axis) embedded in the inner wall of the mold, and the centrifugal final state temperature distribution field T_dist = {T1, T2, ..., T6} (unit: ℃) is constructed as the input parameter for subsequent maintenance procedures.
[0073] (vii) Implementation of the adaptive closed-loop control unit The adaptive closed-loop control unit generates corresponding control commands every 5 to 30 seconds based on the parameters output by the rheological and distribution state analysis unit and the hardening state assessment unit.
[0074] The control strategy is as follows: Strategy 1 (Emergency Braking): When the dimensionless eccentricity factor ε > ε_max (ε_max is 0.12~0.18, preferably 0.15), it indicates that the material distribution is severely uneven, and continued rotation may cause severe vibration or even damage to the equipment. At this time, an emergency braking command is output to control the centrifuge to decelerate at the maximum safe speed (not exceeding -0.5 rpm / s). 2 It will quickly stop the machine and issue an audible and visual alarm signal.
[0075] Strategy Two (Constant-rate Shear): When the equivalent dynamic viscosity η > η_max (η_max is 400~600 Pa·s, preferably 500 Pa·s), it indicates that the concrete material is too viscous, lacks fluidity, and has not been fully liquefied, which is not conducive to uniform distribution under centrifugal force. At this time, a constant-rate shear command is output to maintain the current speed for a preset duration (30~90 seconds, preferably 60 seconds), so that the material gradually liquefies and reduces viscosity under continuous shearing action.
[0076] Strategy 3 (Adaptive Acceleration): When the equivalent dynamic viscosity η ≤ η_max and the dimensionless eccentricity factor ε ≤ ε_max, it indicates that the material has been fully liquefied and its distribution uniformity is within a safe range, meeting the conditions for acceleration. At this point, an adaptive acceleration command is output to accelerate to the next stage target speed at a preset acceleration (0.3~0.8 rpm / s, preferably 0.5 rpm / s). The target speed is read from a preset centrifugation rate table based on the pole specifications.
[0077] Strategy 4 (Centrifugal Termination): When the hardening state grade H ≥ 0.85 and the duration of this state exceeds 60 seconds, it indicates that the concrete has been sufficiently compacted, and the marginal benefit of continuing rotation for quality improvement approaches zero. At this point, a centrifugal termination command is output to control the centrifuge to decelerate and stop, ending the centrifugal molding process.
[0078] (viii) Centrifugal force of poles of different specifications For concrete poles of different specifications, the preset centrifugal force regime is shown in the table below. The setting of the rotation speed and duration for each stage is based on the following: the low-speed stage allows the concrete to be initially spread at the bottom of the mold; the medium-speed stage allows the concrete to be evenly distributed along the inner wall of the mold; the medium-high speed stage allows the concrete to begin to compact under the action of centrifugal force; and the high-speed stage completes the final compaction and molding.
[0079]
[0080] The above centrifugation regime is the initial setting value. The adaptive closed-loop control unit can dynamically adjust the target speed within ±15% of each stage to adapt to the actual rheological characteristics of different batches of concrete materials.
[0081] (ix) Calibration test basis for key parameters The values of key parameters in this system were determined based on the following calibration tests: 1. Calibration test for suitable range of equivalent dynamic viscosity η C50 concrete mixtures with different water-cement ratios (0.32, 0.35, 0.38, 0.41, 0.44) were used for centrifugal molding tests under standard centrifugal conditions. For each water-cement ratio, 30 Φ190mm×12m poles were fabricated. Net load torque was collected, and the η value was calculated. The test results are as follows:
[0082] Experimental results show that when η > 600 Pa·s, the honeycomb and void defect rate reaches 15.2%; when η < 400 Pa·s, the aggregate segregation and delamination defect rate reaches 12.8%; and when η is in the range of 400~600 Pa·s, the defect rate is less than 2.0%. Therefore, η_max = 600 Pa·s is determined as the upper limit threshold, 400 Pa·s as the lower limit reference value, and 500 Pa·s as the preferred control target.
[0083] 2. Calibration test of temperature gradient safety threshold G_max In the curing kiln, Φ190mm×12m poles were cured using different heating rates (5, 10, 15, 20, 25℃ / h). 30 poles were cured under each heating rate condition. The temperature gradient G(t) in the wall thickness direction was monitored throughout the process. After curing, surface crack detection was performed on all poles (using dye penetrant testing).
[0084] Experimental results show that when G ≤ 2.0℃ / cm, the crack incidence rate is 1.7%; when G = 2.0~3.0℃ / cm, the crack incidence rate is 3.2%; when G = 3.0~4.0℃ / cm, the crack incidence rate rises to 9.4%; and when G > 4.0℃ / cm, the crack incidence rate reaches as high as 18.5%. Considering both production efficiency and product quality, the range of G_max is determined to be 2.0~3.0℃ / cm, with an optimal value of 2.5℃ / cm.
[0085] 3. Calibration test of the eccentric safety threshold ε_max Centrifugal forming models with different degrees of eccentricity (ε=0.04, 0.08, 0.12, 0.15, 0.18, 0.22) were established using finite element simulation, and the dynamic load eccentricity coefficient of the mold support points under each working condition was calculated. Simultaneously, statistical analysis was performed on actual production data to determine the centrifuge bearing failure rate and pole wall thickness deviation rate corresponding to each ε value range.
[0086]
[0087] Taking into account both equipment safety and product quality, the value range of ε_max is determined to be 0.12~0.18, with an optimal value of 0.15.
[0088] II. Specific Implementation Methods of the Maintenance Subsystem (a) Reception and parsing of linkage data packets After the centrifugal molding process is completed, the forward linkage control module of the linkage control center encapsulates the centrifugal final state data into a linkage data packet and sends it to the maintenance subsystem via industrial Ethernet or fieldbus.
[0089] The linked data packet contains the following fields: Pole type and specifications (e.g., Φ190mm×12m prestressed pole). Centrifugation final hardening state level H_final (a value between 0 and 1); Centrifugal final-state temperature distribution field T_dist = {T1, T2, ..., T6} (unit: °C); The centrifugal final-state eccentricity factor ε_final; Records of actual rotation speed and duration at each stage of centrifugal molding.
[0090] The maintenance status receiving unit receives and parses the linkage data packet, stores the data of each field into the local database for subsequent units to call.
[0091] (II) Implementation of the initial generation unit of the maintenance system The initial maintenance regime generation unit matches an initial maintenance scheme from a preset maintenance rule base based on the received centrifugation final hardening state level H_final and temperature distribution field T_dist.
[0092] The pre-set maintenance rule base was constructed based on a large amount of process test data, and the optimal maintenance regime was calibrated for different pole specifications, different concrete mix proportions, and different combinations of centrifugal final hardening state levels. The rule base is stored in the form of lookup tables.
[0093] The matching logic is as follows: For the hardening state level H_final: If H_final ≥ 0.85 (compactness meets the standard), it indicates that the centrifugal molding quality is excellent and the concrete has a high initial density. The standard curing system can be adopted: heating rate 10~15℃ / h, constant temperature 55~65℃, constant temperature duration 4~6h, cooling rate 15~20℃ / h. If 0.60 ≤ H_final < 0.85 (medium density), it indicates that the density of centrifugal molding is slightly insufficient, and a gentler heating is needed to avoid thermal shock. At the same time, the isothermal time should be appropriately extended to compensate for insufficient hardening: heating rate 6~10℃ / h, isothermal temperature 55~65℃, isothermal time 5~8h, cooling rate 12~18℃ / h. If H_final < 0.60 (partially dense), it indicates that the density of the centrifugal molding is seriously insufficient, and the curing regime needs to be significantly adjusted: heating rate 4~8℃ / h, constant temperature 50~60℃, constant temperature duration 7~10h, cooling rate 10~15℃ / h.
[0094] For the temperature distribution field T_dist: Calculate the maximum temperature difference ΔT_max in the temperature distribution field: ΔT_max = max(T_dist) - min(T_dist); If ΔT_max > 15℃, it indicates that the internal temperature distribution of the pole is extremely uneven. The heating rate should be reduced to the lower limit (e.g., 6℃ / h) and the heating time should be extended to avoid thermal stress causing cracks. If ΔT_max ≤ 15℃, the standard heating rate can be used.
[0095] For the eccentricity factor ε_final: If ε_final > 0.10, it indicates that the pole has a certain mass eccentricity. During the maintenance process, the constant temperature should be appropriately reduced (by 2~5℃) to reduce the amplification effect of thermal stress on the eccentricity.
[0096] The initial maintenance regime generation unit generates an initial maintenance plan based on the above matching logic, including: initial heating rate (°C / h), initial constant temperature (°C), initial constant temperature duration (h), and initial cooling rate (°C / h).
[0097] (III) Implementation of the Maintenance Process Monitoring Unit The maintenance process monitoring unit operates continuously throughout the entire maintenance process, and the monitoring methods include: Internal temperature monitoring of the pole: The temperature of each measuring point inside the pole is collected once per minute by thermocouples (arranged at least 3 measuring points along the pole axis and at least 2 measuring points along the wall thickness direction) pre-embedded in the concrete before centrifugal molding. Pole surface temperature monitoring: The surface temperature of the pole is collected once per minute (at no less than 3 measuring points) using an infrared thermometer or a patch-type resistance temperature detector. Temperature and humidity monitoring of the curing environment: The ambient temperature and relative humidity inside the curing kiln are collected once per minute using temperature and humidity sensors.
[0098] Based on the collected data, the maintenance process monitoring unit calculates the following two key indicators in real time: Real-time hardening degree S(t): Based on concrete maturity theory, the time-temperature history during curing is converted into an equivalent age. Then, the degree of concrete strength development at the current moment is calculated according to a pre-calibrated strength-maturity relationship curve, expressed as a value between 0 and 1, where 0 represents unhardened and 1 represents fully hardened. The maturity calculation formula uses the Nurse-Saul function:
[0099] Where T_a is the actual temperature of the concrete (°C), T_0 is the reference temperature (usually -10°C), and Δt is the time interval.
[0100] Real-time temperature gradient G(t): Calculates the ratio of the temperature difference between the inner and outer measuring points along the pole's wall thickness to the wall thickness.
[0101] Where T_inside is the internal measuring point temperature, T_outside is the external surface temperature, and d is the pole wall thickness (unit: cm). The unit of G(t) is ℃ / cm.
[0102] (iv) Implementation of the dynamic adjustment unit for maintenance parameters The curing parameter dynamic adjustment unit adjusts the curing parameters dynamically every 5 to 20 minutes (preferably 10 minutes) based on the deviation between the real-time hardening degree S(t) and the expected hardening curve S_target(t), as well as the magnitude of the real-time temperature gradient G(t).
[0103] The expected hardening curve S_target(t) is pre-calibrated based on the pole specifications and concrete mix proportions, and stored in the curing subsystem as a lookup table or function curve. This curve reflects the ideal trajectory of concrete strength development over time under standard curing conditions.
[0104] The adjustment strategy is as follows: Strategy 1 (Compensation for Slow Hardening Speed): Calculate the deviation ΔS(t) = S_target(t) - S(t). When ΔS(t) > the preset positive deviation threshold (0.03~0.08, preferably 0.05), it indicates that the actual hardening speed is significantly slower than expected, and acceleration measures need to be taken. Increase the current heating rate by 5% to 15% (preferably 10%) to accelerate the hydration reaction rate; or Increase the current constant temperature by 1~3℃ (preferably 2℃) to increase the activation energy of the hydration reaction.
[0105] Strategy 2 (Compensation for Overly Fast Hardening Speed): When ΔS(t) < the preset negative deviation threshold (-0.08~-0.03, preferably -0.05), it indicates that the actual hardening speed is significantly faster than expected, and deceleration measures need to be taken. Reduce the current heating rate by 5% to 15% (preferably 10%) to slow down the hydration reaction rate; or The current constant temperature is lowered by 1~3℃ (preferably 2℃) to reduce the activation energy of the hydration reaction.
[0106] Strategy 3 (Temperature Gradient Suppression): When the real-time temperature gradient G(t) > the preset safety threshold G_max (2.0~3.0℃ / cm, preferably 2.5℃ / cm), it indicates that the temperature difference in the pole wall thickness direction is too large, posing a risk of thermal cracking. The determination of G_max is based on the calibration test in Section (IX) of Chapter 1: Pause the current heating operation and maintain the current constant temperature; Extend the duration of the current temperature hold-up phase (by 10-30 minutes each time). Re-evaluate G(t) every 10 minutes until G(t) ≤ G_max, then continue heating at a reduced rate (reduced to 50%~70% of the original rate).
[0107] (v) Implementation of the final state assessment unit for maintenance After the maintenance enters the cooling stage, when the internal temperature of the pole drops to the range of 40~50℃, the maintenance final state assessment unit performs the final state assessment.
[0108] Evaluation steps: The first step is to determine the final hardening degree, S_final_cure. Based on the temperature-time records throughout the curing process, the final strength development degree is estimated using the maturity method, or the actual strength value is obtained through compressive strength tests of test blocks cured under the same conditions (concrete test blocks cured together with the pole), and converted into a hardening degree value between 0 and 1. According to GB / T 4623, the demolding strength of prestressed concrete poles should not be lower than 70% of the design strength grade. For prestressed poles with a design strength grade of C50, the demolding strength should not be lower than 35 MPa, and the corresponding hardening degree, S_final_cure, should reach 0.70 or higher.
[0109] The second step is to determine the final temperature T_final_cure. The average temperature of multiple measuring points inside the pole is taken as the final temperature. The optimal demolding temperature range is usually 30~45℃. If the temperature is too high, thermal cracks are likely to occur on the surface of the pole after demolding. If the temperature is too low, the adhesion between the concrete and the mold increases, making demolding more difficult.
[0110] The third step is to calculate the demolding readiness index R_demold: R_demold = 0.6 × S_final_cure + 0.4 × f(T_final_cure) Where f(T_final_cure) is the temperature suitability function: when T_final_cure is in the range of 30~45℃, f=1; when T_final_cure is in the range of 45~55℃ or 20~30℃, f=0.5; and in other temperature ranges, f=0.
[0111] When R_demold ≥ the preset demolding threshold (0.75~0.90, preferably 0.85), it is determined that the demolding condition is met.
[0112] (vi) Implementation of the first linkage triggering unit When the final state assessment unit determines that R_demold ≥ the preset demolding threshold, the first linkage trigger unit performs the following operations: A demolding start signal is generated and sent to the demolding subsystem through the positive linkage control module of the linkage control center; The final curing degree S_final_cure and the final temperature T_final_cure are used as pre-demolding parameters, packaged and sent to the demolding subsystem. Record complete data of this maintenance process (including temperature-time curves for each stage, adjustment operation records, etc.) and store them in the database for subsequent analysis.
[0113] III. Specific Implementation Methods of the Demolding Subsystem (a) Implementation of the demolding parameter pre-calibration unit After receiving the pre-demolding parameters from the curing subsystem, the demolding parameter pre-calibration unit performs the following calibration procedure: The first step is to obtain the input parameters: Final curing degree S_final_cure (value range 0~1); Final temperature T_final_cure (unit: °C); The centrifugal final state eccentricity factor ε_final (obtained from the linkage data packet and stored in the linkage control center database).
[0114] The second step is to calculate the deviations for each item: Obtain the target curing hardening degree S_target_cure (determined according to the pole design strength grade; for C50 grade prestressed poles, S_target_cure is taken as 0.75~0.85) and the median value of the optimal demolding temperature range T_target_demold (usually taken as 35~40℃).
[0115] Calculate the hardening degree deviation: ΔS = S_target_cure - S_final_cure ΔS > 0 indicates that the actual hardening degree is lower than the target value (insufficient hardening), and ΔS < 0 indicates that the actual hardening degree is higher than the target value (over-hardening).
[0116] Calculate temperature deviation: ΔT = T_final_cure - T_target_demold ΔT > 0 indicates that the actual temperature is higher than the optimal demolding temperature, and ΔT < 0 indicates that the actual temperature is lower than the optimal demolding temperature.
[0117] Calculate the relative eccentricity: ε_rel = ε_final / ε_max Where ε_max is the eccentricity safety threshold (0.12~0.18, preferably 0.15), and ε_rel reflects the relative severity of eccentricity.
[0118] The third step is to calculate the overall calibration coefficient:
[0119] Wherein α, β, and γ are preset weighting coefficients, satisfying α + β + γ = 1. Preferably, α is 0.4~0.6 (the degree of hardening has the greatest impact on demolding), β is 0.15~0.35, and γ is 0.2~0.4.
[0120] λ is a dimensionless comprehensive calibration coefficient, which can be positive or negative: λ > 0 indicates that the demolding time needs to be extended, the demolding speed needs to be reduced, or the demolding force needs to be reduced (conservative demolding); λ < 0 indicates that the demolding time can be shortened, the demolding speed needs to be increased, or the demolding force needs to be increased (active demolding).
[0121] Step 4: Calibrate the demolding parameters: Read the initial demolding parameters from the demolding parameter database: initial demolding time t0 (usually 120~300 seconds), initial demolding speed v0 (usually 0.3~0.8 m / min), and initial demolding force F0 (usually 15~30 kN).
[0122] Calibration is performed using the comprehensive calibration coefficient λ: Calibrate demolding time: t_demold = t0 × (1 + k_t·λ), where k_t is a preset time calibration coefficient, which is 0.2~0.5; Calibrate demolding speed: v_demold = v0 × (1 - k_v·λ), where k_v is the preset speed calibration coefficient, which is 0.3~0.6; Calibrate demolding force: F_demold = F0 × (1 - k_F·λ), where k_F is the preset force calibration coefficient, which is 0.2~0.5.
[0123] The design logic for the sign of the calibration coefficient is as follows: when λ>0 (insufficient hardening or large eccentricity), the demolding time is prolonged (1+k_t·λ>1), the demolding speed is reduced (1-k_v·λ<1), the demolding force is reduced (1-k_F·λ<1), and the overall trend is conservative; when λ<0 (excessive hardening or small eccentricity), the demolding time is shortened, the demolding speed is increased, the demolding force is increased, and the overall trend is positive.
[0124] The calibrated demolding parameters {t_demold, v_demold, F_demold} are sent to the demolding actuator.
[0125] (II) Implementation of the demolding process monitoring unit The demolding process monitoring unit works continuously during the demolding process, and the monitoring methods include: Demolding force monitoring: Install a tension sensor (range 0~50kN, accuracy ±0.5%FS) on the demolding traction device to collect the demolding force F(t) in real time at a sampling frequency of not less than 100Hz; Demolding displacement monitoring: Install a displacement sensor (range 0~2m, accuracy ±0.1mm) on the demolding traction device to collect the demolding displacement D(t) in real time at a sampling frequency of not less than 100Hz; Surface strain monitoring of utility poles: Strain gauges are attached to key locations (at least 3 measuring points, evenly distributed along the axial direction) on the outer surface of the utility pole, and surface strain data ε_strain(t) is collected at a sampling frequency of not less than 100Hz.
[0126] Calculation of real-time adhesive force B(t): In the initial stage of demolding (displacement D < D_initial, where D_initial is typically 50~100mm), the demolding force is mainly used to overcome the static adhesion between the pole and the inner wall of the mold. The average demolding force F_avg in this stage is calculated as a characterization of the static adhesion force.
[0127] During the demolding continuation stage (displacement D ≥ D_initial), the real-time adhesive force B(t) is calculated as follows: B(t) = F(t) - F_friction(D(t)) F_friction(D(t)) is the frictional resistance component estimated based on the displacement D(t), calculated based on a pre-calibrated friction model (the relationship curve between frictional force and displacement).
[0128] Calculation of the rate of change of demolding resistance ΔR(t): Within a preset time window (usually 2-5 seconds), calculate the rate of change of the demolding force F(t):
[0129] When ΔR(t) exceeds the preset threshold (usually 2~5kN / s) for three consecutive sampling periods, it is determined that the demolding resistance has an abnormally increasing trend.
[0130] (III) Implementation of the Demolding Dynamic Adjustment Unit The demolding dynamic adjustment unit executes the following dynamic adjustment strategy based on the real-time adhesive force B(t) and the rate of change of demolding resistance ΔR(t): Strategy 1 (Handling of Overly Tight Bonding): When the real-time bonding force B(t) > B_max (B_max is 20~30kN, preferably 25kN), it is determined that the pole is too tightly bonded to the inner wall of the mold, and the following operations are performed: Reduce the current demolding speed to 50%~80% of the original speed; Pause demolding for 2-5 seconds, then start the auxiliary vibration device (vibrator installed on the outer wall of the mold, frequency 30-80Hz, amplitude 0.3-1.0mm) for 3-8 seconds; After demolding is restored, continue the process at a reduced speed while closely monitoring changes in adhesive strength; If the adhesion remains too high, repeat the above operation, reducing the speed by 10% to 20% each time, until the adhesion returns to a safe range.
[0131] Strategy 2 (Abnormal Demolding Resistance Handling): When the rate of change of demolding resistance ΔR(t) continuously increases within 3 seconds and exceeds the preset threshold (2~5kN / s), it is determined that there is a risk of demolding jamming. Immediately trigger the warning signal (audio and visual alarm); Reduce the demolding speed to 40%~60% of the current speed; If ΔR(t) continues to increase, suspend demolding and notify the operator to check.
[0132] Strategy 3 (Loose Bond Record): When the real-time bond force B(t) < B_min (B_min is 8~15kN, preferably 12kN) and the demolding displacement D(t) > the preset normal demolding displacement D_normal (usually 200~500mm, depending on the pole length), the bond is judged to be abnormally loose. Record the status and the corresponding batch's process parameters; This information is sent to the maintenance subsystem via the reverse linkage control module of the linkage control center, serving as a reference for reducing maintenance energy consumption in subsequent batches.
[0133] (iv) Implementation of the demolding anomaly diagnosis unit When the demolding dynamic adjustment unit triggers an early warning or the demolding process monitoring unit detects the following abnormalities, the demolding abnormality diagnosis unit initiates diagnosis: Abnormal triggering conditions: The demolding force F(t) continues to exceed the preset upper limit F_max (35~45kN) for more than 5 seconds; or, the surface strain ε_strain(t) of the pole exceeds the preset strain threshold (500~1000με); or, visible cracks or surface damage appear on the pole during demolding.
[0134] Diagnostic process: The first step is data collection. Collect complete data within 10 seconds before and after the anomaly occurred: Demolding force F_abnormal at abnormal moments (unit: kN); Demolding displacement D_abnormal at abnormal moments (unit: mm); Strain data ε_strain of pole surface at abnormal moment (unit: με); The position information of the inner wall of the mold at the time of the abnormality (obtained from the displacement-position mapping relationship).
[0135] The second step is anomaly type determination. The demolding force F_abnormal is compared with the preset upper and lower limits of the adhesive force B_max and B_min: If F_abnormal > B_max, the abnormality type is determined to be an over-adhesion abnormality. If F_abnormal < B_min and demolding displacement D_abnormal > preset normal demolding displacement D_normal, the abnormality type is determined to be abnormally loose adhesion; If F_abnormal is within the normal range but ε_strain exceeds the limit, the abnormality type is determined to be surface damage abnormality.
[0136] The third step is to diagnose the cause of the anomaly. Based on the type of anomaly and the collected data, possible causes are diagnosed:
[0137] The fourth step is to generate an anomaly feedback report. The report content includes: Pole model and batch number; Exception type; Abnormal location (specific location on the inner wall of the mold, accuracy ±50mm); Severity of abnormality (mild / moderate / severe); Preliminary diagnosis of the cause; Recommended corrective measures.
[0138] (v) Implementation of the second linkage triggering unit After the demolding anomaly diagnosis unit generates an anomaly feedback report, the second linkage trigger unit performs the following reverse optimization operation through the reverse linkage control module of the linkage control center: Reverse linkage to the maintenance subsystem: The anomaly feedback report is sent to the maintenance parameter dynamic adjustment unit. The maintenance parameter dynamic adjustment unit adjusts the maintenance parameters for subsequent batches of poles of the same specification based on the anomaly type. If the abnormality type is "bonding too tight", it indicates that the bonding force between the concrete and the inner wall of the mold is too large after curing. It is recommended to reduce the constant temperature of subsequent batches by 2-5℃ or shorten the constant temperature time by 0.5-1.5 hours. If the abnormality type is insufficient local hardening of the pole, it indicates that the curing temperature distribution is uneven or the total curing time is insufficient. It is recommended to increase the constant temperature of subsequent batches by 2-5℃ or extend the constant temperature time by 1-2 hours. If the anomaly type is surface damage and the diagnosis is excessively fast demolding speed, it is recommended to reduce the initial demolding speed setting in subsequent batches.
[0139] Reverse linkage to the centrifugal forming subsystem: The anomaly feedback report is sent to the adaptive closed-loop control unit. The adaptive closed-loop control unit adjusts the centrifugal control strategy for subsequent batches of poles of the same specification based on the anomaly type. If the anomaly type is insufficient local hardening of the pole and the anomaly location corresponds to a specific section of the pole, it indicates that the material distribution or density of that section is insufficient during centrifugal molding. It is recommended to extend the centrifugal stage time corresponding to that section by 10% to 20%, or increase the rotation speed of that stage by 5% to 10%. If the abnormality type is excessive adhesion and the abnormality location corresponds to a specific area on the inner wall of the mold, in addition to adjusting the curing parameters, it is also recommended to appropriately extend the low-speed centrifugation time (extend by 0.5~1 minute) during the centrifugal molding stage to improve the interface structure of the concrete in that area.
[0140] IV. Specific Implementation Methods of the Linkage Control Center The linkage control center, serving as the core hub connecting the three subsystems of centrifugal molding, curing, and demolding, is implemented as follows: (I) Implementation of the forward linkage control module The forward linkage control module is responsible for executing the information transmission between "upstream" and "downstream", specifically including two transmission chains: Transfer chain one (centrifugation, curing): After the centrifugal forming subsystem completes centrifugal forming, the forward linkage control module performs the following operations: The final hardening state level H_final is obtained from the hardening state evaluation unit of the centrifugal molding subsystem; The final temperature distribution field T_dist is obtained from the hardening state evaluation unit of the centrifugal molding subsystem; The final-state eccentricity factor ε_final is obtained from the rheological and distribution state analysis unit of the centrifugal molding subsystem; The above three data items, along with information such as pole model, batch number, and actual parameters of each stage of centrifugal forming, are packaged into a linkage data package; The linkage data packets are sent to the maintenance status receiving unit of the maintenance subsystem via industrial Ethernet. Record the sending time and data packet content, and store them in the linked log database.
[0141] Transfer chain two (curing, demolding): After the curing subsystem completes curing and determines that the conditions for demolding are met, the forward linkage control module performs the following operations: The final hardening degree S_final_cure is obtained from the final state assessment unit of the maintenance subsystem; The final temperature T_final_cure is obtained from the final state assessment unit of the maintenance subsystem; Retrieve the centrifugal final state eccentricity factor ε_final for the corresponding batch from the linkage log database; The above three data items are encapsulated into a pre-demolding parameter data package; The pre-demolding parameter data packet is sent to the demolding parameter pre-calibration unit of the demolding subsystem via industrial Ethernet. Record the sending time and data packet content, and store them in the linked log database.
[0142] (II) Implementation of the reverse linkage control module The reverse linkage control module is responsible for executing information feedback between "downstream" and "upstream", specifically including the following feedback paths: Feedback Path 1 (Demolding, Curing): After the demolding subsystem completes the demolding anomaly diagnosis, the reverse linkage control module performs the following operations: Receive anomaly feedback reports from the demolding anomaly diagnosis unit; Analyze the anomaly types and preliminary diagnoses of causes in the anomaly feedback reports; Send the anomaly feedback report to the maintenance parameter dynamic adjustment unit; The dynamic adjustment unit for maintenance parameters automatically generates maintenance regime adjustment suggestions (such as constant temperature adjustment value, constant temperature duration adjustment value, etc.) based on the type of abnormality. The adjustment recommendations will take effect after being confirmed by the operator (or automatically executed by the system after authorization) and will be applied to the maintenance and production of subsequent batches of poles of the same specifications. Record the time, content, and basis of the adjustment operation, and store it in the linkage log database.
[0143] Feedback path two (demolding, centrifugation): After the demolding subsystem completes the demolding anomaly diagnosis, the reverse linkage control module simultaneously performs the following operations: Send the anomaly feedback report to the adaptive closed-loop control unit; The adaptive closed-loop control unit automatically generates centrifugation strategy adjustment suggestions (such as speed adjustment values and duration adjustment values for specific stages) based on the anomaly type and location. The adjustment recommendations will take effect after being confirmed by the operator (or automatically executed by the system after authorization) and will be applied to the centrifugal molding production of subsequent batches of poles of the same specifications. Record the time, content, and basis of the adjustment operation, and store it in the linkage log database.
[0144] (III) Implementation of the Linked Data Management Unit The linkage control center includes a linkage data management unit, which is responsible for the unified storage, management and traceability of data throughout the entire process.
[0145] Data table structure: The linked log database uses a relational database (such as MySQL or PostgreSQL), and the main data table contains the following fields:
[0146] Data storage period: The raw data storage period in the linked log database shall be no less than 1 year, and the storage period for summarized statistical data and optimization model parameters shall be no less than 3 years. Raw data exceeding the storage period may be transferred to offline storage media, but summarized data and key batch data shall be permanently retained.
[0147] Data Traceability and Query: The linked data management unit provides data query interfaces based on multiple dimensions such as batch number, pole specifications, time range, and anomaly type. Operators can input query conditions through the human-machine interface, and the system returns the corresponding complete linked data chain. It also supports exporting data reports by specification and time period for offline analysis and quality auditing.
[0148] (iv) Implementation of multi-batch linkage optimization units The multi-batch linkage optimization unit performs the following cross-batch optimization process, with each analysis cycle consisting of 50 to 200 production batches: The first step is data aggregation. Complete data for all batches within the analysis period is extracted from the linked log database, including: Centrifugation parameters for each batch (speed, duration, eccentricity factor ε, equivalent dynamic viscosity η, etc. at each stage); Curing parameters for each batch (heating rate, constant temperature, constant temperature duration, cooling rate, temperature gradient G, etc.); Demolding parameters for each batch (demolding time, demolding speed, demolding force, real-time adhesion force B, etc.); Product quality inspection results for each batch (appearance defects, dimensional deviations, mechanical properties, etc.).
[0149] The second step is association analysis. Machine learning methods such as multiple linear regression or random forest are used to build a multivariate association model. Quality = f(ε, η, H_final, S_final_cure, T_final_cure, t_demold, v_demold, F_demold, ...) Through model analysis, the process parameters that have the most significant impact on product quality are identified (ranked by feature importance).
[0150] The third step is to generate optimization suggestions. Based on the association model, and with the goal of optimizing product quality, the optimal combination of process parameters is searched using grid search or Bayesian optimization methods. An optimization suggestion report is generated, including: Recommended parameter names to adjust; Comparison of current value and suggested value; Expected improvement in quality.
[0151] The fourth step is implementation and effect verification. After confirmation by operators, the optimization suggestions are automatically distributed to the control units of each subsystem for execution. Following implementation, product quality data before and after optimization are compared to verify the optimization effect. The verification results are then fed back to the model for continuous iterative optimization. Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A fully intelligent quality control system for concrete utility poles, comprising a centrifugal molding subsystem, a curing subsystem, and a demolding subsystem, characterized in that, It also includes a central control hub that connects the various subsystems; The centrifugal forming subsystem includes: The reference parameter identification unit is used to perform a stepped speed-up test under the no-load state of the centrifuge, collect the real-time operating data of the drive motor, and calculate the empty mold impedance parameters that characterize the inherent mechanical resistance of the mold through regression analysis. The empty mold impedance parameters include the empty mold rotational inertia J, the mold viscous friction coefficient B, and the Coulomb friction torque T_c. The mixed signal acquisition unit is used to simultaneously acquire the electromagnetic drive signal of the drive motor, the vibration signal of the mold, and the sound signal inside the cavity during the operation of the centrifuge under load, forming a mixed signal. The signal separation and feature extraction unit is used to perform blind source separation processing on the mixed signal, extract the pure material state signal component, and perform multi-scale permutation entropy analysis on the component to extract feature parameters characterizing the ordered evolution of the internal structure of concrete. The feature parameters include the entropy change rate ΔH / Δs, the entropy inflection point position s_inflection point, and the entropy stability interval. The load torque observation unit is used to acquire the electromagnetic driving torque T_e in real time, and to calculate the mechanical consumption torque T_m of the mold itself at the current speed using the empty mold impedance parameters. The electromagnetic driving torque is subtracted from the mechanical consumption torque to extract the net load torque T_net that only acts on the concrete material. The rheological and distribution state analysis unit is used to perform micro-perturbation response analysis on the net load torque T_net, calculate the equivalent dynamic viscosity η characterizing the flow state of concrete, and extract the frequency domain features of the net load torque T_net to calculate the dimensionless eccentricity factor ε characterizing the uniformity of material distribution. The hardening state assessment unit is used to comprehensively assess the internal density and hardening process of concrete based on the real-time change trend of the characteristic parameters, and generate the centrifugal final state hardening state level H_final and the centrifugal final state temperature distribution field T_dist. An adaptive closed-loop control unit is used to dynamically adjust the centrifuge speed or stage switching timing based on the preset value range of the dimensionless eccentricity factor ε, the equivalent dynamic viscosity η, and the hardening state level. The linkage control center includes a positive linkage control module, which is used to send the centrifugation final hardening state level H_final, centrifugation final temperature distribution field T_dist, and centrifugation final eccentricity factor ε_final as linkage data packets to the maintenance subsystem. The maintenance subsystem includes: A maintenance status receiving unit is used to receive the linkage data packet; The initial maintenance regime generation unit is used to match an initial maintenance scheme from a preset maintenance rule base according to the centrifugation final hardening state level H_final and the centrifugation final temperature distribution field T_dist. The initial maintenance scheme includes an initial heating rate, an initial isothermal temperature, an initial isothermal duration, and an initial cooling rate. The maintenance process monitoring unit is used to continuously collect data on the internal temperature of the pole, the surface temperature of the pole, and the temperature and humidity of the maintenance environment during the maintenance process, and to calculate the real-time hardening degree S(t) and real-time temperature gradient G(t) inside the pole based on these data. The curing parameter dynamic adjustment unit is used to dynamically adjust the curing parameters according to the real-time hardening degree S(t) and the real-time temperature gradient G(t), including: when the real-time hardening degree S(t) deviates from the expected hardening curve S_target(t), adjusting the heating rate or the constant temperature duration for compensation; when the real-time temperature gradient G(t) exceeds the preset safety threshold G_max, reducing the heating rate or extending the constant temperature duration to reduce the temperature gradient. The final state evaluation unit is used to evaluate the final hardening degree S_final_cure and final temperature T_final_cure of the pole at the end of the curing process, and generate the demolding readiness index R_demold accordingly. The positive linkage control module of the linkage control center is also used to send the final hardening degree S_final_cure and the final temperature T_final_cure as demolding pre-parameters to the demolding subsystem; The demolding subsystem includes: The demolding parameter pre-calibration unit is used to receive the demolding pre-parameters, and calculate the demolding parameter calibration coefficient λ in combination with the centrifugal final state eccentricity factor ε_final, to perform pre-calibration on the initial demolding parameters, and generate calibrated demolding parameters. The calibrated demolding parameters include calibrated demolding time t_demold, calibrated demolding speed v_demold, and calibrated demolding force F_demold. The demolding process monitoring unit is used to collect demolding force F(t), demolding displacement D(t) and pole surface strain data in real time during the demolding process, and calculate the real-time bonding force B(t) and demolding resistance change rate ΔR(t) based on these data. The demolding dynamic adjustment unit is used to dynamically adjust the demolding speed and demolding force according to the real-time adhesive force B(t) and the demolding resistance change rate ΔR(t): when the real-time adhesive force B(t) exceeds the preset safety range, the demolding speed is reduced or demolding is paused and auxiliary vibration is started; when the demolding resistance change rate ΔR(t) increases abnormally, an early warning is triggered and the demolding parameters are adjusted. The demolding anomaly diagnosis unit is used to diagnose the cause of the anomaly and generate an anomaly feedback report when abnormal demolding resistance or excessive strain on the pole surface is detected. The linkage control center also includes a reverse linkage control module, which is used to send the abnormal feedback report to the maintenance parameter dynamic adjustment unit of the maintenance subsystem to calibrate the maintenance regime parameters of subsequent batches of poles, and to send it to the adaptive closed-loop control unit of the centrifugal forming subsystem to adjust the centrifugal control strategy of subsequent batches of poles.
2. The system according to claim 1, characterized in that, The specific process by which the reference parameter identification unit calculates the empty-mode impedance parameters is as follows: Data on the mechanical angular velocity ω, angular acceleration α, and stator torque current component I_q of the centrifuge during the no-load acceleration process were collected. Construct a system of linear equations based on rigid body dynamics: T_e=J·α+B·ω+T_c Where T_e is the electromagnetic drive torque, T_e=K_t·I_q, and K_t is the motor torque constant; Using the stator torque current component I_q as the input variable and the mechanical angular velocity ω and angular acceleration α as the state variables, the least squares method is used to fit three constant terms J, B, and T_c, which are defined as the moment of inertia of the empty mold, the viscous friction coefficient of the mold, and the Coulomb friction torque, respectively. The combination of J, B, and T_c is set as the empty-mode impedance parameter and stored in the reference parameter identification unit.
3. The system according to claim 1, characterized in that, The specific process of blind source separation and multi-scale permutation entropy analysis performed by the signal separation and feature extraction unit is as follows: The mixed signal is preprocessed by filtering, denoising, and normalization; The independent component analysis algorithm was used to perform blind source separation on the preprocessed mixed signal, separating the material state signal component y_1(t), the mechanical vibration signal component y_2(t), and the environmental noise signal component y_3(t); The material state signal component y_1(t) is extracted as an effective signal characterizing the internal state of the concrete; The material state signal component y_1(t) is subjected to multi-scale coarsening processing. For the scale factor s, a coarsened sequence is constructed as follows: Where N is the signal length, and s takes the value of a positive integer from 1 to 20; For each coarse-grained sequence at scale s, calculate its permutation entropy value H_s with embedding dimension m=4 and delay time τ=1, and construct a multi-scale permutation entropy spectrum {H_s|s=1,2,...,20}; Feature parameters are extracted from the multi-scale entropy spectrum: entropy change rate ΔH / Δs, entropy inflection point position s_inflection point, and entropy stability interval.
4. The system according to claim 1, characterized in that, The specific process by which the load torque observation unit obtains the net load torque T_net is as follows: The real-time acquired stator current q-axis component I_q is converted into electromagnetic drive torque T_e=K_t·I_q; Based on the current mechanical angular velocity ω and angular acceleration α, and combined with the empty mold impedance parameters J, B, and T_c, the theoretical mechanical torque required for the mold to overcome inertia, viscous friction, and static friction is calculated as follows: T_m = J·α + B·ω + T_c; Calculate the difference between the electromagnetic drive torque and the theoretical mechanical torque, ΔT = T_e - T_m; The difference is superimposed with the compensation value ΔT_comp generated based on the speed observation error, and the final value after superposition is determined as the net load torque T_net = ΔT + ΔT_comp.
5. The system according to claim 1, characterized in that, The specific process of the rheological and distribution state analysis unit performing micro-perturbation response analysis and equivalent dynamic viscosity η calculation is as follows: A sinusoidal wave signal with a preset frequency f and amplitude A is superimposed on the basic speed command of the centrifuge. The frequency f ranges from 1 to 5 Hz, and the amplitude A ranges from 1% to 5% of the basic speed command value. The collected net load torque data and mechanical angular velocity data are subjected to high-pass filtering to remove the DC component and extract the torque fluctuation component ΔT_f and velocity fluctuation component Δω_f, which are consistent with the frequency of the sinusoidal wave signal. The root mean square value (RMS) of the torque fluctuation component (ΔT_f) and the root mean square value (RMS) of the speed fluctuation component (Δω_f) are calculated respectively within a preset time window, wherein the preset time window ranges from 3 to 10 seconds. Calculate the ratio R = RMS(ΔT_f) / RMS(Δω_f); Obtain the inner radius r and effective length L of the mold, and calculate the geometric constant G = 2π·r. 2 ·L, divide the ratio by the geometric constant to obtain the equivalent dynamic viscosity η=R / G, where the unit of the equivalent dynamic viscosity η is Pascal-second.
6. The system according to claim 1, characterized in that, The specific process of frequency domain feature extraction and eccentricity factor ε calculation performed by the rheological and distribution state analysis unit is as follows: Perform a Fast Fourier Transform on the net load torque T_net to obtain its frequency domain spectrum; Extract the amplitudes A_1, A_2, A_3, and A_4 from the spectrum corresponding to the centrifuge's fundamental frequency f_0 and its second, third, and fourth harmonics; Calculate the dimensionless eccentricity factor ε: Wherein, w_2, w_3, and w_4 are preset weight coefficients, and w_2+w_3+w_4=1. The value range of the preset weight coefficients is: w_2 is 0.4~0.6, w_3 is 0.2~0.4, and w_4 is 0.1~0.
3.
7. The system according to claim 1, characterized in that, The specific strategy for the adaptive closed-loop control unit to generate control commands based on each evaluation parameter is as follows: When the dimensionless eccentricity factor ε exceeds the preset safety threshold ε_max, an emergency braking command is output, wherein the value of ε_max is in the range of 0.12 to 0.
18. When the equivalent dynamic viscosity η is higher than the preset rheological threshold η_max, it is determined that the rheological state of the concrete has not met the standard, and a constant speed shear command is output to maintain the current speed for a preset time. The value range of η_max is 400~600 Pascals per second, and the value range of the preset time is 30~90 seconds. When the equivalent dynamic viscosity η is lower than or equal to the preset rheological threshold η_max and the dimensionless eccentricity factor ε is less than or equal to the preset safety threshold ε_max, it is determined that the rheological state meets the standard and the eccentricity is within the safe range. An adaptive acceleration command is output to accelerate to the target speed of the next stage with a preset acceleration. The value range of the preset acceleration is 0.3~0.8 rpm / s. When the hardening state level reaches the preset density level, a centrifugation termination command is output.
8. The system according to claim 1, characterized in that, The specific strategy for the dynamic adjustment unit of maintenance parameters to dynamically adjust the maintenance parameters is as follows: Set the expected hardening curve S_target(t), which is pre-calibrated and stored in the curing subsystem according to the pole specifications and concrete mix ratio; During the maintenance process, the deviation between the current actual hardening degree S(t) and the expected hardening degree S_target(t) is calculated at preset time intervals Δt, where the value of the preset time interval Δt ranges from 5 to 20 minutes. When ΔS(t) is greater than the preset positive deviation threshold, it is determined that the hardening speed is too slow, and a heating rate compensation operation is performed: the current heating rate is increased by a first preset percentage, or the current constant temperature is increased by a first preset temperature value. The preset positive deviation threshold ranges from 0.03 to 0.08, the first preset percentage ranges from 5% to 15%, and the first preset temperature value ranges from 1 to 3℃. When ΔS(t) is less than the preset negative deviation threshold, it is determined that the hardening speed is too fast, and a cooling rate compensation operation is performed: the current heating rate is reduced by a second preset percentage, or the current constant temperature is reduced by a second preset temperature value. The preset negative deviation threshold ranges from -0.08 to -0.03, the second preset percentage ranges from 5% to 15%, and the second preset temperature value ranges from 1 to 3℃. When the real-time temperature gradient G(t) exceeds the preset safety threshold G_max, a gradient suppression operation is performed: the heating is paused and the isothermal duration of the current temperature stage is extended until the temperature gradient falls back to the safe range. The value of G_max is 2.0~3.0℃ / cm.
9. The system according to claim 1, characterized in that, The specific process by which the demolding parameter pre-calibration unit calculates the demolding parameter calibration coefficient λ and calibrates the initial demolding parameters is as follows: Obtain the final curing hardening degree S_final_cure, the final temperature T_final_cure, and the centrifugal final eccentricity factor ε_final, where the value of S_final_cure ranges from 0 to 1; Obtain the target curing hardening degree S_target_cure and the median value of the optimal demolding temperature range T_target_demold; Calculate the hardening degree deviation ΔS = S_target_cure - S_final_cure; Calculate the temperature deviation ΔT = T_final_cure - T_target_demold; Calculate the overall calibration coefficient λ: Wherein, α, β, and γ are preset weight coefficients, and α + β + γ = 1, the value of α ranges from 0.4 to 0.6, the value of β ranges from 0.15 to 0.35, and the value of γ ranges from 0.2 to 0.4; The initial demolding parameters are calibrated using the comprehensive calibration coefficient λ: Calibrate demolding time: t_demold=t_0×(1+k_t·λ), where t_0 is the initial demolding time and k_t is the preset time calibration coefficient; Calibrate demolding speed: v_demold=v_0×(1-k_v·λ), where v_0 is the initial demolding speed and k_v is the preset speed calibration coefficient; Calibrate demolding force: F_demold=F_0×(1-k_F·λ), where F_0 is the initial demolding force, k_F is the preset force calibration coefficient; the value range of k_t is 0.2~0.5, the value range of k_v is 0.3~0.6, and the value range of k_F is 0.2~0.
5.
10. The system according to claim 1, characterized in that, The specific process by which the demolding anomaly diagnosis unit diagnoses the cause of the anomaly and generates an anomaly feedback report is as follows: When abnormal demolding resistance is detected, the demolding force F_abnormal, demolding displacement D_abnormal, and pole surface strain data ε_strain are collected at the moment the abnormality occurs. The demolding force F_abnormal is compared with the preset upper limit of adhesive force B_max and the preset lower limit of adhesive force B_min: If F_abnormal > B_max, the abnormality type is determined as over-tight bonding abnormality, wherein the value range of B_max is 20~30kN; If F_abnormal < B_min and the demolding displacement D_abnormal is greater than the preset normal demolding displacement D_normal, the abnormality type is determined as slightly loose bonding abnormality, wherein the value range of B_min is 8~15kN; The abnormality cause is diagnosed according to the abnormality type and the electric pole surface strain data ε_strain, and the abnormality cause includes at least one of local damage to the inner wall of the mold, residues on the inner wall of the mold, insufficient local hardening of the electric pole, and over-curing of the local part of the electric pole; An abnormality feedback report is generated, and the abnormality feedback report includes the abnormality type, abnormality position, abnormality degree and preliminary diagnosis cause; The reverse linkage control module sends the abnormality feedback report to the curing parameter dynamic adjustment unit of the curing subsystem, and the curing parameter dynamic adjustment unit adjusts the curing system parameters of subsequent batches of electric poles according to the abnormality type in the abnormality feedback report: when the abnormality type is over-tight bonding abnormality, reduce the constant temperature of subsequent batches of electric poles or shorten the constant temperature duration; when the abnormality type is insufficient local hardening of the electric pole, extend the constant temperature duration of subsequent batches of electric poles or increase the constant temperature; The reverse linkage control module sends the abnormality feedback report to the adaptive closed-loop control unit of the centrifugal forming subsystem, and the adaptive closed-loop control unit adjusts the centrifugal control strategy of subsequent batches of electric poles according to the abnormality type in the abnormality feedback report: when the abnormality type is insufficient local hardening of the electric pole, extend the operation duration of the low-speed centrifugation stage or increase the rotational speed of the low-speed centrifugation stage.
