NL2029619B1 - Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current - Google Patents

Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current Download PDF

Info

Publication number
NL2029619B1
NL2029619B1 NL2029619A NL2029619A NL2029619B1 NL 2029619 B1 NL2029619 B1 NL 2029619B1 NL 2029619 A NL2029619 A NL 2029619A NL 2029619 A NL2029619 A NL 2029619A NL 2029619 B1 NL2029619 B1 NL 2029619B1
Authority
NL
Netherlands
Prior art keywords
ion current
early combustion
engine
neural network
early
Prior art date
Application number
NL2029619A
Other languages
Dutch (nl)
Inventor
Wang Jinqiu
Ding Weiqi
Deng Jun
Li Liguang
Hu Zongjie
Original Assignee
Univ Tongji
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Univ Tongji filed Critical Univ Tongji
Priority to NL2029619A priority Critical patent/NL2029619B1/en
Application granted granted Critical
Publication of NL2029619B1 publication Critical patent/NL2029619B1/en

Links

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02P—IGNITION, OTHER THAN COMPRESSION IGNITION, FOR INTERNAL-COMBUSTION ENGINES; TESTING OF IGNITION TIMING IN COMPRESSION-IGNITION ENGINES
    • F02P17/00—Testing of ignition installations, e.g. in combination with adjusting; Testing of ignition timing in compression-ignition engines
    • F02P17/12—Testing characteristics of the spark, ignition voltage or current
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02P—IGNITION, OTHER THAN COMPRESSION IGNITION, FOR INTERNAL-COMBUSTION ENGINES; TESTING OF IGNITION TIMING IN COMPRESSION-IGNITION ENGINES
    • F02P5/00—Advancing or retarding ignition; Control therefor
    • F02P5/04—Advancing or retarding ignition; Control therefor automatically, as a function of the working conditions of the engine or vehicle or of the atmospheric conditions
    • F02P5/145—Advancing or retarding ignition; Control therefor automatically, as a function of the working conditions of the engine or vehicle or of the atmospheric conditions using electrical means
    • F02P5/15—Digital data processing
    • F02P5/152—Digital data processing dependent on pinking
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02P—IGNITION, OTHER THAN COMPRESSION IGNITION, FOR INTERNAL-COMBUSTION ENGINES; TESTING OF IGNITION TIMING IN COMPRESSION-IGNITION ENGINES
    • F02P17/00—Testing of ignition installations, e.g. in combination with adjusting; Testing of ignition timing in compression-ignition engines
    • F02P17/12—Testing characteristics of the spark, ignition voltage or current
    • F02P2017/125—Measuring ionisation of combustion gas, e.g. by using ignition circuits
    • F02P2017/128—Measuring ionisation of combustion gas, e.g. by using ignition circuits for knock detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • General Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)

Abstract

The present invention relates to a method for detecting and distinguishing early combustion and carbon deposition in an engine based on ion current, comprising the steps of: Sl: collecting an ion current signal and a cylinder pressure signal under various engine operating conditions; SZ: screening carbon deposition cycle and early combustion cycle of the engine through the ion current signal, and marking the early combustion cycle and early combustion timing on the ion current signal using the cylinder pressure signal; S3: constructing training data set and training a cyclic neural network model using the ion current signals marked; S4: transplanting a trained cyclic neural network model into an engine ECU for real-time detection of early combustion and carbon deposition of the engine. The present invention has the advantages of higher detection success rate; lower probability of misjudgment and so on compared with the prior art.

Description

METHOD FOR DETECTING AND DISTINGUISHING EARLY COMBUSTION
AND CARBON DEPOSIT OF ENGINE BASED ON ION CURRENT
TECHNICAL FIELD
[01] The present invention relates to the field of internal combustion engines and, more particularly, to a method for detecting and distinguishing early combustion and carbon deposition of an engine based on ion current.
BACKGROUND ART
[02] When the supercharged engine works in the low speed and large load range, it is easy to produce occasional early combustion, and the occurrence of early combustion is more random and has no obvious symptoms. The main hazard of early combustion is that it may cause super knock, which has a very high peak cylinder pressure (greater than 20 MPa) and a very strong cylinder pressure oscillation (amplitude greater than 2
MPa), with the potential to damage the engine within a single combustion cycle.
[03] Therefore, there is a need for a low-cost, real-time diagnostic strategy that can quickly determine early in the onset of early combustion, thereby facilitating subsequent measures to suppress super knock for better engine performance.
[04] In the existing research, Shanghai Automobile Group in Patent "System and
Method for Detecting early Combustion of Engine" (CN 103850852 A) detected the ion current signal in the engine cylinder, calculated the ion current integral value, compared it with the early combustion threshold value, so as to judge whether early combustion occurred in the current combustion cycle. However, this method is slow to judge due to the way of calculating the ion current integral value, and can not achieve rapid diagnosis in the early stage of early combustion. Li Liguang, Tong Sunyu, etc. disclosed a method for detecting early combustion through an ion current in the patent “detection device and detection method for detecting early combustion of spark plug ignition type engine (CN 104564483 A), wherein the early combustion 1s determined by comparing the amplitude of an ion current signal before ignition with a early combustion threshold value.
However, in practical applications, it is found that if carbon deposition exists in the cylinder, especially near the spark plug, a strong ion current signal will also be formed before ignition, the two signals are relatively similar, and it is very easy to cause misjudgment with a simple threshold judgment method .
SUMMARY
[05] It is an object of the present invention to overcome the above-mentioned disadvantages of the prior art and to provide a method for detecting and distinguishing early combustion and carbon deposition of an engine based on 10n current, which is rapid and effective, has a higher detection success rate, a lower probability of misjudgment, and can complete a diagnosis at an early stage of early combustion development, thereby providing sufficient time for suppression of super knock in a cycle.
[06] The object of the present invention can be achieved by the following technical solutions:
[07] A method for detecting and distinguishing early combustion and carbon deposition of an engine based on 10n current, comprising the steps of:
[08] S1: collecting an ion current signal and a cylinder pressure signal under various engine operating conditions;
[09] S2: screening carbon deposition cycle and early combustion cycle of the engine through the ion current signal, and marking the early combustion cycle and early combustion time on the ion current signal using the cylinder pressure signal;
[10] S3: constructing a training data set and training a cyclic neural network model using the ion current signal marked; and
[11] S4: transplanting a trained cyclic neural network model to an engine ECU for the real-time detection of early combustion and carbon deposition.
[12] Further, in step S1, the ion current signal and the cylinder pressure signal of the engine in the normal operation condition, the early combustion condition and the carbon deposition condition are respectively collected.
[13] Further, the ion current signal is derived from an electric field applied between the positive and negative electrodes of an engine spark plug.
[14] Further, screening carbon deposition cycle and early combustion cycle of the engine through the ion current signal specifically comprises:
[15] screening the early combustion and carbon deposition cycles from all data by whether the magnitude of the ion current signal in the early combustion detection window before the ignition time is above a set threshold.
[16] Further, marking the early combustion cycle and early combustion time on the ion current signal using the cylinder pressure signal specifically comprises:
[17] calculating a heat release rate and a combustion phase from the cylinder pressure signal, and marking the early combustion cycle and early combustion time by comparing the combustion start phase with the ignition time.
[18] Further, the training data set comprises input data and output data, wherein the input data is an array of ion current mean values in the early combustion detection window arranged in time, and the output data is a three-element array (x, y, z) used for indicating that the combustion type at the current moment is normal combustion, and the probability of carbon deposition or early combustion exists, wherein x is the probability of normal combustion, y is the probability of carbon deposition and z is the probability of early combustion.
[19] Furthermore, in step S4, when cyclic neural network model is used for real-time detection of early combustion and carbon deposition of the engine , the input variable thereof is the average value of the ion current signal in each degree of crank angle within the early combustion detection window, and the output variable thereof is a three- element array (x,y, z).
[20] Further, the early combustion detection window is the time between the ignition coil storing energy and the ignition timing.
[21] Further preferably, cyclic neural network is a LSTM cyclic neural network with a five-layer structure, wherein a first layer and a third layer are LSTM layers, a second layer and a fourth layer are rejection layers with a given neuron rejection probability, and a fifth layer is a fully connected layer; and cyclic neural network model is iteratively trained using an Adam optimizer, and the neural network structure is adjusted according to the training result.
[22] Furthermore, step S4 specifically comprises:
[23] S41: acquiring the ion current signal of the engine in real time, and judging whether it is currently located in an early combustion detection window, and if so, executing step S42, otherwise, continuing to execute step S41;
[24] S42: calculating an average value of the ion current signal in each degree of crank angle and inputting same into a cyclic neural network model;
[25] S43: the cyclic neural network model judging whether an early combustion occurs, and if so, executing step S44, otherwise, returning to step S45;
[26] S44: taking measures to inhibit super knock;
[27] S45: judging whether it is still located in the early combustion detection window currently, and if so, returning to execute step S42, otherwise, executing step S46;
[28] S46: the cyclic neural network model judging whether a carbon deposition occurs, and if so, executing step S47, otherwise, it is Judged that the engine combustion is normal; and
[29] S47: adjusting subsequent engine operating conditions to eliminate carbon deposition.
[30] Compared to the prior art, the present invention has the following advantages: [BI] 1) In the process of cylinder pressure sensor training, it 1s used to build model training data set, while in on-line diagnosis of batch applications, only ion current sensor is needed instead of cylinder pressure sensor, which greatly saves costs;
[32] 2) The method of the present invention has a higher timeliness than the previous methods of using the ion current integral value to judge the early combustion, and according to the output result of the neural network prediction model, it is possible to judge whether the early combustion occurs at the current moment in time, and the judgment can be made at the early stage of the occurrence of the early combustion, so as to provide sufficient time for subsequent super knock suppression, which is beneficial to protecting the engine structure and greatly improving the detection efficiency;
[33] 3) The method of the present invention can accurately distinguish between early combustion and carbon deposition compared with the conventional method of determining early combustion using the ion current threshold value, can greatly reduce the false alarm rate of early combustion, and can reduce the additional emission generated to suppress the super knock of the engine. In addition, if carbon deposition is detected, it is also possible to eliminate the carbon deposition by subsequently adjusting the engine operating conditions, thereby greatly improving the detection accuracy.
BRIEFT DESCRIPTION OF THE DRAWINGS
[34] FIG.1 is a schematic diagram showing the principle of the method for detecting and distinguishing early combustion and carbon deposition according to the present invention;
[35] FIG.2 is a graph of ion current, cylinder pressure and heat release rate under normal operating conditions;
[36] FIG.3 is a graph of ion current, cylinder pressure, and heat release rate under typical carbon deposition conditions.
[37] FIG.4 is a graph of ion current, cylinder pressure, and heat release rate under typical early combustion conditions.
[38] FIG.5 is a schematic diagram of ion current signal processing;
[39] FIG6 is a flowchart showing the detection of early combustion and carbon deposition when the model is applied in a practical large-scale application.
DETAILED DESCRIPTION OF THE EMBODIMENTS
[40] Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments. It is to be understood that the described embodiments are part of, but not all of, the present invention. Based on the embodiments of the present invention, all other embodiments obtained by a person of ordinary skill in the art without inventive effort should fall within the scope of the present invention.
[41] Examples
[42] As shown in FIG.1, the present invention provides a method for detecting early combustion of an engine based on ion current, comprising the steps of:
[43] A. collecting an ion current signal and a cylinder pressure signal under different operating conditions (normal, early combustion and carbon deposition) of the engine, wherein the ion current signal is from an electric field applied between the positive electrode and the negative electrode of the spark plug.;
[44] B. processing the collected signal, and using the ion current signal to preliminarily distinguish between an early combustion cycle and a carbon deposition cycle, wherein, specifically, early combustion and carbon deposition cycles are distinguished from all data by the amplitude of the ion current signal over a period of time prior to the ignition time;
[45] CC. Marking the early-burn cycle and the early-burn time therein using the cylinder pressure signal, wherein, specifically, the heat release rate and combustion phase are calculated by the cylinder pressure signal, and the early combustion cycle 1s further marked by comparing the combustion start phase with the ignition time;
[46] D. establishing a training data set of the neural network according to the marked result, training the LSTM cyclic neural network, and optimizing the structure of the neural network model at the same time, wherein the input variable of the training data setis an array of ion current mean values arranged according to time within a specified time, and the output result indicates that the combustion type at the current time is normal combustion, and the probability of carbon deposition or the probability of early combustion through a three-element array;
[47] E. using the trained model for on-line real-time diagnosis of carbon deposition and early combustion of the engine, wherein, specifically, the off-line trained model can be directly integrated into a product electronic control unit (ECU) so as to achieve real- time detection and differentiation of early combustion and carbon deposition without adding other controllers.
[48] The method first collects ion current and cylinder pressure data for early combustion and carbon deposition cycles on an engine bench, and then post-processes the collected signals on-line. The ion current and cylinder pressure signals for normal operation, carbon deposition operation and early combustion operation are shown in
FIGS.2, 3 and 4, respectively. the time between the ignition coil storing energy and the ignition time is set as an early combustion detection window, and an early combustion cycle or a carbon deposition cycle is selected according to whether the amplitude of an ion current signal in the window is higher than a set threshold value. The early combustion cycle is further distinguished by calculating the heat release rate and the combustion phasing from the cylinder pressure signal and the early combustion time is marked. The training samples are divided according to the labeling results, and the ion current signal is further processed as shown in FIG.5, and a cyclic neural network based on long-short term memory (LSTM) is trained using the sample data, thereby establishing an optimal neural network model. In the engine on-line combustion diagnosis, after judging that the engine crankshaft position enters the early combustion detection window, the ECU processes the ion current signal and inputs it into the neural network model, and judges the normal combustion, carbon deposition or early combustion in real time according to the output result, as shown in FIG.6.
[49] Specific implementation steps may be as follows:
[50] In the first step, an ion current and cylinder pressure acquisition device is mounted on the dynamometer bench to make the engine run in the conditions easy for early combustion and carbon deposition, data is collected and record is made.
[51] In the second step, the data collected in the first step is analyzed off-line. Firstly, the time between the ignition coil storing energy and the ignition time is set as an early combustion detection window, and the early combustion or carbon deposition cycle are screened out according to whether the amplitude of the ion current signal in the window is higher than a set threshold value. The early combustion cycle is further distinguished by calculating the heat release rate and the combustion phasing from the cylinder pressure signal, and the early combustion time 1s marked.
[52] In the third step, the data is further processed, starting from the peak value of the ion current in the detection window,the mean value of the signal in each degree of the crank angle is calculated as an input array, the mean value is calculated at every 10 collection points, and the length of the input array is designated in advance. If the designated length is not reached, a zero value is filled on the left side of the array. The target output constitutes a three-element array (x,y,z) representing the current probability of normal combustion, carbon deposition, or early combustion, respectively. x, y, zin the training data are filled with 1 or 0 depending on the actual result. The generated database is randomly divided into training group and validation group according to proportion.
[53] In the fourth step, the neural network structure is set, a first layer and a third layer are LSTM layers, a second layer and a fourth layer are rejection layers given the neuron rejection probability, and a fifth layer is a fully connected layer. Adam optimizer 1s used for iterative training, and the neural network structure is adjusted appropriately according to the training results, so as to obtain the optimal discrimination effect.
[54] Inthe fifth step, the neural network prediction model that is trained in the fourth step is transplanted into the ECU. In actual operation, once the current engine crankshaft position is detected and enters the early combustion detection window, the ECU starts to process the ion current signal in real time and input the neural network prediction model. The model determines the current most likely combustion condition based on the output three-element array. If the model determines that early combustion is currently occurring, the determination is aborted and measures are taken to suppress super knock, including but not limited to additional in-cylinder injection or water spray cooling, etc.
If the engine crankshaft position has not exceeded the detection interval by more than the set probability threshold, the current cycle is determined to be normal combustion or carbon deposition based on the last set of output states. If soot is detected, it can be eliminated in subsequent cycles by adjusting engine operating conditions, including but not limited to high speed, lean burn, etc. The test flow is shown in FIG.6.
[55] With regard to the ECU described in the fifth step, it should be noted that the
ECU is an existing product, the current product ECU basically has a certain open function, and it is only necessary to integrate the logic and method described in the present method into the product ECU in the present invention. [S6] The above-mentioned contents are only specific embodiments of the present invention, but the protection scope of the present invention is not limited to this.
Any person familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention.
These modifications or replacements should be covered within the protection scope of the present invention.
Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims (2)

Conclusies l. Werkwijze voor het detecteren en het onderscheiden van vroege verbranding en koolafzetting van een motor op basis van ionenstroom, gekenmerkt doordat deze de volgende stappen omvat van: S1: het verzamelen van een ionenstroomsignaal en een cilinderdruksignaal onder verscheidene motorwerkingsomstandigheden; S2: het screenen van koolafzettingscyclus en vroegeverbrandingscyclus van de motor middels het ionenstroomsignaal, en het markeren van de vroegeverbrandingscyclus en vroegeverbrandingstijd op het ionenstroomsignaal met behulp van het cilinderdruksignaal; S3: het construeren van een trainingsdataset en het trainen van cyclischneuraalnetwerkmodel met behulp van het gemarkeerde ionenstroomsignaal, waarbij het cyclischneuraalnetwerkmodel een LSTM-cyclischneuraalnetwerkmodel met een vijflaagse structuur is, waarbij een eerste laag en een derde laag daarvan LSTM- lagen zijn, een tweede lagen een vierde laag verwerpingslagen met een gegeven neuronverwerpingswaarschijnlijkheid zijn, en een vijfde laag een volledig verbonden laag 1s, waarbij het cyclischneuraalnetwerkmodel met behulp van een Adamoptimaliseermiddel iteratieve training en aanpassing van een neuraalnetwerkstructuur volgens een trainingsresultaat uitvoert; en S4: het transplanteren van een getraind cyclischneuraalnetwerkmodel naar een motor-ECU voor real-timedetectie van vroege verbranding en koolaf zetting.Conclusions l. A method of detecting and distinguishing early combustion and carbon deposit of an engine based on ion current, characterized in that it comprises the steps of: S1: collecting an ion current signal and a cylinder pressure signal under various engine operating conditions; S2: screening carbon deposition cycle and early combustion cycle of the engine by the ion current signal, and marking the early combustion cycle and early combustion time on the ion current signal by using the cylinder pressure signal; S3: Constructing a training dataset and training cyclic neural network model using the marked ion current signal, where the cyclic neural network model is an LSTM cyclic neural network model with a five-layer structure, where a first layer and a third layer thereof are LSTM layers, a second layer is a fourth layer are rejection layers with a given neuron rejection probability, and a fifth layer is a fully connected layer 1s, wherein the cyclic neural network model performs iterative training and adaptation of a neural network structure according to a training result using an Adam optimizer; and S4: transplanting a trained cyclic neural network model to an engine ECU for real-time detection of early combustion and carbon deposition. 2. Werkwijze voor het detecteren en het onderscheiden van vroege verbranding en koolafzetting van een motor op basis van ionenstroom volgens conclusie 1, met het kenmerk dat de stap S4 specifiek het volgende omvat: S41: het verwerven van het ionenstroomsignaal van de motor in real-time, en het beoordelen of het zich momenteel in een detectievenster van vroege verbanding bevindt, en indien dit het geval is, het uitvoeren van stap S42, anders, het voortzetten van het uitvoeren van stap S41; S42: het berekenen van een gemiddelde waarde van het ionenstroomsignaal in elke graad van krukashoek en het invoeren van dezelfde in een cyclischneuraalnetwerkmodel,The method for detecting and distinguishing early combustion and carbon deposit of an engine based on ion current according to claim 1, characterized in that the step S4 specifically comprises: S41: acquiring the engine ion current signal in real-time time, and judging whether it is currently in an early connection detection window, and if so, performing step S42, otherwise, continuing to perform step S41; S42: Calculating an average value of the ion current signal in each degree of crankshaft angle and inputting the same into a cyclical neural network model, S43: het, door het cyclischneuraalnetwerkmodel, beoordelen of een vroege verbranding plaatsvindt, en indien dit het geval is, het uitvoeren van stap S44, anders, het teruggaan naar stap S45;S43: judging, by the cyclic neural network model, whether an early burn occurs, and if so, performing step S44, otherwise, returning to step S45; S44: het nemen van maatregelen om super kloppen (“super knock™) te verminderen;S44: taking measures to reduce super knock™; S45: het beoordelen of het zich momenteel nog steeds in de detectievenster van vroege verbranding bevindt, en indien dit het geval is, het teruggaan om stap S42 uit te voeren, anders, het uitvoeren van stap S46;S45: judging whether it is currently still in the early combustion detection window, and if so, going back to perform step S42, otherwise, performing step S46; S46: het, door het cyclischneuraalnetwerkmodel, beoordelen of een koolafzetting plaatsvindt, en indien dit het geval 1s, het uitvoeren van S47, anders, wordt beoordeeld dat de motorverbranding normaal is; enS46: Judging, by the cyclic neural network model, whether a carbon deposit occurs, and if so, performing S47, otherwise, it is judged that the engine combustion is normal; and S47: het aanpassen van opeenvolgende motorwerkingsomstandigheden om koolafzetting te elimineren.S47: Adjusting sequential engine operating conditions to eliminate carbon deposits.
NL2029619A 2021-11-03 2021-11-03 Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current NL2029619B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
NL2029619A NL2029619B1 (en) 2021-11-03 2021-11-03 Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
NL2029619A NL2029619B1 (en) 2021-11-03 2021-11-03 Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current

Publications (1)

Publication Number Publication Date
NL2029619B1 true NL2029619B1 (en) 2023-06-02

Family

ID=86646815

Family Applications (1)

Application Number Title Priority Date Filing Date
NL2029619A NL2029619B1 (en) 2021-11-03 2021-11-03 Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current

Country Status (1)

Country Link
NL (1) NL2029619B1 (en)

Similar Documents

Publication Publication Date Title
US5392642A (en) System for detection of low power in at least one cylinder of a multi-cylinder engine
US7347081B2 (en) Knock detecting apparatus and method for internal combustion engine
CN108915883B (en) Spark ignition type engine super detonation monitoring system
Leonhardt et al. Methods for engine supervision and control based on cylinder pressure information
CN100476182C (en) Misfire detection apparatus
CN112629648B (en) Ignition timing control device for internal combustion engine
US10809155B2 (en) Method for detecting pre-ignition of internal combustion engine
DE112016001180T5 (en) A control device for an internal combustion engine and an abnormal combustion detection method
Tamura et al. Misfire detection on internal combustion engines using exhaust gas temperature with low sampling rate
JP2587543B2 (en) Knock detection device
US11346316B2 (en) Ignition timing control device for internal combustion engine
CN101907520B (en) Engine fire detection method and detection device thereof
Pla et al. Knock analysis in the crank angle domain for low-knocking cycles detection
Hosseini et al. Knock probability determination in a turbocharged gasoline engine through exhaust gas temperature and artificial neural network
CN112160842B (en) Method for detecting and distinguishing pre-ignition and carbon deposition of engine based on ion current
NL2029619B1 (en) Method for detecting and distinguishing early combustion and carbon deposit of engine based on ion current
CN117877256A (en) Vehicle fault prediction method and system based on digital twin
JP4155807B2 (en) Knock detection device
Borg et al. Characterization of autoignition in a knocking SI engine using heat release analysis
CN108625983B (en) Preignition detecting method and device based on transient rotative speed
Siano et al. A nonlinear black-box modeling method for knock detection in spark-ignition engines
Chao et al. Abnormal combustion diagnosis and combustion period based phase estimation with a modified form tandem ion current detection system on SI gasoline engines
Park et al. Knock Onset Determination with 1D CNN Using Random Search Hyperparameter Optimization and Data Augmentation in SI Engine
JPH11183330A (en) Misfire discriminating method for internal combustion engine
Jianmin et al. Diesel misfire fault diagnosis and misfire cylinder ascertainment based on vibration signal explosion peak identification