Tractor traction characteristic prediction method and system based on LSTM

Abstract

The invention relates to an LSTM-based towing engine traction characteristic prediction method and system, and the method comprises the steps: collecting transmission system parameters and traction characteristic data of different models of tractors under various working conditions, carrying out the data preprocessing, normalization and characteristic selection, constructing a time series data set, carrying out the training through a three-layer LSTM neural network model, and carrying out the prediction of the traction characteristics of the tractors. According to the method, network parameters are adjusted through an Adam optimizer, overfitting is prevented through a Dropout mechanism, finally, traction characteristics can be predicted by inputting transmission system parameters, support is provided for performance optimization, operation scheduling and fault diagnosis of a tractor, the method has the advantages of being high in precision, automatic, suitable for variable working conditions and the like, and a new technical approach is provided for intelligent development of agricultural machinery.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural machinery, and particularly relates to a method and system for predicting the traction characteristics of a tractor based on LSTM. Background Art

[0002] The traction characteristics of a tractor are an important indicator to measure its working performance, directly affecting the operation efficiency and fuel economy. Traditional traction characteristic testing methods usually rely on field tests, which are costly, time-consuming, and greatly affected by environmental factors. With the development of data-driven technologies, using mathematical models and algorithms to predict traction characteristics has become an efficient and economical solution.

[0003] Currently, existing research has predicted traction characteristics by establishing a mathematical model of the tractor power transmission system. However, these methods usually rely on complex physical models, with large computational amounts and poor adaptability. In addition, existing methods fail to fully utilize the non-linear relationship between transmission system parameters and traction characteristics, resulting in limited prediction accuracy.

[0004] Therefore, there is an urgent need for a data-driven algorithm that can quickly and accurately predict the traction characteristics of a tractor based on transmission system parameters. Summary of the Invention

[0005] To solve the problems of the existing technology, the present invention proposes a method and system for predicting the traction characteristics of a tractor based on LSTM, and realizes the efficient prediction of traction characteristics by establishing a data-driven model between transmission system parameters and traction characteristics.

[0006] The purpose of the present invention and the solution to its technical problems are achieved by adopting the following technical solutions. The method for predicting the traction characteristics of a tractor based on LSTM proposed according to the present invention includes the following steps:

[0007] Step 1: Collect the transmission system parameters and traction characteristic data of different models of tractors under different working conditions;

[0008] Step 2: Preprocess the collected data, including removing noise data, scaling the data to the interval [0, 1] using the Min-Max normalization method, and selecting parameters with a correlation coefficient greater than 0.5 with the traction characteristics as input features based on correlation analysis;

[0009] Step 3: Construct a time series data set based on the transmission system parameters and traction characteristic data to ensure the time dependence and continuity of the data;

[0010] Step 4: Adopt a long short-term memory neural network model, input the transmission system parameters, and train the long short-term memory neural network model to learn the non-linear time dependence relationship between the transmission system parameters and the traction characteristics;

[0011] Step 5: Optimize the long short-term memory neural network model. Use the Adam optimizer to optimize the network parameters, set the learning rate to 0.001, and introduce the Dropout regularization mechanism to prevent overfitting. At the same time, adopt the early stopping mechanism to optimize the training process;

[0012] Step 6: Use the trained long short-term memory neural network model, input the parameters of the tractor transmission system, and predict the corresponding traction characteristics. The prediction results can be used for the performance optimization, operation scheduling, and fault diagnosis of the tractor.

[0013] Further, the transmission system parameters include engine speed, transmission setting, and tire size, and the traction characteristic data includes traction force, slip ratio, and traction efficiency.

[0014] Further, the long short-term memory neural network model includes an input layer, a hidden layer, and an output layer network structure; the input layer is used to receive the parameter data of the tractor transmission system, the hidden layer is used to capture the long-term time dependence relationship between the transmission system parameters and the traction characteristics, and the output layer is used to output the predicted values of the traction characteristics.

[0015] Further, the transmission system parameters also include transmission ratio and engine torque.

[0016] The tractor traction characteristic prediction system based on LSTM includes: a data acquisition module, a data preprocessing module, a data construction module, a long short-term memory neural network module, and a prediction module; the data preprocessing module is used to clean and normalize the collected raw data to meet the input requirements of the neural network; the data construction module is used to construct a time series data set according to the transmission system parameters and the traction characteristic data; the long short-term memory neural network module includes an input layer, a hidden layer, and an output layer, and is used to learn and predict the traction characteristics according to the transmission system parameters; the prediction module is used to input the transmission system parameters according to the trained long short-term memory neural network model and output the prediction results of the traction characteristics.

[0017] Further, it also includes an optimization module, which is used to optimize the parameters of the long short-term memory neural network through the Adam optimizer, set the learning rate to 0.001, introduce the Dropout regularization mechanism to prevent overfitting, and at the same time adopt the early stopping mechanism to optimize the training process.

[0018] Further, the data preprocessing module includes a data normalization module, which uses the Min-Max normalization method to scale the data to the interval [0,1] to ensure that the input data meets the input range of the LSTM network.

[0019] In summary, the present invention uses an LSTM neural network to model the non - linear relationship between the parameters of the tractor transmission system and the traction characteristics, and fully considers the dependence of the time series, having the following significant advantages:

[0020] High - precision prediction: LSTM can effectively capture the long - term dependence relationships in the time series, and has higher prediction accuracy compared with traditional physical models or empirical formulas.

[0021] Automation and intelligence: Without complex physical modeling, the system can automatically predict the traction characteristics according to the input transmission system parameters, and has strong adaptive capabilities.

[0022] Adapt to variable working conditions: With the input of different working conditions and parameters, the LSTM model can dynamically adjust the prediction results according to historical data to adapt to the variable working environment.

[0023] The above description is only an overview of the technical solution of the present invention. In order to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And to make the above - mentioned and other purposes, features and advantages of the present invention more obvious and understandable, the following preferred embodiments are specifically given, and in conjunction with the drawings, the details are described as follows. Brief Description of the Drawings

[0024] Figure 1 is the flowchart of the tractor traction characteristic prediction method based on LSTM of the present invention;

[0025] Figure 2 is the module schematic diagram of the tractor traction characteristic prediction system based on LSTM of the present invention;

[0026] Figure 3 is the detailed interaction relationship schematic diagram among the various modules in the tractor traction characteristic prediction method and system based on LSTM of the present invention;

[0027] Figure 4 is the expected effect diagram of using the tractor traction characteristic prediction method and system based on LSTM of the present invention. Detailed Embodiments

[0028] The following further describes the technical solution of the present invention in conjunction with the drawings and preferred embodiments.

[0029] Please refer to Figure 1 , the tractor traction characteristic prediction method based on LSTM specifically includes the following steps:

[0030] Step 1: Collect the transmission system parameters and traction characteristic data of tractors of different models through field tests. The transmission system parameters include engine speed, transmission settings, tire size, transmission ratio, engine torque, etc.; the traction characteristic data includes traction force, traction efficiency, slip rate, etc., ensuring the diversity and representativeness of the data.

[0031] Step 2: Preprocess the collected raw data; First, use a statistical analysis-based method to identify and remove noise data outside the normal range; then use the Min-Max normalization formula to scale the data to the [0,1] interval; finally, select the parameters with a correlation coefficient greater than 0.5 with the traction characteristics as input features based on correlation analysis;

[0032] The expression of the Min-Max normalization formula is:

[0033] Step 3: Construct a time series dataset based on the transmission system parameters and traction characteristic data. The time series dataset can reflect the time dependence of the data, which is crucial for the subsequent training of the LSTM model. The construction of the dataset ensures the continuity and consistency of the input data.

[0034] Step 4: Adopt a long short-term memory neural network (LSTM) model, input the transmission system parameters, and train the LSTM model to learn the non-linear time-dependent relationship between the transmission system parameters and the traction characteristics. In this embodiment, a 3-layer LSTM model with 256 neurons in each layer is adopted, and the number of training epochs is 200.

[0035] Step 5: Optimize the long short-term memory neural network (LSTM) model, use the Adam optimizer to optimize the model parameters, set the learning rate to 0.001, and at the same time introduce the Dropout regularization mechanism (Dropout rate 0.2) to prevent overfitting, and adopt the early stopping mechanism during the training process. When the loss of the validation set no longer decreases in 10 consecutive training epochs, stop the training; at the same time, through the cross-validation method, divide the dataset into a training set (70%), a validation set (15%), and a test set (15%). The validation metrics include:

[0036] Mean Squared Error (MSE):

[0037] Mean Absolute Error (MAE):

[0038] Coefficient of Determination (R 2 ):

[0039] where y i is the true value, is the predicted value, is the mean of the true values, n is the number of samples, the MSE of the model should be less than 0.05, and R 2 should be greater than 0.9 to prove the effectiveness of the model.

[0040] Step 6: Use the trained LSTM model, input the transmission system parameters, and output the prediction results of the traction characteristics.

[0041] Please refer to Figure 2 , the module of the LSTM-based tractor traction characteristic prediction system, including a data acquisition module, which is responsible for collecting the transmission system parameters and traction characteristic data of the tractor under different working conditions. This module obtains data through field tests to ensure the diversity and representativeness of the data;

[0042] A data preprocessing module that can clean and normalize the collected raw data. Data cleaning removes noise and outliers, and normalization scales the data to the range required for neural network input to improve the efficiency and stability of model training;

[0043] A data construction module that can construct a time series data set based on the transmission system parameters and traction characteristic data. This module ensures the time dependence of the data and provides a suitable data format for the training of the LSTM model;

[0044] The long short-term memory neural network (LSTM) module, which includes an input layer, an LSTM hidden layer, and an output layer. The input layer is used to receive the transmission system parameters, the LSTM hidden layer is used to capture the long-term dependencies in the time series data, and the output layer is used to generate the prediction results of the traction characteristics. This module optimizes the model parameters through the training data set to ensure the prediction accuracy;

[0045] The prediction module, according to the trained LSTM model, inputs the transmission system parameters and outputs the prediction results of the traction characteristics. The prediction results can be used for applications such as tractor performance optimization, operation scheduling, and fault diagnosis.

[0046] The optimization module optimizes the parameters of the LSTM neural network through the gradient descent method, namely the Adam optimizer, to minimize the prediction error. This module ensures that the model continuously adjusts the parameters during training to improve the prediction accuracy.

[0047] Please refer to Figure 3 , the interaction relationship between the various modules in the LSTM-based tractor traction characteristic prediction method and system. The specific interaction relationship is as follows:

[0048] The data collected by the data acquisition module is first transmitted to the data preprocessing module for cleaning and normalization. The preprocessed data is then sent to the data construction module to generate a time series dataset. Subsequently, the dataset generated by the data construction module is input into the LSTM neural network module for training and optimization. The trained model is used in the prediction module to output the prediction results of the traction characteristics. Then, the optimization module adjusts the model parameters to ensure the prediction accuracy and feeds the optimized model parameters back to the LSTM neural network module to further improve the model performance. Through the collaborative work among the above modules, the system of the present invention can efficiently predict the traction characteristics of the tractor, providing support for the performance optimization and operation scheduling of agricultural machinery.

[0049] The above are only the preferred embodiments of the present invention. Any person skilled in the art, without departing from the scope of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. The tractor traction characteristics prediction method based on LSTM is characterized by: The following steps are involved: Step 1: Collect transmission system parameters and traction characteristics data of different tractor models under different working conditions; Step 2: Preprocess the collected data, including removing noise data, scaling the data to the [0,1] interval using the Min-Max normalization method, and selecting parameters with a correlation coefficient greater than 0.5 with the traction characteristics as input features based on correlation analysis; Step 3: Construct a time series data set based on the transmission system parameters and traction characteristic data to ensure the time dependence and continuity of the data; Step 4: Using the long short-term memory neural network model, inputting the transmission system parameters, and training the long short-term memory neural network model to learn the nonlinear time dependence between the transmission system parameters and the traction characteristics; Step 5: Optimize the long short-term memory neural network model, use the Adam optimizer to optimize the network parameters, set the learning rate to 0.001, and introduce the Dropout regularization mechanism to prevent overfitting. At the same time, use the early stopping mechanism to optimize the training process; Step 6: Use the trained LSTM neural network model to input the tractor transmission system parameters and predict the corresponding traction characteristics. The prediction results can be used for tractor performance optimization, job scheduling and fault diagnosis.

2. The tractor traction characteristics prediction method based on LSTM according to claim 1 is characterized in that: The transmission system parameters include engine speed, transmission settings, and tire size, and the traction characteristic data include traction force, traction efficiency, and slip rate.

3. The tractor traction characteristic prediction method based on LSTM according to claim 1 is characterized in that: The long short-term memory neural network model includes an input layer, a hidden layer, and an output layer network structure; the input layer is used to receive parameter data of the tractor transmission system, the hidden layer is used to capture the long-term time dependency between the transmission system parameters and the traction characteristics, and the output layer is used to output the predicted value of the traction characteristics.

4. The tractor traction characteristic prediction method based on LSTM according to claim 1 is characterized in that: The transmission system parameters also include transmission ratio and engine torque.

5. The tractor traction characteristics prediction system based on LSTM is characterized by: include: Data acquisition module, data preprocessing module, data construction module, long short-term memory neural network module, prediction module; The data preprocessing module is used to clean and normalize the collected raw data to meet the input requirements of the neural network; The data construction module is used to construct a time series data set according to the transmission system parameters and traction characteristic data; The long short-term memory neural network module includes an input layer, a hidden layer and an output layer, and is used to learn and predict traction characteristics according to transmission system parameters; The prediction module is used to input transmission system parameters and output prediction results of traction characteristics according to the trained long short-term memory neural network model.

6. The tractor traction characteristics prediction system based on LSTM according to claim 5, characterized in that: It also includes an optimization module for optimizing the parameters of the long short-term memory neural network through the Adam optimizer, setting the learning rate to 0.001, and introducing the Dropout regularization mechanism to prevent overfitting, while using the early stopping mechanism to optimize the training process.

7. The tractor traction characteristics prediction system based on LSTM according to claim 5, characterized in that: The data preprocessing module includes a data normalization module, which uses the Min-Max normalization method to scale the data to the [0,1] interval to ensure that the input data conforms to the input range of the LSTM network.

Images