A prestressed electric pole centrifugal forming control method and system based on material state

Abstract

The application discloses a prestressed electric pole centrifugal forming control method and system based on material state, and particularly relates to the technical field of concrete electric pole centrifugal forming manufacturing, and is used for solving the problem that it is difficult to effectively separate and identify the key features which truly reflect the concrete compactness from the sensor mixed signals under strong mechanical disturbance working conditions; mixed signals are formed by synchronously collecting vibration and sound signals during the centrifugal process, the mixed signals are subjected to blind source separation to extract pure material state signal components, the components are subjected to multi-scale permutation entropy analysis to extract feature parameters representing the internal structure ordered evolution, and the non-stationarity measurement indexes are calculated to evaluate the process balance, the material compactness and the process balance are comprehensively evaluated based on the real-time change trends of the above parameters, and the centrifuge speed or the stage switching time is dynamically adjusted accordingly, so that the transformation from the fixed experience program to the adaptive closed-loop control based on the real-time feedback of the material internal state is realized.

Description

Technical Field

[0001] This invention relates to the field of centrifugal molding manufacturing technology for concrete poles, and more specifically, to a control method and system for centrifugal molding of prestressed poles based on material state. Background Technology

[0002] Prestressed concrete poles are typically manufactured using a centrifugal molding process, where centrifugal force within a high-speed rotating mold causes the concrete mixture to distribute evenly and compact along the mold wall. Currently, the control of this process largely relies on preset centrifugal regimes, i.e., setting the rotation speed and time for each stage based on experience. Simultaneously, to improve quality stability, existing technologies also attempt to monitor vibration, sound, and other signals using sensors installed on the equipment or mold, thereby indirectly assessing the real-time state of the concrete within the mold.

[0003] However, under the aforementioned strong mechanical disturbance conditions, the signals collected by the sensors are actually a mixture of various vibration and sound sources coupled together, such as changes in the state of concrete materials, mechanical transmission impact, and mold rotation and oscillation. This makes it difficult to effectively and reliably separate and identify the key state characteristics that uniquely or directly reflect the internal compactness and drainage process of the concrete from the complex mixed signals, thus severely restricting the technical realization of accurate and closed-loop control based on the actual state of the material. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for controlling the centrifugal forming of prestressed poles based on material state to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for controlling the centrifugal forming of prestressed electric poles based on material state includes the following steps:

[0007] S1. During the operation of the centrifuge, the vibration signal and sound signal of the mold are collected simultaneously to obtain a mixed signal;

[0008] S2. Perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials.

[0009] S3. Perform multi-scale coarsening processing on the material state signal components to obtain time series at multiple scales. Calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale. Extract characteristic parameters characterizing the evolution process of the internal structure of concrete from the curve.

[0010] S4. Calculate the non-stationarity measure of the material state signal components;

[0011] S5. Based on the changing trends of characteristic parameters and non-stationarity measurement indices, comprehensively evaluate the real-time compaction state and compaction process uniformity of concrete materials inside the mold.

[0012] S6. Based on the real-time compaction status and the uniformity of the compaction process, adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage.

[0013] Furthermore, vibration and sound signals from the mold are simultaneously acquired during centrifuge operation to obtain a mixed signal, including:

[0014] Vibration sensors are installed at at least two axial positions of the mold, and sound sensors are installed at the ends of the mold or on the centrifuge frame.

[0015] The rotary encoder signal of the centrifuge spindle is used as the synchronous trigger reference to control the vibration sensor and the sound sensor to collect data at equal time intervals.

[0016] The signals from all vibration sensors and sound sensors collected at the same time are combined to form a mixed signal.

[0017] Furthermore, blind source separation processing is performed on the mixed signal to separate the material state signal components related to the changes in the state of concrete materials, including:

[0018] The mixed signal is preprocessed by whitening to eliminate the correlation between the signals of each channel;

[0019] The objective function is constructed based on the negative entropy maximization criterion, and the separation matrix is ​​found through a fixed-point iterative algorithm.

[0020] The separation matrix is ​​used to perform a linear transformation on the whitened mixed signal to obtain several independent source signal estimation components;

[0021] Based on the energy change characteristics of the source signal estimation components during the low-speed stage of the centrifuge startup, the component whose energy increases steadily with the speed and is independent of the mechanical resonance frequency of the mold is selected from several independent source signal estimation components as the material state signal component related to the state change of concrete material.

[0022] Furthermore, the material state signal components are subjected to multi-scale coarsening processing to obtain time series at multiple scales. The permutation entropy of the time series at each scale is calculated to form a curve of permutation entropy changing with scale. Characteristic parameters characterizing the evolution process of the internal structure of concrete are extracted from the curve, including:

[0023] Multiple scale factors are set for the material state signal components, and coarse-grained processing is performed on the material state signal components based on each scale factor to generate a coarse-grained time series corresponding to each scale factor.

[0024] For each coarse-grained time series, the embedding dimension and delay time for permutation entropy calculation are determined based on its sampling characteristics, and the permutation entropy value of the coarse-grained time series is calculated.

[0025] Connect the permutation entropy values ​​corresponding to all scale factors in ascending order of scale factors to form a curve of permutation entropy as a function of scale.

[0026] On the curve of permutation entropy changing with scale, identify the initial and final scales where the permutation entropy value transitions from the peak to a monotonically decreasing state. Calculate the linear fitting slope of the permutation entropy value relative to scale change between the initial and final scales, and use this linear fitting slope as a characteristic parameter characterizing the evolution process of the internal structure of concrete.

[0027] Furthermore, the range of multiple scale factors is set based on the dominant frequency period of the material state signal component. The minimum value of the scale factor is set to an integer multiple of the dominant frequency period, and the maximum value is set to the longest characteristic time scale required to cover the entire process of concrete evolution from fluid state to compact state.

[0028] Furthermore, the non-stationarity measures of the material state signal components are calculated, including:

[0029] The material state signal components are divided into multiple continuous and equal-length analysis windows;

[0030] For each analysis window, calculate the variance of the material state signal components within that analysis window;

[0031] Construct the time series using the variances of all analysis windows;

[0032] Calculate the coefficient of variation or the detrended volatility analysis scaling index for the time series, and use the coefficient of variation or the detrended volatility analysis scaling index as a measure of nonstationarity.

[0033] Furthermore, based on the changing trends of characteristic parameters and non-stationarity metrics, the real-time compaction state and compaction process uniformity of the concrete material within the mold are comprehensively evaluated, including:

[0034] Linear fitting is performed on the historical trend of the feature parameters to obtain the rate of change of the feature parameters;

[0035] Calculate the variance of the characteristic parameters within a specified time window; read the current value of the nonstationarity measure.

[0036] The rate of change of the feature parameter is compared with a preset rate of change threshold, the variance of the feature parameter within a specified time window is compared with a preset variance threshold, and the current value of the nonstationarity metric is compared with a preset stability threshold.

[0037] When the rate of change of the feature parameter is lower than the rate of change threshold, the variance of the feature parameter within the specified time window is lower than the variance threshold, and the current value of the non-stationarity metric is lower than the stability threshold, it is determined that the real-time compaction state has entered the stable compaction stage and the compaction process is of excellent uniformity.

[0038] Otherwise, the real-time compaction state is judged to be in the transitional compaction stage or the compaction process is deemed to be of poor uniformity.

[0039] Furthermore, based on the real-time compaction status and the uniformity of the compaction process, the centrifuge speed during the current centrifugation stage or the timing of transitioning to the next centrifugation stage is adjusted, including:

[0040] If the real-time compaction state is in the transitional compaction stage and the compaction process is of poor uniformity, then reduce the centrifuge speed in the current centrifugation stage.

[0041] If the real-time compaction state is in the transitional compaction stage and the compaction process is of good uniformity, then increase the centrifuge speed in the current centrifugation stage.

[0042] If the real-time compaction state is in the stable compaction stage, then control the centrifuge to switch to the next centrifugation stage.

[0043] Furthermore, the speed of the centrifuge can be increased or decreased during the current centrifugation stage. The speed adjustment amount is obtained by proportional integration based on the difference between the rate of change of the characteristic parameter and the rate of change threshold, and the difference between the current value of the non-stationarity metric and the stability threshold.

[0044] On the other hand, the present invention provides a prestressed pole centrifugal forming control system based on material state, comprising the following modules:

[0045] The signal acquisition module is used to simultaneously acquire vibration and sound signals from the mold during centrifuge operation to obtain a mixed signal;

[0046] The blind source separation module is used to perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials.

[0047] The feature extraction module is used to perform multi-scale coarse-graining processing on the material state signal components to obtain time series at multiple scales, calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale, and extract feature parameters characterizing the evolution process of the internal structure of concrete from the curve.

[0048] The index calculation module is used to calculate the non-stationarity measurement index of the material state signal components.

[0049] The state assessment module is used to comprehensively evaluate the real-time compaction state and compaction process uniformity of concrete materials in the mold based on the changing trends of characteristic parameters and non-stationarity measurement indicators.

[0050] The parameter adjustment module is used to adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage based on the real-time compaction status and the uniformity of the compaction process.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. By effectively extracting and deeply analyzing purified material state characteristics from mixed signals with strong mechanical interference, a direct and accurate perception of the intrinsic evolution law of the concrete centrifugal compaction process is achieved. Blind source separation technology is used to remove interference such as mechanical vibration, ensuring the purity of the state signal. Furthermore, multi-scale permutation entropy analysis can capture the multi-scale ordering process of the concrete's internal structure evolving from disordered flow dynamics to an ordered compacted state from the material state signal components. Combined with the assessment of the signal's time-varying stability using non-stationarity metrics, a three-dimensional characterization of the material's structural order and process stability is formed. This allows the judgment of the compaction state to move beyond reliance on single, superficial signals and instead be based on deep characteristics reflecting the material's internal physical essence, achieving a simultaneous and precise assessment of the concrete's true compaction state and process uniformity.

[0053] 2. The centrifugal molding process has been transformed from time-programmed control to material state-adaptive control. The control system can dynamically decide and adjust the centrifuge speed or switching timing based on the accurate assessment of real-time compaction status and process balance: when the process is unbalanced, it automatically reduces the speed to ease compaction; when the state is good, it increases the speed to optimize efficiency; and when a stable state is reached, it promptly switches to the next stage, ensuring that the molding process always converges to the target compaction state along the optimal path. This effectively overcomes the mismatch problem of fixed-program control caused by factors such as material batch differences and environmental fluctuations. Thus, while improving the consistency and uniformity of prestressed pole molding quality, it also enhances the adaptability and intelligence level of the production process. Attached Figure Description

[0054] Figure 1 This is a flowchart of a centrifugal forming control method for prestressed electric poles based on material state according to the present invention;

[0055] Figure 2 This is a schematic diagram of the structure of a prestressed electric pole centrifugal forming control system based on material state according to the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Example 1: Figure 1 This invention presents a method for controlling the centrifugal forming of prestressed electric poles based on material state, which includes the following steps:

[0058] S1. During the operation of the centrifuge, the vibration signal and sound signal of the mold are collected simultaneously to obtain a mixed signal;

[0059] S2. Perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials.

[0060] S3. Perform multi-scale coarsening processing on the material state signal components to obtain time series at multiple scales. Calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale. Extract characteristic parameters characterizing the evolution process of the internal structure of concrete from the curve.

[0061] S4. Calculate the non-stationarity measure of the material state signal components;

[0062] S5. Based on the changing trends of characteristic parameters and non-stationarity measurement indices, comprehensively evaluate the real-time compaction state and compaction process uniformity of concrete materials inside the mold.

[0063] S6. Based on the real-time compaction status and the uniformity of the compaction process, adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage.

[0064] S1. During the centrifuge operation, the vibration and sound signals of the mold are collected synchronously to obtain a mixed signal. The specific implementation is as follows:

[0065] Vibration sensors are installed at at least two different locations along the axial length of the outer surface of the prestressed pole centrifugal mold. One exemplary installation location is approximately one-quarter of the total mold length from one end, and another exemplary location is approximately one-quarter of the total mold length from the other end. This location selection aims to collect the vibration response of the mold in different axial sections during centrifugation. The vibration sensors are securely attached to the clean, flat surface of the mold's outer wall using a magnetic base or high-strength adhesive. During installation, the sensitive axis of the vibration sensor needs to be adjusted to be perpendicular to the mold's outer wall surface to effectively detect radial vibrations caused by the distribution and compaction of concrete materials under centrifugal force. Acoustic sensors for collecting sound signals are installed near the end flange of the mold or on a rigid structural component of the centrifuge frame. The installation location should avoid direct impact from strong airflow generated by cooling fans or other equipment and should be away from major mechanical noise sources such as the centrifuge motor, gearbox, or drive belt cover. For example, the acoustic sensors should not be directly mounted on the high-speed rotating spindle or drive belt cover. This installation method aims to ensure that the acquired sound signals include more acoustic emission information from the flow of concrete materials, aggregate collisions, and drainage and venting processes within the mold, while reducing interference from background mechanical noise. The sound sensor can be a condenser microphone, whose frequency response range needs to cover the main acoustic emission frequency bands that may be generated during the centrifugal molding of concrete. An example frequency response range is 20 Hz to 20,000 Hz.

[0066] To achieve strictly synchronized acquisition of vibration and sound signals, this embodiment uses the pulse signal output from the rotary encoder of the centrifuge spindle as the synchronous trigger reference and timing control source for the entire data acquisition process. The rotary encoder is coaxially mounted to the centrifuge spindle via a coupling. For each revolution of the centrifuge spindle, the rotary encoder outputs a fixed number of pulses, for example, 1024 pulses. The data acquisition device employs a multi-channel synchronous data acquisition card, connecting its external trigger input to the rotary encoder's pulse output via a shielded cable. The data acquisition card's sampling clock is configured to reference its internal crystal oscillator clock source, but the start of the sampling process is triggered and synchronized by the rising edge of the first pulse emitted by the rotary encoder. After the data acquisition start signal is triggered, the data acquisition card uses its own stable clock frequency to control all enabled analog input channels to perform data sampling at equal time intervals; the reciprocal of this sampling time interval is the sampling frequency. The specific value of the sampling frequency needs to be set according to Shannon's sampling theorem, meaning the sampling frequency must be at least twice the highest frequency component of the vibration and sound signals to be analyzed. Meanwhile, the sampling frequency setting also needs to comprehensively consider the data storage capacity and the computational efficiency of subsequent signal processing algorithms. An example sampling frequency is set to 5000 Hz. The analog signal output lines of each vibration sensor and sound sensor are connected to different analog input channels of the data acquisition card. All channels start their first sampling strictly and simultaneously under the action of the rotary encoder trigger signal, and continue to sample for a preset duration at the same sampling frequency set above throughout the entire acquisition process. For example, the sampling duration is set to 80% of the planned duration of the current centrifugation stage, or sampling continues until a software stop command is received from the control system.

[0067] During data acquisition, at each sampling point determined by the internal clock of the data acquisition card, the analog-to-digital converter of the card synchronously reads the instantaneous voltage values ​​of all connected vibration sensor channels and sound sensor channels. These voltage data collected by sensors of different physical locations and types at the same microscopic sampling point together constitute a multivariate observation data vector uniquely corresponding to that moment. Arranging these multivariate observation data vectors corresponding to all consecutive sampling points within the entire acquisition period in chronological order and storing them in non-volatile memory forms the mixed signal required for subsequent steps. Specifically, the mixed signal is a multi-channel time-series data set, where each channel's data sequence corresponds to the output of a specific sensor, and all channel data are strictly aligned on the time axis, sharing the exact same timestamp sequence and total number of sampling points. This mixed signal can be stored in a custom file format containing header information and binary data bodies. The header information records metadata such as the number of channels, sampling frequency, sensor sensitivity coefficient, and acquisition start time; or it can be stored in a general matrix data format, where each column represents the data of one sensor channel, and each row represents the data of all channels at one sampling point. Through the above implementation methods, it is ensured that multi-source information reflecting the state of the centrifugation process is obtained from the spatial distribution and physical form of the signal, and all information maintains strict synchronicity and comparability in the time dimension, providing input data with clear physical meaning, high spatiotemporal correlation and standardized format for subsequent blind source separation processing.

[0068] S2. Perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials. The specific implementation is as follows:

[0069] A whitening preprocessing is performed on a mixed signal composed of data from multiple sensor channels. The purpose of whitening preprocessing is to eliminate the second-order statistical correlation between the data from different channels of the mixed signal. The process involves calculating the covariance matrix of the mixed signal, where each element represents the covariance value between two specific sensor channel data sequences. Eigenvalue decomposition is then performed on the calculated covariance matrix to obtain an eigenvalue diagonal matrix and a corresponding eigenvector matrix. A whitening transformation matrix is ​​constructed using the eigenvalue diagonal matrix and the eigenvector matrix by right-multiplying the eigenvector matrix by a new diagonal matrix formed by raising each diagonal element of the eigenvalue diagonal matrix to the power of -1 / 2. The original mixed signal data matrix is ​​then left-multiplied by this whitening transformation matrix to obtain the whitened mixed signal. The whitened mixed signal satisfies the condition that its covariance matrix is ​​an identity matrix, i.e., the mean and variance of each channel data are zero, and the covariance between different channels is zero.

[0070] An objective function is constructed based on the criterion of maximizing negative entropy, and a separation matrix is ​​found using a fixed-point iterative algorithm. Negative entropy is an indicator that measures the degree of difference between a random variable and a Gaussian distribution. The objective function is defined as the approximate negative entropy of the output component obtained by multiplying a column vector of the separation matrix by the whitened mixed signal. This approximation is estimated using higher-order statistics of the output component, approximating it by the square of the difference between the expected value of the output component after passing through a specific nonlinear function and the expected value of a Gaussian variable after passing through the same nonlinear function. The fixed-point iterative algorithm is used to solve for the column vector of the separation matrix that maximizes the objective function. The iterative process begins with an initial weight vector, which is a randomly generated unit norm vector. In each iteration, the whitened mixed signal is linearly projected onto the current weight vector to obtain a sequence of output components. Then, the mean of the product of the output component sequence after transformation by the selected nonlinear function and the whitened mixed signal is calculated, and a bias term related to the norm of the current weight vector is subtracted to obtain the update direction. The updated vector is normalized to have a norm of one, and this is used as the weight vector for the next iteration. Repeat the above iterative steps until the absolute value of the difference between the cosine of the angle between the weight vectors obtained from two consecutive iterations and the first value is less than a preset convergence threshold. This convergence threshold is set based on a trade-off between numerical computation accuracy requirements and algorithm efficiency; a typical value is 1 multiplied by 10 to the power of -6. The algorithm is considered convergent when the absolute value of the difference between the cosine of the angle between the weight vectors obtained from two consecutive iterations and the first value is less than the preset convergence threshold. The converged weight vector is then used as a column of the separation matrix. To obtain the complete separation matrix and separate multiple independent source signal estimation components, the above fixed-point iteration process is repeated. Each time, a constraint is applied to the initial weight vector that is orthogonal to the column vectors of the obtained separation matrix to ensure that a new independent direction is found each time. Finally, all converged column vectors are combined into a complete separation matrix.

[0071] A linear transformation is performed on the whitened mixed signal using a separation matrix to obtain several independent source signal estimation components. Specifically, the complete separation matrix is ​​multiplied by the whitened mixed signal data matrix. The result of this multiplication is a new data matrix, where each row represents a sequence of estimated source signal components over the entire sampling time period. These row vectors represent statistically independent source signal estimation components, each corresponding to a potential physical source.

[0072] Based on the energy change characteristics of the source signal estimation components during the low-speed phase of centrifuge startup, a source signal estimation component whose energy steadily increases with rotational speed and is independent of the mechanical resonance frequency of the mold is selected from several independent source signal estimation components as the material state signal component related to the state change of concrete materials. The low-speed phase of centrifuge startup refers to the time interval during which the centrifuge starts rotating from a stationary state, reaches its first stable low-speed feeding speed, and maintains operation. The selection method involves obtaining the rotational speed time series during the low-speed phase of centrifuge startup, which is read from the centrifuge control system. Simultaneously, data for each source signal estimation component within the same time period are extracted. The signal energy of each source signal estimation component is calculated for several equal-length sub-time periods within this low-speed phase. The signal energy is characterized by the sum of squares of the signal sample values ​​within each sub-time period. Correlation analysis is performed between the energy series and the rotational speed time series of each source signal estimation component, and the Pearson correlation coefficient is calculated. The criterion for a significant positive correlation between the energy series and the rotational speed series is that the Pearson correlation coefficient is greater than a preset positive correlation threshold. This positive correlation threshold is set according to the statistical significance level requirements, with a typical setting value of 0.7. When the Pearson correlation coefficient between the energy sequence and the rotational speed sequence of a source signal estimation component is greater than a preset positive correlation threshold, and the energy increases steadily with increasing rotational speed, the source signal estimation component is considered a candidate source signal estimation component. Source signal estimation components related to the mold's mechanical resonance frequency are excluded. The mold's mechanical resonance frequency is pre-determined through experimental modal analysis, and several main natural frequencies of the mold are obtained through impact experiments. The power spectral density of each candidate source signal estimation component is calculated throughout the centrifugation process, and its power spectral density peak is checked to see if it appears near a known mold mechanical resonance frequency point, for example, a significant peak within ±2 Hz of a certain resonance frequency. If a significant peak appears, the candidate source signal estimation component is determined to be a mechanical resonance response component and is excluded. If the energy of a candidate source signal estimation component is positively correlated with the rotational speed, and its power spectral density has no significant peak at the known mold mechanical resonance frequency, then this candidate source signal estimation component is ultimately selected as a material state signal component related to the changes in the state of concrete materials. This material state signal component will serve as the core input data for subsequent in-depth feature analysis. The above screening process ensures that the selected signal components can reflect the state evolution of concrete materials under centrifugal force to the greatest extent.

[0073] S3. Perform multi-scale coarsening processing on the material state signal components to obtain time series at multiple scales. Calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale. Extract characteristic parameters representing the evolution process of the internal structure of concrete from the curve. Specifically, the implementation is as follows:

[0074] Multiple scale factors are set for the material state signal components. The range of these scale factors is based on the dominant frequency period of the material state signal components. The dominant frequency period of a material state signal component is the reciprocal of the frequency corresponding to the maximum peak value in the power spectral density of the material state signal component. By calculating the power spectral density of the material state signal components, the maximum peak value is identified, and the frequency corresponding to this maximum peak value is the dominant frequency. The reciprocal of the dominant frequency is the dominant frequency period. The minimum value of the scale factor is set to an integer multiple of the dominant frequency period, for example, 1 times the dominant frequency period. The maximum value of the scale factor is set to the longest characteristic time scale required to cover the entire evolution of concrete from a fluid state to a compact state. This longest characteristic time scale is determined by combining the initial setting time of the concrete formula with the total duration of the centrifugation process. An exemplary method is to set the longest characteristic time scale to 1 / 5 of the total centrifugation process duration. Multiple scale factors are selected linearly and uniformly between the minimum and maximum scale factors, for example, 10 scale factors are selected, with the scale factor sequence being 1, 2, 3, 4, 5, 6, 7, 8, 9, 10.

[0075] Coarse-grained processing is performed on the material state signal components based on each scale factor to generate a coarse-grained time series corresponding to each scale factor. Coarse-grained processing involves segmenting the material state signal component sequence according to the scale factor and calculating the arithmetic mean of the data points within each segment. For a specific scale factor, the material state signal component sequence is sequentially divided into non-overlapping windows with a length equal to that scale factor. If the last window is shorter than the scale factor, it is discarded. The arithmetic mean of all data points within each window is calculated to obtain a coarse-grained time series. The arithmetic mean is calculated by summing the values ​​of all data points within the window and dividing by the scale factor. This segmentation and averaging process is repeated for each scale factor to generate the coarse-grained time series corresponding to each scale factor.

[0076] For each coarse-grained time series, the embedding dimension and delay time are determined based on its sampling characteristics, and the permutation entropy value of the coarse-grained time series is calculated. Permutation entropy is a metric for measuring the complexity of a time series. For each coarse-grained time series, two parameters need to be determined: embedding dimension and delay time. The embedding dimension represents the dimension of each state vector in the phase space reconstruction. The delay time represents the time interval between adjacent elements in the phase space reconstruction. The embedding dimension typically ranges from 3 to 7, with the specific value chosen based on the length of the coarse-grained time series. For example, when the length of the coarse-grained time series is greater than 1000 points, the embedding dimension is set to 6. The delay time is typically set to 1. After determining the embedding dimension and delay time, the phase space of the coarse-grained time series is reconstructed, resulting in multiple state vectors. Each state vector consists of points in the original coarse-grained time series that are separated by a delay time, and the vector dimension is equal to the embedding dimension. For each state vector, all its elements are sorted in ascending order of value, and the sorting order is recorded to obtain a permutation pattern. Count the occurrences of all permutations in the state vectors. Calculate the probability of each permutation, which is the number of occurrences of that permutation divided by the total number of state vectors. The permutation entropy is equal to the sum of the natural logarithms of the negative probabilities of all permutations. The permutation entropy is maximum when all permutations occur with equal probability; it is zero when the time series is perfectly regular. The magnitude of the permutation entropy reflects the randomness or complexity of the coarse-grained time series.

[0077] By connecting the permutation entropy values ​​corresponding to all scale factors in ascending order of scale factor, a curve representing the change of permutation entropy with scale is formed. Specifically, the scale factor is plotted on the x-axis, and the calculated permutation entropy value is plotted on the y-axis. Points are then plotted on a two-dimensional plane, and these points are connected by line segments in ascending order of scale factor to form a curve. This curve reflects the variation of the complexity of the material state signal components with the analytical scale.

[0078] On the curve of arrangement entropy versus scale, identify the initial and final scales where the arrangement entropy value transitions from a peak to a monotonically decreasing state. Calculate the slope of the linear fit of the arrangement entropy value with respect to scale change between the initial and final scales, and use this linear fit slope as a characteristic parameter characterizing the evolution of the internal structure of concrete. The method for identifying the initial and final scales is to first find the maximum point on the curve of arrangement entropy versus scale, i.e., the peak point. Starting from the scale corresponding to the peak point, look for the scale where the arrangement entropy value begins to continuously decrease along the direction of scale increase as the initial scale. The criterion for determining the initial scale is that, starting from the scale corresponding to the peak point, the arrangement entropy value of the next three consecutive scales is smaller than the arrangement entropy value of the previous scale; these first scales with consecutive decreases are then considered the initial scales. The criterion for determining the final scale is to start from the initial scale and look for the scale where the arrangement entropy value no longer decreases or begins to increase along the direction of scale increase. Specifically, starting from the initial scale, check the arrangement entropy value of each scale; if the arrangement entropy value of a certain scale is greater than or equal to the arrangement entropy value of the previous scale, then that scale is the final scale. After determining the initial and final scales, all scale factors and their corresponding permutation entropy values ​​between the initial and final scales are extracted, and linear fitting is performed using the least squares method. The linear fitting yields a straight line, and the slope of this line is the linear fitting slope of the permutation entropy value relative to scale change. This linear fitting slope reflects the average rate of change of the complexity of the material state signal components with scale within the critical scale range. During the centrifugation process of concrete, as the internal structure evolves from a fluid state to a compact state, the multi-scale complexity characteristics of the material state signal components exhibit regular changes. This linear fitting slope can quantitatively characterize the severity or rate of this structural evolution, and therefore is used as a feature parameter for subsequent comprehensive evaluation. The range of multiple scale factors is set based on the dominant frequency period of the material state signal components. The minimum value of the scale factor is set to an integer multiple of the dominant frequency period, and the maximum value is set to the longest characteristic time scale required to cover the entire process of concrete evolution from a fluid state to a compact state. This setting ensures that the multi-scale analysis can cover the critical time scale range of concrete state evolution. For example, if the dominant frequency of the material state signal component is 10 Hz, the dominant frequency period is 0.1 seconds, and the sampling frequency is 1000 Hz, then the dominant frequency period corresponds to 100 sampling points. The minimum scale factor is set to 100. The longest characteristic time scale is 10 seconds, corresponding to 10,000 sampling points, and the maximum scale factor is set to 10,000. Multiple scale factors are selected logarithmically between the minimum and maximum scale factors, for example, 10 scale factors. With this setting, multi-scale analysis can effectively capture various dynamic characteristics from rapid fluctuations to slow evolutions, laying the foundation for accurately extracting characteristic parameters reflecting the evolution of the internal structure of concrete.

[0079] S4. Calculate the non-stationarity measure of the material state signal components, specifically as follows:

[0080] First, the material state signal components are divided into multiple consecutive analysis windows of equal length. The length of the analysis window needs to balance the ability to capture local fluctuations in the signal with the overall statistical stability. The specific method for determining the window length is based on the sampling frequency of the material state signal components and the duration corresponding to a relatively stable stage in the centrifugal forming process. For example, if the sampling frequency of the material state signal components is 1000 Hz, and a low-speed feeding stage typically lasts 60 seconds, then the length of an analysis window can be set to cover one percent of the total number of sampling points in that stage, i.e., one analysis window contains 600 data points. The windows do not overlap. The entire material state signal component sequence is divided into multiple consecutive segments according to time sequence. If the last segment has fewer data points than the length of an analysis window, the segment is retained and still considered as an analysis window, with a length shorter than the standard window length.

[0081] For each analysis window, the variance of the material state signal component within that window is calculated. Variance is a statistic that measures how much data is dispersed around its mean. For an analysis window, first, the arithmetic mean of all data points within that window is calculated. The arithmetic mean is calculated by summing the values ​​of all data points within the window and dividing by the number of data points contained in the window. Then, the difference between each data point and the arithmetic mean is calculated, and each difference is squared. Finally, all squared differences are summed and divided by the number of data points contained in the window minus one. The result is the variance of the material state signal component within that analysis window. For the last analysis window that is shorter than the required length, the variance is calculated in the same way, with the divisor being the actual number of data points contained in the window minus one. A corresponding variance value is calculated for each analysis window.

[0082] A time series is constructed using the variances of all analysis windows. The construction method involves sequentially arranging the calculated variance values ​​of each analysis window according to their chronological order of appearance in the original material state signal components, forming a new one-dimensional time series. Each data point in this new series represents the energy fluctuation intensity of the material state signal component within a specific time period, and its chronological order remains consistent with the original signal's chronological order. This series is the variance time series.

[0083] Calculate the coefficient of variation (COP) or detrended volatility analysis scaling index for the variance time series, and use this COP or detrended volatility analysis scaling index as a measure of nonstationarity. The COP is the ratio of the standard deviation to the mean, used to measure the relative dispersion of the data, eliminating the influence of measurement scale. To calculate the COP of a variance time series, first calculate the arithmetic mean of all values ​​in the variance time series. Then calculate the standard deviation of the variance time series, which is equal to the square root of the sum of the squares of the differences between each value and the arithmetic mean. Finally, the COP is equal to the standard deviation divided by the arithmetic mean. A larger COP indicates more severe relative fluctuations in the variance time series, meaning the energy distribution of the material state signal components is more unstable over time, and the stronger the nonstationarity.

[0084] The detrended volatility scaling index is another indicator that characterizes the long-term correlation and trend-preserving ability of time series. To calculate the detrended volatility scaling index for a variance time series, the cumulative deviation sequence of the variance time series is first calculated. The first value of the cumulative deviation sequence equals the first value of the variance time series minus the arithmetic mean of the variance time series. Each subsequent value of the cumulative deviation sequence equals the previous cumulative deviation value plus the current variance time series value minus the arithmetic mean of the variance time series. Then, the cumulative deviation sequence is segmented into segments of varying lengths. The segment length starts from a minimum and gradually increases to a maximum, typically with a minimum segment length of no less than 10 points and a maximum segment length no greater than one-quarter of the total length of the cumulative deviation sequence. For each segment length, the cumulative deviation sequence is divided into several non-overlapping segments of equal length, discarding any last segment that is too short. For each segment, a straight line is fitted using the least squares method. The sum of the squares of the differences between all cumulative deviation values ​​within that segment and the corresponding values ​​on the fitted line is calculated, and then divided by the segment length to obtain the local volatility of that segment. For all segments of that length, calculate the average of their local fluctuations, and then take the square root of this average to obtain the fluctuation function value corresponding to that segment length. In a log-log coordinate system, plot the segment length as the x-axis and the corresponding fluctuation function value as the y-axis. These points typically exhibit a linear trend. Use the least squares method to fit this linear trend; the slope of the fitted line is the detrended volatility analysis scaling exponent. The scaling exponent typically ranges from 0 to 1.5. When the scaling exponent is close to 0.5, it indicates that the sequence is close to a random walk; when the scaling exponent is greater than 0.5, it indicates that the sequence has a long-term positive correlation, meaning that past growth trends tend to continue in the future, and the sequence exhibits a certain degree of persistence or trend; when the scaling exponent is less than 0.5, it indicates that the sequence has an inverse correlation, and the fluctuations are relatively violent. During centrifugation, if the energy fluctuations of the material state signal components are continuous, the scaling exponent of its variance time series will be greater than 0.5, reflecting that the compaction process is gradual and continuous. If the energy fluctuations are drastic and trendless, the scaling exponent may be close to or less than 0.5, reflecting that the process may be intermittent or locally unbalanced. Both the coefficient of variation and the detrended volatility analysis scaling exponent quantitatively characterize the time-varying characteristics of the energy distribution of the material state signal components, i.e., non-stationarity, and are therefore used as a measure of non-stationarity. This index provides a quantitative basis for the time-dimensional stability of the compaction process's equilibrium.

[0085] S5. Based on the changing trends of characteristic parameters and non-stationarity metrics, comprehensively evaluate the real-time compaction state and compaction process uniformity of the concrete material within the mold. Specifically, this is implemented as follows:

[0086] Linear fitting is performed on the historical trend of the characteristic parameters to obtain the rate of change of the characteristic parameters. The characteristic parameters are the slopes of the linear fit extracted from the curve of permutation entropy versus scale. The historical trend refers to the sequence of characteristic parameter values ​​over a specific time period, traced back from the current moment. This specific time period is called the fitting time window, and its length is set according to the characteristics of the centrifugation phase. For example, in the high-speed compaction phase, the length of the fitting time window can be set to 30 seconds. The least squares method is used for linear fitting. With time as the independent variable and the characteristic parameter values ​​as the dependent variable, a straight line is fitted to the data points within the fitting time window. The fitted line is a straight line, and the slope of this line is the rate of change of the characteristic parameters. The physical meaning of the rate of change of the characteristic parameters is the amount of change of the characteristic parameters per unit time, and its unit is the characteristic parameter unit divided by the time unit, for example, per second. A positive rate of change of the characteristic parameters indicates that the characteristic parameters increase with time, and a negative rate of change indicates that they decrease with time.

[0087] Calculate the variance of the feature parameter within a specified time window. The specified time window is a period used to evaluate the short-term volatility of the feature parameter; its length can be the same as or different from the fitted time window, for example, both set to 30 seconds. The end time of the specified time window is usually the current time or slightly earlier. Calculate the arithmetic mean of all feature parameter values ​​within the specified time window. Then calculate the difference between each feature parameter value within the window and the arithmetic mean, square each difference, sum them, and divide by the number of data points in the window minus one. The result is the variance of the feature parameter within the specified time window. This variance quantifies the recent volatility of the feature parameter; a larger variance indicates more drastic fluctuations and a more unstable process.

[0088] Read the current value of the nonstationarity metric. The current value of the nonstationarity metric refers to the value obtained after the calculation of a complete analysis window closest to the current time. For example, if the analysis window length is 10 seconds and the current time is at the 55th second, the currently read nonstationarity metric value is the coefficient of variation or detrended volatility scaling index calculated based on the variance time series of the analysis window from the 40th to the 50th second. This value reflects the stability of the energy distribution of the material state signal components within the most recent analysis period.

[0089] The rate of change of the characteristic parameter is compared with a preset rate of change threshold. The preset rate of change threshold is an empirical value derived from a large amount of historical centrifugation test data, used to distinguish whether the characteristic parameter is in a rapid change phase or tending towards stability. This threshold is determined by collecting the numerical distribution of the rate of change of the characteristic parameter during multiple successful centrifugation processes when the concrete is determined to have entered a stable and compacted stage. The statistical mean and standard deviation of these values ​​are calculated, and the rate of change threshold is set as the statistical mean plus one standard deviation. An exemplary rate of change threshold could be 0.01 seconds. The comparison operation determines whether the absolute value of the rate of change of the characteristic parameter is less than the preset rate of change threshold. If the absolute value of the rate of change of the characteristic parameter is less than the preset rate of change threshold, the trend of change of the characteristic parameter is considered to have leveled off.

[0090] The variance of the characteristic parameter within a specified time window is compared with a preset variance threshold. The preset variance threshold is also an empirical value derived from historical data statistics, used to determine whether the fluctuation of the characteristic parameter is within an acceptable stable range. The method for determining the variance threshold is to collect the variance values ​​of the characteristic parameter during the stable compaction stage of multiple successful centrifugal molding processes, calculate their statistical mean and standard deviation, and set the variance threshold as the statistical mean plus twice the standard deviation. An example variance threshold could be 0.0001. The comparison operation determines whether the calculated variance of the characteristic parameter within the specified time window is less than the preset variance threshold. If it is less, the short-term volatility of the characteristic parameter is considered low.

[0091] The current value of the nonstationarity metric is compared with a preset stability threshold. The preset stability threshold is set differently depending on the type of nonstationarity metric used. If the nonstationarity metric is the coefficient of variation, its stability threshold is determined by analyzing the statistical distribution of the coefficient of variation of the variance sequence of the material state signal components in the stable compaction stage; for example, it can be set to the upper 95th percentile of the distribution, with an example value of 0.5. If the nonstationarity metric is a detrended fluctuation analysis scaling exponent, its stability threshold is usually set to a value slightly higher than 0.5, such as 0.7, to determine whether the fluctuation has excessive persistence, as a large scaling exponent may also indicate slow process adjustment or long-range anomalous correlations. The comparison operation determines whether the current value of the nonstationarity metric is less than the preset stability threshold. If it is less, the energy distribution of the material state signal components is considered relatively stable over time.

[0092] When the rate of change of characteristic parameters is lower than the rate of change threshold, the variance of characteristic parameters within a specified time window is lower than the variance threshold, and the current value of the non-stationarity metric is lower than the stability threshold, the real-time compaction state is determined to have entered a stable compaction stage, and the uniformity of the compaction process is considered excellent. The logic behind this determination is that a low rate of change of characteristic parameters indicates that the evolution rate of the internal structure of the material has slowed down and is approaching completion; a low variance of characteristic parameters indicates that the evolution process is smooth and without drastic repetitions; and a low non-stationarity metric indicates that the energy distribution of the material's state signal fluctuates little over time, indirectly reflecting that the compaction process progresses uniformly in the axial or radial direction, without any localized drastic changes or stagnation. When all three conditions are met simultaneously, the concrete material is comprehensively determined to have reached a good and uniform compaction state.

[0093] Otherwise, the real-time compaction state is determined to be in the transitional compaction stage or the compaction process is considered poorly balanced. The "otherwise" conditions include at least one of the following three scenarios: the rate of change of the characteristic parameter is not lower than the rate of change threshold; the variance of the characteristic parameter within a specified time window is not lower than the variance threshold; or the current value of the non-stationarity measure is not lower than the stability threshold. If the rate of change of the characteristic parameter is not lower than the rate of change threshold, it indicates that the internal structure of the material is still evolving rapidly and has not yet reached stability. If the variance of the characteristic parameter within a specified time window is not lower than the variance threshold, it indicates that the structural evolution process is fluctuating or repetitive. If the current value of the non-stationarity measure is not lower than the stability threshold, it indicates that the energy distribution of the material state signal changes significantly over time, possibly reflecting local uneven compaction or intermittent processes. If any one of these conditions is not met, the real-time compaction state is determined to have not yet entered a stable stage, or there is a risk of uneven compaction; these are collectively referred to as the real-time compaction state being in the transitional compaction stage or the compaction process being considered poorly balanced. This determination provides a clear basis for subsequent control decisions.

[0094] S6. Based on the real-time compaction status and the uniformity of the compaction process, adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage. Specifically, this is implemented as follows:

[0095] The corresponding control logic is executed based on the evaluation results. If the real-time compaction state is in the transitional compaction stage and the compaction process uniformity is poor, then the centrifuge speed is reduced in the current centrifugation stage. The purpose of reducing the speed is to mitigate the compaction process and reduce internal stress concentration or structural defects that may be caused by localized over-compaction or uneven material migration. The speed reduction is achieved by sending a command to the centrifuge's frequency converter or DC speed controller to reduce the target speed setpoint. This command contains the new, lower target speed value.

[0096] If the real-time compaction state is in the transitional compaction stage and the compaction process is highly uniform, then the centrifuge speed in the current centrifugation stage is increased. The purpose of increasing the speed is to accelerate the compaction process and improve production efficiency while maintaining good material uniformity. This speed increase is achieved by sending a command to the speed controller to increase the target speed setpoint. If the real-time compaction state is in the stable compaction stage, then the centrifuge is switched to the next centrifugation stage. Switching to the next centrifugation stage means ending the current centrifugation stage and starting the next centrifugation stage at a higher speed according to the preset centrifugation regime, or switching to the final shutdown and maintenance stage. This operation is achieved by sending a stage switching command to the centrifuge control system, which triggers the control system to load and execute the preset speed parameters for the next stage.

[0097] The speed adjustment of the centrifuge during the current centrifugation stage is calculated by performing a proportional-integral (PI) operation on the difference between the rate of change of the characteristic parameter and its threshold, and the difference between the current value of the non-stationarity metric and its stability threshold. The PI operation is a commonly used algorithm in industrial control, used to calculate appropriate control quantities based on the current error and its accumulated historical data. The specific implementation steps are as follows: First, calculate two error signals. The first error signal is the difference between the rate of change of the characteristic parameter and its threshold. This difference is calculated as the absolute value of the rate of change of the characteristic parameter minus the threshold. When the absolute value of the rate of change of the characteristic parameter is less than the threshold, it indicates that the process is stabilizing and no active adjustment is needed; when it is greater than the threshold, it indicates that the change is too rapid and adjustment is required. The second error signal is the difference between the current value of the non-stationarity metric and its stability threshold. This difference is calculated as the current value of the non-stationarity metric minus the stability threshold. A non-stationarity metric value exceeding the threshold indicates that the process is unstable.

[0098] Perform proportional and integral operations on each error signal separately. The output of the proportional operation is equal to the error signal multiplied by a proportional coefficient. The output of the integral operation is equal to the cumulative sum of the error signal over the most recent integration time multiplied by an integral coefficient. The integration time refers to the length of the time window for accumulating the error, for example, set to 30 seconds. The cumulative sum refers to the algebraic sum of the error signal calculated in each control cycle (e.g., once per second) within that integration time. The proportional and integral coefficients need to be determined using engineering tuning methods, such as the critical proportional method. By observing the system response under stepped inputs, adjust the proportional and integral coefficients to make the system response both fast and smooth, with no overshoot or only a small overshoot. An exemplary tuning result is a proportional coefficient of 0.5 rpm per unit error and an integral coefficient of 0.05 rpm per unit error per second.

[0099] The proportional terms of the two error signals are added together to obtain the total proportional term. The integral terms of the two error signals are added together to obtain the total integral term. The speed adjustment is equal to the total proportional term plus the total integral term. The sign of the speed adjustment determines whether the speed is increased or decreased. The sign of the speed adjustment is determined by the control logic: when the speed needs to be increased, the speed adjustment is positive; when the speed needs to be decreased, the speed adjustment is negative. The magnitude of the speed adjustment is dynamically determined through the above proportional-integral calculation. The calculated speed adjustment is a numerical value in units of speed (e.g., revolutions per minute).

[0100] A new target speed is determined based on the calculated speed adjustment. The new target speed equals the centrifuge's original preset speed in the current centrifugation stage plus the speed adjustment. However, the new target speed must be limited to a permissible safe operating range. This range is determined based on the centrifuge's mechanical design, mold strength, and process requirements; for example, a lower limit of 200 rpm and an upper limit of 800 rpm. If the calculated new target speed is lower than the permissible lower limit, the lower limit is used as the actual command; if it is higher than the permissible upper limit, the upper limit is used. The new target speed value, after being limited, is sent to the centrifuge speed controller for execution. This dynamic adjustment method based on proportional-integral calculations allows speed control to depend not only on the current state assessment results but also on the degree and duration of deviation of state parameters from the threshold, thus achieving smoother and more precise adaptive control and effectively guiding the centrifugation process towards a stable and balanced compaction state. The entire control cycle is executed periodically during the centrifugation process, for example, performing an assessment and adjustment once per second until the process ends or transitions to the next stage.

[0101] Example 2: Figure 2 A schematic diagram of a prestressed electric pole centrifugal forming control system based on material state is provided according to the present invention. The prestressed electric pole centrifugal forming control system based on material state includes the following modules:

[0102] The signal acquisition module is used to simultaneously acquire vibration and sound signals from the mold during centrifuge operation to obtain a mixed signal;

[0103] The blind source separation module is used to perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials.

[0104] The feature extraction module is used to perform multi-scale coarse-graining processing on the material state signal components to obtain time series at multiple scales, calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale, and extract feature parameters characterizing the evolution process of the internal structure of concrete from the curve.

[0105] The index calculation module is used to calculate the non-stationarity measurement index of the material state signal components.

[0106] The state assessment module is used to comprehensively evaluate the real-time compaction state and compaction process uniformity of concrete materials in the mold based on the changing trends of characteristic parameters and non-stationarity measurement indicators.

[0107] The parameter adjustment module is used to adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage based on the real-time compaction status and the uniformity of the compaction process.

[0108] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0109] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0110] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0112] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0113] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0114] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for controlling the centrifugal forming of prestressed electric poles based on material state, characterized in that, Includes the following steps: S1. During the operation of the centrifuge, the vibration signal and sound signal of the mold are collected simultaneously to obtain a mixed signal; S2. Perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials. S3. Perform multi-scale coarsening processing on the material state signal components to obtain time series at multiple scales. Calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale. Extract characteristic parameters characterizing the evolution process of the internal structure of concrete from the curve. S4. Calculate the non-stationarity metrics of the material state signal components, including: The material state signal components are divided into multiple continuous and equal-length analysis windows; For each analysis window, calculate the variance of the material state signal components within that analysis window; Construct the time series using the variances of all analysis windows; Calculate the coefficient of variation or the detrended volatility analysis scaling index for the time series, and use the coefficient of variation or the detrended volatility analysis scaling index as a measure of nonstationarity. S5. Based on the changing trends of characteristic parameters and non-stationarity measurement indices, comprehensively evaluate the real-time compaction state and compaction process uniformity of concrete materials inside the mold. S6. Based on the real-time compaction status and the uniformity of the compaction process, adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage.

2. The method for controlling the centrifugal forming of prestressed electric poles based on material state according to claim 1, characterized in that, Vibration and sound signals from the mold are simultaneously acquired during centrifuge operation to obtain a mixed signal, including: Vibration sensors are installed at at least two axial positions of the mold, and sound sensors are installed at the ends of the mold or on the centrifuge frame. The rotary encoder signal of the centrifuge spindle is used as the synchronous trigger reference to control the vibration sensor and the sound sensor to collect data at equal time intervals. The signals from all vibration sensors and sound sensors collected at the same time are combined to form a mixed signal.

3. The method for controlling the centrifugal forming of prestressed electric poles based on material state according to claim 1, characterized in that, Blind source separation processing is performed on the mixed signal to separate the material state signal components related to the changes in the state of concrete materials, including: The mixed signal is preprocessed by whitening to eliminate the correlation between the signals of each channel; The objective function is constructed based on the negative entropy maximization criterion, and the separation matrix is ​​found through a fixed-point iterative algorithm. The separation matrix is ​​used to perform a linear transformation on the whitened mixed signal to obtain several independent source signal estimation components; Based on the energy change characteristics of the source signal estimation components during the low-speed stage of the centrifuge startup, the component whose energy increases steadily with the speed and is independent of the mechanical resonance frequency of the mold is selected from several independent source signal estimation components as the material state signal component related to the state change of concrete material.

4. The method for controlling the centrifugal forming of prestressed electric poles based on material state according to claim 1, characterized in that, Multi-scale coarsening processing is performed on the material state signal components to obtain time series at multiple scales. The permutation entropy of the time series at each scale is calculated to form a curve of permutation entropy changing with scale. Characteristic parameters characterizing the evolution process of the internal structure of concrete are extracted from the curve, including: Multiple scale factors are set for the material state signal components, and coarse-grained processing is performed on the material state signal components based on each scale factor to generate a coarse-grained time series corresponding to each scale factor. For each coarse-grained time series, the embedding dimension and delay time for permutation entropy calculation are determined based on its sampling characteristics, and the permutation entropy value of the coarse-grained time series is calculated. Connect the permutation entropy values ​​corresponding to all scale factors in ascending order of scale factors to form a curve of permutation entropy as a function of scale. On the curve of permutation entropy changing with scale, identify the initial and final scales where the permutation entropy value transitions from the peak to a monotonically decreasing state. Calculate the linear fitting slope of the permutation entropy value relative to scale change between the initial and final scales, and use this linear fitting slope as a characteristic parameter characterizing the evolution process of the internal structure of concrete.

5. The method for controlling the centrifugal forming of prestressed electric poles based on material state according to claim 4, characterized in that, The range of multiple scale factors is set based on the dominant frequency period of the material state signal component. The minimum value of the scale factor is set to an integer multiple of the dominant frequency period, and the maximum value is set to the longest characteristic time scale required to cover the entire process of concrete evolution from fluid state to compact state.

6. A prestressed pole centrifugal forming control system based on material state, used to implement the prestressed pole centrifugal forming control method based on material state as described in any one of claims 1-5, characterized in that, Includes the following modules: The signal acquisition module is used to simultaneously acquire vibration and sound signals from the mold during centrifuge operation to obtain a mixed signal; The blind source separation module is used to perform blind source separation processing on the mixed signal to separate the material state signal component related to the changes in the state of concrete materials. The feature extraction module is used to perform multi-scale coarse-graining processing on the material state signal components to obtain time series at multiple scales, calculate the permutation entropy of the time series at each scale to form a curve of permutation entropy changing with scale, and extract feature parameters characterizing the evolution process of the internal structure of concrete from the curve. The index calculation module is used to calculate the non-stationarity measurement index of the material state signal components. The state assessment module is used to comprehensively evaluate the real-time compaction state and compaction process uniformity of concrete materials in the mold based on the changing trends of characteristic parameters and non-stationarity measurement indicators. The parameter adjustment module is used to adjust the centrifuge speed in the current centrifugation stage or the timing of transitioning to the next centrifugation stage based on the real-time compaction status and the uniformity of the compaction process.

Images