Aggregate loading size identification and detection control system based on radar identification

Abstract

The invention discloses an aggregate loading size identification and detection control system based on radar identification, and particularly relates to the field of constructional engineering, which comprises a multi-source sensing module, a feature extraction module, an index calculation module, a threshold judgment module, a logic decision module and a control execution module. Characteristic parameters such as phase difference and coordinate deviation are extracted through multi-modal processing, four kinds of comprehensive indexes such as volume uniformity and overload risk are generated, risk grading early warning is achieved through a three-level threshold system, a control instruction is generated in combination with a state machine model and a logic gate strategy, and the system is provided with an electro-hydraulic proportional valve and a redundant braking mechanism. The system supports multi-stage response from rate adjustment to emergency braking, has the characteristics of multi-dimensional dynamic monitoring, intelligent decision making and environment self-adaption, effectively improves the safety and automation level of bulk cargo loading operation, and is suitable for continuous operation requirements of complex industrial scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of construction engineering, and more particularly to an aggregate loading size recognition detection control system based on radar recognition. Background Art

[0002] Aggregate loading process monitoring is a core link in the field of construction engineering, which directly affects transportation safety and operation efficiency. Traditional methods mainly rely on manual visual inspection or single sensors, such as floor weighing and two-dimensional visual detection, which are difficult to meet the real-time monitoring needs under complex working conditions. With the development of intelligent construction technology, the industry's requirements for automated and refined control of the loading process are increasing.

[0003] Traditional aggregate loading detection systems usually use a single sensor for data collection, obtain total mass or plane image information through a fixed period, and then calculate the loading capacity based on a linear weighted algorithm. When the value exceeds the preset threshold, the sound and light alarm device is triggered and the static data of key nodes is recorded. The entire process relies on manually set empirical parameters, and the control instructions only include start and stop signals, lacking real-time control of dynamic parameters such as loading rate and material distribution.

[0004] Existing solutions are limited to single-modal data collection and linear judgment logic, making it difficult to accurately identify three-dimensional spatial features such as aggregate particle size mixing and car deformation, resulting in frequent misjudgments of overloads and missed safety hazards. At the same time, the fixed threshold mechanism cannot adapt to nonlinear working conditions during loading, and can easily cause equipment damage due to instantaneous impact during emergency braking. In addition, sensors are easily interfered with in complex environments such as dust and temperature changes, and data drift is a prominent problem, requiring frequent manual calibration and maintenance, which seriously restricts system reliability. Summary of the invention

[0005] In order to overcome the above defects of the prior art, an embodiment of the present invention provides an aggregate loading size identification detection control system based on radar identification, and solves the problems raised in the above background technology through the following scheme.

[0006] To achieve the above purpose, the present invention provides the following technical solution: an aggregate loading size recognition detection control system based on radar recognition, comprising: Multi-source perception module: It realizes dynamic monitoring of the loading process through heterogeneous sensor fusion technology, including millimeter wave radar, inertial measurement unit, optical vision system and laser ranging device, and transmits multi-modal raw data stream to the feature extraction module; Feature extraction module: used to extract features from the multimodal data stream transmitted by the multi-source perception module, including dynamic three-dimensional volume distribution detection data, aggregate particle size mixing detection data, dynamic overload risk detection data and car body deformation detection data, and transmit the extracted feature data to the index calculation module; Index calculation module: It combines multi-dimensional characteristic parameters through nonlinear formulas to calculate four core indicators in real time, namely, volume uniformity index, particle size mixing index, dynamic overload coefficient and structural deformation index. It uses experimentally calibrated parameter system to realize the mapping of physical quantity to risk value, and transmits the calculated indicators to the threshold judgment module. Threshold judgment module: It uses a three-level threshold system to independently evaluate the status of four core indicators, outputs the real-time safety status of each indicator through binary coding, and transmits the judgment results to the logic decision module; Logical decision module: adopts a multi-level state machine architecture, combines the threshold states of various detection indicators through Boolean logic gates, generates hierarchical control instructions, and transmits the control instructions to the control execution module; Control execution module: converts logical decisions into physical control actions and realizes real-time regulation of the loading process through multi-level actuators.

[0007] Technical effects and advantages of the present invention: 1. Multi-dimensional stereoscopic monitoring and precise identification: Through heterogeneous sensor fusion technology, combined with millimeter-wave radar, optical vision and laser ranging devices, multi-dimensional dynamic monitoring of the aggregate loading process is achieved, effectively capturing multi-dimensional features such as material distribution, particle size mixing and compartment deformation, overcoming the limitations of traditional single-point detection, and significantly improving detection accuracy and reliability under complex working conditions; 2. Intelligent risk warning and hierarchical control: Based on nonlinear feature fusion and multi-level threshold judgment mechanism, the system can analyze core indicators such as volume uniformity, overload risk and structural deformation in real time, generate hierarchical control instructions through the logic decision module, and realize intelligent response from early warning to emergency braking, avoiding the lag of manual intervention and ensuring the safety of loading operations; 3. Adaptive control and high robustness design: The control execution module adopts an electro-hydraulic proportional valve and a dual redundant braking mechanism, combined with dynamic rate adjustment and fail-safe mode to ensure that the system can still operate stably under extreme working conditions or when communication is interrupted. At the same time, the self-cleaning air curtain and temperature compensation technology are used to enhance the long-term stability of the sensor in harsh environments such as dust and temperature differences, thereby reducing maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a schematic diagram of the overall structure of the present invention.

[0009] Figure 2 It is a schematic diagram of the structure of the control execution module of the present invention. DETAILED DESCRIPTION

[0010] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0011] refer to Figure 1 The aggregate loading size recognition detection and control system shown in the figure based on radar recognition includes a multi-source perception module, a feature extraction module, an index calculation module, a threshold judgment module, a logic decision module and a control execution module.

[0012] Multi-source perception module: It realizes dynamic monitoring of the loading process through heterogeneous sensor fusion technology, including millimeter wave radar, inertial measurement unit, optical vision system and laser ranging device, and transmits multi-modal raw data stream to the feature extraction module.

[0013] The millimeter-wave radar is installed on the rotating platform at the top of the loading arm. Through three groups of 77GHz radar modules and two groups of 24GHz radar modules with adjustable pitch angles, the three-dimensional point cloud data within the distance range of 0-5m on the surface of the material pile is obtained at a scanning frequency of 30Hz. The inertial measurement unit is installed on the drive motor base, specifically in an XYZ three-axis orthogonal layout, to monitor the vibration acceleration signal in the 0-2kHz frequency band, especially to capture high-frequency impact components above 500Hz. The optical vision system is installed in a protective cover 1.5m above the feed port, using a binocular camera installed at a 45° inclination angle, in conjunction with an 850nm infrared fill light, to capture the motion trajectory of particles with a particle size of 5-25mm in a dusty environment. The laser ranging device is installed on the equidistant mounting brackets of the guardrails on both sides of the carriage, with an interval of 1.2m. Eight groups of laser sensors constitute a deformation monitoring network, which detects the deformation of the side panels of the carriage by real-time comparison of the reference distance, and the initial calibration value is ±1mm.

[0014] All sensors in the multi-mode perception module are installed using a quick-release mechanical interface, the radar and vision systems are equipped with a self-cleaning air curtain device, the accelerometer is dual-fixed using magnetic attraction and threads, and the laser sensor is equipped with a temperature compensation module to ensure measurement stability under working conditions of -20°C ~ 65°C.

[0015] Feature extraction module: used to extract features from the multimodal data stream transmitted by the multi-source perception module, including dynamic three-dimensional volume distribution detection data, aggregate particle size mixing detection data, dynamic overload risk detection data and car body deformation detection data, and transmit the extracted feature data to the indicator calculation module.

[0016] The dynamic three-dimensional volume distribution detection data includes X / Y / Z axis stacking standard deviation, surface curvature mutation point density, reflection intensity gradient entropy and dynamic filling rate ratio, which are marked as σ x , σ y , σ z , C avg 、E g and V ratio The aggregate particle size mixed detection data include dual-frequency reflection intensity difference, attenuation time dispersion, Doppler broadening coefficient and polarization scattering entropy, marked as ΔR, D τ , B d and H p The dynamic overload risk detection data include the frequency domain integral of the impact force, the trajectory deflection acceleration, the mass flow mutation rate and the space momentum vector modulus, which are marked as F int 、a θ , ΔM and |P|, the cabin deformation detection data includes multi-baseline interference phase difference, reflection intensity variation coefficient, dynamic deformation propagation speed and asymmetry index, marked as ΔΦ, CV, respectively. R 、V def and ASI.

[0017] The X / Y / Z axis stacking standard deviation is collected by calculating the standard deviation of the three axial coordinates of the point cloud data in each 77GHz millimeter wave radar scanning cycle. The surface curvature mutation point density is calculated by Delaunay triangulation to construct a surface mesh to calculate the Gaussian curvature of each triangular facet, and the proportion of faces with curvature>K_threshold is counted, where K_threshold=0.5. The reflection intensity gradient entropy is collected by dividing the scanning area into 10×10 grids and calculating the Shannon entropy of the reflection intensity value of each grid, reflecting the degree of chaos of the reflection intensity distribution of the loading surface. The larger the value, the more uneven the material or density. The calculation formula is: , p i It represents the reflection intensity ratio of the i-th grid. The dynamic filling rate ratio is collected by continuous frame point cloud computing instantaneous volume change, which represents the deviation degree of instantaneous volume change rate from the historical average value. The calculation formula is: V r a t i o = V t − V t − Δ t Δ t × V ¯ [ t − 1 0 , t ] , V t Represents the volume at the current moment, V t-Δt represents the volume of the previous time series, V ¯ [ t − 1 0 , t ] It represents the volume average of the latest 10 frames, Δt=0.05s.

[0018] The dynamic three-dimensional volume distribution detection data is jointly analyzed through multi-dimensional spatial characteristics, breaking through the limitation of traditional volume detection that only focuses on the change in total amount. The X / Y / Z axis stacking standard deviation reveals the spatial discrete characteristics of material distribution, the surface curvature mutation point density captures local stacking anomalies, the reflection intensity gradient entropy quantifies the electromagnetic wave scattering chaos of the loading surface, and the dynamic filling rate ratio monitors the time domain instability of the loading process. For the first time, the three-dimensional dynamic characterization of the loading shape is realized, solving the problem that traditional floor scales and single-point laser detection cannot identify three-dimensional defects such as overload and cavity.

[0019] The dual-frequency reflection intensity difference is collected by synchronously transmitting 24GHz and 77GHz linear frequency modulation waves to calculate the difference in echo power between the two frequency bands, and is used to detect the sensitivity difference of aggregates with different particle sizes. The calculation formula is: , P 77 Indicates the echo power in the 77GHz band, P 24 It represents the echo power in the 24GHz frequency band. The attenuation time dispersion is collected by extracting the time series of the echo envelope attenuating to -3dB to calculate the coefficient of variation. It is used to reflect the fluctuation characteristics of the echo signal attenuation time. The calculation formula is: , τ j It represents the -3dB decay time of the jth measurement, std represents the standard deviation operation, mean represents the mean operation, and the Doppler broadening coefficient is collected by FFT spectrum analysis of moving particles, which represents the velocity distribution range of moving particles. The larger the value, the greater the velocity difference. The calculation formula is: , f high Indicates the -3dB upper limit frequency of the spectrum, f low Indicates the spectrum -3dB lower limit frequency, f c Represents the radar carrier frequency. The polarization scattering entropy is used to calculate the eigenvalue entropy through the polarization scattering matrix, which is used to quantify the complexity of the scattering mechanism. A high value indicates the presence of multiple scattering types. The calculation formula is: ,λ u represents the normalized eigenvalue of the coherence matrix.

[0020] The aggregate particle size mixing detection data utilizes the difference in scattered frequency response of different particle sizes to electromagnetic waves through dual-frequency reflection difference, the attenuation time discreteness reflects the particle collision damping characteristics, the Doppler broadening coefficient captures the motion particle size distribution, and the polarization scattering entropy reveals the orientation disorder of non-spherical particles. A joint feature space of particle size, motion and morphology is established, breaking through the limitation of traditional screening methods that can only statically detect samples at specific positions, and realizing dynamic online monitoring of the particle size distribution of the entire section.

[0021] The impact force frequency domain integration is performed by performing FFT on the vibration signal collected by the accelerometer and calculating the energy collection in the 50-200 Hz frequency band, which is used to quantify the high-frequency impact energy and reflect the risk of hard object impact. The calculation formula is: , a(t) represents the time domain acceleration signal, FFT represents fast Fourier transform, f represents frequency, and the trajectory deflection acceleration is collected by calculating the second-order derivative of the tangential acceleration of the typical particle trajectory, which represents the degree of mutation in the particle motion direction. The calculation formula is: , ρ represents the material density, v represents the instantaneous velocity of the particles, t represents the time, and the mass flow mutation rate is collected by differential processing of the weighbridge data to reflect the severity of the change in loading mass. The calculation formula is: , M represents the mass measurement value, t0 represents the current moment, and the spatial momentum vector modulus is used to comprehensively represent the total impact momentum of the moving material by integrating visual and radar data collection. The calculation formula is: , represents the instantaneous mass flow rate, v terminal represents the particle terminal velocity.

[0022] The dynamic overload risk detection data captures the infrasonic energy caused by structural resonance through the impact force frequency domain integration, quantifies the kinetic energy change rate of material impact through trajectory deflection acceleration, monitors the second-order discontinuity of loading rate through mass flow mutation rate, and constructs a comprehensive indicator of energy transfer through spatial momentum vector model. It breaks through the traditional static weighing mode and establishes a dynamic overload warning system for the first time, solving the problem that the existing technology cannot detect dynamic risks such as impact load and momentum overload.

[0023] The multi-baseline interference phase difference is collected by calculating the three-baseline radar array measurement phase standard deviation, which is used to reflect the spatial consistency of the deformation of the car surface. The calculation formula is: , represents the phase measurement value of the mth antenna, It represents the average phase value of three channels. The reflection intensity variation coefficient is collected by dividing the deformation area and the normal area to calculate the contrast value. It is used to quantify the abnormality of the reflection characteristics of the deformation area. The calculation formula is: , σ represents the standard deviation of the regional reflection intensity, μ represents the mean value of the regional reflection intensity, and the dynamic deformation propagation velocity is collected by performing time domain cross-correlation analysis on the deformation time series data to characterize the propagation dynamics characteristics of the deformation wave. The calculation formula is: , Δx represents the distance between monitoring points, R 12 (τ) represents the cross-correlation function of the signals at the two monitoring points, τ represents the time delay parameter, and the asymmetry index is collected by calculating the standardized difference in the distance between the left and right side panels, which is used to quantify the asymmetry of the left and right deformation of the car body. The calculation formula is: , D L Indicates the laser distance measurement value on the left, D R Indicates the laser distance measurement value on the right, and 0.01 is the zero division prevention constant.

[0024] The carriage deformation detection data improves the detection accuracy of micron-level deformation through multi-baseline interference phase difference, identifies the surface roughness change caused by material yield through the reflection intensity variation coefficient, quantifies the mechanical energy transfer efficiency of the deformation wave through the dynamic propagation velocity, and diagnoses the structural imbalance caused by the eccentric load through the asymmetric index. It breaks through the limitation of single-point measurement of traditional strain gauges, realizes the full-chain analysis of deformation generation-propagation-consequences, and solves the problem that the existing technology cannot give early warning of plastic deformation.

[0025] Index calculation module: Through nonlinear formulas, multi-dimensional characteristic parameters are combined to calculate four core indicators, namely volume uniformity index, particle size mixing index, dynamic overload coefficient and structural deformation index, in real time. The parameter system calibrated by experiments is used to realize the mapping of physical quantities to risk values, and the calculated indicators are transmitted to the threshold judgment module.

[0026] The index calculation module calculates the volume uniformity index through dynamic three-dimensional volume distribution detection data, which is specifically expressed as: , VUI represents the volume uniformity index, α represents the logarithmic scaling factor, β represents the anti-zero division correction, γ represents the curvature nonlinear gain, δ represents the entropy suppression coefficient, and η represents the rate sensitivity factor.

[0027] The volume uniformity index adopts a nonlinear combination of logarithmic compression and hyperbolic tangent function. The first term reflects the overall uniformity through the axial dispersion ratio. The second term uses the power law relationship between curvature and reflection entropy to enhance surface anomaly recognition. The last term captures transient mutations through nonlinear mapping of dynamic rate differences. The formula structure realizes the coordinated amplification of abnormal signals by multiple physical quantities, avoiding the sensitivity loss caused by simple weighted average.

[0028] The index calculation module calculates the particle size mixing index through the aggregate particle size mixing detection data, which is specifically expressed as: , PSI represents the particle size mixing index, κ represents the frequency difference enhancement coefficient, μ represents the broadening normalization factor, ν represents the anti-zero division correction, ξ represents the entropy value amplification coefficient, ρ1 represents the Sigmoid steepness coefficient, and θ represents the frequency difference activation threshold.

[0029] The particle size hybrid index constructs a dual-path fusion architecture. The former term amplifies the electromagnetic characteristics of abnormal particle size by the product of frequency difference and attenuation discreteness, and the latter term uses an entropy-driven Sigmoid function to achieve nonlinear grading of the degree of hybridity. The introduction of the exponential term enables the formula to have a characteristic self-enhancement characteristic, which can automatically adapt to the detection sensitivity requirements of aggregates of different materials.

[0030] The indicator calculation module calculates the dynamic overload coefficient through dynamic overload risk detection data, which is specifically expressed as: , DOC represents the dynamic overload coefficient, ω represents the momentum attenuation coefficient, λ represents the dimensional balance factor, ε represents the acceleration nonlinear correction, ζ represents the logarithmic scaling factor, ξ1 represents the mutation rate reference value, and ι represents the exponential sensitivity.

[0031] The dynamic overload coefficient adopts a sub-item coupling design. The first term characterizes the steady-state overload risk through a power-law combination of frequency-domain energy and momentum vectors, and the second term realizes transient impact detection by utilizing logarithmic transformation of acceleration and exponential amplification of mutation rate. The denominator balance factor in the formula eliminates dimensional differences, and the exponential parameter setting enables each physical quantity to produce a synergistic amplification effect when exceeding the limit.

[0032] The index calculation module calculates the structural deformation index through the car body deformation detection data, which is specifically expressed as: , SDI represents the structural deformation index, τ1 represents the phase sensitivity coefficient, υ1 represents the variation suppression factor, ψ represents the denominator equilibrium constant, φ represents the asymmetric amplification index, χ represents the velocity activation steepness, and ω1 represents the propagation velocity threshold.

[0033] The structural deformation index adopts a dual-channel fusion architecture. The former enhances early deformation recognition through a nonlinear combination of phase difference and reflection variation, and the latter uses the exponential relationship between propagation speed and asymmetric index to evaluate the risk of deformation expansion. The Sigmoid function in the denominator realizes soft switching of the speed threshold, so that the formula maintains high sensitivity during low-speed deformation and triggers a strong alarm during high-speed expansion.

[0034] The method for determining the constant parameters such as α, β, γ in the index calculation module includes experimental calibration: based on benchmark test data under standard working conditions, theoretical derivation: based on the mechanical properties of materials and the physical model of sensors, equipment parameters: matching the hardware characteristics of the radar array and the inertial unit used, and engineering experience: referring to the statistical laws of historical loading abnormality cases, which are not specifically introduced in this embodiment.

[0035] Threshold judgment module: A three-level threshold system is used to independently evaluate the status of four core indicators, the real-time safety status of each indicator is output through binary coding, and the judgment results are transmitted to the logic decision module.

[0036] The three-level threshold system presets three-level thresholds of safety, warning and alarm for each indicator. The judgment process adopts the hysteresis comparison method to prevent state oscillation: when the indicator value approaches the threshold from below, it needs to exceed the limit for 3 consecutive 60ms cycles to trigger the state switching. When it falls back from the top, it will be downgraded immediately. Each indicator state is represented by a 2-bit binary code, 00 represents safety, 01 represents warning, and 11 represents alarm. The four indicators generate a total of 8-bit status words. The module integrates a sliding standard deviation check function. When the indicator volatility exceeds the set value, the state output is automatically frozen to avoid misjudgment caused by transient interference.

[0037] The three-level threshold system is expressed as [safety, warning, alarm], the threshold intervals of VUI are divided into [<1.2, 1.2-1.8, ≥1.8], the threshold intervals of PSI are divided into [<0.7, 0.7-1.3, ≥1.3], the threshold intervals of DOC are divided into [<15, 15-28, ≥28], and the threshold intervals of SDI are divided into [<1.5, 1.5-2.8, ≥2.8].

[0038] Logical decision module: It adopts a multi-level state machine architecture, combines the threshold states of various detection indicators through Boolean logic gates, generates hierarchical control instructions, and transmits the control instructions to the control execution module.

[0039] The logic decision module receives four groups of 2-bit state codes from the threshold judgment module and generates hierarchical control instructions through a predefined decision matrix.

[0040] The triggering conditions of the logic decision module include: when (VUI≥1.8∧PSI≥1.3)∨SDI≥2.8, the execution action is to immediately stop loading and alarm; when DOC≥28∧(VUI≥1.8∨PSI≥1.3), the execution action is to suspend loading and start manual review; when any two indicators enter the warning interval, the execution action is to reduce the loading rate and sound and light prompts; when a single indicator is warned, the execution action is data recording and status indicator light.

[0041] The status indicator light is green when the safety state is safe, yellow when the safety state is warning, and red when the safety state is alarm.

[0042] Control execution module: converts logical decisions into physical control actions and realizes real-time regulation of the loading process through multi-level actuators.

[0043] refer to Figure 2 The execution channel of the control execution module includes a flow control channel, an emergency braking channel and a human-computer interaction channel. The flow control channel receives a 0-10V analog signal through an electro-hydraulic proportional valve with a rated pressure of 35MPa, and realizes stepless adjustment of the loading rate within the range of 0-100%, with a response time of ≤20ms. The emergency braking channel uses dual redundant relays to control the electromagnetic brake, with a holding torque of ≥850N·m. The power is cut off and the mechanical lock is started within 10ms after receiving the stop command. The human-computer interaction channel includes a 360° visible 105dB rotating alarm light and a 7-inch industrial touch screen, which synchronously displays the real-time curves of four indicators and the threshold status.

[0044] The control execution module adopts a specially designed dual-mode control mechanism, which automatically switches to a safe mode when communication is interrupted, and gradually reduces the flow at a frequency of 5Hz until it stops completely.

[0045] The present invention realizes multi-dimensional dynamic monitoring of the aggregate loading process through heterogeneous sensor fusion technology, combined with millimeter-wave radar, optical vision and laser ranging devices, effectively captures multi-dimensional characteristics such as material distribution, particle size mixing and compartment deformation, overcomes the limitations of traditional single-point detection, and significantly improves the detection accuracy and reliability under complex working conditions; based on nonlinear feature fusion and multi-level threshold judgment mechanism, the system can analyze core indicators such as volume uniformity, overload risk and structural deformation in real time, and generate hierarchical control instructions through the logic decision module to achieve intelligent response from early warning prompts to emergency braking, avoid the lag of manual intervention, and ensure the safety of loading operations; the control execution module adopts an electro-hydraulic proportional valve and a dual redundant braking mechanism, combined with dynamic rate adjustment and fail-safe mode, to ensure that the system can still operate stably under extreme working conditions or when communication is interrupted. At the same time, through the self-cleaning air curtain and temperature compensation technology, the long-term stability of the sensor in harsh environments such as dust and temperature difference is enhanced, and the maintenance cost is reduced.

[0046] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0047] Finally: 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 in the protection scope of the present invention.

Claims

1. A radar-based aggregate loading size recognition detection and control system, characterized in that: include: Multi-source perception module: It realizes dynamic monitoring of the loading process through heterogeneous sensor fusion technology, including millimeter wave radar, inertial measurement unit, optical vision system and laser ranging device, and transmits multi-modal raw data stream to the feature extraction module; Feature extraction module: used to extract features from the multimodal data stream transmitted by the multi-source perception module, including dynamic three-dimensional volume distribution detection data, aggregate particle size mixing detection data, dynamic overload risk detection data and car body deformation detection data, and transmit the extracted feature data to the index calculation module; Index calculation module: It combines multi-dimensional characteristic parameters through nonlinear formulas to calculate four core indicators in real time, namely, volume uniformity index, particle size mixing index, dynamic overload coefficient and structural deformation index. It uses experimentally calibrated parameter system to realize the mapping of physical quantity to risk value, and transmits the calculated indicators to the threshold judgment module. Threshold judgment module: It uses a three-level threshold system to independently evaluate the status of four core indicators, outputs the real-time safety status of each indicator through binary coding, and transmits the judgment results to the logic decision module; Logical decision module: adopts a multi-level state machine architecture, combines the threshold states of various detection indicators through Boolean logic gates, generates hierarchical control instructions, and transmits the control instructions to the control execution module; Control execution module: converts logical decisions into physical control actions and realizes real-time regulation of the loading process through multi-level actuators.

2. According to the radar-based aggregate loading size recognition detection and control system of claim 1, it is characterized by: The millimeter-wave radar is installed on the rotating platform at the top of the loading arm. Through three groups of 77GHz radar modules and two groups of 24GHz radar modules with adjustable pitch angles, the three-dimensional point cloud data within the distance range of 0-5m on the surface of the material pile is obtained at a scanning frequency of 30Hz. The inertial measurement unit is installed on the drive motor base, specifically in an XYZ three-axis orthogonal layout, to monitor the vibration acceleration signal in the 0-2kHz frequency band, especially to capture high-frequency impact components above 500Hz. The optical vision system is installed in a protective cover 1.5m above the feed port, using a binocular camera installed at a 45° inclination angle, in conjunction with an 850nm infrared fill light, to capture the motion trajectory of particles with a particle size of 5-25mm in a dusty environment. The laser ranging device is installed on the equidistant mounting brackets of the guardrails on both sides of the carriage, with an interval of 1.2m. Eight groups of laser sensors constitute a deformation monitoring network, which detects the deformation of the side panels of the carriage by real-time comparison of the reference distance, and the initial calibration value is ±1mm.

3. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The dynamic three-dimensional volume distribution detection data includes X / Y / Z axis stacking standard deviation, surface curvature mutation point density, reflection intensity gradient entropy and dynamic filling rate ratio, which are marked as σ x , σ y , σ z , C avg 、E g and V ratio The aggregate particle size mixed detection data include dual-frequency reflection intensity difference, attenuation time dispersion, Doppler broadening coefficient and polarization scattering entropy, marked as ΔR, D τ , B d and H p The dynamic overload risk detection data include the frequency domain integral of the impact force, the trajectory deflection acceleration, the mass flow mutation rate and the space momentum vector modulus, which are marked as F int 、a θ , ΔM and |P|, the cabin deformation detection data includes multi-baseline interference phase difference, reflection intensity variation coefficient, dynamic deformation propagation speed and asymmetry index, marked as ΔΦ, CV, respectively. R 、V def and ASI.

4. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The index calculation module calculates the volume uniformity index through dynamic three-dimensional volume distribution detection data, which is specifically expressed as: , VUI represents the volume uniformity index, α represents the logarithmic scaling factor, β represents the anti-zero division correction, γ represents the curvature nonlinear gain, δ represents the entropy suppression coefficient, and η represents the rate sensitivity factor.

5. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The index calculation module calculates the particle size mixing index through the aggregate particle size mixing detection data, which is specifically expressed as: , PSI represents the particle size mixing index, κ represents the frequency difference enhancement coefficient, μ represents the broadening normalization factor, ν represents the anti-zero division correction, ξ represents the entropy value amplification coefficient, ρ1 represents the Sigmoid steepness coefficient, and θ represents the frequency difference activation threshold.

6. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The indicator calculation module calculates the dynamic overload coefficient through dynamic overload risk detection data, which is specifically expressed as: , DOC represents the dynamic overload coefficient, ω represents the momentum decay coefficient, λ represents the dimensional balance factor, ε represents the acceleration nonlinear correction, ζ represents the logarithmic scaling factor, ξ1 represents the mutation rate reference value, and ι represents the exponential sensitivity.

7. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The index calculation module calculates the structural deformation index through the car body deformation detection data, which is specifically expressed as: , SDI represents the structural deformation index, τ1 represents the phase sensitivity coefficient, υ1 represents the variation suppression factor, ψ represents the denominator equilibrium constant, φ represents the asymmetric amplification index, χ represents the velocity activation steepness, and ω1 represents the propagation velocity threshold.

8. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The three-level threshold system presets three-level thresholds of safety, warning and alarm for each indicator. The judgment process adopts the hysteresis comparison method to prevent state oscillation: when the indicator value approaches the threshold from the bottom, it needs to exceed the limit for 3 consecutive 60ms cycles to trigger the state switching. When it falls back from the top, it will be downgraded immediately. Each indicator state is represented by a 2-bit binary code, 00 represents safety, 01 represents warning, and 11 represents alarm. The four indicators generate a total of 8-bit status words. The module integrates a sliding standard deviation check function. When the indicator volatility exceeds the set value, the state output is automatically frozen to avoid misjudgment caused by transient interference; The three-level threshold system is expressed as [safety, warning, alarm], the threshold intervals of VUI are divided into [<1.2, 1.2-1.8, ≥1.8], the threshold intervals of PSI are divided into [<0.7, 0.7-1.3, ≥1.3], the threshold intervals of DOC are divided into [<15, 15-28, ≥28], and the threshold intervals of SDI are divided into [<1.5, 1.5-2.8, ≥2.8].

9. The aggregate loading size identification detection and control system based on radar identification according to claim 1 is characterized by: The triggering conditions of the logic decision module include: when (VUI≥1.8∧PSI≥1.3)∨SDI≥2.8, the execution action is to immediately stop loading and alarm; when DOC≥28∧(VUI≥1.8∨PSI≥1.3), the execution action is to suspend loading and start manual review; when any two indicators enter the warning interval, the execution action is to reduce the loading rate and sound and light prompts; when a single indicator is warned, the execution action is data recording and status indicator light.

10. The aggregate loading size identification detection and control system based on radar identification according to claim 1, characterized in that: The execution channel of the control execution module includes a flow control channel, an emergency braking channel and a human-machine interaction channel. The flow control channel receives a 0-10V analog signal through an electro-hydraulic proportional valve with a rated pressure of 35MPa, realizes stepless adjustment of the loading rate within the range of 0-100%, and the response time is ≤20ms. The emergency braking channel uses dual redundant relays to control the electromagnetic brake, with a holding torque of ≥850N·m. After receiving the stop command, the power is cut off and the mechanical lock is started within 10ms. The human-machine interaction channel includes a 360° visible 105dB rotating alarm light and a 7-inch industrial touch screen, which synchronously displays the real-time curves of four indicators and the threshold status.

Images