A pressure-controlled stable fuel injector testing system

Abstract

The application discloses a pressure control stable fuel injector detection system, and relates to the technical field of pressure control.The system comprises a prediction module, a compensation module and a comparison module.The prediction module collects current and voltage waveform data of a driving fuel injector electromagnetic valve, and constructs a pressure fluctuation prediction model through a machine learning model based on driving fuel injector driving signals and historical pressure fluctuation data, so as to predict the pressure disturbance amplitude and waveform generated in the track by the next injection action.The compensation module generates a reverse compensation instruction through a feedforward-feedback compound controller according to the pressure disturbance amplitude and waveform, and a feedback path processes the residual error that is not offset by the feedforward model through a PID controller.The comparison module is based on a stable pressure environment.The application realizes high-precision and visual positioning and diagnosis of fault types, positions and causes, and thus realizes the ability transition from passive standard detection to active intelligent deep diagnosis in essence.

Description

Technical Field

[0001] This invention relates to the field of pressure control technology, and in particular to a fuel injector detection system with stable pressure control. Background Technology

[0002] In the research and testing of high-pressure common rail fuel injection systems, the injector, as a core precision component, requires precise performance evaluation. Current advanced testing systems typically employ dynamic test benches. These benches apply standard drive signals to the injector and simultaneously acquire multi-dimensional dynamic signals, including its electromagnetic response, mechanical response, and spray pattern. These signals are then compared with standard models or parameters to determine its operating status and any malfunctions. This "excitation-response" testing paradigm forms the mainstream foundation for performance testing and diagnosis in the industry.

[0003] As fuel injectors evolve towards higher pressure and faster speeds, existing testing technologies face a core challenge: maintaining an absolutely stable pressure testing environment at every injection moment. Each injection draws fuel from the common rail, inevitably causing high-frequency transient pressure fluctuations in the pipeline. These disturbances, induced by the actions of the test object itself, directly affect the initial pressure conditions of the next injection test. This interference persists and accumulates, especially during continuous testing or multi-condition scanning. Traditional pressure control methods primarily target steady-state or low-frequency disturbances, and their suppression effectiveness and response speed are limited for rapid pressure fluctuations with specific waveform characteristics directly caused by injection events. This results in a non-constant pressure baseline in the testing environment at the microscopic level, weakening the fundamental reliability of subsequent high-precision measurements and analyses of transient processes such as needle valve movement and spray development. Consequently, this limits the ability of deep fault diagnosis models to accurately identify and locate microscopic fault characteristics in fuel injectors. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a pressure-controlled and stable injector detection system to solve the problem that existing detection technologies cannot effectively compensate for the high-frequency transient pressure disturbances caused by the injection action itself, resulting in unstable pressure reference in the test environment, which in turn restricts the accuracy of subsequent high-precision measurement and in-depth fault diagnosis.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a pressure-controlled and stable injector detection system, which includes a prediction module that collects current and voltage waveform data of the solenoid valve driving the injector, and constructs a pressure fluctuation prediction model based on the driving signal of the injector and historical pressure fluctuation data, and predicts the pressure disturbance amplitude and waveform that the next injection action will generate in the track. The compensation module generates a reverse compensation command based on the amplitude and waveform of the pressure disturbance through a feedforward-feedback composite controller. The feedback path processes the remaining error that was not canceled by the feedforward model through a PID controller. The comparison module, based on a stable pressure environment, initiates an adaptive test process, comparing the real-time collected current waveform and needle valve lift data with the pre-stored digital twin model to generate a deep diagnostic test sequence. The analysis module, during the execution of the deep diagnostic test sequence, monitors the actual lift curve of the injector needle valve and the transient process of fuel spray in real time through a non-contact needle valve lift sensor and a high-speed camera and image analysis module, and analyzes the spray cone angle, penetration distance and atomization uniformity through AI image algorithms. The positioning module uses a pattern recognition algorithm to compare the measured data with the digital twin model in real time, accurately locate the fault type, and generate a visual diagnostic report of the fault location and cause.

[0007] As a preferred embodiment of the pressure control and stable injector detection system of the present invention, the pressure disturbance feedforward compensation subsystem is configured as follows: Real-time current and voltage waveform data of the solenoid valve driving the fuel injector are collected, and combined with the fuel injector drive signal and historical common rail pressure fluctuation data, a pressure fluctuation prediction model is constructed through a pre-trained machine learning model to predict the amplitude and waveform of the pressure disturbance that will be generated in the common rail during the next injection action.

[0008] As a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the pressure disturbance feedforward compensation subsystem further includes a feedforward-feedback composite controller, which is configured as follows: The system receives the predicted disturbance amplitude and waveform output by the pressure fluctuation prediction model, and generates a reverse pressure compensation command accordingly. The remaining common rail pressure error after feedforward compensation is processed by a PID feedback controller to achieve high-precision and stable control of the common rail pressure.

[0009] In a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the adaptive depth diagnostic trigger module is configured as follows: Under the stable pressure environment maintained by the feedforward-feedback composite controller, the adaptive test process is initiated, and the expected behavior of the pre-stored digital twin model of the injector is compared in real time with the real-time collected injector drive current waveform, needle valve lift data and the expected behavior of the injector.

[0010] As a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the adaptive depth diagnostic trigger module is further configured as follows: Based on the results of the real-time comparison, a deep diagnostic test sequence is dynamically generated and executed, which includes a series of special driving conditions designed to stimulate and expose potential fault characteristics.

[0011] In a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the multi-dimensional sensing and monitoring module is configured as follows: During the execution of the deep diagnostic test sequence, the actual movement lift curve of the needle valve is simultaneously acquired through a non-contact needle valve lift sensor, and the transient development process of fuel spray is captured through a high-speed camera and image analysis module.

[0012] As a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the multi-dimensional sensing and monitoring module integrates an AI image analysis unit, which is configured as follows: The captured high-speed spray images are processed, and AI image algorithms are used to automatically analyze and quantify key characteristic parameters of the spray, including spray cone angle, spray penetration distance, and atomization uniformity.

[0013] In a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the intelligent fault diagnosis and location module is configured as follows: The system receives measured data streams from the multi-dimensional sensing and monitoring module, including needle valve lift curves, current waveforms, and spray characteristic parameters. It then uses a pattern recognition algorithm to compare and match these measured data with simulation data from the digital twin model under different fault modes in real time.

[0014] In a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the intelligent fault diagnosis and location module is further configured as follows: Based on the comparison and matching results, the fault type of the fuel injector is accurately located, and associated with the specific fault location and possible cause.

[0015] In a preferred embodiment of the pressure-controlled stable injector detection system of the present invention, the visualization report generation module is configured as follows: The system receives the output from the intelligent fault diagnosis and location module and automatically generates a visual diagnostic report that includes fault type, location information, cause analysis, and comparison charts of measured and model data.

[0016] The beneficial effects of this invention are as follows: By predicting transient pressure disturbances caused by injection through a machine learning model and combining this with feedforward-feedback composite control for active real-time compensation, the fundamental problem of unstable pressure reference caused by injection self-interference is solved, providing a clean and reproducible testing environment for subsequent high-precision measurements. Based on this, the system dynamically compares the real-time acquired preliminary response data with a digital twin model, adaptively generating a deep diagnostic test sequence capable of deeply eliciting fault characteristics. During this process, multi-dimensional data such as the injector's internal characteristic motion parameters and external characteristic spray pattern are simultaneously acquired. Finally, using a pattern recognition algorithm and matching with the twin model library, high-precision, visualized localization and diagnosis of fault type, location, and cause are achieved, thus fundamentally realizing a leap from passive standard detection to proactive intelligent deep diagnosis capabilities. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic diagram of an injector testing system for stable pressure control. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0021] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0022] Reference Figure 1As one embodiment of the present invention, this embodiment provides a pressure-controlled stable injector detection system, comprising the following steps: Real-time current and voltage waveform data of the solenoid valve driving the fuel injector are collected, and combined with the fuel injector drive signal and historical common rail pressure fluctuation data, a pressure fluctuation prediction model is constructed through a pre-trained machine learning model to predict the amplitude and waveform of the pressure disturbance that will be generated in the common rail during the next injection action.

[0023] Furthermore, a high-speed data acquisition system synchronously captures the real-time current and voltage dynamic waveforms during the injector solenoid valve actuation process. This electrical response signal is then time-domain aligned and feature-fused with the given injector actuation command timing and historical common rail pressure fluctuation data streams recorded by sensors. Using a pre-trained machine learning model (e.g., an LSTM-based time series prediction model or an ensemble learning model), deep feature extraction and nonlinear relationship modeling are performed on the aforementioned multi-source time series data, thereby constructing a pressure fluctuation prediction model that accurately reflects the dynamic causal relationship between "drive-pressure disturbance". This model's function is to proactively predict the specific amplitude characteristics, spectral components, and phase information of the pressure disturbance waveform that will be induced in the common rail system before the injector actually executes the next injection action, providing accurate input for feedforward compensation.

[0024] The system receives the predicted disturbance amplitude and waveform output by the pressure fluctuation prediction model, and generates a reverse pressure compensation command accordingly. The remaining common rail pressure error after feedforward compensation is processed by a PID feedback controller to achieve high-precision and stable control of the common rail pressure.

[0025] Furthermore, the system receives the predicted pressure disturbance value with specific waveform details from the aforementioned prediction model. Based on this predicted waveform, the feedforward control channel generates a reverse pressure regulation command with equal amplitude but opposite phase in real time, which is directly applied to the common rail pressure regulation actuator. This aims to proactively and precisely offset injection disturbances before they occur. Simultaneously, the traditional PID feedback control channel operates in parallel, monitoring the common rail pressure in real time through a high-precision pressure sensor. It specifically handles residual errors that the feedforward model cannot completely eliminate, unmodeled system dynamics, and environmental disturbances. The output commands from the feedforward and feedback channels are superimposed to jointly drive the actuator, achieving high-precision, highly dynamic, and stable control of the common rail pressure. Its core purpose is to create a steady-state pressure testing benchmark environment unaffected by the injection action itself.

[0026] Under the stable pressure environment maintained by the feedforward-feedback composite controller, the adaptive test process is initiated, and the expected behavior of the pre-stored digital twin model of the injector is compared in real time with the real-time collected injector drive current waveform, needle valve lift data and the expected behavior of the injector.

[0027] Furthermore, under the ultra-stable pressure environment maintained by the composite controller, the system initiates an adaptive diagnostic process. In the initial stage, the system applies standard test conditions to the injector, simultaneously acquiring its drive current response waveform and needle valve non-contact lift curve in real time. These measured dynamic response data are input in real time into a pre-stored high-fidelity digital twin model of the injector. This digital twin model is a high-precision simulation model integrating electromagnetic, mechanical, and hydraulic multi-physics fields, capable of simulating the dynamic behavior of the injector under ideal, fault-free conditions. By comparing the measured data with the theoretical expected output of the twin model in real time, the system can quickly and sensitively detect whether there are deviations exceeding preset tolerances at characteristic points (such as current peaks, needle valve opening delays, and seat oscillations), thereby triggering a deep diagnostic process.

[0028] Based on the results of the real-time comparison, a deep diagnostic test sequence is dynamically generated and executed, which includes a series of special driving conditions designed to stimulate and expose potential fault characteristics.

[0029] Furthermore, when the initial comparison detects behavioral deviations, the system does not simply report an anomaly but enters an intelligent analysis state. Based on the deviation pattern characteristics identified in the initial comparison, the system calls upon its built-in diagnostic strategy library. This strategy library associates various suspected fault modes with the driving condition combinations that most effectively expose the fault characteristics. The system dynamically combines these to generate a customized deep diagnostic test sequence. This sequence contains a series of specially designed driving signals, such as multi-pulse excitation, variable pulse width scanning, and pressure boundary testing. Its purpose is to proactively and systematically stimulate potential or minor fault characteristics in the tested injector, making them more apparent in subsequent responses, thereby creating conditions for accurate fault location.

[0030] During the execution of the deep diagnostic test sequence, the actual movement lift curve of the needle valve is simultaneously acquired through a non-contact needle valve lift sensor, and the transient development process of fuel spray is captured through a high-speed camera and image analysis module.

[0031] Furthermore, the system initiates high-precision multi-physics data synchronous acquisition. On one hand, using laser or capacitive non-contact displacement sensors, the actual motion lift curve of the injector needle valve throughout the entire diagnostic sequence is captured in real time at an extremely high sampling rate, accurately measuring its opening and closing transient processes, response speed, and vibration characteristics. On the other hand, within a controlled optical environment, an ultra-high-speed camera is used to synchronously capture the transient development process after fuel injection nozzle exit, recording the complete morphological evolution video of the spray. The purpose of this step is to simultaneously acquire complete dynamic response data across two dimensions: the internal mechanical motion of the injector and the external spray performance, under conditions where fault characteristics are actively stimulated.

[0032] The captured high-speed spray images are processed, and AI image algorithms are used to automatically analyze and quantify key characteristic parameters of the spray, including spray cone angle, spray penetration distance, and atomization uniformity.

[0033] Furthermore, advanced AI image processing algorithms are employed for analysis. The algorithm first performs denoising, enhancement, and segmentation on the images, accurately extracting the spray contours from each frame. Subsequently, through machine learning models or deep neural networks, it automatically identifies and quantifies key macroscopic characteristic parameters of the spray, including the spray cone angle (reflecting diffusion characteristics), spray penetration distance (inversely reflecting path characteristics), and the time-varying curves of atomization uniformity / particle size distribution (reflecting breakup quality). This achieves an objective, quantitative, and automated evaluation of the injector's external operating performance, transforming subjective visual observation into precisely comparable data indicators.

[0034] The system receives measured data streams from the multi-dimensional sensing and monitoring module, including needle valve lift curves, current waveforms, and spray characteristic parameters. It then uses a pattern recognition algorithm to compare and match these measured data with simulation data from the digital twin model under different fault modes in real time. Furthermore, data streams from a multi-dimensional sensing system are received, including needle valve lift curves and drive current waveforms from in-depth diagnostic testing, as well as time-series data of spray characteristics obtained from AI visual analysis. These multi-source, heterogeneous time-series data are synchronously aligned and feature extracted to form a multi-dimensional "feature vector" describing the state of the tested injector. This feature vector is input into a pattern recognition system and compared and matched in real time with a "fault feature database" generated by the injector's digital twin model under various preset typical fault modes (such as spring fatigue, sealing surface wear, valve core jamming, and nozzle blockage).

[0035] Based on the comparison and matching results, the fault type of the fuel injector is accurately located, and associated with the specific fault location and possible cause.

[0036] Furthermore, based on the pattern matching results, the system no longer simply determines "whether there is a fault," but performs precise fault type localization and root cause correlation. The system identifies one or more fault modes with the highest matching degree to the currently measured multi-dimensional feature vector. Then, combining the injector's structural principles and failure mechanism knowledge base, the system associates the matched fault modes with specific physical components (such as electromagnets, armatures, pressure regulating springs, control valves, nozzles, etc.) and possible failure causes (such as wear, fatigue, contamination, assembly errors, etc.), and estimates the severity of the fault. This makes the diagnostic results have clear engineering guidance significance.

[0037] Receives the output from the intelligent fault diagnosis and location module, and automatically generates a visual diagnostic report that includes fault type, location information, cause analysis, and comparison charts of measured and model data. Furthermore, the system automatically integrates the analysis results, intermediate data, and comparison charts from all the above steps. The report generation module retrieves the original test conditions, measured curves, model simulation curves, feature comparison tables, matching results, and root cause analysis text from the database, and automatically synthesizes a complete, visually appealing diagnostic report according to a preset professional template. The report intuitively displays the fault phenomena, diagnostic basis, location results, and maintenance suggestions, realizing the automated conversion from raw data to decision-making information, greatly improving the delivery efficiency and readability of diagnostic results, and facilitating engineers' quick understanding and decision-making.

[0038] In summary, this invention uses a machine learning model to predict transient pressure disturbances caused by injection and combines this with feedforward-feedback composite control for proactive real-time compensation. This solves the fundamental problem of unstable pressure reference caused by injection self-interference, providing a clean and reproducible testing environment for subsequent high-precision measurements. Based on this, the system dynamically compares the real-time acquired preliminary response data with a digital twin model, adaptively generating a deep diagnostic test sequence capable of deeply eliciting fault characteristics. During this process, it simultaneously acquires multi-dimensional data such as the injector's internal characteristic motion parameters and external characteristic spray pattern. Finally, it uses a pattern recognition algorithm to match and compare with the twin model library, achieving high-precision, visualized localization and diagnosis of fault type, location, and cause. This fundamentally realizes a leap from passive standard detection to proactive intelligent deep diagnosis capabilities.

[0039] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A pressure-controlled and stable injector detection system, characterized in that: include, The prediction module collects current and voltage waveform data of the solenoid valve that drives the fuel injector. Based on the driving signal of the fuel injector and historical pressure fluctuation data, it constructs a pressure fluctuation prediction model through machine learning to predict the amplitude and waveform of the pressure disturbance that will be generated in the rail during the next injection action. The compensation module generates a reverse compensation command based on the amplitude and waveform of the pressure disturbance through a feedforward-feedback composite controller. The feedback path processes the remaining error that was not canceled by the feedforward model through a PID controller. The comparison module, based on a stable pressure environment, initiates an adaptive test process, comparing the real-time collected current waveform and needle valve lift data with the pre-stored digital twin model to generate a deep diagnostic test sequence. The analysis module, during the execution of the deep diagnostic test sequence, monitors the actual lift curve of the injector needle valve and the transient process of fuel spray in real time through a non-contact needle valve lift sensor and a high-speed camera and image analysis module, and analyzes the spray cone angle, penetration distance and atomization uniformity through AI image algorithms. The positioning module uses a pattern recognition algorithm to compare the measured data with the digital twin model in real time, accurately locate the fault type, and generate a visual diagnostic report of the fault location and cause.

2. The pressure-controlled stable injector detection system as described in claim 1, characterized in that: Also includes: The pressure disturbance feedforward compensation subsystem is configured as follows: Real-time current and voltage waveform data of the solenoid valve driving the fuel injector are collected, and combined with the fuel injector drive signal and historical common rail pressure fluctuation data, a pressure fluctuation prediction model is constructed through a pre-trained machine learning model to predict the amplitude and waveform of the pressure disturbance that will be generated in the common rail during the next injection action.

3. The pressure-controlled stable injector detection system as described in claim 2, characterized in that: The pressure disturbance feedforward compensation subsystem further includes: a feedforward-feedback composite controller, which is configured as follows: The system receives the predicted disturbance amplitude and waveform output by the pressure fluctuation prediction model, and generates a reverse pressure compensation command accordingly. The remaining common rail pressure error after feedforward compensation is processed by a PID feedback controller to achieve high-precision and stable control of the common rail pressure.

4. The pressure-controlled stable injector detection system as described in claim 3, characterized in that: The adaptive depth diagnostic trigger module is configured as follows: Under the stable pressure environment maintained by the feedforward-feedback composite controller, the adaptive test process is initiated, and the expected behavior of the pre-stored digital twin model of the injector is compared in real time with the real-time collected injector drive current waveform, needle valve lift data and the expected behavior of the injector.

5. The pressure-controlled stable injector detection system as described in claim 4, characterized in that: The adaptive depth diagnostic triggering module is further configured as follows: Based on the results of the real-time comparison, a deep diagnostic test sequence is dynamically generated and executed, which includes a series of special driving conditions designed to stimulate and expose potential fault characteristics.

6. The pressure-controlled stable injector detection system as described in claim 5, characterized in that: The multidimensional sensing and monitoring module is configured as follows: During the execution of the deep diagnostic test sequence, the actual movement lift curve of the needle valve is simultaneously acquired through a non-contact needle valve lift sensor, and the transient development process of fuel spray is captured through a high-speed camera and image analysis module.

7. The pressure-controlled stable injector detection system as described in claim 6, characterized in that: The multidimensional sensing and monitoring module integrates an AI image analysis unit, which is configured as follows: The captured high-speed spray images are processed, and AI image algorithms are used to automatically analyze and quantify key characteristic parameters of the spray, including spray cone angle, spray penetration distance, and atomization uniformity.

8. The pressure-controlled stable injector detection system as described in claim 7, characterized in that: The intelligent fault diagnosis and location module is configured as follows: The system receives measured data streams from the multi-dimensional sensing and monitoring module, including needle valve lift curves, current waveforms, and spray characteristic parameters. It then uses a pattern recognition algorithm to compare and match these measured data with simulation data from the digital twin model under different fault modes in real time.

9. The pressure-controlled stable injector detection system as described in claim 8, characterized in that: The intelligent fault diagnosis and location module is further configured as follows: Based on the comparison and matching results, the fault type of the fuel injector is accurately located and associated with the specific fault location and possible cause.

10. The pressure-controlled stable injector detection system as described in claim 9, characterized in that: The visualization report generation module is configured as follows: The system receives the output from the intelligent fault diagnosis and location module and automatically generates a visual diagnostic report that includes fault type, location information, cause analysis, and comparison charts of measured and model data.

Images