A solenoid valve fault diagnosis method and device based on multi-sensor data fusion

Abstract

This invention discloses a solenoid valve fault diagnosis method and device based on multi-sensor data fusion. This method uses current and air pressure signals as characteristic signals for solenoid valve fault diagnosis, further subdividing the solenoid valve fault diagnosis. The characteristic signals are processed using a wavelet packet accumulation (EMD) method to avoid EMD modal aliasing. A BP neural network is used as the fault diagnosis model to effectively diagnose the solenoid valve. To facilitate the implementation of this fault diagnosis method, a corresponding device is also designed. This device can extract the air pressure and current signals of the solenoid valve under test during operation for fault diagnosis. It can also output an adjustable, fluctuating pressure to simulate an environment with sudden changes in air pressure.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a solenoid valve fault diagnosis method and device based on multi-sensor data fusion. Background Art

[0002] With technological advancements, solenoid valves are increasingly being used in automated industrial control equipment. From small industrial robots to large aircraft, solenoid valves are essential components. However, precisely because of their widespread use, a failure to promptly and effectively diagnose solenoid valve faults can lead to a series of safety incidents. Currently, solenoid valve fault diagnosis mostly relies on current signals. This method can only detect valve core movement based on current changes, and cannot further subdivide solenoid valve faults.

[0003] Empirical Mode Decomposition (EMD) is a commonly used method for fault signal processing, suitable for processing various non-stationary signals and exhibiting adaptive properties. This method effectively decomposes the original signal into several intrinsic mode function (IMF) components and a residual signal. Each IMF component represents the local characteristics of the signal. However, when the original signal contains gaps, noise, or two signals with similar frequencies, EMD cannot effectively distinguish them. As a result, an IMF has multiple different frequency characteristics, resulting in modal aliasing and the loss of local characteristics. Summary of the Invention

[0004] The purpose of the present invention is to provide a solenoid valve fault diagnosis method and device based on multi-sensor data fusion. The present invention uses current and air pressure signals as characteristic signals for solenoid valve fault diagnosis, further subdividing the solenoid valve fault diagnosis. The characteristic signals are processed using a wavelet packet accumulation EMD method to avoid EMD modal aliasing. A BP neural network is used as a fault diagnosis model to effectively diagnose the solenoid valve. To facilitate the implementation of this fault diagnosis method, a corresponding device is also designed. This device can extract the air pressure and current signals of the solenoid valve under test during operation to perform fault diagnosis, and can also output an adjustable fluctuating pressure to simulate an environment with sudden changes in air pressure.

[0005] The technical solution of the present invention is a solenoid valve fault diagnosis method based on multi-sensor data fusion, comprising the following steps:

[0006] S1: Using a current acquisition device and an air pressure acquisition device to collect the original current signal and the original air pressure signal of the solenoid valve sample with known state;

[0007] S2: Perform wavelet packet decomposition on the original current signal and the original pressure signal;

[0008] S3: Performing empirical mode decomposition on the current signal and the air pressure signal after wavelet packet decomposition in sequence to obtain several eigenmode function components and a residual;

[0009] S4: Arrange the intrinsic mode function components and the residuals of the current signal and the air pressure signal to form a current signal matrix and an air pressure signal matrix, calculate the energy moments thereof, obtain a current energy matrix and an air pressure energy matrix, and then superimpose the current energy matrix and the air pressure energy matrix to form a state matrix;

[0010] S5: Square and accumulate each row vector parameter of the state matrix to form a state parameter, normalize the state parameter, and arrange them to obtain a state vector of the solenoid valve sample;

[0011] S6: Repeat steps S1 to S5 to collect current signals and air pressure signals of solenoid valves in various states, calculate their different state vectors to form data samples, then use the data samples to train the BP neural network to form a fault diagnosis model, and finally use the fault diagnosis model to diagnose the solenoid valve fault.

[0012] In the above-mentioned solenoid valve fault diagnosis method based on multi-sensor data fusion, in step S1, the solenoid valve status is divided into normal status and fault status, wherein the fault status includes inability to open the valve, abnormal air pressure, abnormal valve core movement, abnormal air pressure but normal valve core, normal air pressure but abnormal valve core movement, valve core blockage and inability to fully open, valve core blockage but can be fully opened, and spring breakage.

[0013] In the aforementioned solenoid valve fault diagnosis method based on multi-sensor data fusion, in step S2, the wavelet packet decomposition adopts binary discrete wavelet, and the wavelet basis function in the binary discrete wavelet adopts Dobesi wavelet.

[0014] In the aforementioned solenoid valve fault diagnosis method based on multi-sensor data fusion, in step S4, the row vectors of the current signal matrix and the air pressure signal matrix are composed of the intrinsic mode function components and residuals of the same time-frequency signal in each frequency band;

[0015] Assume that the maximum number of intrinsic mode function components decomposed from the signal of a certain frequency band is m, then the number of matrix row vector parameters is m+1. If the number of eigenmode function components of other frequency bands is insufficient, 0 is added to express it;

[0016] The column vectors of the current signal matrix and the air pressure signal matrix are the time-frequency signals of each different frequency band;

[0017] The state matrix is the superposition of the current signal matrix and the air pressure signal matrix.

[0018] The aforementioned device for the solenoid valve fault diagnosis method based on multi-sensor data fusion includes a PC all-in-one computer, which is electrically connected to an intelligent language broadcasting device, an NI industrial control box, a constant pressure drive power supply and an air pump; the PC all-in-one computer is connected to an air path system module via the NI industrial control box, and the air path system module is connected to the air pump and the constant pressure drive power supply; the PC all-in-one computer is provided with a current acquisition positive connector and a current acquisition negative connector.

[0019] The device of the aforementioned solenoid valve fault diagnosis method based on multi-sensor data fusion, the gas path system module includes a system base plate, and the system base plate is provided with an electronic pressure regulating valve, a first pressure sensor, a second pressure sensor, a first solenoid switch valve, a second solenoid switch valve, a first one-way valve, a second one-way valve, a first adjustable flow valve, a second adjustable flow valve, a first exhaust valve, a second exhaust valve, a valve block to be tested, a solenoid valve to be tested, a first three-way joint, a second three-way joint, a third three-way joint and a high-pressure gas tank; one end of the electronic pressure regulating valve is connected to the first pressure sensor via an air pipe, and the other end is connected to the second three-way joint via an air pipe; the other end of the first pressure sensor is connected to the first solenoid switch valve, the first one-way valve and the third three-way joint in sequence by an air pipe; one end of the third three-way joint is connected to the first adjustable flow valve and the second three-way joint in sequence an exhaust valve; the installation direction of the first one-way valve is to conduct from the first electromagnetic switch valve to the third three-way joint; one end of the high-pressure gas tank is connected to the second electromagnetic switch valve via the air pipe, and the other end is connected to the second three-way joint; the other end of the second electromagnetic switch valve is connected in sequence to the second adjustable flow valve, the second one-way valve, the first three-way joint, the second pressure sensor, the valve block under test and the second exhaust valve; the installation direction of the second one-way valve is to conduct from the second adjustable flow valve to the first three-way joint; the third three-way joint is connected to the first three-way joint via the air pipe; the second three-way joint is connected to the air pump via the air pipe; the electromagnetic valve under test is installed on the valve block under test through a threaded connection; the first pressure sensor is used to feedback the pressure adjusted by the electronic pressure regulating valve; the second pressure sensor is used to collect the original air pressure signal as an air pressure acquisition device.

[0020] In the aforementioned device for the solenoid valve fault diagnosis method based on multi-sensor data fusion, the intelligent language broadcasting device broadcasts the solenoid valve status according to instructions.

[0021] The aforementioned device for the solenoid valve fault diagnosis method based on multi-sensor data fusion, the NI industrial control box is provided with an A / D conversion module, a D / A conversion module and a current acquisition card; the A / D conversion module is used to convert the analog signals collected by the current acquisition card and the first pressure sensor and the second pressure sensor into digital quantities; the D / A conversion module is used to convert the digital signals emitted by the PC all-in-one into analog signals for controlling the gas circuit system module; the current acquisition card is used to collect the current signal of the solenoid valve under test and convert it into an analog output.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] (1) The present invention designs a solenoid valve fault diagnosis method. This method uses a signal processing method that combines wavelet packets and EMD. By using wavelet packets to preprocess the signal and decompose it into multiple narrowband signals, the phenomenon of modal aliasing during EMD decomposition is avoided, the discrimination between different solenoid valve states is effectively improved, and it is beneficial to the training of the fault diagnosis model.

[0024] (2) The present invention integrates a multi-sensor detection method and combines it with air pressure-assisted detection to provide a more comprehensive diagnosis of solenoid valve failures. By detecting whether the working air pressure of the solenoid valve is normal, it can effectively distinguish whether the cause of the valve core movement is due to pressure fluctuations, and whether the valve core blockage causes the valve core to be unable to open and close completely, or only causes the valve core to open and close slowly.

[0025] (3) The present invention squares and accumulates the row vector parameters of the energy matrix, thereby reducing the number of state parameters, reducing the number of input nodes of the BP neural network, accelerating the training speed and calculation speed of the BP neural network, and reducing redundant calculations.

[0026] (4) The present invention also designs a solenoid valve fault diagnosis device that can effectively extract the current signal and air pressure signal of the solenoid valve to perform fault diagnosis. The device can also output adjustable fluctuating air pressure to simulate working conditions with pressure fluctuations. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a fault diagnosis flow chart of the present invention;

[0028] Figure 2 It is the wavelet packet decomposition diagram of the present invention;

[0029] Figure 3 It is the EMD decomposition flow chart of the present invention;

[0030] Figure 4 This is a flow chart of BP neural network training of the present invention;

[0031] Figure 5 is a schematic diagram of a fault diagnosis device of the present invention;

[0032] Figure 6 Schematic diagram of the gas circuit system module of the present invention. DETAILED DESCRIPTION

[0033] The present invention will be further described below with reference to the examples, but they are not intended to limit the present invention.

[0034] Example 1: A solenoid valve fault diagnosis method based on multi-sensor data fusion, such as Figure 1 As shown, the following steps are included:

[0035] S1: Use a current acquisition device and an air pressure acquisition device to collect the original current signal I(t) and the original air pressure signal P(t) of a solenoid valve sample in a known state. In this step, the solenoid valve state is divided into normal state and fault state. The fault states are further divided into 1. Unable to open the valve; 2. Abnormal air pressure, abnormal valve core movement; 3. Abnormal air pressure, normal valve core; 4. Normal air pressure, abnormal valve core movement; 5. Valve core is blocked and cannot be fully opened; 6. Valve core is blocked and can be fully opened; 7. Spring breakage.

[0036] S2: If Figure 2 As shown, the original current signal and the original pressure signal are subjected to wavelet packet decomposition processing; the wavelet packet decomposition can effectively process non-stationary signals, convert them into signals with time-frequency characteristics, and decompose them according to frequency to form multiple time-frequency signals with different center frequencies but the same bandwidth.

[0037] Furthermore, the wavelet packet decomposition adopts a binary discrete wavelet. The binary discrete wavelet adopts a Dobesi wavelet as a wavelet mother function, and the specific formula is as follows:

[0038]

[0039] in represents the wavelet mother function, a is the scale factor, and b is the translation factor;

[0040] The wavelet packet transform formula is as follows:

[0041]

[0042] Furthermore, the signal frequency is decomposed into four segments, AA2, DA2, AD2 and DD2, from low to high, through wavelet packet transform.

[0043] S3: If Figure 3As shown in the figure, the current signal and the pressure signal after wavelet packet decomposition are sequentially subjected to empirical mode decomposition (EMD) to obtain several intrinsic mode function components and a residual. EMD modal decomposition can effectively process non-stationary time-frequency signals and can decompose the signal into several eigenmode components. Each eigenmode component contains a single characteristic information.

[0044] Furthermore, the EMD decomposition steps are as follows

[0045] A1: Determine the upper and lower extreme points of the signal, calculate the upper envelope u1(t) and lower envelope l1(t) of the signal, and then add the upper and lower envelopes and divide by two to calculate the mean envelope. The specific formula is as follows:

[0046] m1(t)=(u1(t)+l1(t)) / 2;

[0047] A2: Subtract the mean envelope from the original signal to obtain the new signal h1(t). The specific formula is as follows:

[0048] h1(t)=f1(t)-m1(t);

[0049] Where: f1(t) represents the original signal;

[0050] A3: Determine whether h1(t) meets the IMF criteria. If not, repeat the loop until it meets the criteria. The IMF criteria are: 1. The number of extreme points and zero crossings in the entire data length must be equal or differ by at most one; 2. In the signal time domain, the average value of the upper and lower envelopes determined by the maximum and minimum points of the cubic spline fit is 0.

[0051] A4: Let the newly derived IMF component be C1(t). Subtract the IMF component from the original signal to obtain a residue r1(t). The specific formula is as follows:

[0052] r1(t)=f1(t)-C1(t);

[0053] A5: Continue the previous operation on the residual until the residual is a monotonic function or the residual has only one extreme point, and then stop the decomposition.

[0054] S4: Arrange the intrinsic mode function components and the residuals of the current signal and the air pressure signal to form a current signal matrix and an air pressure signal matrix, calculate the energy moments thereof, obtain a current energy matrix and an air pressure energy matrix, and then superimpose the current energy matrix and the air pressure energy matrix to form a state matrix;

[0055] Furthermore, the row vectors of the signal matrix are composed of the IMF components and residuals of the same time-frequency signal in each frequency band. Assuming that the IMF components decomposed from the signal of a certain frequency band are the largest, which is m, then the number of matrix row vector parameters is m+1. If the number of IMF components in other frequency bands is insufficient, 0 is added to represent it. The column vectors are the time-frequency signals of each different frequency band. The details are as follows:

[0056]

[0057] Furthermore, the energy moment is solved for each IMF component and residual in the matrix. The specific formula of the energy moment is as follows:

[0058] E im,j =∫|S im,j (t)| 2 dt;

[0059]

[0060] Furthermore, the current signal energy matrix and the air pressure signal energy matrix are superimposed to obtain the state matrix of the solenoid valve, which is as follows:

[0061]

[0062] S5: Square and add up each row vector parameter of the state matrix to form the state parameter e i , the state parameter e i Normalize and arrange to get the state vector T of the solenoid valve sample m ;

[0063] Furthermore, the state parameter e i The calculation formula is as follows:

[0064] e i =E i,1 2 +E i,2 2 +···+E i,m 2 +E ri 2

[0065] Furthermore, the state vector T m The calculation formula is as follows:

[0066]

[0067] T m =[d1,d2,···d8];

[0068] S6: As Figure 4As shown, steps S1-S5 are repeated to collect current and pressure signals of solenoid valves in various states, calculate their different state vectors to form data samples, and then use the data samples to train a BP neural network to form a fault diagnosis model. Finally, the fault diagnosis model is used to diagnose solenoid valve faults. In this embodiment, 80% of the data samples are imported into the BP neural network for training to form a fault diagnosis model. 20% of the data samples are used to determine the accuracy of the fault diagnosis model. The BP neural network has 8 input nodes for inputting the solenoid valve state vector parameters and 7 output nodes for representing the normal state and six fault states of the solenoid valve. Three hidden layers are set, and the number of hidden layer neurons is selected according to an empirical formula: 23 neuron nodes are set in the first layer, 20 neuron nodes are set in the second layer, and 18 neuron nodes are set in the third layer. The expected error is 0.0005, and the learning rate is 0.001.

[0069] Example 2: A solenoid valve fault diagnosis device based on multi-sensor data fusion, such as Figure 5 As shown, it includes a PC all-in-one 1, which is electrically connected to an intelligent language broadcasting device 2, an NI industrial control box 5, a constant voltage drive power supply 10 and an air pump 11; the PC all-in-one 1 is connected to an air circuit system module 9 via the NI industrial control box 5, and the air circuit system module 9 is connected to the air pump 11 and the constant voltage drive power supply 10; the PC all-in-one 1 is provided with a current acquisition positive connector 3 and a current acquisition negative connector 4.

[0070] Further, if Figure 6As shown, the gas circuit system module 9 includes a system base plate 902, on which are provided an electronic pressure regulating valve 901, a first pressure sensor 903, a second pressure sensor 911, a first electromagnetic switch valve 904, a second electromagnetic switch valve 915, a first one-way valve 905, a second one-way valve 913, a first adjustable throttle valve 906, a second adjustable throttle valve 914, a first exhaust valve 907, a second exhaust valve 908, a valve block to be tested 910, a electromagnetic valve to be tested 909, and a first three-way connector 912. , a second three-way connector 918, a third three-way connector 924 and a high-pressure gas tank 916; one end of the electronic pressure regulating valve 901 is connected to the first pressure sensor 903 via an air pipe, and the other end is connected to the second three-way connector 918 via an air pipe; the other end of the first pressure sensor 903 is connected to the first solenoid switch valve 904, the first one-way valve 905 and the third three-way connector 924 in sequence via an air pipe; one end of the third three-way connector 924 is connected to the first adjustable throttle valve 906 and the first exhaust valve 907 in sequence; The first one-way valve 905 is installed in a direction that is connected from the first electromagnetic switch valve 904 to the third three-way connector 924; one end of the high-pressure gas tank 916 is connected to the second electromagnetic switch valve 915 through the gas pipe, and the other end is connected to the second three-way connector 918; the other end of the second electromagnetic switch valve 915 is connected in sequence to the second adjustable flow valve 914, the second one-way valve 913, the first three-way connector 912, the second pressure sensor 911, the valve block 910 under test and the second exhaust valve 908; the second one-way valve 9 The installation direction is from the second adjustable throttle valve 914 to the first three-way connector 912. The third three-way connector 924 is connected to the first three-way connector 912 via an air pipe. The second three-way connector 918 is connected to the air pump 11 via an air pipe. The solenoid valve under test 909 is threadedly mounted on the valve block under test 910. The first pressure sensor 903 is used to provide feedback on the pressure regulated by the electronic pressure regulating valve 901. The second pressure sensor 911 serves as a pressure acquisition device to collect the raw pressure signal. This air circuit system module can output fluctuating air pressure to simulate an abnormal pressure environment, and can be used in conjunction with fault diagnosis methods to extract signals when the solenoid valve under test is experiencing abnormal pressure. The specific operation is as follows: Open the electronic pressure regulating valve 901, set the rated pressure, and simultaneously begin storing high-pressure gas in the high-pressure gas tank. When testing at the normal rated pressure, open the first solenoid on / off valve 904, and the rated pressure gas is transferred to the valve block under test 910 via the first check valve 905. When an upward pressure fluctuation is required, the second solenoid on-off valve 915 is opened, and the high-pressure gas is transmitted to the valve block 910 under test via the second one-way valve 913 and the second adjustable throttle valve 914. When a downward pressure fluctuation is required, the first exhaust valve 907 is opened, and the rated pressure gas is discharged via the first adjustable throttle valve 906 and the first exhaust valve 907. The first adjustable throttle valve 906 and the second adjustable throttle valve 914 are used to adjust the rate of pressure fluctuation.The first one-way valve 905 is used to prevent the high-pressure gas from flowing back to the electronic pressure regulating valve 901 .

[0071] Furthermore, the intelligent language broadcasting device 2 broadcasts the state of the solenoid valve according to instructions.

[0072] Furthermore, the NI industrial control box 3 is provided with an A / D conversion module, a D / A conversion module and a current acquisition card; the A / D conversion module is used to convert the analog signals collected by the current acquisition card and the first pressure sensor 903 and the second pressure sensor 911 into digital quantities; the D / A conversion module is used to convert the digital signal emitted by the PC all-in-one 1 into an analog signal for controlling the gas circuit system module 9; the current acquisition card is used to collect the current signal of the solenoid valve 909 under test and convert it into an analog output.

[0073] The device of this embodiment works in two parts: the first part is to collect sample data for building a fault diagnosis model; the second part is to collect data of the valve under test and perform fault diagnosis using the fault diagnosis model.

[0074] The working process of the first part is as follows: install the sample solenoid valve on the valve block 910 of the valve to be tested. Turn on the air pump 4. Turn on the constant pressure drive power supply 10 to power the air path system module 9. Send a command to the NI industrial control box 3 through the PC all-in-one 1. The digital signal of the PC all-in-one 1 is converted into analog quantity through the D / A conversion module in the NI industrial control box 3 to control the air path system module 9 to work. The sample solenoid valve is connected to the current acquisition positive connector 3 and the current acquisition negative connector 4 via wires. The current acquisition positive connector 3 and the current acquisition negative connector 4 are connected to the current acquisition card in the NI industrial control box 3 for collecting current. The second pressure sensor 911 and the current acquisition card transmit the air pressure analog signal and the current analog signal to the A / D conversion module in the NI industrial control box 3. The A / D conversion module converts the air pressure analog signal and the current analog signal into digital quantity and transmits them to the PC all-in-one 1 to form the original air pressure signal and the original current signal. Data is collected for different sample solenoid valves under different pressure conditions multiple times to form data samples. The PC all-in-one machine 1 is installed with the algorithm of the electromagnetic valve fault diagnosis method based on multi-sensor data fusion, and data samples are introduced into the algorithm to form a corresponding fault diagnosis model.

[0075] The working process of the second part is as follows: install the electromagnetic valve 909 under test on the valve block 910 of the valve under test, and turn on the air pump 4. Turn on the constant pressure drive power supply 10 to power the air system module 9. Send a command to the NI industrial control box 3 through the PC all-in-one 1. The digital signal of the PC all-in-one 1 is converted into analog quantity through the D / A conversion module in the NI industrial control box 3 to control the air system module 9 to work. At this time, the air system module 9 only works in the rated pressure mode. The electromagnetic valve under test is connected to the current acquisition positive connector 3 and the current acquisition negative connector 4 via a wire for collecting current. The second pressure sensor 911 and the current acquisition card transmit the air pressure analog signal and the current analog signal to the A / D conversion module in the NI industrial control box 3. The A / D conversion module converts the air pressure analog signal and the current analog signal into digital quantity, and transmits them to the PC all-in-one 1 to form the original air pressure signal and the original current signal. The original air pressure signal and the original current signal are imported into the fault diagnosis model. The model diagnoses the state of the solenoid valve under test based on the signals and transmits the diagnosis results to the intelligent language broadcasting device 2. The intelligent language broadcasting device 2 broadcasts the corresponding status based on the diagnosis results.

Claims

1. A solenoid valve fault diagnosis method based on multi-sensor data fusion, characterized by: The following steps are involved: S1: Using a current acquisition device and an air pressure acquisition device to collect the original current signal and the original air pressure signal of the solenoid valve sample with known state; S2: Perform wavelet packet decomposition on the original current signal and the original pressure signal; S3: Performing empirical mode decomposition on the current signal and the air pressure signal after wavelet packet decomposition in sequence to obtain several eigenmode function components and a residual; S4: Arrange the intrinsic mode function components and the residuals of the current signal and the air pressure signal to form a current signal matrix and an air pressure signal matrix, calculate the energy moments thereof, obtain a current energy matrix and an air pressure energy matrix, and then superimpose the current energy matrix and the air pressure energy matrix to form a state matrix; S5: Square and accumulate each row vector parameter of the state matrix to form a state parameter, normalize the state parameter, and arrange them to obtain a state vector of the solenoid valve sample; S6: Repeat steps S1 to S5 to collect current signals and air pressure signals of solenoid valves in various states, calculate their different state vectors to form data samples, then use the data samples to train the BP neural network to form a fault diagnosis model, and finally use the fault diagnosis model to diagnose the solenoid valve fault.

2. The solenoid valve fault diagnosis method based on multi-sensor data fusion according to claim 1 is characterized in that: In step S1, the solenoid valve status is divided into normal status and fault status, wherein the fault status includes inability to open the valve, abnormal air pressure, abnormal valve core movement, abnormal air pressure but normal valve core, normal air pressure but abnormal valve core movement, valve core blockage and inability to fully open, valve core blockage but fully open, and spring breakage.

3. The solenoid valve fault diagnosis method based on multi-sensor data fusion according to claim 1 is characterized in that: In step S2, the wavelet packet decomposition adopts a binary discrete wavelet, and the wavelet basis function in the binary discrete wavelet adopts a Dobesi wavelet.

4. The solenoid valve fault diagnosis method based on multi-sensor data fusion according to claim 1, characterized in that: In step S4, the row vectors of the current signal matrix and the air pressure signal matrix are composed of the eigenmode function components and the residuals of the same time-frequency signal in each frequency band; Assume that the maximum number of intrinsic mode function components decomposed from the signal of a certain frequency band is m, then the number of matrix row vector parameters is m+1. If the number of eigenmode function components of other frequency bands is insufficient, 0 is added to express it; The column vectors of the current signal matrix and the air pressure signal matrix are the time-frequency signals of each different frequency band; The state matrix is the superposition of the current signal matrix and the air pressure signal matrix.

5. A solenoid valve fault diagnosis device based on multi-sensor fusion, characterized by: The fault diagnosis method according to any one of claims 1 to 4 is performed, comprising a PC all-in-one machine (1), wherein the PC all-in-one machine (1) is electrically connected to an intelligent language broadcasting device (2), an NI industrial control box (5), a constant voltage drive power supply (10) and an air pump (11); the PC all-in-one machine (1) is connected to an air path system module (9) via the NI industrial control box (5), and the air path system module (9) is connected to the air pump (11) and the constant voltage drive power supply (10); and the PC all-in-one machine (1) is provided with a current acquisition positive connector (3) and a current acquisition negative connector (4).

6. The solenoid valve fault diagnosis device based on multi-sensor fusion according to claim 5 is characterized in that: The gas path system module (9) comprises a system base plate (902), on which are provided an electronic pressure regulating valve (901), a first pressure sensor (903), a second pressure sensor (911), a first electromagnetic switch valve (904), a second electromagnetic switch valve (915), a first one-way valve (905), a second one-way valve (913), a first adjustable flow valve (906), a second adjustable flow valve (914), a first exhaust valve (907), a second exhaust valve (908), a valve block to be tested (910), a electromagnetic valve to be tested (909), a first three-way connector ( 912), a second three-way joint (918), a third three-way joint (924) and a high-pressure gas tank (916); one end of the electronic pressure regulating valve (901) is connected to the first pressure sensor (903) via a gas pipe, and the other end is connected to the second three-way joint (918) via a gas pipe; the other end of the first pressure sensor (903) is connected to the first electromagnetic switch valve (904), the first one-way valve (905) and the third three-way joint (924) in sequence through the gas pipe; one end of the third three-way joint (924) is connected to the first adjustable throttle valve (906) and the first exhaust valve (9 07); the first one-way valve (905) is installed in a direction from the first electromagnetic switch valve (904) to the third three-way connector (924); one end of the high-pressure gas tank (916) is connected to the second electromagnetic switch valve (915) through the gas pipe, and the other end is connected to the second three-way connector (918); the other end of the second electromagnetic switch valve (915) is connected in sequence to the second adjustable flow valve (914), the second one-way valve (913), the first three-way connector (912), the second pressure sensor (911), the valve block (910) to be tested, and the second exhaust valve (908); the second one-way valve (916) is connected to the second electromagnetic switch valve (915) through the gas pipe, and the other end is connected to the second three-way connector (918); the other end of the second electromagnetic switch valve (915) is connected in sequence to the second adjustable flow valve (914), the second one-way valve (913), the first three-way connector (912), the second pressure sensor (911), the valve block (910) to be tested, and the second exhaust valve (908); The direction of installation of the valve (913) is to conduct from the second adjustable flow valve (914) to the first three-way connector (912); the third three-way connector (924) is connected to the first three-way connector (912) via an air pipe; the second three-way connector (918) is connected to the air pump (11) via an air pipe; the solenoid valve (909) to be tested is installed on the valve block (910) to be tested through a threaded connection; the first pressure sensor (903) is used to feedback the pressure adjusted by the electronic pressure regulating valve (901); the second pressure sensor (911) is used as an air pressure acquisition device to collect the original air pressure signal.

7. The solenoid valve fault diagnosis device based on multi-sensor fusion according to claim 5, characterized in that: The intelligent language broadcasting device (2) broadcasts the state of the solenoid valve according to instructions.

8. The solenoid valve fault diagnosis device based on multi-sensor fusion according to claim 6, characterized in that: The NI industrial control box (5) is provided with an A / D conversion module, a D / A conversion module and a current acquisition card; the A / D conversion module is used to convert analog signals collected by the current acquisition card and the first pressure sensor (903) and the second pressure sensor (911) into digital signals; the D / A conversion module is used to convert digital signals emitted by the PC all-in-one (1) into analog signals for controlling the gas circuit system module (9); the current acquisition card is used to collect the current signal of the electromagnetic valve (909) under test and convert it into analog output.

Images