Concrete slump monitoring method, concrete slump monitoring device, and electronic device

Abstract

The application discloses a concrete slump monitoring method, a concrete slump monitoring device and electronic equipment, and relates to the field of concrete production.The concrete slump monitoring method comprises the following steps: acquiring vehicle-mounted multi-source sensing data; performing filtering processing on motion data to obtain low-frequency attitude parameters; monitoring a tank washing signal to identify an empty load working condition; updating a mechanical wear compensation torque by using an exponential smoothing formula; solving a false torque of a partial load based on the attitude parameters by using a rigid body dynamics partial load compensation formula, removing the false torque of the partial load and the wear compensation torque from an original torque, and obtaining a net rheological driving torque; performing phase locking and segmentation on the net torque according to phase data; decoupling the net torque into a low-frequency fundamental wave and a high-frequency residual component through variational mode decomposition; calculating a rheological state feature matrix comprising a yield impedance, a viscosity slope, a high-frequency entropy and a confidence level; triggering end-side domain adaptive fine adjustment; and outputting a slump monitoring result.The method can improve the accuracy of slump detection.

Description

Technical Field

[0001] This application relates to the field of concrete production technology, and in particular to a method, device and electronic equipment for monitoring concrete slump. Background Technology

[0002] In concrete construction and transportation, slump is a key indicator for measuring the workability and fluidity of concrete mixtures. Currently, traditional concrete slump monitoring mainly relies on manual sampling, which suffers from strong lag, low frequency, and inability to conduct continuous monitoring. In the field of vehicle-mounted dynamic monitoring, existing technologies typically use optical cameras or acoustic sensors installed inside the mixer truck to infer slump by collecting image or sound features. However, mixer trucks operate under harsh conditions, with large amounts of dust, cement slurry splashes, and intense mechanical noise inside the tank, making optical and acoustic sensors highly susceptible to contamination and interference, leading to signal distortion. Some technologies infer slump by detecting operating parameters of the concrete mixer truck motor, such as torque. While this method avoids optical and acoustic interference, the mixer truck motor torque is still affected by various environmental factors, such as the false load torque caused by uneven roads and the additional torque consumption due to vehicle aging and mechanical wear, thus causing deviations in slump detection. Summary of the Invention

[0003] This application provides a concrete slump monitoring method, a concrete slump monitoring device, and an electronic device to address the technical shortcomings of existing vehicle-mounted slump monitoring solutions, such as susceptibility to severe environmental pollution, inability to effectively isolate road dynamic off-center load interference, and mechanical wear reference drift.

[0004] Firstly, this application provides a method for monitoring the slump of concrete, comprising: Acquire multi-source heterogeneous sensor data collected by the vehicle sensor network to obtain vehicle six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system; The vehicle's six-degree-of-freedom motion data is spatiotemporally synchronized and subjected to extended Kalman filtering to filter out high-frequency road vibration interference, thereby obtaining low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and angular acceleration of the mixing tank that characterize the vehicle's macroscopic attitude. The water tank filling signal and the high-speed reverse rotation signal of the mixing tank are detected to identify the no-load condition of the washing tank. The original torque data of the hydraulic drive system under this condition is extracted as the no-load measured torque. Based on the no-load measured torque, the mechanical wear compensation torque of the equipment throughout its entire life cycle is dynamically updated using the exponential smoothing evolution formula. Based on the low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and angular acceleration of the mixing tank rotation, the spurious off-center load torque caused by vehicle dynamics is calculated in reverse using the rigid body dynamics off-center load compensation formula. This spurious off-center load torque, together with the mechanical wear compensation torque, is differentially removed from the original torque data of the hydraulic drive system to obtain the net rheological drive torque that reflects only the Bingham rheological properties of concrete, stripped of road surface interference and mechanical dissipation. Based on the real-time phase data of the mixing tank, the net rheological driving torque is phase-locked and time-domain segmented. The net rheological driving torque is decoupled into a low-frequency fundamental component characterizing macroscopic rheological resistance and a high-frequency residual component characterizing aggregate collision through variational mode decomposition. The macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision are extracted to construct a rheological state feature matrix. Specifically, the extraction of the macroscopic yield resistance index, dynamic plastic viscosity slope, and aggregate collision high-frequency entropy to construct a rheological state feature matrix includes: [the following is a description of the process, not a direct translation] for the net rheological driving torque. Perform variational mode decomposition to obtain Individual eigenmode functions ; Calculate the energy proportion of each intrinsic mode. And calculate the high-frequency entropy of aggregate collisions based on the Shannon entropy formula. Extract the K intrinsic mode functions. The modal component with the lowest center frequency is taken as the low-frequency fundamental component. The low-frequency fundamental component is linearly fitted using the least squares method to extract the macroscopic yield resistance index and the dynamic plastic viscosity slope. The extracted macroscopic yield resistance index, dynamic plastic viscosity slope and the high-frequency entropy of aggregate collision are combined according to a preset dimension to form the rheological state feature matrix. The rheological state feature matrix is ​​dynamically evaluated and corrected. When the confidence level of the stability of the working condition ξ is continuously lower than the safety baseline or the KL divergence of the feature distribution exceeds the drift threshold, domain adaptation fine-tuning is performed, and the concrete slump monitoring results are output.

[0005] For example, the method of dynamically updating the mechanical wear compensation torque throughout the entire life cycle of the equipment using an exponential smoothing evolution formula specifically includes: When the tank washing process is detected to be under no-load operation, the measured value of the pure no-load torque for that period is extracted. And perform adaptive smoothing updates of the time series according to the following formula:

[0006] in, For the mechanical wear compensation torque of the current cycle, This is the mechanical wear compensation torque from the previous cycle. This is the forgetting smoothing coefficient, and its value range is... , The original torque of the washing tank under no-load conditions. The moment of inertia of the mixing tank under no-load conditions. The rated speed of the washing tank, This refers to the mechanical transmission efficiency of the hydraulic system.

[0007] For example, the reverse calculation of the spurious off-center load torque caused by vehicle dynamics using the rigid body dynamics off-center load compensation formula specifically includes: calculating the net rheological drive torque according to the following rigid body dynamics off-center load compensation formula. :

[0008] in, For net rheological drive torque, The original driving torque, Torque compensation for mechanical wear, The equivalent eccentric load center of mass of the concrete. Let the radius of rotation be the center of mass offset. It is the acceleration due to gravity. The instantaneous pitch angle of the vehicle. The instantaneous roll angle of the vehicle. The rotational angular acceleration of the mixing tank. Let the equivalent radius of the mixing tank be . It is the lateral translational acceleration. This refers to the vertical turbulence acceleration.

[0009] For example, the high-frequency entropy of the aggregate collision The calculation formula is: .

[0010] For example, the dynamic evaluation and correction of the rheological state feature matrix specifically includes: Real-time calculation of vehicle driving variance and tank rotation speed fluctuation rate yields the confidence level of operating condition stability. ;when If the current condition is determined to be severe turbulence or a sharp turn, immediately suspend the slump prediction output for the current cycle and lock and maintain the high-confidence inference result from the previous cycle until... And consistently exceeding The suspension will be lifted after one cycle; At the same time, when aggregates collide with high-frequency entropy Higher than the preset reference line At that time, a nonlinear penalty coefficient is introduced. The macroscopic yield resistance index is attenuated and modulated in real time, and the modulated yield resistance index is... satisfy:

[0011] in, The original macroscopic yield resistance index. The mechanism penalty weighting coefficient, This is a preset threshold.

[0012] For example, the execution domain adaptation fine-tuning specifically includes: Calculate the online distribution of the feature vector input to the current vehicle-mounted edge computing unit. Distribution of pre-trained model libraries in the cloud KL divergence between :

[0013] When calculated Exceeding the safe drift threshold When a substantial change is detected in the concrete mix proportion or material source, a manual inspection prompt is sent to the vehicle terminal. After obtaining slump labels from 1 to 2 real manual inspections, the weights of the feature extraction layer of the lightweight neural network are frozen, and backpropagation fine-tuning is performed only on the last fully connected layer.

[0014] Furthermore, the method also includes: Historical sensor data of different concretes under different working conditions were collected, and the corresponding macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision were extracted to generate a training set of rheological state feature matrix. Construct a lightweight neural network that includes an input layer, a feature extraction layer, and a fully connected output layer; Using the rheological state feature matrix training set as input and the corresponding collapse degree ground truth as supervision label, the lightweight neural network is trained until convergence to obtain a pre-trained model, and the pre-trained model is deployed as the initial model.

[0015] Secondly, this application provides a concrete slump monitoring device, comprising: The sensor data acquisition module is used to acquire multi-source heterogeneous sensor data collected by the vehicle sensor network to obtain vehicle six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system. The data processing module is used to perform spatiotemporal synchronization and extended Kalman filtering on the vehicle's six-degree-of-freedom motion data to filter out high-frequency road vibration interference and obtain low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle and mixing tank rotational angular acceleration that characterize the vehicle's macroscopic attitude. The no-load identification and compensation module is used to detect the water tank filling signal and the high-speed reverse rotation signal of the mixing tank to identify the no-load condition of the washing tank, and dynamically update the mechanical wear compensation torque throughout the entire life cycle of the equipment using the exponential smooth evolution formula. The feature extraction module is used to calculate the false load torque caused by vehicle dynamic driving in reverse using the rigid body dynamics eccentric load compensation calculation formula based on the low-frequency translational acceleration, lateral acceleration, pitch angle, side tilt angle and the rotational angular acceleration of the mixing tank. The false load torque is then differentially removed from the original torque data along with the mechanical wear compensation torque to obtain the net rheological driving torque that reflects only the Bingham rheological properties of concrete, stripped of road surface interference and mechanical dissipation. The rheological characteristic determination module is used to perform phase locking and time-domain segmentation of the net rheological driving torque based on the real-time phase data of the mixing tank. The net rheological driving torque is decoupled into a low-frequency fundamental component characterizing macroscopic rheological resistance and a high-frequency residual component characterizing aggregate collision through variational mode decomposition. The macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision are extracted to construct a rheological state characteristic matrix. Specifically, the extraction of the macroscopic yield resistance index, dynamic plastic viscosity slope, and aggregate collision high-frequency entropy to construct a rheological state feature matrix includes: [the following is a description of the process, not a direct translation] for the net rheological driving torque. Variational mode decomposition is performed to obtain K eigenmode functions. ; Calculate the energy proportion of each intrinsic mode. And calculate the high-frequency entropy of aggregate collisions based on the Shannon entropy formula. Extract the K intrinsic mode functions. The modal component with the lowest center frequency is taken as the low-frequency fundamental component. The low-frequency fundamental component is linearly fitted using the least squares method to extract the macroscopic yield resistance index and the dynamic plastic viscosity slope. The extracted macroscopic yield resistance index, dynamic plastic viscosity slope and the high-frequency entropy of aggregate collision are combined according to a preset dimension to form the rheological state feature matrix. The slump output module is used to dynamically evaluate and correct the rheological state feature matrix. When the confidence level of the working condition stability ξ is continuously lower than the safety baseline or the KL divergence of the feature distribution exceeds the drift threshold, the domain adaptation fine-tuning is performed and the concrete slump monitoring results are output.

[0016] Thirdly, this application provides an electronic device including a memory and one or more processors. The memory stores one or more computer programs, each including instructions that, when executed by the processor, cause the electronic device to perform the concrete slump monitoring method as described in the first aspect.

[0017] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed on an electronic device, cause the electronic device to perform the concrete slump monitoring method as described in the first aspect.

[0018] Fifthly, this application provides a computer program product that, when run on an electronic device, causes the electronic device to perform the concrete slump monitoring method as described in the first aspect.

[0019] Understandably, the beneficial effects achievable by the concrete slump monitoring methods, devices, electronic equipment, computer-readable storage media, and computer program products provided above can be referenced as follows: Due to the influence of road inclination, bump acceleration, and tank rotation acceleration during actual operation, the original torque of the drive motor of the mixer truck contains a large amount of spurious off-center load torque caused by gravity and inertia. This solution uses a six-degree-of-freedom IMU to capture the vehicle's attitude in real time and substitutes it into the rigid body dynamics off-center load compensation calculation formula to accurately remove the introduced mechanical dynamic interference from a physical perspective. At the same time, it dynamically subtracts the mechanical wear reference torque by combining the exponential smooth evolution formula under the tank washing no-load condition. The net rheological drive torque obtained after the above purification eliminates the interference of multi-source environmental noise on the machine learning model, which is beneficial to improving the accuracy of slump inversion. On this basis, high-frequency entropy characterizing the collision of aggregates inside the material is extracted by variational mode decomposition, and feature correction is performed using multi-level mechanism judgment logic, so that the end-side lightweight neural network only needs to receive pure physical feature input, thereby improving the accuracy of slump monitoring. This solution overcomes the technical limitations of traditional current methods in accurately measuring slump under dynamic road conditions, avoids the need to install easily contaminated optical or acoustic sensors inside the tank, and achieves a deep integration of physical decoupling and artificial intelligence feature extraction, greatly improving the accuracy of slump detection. Attached Figure Description

[0020] Figure 1 This is a schematic flowchart of the concrete slump monitoring method provided in the embodiments of this application. Detailed Implementation

[0021] To facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with substantially the same function and effect. For example, "first chip" and "second chip" are merely used to distinguish different chips and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different. It should be noted that in the embodiments of this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as exemplary or for example in this application should not be construed as being better or more advantageous than other embodiments or design schemes. Specifically, the use of "exemplary" or "for example" is intended to present related concepts in a specific manner. In the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more.

[0022] It should be noted that in the embodiments of this application, "at..." can refer to the instant when a certain situation occurs, or it can refer to a period of time after a certain situation occurs. The embodiments of this application do not specifically limit this.

[0023] The implementation of this embodiment will now be described in detail with reference to the accompanying drawings.

[0024] This embodiment provides a method for monitoring the slump of concrete. Figure 1 A schematic flowchart of the concrete slump monitoring method provided in the embodiments of this application is shown.

[0025] like Figure 1 As shown, the concrete slump monitoring method may include the following steps: Step 1: Acquire multi-source heterogeneous sensor data collected by the vehicle sensor network to obtain vehicle six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system.

[0026] The vehicle's six-degree-of-freedom motion data includes displacement along the X, Y, and Z translational axes and rotation around these three axes, namely longitudinal, lateral, and vertical translation, as well as pitch, roll / tilt, and yaw / roll rotation. Real-time phase data of the mixing tank can be precisely measured using a rotary encoder, specifically the angular displacement, angular velocity, and angular acceleration of the mixing tank. The raw torque data of the hydraulic drive system is the output torque of the concrete mixer truck's hydraulic motor.

[0027] The sensors installed on the concrete mixer truck include current sensors and the vehicle's IMU (Inertial Measurement Unit) module. The vehicle's sensor network can collect six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system.

[0028] Step 2: Perform spatiotemporal synchronization and extended Kalman filtering on the vehicle's six-degree-of-freedom motion data to filter out high-frequency road vibration interference and obtain low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and angular acceleration of the mixing tank that characterize the vehicle's macroscopic attitude.

[0029] Specifically, the spatiotemporal synchronization and extended Kalman filtering of the vehicle's six-degree-of-freedom motion data includes: resampling and aligning the sampling data of the inertial measurement unit, the rotary encoder of the mixing tank, and the drive system to the same reference time axis through a hardware-level deterministic microsecond clock triggering mechanism; introducing vehicle chassis dynamic constraints into the extended Kalman filter to eliminate mechanical white noise caused by high-frequency road vibrations and extracting low-frequency translational acceleration characterizing the vehicle's macroscopic attitude. lateral acceleration Pitch angle yaw angle and the rotational angular acceleration of the mixing tank .

[0030] Step 3: Detect the water tank filling signal and the high-speed reverse rotation signal of the mixing tank to identify the no-load condition of the washing tank. Extract the original torque data of the hydraulic drive system under this condition as the no-load measured torque, and dynamically update the mechanical wear compensation torque of the equipment throughout its entire life cycle based on the no-load measured torque using the exponential smoothing evolution formula.

[0031] The system determines whether a reversal signal has occurred based on real-time phase data from the mixing tank, and monitors the real-time water injection status using a water flow sensor on the water outlet pipe of the mixer truck's water tank. When both a water tank injection signal and a reversal signal are detected, the system is determined to be in a tank-washing no-load condition. Upon identifying this no-load condition, the measured value of the pure no-load torque for that period is extracted. The time series is then adaptively smoothed and updated according to the following exponential smoothing evolution formula:

[0032] in, For the mechanical wear compensation torque of the current cycle, This is the mechanical wear compensation torque from the previous cycle. This is the forgetting smoothing coefficient, and its value range is... , The original torque of the washing tank under no-load conditions. The moment of inertia of the mixing tank under no-load conditions. The rated speed of the washing tank, This refers to the mechanical transmission efficiency of the hydraulic system.

[0033] Furthermore, 1. The data is based on actual measurements. After confirming that the washing tank has entered the no-load condition and the stirring tank speed has reached a stable stage, data is collected through the pressure sensor in the hydraulic drive system (which is converted into motor output torque) or directly through the torque sensor. Multiple sets of torque data within the time window of this condition are extracted and the time average is calculated to eliminate random interference caused by transient fluid fluctuations and mechanical vibrations, thereby obtaining the original measured torque value representing the current no-load benchmark.

[0034] Historical iteration data is stored in non-volatile memory. After the mixer truck leaves the factory for the first time or undergoes major overhaul and recalibration, a factory no-load baseline value is written as the initial value. After each tank washing process throughout the entire lifecycle of the equipment, the results calculated in this study will be used... It is overwritten to memory and used as input for the next calculation.

[0035] The empirically set values ​​take into account that mechanical wear (such as raceway wear and reducer aging) is a slow, gradual, long-term process, and sudden torque changes during a single tank washing process are often caused by accidental factors (such as poor local lubrication or foreign matter). Therefore, The value is set between 0.85≤α≤0.99, which is intended to give a very high weight to the historical cumulative wear amount, thereby playing a low-pass filtering role to resist interference and prevent the reference torque from being updated erroneously in a step.

[0036] These are inherent mechanical constants, which are theoretically calculated by the mixer truck manufacturer using a 3D CAD model or calibrated offline through a standard no-load acceleration and deceleration drag test before the vehicle leaves the factory, and are configured as fixed parameters in the on-board system.

[0037] It is a real-time operating condition status quantity, which can be obtained by differential differentiation of the real-time phase data collected by the rotary encoder, or by directly reading the washing tank speed control command issued by the vehicle chassis CAN bus.

[0038] To dynamically match variables, since hydraulic transmission efficiency varies with temperature and speed, a transmission efficiency MAP (bench test curve) provided by the hydraulic system supplier can be pre-stored in the onboard system. Based on the current operating temperature data and motor speed, a precise mechanical transmission efficiency is matched in real time using a two-dimensional lookup table and interpolation algorithm.

[0039] For example, in this embodiment, a rotary encoder installed on the drive shaft or raceway of the mixing tank continuously collects real-time phase data of the mixing tank. This phase data specifically includes the real-time phase angle. The system obtains the instantaneous rotational angular velocity of the mixing tank by performing time-domain first derivative calculation on the real-time phase angle. The rotational state of the mixing tank is divided into forward rotation (corresponding to feeding or normal stirring conditions, in which case the angular velocity is positive) and reverse rotation (corresponding to unloading or tank washing conditions, in which case the angular velocity is negative) according to the set reference working direction.

[0040] When no reversal signal occurs: the instantaneous rotational angular velocity extracted in real time is greater than or equal to zero (indicating that it is in forward rotation or stationary state); or, although the instantaneous rotational angular velocity is negative (indicating that it is in reverse state), its absolute value is less than the preset cleaning speed threshold (indicating that the current condition may only be low-speed unloading or reverse fine-tuning). In this case, it is determined that no high-speed reversal signal has occurred, the current monitoring state is maintained, and the reference torque update is not triggered.

[0041] When a high-speed reversal signal occurs: the instantaneous rotational angular velocity extracted in real time is negative, and within a preset continuous time window (e.g., 3 to 5 seconds), the absolute value of this negative angular velocity is continuously greater than or equal to the cleaning speed threshold (this threshold is usually set in the high-speed range between 10 r / min and 14 r / min). At this time, a high-speed reversal signal is determined to have occurred.

[0042] If the water tank filling sensor detects a continuous water tank filling signal at the same time as the high-speed reversal signal is detected, the vehicle is accurately locked into the tank washing no-load condition through logical operation. This triggers the extraction of the no-load measured torque during this period and uses the exponential smooth evolution formula to dynamically update the mechanical wear compensation torque.

[0043] Step 4: Based on the low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and angular acceleration of the mixing tank rotation, the illusory load torque caused by the vehicle's dynamic movement is calculated in reverse using the rigid body dynamics eccentric load compensation formula. This illusory load torque, along with the mechanical wear compensation torque, is differentially removed from the original torque data of the hydraulic drive system to obtain the net rheological drive torque that reflects only the Bingham rheological properties of the concrete, stripped of road surface interference and mechanical dissipation.

[0044] The spurious off-center load torque caused by vehicle dynamics is calculated in reverse using the rigid body dynamics off-center load compensation formula. Specifically, this includes calculating the net rheological drive torque according to the following rigid body dynamics off-center load compensation formula. :

[0045] in, For net rheological drive torque, The original driving torque, Torque compensation for mechanical wear, The equivalent eccentric load center of mass of the concrete. Let the radius of rotation be the center of mass offset. It is the acceleration due to gravity. The instantaneous pitch angle of the vehicle. The instantaneous roll angle of the vehicle. The rotational angular acceleration of the mixing tank. Let the equivalent radius of the mixing tank be . It is the lateral translational acceleration. This refers to the vertical turbulence acceleration.

[0046] During actual road operation, the concrete mixer truck is affected by dynamic conditions such as road slopes, uneven road surfaces, and turns. Due to inertia, the concrete mixture inside the mixer will generate an additional alternating torque relative to the rotation axis of the mixer, i.e., a false off-center load torque. This implementation method, based on rigid body dynamics principles, uses the vehicle's six-degree-of-freedom motion attitude (pitch angle, roll angle, vertical bump acceleration, lateral translational acceleration) and the mixer's kinematic parameters (rotational angular acceleration, equivalent radius R) as state variables to construct a rigid body dynamics off-center load compensation calculation formula. This is achieved by using the center-of-gravity offset distance (m...) r) will affect the gravitational component ( , ) and inertial acceleration components ( , , This is uniformly converted into additional torque acting on the drive shaft. The so-called reverse calculation involves starting from the macroscopically measurable chassis attitude parameters and deriving the illusory load component superimposed on the drive system by external road excitation, then performing differential calculations (…). The off-center load term is removed from the original drive torque in real time, thereby completing the physical-level interference removal.

[0047] Net rheological drive torque obtained after differential elimination It eliminates multi-source dynamic interference such as road bumps, slope inclination and lateral acceleration during turns, and retains only the useful signals that reflect the Bingham rheological properties of concrete itself (macroscopic yield stress, plastic viscosity), thus achieving signal purification from mixed noise to physical purity.

[0048] in, For real-time operating conditions, differential pressure data is collected in real time by a high-pressure sensor in the hydraulic system during heavy-load driving and mixing, and then calculated in conjunction with the hydraulic motor displacement coefficient; or it can be read directly in real time by a dynamic torque sensor connected in series on the drive shaft.

[0049] The latest cycle mechanical wear compensation torque value, which was dynamically updated and stored using the exponential smoothing evolution formula in step 3, is directly called as the deduction benchmark.

[0050] , , , The four parameters are low-frequency dynamic parameters reflecting the vehicle's macroscopic attitude. They are all derived from the high-confidence output results of the vehicle's six-degree-of-freedom motion data after spatiotemporal synchronization and extended Kalman filtering in the previous steps. They are obtained by directly subscribing to the real-time state variables published by the underlying filtering algorithm.

[0051] Derived from the previous steps, the phase angle data collected in real time by the rotary encoder installed on the mixing tank is subjected to a second time derivative and then smoothed by Kalman filtering.

[0052] and These are physical and geometric constants. Preset (Or adjust slightly according to local gravitational acceleration); The parameters are calculated by the mixer truck manufacturer before delivery based on the maximum inner diameter of the tank and the geometric structure of the internal spiral blades, and stored as fixed configuration parameters.

[0053] This represents the actual mass of concrete loaded in the tank. In actual engineering projects, the mass can be calculated by obtaining the total vehicle weight in real time through weighing sensors installed on the vehicle chassis suspension and deducting the empty vehicle weight; or by receiving the actual loading mass of this transportation task issued by the mixing plant through the vehicle networking system (e.g., delivery note data) and assigning a static value.

[0054] As the mixture rotates with the tank, its center of mass will deviate from the tank's central axis. This offset radius is closely related to the lead and tilt angle of the spiral blades inside the tank, as well as the current loading rate (i.e., the percentage of material volume in the tank's volume). A pre-built nonlinear mapping table of loading rate and offset radius, calibrated by fluid dynamics simulation (CFD), allows the current offset rotation radius to be determined by interpolation after calculating the loading rate based on the known load amount before vehicle start-up, by looking up the table. .

[0055] Step 5: Based on the real-time phase data of the mixing tank, the net rheological driving torque is phase-locked and time-domain segmented. The net rheological driving torque is decoupled into a low-frequency fundamental component representing macroscopic rheological resistance and a high-frequency residual component representing aggregate collision through variational mode decomposition. The macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision are extracted to construct the rheological state feature matrix.

[0056] Phase locking refers to using real-time phase data (phase angle) of the mixing tank captured by a rotary encoder to synchronously align the continuously acquired net rheological drive torque timing signal according to each rotation of the tank (360°) or a specific phase interval. Time domain segmentation refers to slicing the aligned timing signal according to the time domain, thereby transforming the continuous flow signal without beginning or end into periodic structured analysis samples.

[0057] The process of constructing the rheological state characteristic matrix includes: [the following steps are taken to determine the net rheological driving torque]. Perform variational mode decomposition to obtain eigenmode functions ; Calculate the energy proportion of each intrinsic mode. And calculate the high-frequency entropy of aggregate collisions based on the Shannon entropy formula. The modal component with the lowest center frequency among the K intrinsic mode functions is extracted as the low-frequency fundamental component. The low-frequency fundamental component is then linearly fitted using the least squares method to extract the macroscopic yield resistance index and the dynamic plastic viscosity slope. The extracted macroscopic yield resistance index, dynamic plastic viscosity slope, and aggregate collision high-frequency entropy are combined according to a preset dimension to construct the rheological state feature matrix. Aggregate collision high-frequency entropy The calculation formula is:

[0058] in, This represents the integration time window, specifically the time domain interval for energy integration. Based on the real-time phase data of the stirred tank locked in the previous steps, [the following steps are performed]. The setting is for the mixing tank to complete one full rotation (i.e., the phase angle change). The corresponding time period is used to ensure the periodic integrity and physical consistency of feature extraction.

[0059] This represents the number of intrinsic mode functions and is also a hyperparameter for the number of decomposition levels in the Variational Mode Decomposition (VMD) algorithm. During initialization, the center frequency distribution under different decomposition levels can be calculated by extracting historical sample data. The optimal number of decomposition levels is determined when the difference between the center frequencies of adjacent modes is lower than a preset aliasing threshold. (The value is usually between 4 and 8) and is fixed in the model preprocessing configuration.

[0060] For the first Each intrinsic mode function is also a representation of the net rheological driving torque. The time-domain signal sequence directly output after VMD decomposition, each Corresponding to the torque fluctuation component within a specific frequency band, the high-frequency mode function mainly characterizes the abrupt change in microscopic resistance caused by the collision of coarse aggregates during the mixing process.

[0061] Step 6: Perform dynamic evaluation and correction on the rheological state feature matrix. When the confidence level ξ of the working condition stability is continuously lower than the safety baseline or the KL divergence of the feature distribution exceeds the drift threshold, perform domain adaptation fine-tuning and output the concrete slump monitoring results.

[0062] The rheological state feature matrix is ​​dynamically evaluated and corrected, specifically including: Level 1: Dynamic hard threshold determination of operating condition stability: Real-time calculation of vehicle driving variance and tank rotation speed fluctuation rate to obtain the confidence level of operating condition stability. ;when If the current condition is determined to be severe turbulence or a sharp turn, immediately suspend the slump prediction output for the current cycle and lock and maintain the high-confidence inference result from the previous cycle until... And consistently exceeding After one cycle, the suspension is lifted; Level 2: Nonlinear penalty modulation: When aggregate collisions with high-frequency entropy Higher than the preset reference line At that time, a nonlinear penalty coefficient is introduced. The macroscopic yield resistance index is attenuated and modulated in real time, and the modulated yield resistance index is... satisfy:

[0063] in, The original macroscopic yield resistance index. The mechanism penalty weighting coefficient, This is a preset threshold.

[0064] Furthermore, and All are directly output from the preceding feature extraction steps. This reflects the baseline flow resistance of the current mixture as a whole, while It reflects in real time the intensity of collisions of large aggregate particles in the current mixture.

[0065] This represents the nominal value of the high-frequency entropy generated by mixing and collision under standard mix proportions (i.e., reference concrete with normal aggregate gradation and no need for trigger domain adaptation). Before the system leaves the factory, the expected value of its high-frequency entropy is calculated by collecting standard concrete rheological data under standard laboratory conditions, and this expected value is calibrated as follows. Stored in the system. In real time... Less than or equal to hour, If the output is 0, modulation is not triggered.

[0066] The attenuation penalty for the yield resistance index is used to control the degree of attenuation when the high-frequency entropy exceeds the limit, and its value range is usually set to [value range missing]. The specific values ​​are derived from the empirical regression curve established during the R&D phase, which compares the increase in resistance caused by excessive aggregate with the actual slump reduction. By introducing this penalty mechanism, the model can be forced to ignore the spurious increase in resistance caused by physical slump in the aggregate, thus reflecting a more realistic slump condition.

[0067] The domain adaptation fine-tuning includes: calculating the online distribution of the feature vector input to the current vehicle edge computing unit. Distribution of pre-trained model libraries in the cloud KL divergence between :

[0068] When calculated Exceeding the safe drift threshold When a substantial change is detected in the concrete mix proportion or material source, a manual inspection prompt is sent to the vehicle terminal. After obtaining slump labels from 1 to 2 real manual inspections, the weights of the feature extraction layer of the lightweight neural network are frozen, and backpropagation fine-tuning is performed only on the last fully connected layer.

[0069] in, For feature statistical intervals / histogram binning, since the features in the rheological state feature matrix (such as macroscopic yield resistance index, aggregate impact high-frequency entropy, etc.) are continuous variables, in order to perform discrete probability distribution calculations, the system predetermines the effective range of each feature and divides it into equal intervals. Each statistical interval (i.e., data bin). That is to represent the first Statistical intervals ( ).

[0070] This represents the baseline prior probability distribution learned by the model during cloud training. During the cloud-based model training phase, the system statistically analyzes the data on the validation set, ensuring they fall within each statistical interval. The sample frequencies within the range are normalized and then solidified into a baseline probability vector. It is then deployed to the system memory along with the initial pre-trained model.

[0071] This represents the real-time posterior probability distribution under the current actual operating conditions. During system operation, a set capacity (e.g., including past data) is maintained. A sliding data window (within one sampling period) is used. Whenever a new rheological state feature matrix is ​​input, the system updates the feature data within this window in real time and counts the data falling within each statistical interval. The real-time frequency within the range is normalized to obtain the online probability distribution. To avoid situations where the denominator or argument is zero in logarithmic operations, [the following will be done]: Apply a very small smoothing constant (e.g.) (i.e., Laplace smoothing).

[0072] Characterizing the current online distribution Deviation from pre-training baseline distribution The magnitude of information entropy cost. The larger the value, the greater the difference between the physical properties of the concrete in the current mixing tank and the standard properties seen during model training.

[0073] This is the decision threshold for triggering adaptive fine-tuning. It is determined by researchers during the model testing phase by artificially introducing abnormal aggregates or changing the water-cement ratio into concrete samples, observing the inflection point of the system's output accuracy decrease, and recording the value corresponding to this inflection point. The value is set to the safe drift threshold. When in real time Furthermore, when the set time period is continuously reached, it can be confirmed that a substantial change has occurred in the environment or materials, and then a spot check prompt is sent to the terminal and the back propagation fine-tuning of the network's last layer weights is initiated.

[0074] This implementation employs a multi-level composite mechanism judgment logic. The first level filters out instantaneous distorted data caused by extreme adverse conditions such as rapid vehicle acceleration or speed bumps through a hard confidence threshold, ensuring the stability of the output results. The second level performs nonlinear correction on the macroscopic impedance using high-frequency entropy, eliminating the interference of pseudo-yield resistance caused by the increase in the proportion of coarse aggregate. Finally, by calculating the KL divergence to monitor the drift of the characteristic distribution, when there is a substantial change in the cement grade or aggregate batch, only 1 to 2 manual sampling labels need to be input to quickly complete the end-side domain adaptation through final-level fine-tuning, avoiding the cumbersome process of returning the equipment to the factory for recalibration.

[0075] This implementation also includes: collecting historical sensing data of different concretes under different working conditions, generating a rheological state feature matrix training set based on the historical sensing data; constructing a lightweight neural network including an input layer, a feature extraction layer, and a fully connected output layer; using the rheological state feature matrix training set as input and the corresponding slump ground truth as supervision label, training the lightweight neural network until convergence to obtain a cloud-based pre-trained model library, and deploying the model library to the edge computing unit as the initial model.

[0076] Furthermore, this embodiment also provides a concrete slump monitoring device, which can be used to execute the above-described concrete slump monitoring method. The concrete slump monitoring device includes: a sensor data acquisition module, used to acquire multi-source heterogeneous sensor data collected by an on-board sensor network to obtain vehicle six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system; a data processing module, used to perform spatiotemporal synchronization and extended Kalman filtering on the vehicle six-degree-of-freedom motion data to filter out high-frequency road vibration interference, obtaining low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and mixing tank rotational angular acceleration characterizing the vehicle's macroscopic attitude; an unloaded identification and compensation module, used to detect water tank filling signals and high-speed reverse rotation signals of the mixing tank to identify the unloaded washing condition, extract the raw torque data of the hydraulic drive system under this condition as the unloaded measured torque, and dynamically update the mechanical wear compensation torque for the entire life cycle of the equipment based on the unloaded measured torque using an exponential smoothing evolution formula; and a feature extraction module, used to extract the low-frequency translational acceleration, lateral acceleration, pitch angle, roll ... The tilt angle and angular acceleration of the mixing tank are used to inversely calculate the spurious off-center load torque caused by vehicle dynamics using rigid body dynamics off-center load compensation formula. This spurious off-center load torque, along with the mechanical wear compensation torque, is differentially removed from the original torque data to obtain the net rheological driving torque, which reflects only the Bingham rheological characteristics of concrete, stripped of road surface interference and mechanical dissipation. A rheological feature determination module is used to perform phase locking and time-domain segmentation of the net rheological driving torque based on the real-time phase data of the mixing tank. Through variational mode decomposition, the net rheological driving torque is decoupled into a low-frequency fundamental component characterizing macroscopic rheological resistance and a high-frequency residual component characterizing aggregate collision. The macroscopic yield resistance index, dynamic plastic viscosity slope, and high-frequency entropy of aggregate collision are extracted to construct a rheological state feature matrix. Specifically, the extraction of the macroscopic yield resistance index, dynamic plastic viscosity slope, and high-frequency entropy of aggregate collision to construct the rheological state feature matrix includes: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] Variational mode decomposition is performed to obtain K eigenmode functions. ; Calculate the energy proportion of each intrinsic mode. And calculate the high-frequency entropy of aggregate collisions based on the Shannon entropy formula. Extract the K intrinsic mode functions. The modal component with the lowest center frequency is taken as the low-frequency fundamental component. The low-frequency fundamental component is linearly fitted using the least squares method to extract the macroscopic yield resistance index and the dynamic plastic viscosity slope. The extracted macroscopic yield resistance index, dynamic plastic viscosity slope, and aggregate collision high-frequency entropy are combined according to a preset dimension to form the rheological state feature matrix. The slump output module is used to dynamically evaluate and correct the rheological state feature matrix. When the working condition stability confidence ξ is continuously lower than the safety baseline or the KL divergence of the feature distribution exceeds the drift threshold, domain adaptation fine-tuning is performed, and the concrete slump monitoring results are output.

[0077] The specific details of each module or unit in the above-mentioned concrete slump monitoring device have been described in detail in the corresponding concrete slump monitoring method, so they will not be repeated here.

[0078] This application also provides an electronic device, including a processor and a memory, wherein the memory stores one or more computer programs, and the one or more computer programs include instructions that, when executed by the electronic device, implement the above-described method.

[0079] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication portion of a network interface card such as a LAN card or modem, and / or installed from a removable medium such as a disk, optical disk, magneto-optical disk, or semiconductor memory. When the computer program is executed by a processor (CPU), it performs the functions defined in the embodiments of this application.

[0080] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0081] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0082] The units described in the embodiments of this disclosure can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the unit itself.

[0083] In another aspect, this application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The computer-readable medium carries one or more programs, which include instructions that, when executed by the electronic device, cause the electronic device to perform the methods described in the above embodiments.

[0084] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0085] 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 within the technical scope 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.

Claims

1. A method for monitoring the slump of concrete, characterized in that, include: Acquire multi-source heterogeneous sensor data collected by the vehicle sensor network to obtain vehicle six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system; The vehicle's six-degree-of-freedom motion data is spatiotemporally synchronized and subjected to extended Kalman filtering to filter out high-frequency road vibration interference, thereby obtaining low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and angular acceleration of the mixing tank that characterize the vehicle's macroscopic attitude. The water tank filling signal and the high-speed reverse rotation signal of the mixing tank are detected to identify the no-load condition of the washing tank. The original torque data of the hydraulic drive system under this condition is extracted as the no-load measured torque. Based on the no-load measured torque, the mechanical wear compensation torque of the equipment throughout its entire life cycle is dynamically updated using the exponential smoothing evolution formula. Based on the low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle, and angular acceleration of the mixing tank rotation, the spurious off-center load torque caused by vehicle dynamics is calculated in reverse using the rigid body dynamics off-center load compensation formula. This spurious off-center load torque, together with the mechanical wear compensation torque, is differentially removed from the original torque data of the hydraulic drive system to obtain the net rheological drive torque that reflects only the Bingham rheological properties of concrete, stripped of road surface interference and mechanical dissipation. Based on the real-time phase data of the mixing tank, the net rheological driving torque is phase-locked and time-domain segmented. The net rheological driving torque is decoupled into a low-frequency fundamental component characterizing macroscopic rheological resistance and a high-frequency residual component characterizing aggregate collision through variational mode decomposition. The macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision are extracted to construct a rheological state feature matrix. The extraction of the macroscopic yield resistance index, dynamic plastic viscosity slope, and high-frequency entropy of aggregate collisions to construct a rheological state feature matrix specifically includes: For the net rheological drive torque Variational mode decomposition is performed to obtain K eigenmode functions. ; Calculate the energy percentage of each intrinsic mode. And calculate the high-frequency entropy of aggregate collisions based on the Shannon entropy formula. ; Extract the K intrinsic mode functions The modal component with the lowest center frequency is taken as the low-frequency fundamental component. The least squares method is used to linearly fit the low-frequency fundamental component to extract the macroscopic yield resistance index and dynamic plastic viscosity slope. The extracted macroscopic yield resistance index, dynamic plastic viscosity slope, and aggregate collision high-frequency entropy are combined according to a preset dimension to form the rheological state feature matrix. The rheological state feature matrix is ​​dynamically evaluated and corrected. When the confidence level of the stability of the working condition ξ is continuously lower than the safety baseline or the KL divergence of the feature distribution exceeds the drift threshold, domain adaptation fine-tuning is performed, and the concrete slump monitoring results are output.

2. The method according to claim 1, characterized in that, The method of dynamically updating the mechanical wear compensation torque throughout the entire life cycle of the equipment using an exponential smoothing evolution formula specifically includes: When the tank washing no-load condition is identified, the measured value of the pure no-load torque under the tank washing no-load condition is extracted. And perform adaptive smoothing updates of the time series according to the following formula: in, For the mechanical wear compensation torque of the current cycle, This is the mechanical wear compensation torque from the previous cycle. This is the forgetting smoothing coefficient, and its value range is... , The original torque of the washing tank under no-load conditions. The moment of inertia of the mixing tank under no-load conditions. The rated speed of the washing tank, This refers to the mechanical transmission efficiency of the hydraulic system.

3. The method according to claim 1, characterized in that, The method of inversely calculating the spurious off-center load torque caused by vehicle dynamics using the rigid body dynamics off-center load compensation formula specifically includes: calculating the net rheological drive torque according to the following rigid body dynamics off-center load compensation calculation formula. : in, For net rheological drive torque, The original driving torque, Torque compensation for mechanical wear, The equivalent eccentric load center of mass of the concrete. Let the radius of rotation be the center of mass offset. It is the acceleration due to gravity. The instantaneous pitch angle of the vehicle. The instantaneous roll angle of the vehicle. The rotational angular acceleration of the mixing tank. Let the equivalent radius of the mixing tank be . It is the lateral translational acceleration. This refers to the vertical turbulence acceleration.

4. The method according to claim 1, characterized in that, The high-frequency entropy of aggregate collision The calculation formula is: 。 5. The method according to claim 1, characterized in that, The dynamic evaluation and correction of the rheological state feature matrix specifically includes: Real-time calculation of vehicle driving variance and tank rotation speed fluctuation rate yields the confidence level of operating condition stability. ;when If the current condition is determined to be severe turbulence or a sharp turn, immediately suspend the slump prediction output for the current cycle and lock and maintain the high-confidence inference result from the previous cycle until... And consistently exceeding The suspension will be lifted after one cycle; At the same time, when aggregates collide with high-frequency entropy Higher than the preset reference line At that time, a nonlinear penalty coefficient is introduced. The macroscopic yield resistance index is attenuated and modulated in real time, and the modulated yield resistance index is... satisfy: in, The original macroscopic yield resistance index. The mechanism penalty weighting coefficient, This is a preset threshold.

6. The method according to claim 1, characterized in that, The execution domain adaptation fine-tuning specifically includes: Calculate the online distribution of the feature vector input to the current vehicle-mounted edge computing unit. Distribution of pre-trained model libraries in the cloud KL divergence between : When calculated Exceeding the safe drift threshold When a substantial change is detected in the concrete mix proportion or material source, a manual inspection prompt is sent to the vehicle terminal. After obtaining slump labels from 1 to 2 real manual inspections, the weights of the feature extraction layer of the lightweight neural network are frozen, and backpropagation fine-tuning is performed only on the last fully connected layer.

7. The method according to claim 1, characterized in that, The method further includes: Historical sensor data of different concretes under different working conditions were collected, and the corresponding macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision were extracted to generate a training set of rheological state feature matrix. Construct a lightweight neural network that includes an input layer, a feature extraction layer, and a fully connected output layer; Using the rheological state feature matrix training set as input and the corresponding collapse degree ground truth as supervision label, the lightweight neural network is trained until convergence to obtain a pre-trained model, and the pre-trained model is deployed as the initial model.

8. A concrete slump monitoring device, characterized in that, include: The sensor data acquisition module is used to acquire multi-source heterogeneous sensor data collected by the vehicle sensor network to obtain vehicle six-degree-of-freedom motion data, real-time phase data of the mixing tank, and raw torque data of the hydraulic drive system. The data processing module is used to perform spatiotemporal synchronization and extended Kalman filtering on the vehicle's six-degree-of-freedom motion data to filter out high-frequency road vibration interference and obtain low-frequency translational acceleration, lateral acceleration, pitch angle, roll angle and mixing tank rotational angular acceleration that characterize the vehicle's macroscopic attitude. The no-load identification and compensation module is used to detect the water tank filling signal and the high-speed reverse rotation signal of the mixing tank to identify the no-load condition of the washing tank, and dynamically update the mechanical wear compensation torque throughout the entire life cycle of the equipment using the exponential smooth evolution formula. The feature extraction module is used to calculate the false load torque caused by vehicle dynamic driving in reverse using the rigid body dynamics eccentric load compensation calculation formula based on the low-frequency translational acceleration, lateral acceleration, pitch angle, side tilt angle and the rotational angular acceleration of the mixing tank. The false load torque is then differentially removed from the original torque data along with the mechanical wear compensation torque to obtain the net rheological driving torque that reflects only the Bingham rheological properties of concrete, stripped of road surface interference and mechanical dissipation. The rheological characteristic determination module is used to perform phase locking and time-domain segmentation of the net rheological driving torque based on the real-time phase data of the mixing tank. The net rheological driving torque is decoupled into a low-frequency fundamental component characterizing macroscopic rheological resistance and a high-frequency residual component characterizing aggregate collision through variational mode decomposition. The macroscopic yield resistance index, dynamic plastic viscosity slope and high-frequency entropy of aggregate collision are extracted to construct a rheological state characteristic matrix. The extraction of the macroscopic yield resistance index, dynamic plastic viscosity slope, and high-frequency entropy of aggregate collisions to construct a rheological state feature matrix specifically includes: For the net rheological drive torque Variational mode decomposition is performed to obtain K eigenmode functions. ; Calculate the energy percentage of each intrinsic mode. And calculate the high-frequency entropy of aggregate collisions based on the Shannon entropy formula. ; Extract the K intrinsic mode functions The modal component with the lowest center frequency is taken as the low-frequency fundamental component. The least squares method is used to linearly fit the low-frequency fundamental component to extract the macroscopic yield resistance index and dynamic plastic viscosity slope. The extracted macroscopic yield resistance index, dynamic plastic viscosity slope, and aggregate collision high-frequency entropy are combined according to a preset dimension to form the rheological state feature matrix. The slump output module is used to dynamically evaluate and correct the rheological state feature matrix. When the confidence level of the working condition stability ξ is continuously lower than the safety baseline or the KL divergence of the feature distribution exceeds the drift threshold, the domain adaptation fine-tuning is performed and the concrete slump monitoring results are output.

9. An electronic device, characterized in that, The device includes a processor and a memory, the memory storing one or more computer programs, the one or more computer programs including instructions that, when executed by the electronic device, cause the electronic device to perform the concrete slump monitoring method according to any one of claims 1-7.

Images