Intelligent control system and method for engine detonation and preignition

Abstract

The invention relates to the technical field of knock control, and discloses an intelligent control system and method for engine knock and preignition, and the method comprises the following steps: collecting engine knock signal data and preignition signal data, and carrying out the signal processing of the knock signal data and the preignition signal data, the method comprises the following steps: generating knock intensity data and preignition intensity data, and during engine knock and preignition identification, setting multi-stage identification thresholds including a basic threshold, a dynamic self-learning threshold and a temperature compensation threshold, and updating the thresholds in real time by utilizing a self-learning algorithm, so that the engine knock and preignition identification accuracy is improved. The method can effectively adapt to sensor signal characteristic change caused by engine wear, fuel quality difference and environment change, solves the problems of misjudgment and missed judgment caused by a fixed recognition threshold value, ensures the accuracy and reliability of knock and preignition recognition, reduces control errors, and improves the reliability of knock and preignition recognition. And the dynamic property, the economical efficiency and the driving smoothness of the engine are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of knock control technology, specifically to an intelligent control system and method for engine knock and pre-ignition. Background Technology

[0002] Knock control is an engine compression ratio control system that suppresses abnormal combustion by dynamically adjusting the ignition advance angle. The system uses dual knock sensors to monitor cylinder block vibration in real time and uses piezoelectric ceramic elements to convert mechanical vibration into electrical signals, which are then transmitted to the DME control unit for analysis and processing.

[0003] Currently, due to the complex and varied operating conditions of engines, the detection of knock and pre-ignition relies on a preset fixed threshold. This cannot adapt in real time to the baseline drift of sensor signals caused by engine wear, fuel quality differences, and environmental changes. When the signal drift is not effectively compensated, it will cause misjudgment and missed judgment, and the accuracy of control cannot be guaranteed.

[0004] Therefore, an intelligent control system and method for engine knock and pre-ignition are proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent control system and method for engine knock and pre-ignition, which solves the problem mentioned in the background art that when signal drift is not effectively compensated, it will cause misjudgment and missed judgment, thus failing to guarantee the accuracy of control.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent control system and method for engine knock and pre-ignition, the method comprising the following steps: S1. Collect engine knock signal data and pre-ignition signal data; S2. Perform signal processing on the detonation signal data and pre-ignition signal data to generate detonation intensity data and pre-ignition intensity data; S3. Based on the detonation intensity data and pre-ignition intensity data, perform detonation and pre-ignition identification processing to generate detonation identification results and pre-ignition identification results; S4. When the detonation identification result indicates that a detonation has occurred, the detonation control strategy is executed according to the detonation level, and a detonation control command is generated. S5. When the pre-ignition identification result indicates that pre-ignition has occurred, execute the pre-ignition control strategy according to the pre-ignition level and generate a pre-ignition control command; S6. Execute the knock control command and pre-ignition control command, adjust the engine operating parameters, perform diagnostic processing, and generate engine status diagnostic data. S7. Based on the engine status diagnostic data, perform adaptive optimization processing, update control parameters and thresholds, and generate optimized control commands.

[0007] Preferably, the acquisition of engine knock signal data and pre-ignition signal data in S1 includes the following steps: S11. Collect engine cylinder vibration signals through a knock sensor and generate knock signal data; S12. Collect engine speed data through a speed sensor, and collect engine intake air temperature data and coolant temperature data through a temperature sensor. S13. Integrate the knock signal data, speed data, and temperature data to output the engine signal acquisition dataset.

[0008] Preferably, the generation of detonation intensity data and pre-ignition intensity data in step S2 includes the following steps: S21. Filter and amplify the detonation signal data, calculate the detonation intensity value (virkr / IKCtl_facKnkInten_u8), and generate detonation intensity data; S22. Perform sliding window statistical processing on the pre-ignition signal data, calculate the pre-ignition intensity value (virkr_2 / IKCtl_facPreIgnInten_u8), and generate pre-ignition intensity data; S23. Combining the speed data and temperature data, perform temperature compensation and speed correction on the knock intensity data and pre-ignition intensity data, and output the corrected intensity data.

[0009] Preferably, the generation of the detonation control command in S4 includes the following steps: S41. Set multi-level knock recognition threshold (ZWAPPL), including basic threshold, dynamic self-learning threshold (wkradap[i] / IKCtl_agAdpn_as8_[i]) and temperature compensation threshold (%ZWWL); S42. Based on the comparison results between the detonation intensity data and the threshold, the detonation level is divided into mild detonation, moderate detonation, and severe detonation. S43. For mild knocking, implement a strategy of slightly delaying the ignition angle; for moderate knocking, implement a strategy of moderately delaying the ignition angle; for severe knocking, implement a strategy of significantly delaying the ignition angle and limiting the engine load. S44. After knocking is eliminated, a gradual recovery mechanism is used to gradually restore the ignition angle, and the recovery rate is dynamically adjusted according to the engine operating conditions.

[0010] Preferably, the generation of the pre-ignition control command in S5 includes the following steps: S51. Set a pre-ignition identification threshold independent of detonation, and use a sliding window statistical method to improve the accuracy of pre-ignition identification; S52. Based on the comparison results between the pre-ignition intensity data and the threshold, the pre-ignition level is divided into primary pre-ignition, intermediate pre-ignition and advanced pre-ignition. S53. For primary pre-ignition, implement a mixture enrichment control strategy; for intermediate pre-ignition, implement a valve overlap reduction control strategy; for advanced pre-ignition, implement a load limiting and fuel cut-off control strategy. S54. Set up a pre-ignition count protection mechanism. When the number of pre-ignitions exceeds the threshold, the engine protection mode is triggered to avoid cylinder damage.

[0011] Preferably, the generation of engine condition diagnostic data in step S6 includes the following steps: S61. The knock control command is executed by the ignition control unit to adjust the ignition angle, the pre-ignition control command is executed by the fuel injection control unit to adjust the fuel injection quantity, and the valve overlap angle is adjusted by the valve control unit. S62. Perform knock sensor diagnostic processing, including rationality diagnosis (DFC_KS1min) and linearity diagnosis (DFC_KnDetSens1PortAMax, etc.), and generate sensor diagnostic data; S63. Perform self-learning verification processing, periodically verify the rationality of the self-learning value (wkradap[i]), trigger relearning when an anomaly occurs, and output engine status diagnostic data.

[0012] Preferably, the generation optimization control instruction in S7 includes the following steps: S71. Based on the engine condition diagnostic data, update the knock detection threshold and pre-ignition detection threshold using a self-learning algorithm; S72. Dynamically adjust control parameters according to changes in engine operating conditions, including ignition angle recovery rate, mixture enrichment rate, and valve overlap angle reduction. S73. Under high-temperature conditions, enhance temperature compensation parameters to ensure the adaptability of knock and pre-ignition control. S74 outputs optimized control commands to optimize engine operating status in real time.

[0013] Preferred, including: The signal acquisition module uses the knock sensor unit to generate engine vibration signals, sets the speed measurement reference through the speed sensor unit, and outputs signals to acquire data through the temperature sensor unit. The signal processing module receives the signal acquisition data, calculates the detonation intensity using the detonation intensity calculation unit, calculates the pre-ignition intensity using the pre-ignition intensity calculation unit, and outputs the processed data using the data filtering unit. The control decision module receives the processed data, generates a knock recognition result through the knock recognition unit, generates a pre-ignition recognition result through the pre-ignition recognition unit, and outputs control commands through the strategy selection unit. The execution control module receives the control commands, adjusts the ignition angle through the ignition control unit, adjusts the fuel injection quantity through the fuel injection control unit, adjusts the valve parameters through the valve control unit, and outputs control data. The diagnostic module receives the signal acquisition data and processed data, performs sensor diagnosis through the rationality diagnosis unit, performs linearity checks through the linearity diagnosis unit, and outputs the diagnostic results through the diagnostic output unit. The adaptive optimization module receives the diagnostic results and control data, updates the control threshold through the self-learning unit, adjusts the control parameters using the operating condition adaptation unit, and outputs optimization instructions through the optimization output unit.

[0014] Preferably, the signal acquisition module further includes a signal calibration unit to perform online calibration of the knock sensor, speed sensor and temperature sensor to ensure data acquisition accuracy. The signal processing module further includes a real-time adaptation unit to dynamically adjust the sampling frequency and filtering parameters according to the engine speed.

[0015] Preferably, the control decision module further includes a multi-level strategy selection unit, which dynamically classifies the knock level and pre-ignition level based on knock intensity data and pre-ignition intensity data, and adaptively selects ignition angle retarding, mixture enrichment and valve control strategies according to the level results. The execution control module further includes a safety interlock unit, which monitors engine operating parameters in real time, and automatically locks control commands and triggers backup control mode when system abnormalities and sensor failures are detected to ensure safe engine operation.

[0016] Compared with the prior art, the present invention provides an intelligent control system and method for engine knock and pre-ignition, which has the following beneficial effects: 1. In this invention, when identifying engine knock and pre-ignition, a multi-level identification threshold including a basic threshold, a dynamic self-learning threshold, and a temperature compensation threshold is set, and the threshold is updated in real time using a self-learning algorithm. This effectively adapts to changes in sensor signal characteristics caused by engine wear, fuel quality differences, and environmental changes, solving the problem of misjudgment and missed judgment caused by fixed identification thresholds, ensuring the accuracy and reliability of knock and pre-ignition identification, and reducing control errors.

[0017] 2. In this invention, when performing knock and pre-ignition control, by setting up independent knock identification units, pre-ignition identification units, and multi-level strategy selection units, it is possible to adaptively select and coordinately execute ignition angle delay, mixture enrichment, and valve overlap angle adjustment control strategies based on the identification results of knock level and pre-ignition level. This solves the problem of lack of coordination and mutual interference between knock control and pre-ignition control, enabling optimal decision-making on control priority when multiple abnormal combustions occur, and ensuring the effectiveness and safety of control.

[0018] 3. In this invention, during the parameter recovery process after knock and pre-ignition suppression, a progressive recovery mechanism is adopted, and the recovery rate of ignition parameters is dynamically adjusted based on engine condition diagnostic data and changes in operating conditions. This solves the problem of engine performance fluctuation caused by a single and fixed recovery strategy, and achieves smooth and adaptive recovery of control parameters, effectively improving the engine's power, economy and driving smoothness. Attached Figure Description

[0019] Figure 1 This is a flowchart of an intelligent control method for engine knock and pre-ignition according to the present invention; Figure 2 This is a schematic diagram of the intelligent control system for engine knock and pre-ignition according to the present invention. Detailed Implementation

[0020] The technical solutions of 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.

[0021] Please see Figure 1 The intelligent control system and method for engine knock and pre-ignition include the following steps: S1. Collect engine knock signal data and pre-ignition signal data; S2. Perform signal processing on the detonation signal data and pre-ignition signal data to generate detonation intensity data and pre-ignition intensity data; S3. Based on the detonation intensity data and pre-ignition intensity data, perform detonation and pre-ignition identification processing to generate detonation identification results and pre-ignition identification results; S4. When the detonation identification result indicates that detonation has occurred, execute the detonation control strategy according to the detonation level and generate a detonation control command; S5. When the pre-ignition identification result indicates that pre-ignition has occurred, execute the pre-ignition control strategy according to the pre-ignition level and generate a pre-ignition control command; S6. Execute knock control command and pre-ignition control command, adjust engine operating parameters, perform diagnostic processing, and generate engine status diagnostic data. S7. Based on engine condition diagnostic data, perform adaptive optimization processing, update control parameters and thresholds, and generate optimized control commands; The acquisition of engine knock signal data and pre-ignition signal data in S1 includes the following steps: S11. Collect engine cylinder vibration signals through a knock sensor and generate knock signal data; S12. Collect engine speed data through a speed sensor, and collect engine intake air temperature data and coolant temperature data through a temperature sensor. S13. Integrate knock signal data, speed data, and temperature data to output engine signal acquisition dataset; The generation of knock intensity data and pre-ignition intensity data in S2 includes the following steps: S21. Filter and amplify the detonation signal data, calculate the detonation intensity value (virkr / IKCtl_facKnkInten_u8), and generate detonation intensity data. The specific process can be broken down as follows: Step 1: Signal Acquisition and Preprocessing This step is the foundation for all subsequent processing. The knock sensor is installed on the engine block and is essentially a piezoelectric accelerometer used to detect wideband vibration signals during engine operation. The sensor converts mechanical vibrations into raw analog electrical signals, which are first sent to the ECU's analog-to-digital converter (ADC) for sampling and conversion into discrete digital signal sequences for microprocessor processing. Before and after sampling, preliminary analog filtering is usually performed to eliminate high-frequency noise interference. Step 2: Digital Filtering and Frequency Band Extraction This is a crucial step in the filtering process, the purpose of which is to separate the characteristic frequency components specific to detonation from a broadband vibration signal containing various mechanical noises. Frequency band selection: When knock occurs, high-frequency oscillations within a specific frequency range are generated. Therefore, digital filtering will apply a bandpass filter to the knock characteristic frequency window calibrated for a specific engine model. Filtering execution: The system applies a digital bandpass filter algorithm to the digital signal sampled by the ADC. This filter attenuates low-frequency and high-frequency noise outside the characteristic frequency band, while amplifying and retaining the signal components within the characteristic frequency band. After this step, the output is a "purified" detonation characteristic frequency band vibration signal. Step 3: Signal Rectification and Integration To quantify the intensity of the knock, the filtered vibration signal needs to be converted into a quantifiable energy index. Signal rectification: The filtered signal is rectified and squared to convert all negative values ​​into positive values, resulting in a pulsating signal that is always positive. Signal integration: This is a conceptual step in "amplification" and energy calculation. The system integrates the rectified signal within a specific time window and calculates its root mean square (RMS) value. This integral value and RMS value directly reflect the total energy and average power of the detonation characteristic frequency band vibration signal within the detonation window, i.e., the intensity of the detonation. Step 4: Calculate the detonation intensity value (virkr / IKCtl_facKnkInten_u8) The raw energy value obtained through integration and RMS calculation is an absolute value. Its dimensions and magnitude will vary greatly depending on the sensor sensitivity and engine model. In order to make unified judgments and controls, it needs to be normalized into a standardized intensity value. Background noise benchmark: Even under non-knock conditions, the engine has background vibration. The system will continuously calculate and track the energy level of this background vibration as a benchmark. Strength calculation: The knock intensity value is usually calculated using the following logic: Intensity = (Signal energy within the detonation window - Background noise energy) / Background noise energy; Another similar ratio form, whose physical meaning is the degree to which the detonation energy exceeds the background noise energy, results in virkr / IKCtl_facKnkInten_u8 being a dimensionless and a standardized value with fixed dimensions, which is convenient for comparison with the preset detonation identification threshold. Step 5: Generate knock intensity data The calculated virkr / IKCtl_facKnkInten_u8 value is the final detonation intensity data, which is updated in real time and passed to subsequent functional modules for detonation identification and judgment. When virkr / IKCtl_facKnkInten_u8 is greater than the set knock threshold, it is determined that knock has occurred, and the ignition angle delay control action is triggered. This data will also be used for the knock self-learning function to adapt to long-term changes in engine condition; S22. Perform sliding window statistical processing on the pre-ignition signal data, calculate the pre-ignition intensity value (virkr_2 / IKCtl_facPreIgnInten_u8), and generate pre-ignition intensity data. The specific process is as follows: Step 1: Signal Acquisition and Preprocessing Similar to detonation signal processing, the system first acquires the original broadband vibration analog signal through the detonation sensor, and converts it into a digital signal through an analog-to-digital converter (ADC). Then, preliminary filtering is performed to remove obvious high-frequency noise. However, unlike detonation processing, a bandpass filter for a specific high frequency is not applied prematurely here, because the frequency mode of pre-ignition is broader and different from that of detonation. Step 2: Sliding window statistical processing This is the most distinctive aspect of pre-ignition detection. The system no longer relies on a fixed crankshaft angle window, but instead uses one or more "sliding windows" to continuously analyze the statistical characteristics of the vibration signal. Window definition: The algorithm sets one or more observation windows in the critical area before the top dead center of ignition. These windows "slide" as the engine crankshaft rotates, covering the latest signal segments in real time. Statistical calculation: Within each sliding window, the system does not directly calculate the energy, but instead calculates the statistical characteristic values ​​of the vibration signal. The most commonly used are the root mean square (RMS) and variance. The RMS value reflects the average power of the signal within the window and can effectively characterize the intensity of the vibration. By calculating the RMS value of each window, a series of statistical intensity sequences that change over time are obtained. Establishing a background baseline: The system will continuously learn the normal fluctuation range of RMS values ​​within these sliding windows under normal operating conditions without pre-ignition, forming a dynamic "background noise baseline". This baseline is not a fixed value, but a statistical distribution that varies with engine speed and load. Step 3: Calculate the pre-ignition intensity value (virkr_2 / IKCtl_facPreIgnInten_u8) The calculation of pre-ignition intensity value is essentially an anomaly detection process, that is, determining whether the current signal statistical characteristics deviate from the normal baseline; Deviation comparison: Once the RMS value of the vibration signal within a new sliding window is calculated, the system will compare it with the normal baseline value under the current operating conditions obtained through learning. Intensity quantification: The calculation logic of the pre-ignition intensity value virkr_2 / IKCtl_facPreIgnInten_u8 can be understood as the degree of deviation between the statistical characteristic value of the current window signal and the characteristic value of the background baseline. Specifically, it is a multiple of a standard deviation and the ratio of the relative deviation. Intensity = (Current window RMS value - Baseline RMS mean) / Standard deviation of baseline RMS; The larger this value, the higher the current vibration intensity, the smaller the normal fluctuation, and thus the more likely it is to be a premature combustion event; Trigger judgment: The key point is that this comparison and calculation process continues before the ignition moment. Once the calculated virkr_2 / IKCtl_facPreIgnInten_u8 value exceeds the preset pre-ignition detection threshold at some point before the spark plug ignition, the system will immediately determine that a pre-ignition event has occurred. Step 4: Generate pre-ignition intensity data The calculated virkr_2 / IKCtl_facPreIgnInten_u8 value is the pre-ignition intensity data. This data is transmitted to the control decision module in real time. Unlike knock control, the pre-ignition control strategy will be triggered immediately. At the same time, the system will start the pre-ignition counting protection mechanism. When the number of pre-ignitions exceeds a certain number in a short period of time, it will enter a higher level of protection mode to prevent engine damage. S23. Combining speed data and temperature data, perform temperature compensation and speed correction on the knock intensity data and pre-ignition intensity data, and output the corrected intensity data. The process of generating knock control commands in S4 includes the following steps: S41. Set multi-level knock recognition threshold (ZWAPPL), including basic threshold, dynamic self-learning threshold (wkradap[i] / IKCtl_agAdpn_as8_[i]) and temperature compensation threshold (%ZWWL); S42. Based on the comparison results between the detonation intensity data and the threshold, the detonation levels are divided into mild detonation, moderate detonation, and severe detonation. S43. For mild knocking, implement a strategy of slightly delaying the ignition angle; for moderate knocking, implement a strategy of moderately delaying the ignition angle; for severe knocking, implement a strategy of significantly delaying the ignition angle and limiting the engine load. S44. After knocking is eliminated, a gradual recovery mechanism is used to gradually restore the ignition angle, and the recovery rate is dynamically adjusted according to the engine operating conditions. The generation of pre-ignition control commands in S5 includes the following steps: S51. Set a pre-ignition detection threshold independent of detonation, and use a sliding window statistical method to improve the accuracy of pre-ignition detection. The specific implementation process is as follows: Step 1: Define the sliding window The system first defines an observation window in the time domain and crankshaft angle domain. Crucially, this window is set in a critical phase interval before the spark plug ignition point to capture abnormal vibrations that occur earlier than normal ignition. Window size: The width of the window is precisely calibrated. It needs to be long enough to contain statistically significant signal samples, and short enough to ensure real-time performance and rapid response to pre-ignition events. Sliding mechanism: The window is not fixed, but "slides" as the engine crankshaft rotates and time goes by. Each time a new data point is collected, the window moves forward by one point, discarding the oldest data point and incorporating the latest data point. This enables continuous and real-time monitoring of the signal. Step 2: Calculate the statistics within the window. Within each sliding window, the system does not simply look for the maximum amplitude, but instead calculates statistical features that characterize the signal energy distribution and intensity. The most commonly used statistic is the root mean square value. Calculating the RMS value of the signal within the window: The RMS value is calculated by first squaring each signal data point within the window, then calculating the average of these squared values, and finally taking the square root of the average. The formula can be simplified as follows: ; in It is the signal value within the window. It is the number of data points within the window; The significance of RMS value: The RMS value reflects the average power of the signal within the window. Compared with the peak value, the RMS value is less sensitive to transient and noise spikes and is more representative of the overall energy level of vibration over a period of time. This is more effective in identifying persistent signs of premature combustion. Step 3: Establish a dynamic statistical baseline and identify deviations This is the core of improving accuracy. Through long-term learning, the system establishes a "fingerprint" and baseline model of the RMS value within the window under normal operating conditions. Baseline learning: Under various operating conditions where pre-ignition is known to be absent, the system continuously records the RMS value calculated for each sliding window and statistically analyzes its distribution characteristics and moving average. and moving standard deviation This forms a statistical baseline that changes dynamically with the operating conditions; Deviation detection and pre-ignition intensity calculation: During real-time operation, the system calculates the measured RMS value of the current window. The pre-ignition intensity value is a quantitative representation of this statistical bias, compared to the learned statistical baseline under the current operating conditions. A common calculation method is: Pre-ignition intensity ; The physical meaning of this value is: how many standard deviations the energy level of the current signal is higher than normal. It is a standardized value that makes the judgment unaffected by changes in the absolute signal amplitude.

[0022] Step 4: Make intelligent judgments based on statistical probability When the calculated pre-ignition intensity value exceeds a preset threshold, the system does not immediately determine that pre-ignition has occurred, but makes a comprehensive judgment based on the timing and duration of its occurrence. Threshold decision: A high statistical threshold greatly reduces false alarms. Pre-ignition is only judged when the vibration energy is abnormal to the point that it does not occur under normal conditions. Improved accuracy: This method effectively transforms the pre-ignition identification problem from a simple "amplitude exceeding the threshold" judgment into a "low-probability event detection" problem. It takes into account the inherent statistical laws of engine vibration, thereby filtering out many isolated interference signals with high amplitude but not pre-ignition, improving the specificity and reliability of the identification. S52. Based on the comparison results between the pre-ignition intensity data and the threshold, the pre-ignition level is divided into primary pre-ignition, intermediate pre-ignition and advanced pre-ignition. S53. For primary pre-ignition, implement a mixture enrichment control strategy; for intermediate pre-ignition, implement a valve overlap reduction control strategy; for advanced pre-ignition, implement a load limiting and fuel cut-off control strategy. S54. Set up a pre-ignition count protection mechanism. When the number of pre-ignitions exceeds the threshold, the engine protection mode is triggered to avoid cylinder damage. Generating engine condition diagnostic data in S6 includes the following steps: S61. The ignition control unit executes the knock control command and adjusts the ignition angle; the fuel injection control unit executes the pre-ignition control command and adjusts the fuel injection quantity; and the valve control unit adjusts the valve overlap angle. S62. Perform knock sensor diagnostic processing, including rationality diagnosis (DFC_KS1min) and linearity diagnosis (DFC_KnDetSens1PortAMax, etc.), and generate sensor diagnostic data; S63. Perform self-learning verification processing, periodically verify the rationality of the self-learning value (wkradap[i]), trigger relearning when abnormal, and output engine status diagnostic data; The generation optimization control instructions in S7 include the following steps: S71. Based on engine condition diagnostic data, update the knock detection threshold and pre-ignition detection threshold through a self-learning algorithm; The self-learning update process of the detonation detection threshold: Step 1: Data Monitoring and Collection: The system continuously monitors key control parameters, the most critical of which is the mean ignition delay angle. (wkrmav / IKCtl_agKnkAvrgGen_s8) and knock self-learning value (wkradap[i] / IKCtl_agAdpn_as8_[i]), these data are stored and distinguished according to engine speed and load range ([i]); Step 2: Effect Evaluation and Analysis Evaluation criteria: The ideal control state is that the average ignition retarding angle is maintained within a small range. This indicates that the knock threshold is set just right, effectively controlling knock without losing too much power and fuel economy due to excessive ignition retarding. Problem identification: When the system finds that the wkrmav of a certain operating range ([i]) is consistently large, it means that the ECU needs to "frequently" and "significantly" delay the ignition angle to suppress knock. This suggests that the current knock detection threshold (ZWAPPL) is too sensitive to the current state of the engine, that is, the threshold is set too low, causing the system to intervene in the control too early. Step 3: Threshold Adjustment Learning logic: When the above-mentioned "over-control" situation is detected, the self-learning algorithm will be activated, and it will slowly and gradually increase the knock detection threshold (ZWAPPL) corresponding to the operating condition range. Update mechanism: This adjustment is usually not a one-time event, but is achieved by modifying the corresponding self-learning value wkradap[i]. wkradap[i] is a correction value that is added to the base threshold. The algorithm adjusts wkradap[i] towards a negative value, which is equivalent to raising the threshold for knock detection in actual use. The goal is to make wkrmav return to the ideal range of [0, -4.5] degrees after learning, while ensuring that there are no audible severe knocks. Step 4, Verification and Locking: The updated threshold will be applied to subsequent control. The system will continue to monitor the control effect under the new threshold to ensure smooth engine operation and stable effect. The learned value will be saved and used for a long time until the engine state changes. The self-learning update process of the premature combustion detection threshold: Step 1, Data Monitoring and Collection: The system continuously records the frequency and intensity (virkr_2) of pre-ignition events and the effectiveness of the control measures taken. At the same time, it also statistically analyzes the background distribution of pre-ignition intensity signals under normal operating conditions. Step 2: Effect Evaluation and Analysis Evaluation criteria: Ideally, pre-ignition events should occur very rarely, and when the system takes primary control measures, it should be able to effectively suppress subsequent pre-ignition. At the same time, under most non-pre-ignition operating conditions, the pre-ignition intensity signal should be far away from the trigger threshold, that is, there should be sufficient "safety margin". Problem identification: If the system frequently triggers pre-ignition control under specific operating conditions, but subsequent analysis reveals that no actual destructive pre-ignition has occurred, it indicates that the current pre-ignition threshold is too sensitive. Conversely, if pre-ignition occurs but the system fails to recognize it in time, requiring frequent activation of higher-level controls, it indicates that the threshold is not sensitive enough. Step 3: Threshold Adjustment Safety First Principle: The self-learning algorithm is very conservative in adjusting the pre-ignition threshold. In general, the decision to lower the threshold is made very cautiously, and will only be made when it is necessary and has been fully verified. Progressive optimization: For overly sensitive situations, the algorithm will slowly increase the premature ignition detection threshold to reduce false triggers. This process will be closely integrated with the "premature ignition count protection" mechanism to ensure that if a real premature ignition occurs during the adjustment process, the system can immediately revert to safe mode. Based on statistical learning: The algorithm learns the statistical distribution of the pre-ignition intensity signal (virkr_2) under normal operating conditions and dynamically adjusts the threshold accordingly to maintain a reasonable safe distance from the background noise. This distance will slowly adapt to the fuel quality and engine carbon buildup. S72. Dynamically adjust control parameters according to changes in engine operating conditions, including ignition angle recovery rate, mixture enrichment rate, and valve overlap angle reduction. S73. Under high-temperature conditions, enhance temperature compensation parameters to ensure the adaptability of knock and pre-ignition control. S74 outputs optimized control commands to optimize engine operating status in real time; include: The signal acquisition module uses the knock sensor unit to generate engine vibration signals, sets the speed measurement reference through the speed sensor unit, and outputs signals to acquire data through the temperature sensor unit. The signal processing module receives signal acquisition data, calculates the detonation intensity through the detonation intensity calculation unit, calculates the pre-ignition intensity through the pre-ignition intensity calculation unit, and outputs the processed data through the data filtering unit. The control decision module receives and processes data, generates a knock recognition result through the knock recognition unit, generates a pre-ignition recognition result through the pre-ignition recognition unit, and outputs control commands through the strategy selection unit. The execution control module receives control commands, adjusts the ignition angle through the ignition control unit, adjusts the fuel injection quantity through the fuel injection control unit, and adjusts the valve parameters through the valve control unit, and outputs control data. The diagnostic module receives and processes signal acquisition data, performs sensor diagnosis through the rationality diagnostic unit, performs linearity checks through the linearity diagnostic unit, and outputs diagnostic results through the diagnostic output unit. The adaptive optimization module receives diagnostic results and control data, updates control thresholds through a self-learning unit, adjusts control parameters using a working condition adaptation unit, and outputs optimization instructions through an optimization output unit. The signal acquisition module also includes a signal calibration unit to perform online calibration of the knock sensor, speed sensor and temperature sensor to ensure data acquisition accuracy. The signal processing module also includes a real-time adaptation unit to dynamically adjust the sampling frequency and filtering parameters according to the engine speed. The control decision module also includes a multi-level strategy selection unit, which dynamically classifies the knock level and pre-ignition level based on knock intensity data and pre-ignition intensity data, and adaptively selects ignition angle retarding, mixture enrichment and valve control strategies according to the level results. The execution control module also includes a safety interlock unit, which monitors engine operating parameters in real time. When system abnormalities and sensor failures are detected, it automatically locks the control commands and triggers the backup control mode to ensure safe engine operation.

[0023] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0024] 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 method for intelligent control of engine knock and pre-ignition, characterized in that, The method includes the following steps: S1. Collect engine knock signal data and pre-ignition signal data; S2. Perform signal processing on the detonation signal data and pre-ignition signal data to generate detonation intensity data and pre-ignition intensity data; S3. Based on the detonation intensity data and pre-ignition intensity data, perform detonation and pre-ignition identification processing to generate detonation identification results and pre-ignition identification results; S4. When the detonation identification result indicates that a detonation has occurred, the detonation control strategy is executed according to the detonation level, and a detonation control command is generated. S5. When the pre-ignition identification result indicates that pre-ignition has occurred, execute the pre-ignition control strategy according to the pre-ignition level and generate a pre-ignition control command; S6. Execute the knock control command and pre-ignition control command, adjust the engine operating parameters, perform diagnostic processing, and generate engine status diagnostic data. S7. Based on the engine status diagnostic data, perform adaptive optimization processing, update control parameters and thresholds, and generate optimized control commands.

2. The intelligent control method for engine knock and pre-ignition according to claim 1, characterized in that, The acquisition of engine knock signal data and pre-ignition signal data in S1 includes the following steps: S11. Collect engine cylinder vibration signals through a knock sensor and generate knock signal data; S12. Collect engine speed data through a speed sensor, and collect engine intake air temperature data and coolant temperature data through a temperature sensor. S13. Integrate the knock signal data, speed data, and temperature data to output the engine signal acquisition dataset.

3. The intelligent control system and method for engine knock and pre-ignition according to claim 1, characterized in that, The generation of knock intensity data and pre-ignition intensity data in S2 includes the following steps: S21. Filter and amplify the detonation signal data, calculate the detonation intensity value (virkr / IKCtl_facKnkInten_u8), and generate detonation intensity data; S22. Perform sliding window statistical processing on the pre-ignition signal data, calculate the pre-ignition intensity value (virkr_2 / IKCtl_facPreIgnInten_u8), and generate pre-ignition intensity data; S23. Combining the speed data and temperature data, perform temperature compensation and speed correction on the knock intensity data and pre-ignition intensity data, and output the corrected intensity data.

4. The intelligent control system and method for engine knock and pre-ignition according to claim 1, characterized in that, The generation of the detonation control command in S4 includes the following steps: S41. Set multi-level knock recognition threshold (ZWAPPL), including basic threshold, dynamic self-learning threshold (wkradap[i] / IKCtl_agAdpn_as8_[i]) and temperature compensation threshold (%ZWWL); S42. Based on the comparison results between the detonation intensity data and the threshold, the detonation level is divided into mild detonation, moderate detonation, and severe detonation. S43. For mild knocking, implement a strategy of slightly delaying the ignition angle; for moderate knocking, implement a strategy of moderately delaying the ignition angle; for severe knocking, implement a strategy of significantly delaying the ignition angle and limiting the engine load. S44. After knocking is eliminated, a gradual recovery mechanism is used to gradually restore the ignition angle, and the recovery rate is dynamically adjusted according to the engine operating conditions.

5. The intelligent control method for engine knock and pre-ignition according to claim 1, characterized in that, The generation of the pre-ignition control command in S5 includes the following steps: S51. Set a pre-ignition identification threshold independent of detonation, and use a sliding window statistical method to improve the accuracy of pre-ignition identification; S52. Based on the comparison results between the pre-ignition intensity data and the threshold, the pre-ignition level is divided into primary pre-ignition, intermediate pre-ignition and advanced pre-ignition. S53. For primary pre-ignition, implement a mixture enrichment control strategy; for intermediate pre-ignition, implement a valve overlap reduction control strategy; for advanced pre-ignition, implement a load limiting and fuel cut-off control strategy. S54. Set up a pre-ignition count protection mechanism. When the number of pre-ignitions exceeds the threshold, the engine protection mode is triggered to avoid cylinder damage.

6. The intelligent control method for engine knock and pre-ignition according to claim 1, characterized in that, The generation of engine condition diagnostic data in S6 includes the following steps: S61. The knock control command is executed by the ignition control unit to adjust the ignition angle, the pre-ignition control command is executed by the fuel injection control unit to adjust the fuel injection quantity, and the valve overlap angle is adjusted by the valve control unit. S62. Perform knock sensor diagnostic processing, including rationality diagnosis (DFC_KS1min) and linearity diagnosis (DFC_KnDetSens1PortAMax, etc.), and generate sensor diagnostic data; S63. Perform self-learning verification processing, periodically verify the rationality of the self-learning value (wkradap[i]), trigger relearning when an anomaly occurs, and output engine status diagnostic data.

7. The intelligent control method for engine knock and pre-ignition according to claim 1, characterized in that, The generation optimization control instructions in S7 include the following steps: S71. Based on the engine condition diagnostic data, update the knock detection threshold and pre-ignition detection threshold using a self-learning algorithm; S72. Dynamically adjust control parameters according to changes in engine operating conditions, including ignition angle recovery rate, mixture enrichment rate, and valve overlap angle reduction. S73. Under high-temperature conditions, enhance temperature compensation parameters to ensure the adaptability of knock and pre-ignition control. S74 outputs optimized control commands to optimize engine operating status in real time.

8. An intelligent control system for engine knock and pre-ignition, used to implement the intelligent control method for engine knock and pre-ignition as described in any one of claims 1-7, characterized in that, include: The signal acquisition module uses the knock sensor unit to generate engine vibration signals, sets the speed measurement reference through the speed sensor unit, and outputs signals to acquire data through the temperature sensor unit. The signal processing module receives the signal acquisition data, calculates the detonation intensity using the detonation intensity calculation unit, calculates the pre-ignition intensity using the pre-ignition intensity calculation unit, and outputs the processed data using the data filtering unit. The control decision module receives the processed data, generates a knock recognition result through the knock recognition unit, generates a pre-ignition recognition result through the pre-ignition recognition unit, and outputs control commands through the strategy selection unit. The execution control module receives the control commands, adjusts the ignition angle through the ignition control unit, adjusts the fuel injection quantity through the fuel injection control unit, adjusts the valve parameters through the valve control unit, and outputs control data. The diagnostic module receives the signal acquisition data and processed data, performs sensor diagnosis through the rationality diagnosis unit, performs linearity checks through the linearity diagnosis unit, and outputs the diagnostic results through the diagnostic output unit. The adaptive optimization module receives the diagnostic results and control data, updates the control threshold through the self-learning unit, adjusts the control parameters using the operating condition adaptation unit, and outputs optimization instructions through the optimization output unit.

9. The intelligent control system for engine knock and pre-ignition according to claim 8, characterized in that: The signal acquisition module also includes a signal calibration unit to perform online calibration of the knock sensor, speed sensor and temperature sensor to ensure data acquisition accuracy. The signal processing module also includes a real-time adaptation unit to dynamically adjust the sampling frequency and filtering parameters according to the engine speed.

10. The intelligent control system for engine knock and pre-ignition according to claim 8, characterized in that: The control decision module also includes a multi-level strategy selection unit, which dynamically classifies the knock level and pre-ignition level based on knock intensity data and pre-ignition intensity data, and adaptively selects ignition angle retarding, mixture enrichment and valve control strategies according to the level results. The execution control module also includes a safety interlock unit, which monitors engine operating parameters in real time. When system abnormalities and sensor failures are detected, it automatically locks the control commands and triggers the backup control mode to ensure safe engine operation.

Images