A subgrade filling quality parameter optimization analysis system

Abstract

The application relates to the technical field of civil engineering, and discloses a roadbed filling quality parameter optimization analysis system, which comprises a full-cycle data acquisition module, a working condition self-identification and model self-calibration module, a soil body state real-time estimation module, a forward prediction module and a multi-target constraint control decision module. The full-cycle data acquisition module is used for collecting vibration time sequence signals and equipment operation space-time data of a road roller in a working process in real time. The working condition self-identification and model self-calibration module is used for controlling the road roller to send a diagnostic excitation signal before compaction work or under a specific working condition. The soil body state real-time estimation module is based on an adaptively configured physical inversion model in conventional compaction work. The forward prediction module is based on an adaptively configured state transition function. The multi-target constraint control decision module is based on a prediction result. The technical scheme that the diagnostic excitation signal is used to actively detect filling materials and automatically identify working condition characteristics, and then adaptively configure a physical model, achieves the technical effect of tailoring a model for a specific soil quality body, and solves the problems of poor model universality and low calculation accuracy when soil quality changes.

Description

Technical Field

[0001] This invention relates to the field of civil engineering technology, specifically to a roadbed filling quality parameter optimization and analysis system. Background Technology

[0002] Earthwork engineering, including roadbed, embankment, and foundation backfilling, is a crucial component of infrastructure construction. The quality of compaction directly affects the load-bearing capacity, stability, and service life of the entire project. Therefore, effective monitoring and real-time optimization of filling quality during the compaction process are of paramount importance for ensuring project safety and improving construction efficiency.

[0003] In the field of roadbed compaction quality control, continuous compaction control (CCC) technology based on the vibration response of road rollers is widely used. However, in long-term engineering practice, the existing technical solutions have gradually revealed their inherent limitations. Existing technologies generally use a fixed physical model or empirical formula to explain the relationship between vibration signals and soil conditions. However, the filling materials at construction sites are diverse, ranging from highly plastic clay to graded crushed stone, with significant differences in their mechanical properties. Rigidly applying a fixed model to all working conditions, ignoring the inherent physical differences between different materials, inevitably leads to significant deviations between the calculated results and the actual conditions when soil conditions change, greatly reducing the accuracy and reliability of the model.

[0004] Existing technologies for evaluating compaction quality largely rely on indirect, empirical indicators such as compaction value (CMV), vibration modulus (Evib), or harmonic response ratio. While these indicators are somewhat correlated with the degree of compaction, they lack a clear and stable physical correspondence with the soil's core physical and mechanical parameters (such as soil stiffness and damping). This decoupling means that these indicators merely reflect the apparent compaction effect, rather than directly measuring the soil's essential state, thus failing to accurately and truthfully reveal the deep evolution of the soil's mechanical state during compaction. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a roadbed filling quality parameter optimization and analysis system, which solves the problems of existing roadbed compaction control technologies relying on empirical indirect indicators and lacking adaptive and forward-looking optimization decision-making capabilities for different working conditions, thus leading to blind, inefficient, and unreliable construction processes and quality assurance.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a roadbed filling quality parameter optimization and analysis system, comprising:

[0007] The full-cycle data acquisition module is used to collect vibration time-series signals and equipment runtime spatial data of the road roller in real time during operation.

[0008] The working condition self-identification and model self-calibration module is connected to the full-cycle data acquisition module. It is used to control the road roller to send out diagnostic excitation signals before compaction operations or under specific working conditions, and automatically identify the working condition characteristics of the current filling material based on the acquired response signals, and then adaptively configure the physical inversion model and state transition function.

[0009] The real-time soil state estimation module, which is connected to the working condition self-identification and model self-calibration module, calculates the collected vibration time series signal into a real-time physical state vector characterizing the physical and mechanical properties of the soil based on the adaptively configured physical inversion model during conventional compaction operations.

[0010] The forward prediction module, which is connected to the real-time soil state estimation module, predicts the physical state vector after the next compaction caused by different combinations of construction parameters based on the adaptively configured state transition function and the real-time physical state vector, and obtains the prediction result.

[0011] The multi-objective constraint control decision module, which is connected to the forward prediction module, solves for and outputs the optimal control vector that enables the current physical state vector to reach the target state based on the prediction results.

[0012] Preferably, the full-cycle data acquisition module is used to synchronously process the vibration time series signal and the equipment runtime space-time data to generate a unified data frame that contains vibration data with time and space stamps and equipment status parameters at each moment.

[0013] Preferably, the step of the full-cycle data acquisition module performing synchronous processing to generate a unified data frame includes: marking the data streams of the acquired vibration time series signal and the equipment runtime space-time data with high-precision timestamps respectively;

[0014] Align data streams from different sources based on high-precision timestamps;

[0015] The aligned data is encapsulated into a predefined structure to form a unified data frame.

[0016] Furthermore, the step of the full-cycle data acquisition module to perform synchronous processing to generate a unified data frame includes: marking the data streams of the acquired vibration time series signal and the equipment runtime space data with high-precision timestamps respectively; aligning the data streams from different sources based on the high-precision timestamps; and encapsulating the aligned data into a predefined structure to form a unified data frame.

[0017] Preferably, the working condition self-identification and model self-calibration module is used to control the vibration system of the road roller to emit a diagnostic excitation signal in the form of frequency sweep excitation;

[0018] Time-frequency analysis is performed on the response signal under diagnostic excitation signal to generate a three-dimensional feature spectrum, and the operating condition feature vector characterizing the resonance peak and energy decay rate is extracted from the three-dimensional feature spectrum. The operating condition feature vector is matched with the benchmark features in the built-in basic physical model library to select or calibrate the physical inversion model and state transition function.

[0019] Preferably, the step of the working condition self-identification and model self-calibration module matching the working condition feature vector with the benchmark features in the built-in basic physical model library includes:

[0020] Calculate the distance between the operating condition feature vector and each benchmark feature vector stored in the basic physical model library;

[0021] Determine the baseline feature vector with the minimum distance;

[0022] The physical inversion model and state transition function associated with the baseline feature vector that has the minimum distance are selected as the result of the adaptive configuration.

[0023] Furthermore, the step of the operating condition self-identification and model self-calibration module matching the operating condition feature vector with the benchmark features in the built-in basic physical model library includes: calculating the distance between the operating condition feature vector and each benchmark feature vector stored in the basic physical model library; determining the benchmark feature vector with the minimum distance; and selecting the physical inversion model and state transition function associated with the benchmark feature vector with the minimum distance as the result of adaptive configuration.

[0024] Preferably, the real-time soil condition estimation module specifically includes the following steps;

[0025] A coupled dynamic model of the vibrating wheel and soil was established as a physical inversion model.

[0026] By running the inverse problem solving algorithm, the error between the theoretical vibration response of the coupled dynamics model and the actual vibration response calculated from the vibration time series signal is minimized, thereby obtaining the real-time physical state vector;

[0027] The real-time physical state vector includes the equivalent dynamic stiffness, which characterizes the degree of soil compaction, and the equivalent damping coefficient, which characterizes the energy dissipation characteristics of the soil.

[0028] Preferably, the real-time soil state estimation module minimizes the error through an inverse problem solving algorithm, and the steps include:

[0029] The actual measured vibration response spectrum is calculated based on the vibration time series signal;

[0030] Based on the coupled dynamics model, a parameter vector to be solved, and the frequency, the theoretical vibration response is calculated.

[0031] Using the sum of squared residuals between the actual measured vibration response spectrum and the theoretical vibration response as the objective function, the parameter vector that minimizes the objective function is solved, and this vector is used as the real-time physical state vector obtained by inversion.

[0032] Furthermore, the real-time soil state estimation module minimizes the error through an inverse problem-solving algorithm. Its core is solving the following optimization problem, as shown in the formula below:

[0033] ;

[0034] In the formula, This represents the parameter vector obtained from the inversion. This represents the parameter vector to be solved. Represents all possible parameter vectors middle, Indicates frequency and parameters conditions, This represents the frequency response function calculated from the actually measured vibration time series signal. Indicates frequency, The squared 2-norm of a vector is used to solve the inverse problem, which enables the transformation from measurable vibration signals to unmeasurable but more physically meaningful soil state parameters.

[0035] Preferably, the forward prediction module is used to calculate the corresponding effective compaction energy based on the selected combination of construction parameters;

[0036] The calculation is performed using a state transition function, which characterizes the relationship between the injection of effective compaction energy and the growth of the equivalent dynamic stiffness in the real-time physical state vector. This relationship exhibits a saturation characteristic with diminishing marginal effects.

[0037] Preferably, the forward prediction module performs calculations using a state transition function, including:

[0038] The real-time physical state vector output by the real-time soil state estimation module is received as the current state;

[0039] Calculate the effective compaction energy based on the selected combinations of construction parameters;

[0040] The current state and effective compaction energy are used as inputs, and the state transition function is used to calculate the physical state vector after the next compaction.

[0041] Furthermore, in the state transition function, the evolution relationship of the equivalent dynamic stiffness is defined as:

[0042] ;

[0043] In the formula, This represents the predicted equivalent dynamic stiffness after the next compaction pass. This represents the equivalent dynamic stiffness in the current state. This indicates the maximum equivalent stiffness that the soil material can achieve. This represents the stiffness growth rate coefficient obtained by fitting the sample compaction curve. Indicates the first The effective compaction energy corresponding to each compaction. Indicates the first The combination of construction parameters for compaction. The base of the natural logarithm is used to accurately simulate the physical process of soil compaction, from loose to dense, with the hardening effect gradually weakening.

[0044] Preferably, the multi-objective constraint control decision module is used to define a cost function that integrates construction time and energy consumption;

[0045] Within the allowable construction parameter space determined by the performance of the road roller, the optimal control vector is obtained by minimizing the cost function.

[0046] The optimal control vector includes the optimal vibration frequency, amplitude level, and compaction speed.

[0047] This invention provides a system for optimizing and analyzing roadbed filling quality parameters. It has the following beneficial effects:

[0048] 1. This invention employs diagnostic excitation signals to actively detect filling materials, automatically identify their working characteristics, and then adaptively configure the physical model. This achieves the technical effect of tailoring modeling to specific soil conditions. Compared to existing technologies that rely on a single, fixed physical model or empirical formula to address all construction scenarios, it solves the shortcomings of poor model universality and low calculation accuracy when soil conditions change.

[0049] 2. This invention establishes a coupled dynamic model of the vibrating wheel and soil and applies an inverse problem-solving algorithm. This scheme can calculate core physical state parameters such as the equivalent dynamic stiffness of the soil in real time, making the compaction quality transparent and quantifiable. Compared with existing technologies that use indirect apparent indicators such as compaction value or frequency response, it solves the shortcomings of these technologies, which have unclear physical meanings and cannot truly reflect the deep changes in the mechanical state of the soil.

[0050] 3. This invention combines the state transition function used for prediction with the multi-objective cost function used for decision-making. This achieves forward-looking optimal control. It not only ensures quality meets standards but also finds the most efficient and economical path to achieve those standards. This overcomes the shortcomings of existing technologies that rely on operator experience for trial-and-error control, such as the blindness of the control process, difficulty in guaranteeing quality uniformity, and inability to balance efficiency and energy consumption. Attached Figure Description

[0051] Figure 1 This is a system framework diagram of the present invention;

[0052] Figure 2 This is a schematic diagram of the data processing flow of the full-cycle data acquisition module of the present invention;

[0053] Figure 3 This is a schematic diagram illustrating the workflow of the working condition self-identification and model self-calibration module of the present invention;

[0054] Figure 4 This is a schematic diagram of the inverse problem solving process of the real-time soil state estimation module of the present invention;

[0055] Figure 5 This is a schematic diagram of the state evolution calculation process of the forward prediction module of the present invention;

[0056] Figure 6 This is a schematic diagram of the optimization decision-making process of the multi-objective constraint control decision module of the present invention. Detailed Implementation

[0057] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see the appendix Figure 1 -Appendix Figure 6 This invention provides a roadbed filling quality parameter optimization and analysis system, comprising:

[0059] The full-cycle data acquisition module is used to collect vibration time-series signals and equipment runtime spatial data of the road roller in real time during operation.

[0060] Specifically, the core function of the full-cycle data acquisition module is to accurately and synchronously collect multi-source heterogeneous data of the road roller during operation, providing high-quality, spatiotemporally consistent raw input for subsequent calculation and analysis of various modules.

[0061] For example, the physical structure of this module includes a vibration signal acquisition unit, a spatiotemporal and equipment status acquisition unit, and a data synchronization and fusion processing unit. Both the vibration signal acquisition unit and the spatiotemporal and equipment status acquisition unit are connected to the data synchronization and fusion processing unit.

[0062] Specifically, the vibration signal acquisition unit preferably employs one or more high-frequency microelectromechanical system (MEMS) accelerometers. These sensors are rigidly mounted on the bearing housings of the roller's vibratory drum to directly capture the most authentic vibration response as the vibratory drum interacts with the soil.

[0063] ;

[0064] In the formula, The sampling frequency of the accelerometer. This is the highest excitation frequency that the road roller's vibration system may reach during operation. This is for the safety factor.

[0065] Setting the sampling frequency in this way ensures that the Nyquist theorem is satisfied in the digital signal processing process and can capture the high-order harmonic components and transient impact responses that reflect the nonlinear mechanical behavior of the soil without distortion, laying the foundation for the accuracy of subsequent physical model inversion.

[0066] The spatiotemporal and equipment status acquisition unit specifically includes a high-precision Global Navigation Satellite System (GNSS) receiver that supports real-time dynamic differential (RTK) technology, and a Controller Area Network (CAN) bus interface.

[0067] The GNSS receiver is used to acquire the road roller's three-dimensional geographic coordinates, speed, and heading angle in real time, forming part of the equipment's operational space-time data. The CAN bus interface, by connecting to the road roller's own control network, reads the equipment status parameters such as vibration frequency and amplitude level set by the driver in real time, forming another part of the equipment's operational space-time data.

[0068] The core technical feature of this module lies in its synchronous processing of vibration time series signals and equipment runtime spatial data to generate a unified data frame containing vibration data and equipment status parameters with time and space stamps at each moment.

[0069] This synchronization process, which involves generating a unified data frame, is executed within the data synchronization and fusion processing unit, and its flow is as follows:

[0070] First, each data stream acquired from the vibration signal acquisition unit and the spatiotemporal and equipment status acquisition unit is marked with a high-precision timestamp from a unified source. Preferably, this timestamp signal is provided by a GNSS receiver to ensure the uniformity and accuracy of the time reference.

[0071] Subsequently, data streams from different sources are time-aligned based on high-precision timestamps. This step resolves the asynchronous issues caused by varying delays in the generation and transmission of data from different sensors, ensuring that at any given moment in the analysis, the vibration response corresponds precisely to the roller's location, speed, and operating parameters.

[0072] Finally, the aligned data is encapsulated into a predefined structure to form a unified data frame. This data frame provides a standardized input format for all subsequent computation modules.

[0073] An exemplary unified data frame The structure can be represented as:

[0074] ;

[0075] In the formula, For timestamps, The sampled value of the vibration time series signal at that moment. Let be the three-dimensional position vector of the road roller at that moment. The speed of the road roller at that moment. Set the vibration frequency for the road roller at that moment. Set the amplitude level for the road roller at that moment.

[0076] Through the above-described structure and processing flow, the full-cycle data acquisition module of this invention can achieve accurate and synchronous acquisition and fusion of multi-source information in the compaction operation process. Its output unified data frame provides a highly relevant and spatiotemporally unbiased data foundation for subsequent working condition identification, state inversion, forward prediction and decision optimization.

[0077] The working condition self-identification and model self-calibration module is connected to the full-cycle data acquisition module. It is used to control the road roller to send out diagnostic excitation signals before compaction operations or under specific working conditions, and automatically identify the working condition characteristics of the current filling material based on the acquired response signals, and then adaptively configure the physical inversion model and state transition function.

[0078] Specifically, the working condition self-identification and model self-calibration module is connected to the full-cycle data acquisition module. Its main function is to actively detect the inherent physical and mechanical properties of the current filling material before compaction operations or under specific working conditions, and based on this, automatically configure the most suitable analysis model and parameters for the subsequent calculation module.

[0079] The functional implementation process of this module specifically includes the steps of performing diagnostic stimulation, analyzing response signals to extract features, and matching the features with a model library to complete self-calibration.

[0080] First, the working condition self-identification and model self-calibration module controls the vibration system of the road roller to emit one or more preset diagnostic excitation signals. For example, a preferred diagnostic excitation signal is a frequency sweep excitation signal, whose vibration frequency varies linearly or logarithmically with time within a preset frequency range (e.g., 20Hz to 60Hz). This aims to comprehensively detect the dynamic response characteristics of the soil at different frequencies.

[0081] While issuing diagnostic excitation signals, the full-cycle data acquisition module simultaneously records the response signal of the roller's vibrating drum. The working condition self-identification and model self-calibration module receives this response signal and performs time-frequency analysis on it, using algorithms such as short-time Fourier transform (STFT) or wavelet transform.

[0082] Time-frequency analysis can generate a three-dimensional characteristic spectrum that shows the distribution of vibrational energy in the time and frequency dimensions. The three dimensions of this spectrum are time, frequency, and amplitude.

[0083] The working condition self-identification and model self-calibration module extracts a set of parameters that can characterize the current working condition of the filling material from the three-dimensional feature spectrum map, and combines these parameters into a working condition feature vector. .

[0084] The vector includes at least:

[0085] ;

[0086] In the formula; The frequencies of the resonance peaks identified from the three-dimensional feature spectrum. The damping ratio represents the rate of decrease in equivalent energy. This indicates the transpose operation.

[0087] The load condition self-identification and model self-calibration module matches the load condition feature vectors with benchmark features in a built-in basic physical model library. This library pre-stores multiple benchmark feature vectors. Each reference vector corresponds to a typical, calibrated filling material condition. Furthermore, each reference eigenvector... They are all associated with a set of optimized model parameters. The parameter set includes a physical inversion model for the real-time soil state estimation module, and a state transition function and its key coefficients for the forward prediction module.

[0088] The matching steps specifically include:

[0089] First, calculate the current real-time extracted feature vector of the operating condition. With each baseline eigenvector stored in the fundamental physics model library Distance between Preferably, the distance is calculated using Euclidean distance.

[0090] ;

[0091] In the formula, Represents the feature vector of the current operating condition. With the Standard eigenvectors The Euclidean distance between them Describing the 2-norm, This represents the currently identified feature vector of the operating condition. Represents the first in the basic physical model library A standard feature vector, and This represents two components in the feature vector of the current operating condition. and Indicates the first Reference values ​​in a standard model, This represents the index number of the standard eigenvectors in the basic physics model library.

[0092] Secondly, by comparing all the calculated distances Determine the baseline feature vector with the minimum distance among them. ,Right now;

[0093] ;

[0094] In the formula, Represents the feature vector of the current operating condition. With the Standard operating condition vectors The Euclidean distance between them Indicates all distance values middle, Indicates obtaining the distance Minimum index , This represents the index of the best standard operating condition feature vector that was finally matched.

[0095] Finally, select the benchmark feature vector that has the minimum distance to it. Associated model parameter set The results of the adaptive configuration are then transmitted to the real-time soil state estimation module and the forward prediction module, respectively.

[0096] The real-time soil state estimation module, which is connected to the working condition self-identification and model self-calibration module, in conventional compaction operations, uses an adaptively configured physical inversion model to invert and calculate the collected vibration time series signals into a real-time physical state vector characterizing the physical and mechanical properties of the soil.

[0097] Specifically, the real-time soil condition estimation module is connected to the working condition self-identification and model self-calibration module, and runs continuously during routine compaction operations. Its core function is based on an adaptively configured physical inversion model, which inverts and calculates the vibration time series signals collected by the full-cycle data acquisition module into a real-time physical state vector characterizing the physical and mechanical properties of the soil.

[0098] Specifically, the real-time soil state estimation module first establishes a coupled dynamic model of "vibrating wheel-soil" as a physical inversion model. For example, this model equates the complex interactions to a single-degree-of-freedom forced vibration system, whose vibration differential equation is:

[0099] ;

[0100] In the formula, Let be the equivalent mass of the vibrating wheel. The periodic excitation force generated by the vibration system These represent the vertical displacement, velocity, and acceleration of the vibrating wheel, respectively. The equivalent dynamic stiffness is used to characterize the degree of soil compaction. The equivalent damping coefficient characterizes the energy dissipation properties of soil.

[0101] In this model, the equivalent dynamic stiffness With equivalent damping coefficient These are the unknown parameters to be solved. These parameters constitute the objective of the inversion, namely the real-time physical state vector.

[0102] The real-time soil state estimation module obtains the real-time physical state vector by running an inverse problem-solving algorithm to minimize the error between the theoretical vibration response of the coupled dynamic model and the actual vibration response calculated from the vibration time series signal.

[0103] The process of minimizing the error using the inverse problem-solving algorithm includes the following steps:

[0104] The real-time soil condition estimation module calculates the actual measured vibration response spectrum based on the vibration time series signal received from the full-cycle data acquisition module using the Fast Fourier Transform (FFT) algorithm. .

[0105] The theoretical vibration response is calculated based on the coupled dynamics model. The theoretical response is frequency. and a parameter vector to be solved The function. Parameter vector Defined as:

[0106] ;

[0107] Then, the objective function is the sum of squared residuals between the actual measured vibration response spectrum and the theoretical vibration response. The goal is to find the parameter vector that minimizes the objective function. This optimization problem is defined as follows:

[0108] ;

[0109] In the formula, This represents the parameter vector obtained from the final identification. Represents the parameter that minimizes the value of the subsequent function. The value of , Describe the objective function. Indicates based on parameters The calculated frequency response function, This represents the measured value of the frequency response spectral function. This represents the square of the vector norm.

[0110] A nonlinear optimization algorithm, such as the Levenberg-Marquardt algorithm, is used to iteratively solve the above optimization problem to obtain the objective function. Minimum parameter vector .

[0111] The optimal parameter vector obtained by the solution The resulting real-time physical state vector is output as an inversion result. This vector, along with the corresponding high-precision geographic coordinates, is then transmitted to the forward prediction module.

[0112] In this way, the real-time soil state estimation module can convert vibration signals, which have a relatively indirect physical meaning, into equivalent dynamic stiffness that can directly characterize the soil's density and bearing capacity in real time and quantitatively. And the equivalent damping coefficient characterizing energy dissipation properties. .

[0113] The forward prediction module, which is connected to the real-time soil state estimation module, predicts the physical state vector after the next compaction caused by different combinations of construction parameters based on the real-time physical state vector, according to the adaptively configured state transition function, and obtains the prediction results.

[0114] Specifically, the core function of the forward prediction module is based on the state transition function adaptively configured by the working condition self-identification and model self-calibration module. According to the real-time physical state vector output by the real-time soil state estimation module, it quantitatively predicts the evolution of the soil physical state after the next compaction operation using different combinations of construction parameters, thereby obtaining the prediction results.

[0115] The forward prediction module provides the decision-making basis for the subsequent multi-objective constraint control decision module. Its specific implementation steps are as follows:

[0116] The forward prediction module receives the real-time physical state vector representing the current compaction location, output by the real-time soil state estimation module, and uses it as the current state for the prediction calculation. This vector must include at least the current equivalent dynamic stiffness value. .

[0117] The forward prediction module is based on a candidate combination of construction parameters. Calculate the corresponding effective compaction energy. Construction parameter combinations Preferably, it includes vibration frequency, amplitude level, and compaction speed.

[0118] Effective compaction energy It is used to quantify the effective work injected into the soil during a single compaction process, and its value is related to the vibration power of the roller and the construction coverage efficiency.

[0119] The forward prediction module uses a preset state transition function for calculation. This state transition function characterizes the relationship between the injection of effective compaction energy and the growth of the equivalent dynamic stiffness in the real-time physical state vector, and this relationship preferably exhibits a saturation characteristic with diminishing marginal effects.

[0120] Furthermore, the forward prediction module's calculation using the state transition function specifically includes:

[0121] Receive the real-time physical state vector output by the real-time soil state estimation module as the current state. And extract the current equivalent dynamic stiffness from it. .

[0122] Based on a candidate combination of construction parameters Calculate the corresponding effective compaction energy. The equivalent dynamic stiffness in the current state. With the calculated effective compaction energy As input, the state transition function is used to calculate the physical state vector after the next compaction. This includes the predicted equivalent dynamic stiffness. .

[0123] For example, the state transition function can be characterized by the following mathematical form to represent the evolution of the equivalent dynamic stiffness:

[0124] ;

[0125] In the formula, This indicates the predicted next compaction. Indicates the equivalent dynamic stiffness under the current operating conditions. This represents the theoretically maximum equivalent dynamic stiffness that the soil can achieve under current conditions. This indicates that the construction parameters are based on the selected combination. This represents the calculated effective compaction energy per unit volume. This represents the spring rate growth coefficient.

[0126] It should be noted that the parameters and It is not a fixed value, but rather an adaptively configured value based on the characteristics of the specific filling material by the working condition self-identification and model self-calibration module. This connection ensures the accuracy and working condition adaptability of the prediction model.

[0127] The multi-objective constraint control decision module, connected to the forward prediction module, solves for and outputs the optimal control vector that enables the current physical state vector to reach the target state based on the prediction results.

[0128] Specifically, the multi-objective constraint control decision module performs multi-objective constraint optimization based on the prediction results obtained by the forward prediction module, solves and outputs the optimal control vector that enables the current physical state vector to reach the target state through the optimal path.

[0129] The multi-objective constraint control decision module first defines a cost function that integrates construction time and energy consumption. This function is used to quantify the economic cost required to complete a single compaction operation using a specific combination of construction parameters.

[0130] For example, the cost function It can be represented as:

[0131] ;

[0132] In the formula, For a possible combination of construction parameters, This represents the cost value corresponding to this parameter combination. The time required to complete a single pass using this parameter combination. The energy consumption required to complete a single pass of work, and These are the weighting coefficients for time and energy consumption, respectively.

[0133] The multi-objective constraint control decision module optimizes the system within the allowable construction parameter space determined by the performance of the road roller itself, with the goal of minimizing the cost function, and solves for the optimal control vector. The optimal control vector includes the optimal vibration frequency, amplitude level, and compaction speed.

[0134] Furthermore, the step of the multi-objective constraint control decision module performing the optimization to obtain the optimal control vector specifically includes:

[0135] Obtain the current physical state vector output by the real-time soil state estimation module, and a preset target state. The current physical state vector includes at least the current equivalent dynamic stiffness. The target state includes at least the target equivalent dynamic stiffness. .

[0136] Iterate through a series of discretized combinations of construction parameters within the allowed construction parameter space. ;

[0137] For each combination of construction parameters The forward prediction module is invoked, taking the current state and the combination as input, to obtain the corresponding prediction result, such as the predicted stiffness for the next pass. And calculate its cost value based on the stated cost function. .

[0138] Select an optimal control vector from all combinations of construction parameters. The optimal control vector The following two conditions must be met simultaneously: the prediction result meets the requirement of achieving the target state, and the corresponding cost value is the minimum among all combinations that meet the former condition.

[0139] This selection process can be formally described as solving the following constrained optimization problem.

[0140] ;

[0141] In the formula, This represents the optimal combination of compaction parameters. Indicates the first A number of optional construction parameter combinations Represents the cost function, Indicates the use of parameter combinations hour, This indicates the set target stiffness value. This indicates that among all candidate parameter combinations, the one that satisfies the stiffness requirement is... The index of those schemes.

[0142] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A roadbed filling quality parameter optimization and analysis system, characterized in that, The application relates to a self-adaptive compaction quality control system for road construction, which comprises: a full-cycle data acquisition module for acquiring vibration time series signals and device operation space-time data of a road roller in a working process in real time; a working condition self-identification and model self-calibration module connected with the full-cycle data acquisition module, which is used for controlling the road roller to emit a diagnostic excitation signal before compaction work or under a specific working condition, automatically identifying the working condition characteristics of the filling material according to the acquired response signal, and then adaptively configuring a physical inversion model and a state transition function; the working condition self-identification and model self-calibration module is used for controlling the road roller vibration system to emit a diagnostic excitation signal in the form of a sweep frequency excitation; the response signal under the action of the diagnostic excitation signal is subjected to time-frequency analysis to generate a three-dimensional feature spectrum, and a working condition characteristic vector representing a resonance peak and an energy attenuation rate is extracted from the three-dimensional feature spectrum; the working condition characteristic vector is matched with reference features in a built-in basic physical model library to select or calibrate the physical inversion model and the state transition function; the working condition self-identification and model self-calibration module comprises the following steps of matching the working condition characteristic vector with the reference features in the built-in basic physical model library: calculating the distance between the working condition characteristic vector and each reference characteristic vector stored in the basic physical model library; determining the reference characteristic vector with the minimum distance; selecting the physical inversion model and the state transition function associated with the reference characteristic vector with the minimum distance as the adaptive configuration result; a soil state real-time estimation module connected with the working condition self-identification and model self-calibration module, which is used for inversely calculating the vibration time series signal acquired in the conventional compaction work into a real-time physical state vector representing the physical and mechanical properties of the soil based on the adaptively configured physical inversion model; the soil state real-time estimation module comprises the following steps: establishing a vibration wheel-soil coupling dynamics model as the physical inversion model; obtaining the real-time physical state vector by minimizing the error between the theoretical vibration response of the coupling dynamics model and the actual vibration response calculated according to the vibration time series signal through an inverse problem solving algorithm; wherein the real-time physical state vector comprises an equivalent dynamic stiffness representing the compaction degree of the soil and an equivalent damping coefficient representing the energy dissipation characteristics of the soil; the process of minimizing the error of the soil state real-time estimation module through the inverse problem solving algorithm comprises the following steps: calculating the actual measured vibration response spectrum based on the vibration time series signal; calculating the theoretical vibration response based on the coupling dynamics model, a to-be-solved parameter vector and a frequency; solving the to-be-solved parameter vector that minimizes the objective function of the residual sum of squares between the actual measured vibration response spectrum and the theoretical vibration response, and taking it as the inversely obtained real-time physical state vector; a forward prediction module connected with the soil state real-time estimation module, which is used for predicting the physical state vector after the next compaction caused by different combinations of construction parameters based on the adaptively configured state transition function and the real-time physical state vector, and obtaining a prediction result. A multi-objective constrained control decision module is connected with the forward prediction module, and based on the prediction result, an optimal control vector is solved and output, which can make the current physical state vector reach the target state.

2. The system for optimizing and analyzing the quality parameter of embankment filling according to claim 1, wherein, The full-cycle data acquisition module is used for synchronously processing the vibration time series signal and the equipment operation space-time data to generate a unified data frame containing vibration data and equipment state parameters with space-time stamps at each time.

3. The system for optimizing and analyzing the quality parameter of embankment filling according to claim 1, wherein, The step of synchronously processing the full-cycle data acquisition module to generate a unified data frame includes marking high-precision time stamps for the data streams of the collected vibration time series signal and the equipment operation space-time data respectively. Aligning the data streams of different sources based on the high-precision time stamps. The aligned data is encapsulated into a predefined structure to form a unified data frame.

4. The system for optimizing and analyzing the quality parameter of embankment filling according to claim 1, wherein, The forward prediction module is used for calculating the corresponding effective compaction energy based on the selected construction parameter combination. The state transition function is used for calculation, which represents the relationship between the injection of effective compaction energy and the growth of equivalent dynamic stiffness in the real-time physical state vector, and the relationship presents a saturation characteristic of diminishing marginal effect.

5. The system for optimization analysis of roadbed filling quality parameters according to claim 1, characterized in that, The step of using the state transition function for calculation by the forward prediction module includes: Receiving the real-time physical state vector output by the soil state real-time estimation module as the current state; Calculating the effective compaction energy based on the selected construction parameter combination; Taking the current state and the effective compaction energy as inputs, and performing operation through the state transition function to obtain the physical state vector after the next compaction.

6. The system for optimization analysis of roadbed filling quality parameters according to claim 1, characterized in that, The multi-objective constrained control decision module is used for defining a cost function which integrates construction time and energy consumption; Within the allowable construction parameter space determined by the roller performance, optimization is performed to minimize the cost function, and the optimal control vector is solved; The optimal control vector includes optimal vibration frequency, amplitude level and rolling speed.

Images