Marine diesel engine alarm system

Abstract

The invention relates to the field of monitoring, sensing and early warning, and discloses a marine diesel engine alarm system which comprises a data acquisition module used for acquiring various physical parameters in the operation process of a marine diesel engine; the data preprocessing module is used for carrying out denoising, normalization and time synchronization processing on the physical parameters; the state modeling module is used for constructing a mathematical model reflecting the running state of the diesel engine based on the preprocessed physical parameters; the feature extraction module is used for extracting feature information reflecting a fault trend or disturbance from the mathematical model; the alarm judgment module is used for generating an alarm signal according to the feature information extracted by the mathematical model; and the alarm output module is used for outputting the alarm signal, recording alarm information and feeding back the alarm information to a system management end. According to the invention, through multi-dimensional parameter fusion and a self-adaptive alarm mechanism, the accuracy and response speed of fault early warning of the marine diesel engine are improved, and false alarms and missing alarms are reduced.

Description

Technical Field

[0001] This invention relates to the field of monitoring, sensing, and early warning, specifically to a marine diesel engine alarm system. Background Technology

[0002] With the widespread application of marine diesel engines in navigation and operations, their operating status plays a crucial role in the safety, reliability, and economy of ships. However, due to the long-term high-load operation of marine diesel engines and their frequent exposure to various external disturbances and complex operating conditions, fluctuations in physical parameters such as temperature, pressure, and vibration often occur during operation. Once anomalies occur, these problems can lead to system failures, and in severe cases, even equipment shutdowns or accidents, resulting in increased maintenance costs, reduced operational efficiency, and even affecting ship safety.

[0003] In existing technologies, fault warning systems for marine diesel engines typically rely on monitoring methods based on a single physical parameter. For example, traditional methods depend on threshold settings for temperature or vibration signals; the system issues an alarm when it detects that a certain parameter exceeds a preset threshold. However, this single-parameter monitoring approach has several drawbacks. First, traditional methods often neglect the comprehensive analysis of multi-dimensional characteristics, easily leading to the overlooking of subtle but potential faults. Second, in complex operating environments, the system often struggles to accurately determine whether fluctuations in certain physical parameters are normal operating condition changes or merely transient fluctuations caused by temporary disturbances, resulting in false alarms or missed alarms. Furthermore, because existing technologies are relatively slow to detect abnormal changes, they often only respond after a fault occurs, lacking the ability to provide early warnings of potential faults.

[0004] Meanwhile, existing fault diagnosis and alarm mechanisms are usually based on simple threshold judgments, ignoring the influence and interrelationships of multiple factors. Although some methods attempt to introduce model prediction or multi-dimensional data fusion, most methods still have limitations in terms of computational complexity and model adaptability. For example, some schemes lack adaptive adjustment capabilities, making it difficult to automatically update judgment criteria according to changes in equipment status, and can only rely on fixed set parameters, which is particularly inadequate in long-term ship operation.

[0005] Therefore, this invention proposes a marine diesel engine alarm system to address the shortcomings of existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a marine diesel engine alarm system that solves the problems of monitoring based on a single physical parameter, serious false alarms and missed alarms, lack of adaptive adjustment, and insufficient early warning capability for faults under complex operating conditions.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a marine diesel engine alarm system, comprising: The data acquisition module is used to collect various physical parameters during the operation of the ship's diesel engine; The data preprocessing module is used to perform noise reduction, normalization, and time synchronization processing on the physical parameters; The state modeling module is used to construct a mathematical model reflecting the operating state of the diesel engine based on the preprocessed physical parameters; The feature extraction module is used to extract feature information reflecting fault trends or disturbances from mathematical models; An alarm determination module is used to generate an alarm signal based on the feature information extracted from the mathematical model. The alarm output module is used to output the alarm signal, record the alarm information, and send feedback to the system management terminal.

[0008] Preferably, the data acquisition module includes: Temperature acquisition unit is used to acquire temperature parameters at the measuring point location of the diesel engine; The pressure acquisition unit is used to acquire pressure parameters of the fuel system and lubrication system; The vibration acquisition unit is used to collect vibration parameters of the rotating parts of the diesel engine.

[0009] Preferably, the data preprocessing module includes: The noise reduction module is used to filter out noise from the acquired physical parameters; The normalization module is used to unify the dimensions of the denoised physical parameters. The time synchronization module is used to perform time alignment processing on the normalized physical parameters.

[0010] Preferably, the state modeling module includes: The thermal modeling unit is used to establish a mathematical model characterizing the heat transfer characteristics of a diesel engine based on temperature parameters. The pressure modeling unit is used to establish a mathematical model characterizing the working pressure characteristics of a diesel engine based on pressure parameters. The vibration modeling unit is used to establish a mathematical model characterizing the dynamic response characteristics of a diesel engine based on vibration parameters.

[0011] Preferably, the feature extraction module includes: The trend feature extraction unit is used to extract feature information reflecting the changing trend of the diesel engine's operating state from the mathematical model; the disturbance feature extraction unit is used to extract feature information reflecting the operating disturbance of the diesel engine from the mathematical model.

[0012] Preferably, the step of extracting feature information reflecting the changing trend of diesel engine operating status from the mathematical model is as follows: Calculate the rate of change of the diesel engine's operating state based on the temperature, pressure, and vibration parameters in the mathematical model. The rate of change is smoothed to obtain the trend change signal; Based on a preset threshold, abrupt change points in the trend change signal are extracted to reflect the abrupt change trend of the diesel engine's operating status. The formula for calculating the rate of change is: Where r(t) is the rate of change at time t, X(t) is the parameter value at time t, X(t-Δt) is the parameter value of the previous sampling period, and Δt is the sampling time interval.

[0013] Preferably, the step of extracting feature information reflecting diesel engine operating disturbances from the mathematical model is as follows: Calculate the residuals of each parameter in the mathematical model; Perform a Fourier transform on the residuals to obtain the perturbation characteristic signal in the frequency domain; Extract the main frequency components in the frequency domain and identify the characteristic information reflecting the disturbance; The formula for calculating the residual is: ε(t)=X meas (t)-X model (t); Where ε(t) is the residual at time t, X meas (t) represents the actual measured value at time t, X model (t) represents the model prediction at time t.

[0014] Preferably, the alarm determination module includes: The fault determination unit is used to determine whether the diesel engine has malfunctioned based on the trend feature information and disturbance feature information extracted by the feature extraction module. The threshold judgment unit is used to determine whether to trigger an alarm signal based on a preset alarm threshold. The alarm generation unit is used to generate an alarm signal when a diesel engine malfunction is detected.

[0015] Preferably, the step of determining whether to trigger an alarm signal based on a preset alarm threshold includes: Calculate the alarm index based on the characteristic information of the trend and the characteristic information of the disturbance; Compare the alarm index with the preset alarm threshold; When the alarm index exceeds the preset alarm threshold, an alarm signal is triggered. The formula for calculating the alarm index is as follows: A(t)=α·∣r(t)∣+β·|ε(t)|; Where A(t) is the alarm index at time t, r(t) is the rate of change at time t, ε(t) is the residual at time t, and α and β are weighting coefficients.

[0016] Preferably, the alarm output module includes: Alarm signal output unit, used to output alarm signals to ship control system or terminal equipment; Alarm recording unit, used to record alarm events; The feedback control unit is used to send alarm information to the system management terminal and trigger corresponding emergency response operations.

[0017] This invention provides a marine diesel engine alarm system. It has the following beneficial effects: 1. This invention achieves more accurate diesel engine fault identification by constructing an alarm judgment mechanism that fuses "trend + disturbance" dual-dimensional features. Compared with traditional techniques that rely solely on static threshold judgment, it effectively avoids false alarms and missed alarms, and solves the problem of poor robustness of existing technologies under complex operating conditions.

[0018] 2. A moving average-processed rate of change signal extraction strategy was adopted, making trend identification smoother and the judgment of abnormal change points more controllable. This processing makes the system less susceptible to interference from instantaneous fluctuations. Previous methods often directly compared values, which was prone to errors, especially under high-frequency disturbances.

[0019] 3. This invention deeply couples the state modeling module with the feature extraction module, establishing a multi-dimensional mathematical model of temperature, pressure, and vibration, and constructing disturbance features based on residual frequency domain analysis. This design approach is more comprehensive than existing strategies that use single-parameter monitoring, and solves the problem of poor early-stage weak fault identification capabilities of traditional methods.

[0020] 4. By introducing a dynamic alarm index calculation mechanism and setting adjustable multi-level alarm thresholds, alarm judgment becomes adaptive and does not rely on manual experience settings. This approach is more flexible than the previous static logic judgment and solves the technical shortcomings of existing systems, such as poor adaptability, complex parameter tuning, and susceptibility to misjudgment. Attached Figure Description

[0021] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a diagram of the data acquisition module architecture of the present invention; Figure 3 This is a diagram of the data preprocessing module architecture of the present invention; Figure 4 This is an architecture diagram of the state modeling module of the present invention; Figure 5 This is a diagram of the feature extraction module architecture of the present invention; Figure 6 This is a diagram of the alarm determination module architecture of the present invention; Figure 7 This is a diagram of the alarm output module architecture of the present invention. Detailed Implementation

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see the appendix Figure 1 - Appendix Figure 7 This invention provides a marine diesel engine alarm system, comprising: The data acquisition module is used to collect various physical parameters during the operation of the ship's diesel engine; The data acquisition module, through multiple sensors and measurement units, can accurately and promptly capture various physical parameters of the diesel engine during operation. This data will provide crucial information for subsequent fault warnings, condition analysis, and alarm determination.

[0024] In this embodiment, the data acquisition module comprises multiple sub-units, each responsible for acquiring different types of physical parameters. Specifically, the data acquisition module includes a temperature acquisition unit, a pressure acquisition unit, and a vibration acquisition unit. Each unit collaborates with the acquisition device through specific sensors to acquire key data regarding the diesel engine's operating status in real time. This data will undergo further processing by a preprocessing module to ensure its accuracy and reliability, and to provide high-quality input for subsequent model building and feature extraction.

[0025] Temperature acquisition unit: In one possible implementation, the temperature acquisition unit monitors the temperature changes of various components of the diesel engine in real time by installing temperature sensors at multiple measuring points. These temperature sensors can reflect the thermal state changes of the diesel engine during operation, especially the temperatures of key components such as engine cylinders, combustion chambers, cooling systems, and exhaust systems.

[0026] Specifically, the temperature acquisition unit can collect temperature data through thermocouples, infrared sensors, and other means. The real-time changes in temperature data reflect the thermal load and operating status of the diesel engine, thus providing strong support for subsequent state modeling and feature extraction.

[0027] Pressure acquisition unit: The pressure acquisition unit is responsible for collecting pressure parameters from the diesel engine's internal fuel and lubrication systems. During the diesel engine's operation, pressure fluctuations in the fuel and lubrication systems can reveal the engine's operating status, system sealing, and lubrication conditions.

[0028] For example, the pressure in the fuel system may reflect the operating status of the fuel pump, while the pressure in the lubrication system can reflect whether the oil pump is working properly. The pressure acquisition unit collects real-time data through pressure sensors (such as pressure transmitters) and transmits this data to the data preprocessing module for further analysis.

[0029] Vibration acquisition unit: The vibration acquisition unit is used to collect vibration parameters of the rotating components of a diesel engine. The rotating components of a diesel engine generate vibrations during operation, especially over long periods, as wear and tear and malfunctions can increase the vibration amplitude. By monitoring vibration parameters, potential abnormalities or malfunctions in the diesel engine can be detected early.

[0030] Vibration acquisition units are typically equipped with accelerometers or vibration sensors, which are installed on critical components of the diesel engine, such as the crankshaft, connecting rods, and pistons. Vibration signals are acquired in real time and transmitted to a data processing module for further analysis of the diesel engine's operating status.

[0031] In this embodiment, the parameters (such as temperature, pressure, and vibration data) collected by the data acquisition module need to be preprocessed according to specific formulas. The temperature acquisition unit may output temperature data T(t), the pressure acquisition unit may output pressure data P(t), and the vibration acquisition unit may output vibration data V(t). These data are associated with timestamps to form a continuous data stream.

[0032] Specifically, the collected data T(t), P(t), and V(t) can be standardized using the following formula: Among them, T norm (t) represents the normalized temperature result at time t, which is a dimensionless standard value; P norm (t) represents the time P. norm The pressure normalization result of (t); V norm (t) represents the vibration normalization result at time t; T(t) represents the temperature, pressure, and vibration data at time t, respectively. min T max P min P max V min V maxThese represent the minimum and maximum values ​​of the parameter, respectively. Through the normalization process, parameters from different measurement units can be transformed into a unified standard form, providing consistent data input for the subsequent feature extraction module.

[0033] In this embodiment, the data acquisition module and the data preprocessing module work closely together. During the data acquisition process, the acquired temperature, pressure, and vibration data undergo preprocessing steps, including noise reduction, normalization, and time synchronization. These steps ensure that the raw data can be accurately and reliably analyzed in subsequent state modeling and feature extraction processes.

[0034] The collaborative work between the data acquisition module and the condition modeling module is also crucial. Data such as temperature, pressure, and vibration will serve as inputs, providing a foundation for diesel engine condition modeling. Using this data, the system can construct a mathematical model reflecting the diesel engine's operating state, thereby enabling effective prediction of diesel engine fault trends and disturbances.

[0035] In one possible implementation, the data acquisition module can use different types of sensors depending on the specific application scenario. For the temperature acquisition unit, in addition to thermocouples and infrared sensors, RTDs (Resistance Temperature Detectors) can also be used to achieve high-precision temperature measurement. For the pressure acquisition unit, in addition to common pressure sensors, strain gauge-based pressure sensors can be selected to obtain more accurate pressure data. The vibration acquisition unit can choose to use higher-precision vibration sensors, such as piezoelectric vibration sensors, to improve monitoring sensitivity.

[0036] The marine diesel engine alarm system in this embodiment achieves multi-dimensional real-time monitoring of the diesel engine through a data acquisition module, ensuring the real-time acquisition of key physical parameters such as temperature, pressure, and vibration. These parameters provide a solid data foundation for subsequent fault warning, status analysis, and alarm determination. Through precise data acquisition and processing, the system can promptly detect abnormal states, improving the safety and reliability of the marine diesel engine.

[0037] The data preprocessing module is used to perform noise reduction, normalization, and time synchronization processing on the physical parameters; The main task of the data preprocessing module is to standardize and normalize the raw sensor data to ensure the stability, uniformity, and comparability of the input data required for subsequent analysis and modeling. This module is closely integrated with the data acquisition module, and its output will serve as the data input basis for the state modeling module.

[0038] In this embodiment, the data preprocessing module mainly includes the following three sub-modules: a noise reduction module, a normalization module, and a time synchronization module. Each sub-module has a clear function and independent structure, while forming an organic synergy within the overall system architecture.

[0039] In this embodiment, the noise reduction module is used to filter out noise from the collected physical parameters.

[0040] In general, raw data collected from sensors such as temperature, pressure and vibration often contains abnormal signal fluctuations caused by high-frequency interference, electromagnetic noise or mechanical shock.

[0041] Specifically, in one possible implementation, the denoising module uses a moving average filter to smooth the original signal, with the basic formula as follows: Among them, X smooth X(t) represents the smoothed data result at the current time t, i.e. the signal output value after noise reduction; X(ti) represents the original data point collected at time ti; i is the offset step index in the sliding window, with a value range from 0 to N-1; N is the size of the sliding window, representing the number of data points used to calculate the moving average, with a value of a positive integer, generally set according to the sampling frequency, signal characteristics, or target response speed.

[0042] Alternatively, wavelet transform can be used to remove high-frequency noise in the frequency domain, which is particularly suitable for denoising non-stationary signals, such as vibration data. This method can effectively suppress abnormal pulses or sensor drift errors while preserving the signal's trend.

[0043] In this embodiment, the normalization module is used to perform dimensional unification processing on the denoised physical parameters.

[0044] As mentioned above, in order to facilitate the input of parameters of different units and magnitudes into the same modeling framework, it is necessary to normalize various types of physical data.

[0045] Specifically, the system employs a minimum-maximum normalization method to process parameters such as temperature, pressure, and vibration as follows: Among them, X norm X(t) is the normalized standard value, generally limited to [0,1]; X(t) is the original value of the input parameter at time t, taken from the output of the denoising module; X min This is the preset or historical minimum value of this parameter within the monitoring period; X max This is the preset or historical maximum value of the parameter within the monitoring period.

[0046] In some embodiments, to prevent extreme outliers from interfering with the normalization results, a pruning mechanism can be introduced before normalization, that is, restricting X(t) to [X... min ,X max Process within the interval: X′(t)=min(max(X(t),X min ),X max ); Where X′(t) represents the clipped data value at time t.

[0047] This method is particularly effective in vibration parameter processing and is suitable for stable input control when the data contains mechanical shock interference or spike signals.

[0048] In this embodiment, the time synchronization module is used to perform time alignment processing on the normalized physical parameters.

[0049] Considering that multiple sensors have different acquisition frequencies and inconsistent response delays, direct fusion analysis may introduce time deviations and affect the quality of modeling.

[0050] In general, the time synchronization module uses an interpolation resampling method to synchronize each data stream according to a unified time axis.

[0051] In one possible implementation, a unified set of time nodes {t0, t1, ..., t} is defined. n All data streams will be resampled into this set. Original data values ​​not located at this node will be processed using linear interpolation. Where X(t) k ) represents the target time point t k The interpolation result on t is the estimated data value at that time after interpolation processing; k For the target interpolation time point; X(t) i ) is at time t i The collected raw data values ​​are used as the starting point for interpolation; X(t) i+1 ) is at time t i+1 The collected raw data values ​​are used as the endpoint values ​​for interpolation; t i t i+1 To be with X(t) i ), X(t) i+1 The corresponding timestamp; t k -t i Indicates starting from time t i to the target interpolation point t k The time difference.

[0052] In some high-frequency sampling systems, synchronization can also be achieved through a timestamp-based caching mechanism, which involves using a circular buffer to cache the most recent data frames at different sampling frequencies and selecting the data closest to the target time point as input.

[0053] As a further extension, in some embodiments, the data preprocessing module may also integrate a data integrity verification submodule to determine whether there are breakpoints or continuous anomalies in the acquired signal. If the data is determined to be untrustworthy, it is shielded by marking or setting it to null to prevent it from affecting the subsequent processing flow.

[0054] In summary, the data preprocessing module in this embodiment of the invention performs noise suppression, dimension unification, and time alignment on the raw collected data through a multi-step, multi-level processing flow, laying a highly consistent and high-quality data foundation for state modeling and feature extraction. This structural setup has good versatility and engineering adaptability, and can adapt to the data input requirements of various marine diesel engine operating scenarios.

[0055] The state modeling module is used to construct a mathematical model reflecting the operating state of the diesel engine based on the preprocessed physical parameters; The state modeling module is responsible for converting physical layer signals into state representation models with structural semantics, and is the core component for realizing diesel engine fault identification and trend analysis.

[0056] This module receives multi-dimensional physical parameters such as temperature, pressure, and vibration after denoising, normalization, and time synchronization processing, and constructs mathematically expressible physical behavior models for different types of signals. Through this modeling process, the system can abstract information such as heat conduction characteristics, pressure change behavior, and dynamic response laws from time-series data, laying a mathematical foundation for subsequent feature extraction and alarm determination.

[0057] In this embodiment, the thermal modeling unit is used to establish a mathematical model characterizing the thermal conduction characteristics of a diesel engine based on temperature parameters.

[0058] Under normal circumstances, diesel engines exhibit thermal coupling behavior between multiple components and materials during operation. Therefore, the temperature distribution changes are not only related to the current load and operating conditions, but also affected by the structural heat dissipation path and heat exchange efficiency.

[0059] Specifically, in one possible implementation, the thermal modeling unit mathematically describes temperature changes based on a one-dimensional transient heat conduction model, the basic form of which can be expressed as: Where T(x,t) represents the temperature distribution function at position x at time t, in °C; α is the thermal diffusivity, in m³ / s. 2 / s; t is the time variable; This represents the partial derivative operation.

[0060] As an alternative, in scenarios with well-defined boundary conditions, an initial boundary value T(x,0) and a convection boundary term can be introduced to fit the influence of the cooling system or exhaust environment on the heat distribution.

[0061] In some embodiments, if there are few measuring points, a temperature state vector model can be used instead of a distribution model, defined as: T(t)=[T1(t),T2(t),...,T n (t)]; Among them, T i (t) represents the temperature value of the i-th measuring point at time t. The system can use this temperature vector to perform state tracking and residual diagnosis of the diesel engine heat transfer process.

[0062] In this embodiment, the pressure modeling unit is used to establish a mathematical model characterizing the working pressure characteristics of the diesel engine based on pressure parameters.

[0063] During diesel engine operation, the internal pressure of its fuel injection and lubrication systems fluctuates periodically and is closely related to factors such as load and speed. Establishing a pressure model helps to characterize the system's stability and hydraulic response.

[0064] Specifically, in one possible implementation, pressure modeling employs a state-space modeling approach, simplifying pressure changes into a linear dynamic system, expressed as: Where P(t) represents the pressure state vector; A is the state transition matrix, representing the internal pressure change trend; B is the input gain matrix; u(t) is the external input variable, such as rotational speed, load, etc.; y p (t) represents the observable pressure output; C is the output mapping matrix; P(t) represents the pressure value at time t.

[0065] As an alternative, a pulsating pressure modeling method can be introduced into the diesel engine fuel supply system to fit the periodic fluctuations as: Where P(t) represents the pressure value at time t; P0 represents the average pressure or reference pressure of the system; K represents the number of harmonics in the fluctuation mode, and K is a positive integer; A k f represents the amplitude of the k-th harmonic component; k φ represents the frequency of the k-th harmonic; k represents the phase angle of the k-th harmonic; t represents the time variable.

[0066] This approach is suitable for analyzing high-frequency disturbances such as unstable oil injection and pipeline resonance.

[0067] In this embodiment, the vibration modeling unit is used to establish a mathematical model characterizing the dynamic response characteristics of the diesel engine based on vibration parameters.

[0068] Vibration signals exhibit strong nonlinear and non-stationary characteristics, reflecting the microscopic dynamics of the internal mechanical structure of a diesel engine (such as the crankshaft and connecting rod). By establishing a dynamic response model, the system can identify vibration changes caused by bearing wear, imbalance, or structural loosening.

[0069] In one possible implementation, vibration modeling is based on a time-domain autoregressive model (AR model), the mathematical form of which is as follows: Where V(t) represents the vibration signal value at the current time t; V(tj) represents the historical vibration signal value at time tj, i.e., the j-th lag term; p represents the model order, i.e., the number of historical data points involved in the modeling, which is a positive integer; a j denoted as the j-th order autoregressive coefficient, reflecting the degree of influence of the lag term on the current vibration value; ∈(t) represents the residual term or white noise interference, representing unpredictable disturbances or system noise; j is the cyclic summation index variable.

[0070] In some embodiments, frequency domain modeling methods can also be used, such as Fast Fourier Transform (FFT) to extract the main frequency response features, and then combined with an amplitude-frequency curve fitting model to monitor the resonance intensity corresponding to the rotational speed.

[0071] Furthermore, if the system has high computing power, nonlinear system modeling methods, such as support vector regression (SVR) or recurrent neural networks (RNN), can be used to dynamically model and predict the changing trends in long-term series.

[0072] In summary, the state modeling module in this embodiment of the invention achieves a structural mapping from sensor data to state representation by separately modeling the signals in three dimensions: temperature, pressure, and vibration. This module not only supports the input standardization of the subsequent feature extraction module but also provides the mathematical foundation for the physical behavior and operational state of the entire alarm mechanism. The model structure is clear, the formulas are fully defined, and it possesses both engineering feasibility and algorithm reproducibility.

[0073] The feature extraction module is used to extract feature information reflecting fault trends or disturbances from mathematical models; The feature extraction module operates on top of the state modeling module. It takes the mathematical models generated by thermal modeling, pressure modeling, and vibration modeling as input and extracts key indicator information that characterizes the diesel engine's operating trend changes and disturbance characteristics. Through the feature extraction module, a structured mapping of the system's operating state from the "physical model" to the "feature space" can be achieved, serving as a crucial intermediate step for the alarm system to complete risk identification and classification assessment.

[0074] In this embodiment, the trend feature extraction unit is used to extract feature information reflecting the changing trend of the diesel engine's operating status from the mathematical model.

[0075] Generally, slow changes in parameters such as temperature, pressure, and vibration reflect long-term trends in system operating conditions over time, such as changes in power output, decreased cooling efficiency, and load fluctuations. To identify such trend behaviors, the system first calculates the rate of change of various physical parameters output from the modeling module.

[0076] Specifically, the trend feature extraction unit is based on the following formula for calculating the rate of change: Where r(t) is the rate of change at time t, X(t) is the value of a certain physical parameter (temperature, pressure or vibration) at time t, X(t-Δt) is the parameter value of the previous sampling period, and Δt is the sampling time interval.

[0077] As an alternative, to reduce the interference of high-frequency fluctuations on trend judgment, a moving average method can be applied to smooth the rate of change series. The formula for the moving average is as follows: Where, r smooth (t) represents the smoothed rate of change at the current time t, which is the trend indicator obtained after denoising the rate of change signal; r(ti) represents the original rate of change value at time point ti; i represents the index in the sliding window; N represents the length of the sliding window, which is the number of data points participating in the average calculation. It is a positive integer and is usually set according to the sampling frequency and the target trend response duration; t represents the current time.

[0078] In one possible implementation, to further determine abnormal trends or accelerated changes, the system introduces a threshold determination mechanism to detect abrupt changes in the trend change signal. Abrupt changes can be determined by comparing r... smooth (t) is achieved by setting a threshold θ: If |r smooth If (t)|>θ, then it is determined to be a trend inflection point.

[0079] In some embodiments, multi-level thresholds θ1 and θ2 under different operating conditions can be set to achieve hierarchical alarms.

[0080] In this embodiment, the disturbance feature extraction unit is used to extract feature information reflecting the operating disturbance of the diesel engine from the mathematical model.

[0081] Diesel engines may experience short-term or high-frequency disturbances during normal operation, such as rapid vibrations or pressure fluctuations caused by mechanical shocks, cylinder knocking, or pipeline resonance. These disturbances are usually difficult to reflect directly from trend changes, so it is necessary to introduce a combination of residual and frequency domain analysis to capture them.

[0082] Specifically, the perturbation feature extraction unit first calculates the model residuals for various parameters. The formula for calculating the residuals is as follows: ε(t) = X meas (t)-X model (t); Where ε(t) is the residual at time t, X meas (t) represents the actual measured value at time t, X model (t) represents the model prediction obtained through thermal modeling, pressure modeling, or vibration modeling.

[0083] Alternatively, after the residual calculation is complete, the system applies a Fast Fourier Transform (FFT) to the residual signal to transform it into the frequency domain: Where E(f) is the magnitude of the residual spectrum at frequency f; ε(t) represents the Fourier transform operation; f is the frequency variable; ε(t) represents the residual value at time t, that is, the difference between the actual measured value and the model prediction value.

[0084] In one possible implementation, the system further extracts the dominant frequency component (i.e., the frequency point with the maximum energy) in the frequency domain to identify specific disturbance sources: In a fuel supply system, the presence of periodic residuals at a certain frequency may indicate abnormal fuel pressure fluctuations. In a vibration channel, if high-frequency energy accumulates in the spectrum, it may indicate structural resonance or early bearing damage.

[0085] In some embodiments, the system combines multiple frequency components with their amplitude ratios to construct a disturbance feature vector, which serves as the input for subsequent alarm judgment and classification models.

[0086] In summary, the feature extraction module in this embodiment achieves multi-dimensional structural feature extraction of the diesel engine's operating state by combining trend change rate analysis with frequency domain disturbance detection. This method not only covers long-term trend behavior but also encompasses the ability to analyze short-term disturbance signals, forming the core foundation of the alarm decision logic.

[0087] An alarm determination module is used to generate an alarm signal based on the feature information extracted from the mathematical model. The alarm determination module generates a key indicator variable, namely the alarm index, by comprehensively analyzing trend characteristics and disturbance characteristics, to determine whether an alarm signal is triggered, thus forming a complete fault identification and alarm chain.

[0088] In this embodiment, the alarm determination module includes a fault determination unit, a threshold determination unit, and an alarm generation unit.

[0089] Under normal circumstances, diesel engines may exhibit trend-based variations (such as slow temperature increases or long-term vibration deviations from the baseline) or disturbance anomalies (such as sudden short-term pressure changes or localized impacts) during operation. Both types of characteristics can serve as important bases for determining the fault state. Therefore, the alarm determination module combines the aforementioned trend feature extraction unit and smoothing rate of change analysis method with the residual frequency domain information from the disturbance feature extraction unit to comprehensively analyze and make an operational status judgment in the fault judgment unit.

[0090] In this embodiment, the fault judgment unit is used to determine whether the diesel engine has any abnormal operation or potential faults based on the trend feature information and disturbance feature information extracted by the feature extraction module.

[0091] In one possible implementation, the system introduces the concept of an alarm index, fusing feature information from different dimensions to unify the judgment criteria. The formula for calculating the alarm index is as follows: A(t)=α·∣r(t)∣+β·|ε(t)|; Where A(t) is the alarm index at time t, r(t) is the rate of change at time t, ε(t) is the residual at time t, and α and β are weighting coefficients.

[0092] Alternatively, in certain applications, a signal r(t) that has undergone sliding smoothing can also be introduced. smooth (t) is used to enhance anti-interference capabilities. The alarm index can then be extended to the following form: A(t) = α·|r smooth (t)|+β·|ε(t)|; in: This is the result of a moving average of length N.

[0093] This approach can improve the system's ability to respond to periodic fault characteristics, and is particularly suitable for identifying anomalies caused by structural vibration or mechanical shock.

[0094] In this embodiment, the threshold determination unit is used to determine whether to trigger an alarm signal based on a preset alarm threshold.

[0095] Specifically, the system compares the alarm index A(t) with the preset alarm threshold θ to determine whether the alarm condition is met: if A(t)>θ, it is considered that there is an operational abnormality and the alarm condition is triggered.

[0096] Generally, the alarm threshold θ can be set during initial system debugging or adaptively adjusted based on historical operating data. In one feasible implementation, to achieve tiered response, multiple thresholds θ1 and θ2 can be set, corresponding to different response levels such as early warning and severe alarm.

[0097] In some embodiments, to suppress the impact of transient anomalies on system judgment, a time window smoothing mechanism can be introduced into the alarm exponential sequence. A valid fault is only identified when A(t) > θ at multiple consecutive time points. The specific expression is as follows: like This will trigger an alarm.

[0098] Where M is the window length; k is the minimum number of times the trigger threshold is satisfied; This is an indicator function; it is 1 if the condition is met, and 0 otherwise.

[0099] In this embodiment, the alarm generation unit generates an alarm signal when the alarm conditions are met.

[0100] Alarm signals can be output in several ways, including: The control signal is sent to the host computer system. Trigger the audible and visual alarm device; Write to the system log for subsequent source tracing and analysis; Display the anomaly type and characteristic value in the human-machine interface.

[0101] Specifically, the alarm signal structure may include the following: Alarm timestamp; Alarm level; Trigger parameters and their values; Module source (temperature, pressure, or vibration); Alarm index value and corresponding threshold.

[0102] In some embodiments, the system may include an output alarm index curve and threshold line to help maintenance personnel quickly identify abnormal trends.

[0103] In summary, the alarm determination module in this embodiment is based on the preceding model construction and feature extraction, uses a multi-feature fusion mechanism to construct an alarm index, and determines whether an alarm signal is triggered by a fixed or dynamic threshold.

[0104] The alarm output module is used to output the alarm signal, record the alarm information, and send feedback to the system management terminal.

[0105] 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 marine diesel engine alarm system, characterized in that, include: The data acquisition module is used to collect various physical parameters during the operation of the ship's diesel engine; The data preprocessing module is used to perform noise reduction, normalization, and time synchronization processing on the physical parameters; The state modeling module is used to construct a mathematical model reflecting the operating state of the diesel engine based on the preprocessed physical parameters; The feature extraction module is used to extract feature information reflecting fault trends or disturbances from mathematical models; An alarm determination module is used to generate an alarm signal based on the feature information extracted from the mathematical model. The alarm output module is used to output the alarm signal, record the alarm information, and send feedback to the system management terminal.

2. The marine diesel engine alarm system according to claim 1, characterized in that, The data acquisition module includes: Temperature acquisition unit is used to acquire temperature parameters at the measuring point location of the diesel engine; The pressure acquisition unit is used to acquire pressure parameters of the fuel system and lubrication system; The vibration acquisition unit is used to collect vibration parameters of the rotating parts of the diesel engine.

3. The marine diesel engine alarm system according to claim 1, characterized in that, The data preprocessing module includes: The noise reduction module is used to filter out noise from the acquired physical parameters; The normalization module is used to unify the dimensions of the denoised physical parameters. The time synchronization module is used to perform time alignment processing on the normalized physical parameters.

4. The marine diesel engine alarm system according to claim 1, characterized in that, The state modeling module includes: The thermal modeling unit is used to establish a mathematical model characterizing the heat transfer characteristics of a diesel engine based on temperature parameters. The pressure modeling unit is used to establish a mathematical model characterizing the working pressure characteristics of a diesel engine based on pressure parameters. The vibration modeling unit is used to establish a mathematical model characterizing the dynamic response characteristics of a diesel engine based on vibration parameters.

5. A marine diesel engine alarm system according to claim 1, characterized in that, The feature extraction module includes a trend feature extraction unit, used to extract feature information reflecting the changing trend of the diesel engine's operating status from the mathematical model; The disturbance feature extraction unit is used to extract feature information reflecting the operating disturbance of the diesel engine from the mathematical model.

6. A marine diesel engine alarm system according to claim 5, characterized in that, The steps for extracting feature information reflecting the changing trend of diesel engine operating status from the mathematical model are as follows: Calculate the rate of change of the diesel engine's operating state based on the temperature, pressure, and vibration parameters in the mathematical model. The rate of change is smoothed to obtain the trend change signal; Based on a preset threshold, abrupt change points in the trend change signal are extracted to reflect the abrupt change trend of the diesel engine's operating status. The formula for calculating the rate of change is: Where r(t) is the rate of change at time t, X(t) is the parameter value at time t, X(t-Δt) is the parameter value of the previous sampling period, and Δt is the sampling time interval.

7. A marine diesel engine alarm system according to claim 5, characterized in that, The step of extracting feature information reflecting diesel engine operating disturbances from the mathematical model is as follows: Calculate the residuals of each parameter in the mathematical model; Perform a Fourier transform on the residuals to obtain the perturbation characteristic signal in the frequency domain; Extract the main frequency components in the frequency domain and identify the characteristic information reflecting the disturbance; The formula for calculating the residual is: ε(t)=X meas (t)-X model (t); Where ε(t) is the residual at time t, X meas (t) represents the actual measured value at time t, X model (t) represents the model prediction at time t.

8. A marine diesel engine alarm system according to claim 1, characterized in that, The alarm determination module includes a fault determination unit, used to determine whether the diesel engine has malfunctioned based on the trend feature information and disturbance feature information extracted by the feature extraction module. The threshold judgment unit is used to determine whether to trigger an alarm signal based on a preset alarm threshold. The alarm generation unit is used to generate an alarm signal when a diesel engine malfunction is detected.

9. A marine diesel engine alarm system according to claim 8, characterized in that, The step of determining whether to trigger an alarm signal based on a preset alarm threshold includes: Calculate the alarm index based on the characteristic information of the trend and the characteristic information of the disturbance; Compare the alarm index with the preset alarm threshold; When the alarm index exceeds the preset alarm threshold, an alarm signal is triggered. The formula for calculating the alarm index is as follows: A(t)=α·∣r(t)∣+β·|ε(t)|; Where A(t) is the alarm index at time t, r(t) is the rate of change at time t, ε(t) is the residual at time t, and α and β are weighting coefficients.

10. A marine diesel engine alarm system according to claim 1, characterized in that, The alarm output module includes: an alarm signal output unit, used to output alarm signals to the ship control system or terminal equipment; Alarm recording unit, used to record alarm events; The feedback control unit is used to send alarm information to the system management terminal and trigger corresponding emergency response operations.

Images