Electromagnetic valve fault diagnosis method for industrial control

Abstract

The invention discloses a fault diagnosis method for an electromagnetic valve used for industrial control. The method comprises the steps that valve element displacement simulation data and valve element displacement real data are obtained respectively; the valve element displacement simulation data and the valve element displacement real data are compared and verified, and the verified valve element displacement data serve as a data set of an electromagnetic valve fault diagnosis model; an LSTM-CNN hybrid neural network is constructed as an electromagnetic valve fault diagnosis model; inputting the obtained data set of the electromagnetic valve fault diagnosis model into the electromagnetic valve fault diagnosis model for training to obtain a trained electromagnetic valve fault diagnosis model; and utilizing the trained solenoid valve fault diagnosis model to predict the acquired real data of the solenoid valve. According to the method and the related system, the efficiency and the prediction accuracy of fault diagnosis of the electromagnetic valve are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and more particularly to a solenoid valve fault diagnosis method and a related system used in industrial control. Background Art

[0002] Solenoid valves are key control components of industrial control systems. They are widely used in industrial automation, such as the petrochemical industry. Solenoid valves are widely used in all walks of life because of their accurate action, high degree of automation, stable and reliable operation, and low cost under high-efficiency working conditions. However, various problems are inevitable during use. Due to the limitations of solenoid valves under complex working conditions, the current fault diagnosis technology is still insufficient in monitoring the state of solenoid valves, making it difficult to detect some faults in the early stages, and it is not easy to obtain complete fault feature information in real time, resulting in a lack of fault sample data, and it is difficult to complete the training of the fault diagnosis model for solenoid valves on this basis.

[0003] Therefore, how to more effectively detect and diagnose the solenoid valves in industrial control systems is a technical problem that technical personnel in this field urgently need to solve. Summary of the invention

[0004] In view of this, the present invention provides a solenoid valve fault diagnosis method and a related system used in industrial control, which are used to accurately detect and diagnose the solenoid valve in the industrial control system.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] The present invention first discloses a solenoid valve fault diagnosis method used in industrial control, comprising the following steps:

[0007] Use AMESim simulation software to set solenoid valve parameters and build a solenoid valve simulation model;

[0008] Using the solenoid valve simulation model to simulate different fault conditions of the solenoid valve, and obtaining current simulation curves under different fault conditions;

[0009] Maxwell simulation software is used to simulate the valve core displacement simulation data of the solenoid valve simulation model under different current magnitudes;

[0010] A solenoid valve test bench is built, and under experimental conditions, current is applied to a real solenoid valve having the same solenoid valve parameters as the solenoid valve simulation model, to test the real valve core displacement data of the real solenoid valve under different current magnitudes;

[0011] Compare and verify the valve core displacement simulation data and the valve core displacement real data, and use the verified valve core displacement data as a data set of the solenoid valve fault diagnosis model;

[0012] Construct LSTM-CNN hybrid neural network as solenoid valve fault diagnosis model;

[0013] The obtained data set of the solenoid valve fault diagnosis model is input into the solenoid valve fault diagnosis model for training to obtain a trained solenoid valve fault diagnosis model;

[0014] The trained solenoid valve fault diagnosis model is used to predict the collected real data of the solenoid valve.

[0015] Furthermore, the solenoid valve parameters include coil turns, coil resistance, solenoid valve core weight, spring preload, drive voltage, magnetic permeability, maximum valve core stroke, spring stiffness coefficient and damping parameters.

[0016] Furthermore, the different fault conditions of the solenoid valve include coil short circuit, valve core stuck, valve core wear, and spring breakage or deformation.

[0017] Furthermore, before inputting the data set into the solenoid valve fault diagnosis model for training, the data set is also preprocessed, specifically including standardizing the data in the data set.

[0018] Furthermore, the LSTM-CNN hybrid neural network includes an input layer, an LSTM layer, a CNN layer, a fully connected layer and an output layer which are connected in sequence.

[0019] Based on the above method, the present invention also discloses a solenoid valve fault diagnosis system used in industrial control, comprising:

[0020] Simulation data acquisition module: Use AMESim simulation software to set solenoid valve parameters and build a solenoid valve simulation model;

[0021] Using the solenoid valve simulation model to simulate different fault conditions of the solenoid valve, and obtaining current simulation curves under different fault conditions;

[0022] Maxwell simulation software is used to simulate the valve core displacement simulation data of the solenoid valve simulation model under different current magnitudes;

[0023] Experimental data acquisition module: acquires the real data of valve core displacement of the real solenoid valve under different current magnitudes in the implementation state, and the real solenoid valve and the solenoid valve simulation model have the same solenoid valve parameters;

[0024] A data set acquisition module: compares and verifies the valve core displacement simulation data and the valve core displacement real data, and uses the verified valve core displacement data as a data set of the solenoid valve fault diagnosis model;

[0025] Model training module: inputting the data set of the solenoid valve fault diagnosis model into the solenoid valve fault diagnosis model for training to obtain a trained solenoid valve fault diagnosis model, wherein the solenoid valve fault diagnosis model includes an LSTM-CNN hybrid neural network;

[0026] Fault diagnosis module: Use the trained solenoid valve fault diagnosis model to predict the collected real data of the solenoid valve.

[0027] Preferably, the same solenoid valve parameters in the experimental data acquisition module include coil turns, coil resistance, solenoid valve core weight, spring preload, drive voltage, magnetic permeability, maximum valve core stroke, spring stiffness coefficient and damping parameters.

[0028] Preferably, in the simulation data acquisition module, the different fault conditions of the solenoid valve include coil short circuit, valve core jamming, valve core wear, and spring breakage or deformation.

[0029] Preferably, in the model training module, before inputting the data set into the solenoid valve fault diagnosis model for training, the model training module also includes preprocessing the data set, specifically including standardizing the data in the data set.

[0030] Preferably, in the model training module, the LSTM-CNN hybrid neural network includes an input layer, an LSTM layer, a CNN layer, a fully connected layer and an output layer connected in sequence.

[0031] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a solenoid valve fault diagnosis method and a related system for industrial control, which have the following beneficial effects:

[0032] The present invention can realize rapid and accurate diagnosis of solenoid valve faults, improve the reliability and stability of the solenoid valve, and improve the fault diagnosis efficiency of the solenoid valve. The prediction accuracy is greatly improved compared with the previous methods. At the same time, the data obtained by simulation can be used for model training, which reduces the time spent on obtaining data through a large amount of experiments. At the same time, it can also provide alternative methods for obtaining data under some complex working conditions where fault conditions are difficult to simulate. At the same time, it can be trained and accurately diagnosed according to different solenoid valves. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] 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 only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0034] Figure 1This is a schematic diagram of the overall process of the industrial control system solenoid valve diagnosis method provided by the present invention.

[0035] Figure 2 This is the overall structure diagram of the LSTM-CNN network fault prediction model provided by the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0037] The electromagnet is the core component of the solenoid valve, and its main related parameters are the driving current and the valve core stroke size. The embodiment of the present invention discloses a solenoid valve fault diagnosis method used in industrial control, such as Figure 1 As shown, the method comprises the following steps:

[0038] First, through the structural principle of the solenoid valve, a mathematical model of the electromagnetic part and the mechanical part of the driving current and the displacement of the solenoid valve core can be established. Under different fault conditions of the solenoid valve, there is an actual correlation between the displacement of the valve core and its working state. For example, in the normal working state and the valve core blocked state, the movement stroke of the valve core is different, resulting in different changes in the driving current. Therefore, electromagnetic theory can be used to explain how the change in driving current reflects the actual working state of the solenoid valve under different working conditions. By preliminarily analyzing the changing characteristics of the driving current when the valve core has different strokes, the relationship between the driving current size of the solenoid valve and its working state can be obtained.

[0039] AMESim and Maxwell are used to simulate the solenoid valve respectively. Maxwell is used to simulate the displacement of the solenoid valve core and the corresponding driving current under different driving voltages. The feasibility of different working states of the solenoid valve is explored by combining theoretical analysis and simulation explanations with the changing law of driving current. At the same time, the simulation results can provide theoretical guidance for subsequent experimental methods. AMESim is used to establish a simulation model for the core part of the solenoid valve. The number of coil turns, coil resistance, weight of the solenoid valve core, spring preload, driving voltage, magnetic permeability, maximum valve core stroke, spring stiffness coefficient, and damping parameters are set. The solenoid valve is simulated under different fault conditions to obtain the simulation curves of valve core displacement and driving current. These simulation results are saved in the database for subsequent training of the LSTM-CNN hybrid neural network model. The input of the neural network model is the current simulation curve under different fault conditions after data processing, and the output is the specific fault type.

[0040] Build a solenoid valve test bench. Since solenoid valves may use a variety of driving voltages in actual work, in order to ensure the universality of the collected data, it is necessary to apply driving voltages of different magnitudes to the solenoid valve. By applying a step signal to the solenoid valve for control, while simulating its actual fault state, test the driving current curve corresponding to different valve core displacements under different working states (normal and faulty) of the solenoid valve. Compare and verify the collected experimental data with the results obtained by the previous simulation to observe whether there are obvious differences between the simulation data and the experimental data. If there are obvious differences, it is necessary to confirm whether the simulation is accurately simulated according to the actual working conditions; if the difference is not obvious, this set of experimental data will be used for the training set and test set of the subsequent LSTM-CNN neural network.

[0041] According to the current characteristics of the solenoid valve under the actual fault state (coil short circuit, valve core stuck, valve core wear, spring breakage or deformation) obtained in the previous experiment, the multiple sets of data collected in the experiment are labeled and classified, and the experimental data are preprocessed. First, the sampling frequency of the experimental data is determined to obtain 200 feature points, reflecting the characteristics that only part of the signal information can be obtained in actual applications. The data is standardized, 75% of the data is used for the training set, and the rest of the data is used for the test set. LSTM-CNN hybrid neural network model is constructed, and a neural network model with input layer, LSTM layer, CNN, and output layer is created. By setting the value of the hyperparameter convolution kernel size (4, 8, 12) and batchsize (32, 64, 128), the model is trained, and the trained neural network model is used for the test set to verify the results. When the verification error is less than 1%, it means that the network model is effective and feasible, and the verification accuracy is high. Otherwise, it is re-divided. The structure diagram of the LSTM-CNN hybrid neural network model is as follows Figure 2 shown.

[0042] Since the simulation data is too smooth and deviates from the actual situation, random noise is generated. 200 feature points are obtained by sampling the simulation data at the same frequency. A random noise value is added to each data point as a new data point. Multiple noises are applied to a set of data multiple times to obtain multiple sets of data. At the same time, the experimental data is truncated. Under the same fault state, different numbers of truncated data are replaced with simulation data to obtain multiple sets of data. Part of the obtained multiple sets of data are divided into training sets and test sets at a ratio of 3 to 1, and part of the data is divided into validation sets. The subsequent steps are the same as 4. The final trained model and the model generated in step 4 are used to predict the validation set. By comparing the prediction results of the simulation data with the actual data, it is found that the prediction accuracy of the two is similar. This shows that the features captured by the simulation data are very close to the actual data, and the model trained using the simulation curve is basically the same as the model trained using the actual curve in terms of effect. This means that the model can learn effective representations from the simulation data. Therefore, the simulated drive current data can replace the real data to a certain extent. The LSTM-CNN neural network model trained using simulation data can also be used for fault diagnosis.

[0043] The embodiment of the present invention also discloses a solenoid valve fault diagnosis system used in industrial control, comprising:

[0044] Simulation data acquisition module: Use AMESim simulation software to set solenoid valve parameters and build a solenoid valve simulation model;

[0045] Using the solenoid valve simulation model to simulate different fault conditions of the solenoid valve, and obtaining current simulation curves under different fault conditions;

[0046] Maxwell simulation software is used to simulate the valve core displacement simulation data of the solenoid valve simulation model under different current magnitudes;

[0047] Experimental data acquisition module: acquires the real data of valve core displacement of the real solenoid valve under different current magnitudes in the implementation state, and the real solenoid valve and the solenoid valve simulation model have the same solenoid valve parameters;

[0048] A data set acquisition module: compares and verifies the valve core displacement simulation data and the valve core displacement real data, and uses the verified valve core displacement data as a data set of the solenoid valve fault diagnosis model;

[0049] Model training module: inputting the data set of the solenoid valve fault diagnosis model into the solenoid valve fault diagnosis model for training to obtain a trained solenoid valve fault diagnosis model, wherein the solenoid valve fault diagnosis model includes an LSTM-CNN hybrid neural network;

[0050] Fault diagnosis module: Use the trained solenoid valve fault diagnosis model to predict the collected real data of the solenoid valve.

[0051] The detailed functions and data flows of each module are described in the method embodiment and will not be repeated here.

[0052] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0053] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A solenoid valve fault diagnosis method for industrial control, characterized in that: The following steps are involved: Use AMESim simulation software to set solenoid valve parameters and build a solenoid valve simulation model; Using the solenoid valve simulation model to simulate different fault conditions of the solenoid valve, and obtaining current simulation curves under different fault conditions; Maxwell simulation software is used to simulate the valve core displacement simulation data of the solenoid valve simulation model under different current magnitudes; A solenoid valve test bench is built, and under experimental conditions, current is applied to a real solenoid valve having the same solenoid valve parameters as the solenoid valve simulation model, to test the real valve core displacement data of the real solenoid valve under different current magnitudes; Compare and verify the valve core displacement simulation data and the valve core displacement real data, and use the verified valve core displacement data as a data set of the solenoid valve fault diagnosis model; Construct LSTM-CNN hybrid neural network as solenoid valve fault diagnosis model; The obtained data set of the solenoid valve fault diagnosis model is input into the solenoid valve fault diagnosis model for training to obtain a trained solenoid valve fault diagnosis model; The trained solenoid valve fault diagnosis model is used to predict the collected real data of the solenoid valve.

2. The solenoid valve fault diagnosis method used in industrial control according to claim 1, characterized in that: The solenoid valve parameters include the number of coil turns, coil resistance, solenoid valve core weight, spring preload, drive voltage, magnetic permeability, maximum valve core stroke, spring stiffness coefficient and damping parameters.

3. The solenoid valve fault diagnosis method for industrial control according to claim 1, characterized in that: The different fault conditions of the solenoid valve include coil short circuit, valve core stuck, valve core wear, and spring breakage or deformation.

4. The solenoid valve fault diagnosis method for industrial control according to claim 1, characterized in that: Before the data set is input into the solenoid valve fault diagnosis model for training, the data set is also preprocessed. Including standardizing the data in the dataset.

5. The solenoid valve fault diagnosis method for industrial control according to claim 1, characterized in that: The LSTM-CNN hybrid neural network includes an input layer, an LSTM layer, a CNN layer, a fully connected layer and an output layer which are connected in sequence.

6. A solenoid valve fault diagnosis system for industrial control, characterized in that: include: Simulation data acquisition module: Use AMESim simulation software to set solenoid valve parameters and build a solenoid valve simulation model; Using the solenoid valve simulation model to simulate different fault conditions of the solenoid valve, and obtaining current simulation curves under different fault conditions; Maxwell simulation software is used to simulate the valve core displacement simulation data of the solenoid valve simulation model under different current magnitudes; Experimental data acquisition module: acquires the real data of valve core displacement of the real solenoid valve under different current magnitudes in the implementation state, and the real solenoid valve and the solenoid valve simulation model have the same solenoid valve parameters; A data set acquisition module: compares and verifies the valve core displacement simulation data and the valve core displacement real data, and uses the verified valve core displacement data as a data set of the solenoid valve fault diagnosis model; Model training module: inputting the data set of the solenoid valve fault diagnosis model into the solenoid valve fault diagnosis model for training to obtain a trained solenoid valve fault diagnosis model, wherein the solenoid valve fault diagnosis model includes an LSTM-CNN hybrid neural network; Fault diagnosis module: Use the trained solenoid valve fault diagnosis model to predict the collected real data of the solenoid valve.

7. The solenoid valve fault diagnosis system for industrial control according to claim 6, characterized in that: The same solenoid valve parameters in the experimental data acquisition module include the number of coil turns, coil resistance, solenoid valve core weight, spring preload, drive voltage, magnetic permeability, maximum valve core stroke, spring stiffness coefficient and damping parameters.

8. The solenoid valve fault diagnosis system for industrial control according to claim 6, characterized in that: In the simulation data acquisition module, the different fault conditions of the solenoid valve include coil short circuit, valve core stuck, valve core wear, and spring breakage or deformation.

9. The solenoid valve fault diagnosis system for industrial control according to claim 6, characterized in that: In the model training module, before inputting the data set into the solenoid valve fault diagnosis model for training, the data set is also preprocessed, specifically including standardizing the data in the data set.

10. The solenoid valve fault diagnosis system for industrial control according to claim 6, characterized in that: In the model training module, the LSTM-CNN hybrid neural network includes an input layer, an LSTM layer, a CNN layer, a fully connected layer and an output layer connected in sequence.

Images