A quick density determination system for automotive urea

Abstract

The application relates to the technical field of physical property measurement of chemical substances, and particularly discloses a rapid density determination system for vehicle urea, which comprises a vibrating tube type density sensing unit, a multi-mode cooperative driving and signal processing unit, an embedded real-time calculation and compensation unit and a system control and communication interface unit. The multi-mode cooperative driving and signal processing unit simultaneously excites multiple vibration modes based on a pre-stored resonance spectrum, and quickly solves the initial value of density by analyzing the frequency offset; the embedded real-time calculation and compensation unit compensates the initial value of density in real time in terms of temperature, viscosity and other multi-physical fields by using a neural network model. The application realizes millisecond-level rapid density determination and high-precision output, and meets the real-time regulation and control requirements of an SCR system.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of physical property measurement of chemical substances, and particularly relates to a rapid density determination system for vehicle urea. BACKGROUND

[0002] In the field of diesel engine exhaust aftertreatment, selective catalytic reduction technology is the core means to reduce nitrogen oxide emissions, which relies on the accurate injection and decomposition of vehicle urea solution. The density of urea solution is a key parameter to characterize its concentration and quality, which is directly related to the accuracy of reductant supply and the purification efficiency of the SCR system, so the rapid and accurate determination of the density of urea solution has important engineering application value.

[0003] The density sensor based on the principle of vibrating tube is a common technical solution for online detection of urea solution. This technology indirectly calculates the density of the liquid by measuring the change of the natural frequency of the vibrating tube filled with the liquid to be measured. The basic principle is that there is a certain functional relationship between the vibration frequency and the mass of the medium in the tube.

[0004] The piezoelectric driving circuit inside the traditional vibrating tube density sensor needs to go through a frequency search and locking process when starting measurement or changing working conditions, so that the driving signal matches the mechanical resonance frequency of the vibrating tube under the current medium. This process usually requires a stable time of several seconds, which causes the sensor to be unable to provide instantaneous density readings. Under the actual operating conditions of diesel engines, the SCR system requires millisecond-level control of urea injection amount, and the lag of sensor response seriously restricts the real-time performance and accuracy of the closed-loop control system, which may cause excessive or insufficient urea injection, thereby affecting the emission control effect and increasing the operating cost. Therefore, how to significantly shorten the response time of the vibrating tube density sensor and make its output meet the real-time and rapid regulation and control requirements of the SCR system has become a technical problem to be solved in the field. SUMMARY

[0005] The purpose of the present application is to provide a rapid density determination system for vehicle urea, to solve the technical contradiction that the existing density sensor based on the principle of vibrating tube has a too long response time when starting or changing working conditions, and cannot meet the millisecond-level real-time regulation and control requirements of the SCR system due to the need to search and lock the mechanical resonance frequency.

[0006] To achieve the above purpose, the present application provides a rapid density determination system for vehicle urea, comprising:

[0007] A vibrating tube type density sensor unit for containing the vehicle urea solution to be measured and generating a mechanical vibration signal related to the density of the solution;

[0008] A multi-modal collaborative driving and signal processing unit is used to generate driving signals to excite the vibrating tube and to process the pick-up signals synchronously to extract frequency and amplitude information;

[0009] An embedded real-time computation and compensation unit is used to compensate and correct the multi-physical field coupling effects such as temperature, viscosity and tube stress on the density initial value output by the multi-modal collaborative driving and signal processing unit, to output the final high-precision density value;

[0010] A system control and communication interface unit is used to coordinate the working timing of the units inside the system, to process external instructions, and to output measurement results according to the preset protocol;

[0011] The multi-modal collaborative driving and signal processing unit is internally integrated with a high-precision direct digital frequency synthesizer, a programmable gain amplifier array, a high-speed analog-to-digital converter and a digital signal processor, and the working process of the multi-modal collaborative driving and signal processing unit is as follows:

[0012] After the system is powered on or receives a measurement instruction, the digital signal processor calls the pre-stored reference resonance spectrum of the vibrating tube in a reference medium from the non-volatile memory, and the reference resonance spectrum contains the resonance frequency values of multiple order modes of the vibrating tube in the reference medium, the corresponding quality factors and the frequency interval relationship between the modes;

[0013] The direct digital frequency synthesizer simultaneously generates multiple-frequency sinusoidal wave signals of 2 or more different frequencies according to the reference resonance spectrum, and these frequencies correspond to the resonance frequency points of the 1st and 2nd bending modes predicted in the reference resonance spectrum;

[0014] The vibrating tube is pasted with a driving piezoelectric ceramic sheet, and the multiple-frequency sinusoidal wave signals are power amplified by the programmable gain amplifier and then applied to the driving piezoelectric ceramic sheet together;

[0015] At the same time, the vibration response signals sensed by the driving piezoelectric ceramic sheet are preliminarily conditioned by the programmable gain amplifier array and then synchronously sampled by the high-speed analog-to-digital converter;

[0016] The digital signal processor performs fast Fourier transform on the sampled time-domain signals to obtain a real-time vibration response spectrum;

[0017] The digital signal processor compares and analyzes the real-time vibration response spectrum with the reference resonance spectrum, calculates the relative change of the response amplitude near each excitation frequency point and the absolute shift of the resonance peak frequency of each mode;

[0018] Based on the vibration beam theory, the digital signal processor constructs an overdetermined set of equations by synchronously monitoring the frequency offset of at least two modes, thereby quickly calculating the equivalent additional mass of the current urea solution acting on the vibration tube, and then deriving the initial density value.

[0019] After calculating the initial density value, the digital signal processor synchronously starts the background refinement tracking thread.

[0020] The refined tracking thread uses the calculated initial density value as a basis, combined with the real-time compensation temperature value obtained from the embedded real-time calculation and compensation unit, to back-calculate the accurate resonant frequency prediction value of the first-order main mode of the vibrating tube under the current working condition through the theoretical model.

[0021] Subsequently, the multi-modal collaborative driving and signal processing unit smoothly transitions the driving mode from multi-frequency collaborative to a single-frequency phase-locked loop tracking mode centered on the precise resonant frequency.

[0022] At the moment the system starts up, the embedded real-time calculation and compensation unit first reads the initial temperature value of the high-precision temperature sensor;

[0023] The digital signal processor calculates the expected frequency drift of the current temperature relative to the reference calibration temperature based on the pre-stored thermal expansion coefficient and elastic modulus temperature coefficient of the vibrating tube material. The digital signal processor then adds this drift to the resonant frequency recorded in the reference resonant spectrum, which serves as the frequency setting value for the initial multi-frequency sine wave signal generated by the direct digital frequency synthesizer.

[0024] Preferably, the embedded real-time computing and compensation unit includes a microcontroller, a high-precision temperature sensor, and a data module storing a complete compensation algorithm model. The compensation process of the embedded real-time computing and compensation unit is as follows:

[0025] The high-precision temperature sensor measures the temperature of the urea solution vibrating against the pipe wall or in close contact with the pipe wall in real time; the microcontroller receives the temperature data and the initial density value; the data module pre-stores a compensation model established through extensive experimental calibration. This compensation model is a multidimensional nonlinear function mapping relationship with the initial density value and temperature as input and the corrected density value as output. This multidimensional nonlinear function mapping relationship is implemented through a trained three-layer feedforward neural network, whose hidden layer contains 8 neurons and the activation function uses a modified linear unit; the microcontroller executes the neural network model to compensate the initial density value in real time and outputs the final calibrated urea solution density value.

[0026] Preferably, the system control and communication interface unit receives a measurement request signal from the vehicle SCR controller, triggers and synchronizes the start of the multi-modal collaborative driving and signal processing unit and the embedded real-time calculation and compensation unit, and after completing a measurement cycle, encapsulates the final density value, solution temperature value and system status word into a data frame conforming to the controller area network or vehicle Ethernet protocol, and sends it to the SCR controller through the physical layer interface.

[0027] Preferably, in the single-frequency phase-locked loop tracking mode, the direct digital frequency synthesizer generates a single-frequency sinusoidal driving signal with the accurate resonant frequency prediction value as the center frequency.

[0028] The digital signal processor internally enables a high-bandwidth digital phase-locked loop algorithm that continuously compares the phase difference between the driving signal and the pickup signal and dynamically adjusts the output frequency of the direct digital frequency synthesizer so that the phase difference is always locked around the 90-degree resonance point.

[0029] Preferably, the method for obtaining and updating the reference resonant spectrum is as follows: before the sensor is shipped or during periodic maintenance, fill the vibration tube cavity with standard reference medium and run a high-precision full-spectrum scanning calibration program at a standard temperature; the full-spectrum scanning calibration program controls the direct digital frequency synthesizer to scan in a wide frequency band with small steps, while recording the amplitude of the pickup signal, thereby accurately plotting the complete resonant curve of the vibration tube under the reference medium, identifying and recording the accurate resonant frequencies and quality factors of the first three modes, and storing them as non-volatile reference data.

[0030] Preferably, the system further comprises an online self-diagnosis and early warning unit; the online self-diagnosis and early warning unit continuously monitors the signal quality indicators output by the multi-modal collaborative driving and signal processing unit, including the signal-to-noise ratio, harmonic distortion and stability of the driving current of each mode; at the same time, it monitors the input and output rationality of the neural network model in the embedded real-time calculation and compensation unit.

[0031] Preferably, the online self-diagnosis and early warning unit is also used to: when it is monitored that the signal quality continuously falls below the preset threshold, or the final density value after compensation exceeds the reasonable physical range of urea solution, it is determined that there is a potential fault in the sensor; the online self-diagnosis and early warning unit immediately sends a specific fault code and early warning information to the SCR controller through the system control and communication interface unit.

[0032] Compared with the prior art, the present application has the following advantages:

[0033] 1、The present application fundamentally changes the working mode of the traditional vibrating tube density sensor which must undergo a time-consuming frequency search process by adopting a multi-modal collaborative driving strategy based on pre-stored resonance spectrum. The system directly calculates the initial density value within milliseconds by simultaneously exciting and analyzing the frequency shift of multiple modes using an overdetermined equation, shortening the response time from seconds to milliseconds from the start to the first effective output of the sensor, significantly improving the dynamic response speed, and meeting the stringent real-time requirements of the SCR system for urea solution density information, laying a key foundation for more accurate urea injection closed-loop control.

[0034] 2、The present application constructs a double-layer measurement architecture combining multi-modal rapid measurement with high-precision single-frequency tracking. The system uses multi-modal collaboration to achieve instantaneous response at the moment of starting, and seamlessly switches to the phase-locked loop mode for continuous high-precision tracking after stabilization. This architecture takes into account the dual needs of rapid startup and long-term stable high-precision monitoring, enabling the system to quickly follow up when the working conditions change, and to provide better measurement accuracy and stability than traditional methods in steady state, thereby comprehensively improving the overall performance indicators of the sensor.

[0035] 3、The present application integrates an embedded real-time multi-physical field compensation unit based on neural networks. The embedded real-time multi-physical field compensation unit uses a pre-stored, nonlinear compensation model trained with a large amount of experimental data to perform real-time, online comprehensive compensation of cross-sensitive factors such as temperature and viscosity. Compared to traditional segmented linear or simple polynomial compensation methods, the neural network model can more accurately fit complex coupling effects, effectively eliminating the influence of non-density factors on the measurement results, thereby ensuring high accuracy and reliability of the final output density value under a wide temperature range and different solution qualities.

[0036] 4、The present application introduces an initial frequency setting mechanism with temperature pre-compensation and an online self-diagnosis function. The temperature pre-compensation mechanism makes the initial driving frequency of the system closer to the actual working conditions, further optimizing the startup performance. The online self-diagnosis function continuously monitors the health status and output reasonableness of the sensor core signal chain, can timely detect potential faults such as crystallization and failure, and reports them through standard vehicle communication protocols, enhancing the reliability and maintainability of the system and ensuring long-term stable operation of the SCR aftertreatment system. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is the overall technical scheme architecture diagram of the urea density rapid measurement system for vehicles according to the present application;

[0038] Figure 2 is the core principle framework diagram of the multi-modal collaborative driving strategy based on pre-stored resonance spectrum and real-time dynamic tracking in the present application;

[0039] Figure 3 is the logical flow framework diagram of the multi-modal collaborative driving and signal processing unit in the application;

[0040] Figure 4 is the multi-physical field coupling effect compensation principle framework diagram of the embedded real-time calculation and compensation unit in the application;

[0041] Figure 5 is the double-layer measurement architecture diagram combining fast start and continuous high-precision monitoring in the application. DETAILED DESCRIPTION

[0042] The application provides a vehicle urea density rapid determination system, and the overall technical scheme architecture of the vehicle urea density rapid determination system is shown in FIG. 1. Figures 1 to 5 The system aims to solve the core technical contradiction that the response time of the existing density sensor based on the vibration tube principle is too long due to the need to search for and lock the mechanical resonance frequency, and the real-time regulation and control demand of the selective catalytic reduction system cannot be met. The system realizes millisecond-level fast determination and continuous high-precision monitoring of the density of vehicle urea solution from start to output through an innovative multi-modal collaborative driving and signal processing strategy combined with embedded real-time multi-physical field compensation.

[0043] The vehicle urea density rapid determination system is composed of four core functional units: a vibration tube type density sensing unit, a multi-modal collaborative driving and signal processing unit, an embedded real-time calculation and compensation unit, and a system control and communication interface unit. These units interact through precise electrical connections and data buses to form a complete closed-loop measurement and control system.

[0044] First, the specific implementation details of the vibration tube type density sensing unit are described. The specific implementation details of the vibration tube type density sensing unit are the mechanical sensing core of the system, and its function is to contain the vehicle urea solution to be measured and generate a mechanical vibration signal directly related to the density of the solution. The main body of the specific implementation details of the vibration tube type density sensing unit is a hollow thin-walled vibration tube made of a specific elastic alloy material. The specific elastic alloy material is usually selected from a type with high elastic modulus, low temperature coefficient and excellent urea corrosion resistance, such as nickel-based high-temperature alloy or titanium alloy with special surface treatment.

[0045] The geometric dimensions of the vibration tube are precisely designed, and the ratio of its length, outer diameter and wall thickness needs to meet specific stiffness and mass distribution requirements to ensure that it has clearly separated and stable first few order bending vibration modes in the target frequency range, for example, 100 Hz to 5000 Hz. The two ends of the vibration tube are sealed and fixed by high-strength and high-rigidity metal flanges, which are fastened to the inner wall of the stainless steel or engineering plastic shell of the sensor by bolts. This mounting method is equivalent to a beam structure with both ends fixed in mechanics, which provides clear boundary conditions for vibration.

[0046] At specific axial and circumferential positions of the outer wall of the vibrating tube, two groups of piezoelectric ceramic sheets are symmetrically pasted by high-precision patching process. One group of piezoelectric ceramic sheets serves as a driving element, responsible for converting electrical signals into mechanical force to excite the vibrating tube to vibrate; the other group serves as a vibration pickup element, responsible for converting the mechanical vibration of the vibrating tube into electrical signals.

[0047] The selection of piezoelectric ceramic sheets requires high electromechanical coupling coefficient, low aging rate and wide temperature stability. The driving and pickup ceramic sheets usually appear in pairs and are pasted at the antinode positions corresponding to the maximum modal strain of the vibrating tube to maximize the excitation and detection efficiency. The entire vibrating tube assembly, piezoelectric ceramic sheets and the sensor housing which provides mechanical protection and environmental sealing together constitute a complete mechanical resonance cavity. The sensor housing is designed with urea solution inlet and outlet flow channels, and usually has a rectifier structure inside to reduce the interference of fluid flow on the vibrating tube and ensure that the urea solution can quickly and uniformly fill the internal cavity of the vibrating tube.

[0048] Next, the specific implementation of the multi-modal cooperative driving and signal processing unit will be described in detail, and the core principle framework is shown in FIG. 1, and the logical flow framework is shown in FIG. 2. Figure 2 Figure 3 The multi-modal cooperative driving and signal processing unit is the key innovation of the present application, which abandons the single frequency search and phase-locked loop locking mode adopted by traditional vibrating tube density sensors, and instead implements a multi-modal cooperative driving strategy based on pre-stored resonance spectrum and real-time dynamic tracking.

[0049] The multi-modal cooperative driving and signal processing unit integrates high-precision direct digital frequency synthesizer, multi-channel programmable gain amplifier array, synchronous sampling high-speed analog-to-digital converter and high-performance digital signal processor in hardware. These hardware components are interconnected through internal high-speed parallel bus and serial peripheral interface.

[0050] The working process of the multi-modal cooperative driving and signal processing unit starts from system power-on initialization or receiving a measurement trigger instruction from the outside. The digital signal processor first calls the pre-stored reference resonance spectrum of the specific vibrating tube in the reference medium from its internal or externally connected non-volatile memory. The reference medium is usually deionized water or standard concentration, such as 32.5% urea solution for vehicle, with known density and viscosity. The reference resonance spectrum is obtained on a high-precision calibration platform before the sensor is shipped or periodically maintained. The calibration process is as follows: fill the vibrating tube cavity with reference medium and place it in a constant temperature environment, such as 25 degrees Celsius.

[0051] ​The system runs a full spectrum scan calibration program, which controls the direct digital frequency synthesizer to scan in a preset wide frequency band, for example, 100 Hz to 5000 Hz, with a very small frequency step, for example, 0.1 Hz. At each frequency point, the direct digital frequency synthesizer outputs a fixed amplitude sinusoidal wave drive signal, while the high-speed analog-to-digital converter synchronously collects the response signal output by the pick-up element.

[0052] The digital signal processor calculates the amplitude of the collected time domain signal or performs a fast Fourier transform analysis, and records the vibration response amplitude corresponding to each frequency point. After the scan is completed, a complete amplitude-frequency characteristic curve, i.e., a resonance curve, is obtained. From the resonance curve, the system automatically identifies and extracts the accurate resonance frequency values of the first several orders, usually the first three orders, of bending vibration modes, the quality factors corresponding to each resonance peak, the half-power bandwidth, and the frequency interval relationship between each modal resonance peak.

[0053] After these data are formatted, they are permanently stored in the non-volatile memory as a reference data package. The system can perform a quick self-check program at each power-up, which verifies the integrity of the signal link by injecting a signal of a known frequency and checking the response, but usually does not perform a full spectrum rescan.

[0054] After the reference spectrum is called, the digital signal processor immediately sends control instructions to the direct digital frequency synthesizer according to the spectrum data. The direct digital frequency synthesizer simultaneously generates two or more sinusoidal signals of different frequencies according to the instructions. In a typical implementation, this set of initial multi-frequency sinusoidal signals contains two frequency components, which accurately correspond to the resonance frequency values of the first order bending mode and the resonance frequency value of the second order bending mode .

[0055] The direct digital frequency synthesizer has the ability to simultaneously generate multiple independent frequency signals, with a frequency resolution of 0.001 Hz and very low phase noise. This set of multi-frequency composite signals is sent to the programmable gain amplifier for power amplification. The gain value of the programmable gain amplifier is preset by the digital signal processor according to the typical drive efficiency of each mode corresponding to the reference spectrum, to ensure that each frequency component can excite the vibration tube with appropriate energy. The amplified multi-frequency drive voltage is applied to the drive piezoelectric ceramic piece pasted on the vibration tube.

[0056] The vibration response signal sensed by the driving piezoelectric ceramic attached to the vibrating tube is a complex voltage signal containing multiple frequency components when the driving is applied. The complex voltage signal first enters the independent conditioning channel in the programmable gain amplifier array. The independent conditioning channel usually includes an anti-aliasing low-pass filter and a programmable gain amplifier for adjusting the signal amplitude to the input range of the analog-to-digital converter. The conditioned analog signal is sent to the high-speed analog-to-digital converter for synchronous sampling.

[0057] The sampling rate of the high-speed analog-to-digital converter must meet the Nyquist sampling theorem, usually set to more than 4 times the highest frequency of interest, for example 20 kHz. Its sampling clock is synchronized with the clock of the direct digital frequency synthesizer to ensure the coherence of signal processing.

[0058] The high-speed analog-to-digital converter transmits the collected time-domain voltage sequence to the digital signal processor. The digital signal processor performs a fast Fourier transform on this time-domain data, converting it to the frequency domain, thereby obtaining the real-time vibration response spectrum of the vibrating tube under the complex driving at the current time. The real-time vibration response spectrum contains resonance peaks near the driving frequencies and The digital signal processor then performs the core comparison and analysis algorithm. The comparison and analysis algorithm first locates the peak frequency and of the current actual resonance peak in the real-time vibration response spectrum within a narrow search window, for example ±10 Hz, centered on the pre-stored reference frequency and and reads its corresponding amplitude and . At the same time, the corresponding reference amplitude and of the reference medium is read from the reference data.

[0059] Based on the theory of vibrating beams, for a beam structure with both ends fixed, the resonance frequency of the first bending mode has a certain functional relationship with the effective mass of the beam itself, including the mass of the tube and the additional mass of the medium in the tube. When the density of the medium in the tube changes, it is equivalent to adding additional mass to the vibrating tube, resulting in a decrease in the resonance frequency of all modes.

[0060] Different modes have different sensitivities to additional mass, and high-order modes usually have higher sensitivity. Therefore, the frequency shift of the first mode and the medium density There is a correlation. By simultaneously monitoring the frequency shifts of at least two modes, a system of equations can be constructed. Assuming the vibrating tube is a uniform beam and the medium is uniformly attached, the relationship between its frequency and density can be approximately described by the following formula:

[0061]

[0062] in, The currently measured number First-order modal frequency, Based on the first First-order modal frequency, Let the density of the urea solution be the unknown. For the first The sensitivity coefficient of the first mode is a constant related to the geometry, material properties and modal order of the vibrating tube, and is determined through calibration.

[0063] For 2 modes =1 and =2, we can obtain two equations. The digital signal processor uses the pre-stored calibration sensitivity coefficients C1 and C2 to measure the... and Substituting into the above formula, we can construct about The system consists of an overdetermined set of equations. Due to minute measurement errors, the system employs optimization algorithms such as the least squares method to solve for the optimal density value. .

[0064] this This means that the initial density value can be rapidly calculated in milliseconds, typically less than 10 milliseconds, using a multimodal collaborative driving strategy. This process completely avoids the time-consuming steps of slowly scanning or iteratively locking a single frequency in traditional methods, achieving instantaneous and rapid density determination.

[0065] Furthermore, in the multimodal collaborative driving strategy, the frequency setting of the initial multi-frequency sine wave signal can introduce a pre-offset based on temperature prediction to further optimize startup performance. Specifically, at the instant the system starts, the embedded real-time calculation and compensation unit first reads the initial temperature value from the high-precision temperature sensor. The digital signal processor uses the pre-stored coefficient of thermal expansion of the vibrating tube material. and temperature coefficient of elastic modulus Calculate the current temperature Relative to reference calibration temperature For example, the expected frequency drift scaling factor at 25 degrees Celsius.

[0066] For the The temperature drift of the frequency of the first mode can be approximated as: The digital signal processor uses this calculated frequency drift. Resonant frequency of the reference The temperature pre-compensated frequency setting value is obtained The direct digital frequency synthesizer generates the initial multi-frequency sinusoidal signal according to This makes the initial driving frequency closer to the actual resonant point of the vibrating tube at the current temperature at the moment of starting, effectively reducing the transient establishment time of the system from power-on to stable signal output.

[0067] Then the specific implementation of the embedded real-time calculation and compensation unit is described, and the compensation principle framework is described in the attached Figure 4 The core function of the embedded real-time calculation and compensation unit is to perform real-time compensation and correction of the density initial value output by the multi-modal cooperative driving and signal processing unit in terms of temperature, viscosity, and pipe stress, etc. multi-physical field coupling effects, and finally output a high-precision urea solution density value. The embedded real-time calculation and compensation unit is packaged with a microcontroller, or several high-precision temperature sensors, and a data module storing the complete compensation algorithm model on the hardware. The microcontroller and the digital signal processor exchange data through a high-speed serial communication interface such as a serial peripheral interface.

[0068] The high-precision temperature sensor is used to measure the temperature of the vibrating tube wall or the urea solution flowing close to the pipe wall in real time. The temperature sensor usually uses platinum resistance or high-precision negative temperature coefficient thermistor, and the probe realizes good thermal contact with the outer wall of the vibrating tube through heat-conducting silicone grease or mechanical clamps, ensuring the real-time and accuracy of temperature measurement. The analog signal of the temperature sensor is sampled by the analog-to-digital converter inside the microcontroller or the external analog-to-digital converter after passing through the signal conditioning circuit, and converted into a digital temperature value .

[0069] The data module pre-stores a mathematical model for multi-physical field effect compensation established through a large number of experiments. The mathematical model is a multi-dimensional nonlinear function mapping relationship with the density initial value and the real-time temperature value as input, and the corrected final density value as output. Since temperature changes will cause changes in the volume expansion density of the urea solution itself, changes in the elastic modulus of the vibrating tube material affecting the resonant frequency, and changes in the viscosity of the solution affecting the vibration damping, these effects are coupled with each other, and traditional segmented linear compensation or simple polynomial fitting cannot meet the high-precision requirement. Therefore, in this embodiment, the mapping relationship is realized through a three-layer feedforward artificial neural network specially trained.

[0070] The structure of the three-layer feedforward artificial neural network is as follows: the input layer contains 2 neurons corresponding to the input parameters and The output layer contains 1 neuron, corresponding to the output parameter The hidden layer contains 8 neurons. The activation function of the hidden layer neurons adopts the rectified linear unit, whose mathematical expression is The output layer neuron adopts the linear activation function.

[0071] The weight matrix and bias vector parameters of the three-layer feedforward artificial neural network are obtained through a large number of experimental calibration data training. The training data set covers the full concentration range of the urea solution for vehicle, such as 0% to 40%, the full working temperature range, such as -30 degrees Celsius to 80 degrees Celsius, and the viscosity changes that different quality urea solutions may bring. The training process adopts the back propagation algorithm and the gradient descent optimizer, and the goal is to minimize the mean square error between the predicted value of the three-layer feedforward artificial neural network and the measured value of the high-precision reference densimeter. After training and verification, all the weight and bias parameters of the three-layer feedforward artificial neural network are solidified and stored in the read-only memory or flash memory of the data module.

[0072] In each measurement period, the microcontroller synchronously receives the density initial value from the digital signal processor and the real-time temperature value from the temperature sensor . The firmware program of the microcontroller performs forward propagation calculation of the three-layer feedforward artificial neural network. The specific calculation process is: first, the input value is normalized for pretreatment, so that it falls within the range set during the training of the three-layer feedforward artificial neural network, such as 0 to 1.

[0073] Then the normalized input vector is input into the network, the weighted sum of each neuron in the hidden layer is calculated according to the formula and the rectified linear unit activation is applied, the output of the hidden layer is weighted to obtain the value of the output layer, and finally the output value is denormalized to obtain the final compensated density value The whole calculation process is highly optimized and can be completed in the microsecond level on the microcontroller, meeting the real-time requirements. Through compensation, the system can effectively strip the influence of temperature, viscosity and other non-density factors on the measurement results, so that the final output density value remains high precision in a wide temperature range and under different solution states, and the typical precision is better than ±0.0005 grams per cubic centimeter.

[0074] Finally, the specific implementation of the system control and communication interface unit is described. The system control and communication interface unit serves as the hub for the system to interact with the external vehicle network, responsible for coordinating the working time sequence of the internal units, analyzing external instructions and formatting the output measurement results. The system control and communication interface unit is usually realized with a microcontroller or a field programmable gate array as the core, and it runs a real-time task scheduling program internally.

[0075] The system control and communication interface unit continuously monitors the measurement request signal from the vehicle selective catalytic reduction controller through an isolated digital input channel. The measurement request signal is usually a pulse width modulation signal or a specific controller area network remote frame.

[0076] Upon receiving a valid measurement request, the system control and communication interface unit immediately sends a hardware interrupt or command word to the digital signal processor of the multi-modal cooperative driving and signal processing unit, triggering it to start a new measurement cycle. Meanwhile, it informs the microcontroller of the embedded real-time computation and compensation unit to prepare for data reception. The system control and communication interface unit maintains an accurate timing state machine internally, ensuring the orderly connection of driving, sampling, computation, and compensation.

[0077] After a complete measurement cycle, the system control and communication interface unit reads the final compensated density value from the embedded real-time computation and compensation unit and the current solution temperature value T from the temperature sensor. In addition, it also collects system status information such as signal quality flags, fault codes, etc. from the multi-modal cooperative driving and signal processing unit and the self-diagnosis unit.

[0078] The system control and communication interface unit encapsulates these data into standard data frames according to the pre-set communication protocol. For vehicle applications, the communication protocol is usually the CANopen device protocol in the controller area network protocol or the SAE J1939 protocol, or the SOME / IP or DoIP protocol in the vehicle Ethernet. The data frame at least contains the density value, temperature value, status word, and cyclic redundancy check code.

[0079] After encapsulation, the system control and communication interface unit sends the data frame to the vehicle network bus through its integrated controller area network controller physical layer interface or Ethernet media access controller physical layer interface, for the selective catalytic reduction controller to receive and use for closed-loop calculation of urea injection quantity.

[0080] Further, the digital signal processor, after quickly solving the initial density value by the multi-modal cooperative strategy, will simultaneously start a background refinement tracking thread. Its architecture diagram is shown in the attached Figure 5 . The purpose of this refinement tracking thread is to seamlessly transition to a higher precision and higher update rate continuous monitoring mode after the quick start is completed.

[0081] The specific implementation is as follows: the digital signal processor, based on the solved and the real-time compensation temperature T obtained from the microcontroller, combines the theoretical model and calibration parameters of the vibrating tube to reversely calculate the accurate resonant frequency prediction value of the first-order main modal of the vibrating tube under the current density and temperature conditions The calculation utilizes an inverse function model of the aforementioned frequency-density-temperature relationship. Once the density value is calculated , the digital signal processor issues a mode switching command to the multi-modal cooperative driving and signal processing unit. The direct digital frequency synthesizer stops outputting the multi-frequency signal and instead generates a single-frequency sinusoidal signal with as the center frequency.

[0082] Meanwhile, the high-bandwidth digital phase-locked loop algorithm is enabled inside the digital signal processor. The digital phase-locked loop algorithm continuously compares the phase difference between the driving signal and the pick-up signal and dynamically adjusts the output frequency of the direct digital frequency synthesizer so that the phase difference is always locked near the 90-degree resonance point. The phase-locked loop has extremely high frequency tracking accuracy, typically better than 0.01 hertz, and extremely fast tracking speed, capable of following the resonance frequency changes caused by small changes in density or temperature drift in real time.

[0083] In this mode, the system measures the density of the urea solution at the frequency locked by the phase-locked loop as the measurement value, and calculates a high-update-rate density data stream through a lookup table or a simplified single-mode density-frequency relationship model. This two-layer measurement architecture that combines multi-modal fast startup and single-mode high-precision tracking takes into account both the instantaneous response capability and the steady-state measurement performance of the system.

[0084] In addition, the system also includes an online self-diagnosis and early warning unit integrated into the system control and communication interface unit or running as an independent coprocessor. The online self-diagnosis and early warning unit continuously monitors the health status of the entire signal chain through software algorithms. Its diagnosis mainly includes the following aspects:

[0085] First, monitor the real-time vibration response frequency spectrum quality output by the multi-modal cooperative driving and signal processing unit, calculate the signal-to-noise ratio, total harmonic distortion of each modal signal, and stability indicators of the driving current.

[0086] Second, monitor the basic noise level of the pick-up signal to determine whether the signal attenuation is caused by the failure or poor connection of the piezoelectric ceramic sheet.

[0087] Third, monitor the input and output rationality of the neural network model embedded in the embedded real-time calculation and compensation unit, such as checking whether the input density initial value and temperature T are within the effective range covered by the training data set, and checking whether the output final density value is within the reasonable physical density range of the urea solution for vehicles, such as 1.0 to 1.2 grams per cubic centimeter.

[0088] Fourth, combined with the temperature and density output, the concentration and freezing point of the current solution are estimated by the built-in urea solution property model, and cross-verified with reasonable values. The self-diagnosis unit sets a series of dynamic or static thresholds for these monitoring indicators.

[0089] When any indicator exceeds the threshold for a preset length of time, such as the signal-to-noise ratio being below 20 decibels for 5 consecutive cycles or the output density value exceeding 1.15 grams per cubic centimeter for 3 consecutive times, the self-diagnosis unit determines that the sensor has a potential fault. Possible fault types include crystallization blockage inside the vibrating tube, delamination or aging of the piezoelectric ceramic sheet, and abnormal fluid medium that is not urea solution, etc.

[0090] Once a fault is confirmed, the self-diagnosis unit generates a specific diagnostic fault code and, through the system control and communication interface unit, encapsulates it into the status word of the output data frame according to the standard on-board diagnostic protocol or actively sends a controller area network error frame to the selective catalytic reduction controller to issue a warning message. The controller can decide whether to enable the backup urea density estimation strategy, limit engine torque, or turn on the fault indicator light on the driver's dashboard according to the fault level, thereby ensuring the safe operation of the aftertreatment system and prompting maintenance needs.

Claims

1. A rapid density determination system for automotive urea, characterized in that, include: A vibrating tube density sensing unit is used to contain the automotive urea solution to be tested and generate a mechanical vibration signal related to the solution density. The multimodal collaborative drive and signal processing unit is used to generate drive signals to excite the vibrating tube and synchronously process the vibration pickup signals to extract frequency and amplitude information. An embedded real-time calculation and compensation unit is used to perform real-time compensation and correction on the initial density value output by the multi-modal collaborative driving and signal processing unit for multi-physical field coupling effects such as temperature, viscosity and tube stress, so as to output the final high-precision density value. The system control and communication interface unit is used to coordinate the working timing of various units within the system, process external commands, and output measurement results according to a preset protocol. The multimodal collaborative driving and signal processing unit integrates a high-precision direct digital frequency synthesizer, a programmable gain amplifier array, a high-speed analog-to-digital converter, and a digital signal processor. The workflow of the multimodal collaborative driving and signal processing unit is as follows: After the system is powered on or a measurement command is received, the digital signal processor retrieves the pre-stored reference resonance spectrum of the vibrating tube in the reference medium from the non-volatile memory. The reference resonance spectrum includes the resonant frequency values ​​of multiple order modes of the vibrating tube in the reference medium, the corresponding quality factors, and the frequency interval relationship between each mode. The direct digital frequency synthesizer generates two or more multi-frequency sinusoidal signals of different frequencies simultaneously based on the reference resonant spectrum. These frequencies correspond to the resonant frequency points of the first and second order bending modes predicted in the reference resonant spectrum, respectively. A driving piezoelectric ceramic sheet is attached to the vibrating tube. The multi-frequency sine wave signal is amplified by the programmable gain amplifier and then applied to the driving piezoelectric ceramic sheet. Meanwhile, the vibration response signal sensed by the driving piezoelectric ceramic sheet is initially conditioned by the programmable gain amplifier array and then synchronously sampled by the high-speed analog-to-digital converter. The digital signal processor performs a fast Fourier transform on the sampled time-domain signal to obtain the real-time vibration response spectrum; The digital signal processor compares and analyzes the real-time vibration response spectrum with the reference resonance spectrum, and calculates the relative change of the response amplitude near each excitation frequency point and the absolute offset of the resonant peak frequency of each mode. Based on the vibration beam theory, the digital signal processor constructs an overdetermined set of equations by synchronously monitoring the frequency offset of at least two modes, thereby quickly calculating the equivalent additional mass of the current urea solution acting on the vibration tube, and then deriving the initial density value. After calculating the initial density value, the digital signal processor synchronously starts the background refinement tracking thread. The refined tracking thread uses the calculated initial density value as a basis, combined with the real-time compensation temperature value obtained from the embedded real-time calculation and compensation unit, to back-calculate the accurate resonant frequency prediction value of the first-order main mode of the vibrating tube under the current working condition through the theoretical model. Subsequently, the multi-modal collaborative driving and signal processing unit smoothly transitions the driving mode from multi-frequency collaborative to a single-frequency phase-locked loop tracking mode centered on the precise resonant frequency. At the moment the system starts up, the embedded real-time calculation and compensation unit first reads the initial temperature value of the high-precision temperature sensor; The digital signal processor calculates the expected frequency drift of the current temperature relative to the reference calibration temperature based on the pre-stored thermal expansion coefficient and elastic modulus temperature coefficient of the vibrating tube material. The digital signal processor then adds this drift to the resonant frequency recorded in the reference resonant spectrum, which serves as the frequency setting value for the initial multi-frequency sine wave signal generated by the direct digital frequency synthesizer.

2. The rapid density determination system for automotive urea according to claim 1, characterized in that, The embedded real-time computing and compensation unit includes a microcontroller, a high-precision temperature sensor, and a data module storing a complete compensation algorithm model. The compensation process of the embedded real-time computing and compensation unit is as follows: The high-precision temperature sensor measures the temperature of the urea solution vibrating against the pipe wall or in close contact with the pipe wall in real time; the microcontroller receives the temperature data and the initial density value; the data module pre-stores a compensation model established through extensive experimental calibration. This compensation model is a multidimensional nonlinear function mapping relationship with the initial density value and temperature as input and the corrected density value as output. This multidimensional nonlinear function mapping relationship is implemented through a trained three-layer feedforward neural network, whose hidden layer contains 8 neurons and the activation function uses a modified linear unit; the microcontroller executes the neural network model to compensate the initial density value in real time and outputs the final calibrated urea solution density value.

3. The rapid density determination system for automotive urea according to claim 2, characterized in that, The system control and communication interface unit receives a measurement request signal from the vehicle SCR controller, triggers and synchronizes the startup of the multimodal collaborative drive and signal processing unit and the embedded real-time calculation and compensation unit. After completing a measurement cycle, the system control and communication interface unit encapsulates the final density value, solution temperature value and system status word into a data frame conforming to the controller LAN or vehicle Ethernet protocol, and sends it to the SCR controller through the physical layer interface.

4. The rapid density determination system for automotive urea according to claim 3, characterized in that, In the single-frequency phase-locked loop tracking mode, the direct digital frequency synthesizer generates a single-frequency sinusoidal drive signal with the predicted accurate resonant frequency as the center frequency; The digital signal processor internally employs a high-bandwidth digital phase-locked loop algorithm to continuously compare the phase difference between the drive signal and the pickup signal, and dynamically adjusts the output frequency of the direct digital frequency synthesizer so that the phase difference is always locked near the 90-degree resonant point.

5. The rapid density determination system for automotive urea according to claim 4, characterized in that, The method for obtaining and updating the reference resonance spectrum is as follows: Before the sensor leaves the factory or during periodic maintenance, the vibration tube cavity is filled with a standard reference medium, and a high-precision full-spectrum scanning calibration program is run at a standard temperature; the full-spectrum scanning calibration program controls the direct digital frequency synthesizer to scan in small steps within a wide frequency band, while recording the amplitude of the vibration pickup signal, thereby accurately plotting the complete resonance curve of the vibration tube under the reference medium, identifying and recording the accurate resonant frequencies and quality factors of the first three modes, and storing them as non-volatile reference data.

6. The rapid density determination system for automotive urea according to claim 5, characterized in that, The system also includes an online self-diagnosis and early warning unit; the online self-diagnosis and early warning unit continuously monitors the signal quality indicators output by the multimodal collaborative driving and signal processing unit, including the signal-to-noise ratio, harmonic distortion, and stability of the driving current of each mode; at the same time, it monitors the rationality of the input and output of the neural network model in the embedded real-time calculation and compensation unit.

7. The rapid density determination system for automotive urea according to claim 6, characterized in that, The online self-diagnosis and early warning unit is also used to: determine that the sensor has a potential fault when the signal quality is continuously lower than a preset threshold, or the final density value after compensation exceeds the reasonable physical range of the urea solution; the online self-diagnosis and early warning unit then sends a specific fault code and early warning information to the SCR controller through the system control and communication interface unit.

Images