Marine diesel engine parameter datum line prediction method based on random forest algorithm

Abstract

The invention provides a marine diesel engine parameter datum line prediction method based on a random forest algorithm. The marine diesel engine parameter datum line prediction method based on the random forest algorithm comprises the following steps that a Waxtin 9L34DF dual-fuel supercharged engine is modeled and calibrated according to operation data of a real ship, and datum line data of an engine model under different working conditions are obtained in AVL-BOOST software; carrying out normalization processing on the datum line data; dividing the normalized data into a training set and a test set according to a certain proportion; inputting the training set data into the random forest prediction model, training the model, and stopping training and storing the training set data into the random forest prediction model after the random forest prediction model reaches a set number of iterations; and inputting the data of the test set into the random forest prediction model stored through training, classifying the data of the test set, and performing datum line prediction under a dynamic working condition to obtain a corresponding prediction result. According to the invention, the real environment in the cabin can be simulated by adopting a noise increasing mode according to the data collected in the model.

Description

Technical Field

[0001] The present invention relates to the technical field of prediction methods, and in particular to a parameter baseline prediction method for a dual-fuel engine under variable operating conditions based on random forests. Background Art

[0002] The acceleration of globalization and the frequent interaction of international trade have made the shipping industry the world's largest mode of trade and transportation, and it is also developing rapidly. At present, about 90% of China's import and export cargo volume relies on maritime transportation, and the shipping industry plays a vital supporting role in my country's economic development. With the rapid advancement of intelligent technologies such as the Internet of Things, big data and artificial intelligence, digitalization and intelligence are gradually penetrating into all aspects of the full life cycle management of ships, such as design, manufacturing and operation and maintenance. Diesel engines are the main power source of ships. Once a failure occurs, it will not only interfere with the normal operation of the entire power system, causing serious economic losses, but also threaten personal safety and cause major accidents. Therefore, accurate and efficient health status assessment is crucial to prevent ship power system failures and ensure safe and stable navigation of ships. It is also a key way to achieve intelligent upgrades of marine diesel engines. However, due to the complexity of the system and the immaturity of the technology, the health status assessment and maintenance of diesel engines on ships currently still mainly rely on parameter limit alarms, manual inspections and digital instrument monitoring. Such methods rely too much on high-level technicians, and it is difficult to accurately grasp the changes in the health status of diesel engines, and it is difficult to achieve effective early warning of failures.

[0003] As ship power systems become increasingly complex and develop in an intelligent direction, traditional fault monitoring methods can no longer meet the needs of modern ships for diesel engine operation and maintenance systems. This limitation has prompted fault monitoring to gradually shift to fault prediction. By introducing advanced fault prediction technology, potential fault hazards can be identified before diesel engine problems occur, thereby effectively improving the ship's health status assessment capabilities. This predictive maintenance not only helps to identify and solve potential fault problems in advance, but also significantly reduces economic losses and waste of human resources. Through accurate fault prediction, ships can be maintained and repaired before faults occur, avoiding ship shutdowns, operational interruptions and related high costs caused by faults. In addition, the intelligent prediction system can monitor the operating status of the diesel engine in real time and provide data-driven decision support, thereby optimizing maintenance plans and resource allocation, and improving overall operation and maintenance efficiency and safety. Summary of the invention

[0004] According to the technical problems mentioned in the above background technology, a parameter baseline prediction method for a dual-fuel engine under variable working conditions based on random forest is provided. The present invention uses random forest to predict the parameter baseline. This machine learning algorithm based on ensemble learning improves the accuracy and robustness of the overall prediction by integrating the prediction results of multiple decision trees.

[0005] The technical means adopted by the present invention are as follows:

[0006] The method for predicting the parameter baseline of a marine diesel engine based on a random forest algorithm comprises the following steps:

[0007] Step 1: Model and calibrate the Wärtsilä 9L34DF dual-fuel supercharged engine based on the actual ship's operating data, and obtain the baseline data of the engine model under different operating conditions in the AVL-BOOST software;

[0008] Step 2: Noise the baseline data collected in the AVL-BOOST software and normalize the baseline data;

[0009] Step 3: Divide the normalized data into training set and test set according to a certain ratio; input the training set data into the random forest prediction model to train the model. When the random forest prediction model reaches the set number of iterations, stop training and save the random forest prediction model;

[0010] Step 4: Input the test set data into the random forest prediction model saved through training, classify the test set data, perform baseline prediction under dynamic conditions, and obtain the corresponding prediction results.

[0011] Furthermore, the number of iterations, i.e. the number of decision trees, is 100.

[0012] Furthermore, in the step 1, a baseline of the prediction parameters is determined based on a random forest, that is, by integrating the prediction results of multiple decision trees.

[0013] Furthermore, the decision trees are trained using randomly sampled training data and randomly selected features.

[0014] Furthermore, the Bayesian optimization algorithm is used to optimize the random forest prediction model during the training process to find the optimal parameters.

[0015] Furthermore, the Bayesian optimization to automatically adjust the hyperparameters of the decision tree regression model includes the following steps:

[0016] S1: Randomly set the seed to ensure repeatable results;

[0017] S2: Define the hyperparameter space for the seeds in S1, i.e., define the hyperparameters that need to be optimized; the hyperparameters are the number of trees and the size of leaf nodes;

[0018] S3: Bayesian optimization execution: Use Bayesian optimization method to evaluate different hyperparameter combinations and select the best combination by expected improvement criterion;

[0019] S4: Training model: Use the obtained optimal hyperparameters to train a decision tree model;

[0020] S5: Prediction and denormalization: predict the test data and denormalize the results;

[0021] S6: Evaluate the performance of the model on the test set by calculating the root mean square error.

[0022] Furthermore, in step three, the certain ratio is 7:3.

[0023] Furthermore, the normalization process refers to mapping the prediction baseline coefficient to between [0, 1].

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] The parameter baseline prediction method of the dual-fuel engine under variable operating conditions based on RF of the present invention has the following advantages:

[0026] 1. Strong applicability: By adding noise to the data collected in the model, the real environment in the cabin can be simulated, the authenticity of the data can be improved, the data can be normalized, and the influence of different dimensions and orders of magnitude in each parameter on the diagnostic results can be eliminated.

[0027] 2. Efficient and accurate: Using (RF) random forest for feature extraction and training reduces the dependence on cabin staff, fully learns the information in the extracted data, and improves the accuracy and efficiency of baseline prediction.

[0028] 3. Good real-time performance: Compared with the output of parameter baselines under different working conditions in the model, the use of deep learning to predict the baseline can achieve real-time, fast and accurate calculation of the baseline values ​​and deviation rates of each parameter under any working condition. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 This is the simulation model of the Wärtsilä 9L34DF engine of the present invention.

[0031] Figure 2 This is a flowchart of the ship diesel engine baseline prediction of the present invention.

[0032] Figure 3 This is the original sample data set of the present invention.

[0033] Figure 4 It is the technical roadmap of the present invention.

[0034] Figure 5 This is the random forest structure diagram of the present invention.

[0035] Figure 6 Schematic diagram of prediction error of the present invention.

[0036] Figure 7 Schematic diagram of BMEP prediction results of the present invention; (a) is a diagram of BMEP training set prediction results; (b) is a diagram of BMEP test set prediction results.

[0037] Figure 8 Schematic diagram of the air cooler outlet temperature prediction result of the present invention; wherein (a) is the air cooler outlet temperature training set prediction result diagram; (b) is the air cooler outlet temperature test set prediction result diagram.

[0038] Fig. 9 Schematic diagram of the prediction result of the supercharger outlet pressure of the present invention; wherein (a) is the prediction result diagram of the supercharger outlet pressure training set; (b) is the prediction result diagram of the supercharger outlet pressure test set.

[0039] Fig.10 Schematic diagram of the cylinder exhaust temperature prediction result of the present invention; (a) is the cylinder exhaust temperature training set prediction result diagram; (b) is the cylinder exhaust temperature test set prediction result diagram.

[0040] Fig.11 Schematic diagram of the intake flow prediction results of the present invention; (a) is the intake flow training set prediction result diagram; (b) is the intake flow test set prediction result diagram.

[0041] Fig.12 Schematic diagram of the single-cylinder fuel consumption prediction result of the present invention; (a) is a diagram of the single-cylinder fuel consumption training set prediction result; (b) is a diagram of the single-cylinder fuel consumption test set prediction result. DETAILED DESCRIPTION

[0042] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0043] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0044] like Figures 1 to 12 As shown, the present invention provides a method for predicting a marine diesel engine parameter baseline based on a random forest algorithm, comprising the following steps:

[0045] Step 1: Model and calibrate the Wärtsilä 9L34DF dual-fuel supercharged engine based on the operating data of the actual ship, and obtain the baseline data of the engine model under different working conditions in the AVL-BOOST software; predict the baseline of the parameters based on random forest, that is, by integrating the prediction results of multiple decision trees. The decision trees are trained with randomly sampled training data and randomly selected features. During the training process, the Bayesian optimization algorithm is used to optimize the random forest prediction model to seek the optimal parameters. The Bayesian optimization to automatically adjust the hyperparameters of the decision tree regression model includes the following steps:

[0046] S1: Randomly set the seed to ensure repeatable results;

[0047] S2: Define the hyperparameter space for the seeds in S1, i.e., define the hyperparameters that need to be optimized; the hyperparameters are the number of trees and the size of leaf nodes;

[0048] S3: Bayesian optimization execution: Use Bayesian optimization method to evaluate different hyperparameter combinations and select the best combination by expected improvement criterion;

[0049] S4: Training model: Use the obtained optimal hyperparameters to train a decision tree model;

[0050] S5: Prediction and denormalization: predict the test data and denormalize the result; the normalization process refers to mapping the prediction baseline coefficient to between [0, 1].

[0051] S6: Evaluate the performance of the model on the test set by calculating the root mean square error.

[0052] Step 2: Noise the baseline data collected in the AVL-BOOST software and normalize the baseline data;

[0053] Step 3: Divide the normalized data into a training set and a test set according to a certain ratio (7:3); input the training set data into the random forest prediction model to train the model. When the random forest prediction model reaches the set number of iterations (the number of iterations, i.e., the number of decision trees, is 100), stop training and save the random forest prediction model;

[0054] Step 4: Input the test set data into the random forest prediction model saved through training, classify the test set data, perform baseline prediction under dynamic conditions, and obtain the corresponding prediction results.

[0055] Example 1

[0056] In this embodiment, the step of data collection is to collect data from a 9L34DF dual-fuel engine built in the AVL-BOOST engine simulation software, and compare the simulated engine model built in the AVL-BOOST with the test report of the bench engine. It is found that the error is within 5%, which meets the accuracy requirements and can truly reflect the actual state of the ship engine. In the BOOST software, by setting different engine working conditions, the baseline data under different working conditions are collected. Due to the particularity of the ship's operating environment, its data is easily affected by temperature, vibration, etc., so random noise is added to the data obtained by the simulator to improve the authenticity of the ship data. At the same time, in order to avoid the problem of slow algorithm convergence data due to different value ranges of different data types, all the collected ship engine data are normalized and the predicted baseline coefficient is mapped to between [0, 1].

[0057] 1. Data preprocessing

[0058] (1) Four operating conditions of 25%, 50%, 75%, and 100% were set in BOOST to compare and calibrate with the bench test data. Table 1 shows the actual and simulated values ​​of various parameters under different operating conditions in the fuel mode.

[0059] Table 1 shows the actual and simulated values ​​of various parameters under different working conditions in fuel mode

[0060]

[0061]

[0062] (2) Noise removal and normalization of data samples

[0063] The purpose of adding noise is to simulate the uncertainty and randomness of data in the real world to make the model more robust. Noise can help test the performance of the model when dealing with imperfect data, prevent the model from overfitting, and improve the generalization ability of the model.

[0064] Normalization processing uses a common normalization method called Min-Max Normalization, which aims to compress the data into a specified range, usually [0,1].

[0065] 2. Establish (RF) random forest baseline prediction model

[0066] In the MATLAB script file, first use the xlsread command to read the processed data set Excel into the workspace. There are a total of 4000 groups of data, which are divided into 0.7 for training set and 0.3 for test set. num_outputs is the number of models to be trained, and models is the model for storing each target variable. The Bayesian optimization algorithm is used to find the optimal number of decision trees in the random forest model. The number of decision trees is 100, the minimum number of leaf nodes is 1, and training is stopped when the number of decision trees reaches 100.

[0067] The overall random forest structure is as follows Figure 4 As shown in the figure, it consists of one hundred decision trees. The results of each decision tree are summarized and the final prediction result is determined by voting.

[0068] 3. Experiment and result analysis

[0069] After building the (RF) random forest model, the processed 2800 sets of data were input into the model for training, and the error curve obtained was as follows Figure 6 As shown, the training is terminated when the number of decision trees reaches 100. 1200 sets of test data are input into the (RF) random forest model to predict the parameter baseline under different working conditions. Figure 6 to Figure 11, the R2 of the model is very close to 1, indicating that the model explains almost all the variance and has an excellent fitting effect. The MAE is very small, indicating that the prediction error of the model is very small. The MBE is close to 0, indicating that the model has no significant systematic deviation. From the results of R2, MAE and MBE, the performance of the model is almost optimal, indicating that the present invention can accurately predict the baseline of different parameters of the ship engine under different working conditions.

[0070] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0071] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0072] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0073] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0074] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0075] If the integrated unit is implemented in the form of 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 the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A marine diesel engine parameter baseline prediction method based on random forest algorithm, characterized in that: The following steps are involved: Step 1: Model and calibrate the Wärtsilä 9L34DF dual-fuel supercharged engine based on the actual ship's operating data, and obtain the baseline data of the engine model under different operating conditions in the AVL-BOOST software; Step 2: Noise the baseline data collected in the AVL-BOOST software and normalize the baseline data; Step 3: Divide the normalized data into training set and test set according to a certain ratio; input the training set data into the random forest prediction model to train the model. When the random forest prediction model reaches the set number of iterations, stop training and save the random forest prediction model; Step 4: Input the test set data into the random forest prediction model saved through training, classify the test set data, perform baseline prediction under dynamic conditions, and obtain the corresponding prediction results.

2. The method for predicting the parameter baseline of a marine diesel engine based on a random forest algorithm according to claim 1, characterized in that: The number of iterations, i.e. the number of decision trees, is 100.

3. The method for predicting the parameter baseline of a marine diesel engine based on a random forest algorithm according to claim 2, characterized in that: In the step 1, a baseline of the prediction parameters is determined based on a random forest, that is, by integrating the prediction results of multiple decision trees.

4. The method for predicting the parameter baseline of a marine diesel engine based on a random forest algorithm according to claim 2, characterized in that: The decision trees are trained using randomly sampled training data and randomly selected features.

5. The method for predicting the parameter baseline of a marine diesel engine based on a random forest algorithm according to claim 2, characterized in that: During the training process, the Bayesian optimization algorithm is used to optimize the random forest prediction model and seek the optimal parameters.

6. The method for predicting the parameter baseline of a marine diesel engine based on a random forest algorithm according to claim 4, characterized in that: The Bayesian optimization to automatically adjust the hyperparameters of the decision tree regression model includes the following steps: S1: Randomly set the seed to ensure repeatable results; S2: Define the hyperparameter space for the seeds in S1, i.e., define the hyperparameters that need to be optimized; the hyperparameters are the number of trees and the size of leaf nodes; S3: Bayesian optimization execution: Use Bayesian optimization method to evaluate different hyperparameter combinations and select the best combination by expected improvement criterion; S4: Training model: Use the obtained optimal hyperparameters to train a decision tree model; S5: Prediction and denormalization: predict the test data and denormalize the results; S6: Evaluate the performance of the model on the test set by calculating the root mean square error.

7. The parameter baseline prediction method for a dual-fuel engine under variable operating conditions based on random forest according to claim 1, characterized in that: In the step three, the certain ratio is 7:

3.

8. The parameter baseline prediction method for a dual-fuel engine under variable operating conditions based on random forest according to claim 1, characterized in that: The normalization process refers to mapping the prediction baseline coefficient to between [0, 1].

Images