An intelligent agricultural machinery status monitoring method and system

Abstract

The present invention relates to the field of intelligent monitoring, specifically to an intelligent agricultural machinery status monitoring method and system. The method comprises the following steps: agricultural machinery operation data acquisition, signal preprocessing, feature extraction, deep feature learning, status monitoring, and fault diagnosis. During feature extraction, the present invention uses a GhostNet lightweight convolutional network to extract high-frequency vibration signal features, and a statistical feature extraction method for low-frequency operation signals, which are combined with high-frequency features for fusion. During status monitoring, a dynamic graph convolutional neural network plus a bidirectional LSTM model are used to model the status monitoring process, capturing the spatial dependency and temporal dynamic evolution of the agricultural machinery status. A hyperparameter search mechanism based on differential evolution optimization is designed, and high-fitness individuals are stored in a PCM memory pool, reducing manual intervention and improving generalization capabilities. The system comprises a data acquisition module, a signal preprocessing module, a feature extraction module, a deep feature learning module, a status monitoring module, and a fault diagnosis module.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring, and in particular to an intelligent agricultural machinery status monitoring method and system. Background Art

[0002] With the continuous improvement of the level of modern agricultural mechanization, the widespread application of agricultural machinery and equipment has greatly improved agricultural production efficiency. However, the operating status of agricultural machinery and equipment directly affects the efficiency and economic benefits of agricultural operations. Most of the existing agricultural machinery status monitoring methods rely on traditional signal processing methods, such as Fourier transform or wavelet transform, which can only extract limited frequency domain or time-frequency features and cannot fully explore the deep-level features in high-frequency vibration signals and low-frequency operation signals; traditional status monitoring methods are based on simple machine learning algorithms, which are difficult to capture complex time dependencies and cannot effectively model the dynamic evolution process of agricultural machinery operating status. In addition, the monitoring model lacks an intelligent optimization mechanism, the training process converges slowly, and the monitoring accuracy is limited. Summary of the Invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an intelligent agricultural machinery status monitoring method and system. During feature extraction, the GhostNet lightweight convolutional network is used to extract high-frequency vibration signal features, and the statistical feature extraction method is used for low-frequency operation signals, which are combined with high-frequency features for fusion to ensure the effective utilization of multi-source signals and improve the comprehensive performance of the monitoring model. During status monitoring, a dynamic graph convolutional neural network + bidirectional LSTM modeling state monitoring process is adopted to capture the spatial dependence and temporal dynamic evolution of the agricultural machinery status, improve the accuracy of status monitoring, design a hyperparameter search mechanism based on differential evolution optimization, store high-fitness individuals through the PCM memory pool, adaptively adjust hyperparameters such as learning rate, number of network layers, and number of neurons, reduce manual intervention, improve generalization ability, and introduce an archive disturbance mechanism. When the PCM memory pool has no significant optimization within multiple iteration cycles, a global search is triggered to improve the efficiency of hyperparameter search and make the status monitoring model more stable.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides an intelligent agricultural machinery status monitoring method, which includes the following steps:

[0005] Step S1: collecting agricultural machinery operation data, installing high-frequency vibration sensors and low-frequency signal sensors at key parts of the agricultural machinery to collect high-frequency vibration signals and low-frequency operation signals of the agricultural machinery respectively. The key parts include bearings, gears and engines;

[0006] Step S2: Signal preprocessing: using wavelet transform to perform multi-scale decomposition on the high-frequency vibration signal, and using smoothing filtering and denoising on the low-frequency operation signal to obtain the preprocessed high-frequency vibration signal and low-frequency operation signal;

[0007] Step S3: Feature extraction: GhostNet lightweight convolution is used to extract high-frequency features from the preprocessed high-frequency vibration signal, and the features are enhanced by combining ensemble learning and coordinate attention mechanism. Statistical feature extraction methods are used to extract low-frequency features from the preprocessed low-frequency operation signal, and the low-frequency features are fused to obtain the agricultural machinery operation features.

[0008] Step S4: Deep feature learning, using deep learning and self-supervised learning framework, combined with autoencoders to conduct deep learning on the operation characteristics of agricultural machinery to obtain deep features;

[0009] Step S5: Condition monitoring: construct a condition monitoring model based on deep features, analyze the operating status of agricultural machinery in real time, and obtain condition monitoring results;

[0010] Step S6: Fault diagnosis, building a fault diagnosis model, analyzing the status detection results, and obtaining analysis results and risk prediction.

[0011] Furthermore, step S4 specifically includes the following steps:

[0012] Step S41: Data preparation: Arrange the agricultural machinery operation characteristics into a feature matrix in chronological order, where rows represent time steps and columns represent different features. The agricultural machinery operation characteristics of each time step are regarded as a node.

[0013] Step S42: self-supervised target construction, generating two sets of positive samples through data augmentation, and using contrastive loss function to learn the node discrimination ability;

[0014] Step S43: constructing a spatiotemporal dependency structure, using a sliding window method to divide the agricultural machinery operation characteristics, constructing a time series data set, calculating the similarity matrix between time steps, and constructing a dynamic adjacency matrix;

[0015] Step S44: Autoencoder reconstruction: construct an autoencoder including an encoder and a decoder, perform dimensionality reduction on the encoder, and reconstruct the decoder, with the optimization objective being to minimize the reconstruction loss.

[0016] Step S45: deep feature extraction, using dynamic graph convolutional neural network combined with LSTM to obtain deep features;

[0017] Step S46: Output depth features.

[0018] Furthermore, step S45 specifically includes the following contents:

[0019] Dynamic graph convolutional neural network: This part takes in the dynamic adjacency matrix and feature matrix, updates node features through dynamic graph convolution operations, extracts spatial dependencies, and calculates the characteristic representation of agricultural machinery operation at each time step.

[0020] LSTM module: uses a two-layer bidirectional LSTM, inputs the time series dataset extracted from the dynamic graph convolutional neural network, models temporal dependencies, calculates long-term and short-term features, and extracts deep features;

[0021] Joint training: An end-to-end training strategy is adopted to partially extract spatial dependencies through a dynamic graph convolutional neural network, and then model temporal dependencies through LSTM.

[0022] Furthermore, step S5 is to construct a condition monitoring model based on the deep features, which specifically includes the following steps:

[0023] Step S51: constructing state features, combining the deep features of step S4 to construct a time series feature vector, and performing normalization processing;

[0024] Step S52: Initialize the condition monitoring model, input the time series feature vector processed in step S51, and use LSTM to build the condition monitoring model;

[0025] Step S53: Population initialization: randomly initialize an initial population, where each individual represents a set of hyperparameter configurations, and each individual is represented as follows:

[0026] ;

[0027] in, For the Individuals, is the learning rate, is the number of network layers, is the number of neurons in each layer, and is the control parameter of differential evolution;

[0028] Calculate the fitness of each individual in the initial population and obtain the fitness value;

[0029] Step S54: Establish a PCM memory pool, set the storage capacity to M, store the top M individuals with the highest current fitness values ​​into the PCM memory pool, set the maximum number of iterations and the global iteration variable, as shown below:

[0030] ;

[0031] when When , execute step S55 to step S57;

[0032] in, is a global iteration variable, is the maximum number of iterations, set to 200;

[0033] Step S55: Adaptive search, performing global search and local search, specifically including the following:

[0034] Global search: Perform standard differential evolution on each individual in the initial population. The formula used is as follows:

[0035] ;

[0036] in, 、 and is an individual randomly selected from the initial population, New individuals generated by global search;

[0037] Local search: Guide mutation of individuals in the PCM memory pool with a probability of 50%. The formula used is as follows:

[0038] ;

[0039] in, is an individual randomly selected from the PCM memory pool, New individuals generated by local search;

[0040] Step S56: Population update, calculation and The fitness value of Fitness value higher than , then replace Entering the initial population, if If the fitness of is higher than that of any individual in the PCM memory pool, it is replaced to obtain the replaced initial population and the replaced PCM memory pool;

[0041] Step S57: archive disturbance mechanism, monitoring the changes in the PCM memory pool, and triggering the archive disturbance mechanism when there is no optimization after 50 iterations;

[0042] Step S58: Output the result. When , the iteration ends and the individual with the highest fitness in the final PCM pool is output as the optimal hyperparameter combination.

[0043] The present invention provides an intelligent agricultural machinery status monitoring system, which includes a data acquisition module, a signal preprocessing module, a feature extraction module, a deep feature learning module, a status monitoring module, and a fault diagnosis module. Specifically, it includes the following contents:

[0044] The data acquisition module collects high-frequency vibration signals and low-frequency operation signals of the agricultural machinery;

[0045] The signal preprocessing module preprocesses the high-frequency vibration signal and the low-frequency operation signal;

[0046] The feature extraction module uses GhostNet lightweight convolution to extract high-frequency features from the preprocessed high-frequency vibration signal, and combines ensemble learning and coordinate attention mechanism to enhance features. It uses statistical feature extraction method to extract low-frequency features from the preprocessed low-frequency operation signal, and fuses them to obtain agricultural machinery operation features.

[0047] The deep feature learning module uses a deep learning and self-supervised learning framework, combined with an autoencoder, to perform deep learning on the operation characteristics of agricultural machinery to obtain deep features;

[0048] The state monitoring module builds a state monitoring model based on deep features, analyzes the operating state of agricultural machinery in real time, and obtains state monitoring results;

[0049] The fault diagnosis module constructs a fault diagnosis model, analyzes the status detection results, and obtains analysis results and risk predictions.

[0050] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0051] (1) When extracting features, the GhostNet lightweight convolutional network is used to extract high-frequency vibration signal features. The low-frequency operation signal adopts the statistical feature extraction method and is combined with high-frequency features for fusion to ensure the effective use of multi-source signals and improve the comprehensive performance of the monitoring model.

[0052] (2) In the state monitoring process, a dynamic graph convolutional neural network + bidirectional LSTM model is used to model the state monitoring process, capture the spatial dependence and temporal dynamic evolution of the agricultural machinery state, improve the accuracy of state monitoring, and design a hyperparameter search mechanism based on differential evolution optimization. The PCM memory pool is used to store high-fitness individuals, and hyperparameters such as the learning rate, number of network layers, and number of neurons are adaptively adjusted to reduce manual intervention and improve generalization ability. An archive disturbance mechanism is introduced. When the PCM memory pool has no significant optimization within multiple iteration cycles, a global search is triggered to improve the efficiency of the hyperparameter search and make the state monitoring model more stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A schematic flow chart of an intelligent agricultural machinery status monitoring method provided by the present invention;

[0054] Figure 2 Schematic diagram of the process of step S4;

[0055] Figure 3 This is a module schematic diagram of an intelligent agricultural machinery status monitoring system provided by the present invention.

[0056] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0057] The technical solutions 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, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0058] In the description of the present invention, it should be understood that terms such as "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0059] Example 1, see Figure 1 The present invention provides an intelligent agricultural machinery status monitoring method, which includes the following steps:

[0060] Step S1: collecting agricultural machinery operation data, installing high-frequency vibration sensors and low-frequency signal sensors at key parts of the agricultural machinery to collect high-frequency vibration signals and low-frequency operation signals of the agricultural machinery respectively. The key parts include bearings, gears and engines;

[0061] Step S2: Signal preprocessing: using wavelet transform to perform multi-scale decomposition on the high-frequency vibration signal, and using smoothing filtering and denoising on the low-frequency operation signal to obtain the preprocessed high-frequency vibration signal and low-frequency operation signal;

[0062] Step S3: Feature extraction: GhostNet lightweight convolution is used to extract high-frequency features from the preprocessed high-frequency vibration signal, and the features are enhanced by combining ensemble learning and coordinate attention mechanism. Statistical feature extraction methods are used to extract low-frequency features from the preprocessed low-frequency operation signal, and the low-frequency features are fused to obtain the agricultural machinery operation features.

[0063] Step S4: Deep feature learning, using deep learning and self-supervised learning framework, combined with autoencoders to conduct deep learning on the operation characteristics of agricultural machinery to obtain deep features;

[0064] Step S5: Condition monitoring: construct a condition monitoring model based on deep features, analyze the operating status of agricultural machinery in real time, and obtain condition monitoring results;

[0065] Step S6: Fault diagnosis, building a fault diagnosis model, analyzing the status detection results, and obtaining analysis results and risk prediction.

[0066] Example 2, see Figure 2 This embodiment is based on the above embodiment, and step S4 specifically includes the following steps:

[0067] Step S41: Data preparation: Arrange the agricultural machinery operation characteristics into a feature matrix in chronological order, where rows represent time steps and columns represent different features. The agricultural machinery operation characteristics of each time step are regarded as a node.

[0068] Step S42: self-supervised target construction, generating two sets of positive samples through data augmentation, and using contrastive loss function to learn the node discrimination ability;

[0069] Step S43: constructing a spatiotemporal dependency structure, using a sliding window method to divide the agricultural machinery operation characteristics, constructing a time series data set, calculating the similarity matrix between time steps, and constructing a dynamic adjacency matrix;

[0070] Step S44: Autoencoder reconstruction: construct an autoencoder including an encoder and a decoder, perform dimensionality reduction on the encoder, and reconstruct the decoder, with the optimization objective being to minimize the reconstruction loss.

[0071] Step S45: deep feature extraction, using dynamic graph convolutional neural network combined with LSTM to obtain deep features;

[0072] Step S46: Output depth features.

[0073] In this embodiment, a "John Deere 6M Series" tractor is used as the experimental object, and engine speed, fuel consumption rate, hydraulic system pressure, transmission temperature, vehicle speed and ambient temperature are collected;

[0074] Data was collected through sensors on the tractor for 2 hours, with a total of 72,000 time steps collected;

[0075] Data storage is represented as follows:

[0076] Time step Engine speed Fuel consumption rate Hydraulic system pressure Transmission temperature Speed Ambient temperature 1 1500 7.5 8.0 50 6.2 18 2 1510 7.6 8.2 50.2 6.3 18.1 3 1498 7.4 8.1 50.1 6.1 18 4 1520 7.8 8.3 50.3 6.4 18.2 5 1480 7.3 8.0 50.0 6.0 18

[0077] Set the window size to 5 seconds (50 time steps) and the sliding step size to 1 second;

[0078] We get 72000-50+1 = 71951 time segments.

[0079] Add Gaussian noise to the engine speed and time-shift some data to simulate clock drift;

[0080] Using NT-Xent;

[0081] Calculate the Euclidean distance between different time steps and set a threshold: establish a connection when the Euclidean distance is < 0.5, and construct a dynamic adjacency matrix;

[0082] Autoencoder reconstruction:

[0083] Encoder: 2 fully connected layers: FC1: 6×128, FC2: 128×64;

[0084] Decoder: 2 fully connected layers: FC1: 64×128, FC2: 128×6;

[0085] The optimization objective adopts the mean square error function;

[0086] A 3-layer dynamic graph convolutional neural network is used: the hidden layer sizes are 64, 32, and 16;

[0087] 2-layer LSTM: 64 hidden units, 50 time steps, and 128 output dimensions;

[0088] LSTM output is 128-dimensional;

[0089] The dynamic graph convolutional neural network output is 16-dimensional;

[0090] The final depth feature vector is 144-dimensional;

[0091] Using K-Means clustering:

[0092] K=3

[0093] Category: normal state, mild abnormality (e.g. fuel consumption increased by 5%), severe abnormality (e.g. hydraulic system pressure decreased by 20%);

[0094] Calculate the Mahalanobis distance and generate an alarm when the Mahalanobis distance is greater than 3.

[0095] Embodiment 3: This embodiment is based on the above embodiment. Step S45 specifically includes the following contents:

[0096] Dynamic graph convolutional neural network: This part takes in the dynamic adjacency matrix and feature matrix, updates node features through dynamic graph convolution operations, extracts spatial dependencies, and calculates the characteristic representation of agricultural machinery operation at each time step.

[0097] LSTM module: uses a two-layer bidirectional LSTM, inputs the time series dataset extracted from the dynamic graph convolutional neural network, models temporal dependencies, calculates long-term and short-term features, and extracts deep features;

[0098] Joint training: An end-to-end training strategy is adopted to partially extract spatial dependencies through a dynamic graph convolutional neural network, and then model temporal dependencies through LSTM.

[0099] Example 4: This example is based on the above example. In step S5, a condition monitoring model is constructed based on deep features, specifically including the following steps:

[0100] Step S51: constructing state features, combining the deep features of step S4 to construct a time series feature vector, and performing normalization processing;

[0101] Step S52: Initialize the condition monitoring model, input the time series feature vector processed in step S51, and use LSTM to build the condition monitoring model;

[0102] Step S53: Population initialization: randomly initialize an initial population, where each individual represents a set of hyperparameter configurations, and each individual is represented as follows:

[0103] ;

[0104] in, For the Individuals, is the learning rate, is the number of network layers, is the number of neurons in each layer, and is the control parameter of differential evolution;

[0105] Calculate the fitness of each individual in the initial population and obtain the fitness value;

[0106] Step S54: Establish a PCM memory pool, set the storage capacity to M, store the top M individuals with the highest current fitness values ​​into the PCM memory pool, set the maximum number of iterations and the global iteration variable, as shown below:

[0107] ;

[0108] when When , execute step S55 to step S57;

[0109] in, is a global iteration variable, is the maximum number of iterations, set to 200;

[0110] Step S55: Adaptive search, performing global search and local search, specifically including the following:

[0111] Global search: Perform standard differential evolution on each individual in the initial population. The formula used is as follows:

[0112] ;

[0113] in, 、 and is an individual randomly selected from the initial population, New individuals generated by global search;

[0114] Local search: Guide mutation of individuals in the PCM memory pool with a probability of 50%. The formula used is as follows:

[0115] ;

[0116] in, is an individual randomly selected from the PCM memory pool, New individuals generated by local search;

[0117] Step S56: Population update, calculation and The fitness value of Fitness value higher than , then replace Entering the initial population, if If the fitness of is higher than that of any individual in the PCM memory pool, it is replaced to obtain the replaced initial population and the replaced PCM memory pool;

[0118] Step S57: archive disturbance mechanism, monitoring the changes in the PCM memory pool, and triggering the archive disturbance mechanism when there is no optimization after 50 iterations;

[0119] Step S58: Output the result. When , the iteration ends and the individual with the highest fitness in the final PCM pool is output as the optimal hyperparameter combination.

[0120] In this embodiment, the codes used are as follows:

[0121] import numpy as np

[0122] import tensorflow as tf

[0123] from tensorflow.keras.models import Sequential

[0124] from tensorflow.keras.layers import LSTM, Dense

[0125] from sklearn.preprocessing import MinMaxScaler

[0126] # Step S51: State feature construction, time series feature vector normalization

[0127] def normalize_data(data):

[0128] scaler = MinMaxScaler(feature_range=(0, 1))

[0129] return scaler.fit_transform(data)

[0130] # Step S52: Build LSTM model

[0131] def create_lstm_model(input_shape, layers, neurons):

[0132] model = Sequential()

[0133] model.add(LSTM(neurons, input_shape=input_shape, return_sequences=True))

[0134] for _ in range(layers - 1):

[0135] model.add(LSTM(neurons, return_sequences=True))

[0136] model.add(Dense(1))

[0137] model.compile(optimizer='adam', loss='mean_squared_error')

[0138] return model

[0139] # Step S53: Population initialization, generating the initial population

[0140] def initialize_population(pop_size, param_bounds):

[0141] population = []

[0142] for _ in range(pop_size):

[0143] # Randomly initialize each hyperparameter individual

[0144] l_r = np.random.uniform(param_bounds['l_r'][0], param_bounds['l_r'][1]) # learning rate

[0145] layers = np.random.randint(param_bounds['layers'][0], param_bounds['layers'][1]) # Number of network layers

[0146] neurons = np.random.randint(param_bounds['neurons'][0],param_bounds['neurons'][1]) # Number of neurons in each layer

[0147] F = np.random.uniform(param_bounds['F'][0], param_bounds['F'][1]) # Differential evolution control parameter F

[0148] CR = np.random.uniform(param_bounds['CR'][0], param_bounds['CR'][1]) # Differential evolution control parameter CR

[0149] individual = (l_r, layers, neurons, F, CR)

[0150] population.append(individual)

[0151] return np.array(population)

[0152] # Calculate fitness function: based on model training error

[0153] def fitness_function(model, data, labels, l_r, layers, neurons):

[0154] model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=l_r), loss='mean_squared_error')

[0155] model.fit(data, labels, epochs=10, batch_size=32, verbose=0)

[0156] loss = model.evaluate(data, labels, verbose=0)

[0157] return loss

[0158] # Step S54: Create a PCM memory pool

[0159] def initialize_pcm(population, data, labels):

[0160] fitness_values ​​= []

[0161] for individual in population:

[0162] l_r, layers, neurons, F, CR = individual

[0163] model = create_lstm_model((data.shape[1], data.shape[2]),layers, neurons)

[0164] fitness = fitness_function(model, data, labels, l_r, layers,neurons)

[0165] fitness_values.append(fitness)

[0166] pcm = sorted(zip(population, fitness_values), key=lambda x: x[1])[:10] # Select the top 10 individuals with the best fitness

[0167] return pcm

[0168] # Differential evolution global search

[0169] def global_search(population, F):

[0170] r1, r2, r3 = np.random.choice(len(population), 3, replace=False)

[0171] v_global = population[r1][:-1] + F * (population[r2][:-1] -population[r3][:-1]) # Global mutation

[0172] return v_global

[0173] # Differential evolution local search

[0174] def local_search(population, F, pcm):

[0175] r1, r2, r3 = np.random.choice(len(pcm), 3, replace=False)

[0176] v_local = pcm[r1][:-1] + F * (pcm[r2][:-1] - pcm[r3][:-1]) # Local mutation

[0177] return v_local

[0178] # Step S55: Adaptive search

[0179] def adaptive_search(population, F, pcm):

[0180] new_population = []

[0181] for individual in population:

[0182] v_global = global_search(population, F)

[0183] v_local = local_search(population, F, pcm)

[0184] new_population.append(v_global if np.random.rand() < 0.5 elsev_local)

[0185] return np.array(new_population)

[0186] # Step S56: Population update

[0187] def update_population(population, pcm, new_population):

[0188] updated_population = []

[0189] for i in range(len(population)):

[0190] if new_population[i][-1] < population[i][-1]: # If the new individual has a higher fitness, replace

[0191] updated_population.append(new_population[i])

[0192] else:

[0193] updated_population.append(population[i])

[0194] return updated_population

[0195] # Step S57: Archive Disturbance Mechanism

[0196] def archive_disturbance(pcm, t, max_no_improvement=50):

[0197] if t % max_no_improvement == 0:

[0198] # Trigger the perturbation mechanism

[0199] pcm = np.random.shuffle(pcm)

[0200] return pcm

[0201] # Step S58: Output the optimal hyperparameter combination

[0202] def output_best_parameters(pcm):

[0203] best_individual = min(pcm, key=lambda x: x[1]) # Select the individual with the best fitness

[0204] return best_individual

[0205] # Main execution process

[0206] def main():

[0207] # Assume data and labels have been loaded

[0208] data = np.random.rand(100, 10, 5) # Sample data, 100 samples, 10 time steps, 5 features

[0209] labels = np.random.rand(100, 1) # Example labels

[0210] param_bounds = {

[0211] 'l_r': (0.0001, 0.01), # learning rate range

[0212] 'layers': (1, 5), # Network layer range

[0213] 'neurons': (50, 200), # The number of neurons in each layer

[0214] 'F': (0.5, 1.0), # Differential evolution F parameter range

[0215] 'CR': (0.5, 1.0) # Differential evolution CR parameter range

[0216] }

[0217] population_size = 30

[0218] population = initialize_population(population_size, param_bounds)

[0219] pcm = initialize_pcm(population, data, labels)

[0220] max_iterations = 200

[0221] for t in range(1, max_iterations + 1):

[0222] new_population = adaptive_search(population, 0.8, pcm) # Set F to 0.8

[0223] population = update_population(population, pcm, new_population)

[0224] pcm = archive_disturbance(pcm, t)

[0225] if t == max_iterations:

[0226] best_parameters = output_best_parameters(pcm)

[0227] print(f"Optimal hyperparameter combination: {best_parameters}")

[0228] if __name__ == "__main__":

[0229] main().

[0230] Example 5, see Figure 3 The present invention provides an intelligent agricultural machinery status monitoring system, which includes a data acquisition module, a signal preprocessing module, a feature extraction module, a deep feature learning module, a status monitoring module, and a fault diagnosis module, specifically including the following contents:

[0231] The data acquisition module collects high-frequency vibration signals and low-frequency operation signals of the agricultural machinery;

[0232] The signal preprocessing module preprocesses the high-frequency vibration signal and the low-frequency operation signal;

[0233] The feature extraction module uses GhostNet lightweight convolution to extract high-frequency features from the preprocessed high-frequency vibration signal, and combines ensemble learning and coordinate attention mechanism to enhance features. It uses statistical feature extraction method to extract low-frequency features from the preprocessed low-frequency operation signal, and fuses them to obtain agricultural machinery operation features.

[0234] The deep feature learning module uses a deep learning and self-supervised learning framework, combined with an autoencoder, to perform deep learning on the operation characteristics of agricultural machinery to obtain deep features;

[0235] The state monitoring module builds a state monitoring model based on deep features, analyzes the operating state of agricultural machinery in real time, and obtains state monitoring results;

[0236] The fault diagnosis module constructs a fault diagnosis model, analyzes the status detection results, and obtains analysis results and risk predictions.

[0237] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0238] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

[0239] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. An intelligent agricultural machinery status monitoring method, characterized by: The method comprises the following steps: Step S1: collecting agricultural machinery operation data, installing high-frequency vibration sensors and low-frequency signal sensors at key parts of the agricultural machinery to collect high-frequency vibration signals and low-frequency operation signals of the agricultural machinery respectively. The key parts include bearings, gears and engines; Step S2: Signal preprocessing: using wavelet transform to perform multi-scale decomposition on the high-frequency vibration signal, and using smoothing filtering and denoising on the low-frequency operation signal to obtain the preprocessed high-frequency vibration signal and low-frequency operation signal; Step S3: Feature extraction: GhostNet lightweight convolution is used to extract high-frequency features from the preprocessed high-frequency vibration signal, and the features are enhanced by combining ensemble learning and coordinate attention mechanism. Statistical feature extraction methods are used to extract low-frequency features from the preprocessed low-frequency operation signal, and the low-frequency features are fused to obtain the agricultural machinery operation features. Step S4: Deep feature learning, using deep learning and self-supervised learning framework, combined with autoencoders to conduct deep learning on the operation characteristics of agricultural machinery to obtain deep features; Step S5: Condition monitoring: construct a condition monitoring model based on deep features, analyze the operating status of agricultural machinery in real time, and obtain condition monitoring results; Building a condition monitoring model based on deep features includes the following steps: Step S51: constructing state features, combining the deep features of step S4 to construct a time series feature vector, and performing normalization processing; Step S52: Initialize the condition monitoring model, input the time series feature vector processed in step S51, and use LSTM to build the condition monitoring model; Step S53: Population initialization: randomly initialize an initial population, where each individual represents a set of hyperparameter configurations, and each individual is represented as follows: ； in, For the Individuals, is the learning rate, is the number of network layers, is the number of neurons in each layer, and is the control parameter of differential evolution; Calculate the fitness of each individual in the initial population and obtain the fitness value; Step S54: Establish a PCM memory pool, set the storage capacity to M, store the top M individuals with the highest current fitness values ​​into the PCM memory pool, set the maximum number of iterations and the global iteration variable, as shown below: ； when When , execute step S55 to step S57; in, is a global iteration variable, is the maximum number of iterations, set to 200; Step S55: Adaptive search, performing global search and local search, specifically including the following: Global search: Perform standard differential evolution on each individual in the initial population. The formula used is as follows: ； in, 、 and is an individual randomly selected from the initial population, New individuals generated by global search; Local search: Guide mutation of individuals in the PCM memory pool with a probability of 50%. The formula used is as follows: ； in, is an individual randomly selected from the PCM memory pool, New individuals generated by local search; Step S56: Population update, calculation and The fitness value of Fitness value higher than , then replace Entering the initial population, if If the fitness of is higher than that of any individual in the PCM memory pool, it is replaced to obtain the replaced initial population and the replaced PCM memory pool; Step S57: archive disturbance mechanism, monitoring the changes in the PCM memory pool, and triggering the archive disturbance mechanism when there is no optimization after 50 iterations; Step S58: Output the result. When , the iteration ends and the individual with the highest fitness in the final PCM pool is output as the optimal hyperparameter combination; Step S6: Fault diagnosis, building a fault diagnosis model, analyzing the status detection results, and obtaining analysis results and risk prediction.

2. An intelligent agricultural machinery status monitoring system, used to implement the intelligent agricultural machinery status monitoring method according to claim 1, characterized in that: It includes data acquisition module, signal preprocessing module, feature extraction module, deep feature learning module, status monitoring module and fault diagnosis module.

3. The intelligent agricultural machinery status monitoring system according to claim 2, characterized in that: The data acquisition module collects high-frequency vibration signals and low-frequency operation signals of the agricultural machinery; The signal preprocessing module preprocesses the high-frequency vibration signal and the low-frequency operation signal; The feature extraction module uses GhostNet lightweight convolution to extract high-frequency features from the preprocessed high-frequency vibration signal, and combines ensemble learning and coordinate attention mechanism to enhance features. It uses statistical feature extraction method to extract low-frequency features from the preprocessed low-frequency operation signal, and fuses them to obtain agricultural machinery operation features. The deep feature learning module uses a deep learning and self-supervised learning framework, combined with an autoencoder, to perform deep learning on the operation characteristics of agricultural machinery to obtain deep features; The state monitoring module builds a state monitoring model based on deep features, analyzes the operating state of agricultural machinery in real time, and obtains state monitoring results; The fault diagnosis module constructs a fault diagnosis model, analyzes the status detection results, and obtains analysis results and risk predictions.

Images