Real-time prediction method, system, medium and device for marine engine operating parameters based on DLinear algorithm

Abstract

This application provides a method, system, medium, and device for real-time prediction of marine engine operating parameters based on the DLinear algorithm. This application replaces explicit sequence decomposition with feature decoupling, simplifying multivariate prediction into multiple univariate prediction problems through feature decomposition, and uses linear mapping instead of complex time series modeling. This method simplifies the Dlinear algorithm, making it suitable for time series prediction of parameters with weak correlations. For parameters with stronger correlations, the original Dlinear algorithm structure can be used. This application collects parameter data from marine engines across different load ranges, establishing a feature decoupling prediction model and a trend / seasonal bilinear prediction model. This avoids the impact of full-load testing on engine lifespan, solves the problem of insufficient effective data samples for modeling, and can quickly capture the nonlinear characteristics of multivariate co-evolution.

Description

Technical Field

[0001] This application relates to the field of intelligent monitoring technology for marine power systems, and in particular to a method, system, medium and device for real-time prediction of marine engine operating parameters based on the DLinear algorithm, which is applicable to marine engine condition monitoring, fault early warning and energy efficiency optimization. Background Technology

[0002] Marine engines face significant challenges in full-load testing during actual operation. Traditional methods for 100% load testing have significant limitations: First, full-load testing requires the engine to operate at its limits for extended periods, accelerating wear and increasing the risk of failure, irreversibly impacting engine lifespan. Second, full-load conditions occur infrequently in actual navigation, resulting in a severe shortage of effective data samples for modeling, failing to meet the basic training data requirements of data-driven models. Furthermore, complex dynamic coupling relationships exist among engine operating parameters; changes in a single parameter can trigger chain reactions in multiple parameters, making it difficult for traditional models to accurately capture the nonlinear characteristics of this multivariate co-evolution. Existing prediction methods, based on physical mechanism simulation models, require precise understanding of the differential equations governing the engine's internal workings, resulting in high modeling complexity and limited generalization ability. Conventional time-series models, when handling multi-parameter predictions, suffer from an excessive number of parameters and are prone to overfitting under limited data conditions, making it difficult to meet the stringent real-time requirements of engineering scenarios while maintaining prediction accuracy.

[0003] Meanwhile, the high risk of wear makes 100% load testing unsustainable, leading to a scarcity of effective data. This scarcity further amplifies the difficulty of modeling, and the complex coupling relationships prevent both traditional models (physical mechanism simulation models and time-series models) from providing high-precision predictions under the constraints of small data sets and real-time requirements, thus forming a closed loop of technical bottlenecks. Therefore, full-load testing of marine engines faces three core obstacles: lifespan risks, data scarcity, and modeling difficulties. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the purpose of this application is to provide a method, system, medium and device for real-time prediction of marine engine operating parameters based on the DLinear algorithm, in order to solve the technical problems of life risk, data scarcity and modeling difficulties faced by marine engine full load condition testing.

[0005] To achieve the above and other related objectives, the first aspect of this application provides a real-time prediction method for marine engine operating parameters based on the DLinear algorithm, comprising: acquiring multi-load operating condition data of the marine engine to obtain engine parameter time series data X∈R. T×mWhere T is the time step and m is the parameter dimension; calculate the Pearson correlation coefficient between engine parameter feature variables; if the Pearson correlation coefficient is greater than a preset threshold, the engine parameter feature variables are strongly correlated, otherwise they are weakly correlated; for weakly correlated parameters, a non-overlapping sliding window is used to truncate the original sequence, and the window sample is input into the feature decoupling prediction model based on the DLinear algorithm. After the linear prediction layer processes the input data, single-channel prediction values ​​are obtained, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of marine engine operating parameters; for strongly correlated parameters, each parameter is decomposed into trend components and seasonal components by moving average; the trend component is enhanced based on the gradual change effect, and the seasonal component is noise suppressed based on the marine environment; the parameter matrix is ​​regularized based on physical constraint coupling, so as to dynamically weight and fuse with ship operating condition factors to output the predicted value of each parameter.

[0006] In some embodiments of the first aspect of this application, the multi-load operating condition data acquisition of the marine engine includes: real-time acquisition of the marine engine's operating data under various different engine loads; the marine engine's operating data includes any one or more of the following: combustion system operating data, turbocharging / intake system operating data, fuel injection / common rail system operating data, and exhaust / turbine system operating data.

[0007] In some embodiments of the first aspect of this application, for weakly correlated parameters, a non-overlapping sliding window is used to truncate the original sequence, the window samples are input into a feature decoupling prediction model based on the DLinear algorithm, the linear prediction layer processes the input data to obtain single-channel prediction values, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of the marine engine operating parameters. This includes: performing data fusion processing on the collected multi-load operating condition data; standardizing the non-full-load operating conditions as the training and testing sets; and using the full-load operating conditions as the input to the prediction model; constructing a time series window to divide the continuous operating condition data of the marine engine into several window samples; inputting the window samples into a feature decoupling prediction model based on the DLinear algorithm; the feature decoupling prediction model decouples the input data to obtain single-channel prediction values, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of the marine engine operating parameters.

[0008] In some embodiments of the first aspect of this application, the feature decoupling prediction model includes: an input layer, the size of which is batch_size × seq_len × num_features; where batch_size represents the batch size, seq_len represents the time step size, and num_features represents the number of variables; a decomposition layer, connected after the input layer, for performing feature dimension decomposition along the last dimension of the input data to obtain num_features univariate sequences; a univariate linear prediction layer, connected after the decomposition layer, for independently feeding each univariate sequence into the corresponding one-dimensional linear prediction layer for prediction, and obtaining the corresponding univariate prediction results; and an output concatenation layer, connected after the univariate linear prediction layer, for concatenating all univariate prediction results and outputting a total prediction result with a tensor shape of batch_size × pred_len × num_features.

[0009] In some embodiments of the first aspect of this application, the step of decomposing each strongly correlated parameter into a trend component and a seasonal component by moving average; enhancing the trend component based on the gradual change effect and suppressing noise in the seasonal component based on the marine environment includes: in the trend term branch, firstly extracting the trend component by moving average decomposition, then enhancing the gradual change effect by a trend enhancement module, and sending the enhanced trend component into a multivariate trend linear layer for prediction to obtain a predicted value of trend change; in the seasonal term branch, firstly extracting the seasonal component by moving average decomposition, then suppressing environmental noise in the seasonal component; and sending the noise-suppressed seasonal component into a seasonal linear layer for prediction to obtain a predicted value of seasonal change.

[0010] In some embodiments of the first aspect of this application, the step of regularizing the parameter matrix based on physical constraint coupling to dynamically weight and fuse it with ship operating condition factors to output the predicted value of each parameter includes: inputting the predicted value of the trend change of the trend term branch and the predicted value of the seasonal change of the seasonal term branch into the parameter matrix regularization, and dynamically weighting and fusing the regularized prediction results with the ship operating condition factors to obtain the final prediction result of the ship engine parameters.

[0011] In some embodiments of the first aspect of this application, the loss function used in the backpropagation of parameter prediction for strongly correlated parameters is: L total =L pred +λ·L reg Among them, L pred It is the prediction loss, representing the difference between the model's output value and the true value; L regλ is the regularization loss, representing a penalty term used to limit the complexity of the model parameters; λ is a hyperparameter used to balance the importance of prediction loss and regularization loss.

[0012] To achieve the above and other related objectives, a second aspect of this application provides a real-time prediction system for marine engine operating parameters based on the DLinear algorithm, comprising: an operating condition data acquisition module for acquiring multi-load operating condition data of the marine engine and obtaining engine parameter time series data X∈R. T×m The system is structured as follows: T is the time step, and m is the parameter dimension. The correlation calculation module calculates the Pearson correlation coefficient between engine parameter feature variables. If the Pearson correlation coefficient is greater than a preset threshold, the engine parameter feature variables are strongly correlated; otherwise, they are weakly correlated. The weak correlation parameter prediction module uses a non-overlapping sliding window to extract the original sequence for weakly correlated parameters. The windowed samples are input into a feature decoupling prediction model based on the DLinear algorithm. The linear prediction layer processes the input data to obtain single-channel prediction values, which are then fused to obtain the predicted values ​​of the marine engine operating parameters. The strong correlation parameter prediction module decomposes each strongly correlated parameter into a trend component and a seasonal component using a moving average. The trend component is enhanced based on the gradual change effect, and the seasonal component is noise-suppressed based on the marine environment. The parameter matrix is ​​regularized based on physical constraint coupling to dynamically weight and fuse with ship operating condition factors to output the predicted value of each parameter.

[0013] To achieve the above and other related objectives, a third aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the real-time prediction method for marine engine operating parameters based on the DLinear algorithm.

[0014] To achieve the above and other related objectives, a fourth aspect of this application provides a computer program product comprising computer program code that, when executed on a computer, enables the computer to implement the real-time prediction method for marine engine operating parameters based on the DLinear algorithm.

[0015] To achieve the above and other related objectives, a fifth aspect of this application provides a computer device, including a memory, a processor, and a computer program stored in the memory; the processor executes the computer program to implement the real-time prediction method for marine engine operating parameters based on the DLinear algorithm.

[0016] As described above, the real-time prediction method, system, medium, and device for marine engine operating parameters based on the DLinear algorithm of this application have the following beneficial effects: This application replaces explicit sequence decomposition with feature decoupling, simplifying multivariate prediction into multiple univariate prediction problems through feature decomposition, and using linear mapping instead of complex time series modeling. This method simplifies the Dlinear algorithm, making it suitable for time series prediction of parameters with weak correlations, while the original Dlinear algorithm structure can be used for predicting parameters with stronger correlations. By collecting parameter data from different load ranges of marine engines, this application establishes a feature decoupling prediction model and a trend / seasonal bilinear prediction model, which can avoid the impact of full-load testing on engine life, solve the problem of insufficient effective data samples for modeling, and quickly capture the nonlinear characteristics of multivariate co-evolution. Attached Figure Description

[0017] Figure 1 The diagram shown is a flowchart illustrating a real-time prediction method for marine engine operating parameters based on the DLinear algorithm in one embodiment of this application.

[0018] Figure 2 This diagram illustrates a real-time prediction of engine operating parameters applicable to situations where the correlation between features is weak, as shown in one embodiment of this application.

[0019] Figure 3 The diagram shown is an architectural schematic of a DLinear feature decoupling prediction model according to an embodiment of this application.

[0020] Figure 4 The diagram shown is an embodiment of this application applicable to real-time prediction of engine operating parameters with strong correlations between features.

[0021] Figure 5 The diagram shows the flow of data in a strongly correlated parameter prediction model according to one embodiment of this application.

[0022] Figure 6 The diagram shown is a schematic representation of a real-time prediction system for marine engine operating parameters based on the DLinear algorithm, as described in one embodiment of the application.

[0023] Figure 7 The diagram shown is a structural schematic of a computer device according to an embodiment of this application. Detailed Implementation

[0024] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0025] Before providing a further detailed description of the present invention, the nouns and terms used in the embodiments of the present invention are explained, and the nouns and terms used in the embodiments of the present invention are subject to the following interpretations:

[0026] <1> DLinear (Decoupled Linear Network) is a lightweight architecture for multivariate time series prediction. Its core design lies in feature decoupling, that is, by explicitly separating the temporal dynamics and cross-variable static associations in the sequence, it can achieve high-precision and high-real-time prediction with a very low number of parameters under limited data conditions.

[0027] <2> Full load test: A standard test procedure that applies 100% load at the engine's rated speed and continues to operate until thermal equilibrium is reached in order to determine its maximum output power, fuel consumption, and emission characteristics.

[0028] To facilitate understanding of the embodiments of this application, firstly, in conjunction with Figure 1 Detailed explanation. Figure 1 The flowchart of a real-time prediction method for marine engine operating parameters based on the DLinear algorithm, as illustrated in an embodiment of this application, includes the following steps:

[0029] Step S11: Perform multi-load operating condition data acquisition on the marine engine to obtain engine parameter time series data X∈R T×m , where T is the time step and m is the parameter dimension.

[0030] It should be noted that marine engine operating data includes, but is not limited to: combustion system operating data, turbocharging / intake system operating data, fuel injection / common rail system operating data, and exhaust / turbine system operating data. Combustion system operating data includes, for example, in-cylinder temperature and cylinder pressure data; turbocharging / intake system operating data includes, for example, scavenging air pressure, scavenging air temperature, and scavenging air box dew point data; fuel injection / common rail system operating data includes, for example, fuel pressure and common rail pressure fluctuation data; exhaust / turbine system operating data includes, for example, exhaust temperature and exhaust temperature difference between cylinders; and cooling system operating data includes, for example, coolant temperature and coolant inlet / outlet temperature difference data. It should be understood that marine engine operating data encompasses a considerable amount of data, which will not be listed exhaustively here.

[0031] In the embodiments of this application, the multi-load operating condition data acquisition of marine engines refers to the real-time acquisition of marine engine operating data under various different engine loads. This obtains raw data covering the typical operating range of the engine, providing rich samples for subsequent modeling. For example, marine engine operating data can be acquired in real time under three different engine loads: 25%, 50%, and 75%, such as in-cylinder temperature data, scavenging air pressure data, fuel pressure data, exhaust temperature data, and coolant temperature data.

[0032] The final obtained engine parameter time series data is represented as a matrix X∈R T×m In this matrix, T represents the time step (i.e., the number of time points for data acquisition), and m represents the parameter dimension (i.e., the number of engine parameters recorded at each time point). Such data acquisition is crucial for analyzing engine performance, predicting potential faults, and optimizing its operating efficiency.

[0033] Step S12: Calculate the Pearson correlation coefficient between engine parameter characteristic variables; if the Pearson correlation coefficient is greater than the preset threshold, the engine parameter characteristic variables are strongly correlated, otherwise they are weakly correlated.

[0034] For example, in this embodiment, the preset threshold for the Pearson correlation coefficient is set to 0.7. Therefore, if the Pearson correlation coefficient between two feature variables is greater than 0.7, it indicates a strong correlation between the two feature variables; while if the Pearson correlation coefficient between two feature variables is less than or equal to 0.7, it indicates a weak correlation between the two feature variables. In this way, it is possible to identify which engine parameter feature variables have a significant linear relationship.

[0035] It should be noted that the Pearson correlation coefficient is a statistic that measures the strength and direction of the linear relationship between two variables. For time series data of marine engine parameters, calculating the Pearson correlation coefficient between characteristic variables can determine the relationship between different parameters. The Pearson correlation coefficient ranges from (-1, 1), where 1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no linear correlation. In practice, this can be achieved using correlation functions in statistical software or programming languages. The calculated coefficient matrix will show the correlation strength between all characteristic variables.

[0036] Step S13: For weakly correlated parameters, the original sequence is truncated using a non-overlapping sliding window. The window sample is input into the feature decoupling prediction model based on the DLinear algorithm. The linear prediction layer processes the input data to obtain single-channel prediction values, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of the marine engine operating parameters.

[0037] Specifically, firstly, the original sequence is truncated using a non-overlapping sliding window, dividing the original data into multiple independent sample windows. Each window contains a certain number of data points, and there is no data overlap between adjacent windows. This non-overlapping partitioning method ensures that the data within each sample window has a certain degree of independence, avoiding redundant information interference caused by data duplication. Next, these window samples are input into a feature decoupling prediction model based on the DLinear algorithm. In this model, the input sample data first undergoes feature decoupling processing, separating different features in the data to better capture the independence and interactions between each feature, providing a clearer data structure for subsequent predictions. Subsequently, a linear prediction layer processes the decoupled feature data. The role of the linear prediction layer is to generate single-channel prediction values ​​based on the input feature data through linear combination. Each channel corresponds to a specific feature or parameter, and the linear prediction layer can independently predict for each channel, thus obtaining the prediction results for each single channel. This single-channel prediction method can fully utilize the independent information of each feature, avoiding prediction errors caused by complex interactions between features. Finally, the predicted values ​​from each individual channel are fused to obtain the final predicted values ​​for the marine engine's operating parameters. The fusion process combines multiple individual channel predictions into a single, complete prediction result. This fusion method fully utilizes the information from each channel while balancing the prediction results between different channels, thereby improving the overall accuracy and reliability of the prediction. Using this method, even with weakly correlated parameters, the operating parameters of marine engines can be effectively predicted, providing strong support for engine operation monitoring and maintenance.

[0038] For example, the real-time prediction step for engine operating parameters, which is applicable to features with weak correlations, is as follows: Figure 2 As shown, it includes steps 2a to 2d.

[0039] Step 2a) Collect data on the marine engine under multiple load conditions, such as 25% load, 50% load, 75% load, and a very small amount of 100% load.

[0040] Step 2b) Perform data fusion processing on the collected multi-load condition data, standardize the non-full load condition data as the training set and test set, and use the full load condition data as the input of the prediction model.

[0041] Specifically, data from different sensors, sampling frequencies, and even different load conditions undergo time alignment, missing value imputation, anomaly removal, and unit standardization to form a computable fused dataset. This solves the problems of data silos and heterogeneity, ensuring the consistency of subsequent algorithm inputs. The non-full load standardization refers to normalizing or recalibrating the data under different load conditions (excluding full load) to make the statistical distributions of data at 25%, 50%, and 75% loads comparable and concatenable. This eliminates the differences in dimensions or ranges caused by load variations, enabling the model to generalize across different load conditions.

[0042] Step 2c) Construct a time series window to divide the continuous operating data of the marine engine into several window samples.

[0043] Specifically, in the real-time parameter prediction process for marine engines, the construction of time-series windows plays a crucial role in transforming continuous data into supervised learning samples. For example, assuming the operating parameters of a marine engine are recorded in a time series, such as engine speed and temperature recorded once per second, if a 30-second non-overlapping sliding window is used, then each window contains 30 data points. That is, the first window contains data from the first second to the 30th second, the second window contains data from the 31st second to the 60th second, and so on. There is no overlap between windows, and the last data point of the previous window and the first data point of the next window are continuous but not repetitive.

[0044] Step 2d) Input the window sample into the feature decoupling prediction model based on the DLinear algorithm. The feature decoupling prediction model decouples the input data and obtains single-channel prediction values. The single-channel prediction values ​​are then fused to obtain the predicted values ​​of the marine engine operating parameters.

[0045] Specifically, the feature decoupling prediction model decomposes the input data into feature dimensions and establishes an independent channel for each variable, such as... Figure 2The parameter 1, ..., parameter N are independent channels. These independent channels can be, for example, independent channels for in-cylinder temperature, scavenging pressure, fuel pressure, exhaust temperature, coolant temperature, etc. The variables of each independent channel are fed into their respective independent linear prediction layers for processing, outputting the corresponding single-channel predicted value. Finally, all the single-channel predicted values ​​are fused to obtain the final predicted values ​​of the marine engine's operating parameters, such as... Figure 2 The predicted values ​​for parameter 1, ..., parameter N are shown. Correspondingly, these predicted values ​​can be, for example, predicted values ​​for in-cylinder temperature, scavenging pressure, fuel pressure, exhaust temperature, coolant temperature, etc.

[0046] The architecture of the DLinear feature decoupling prediction model is as follows: Figure 3 As shown, the model consists of an input layer, a hierarchical sub-layer, a variable-wise linear prediction layer, and an output concatenation layer. The main process involves independent linear mapping and output concatenation for each variable. The core idea is channel (variable) decoupling, i.e., breaking down a multivariate time series into N univariate sub-sequences, predicting each sub-sequence using a simplified linear prediction layer, and finally concatenating them back into a multivariate output. This model addresses the unique characteristics of multivariate time series prediction by establishing an independent linear learning channel for each key operating parameter. This design decomposes the complex multi-parameter coupled prediction problem into multiple parallel univariate prediction subtasks, effectively reducing model complexity. Each sub-channel learns the unique temporal evolution of that parameter through a dedicated linear transformation layer, avoiding interference between the dynamic characteristics of different parameters.

[0047] The input layer has an input data size of batch_size × seq_len × num_features, meaning that each sample in a batch has seq_len time steps and a total of num_features variables. Taking a marine engine input data of shape [128, 300, 12] as an example: batch_size of 128 indicates that 128 continuous navigation data segments are extracted at once during training; seq_len of 300 indicates that each slice contains 300 historical time steps. If sampling is done at 10Hz, seq_len of 300 means 30 seconds of continuous recording; num_features of 12 indicates the number of columns of the corresponding engine monitoring variables. For example, a typical marine low-speed two-stroke diesel engine will provide the following monitoring variables: speed characteristics (such as crankshaft speed, turbocharger speed), temperature characteristics (such as exhaust manifold temperature, coolant outlet temperature, piston ring temperature), pressure characteristics (such as maximum in-cylinder burst pressure, scavenging pressure, common rail fuel pressure), flow / valve position characteristics (such as fuel injection quantity, variable exhaust valve opening), and vibration / torque characteristics (such as shaft power, cylinder vibration acceleration). If all 12 types of signals are connected to the system, then num_features is 12.

[0048] The decomposition layer, connected after the input layer, is used to decompose the feature dimension along the last dimension of the input data to obtain num_features univariate sequences: the shape of the feature 1 sequence is batch_size×seq_len×1, the shape of the feature 2 sequence is batch_size×seq_len×1, ..., the shape of the feature N sequence is batch_size×seq_len×1.

[0049] A univariate linear prediction layer, connected after the decomposition layer, is used to independently feed each univariate sequence into the corresponding one-dimensional linear prediction layer for prediction. For example, feature sequence 1 is fed into linear prediction layer 1 and predicted value 1 is output; feature sequence 2 is fed into linear prediction layer 2 and predicted value 2 is output; ..., feature sequence N is fed into linear prediction layer N and predicted value N is output. The shape of predicted values ​​1, 2, ..., N is batch_size × pred_len × 1, where pred_len refers to the future time step to be predicted.

[0050] The output concatenation layer, connected after the univariate linear prediction layer, concatenates the N univariate prediction results in the last dimension to restore the multivariate form. The final output tensor has a shape of batch_size × pred_len × num_features. Specifically, each channel passes through its respective linear prediction layer to obtain the univariate future sequence: Where B (batch_size) represents the batch size, H (pred_len) represents the prediction step size, and C is the number of channels. In channel-level concatenation, C channels are... In the last dimension (channel dimension), directly splice them in sequence:

[0051] dim = -1); This will give you the final output. If downstream tasks require comprehensive decision-making (such as remaining lifetime estimation or fault alarm), it can be simply... The channel dimensions can be weighted by average, maximum value, or logical rule fusion, or the following can also be used: A lightweight post-processing network (such as a 1×1 convolutional or fully connected layer) is directly input for secondary fusion. However, it should be understood that all these operations occur outside of the DLinear main body; the model itself does not perform any cross-channel weight sharing or attention interaction internally. That is, DLinear's independent channel strategy ensures complete decoupling between variables.

[0052] In summary, DLinear's feature decoupling prediction process involves first splitting a multivariate sequence into univariate sequences, then inputting each univariate sequence into a linear prediction layer, and finally concatenating the individual predictions back into a multivariate sequence. The entire process involves zero convolution, zero attention, and zero recursion, achieving multivariate time series prediction with only N independent linear mappings. The number of parameters grows linearly with the number of variables, meeting the needs of small-data and real-time scenarios for ship engine parameter prediction.

[0053] Step S14: For strongly correlated parameters, decompose each parameter into trend component and seasonal component by moving average; enhance the trend component based on the gradual change effect, and suppress the noise of the seasonal component based on the marine environment; regularize the parameter matrix based on physical constraint coupling, and dynamically weight and fuse it with the ship operating condition factor to output the predicted value of each parameter.

[0054] Specifically, when processing time series data, it is often necessary to decompose the data into different components to better understand and analyze its inherent patterns. In this embodiment, each parameter is decomposed using a moving average, separating it into trend and seasonal components. This aims to more clearly identify long-term trends and periodic changes in the data, thus providing a foundation for further analysis and prediction. It should be understood that moving average decomposition is a method based on moving averages used to extract trend and seasonal information from time series data. The moving average method smooths data by calculating the average of a series of data points, thereby removing short-term fluctuations and highlighting long-term trends. In moving average decomposition, moving averages are first calculated, and then these averages are used to separate the trend and seasonal components. The trend component reflects the long-term trend of the data, such as an increase or decrease over time; the seasonal component reflects the recurring changes in the data within a fixed period, such as annual seasonal fluctuations or weekly periodic changes.

[0055] The real-time prediction steps for engine operating parameters with strong correlations between features are as follows: Figure 4 As shown:

[0056] In the trend term branch, the trend component T is first extracted by moving average decomposition. Then, the gradual change effect is enhanced by the trend enhancement module. The enhanced trend component is sent to the multivariate trend linear layer for prediction to obtain the predicted value of trend change.

[0057] It should be understood that the trend component is dominated by the slowly varying physical processes and cumulative effects of the ship engine, and the prediction accuracy can be improved by monitoring the curvature of the trend term. The trend enhancement module is a data analysis tool used to process and analyze the trend component in time series data. It improves the model's sensitivity to trend changes and prediction accuracy by extracting and strengthening trend features. Taking the cylinder temperature of a ship engine as an example, the trend enhancement module first uses the moving average method to calculate the average value of the cylinder temperature over the past few hours to extract the long-term trend of the cylinder temperature. Then, the extracted trend data is further processed, such as calculating the first or second difference of the cylinder temperature, thereby identifying the acceleration or deceleration of the cylinder temperature trend. It can also enhance the trend features by identifying the periodic changes in the cylinder temperature and large cylinder diameter compensation.

[0058] Furthermore, the multivariate trend linear layer is a statistical model used in time series analysis to analyze and predict trend changes in marine engine parameters. This model establishes a linear relationship by considering the values ​​of the marine engine parameter at different time points, as well as other engine parameters that may affect it. Taking the cylinder temperature of a marine engine as an example, the trend component of the cylinder temperature, after enhanced trend processing, is fed into the multivariate trend linear layer. The trend component of the cylinder temperature serves as the primary prediction target, while other engine parameters (such as engine speed and load) are used as explanatory variables. By analyzing historical data of these variables, the linear relationship between them and the cylinder temperature is learned. The purpose of this linear model is to learn the interactions between these variables and use these relationships to predict future changes in the target parameter.

[0059] In the seasonal component branch, the seasonal component S is first extracted by moving average decomposition. Then, environmental noise suppression is performed on the seasonal component (such as using filtering techniques or other signal processing methods to reduce noise interference caused by external marine environmental factors). The seasonal component after noise suppression is sent to the seasonal linear layer for prediction to obtain the predicted value of seasonal change.

[0060] Specifically, environmental noise suppression of the extracted seasonal components is crucial. External marine environmental factors, such as sudden storms or changes in ocean currents, can introduce noise into the data, interfering with the accurate identification of seasonal trends. Therefore, to reduce this noise interference, filtering techniques or other signal processing methods (such as low-pass filters, median filters, or wavelet exchange) can be used to remove or reduce this noise. The noise-suppressed seasonal components are then fed into a seasonal linear layer. Using the seasonal components as the primary input variables, a linear relationship is established to predict future trend changes. For example, if certain performance parameters of a ship's engine have been identified as exhibiting specific patterns of change at certain times of the year, the seasonal linear layer can predict the trends of these parameters at similar times in the future.

[0061] Subsequently, the predicted values ​​of trend changes in the trend term branch and seasonal changes in the seasonal term branch are both input into the parameter matrix regularization. The regularized prediction results are then dynamically weighted and fused with ship operating condition factors to obtain the final prediction results of the ship engine parameters. It should be noted that parameter matrix regularization is used to optimize model parameters, reduce the risk of overfitting, and improve the model's generalization ability. By applying regularization techniques, such as L1 (Lasso) or L2 (Ridge) regularization, model parameters can be adjusted to ensure that the model is not overly sensitive to training data, thus enabling accurate predictions even on new and unseen data.

[0062] After parameter regularization, the result is an optimized prediction of trend and seasonal variations. These predictions are then dynamically weighted and fused with ship operating condition factors. Ship operating condition factors refer to various conditions affecting the operating status of ship engines, such as sea state, weather, ship speed, and load. Dynamic weighted fusion weights the trend and seasonal predictions based on the current ship operating condition factors to generate the final prediction output. Therefore, dynamic weighted fusion allows the model to adjust the weights of different predictions according to actual conditions, thereby improving the accuracy and reliability of the predictions. Ultimately, the fused result serves as the predicted output for ship engine parameters. This output considers not only the long-term trend and seasonal variations of the data but also the current ship operating conditions, providing a comprehensive and accurate prediction for ship operations.

[0063] Furthermore, during model training, patterns in the data are learned by minimizing a loss function, which measures the difference between the model's predictions and the actual values. During backpropagation, gradients are calculated on the model parameters based on the total loss function, and these gradients are then used to update the model's weights. Through this iterative process, the model gradually learns patterns in the data, thereby improving its predictive accuracy. The loss function in this embodiment can be expressed as:

[0064] L total =L pred +λ·L reg ;Formula (1)

[0065] Among them, L pred It is the prediction loss, representing the difference between the model's output value and the true value; L reg λ is the regularization loss, representing a penalty term used to limit the complexity of model parameters, such as through L1 or L2 regularization. λ is a hyperparameter used to balance the importance of prediction loss and regularization loss.

[0066] The flow diagram of the data stream in the strongly correlated parameter prediction model is as follows: Figure 5 As shown:

[0067] First, forward propagation: Input ship engine parameters, and decompose each parameter using a moving average to obtain a trend component T and a seasonal component S. For the trend component T, trend enhancement is applied before inputting it into a multivariate trend linear layer for prediction; the predicted trend change value is then input into the parameter matrix. For the seasonal component S, noise suppression is applied before inputting it into a seasonal linear layer for prediction; the predicted seasonal change value is also input into the parameter matrix. After regularization of the parameter matrix, the regularized prediction values ​​are dynamically fused with ship operating condition factors and then output.

[0068] Secondly, backpropagation: a prediction loss function is constructed based on the output of the dynamic fusion, and a regularization loss function is constructed based on the parameter matrix. Then, the total loss function is constructed based on the prediction loss function and the regularization loss function. The total loss function is shown in formula (1) above. During backpropagation, the gradient of the model parameters is calculated based on the total loss function, and then these gradients are used to update the model weights.

[0069] The foregoing section provided a detailed explanation of a real-time prediction method for marine engine operating parameters based on the DLinear algorithm, as described in the embodiments of this application. The following section will further illustrate the real-time prediction system, device, and medium for marine engine operating parameters based on the DLinear algorithm, with reference to specific embodiments.

[0070] like Figure 6 The diagram illustrates the structure of a real-time prediction system for marine engine operating parameters based on the DLinear algorithm, according to an embodiment of this application. The real-time prediction system 600 for marine engine operating parameters includes: a condition data acquisition module 601, a correlation calculation module 602, a weak correlation parameter prediction module 603, and a strong correlation parameter prediction module 604.

[0071] The operating condition data acquisition module 601 is used to acquire multi-load operating condition data of marine engines and obtain engine parameter time series data X∈R. T×m , where T is the time step and m is the parameter dimension.

[0072] The correlation calculation module 602 is used to calculate the Pearson correlation coefficient between engine parameter feature variables; if the Pearson correlation coefficient is greater than a preset threshold, the engine parameter feature variables are strongly correlated, otherwise they are weakly correlated.

[0073] The weak correlation parameter prediction module 603 is used to extract the original sequence using a non-overlapping sliding window for weak correlation parameters, input the window sample into the feature decoupling prediction model based on the DLinear algorithm, and obtain single-channel prediction values ​​by the linear prediction layer after processing the input data. The single-channel prediction values ​​are then fused to obtain the predicted values ​​of the marine engine operating parameters.

[0074] The strongly correlated parameter prediction module 604 is used to decompose each strongly correlated parameter into a trend component and a seasonal component by performing a moving average; enhance the trend component based on the gradual change effect, and suppress the noise of the seasonal component based on the marine environment; and regularize the parameter matrix based on physical constraint coupling to dynamically weight and fuse it with the ship operating condition factor to output the predicted value of each parameter.

[0075] It should be understood that the specific processes by which each module performs the corresponding steps described above have been detailed in the above method embodiments, and will not be repeated here for the sake of brevity. It should also be understood that the module division in the embodiments of this application is illustrative and merely a logical functional division; other division methods may exist in actual implementation. Furthermore, the functional modules in the various embodiments of this application can be integrated into a single processor, exist as separate physical entities, or have two or more modules integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0076] It should be noted that, in the embodiments of this application, the words "exemplary" or "for example" indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0077] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0078] Figure 7 This is a schematic block diagram of a computer device / equipment / system provided in an embodiment of this application. For example... Figure 7As shown, the computer device includes at least one processor 701, a memory 702, at least one network interface 703, and a user interface 705. The various components in the device are coupled together via a bus system 704. It is understood that the bus system 704 is used to implement communication between these components. In addition to a data bus, the bus system 704 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 7 The general will label all buses as bus systems.

[0079] The user interface 705 may include a monitor, keyboard, mouse, trackball, clicker, button, touchpad, or touch screen.

[0080] It is understood that memory 702 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM) or programmable read-only memory (PROM), used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM) and synchronous static random access memory (SSRAM). The memories described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable categories of memory.

[0081] In this embodiment of the invention, the memory 702 is used to store various types of data to support the operation of the electronic terminal 700. Examples of this data include: any executable program for operation on the electronic terminal 700, such as operating system 7021 and application program 7022; operating system 7021 includes various system programs, such as framework layer, core library layer, driver layer, etc., for implementing various basic services and handling hardware-based tasks. Application program 7022 may include various applications, such as media player, browser, etc., for implementing various application services. The implementation of the XX method provided in this embodiment of the invention may be included in application program 7022.

[0082] The methods disclosed in the above embodiments of the present invention can be applied to or implemented by processor 701. Processor 701 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 701 or by instructions in software form. The processor 701 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 701 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. General-purpose processor 701 may be a microprocessor or any conventional processor, etc. The steps of the accessory optimization method provided in the embodiments of the present invention can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in memory. The processor reads the information in the memory and combines it with its hardware to complete the steps of the aforementioned method.

[0083] In an exemplary embodiment, the electronic terminal 700 may be used by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs) to execute the aforementioned method.

[0084] According to the method provided in the embodiments of this application, this application also provides a computer program product, which includes: computer program code, which, when run on a computer, causes the computer to execute the real-time prediction method for marine engine operating parameters based on the DLinear algorithm of any embodiment.

[0085] According to the method provided in the embodiments of this application, this application also provides a computer-readable storage medium storing program code, which, when run on a computer, causes the computer to perform the above-described method.

[0086] As used in this specification, the terms "component," "module," "system," etc., are used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, and / or a computer. As illustrated, applications running on computing devices and computing devices can both be components. One or more components may reside in a process and / or an execution thread, and components may be located on a single computer and / or distributed among two or more computers. Furthermore, these components can be executed from various computer-readable media on which various data structures are stored. Components can communicate, for example, via local and / or remote processes based on signals having one or more data packets (e.g., data from two components interacting with another component between a local system, a distributed system, and / or a network, such as the Internet interacting with other systems via signals).

[0087] Those skilled in the art will recognize that the various illustrative logical blocks and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0088] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0089] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0090] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0091] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0092] In the above embodiments, the functions of each functional unit can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. A computer program product includes one or more computer instructions (programs). When the computer program instructions (programs) are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs, DVDs), or semiconductor media (e.g., solid-state disks, SSDs, etc.).

[0093] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0094] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0095] In summary, this application provides a method, system, medium, and device for real-time prediction of marine engine operating parameters based on the DLinear algorithm. This application replaces explicit sequence decomposition with feature decoupling, simplifying multivariate prediction into multiple univariate prediction problems through feature decomposition, and uses linear mapping instead of complex time series modeling. This method simplifies the DLinear algorithm, making it suitable for time series prediction of parameters with weak correlations. For parameters with stronger correlations, the original DLinear algorithm structure can be used. By collecting parameter data from marine engines across different load ranges, this application establishes a feature decoupling prediction model and a trend / seasonal bilinear prediction model. This avoids the impact of full-load testing on engine lifespan, solves the problem of insufficient effective data samples for modeling, and can quickly capture the nonlinear characteristics of multivariate co-evolution. Therefore, this application effectively overcomes the various shortcomings of existing technologies and has high industrial application value.

[0096] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for real-time prediction of marine engine operating parameters based on the DLinear algorithm, characterized in that, include: Multi-load operating condition data acquisition of marine engines to obtain engine parameter time series data. , where T is the time step and m is the parameter dimension; Calculate the Pearson correlation coefficient between engine parameter characteristic variables; if the Pearson correlation coefficient is greater than a preset threshold, the engine parameter characteristic variables are strongly correlated; otherwise, they are weakly correlated. For weakly correlated parameters, a non-overlapping sliding window is used to truncate the original sequence. The window sample is input into the feature decoupling prediction model based on the DLinear algorithm. The linear prediction layer processes the input data to obtain single-channel prediction values, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of the marine engine operating parameters. For strongly correlated parameters, each parameter is decomposed into trend and seasonal components by moving average. The trend component is enhanced based on the gradual change effect, and the seasonal component is suppressed based on the marine environment. The parameter matrix is ​​regularized based on physical constraint coupling, and dynamically weighted and fused with ship operating condition factors to output the predicted value of each parameter. The aforementioned multi-load operating condition data acquisition for marine engines includes: Real-time acquisition of marine engine operating data under various engine loads; the marine engine operating data includes any one or more of the following: combustion system operating data, turbocharging / intake system operating data, fuel injection / common rail system operating data, and exhaust / turbine system operating data; the feature decoupling prediction model includes: an input layer, the size of which is batch_size×seq_len×num_features; batch_size represents the batch size, seq_len represents the time step size, and num_features represents the number of variables; a decomposition layer, connected after the input layer, used to decompose the input data along the last dimension of the feature dimension to obtain num_features univariate sequences; a univariate linear prediction layer, connected after the decomposition layer, used to independently feed each univariate sequence into the corresponding one-dimensional linear prediction layer for prediction, and obtain the corresponding univariate prediction results; and an output concatenation layer, connected after the univariate linear prediction layer, used to concatenate all univariate prediction results and output a total prediction result with a tensor shape of batch_size×pred_len×num_features.

2. The method for real-time prediction of marine engine operating parameters based on the DLinear algorithm according to claim 1, characterized in that, For weakly correlated parameters, a non-overlapping sliding window is used to truncate the original sequence. The windowed samples are input into a feature decoupling prediction model based on the DLinear algorithm. After the linear prediction layer processes the input data, single-channel prediction values ​​are obtained. The single-channel prediction values ​​are then fused to obtain the predicted values ​​of the marine engine operating parameters, which include: The collected multi-load operating condition data are fused, and the non-full-load operating conditions are standardized as training and testing sets, while the full-load operating conditions are used as the input to the prediction model. A time series window is constructed to divide the continuous operating condition data of the marine engine into several window samples. The window samples are input into a feature decoupling prediction model based on the DLinear algorithm. The feature decoupling prediction model decouples the input data to obtain single-channel prediction values, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of the marine engine operating parameters.

3. The method for real-time prediction of marine engine operating parameters based on the DLinear algorithm according to claim 1, characterized in that, For the strongly correlated parameters, each parameter is decomposed into a trend component and a seasonal component by a moving average. The method enhances the trend component based on the gradual variation effect and suppresses noise in the seasonal component based on the marine environment, including: In the trend term branch, the trend component is first extracted by moving average decomposition, and then the gradual change effect is enhanced by the trend enhancement module. The enhanced trend component is sent to the multivariate trend linear layer for prediction to obtain the predicted value of trend change. In the seasonal component branch, the seasonal component is first extracted by moving average decomposition, and then environmental noise is suppressed on the seasonal component. The seasonal component after noise suppression is sent to the seasonal linear layer for prediction to obtain the predicted value of seasonal change.

4. The real-time prediction method for marine engine operating parameters based on the DLinear algorithm according to claim 3, characterized in that, The method of regularizing the parameter matrix based on physical constraint coupling to dynamically weight and fuse it with ship operating condition factors to output the predicted value of each parameter includes: The predicted values ​​of the trend changes in the trend term branch and the predicted values ​​of the seasonal changes in the seasonal term branch are both input into the parameter matrix regularization. The regularized prediction results are then dynamically weighted and fused with the ship operating condition factors to obtain the final prediction results of the ship engine parameters.

5. The real-time prediction method for marine engine operating parameters based on the DLinear algorithm according to claim 4, characterized in that, The loss function used in backpropagation for parameter prediction of strongly correlated parameters is: ； in, Predictive loss represents the difference between the model's output value and the true value. It is the regularization loss, which represents a penalty term used to limit the complexity of the model parameters; It is a hyperparameter used to balance the importance of prediction loss and regularization loss.

6. A real-time prediction system for marine engine operating parameters based on the DLinear algorithm, characterized in that, include: The operating condition data acquisition module is used to acquire multi-load operating condition data of marine engines and obtain time series data of engine parameters. , where T is the time step and m is the parameter dimension; The correlation calculation module is used to calculate the Pearson correlation coefficient between engine parameter feature variables. If the Pearson correlation coefficient is greater than a preset threshold, the engine parameter feature variables are strongly correlated; otherwise, they are weakly correlated. The weak correlation parameter prediction module is used to extract the original sequence using a non-overlapping sliding window for weak correlation parameters. The window sample is input into the feature decoupling prediction model based on the DLinear algorithm. The linear prediction layer processes the input data to obtain single-channel prediction values, and the single-channel prediction values ​​are fused to obtain the predicted values ​​of marine engine operating parameters. The strong correlation parameter prediction module is used to decompose each strongly correlated parameter into a trend component and a seasonal component by performing a moving average. The trend component is enhanced based on the gradual change effect, and the seasonal component is suppressed based on the marine environment. The parameter matrix is ​​regularized based on physical constraint coupling, and dynamically weighted and fused with ship operating condition factors to output the predicted value of each parameter. The aforementioned multi-load operating condition data acquisition for marine engines includes: Real-time acquisition of marine engine operating data under various engine loads; the marine engine operating data includes any one or more of the following: combustion system operating data, turbocharging / intake system operating data, fuel injection / common rail system operating data, and exhaust / turbine system operating data; the feature decoupling prediction model includes: an input layer, the size of which is batch_size×seq_len×num_features; batch_size represents the batch size, seq_len represents the time step size, and num_features represents the number of variables; a decomposition layer, connected after the input layer, used to decompose the input data along the last dimension of the feature dimension to obtain num_features univariate sequences; a univariate linear prediction layer, connected after the decomposition layer, used to independently feed each univariate sequence into the corresponding one-dimensional linear prediction layer for prediction, and obtain the corresponding univariate prediction results; and an output concatenation layer, connected after the univariate linear prediction layer, used to concatenate all univariate prediction results and output a total prediction result with a tensor shape of batch_size×pred_len×num_features.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the real-time prediction method for marine engine operating parameters based on the DLinear algorithm as described in any one of claims 1 to 5.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the real-time prediction method for marine engine operating parameters based on the DLinear algorithm as described in any one of claims 1 to 5.

Images