Intelligent multi-way valve comprehensive training test system and method

Abstract

The invention provides an intelligent multi-way valve comprehensive training test system which comprises a fault simulation module which comprises a pipeline leakage simulation module, a valve element clamping stagnation simulation module, an open circuit simulation module and a pipeline damage simulation module. The system further comprises a comprehensive load simulation module. According to the method, the intelligent multi-way valve to be tested is installed in the intelligent multi-way valve comprehensive training and testing system, and fault simulation testing and load simulation testing are conducted on the intelligent multi-way valve to be tested. The comprehensive training and testing system and method for the intelligent multi-way valve can be used for fault mode recognition function training and controller training optimization of an intelligent control system in the intelligent multi-way valve at the same time, and diversified testing and training of the intelligent multi-way valve are achieved in a laboratory environment; and time and economic cost caused by switching among different test benches can be greatly saved.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-way valves, relates to intelligent hydraulic valve components, and in particular to an intelligent multi-way valve comprehensive training and testing system and method. Background Art

[0002] Multi-way valves are widely used in industrial and agricultural fields such as assembly lines, mining engineering machinery, and agricultural engineering machinery. With the improvement of intelligence and informatization levels, as well as the increase in demand for refined operations, traditional multi-way valves have been unable to meet the needs of various types of engineering machinery and assembly lines due to their low automation / intelligence and less work status information collection. The demand for intelligent multi-way valves is increasing day by day. However, at this stage, there is relatively little research on intelligent multi-way valves. The intelligent control of intelligent multi-way valves is different from that of traditional multi-way valves, and has higher requirements in fault mode recognition and automatic identification of working conditions. With the rise of artificial intelligence, integrating artificial intelligence models in intelligent multi-way valves has become a means to improve the accuracy of fault mode recognition and valve control performance under different working conditions, which requires a large amount of training data to be provided for intelligent multi-way valves. At the same time, for specific application scenarios, due to different load and motion requirements, the use of the same set of intelligent multi-way valve controllers will result in large system errors under certain working conditions, which requires intelligent proportional valves to optimize control parameters for application scenarios. On the other hand, the existing multi-way valve test bench has a single function. The prior art provides a load simulation system and control method for a multi-way valve, which realizes the load simulation of the multi-way valve port by connecting solenoid valves in series. However, this method can only realize switch simulation, and cannot realize proportional valve state simulation under different load pressures and flow rates; other multi-way valve test platforms are also unable to realize intelligent multi-way valve fault simulation, control optimization and comprehensive testing combining the two. Summary of the invention

[0003] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a comprehensive training and testing system and method for intelligent multi-way valves, so as to solve the technical problem in the prior art that it is difficult to perform diversified testing and training on intelligent multi-way valves in the same system.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions to achieve the above problems:

[0005] An intelligent multi-way valve comprehensive training and testing system comprises a fault simulation module, wherein the fault simulation module comprises a pipeline leakage simulation module, a valve core sticking simulation module, a circuit breaking simulation module and a pipeline damage simulation module.

[0006] Also includes a comprehensive load simulation module.

[0007] The present invention also has the following technical features:

[0008] The pipeline leakage simulation module comprises a pipeline connected to the oil port of the intelligent multi-way valve to be tested, the end face of the pipeline is fixed by a frame, and the end face of the pipeline is sealed and installed with a leakage gland through a sealing member.

[0009] The leakage gland provides pressure through a hydraulic cylinder, and the hydraulic cylinder is connected to the servo valve and the accumulator respectively through a hydraulic oil pipe with a pressure sensor.

[0010] An annular leakage oil collecting groove is provided on the inner side wall of the leakage gland along the circumferential direction. The leakage oil collecting groove is connected with an oil discharge hole penetrating the side wall of the leakage gland, and a measuring cup is arranged at the outer end of the oil discharge hole.

[0011] The valve core stuck simulation module includes a valve body, in which a valve core is installed, one end of the valve core is connected to one end of an extension rod, and the other end of the extension rod extends into the electromagnetic brake. The movement of the wedge block in the electromagnetic brake can achieve the jamming or blocking of the extension rod.

[0012] The circuit breaking simulation module is connected to the electrical circuits of the intelligent multi-way valve to be tested respectively through multi-way relays, and the electrical circuit breaking fault of the intelligent multi-way valve to be tested is simulated by turning on and off the relays; the electrical circuit breaking fault includes power supply circuit breaking fault, communication cable circuit breaking fault and sensor signal line circuit breaking fault.

[0013] The pipeline damage simulation module includes a pipeline, a damage opening is opened on the side wall of the pipeline, a sealing tube core that can move in the pipeline is installed in the pipeline, and there are three working states between the sealing tube core and the damage opening, namely, a sealed state, a semi-damaged state and an open state.

[0014] The comprehensive load simulation module includes a two-position two-way proportional valve connected to the intelligent multi-way valve to be tested, the oil inlet end of the two-position two-way proportional valve is connected to the first flow valve, and the oil outlet end of the two-position two-way proportional valve is connected to the second flow valve.

[0015] The present invention also protects a comprehensive training and testing method for an intelligent multi-way valve, which installs the intelligent multi-way valve to be tested in the intelligent multi-way valve comprehensive training and testing system as described above, and performs a fault simulation test and a load simulation test on the intelligent multi-way valve to be tested.

[0016] Step 1: Fault simulation test:

[0017] During the fault simulation test phase, different fault simulation modules are controlled to generate corresponding fault actions. At the same time, the fault mode recognition function is trained through the convolutional neural network embedded in the controller of the intelligent multi-way valve to be tested.

[0018] Step 2: Load simulation test:

[0019] During the load simulation stage, the working port pressure of the intelligent multi-way valve to be tested is controlled by the two-position two-way proportional control in the comprehensive load simulation module, the outlet flow of the overflow valve is measured by the flowmeter, and the flow through the intelligent multi-way valve to be tested is controlled by the intelligent multi-way valve to be tested, thereby realizing flow closed-loop control training and testing under different working pressures.

[0020] At the same time, the controller optimization training is carried out through the neural network embedded in the controller of the intelligent multi-way valve to be tested. According to different specific application scenarios and working conditions, the control parameters are adjusted in real time through the neural network, thereby realizing the optimization of the controller of the intelligent multi-way valve to be tested.

[0021] In step 1, the training method of the fault mode recognition function comprises the following steps:

[0022] Step 101, setting the load state:

[0023] Set the key load parameter information according to the working conditions of the intelligent multi-way valve to be tested.

[0024] The key load parameter information includes maximum / minimum load force, hydraulic cylinder bore / rod diameter, motor displacement, and load motion target curve.

[0025] Step 102, calculate the hydraulic theoretical flow:

[0026] According to the load state set in step 101, the theoretical hydraulic flow required for load movement is calculated.

[0027] Step 103, training the fault mode recognition function:

[0028] Taking the theoretical flow in step 102 and the key information collected by the controller of the intelligent multi-way valve to be tested according to the built-in sensor of the valve plate of the intelligent multi-way valve to be tested as input, and the corresponding probability of the failure mode of the intelligent multi-way valve to be tested as output, the convolutional neural network in the controller is trained and learned based on the BP algorithm and the gradient descent algorithm to form a fault judgment rule.

[0029] The key information includes P port pressure, T port pressure, A port pressure, B port pressure, valve core displacement, valve core control voltage, valve power supply voltage and communication level.

[0030] The failure modes include a no-fault mode, a pipeline leakage mode, a valve core stuck mode, a circuit break mode and a pipeline damage mode.

[0031] In step 2, the controller optimization method of the intelligent multi-way valve to be tested comprises the following steps:

[0032] Step 201, setting the load state:

[0033] Step 201 is the same as step 101 .

[0034] Step 202, calculating the hydraulic theoretical flow:

[0035] Step 202 is the same as step 102 .

[0036] Step 203, controller optimization:

[0037] The neural network embedded in the controller takes the target pressure, target flow, current pressure of each working port, valve core displacement and measured flow as input; the neural network takes the control parameters of the controller of the intelligent multi-way valve to be tested as output, and the neural network uses the mean square error of the control error e as the loss function of the neural network. Based on the BP algorithm, the control parameters of the controller of the multi-way valve to be tested are adjusted in real time through the neural network, and the controller of the intelligent multi-way valve to be tested is trained and optimized.

[0038] Compared with the prior art, the present invention has the following technical effects:

[0039] (I) The intelligent multi-way valve comprehensive training and testing system and method of the present invention can be used for both the training of the fault mode recognition function of the intelligent control system in the intelligent multi-way valve and the optimization of the controller training, thereby realizing diversified testing and training of the intelligent multi-way valve in a laboratory environment, and greatly saving the time and economic cost of switching between different test benches.

[0040] (II) On the other hand, the training and testing functions of the intelligent multi-way valve include both fault simulation and load simulation, which can realize the simulation of common faults of the intelligent multi-way valve and the simulation of different load conditions of the intelligent multi-way valve on the same test bench; and since all experiments can be completed in a laboratory environment, simulations of different faults, loads and complex conditions can be widely carried out, so that the intelligent multi-way valve control system can be fully trained, which can greatly improve the fault identification accuracy and valve core control performance of the intelligent multi-way valve control system.

[0041] (III) To improve the control performance of the intelligent multi-way valve, the fault recognition of the intelligent multi-way valve is realized through the convolutional neural network with more advantages in fault mode recognition, and the training of the intelligent multi-way valve fault recognition network is completed through simulation data; the parameters of the multi-way valve controller are intelligently adjusted through a simple neural network with higher real-time performance, and the intelligent multi-way valve is optimized in future application scenarios through the load simulation control process; by applying fault simulation in normal working scenarios, the emergency handling capabilities of the intelligent multi-way valve under different fault modes are trained. Through the above three operations, the fault mode recognition performance of the intelligent multi-way valve, the optimization of the control performance for the application scenario and the emergency handling performance under the fault mode are finally realized to improve the comprehensive performance of the intelligent multi-way valve.

[0042] (IV) The present invention can simulate the sudden failure condition when the proportional valve is working normally, and take timely response measures when a failure occurs to reduce system leakage and prevent the system from getting out of control. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the composition of the intelligent multi-way valve comprehensive training and testing system.

[0044] Figure 2 It is a schematic diagram of the composition of the pipeline leakage simulation module.

[0045] FIG. 3( a ) is a schematic diagram of the free movement state of the valve core of the valve core stuck simulation module.

[0046] FIG3( b ) is a schematic diagram of the valve core stuck state of the valve core stuck simulation module.

[0047] Figure 4 It is a schematic diagram of the sealing status of the pipeline damage simulation module.

[0048] Figure 5 It is a schematic diagram of the semi-damaged state of the pipeline damage simulation module.

[0049] Figure 6 It is a schematic diagram of the open state of the pipeline damage simulation module.

[0050] Figure 7 It is a schematic diagram of the hydraulic composition of the intelligent multi-way valve comprehensive training and testing system.

[0051] Figure 8 It is the structure diagram of the convolutional neural network for intelligent multi-way valve fault pattern recognition.

[0052] Fig. 9 It is a neural network structure diagram for optimizing the controller of the intelligent multi-way valve.

[0053] The meanings of the numbers in the figure are: 1- pipeline leakage simulation module, 2- valve core sticking simulation module, 3- circuit break simulation module, 4- pipeline damage simulation module, 5- integrated load simulation module, 6- oil source module, 7- intelligent multi-way valve to be tested.

[0054] 101- pipeline, 102- frame, 103- seal, 104- leakage gland, 105- hydraulic cylinder, 106- pressure sensor, 107- servo valve, 108- accumulator, 109- leakage oil collecting groove, 110- oil unloading hole, 111- measuring cup.

[0055] 201-valve body, 202-valve core, 203-extension rod, 204-electromagnetic brake, 205-wedge block.

[0056] 401- pipeline, 402- damaged port, 403- sealed tube core.

[0057] 501 - two-position two-way proportional valve, 502 - first flow valve, 503 - second flow valve.

[0058] 601- oil tank, 602- hydraulic pump, 603- one-way valve, 604- pressure sensor, 605- accumulator, 606- safety valve.

[0059] The specific contents of the present invention are further explained in detail below in conjunction with embodiments. DETAILED DESCRIPTION

[0060] It should be noted that, unless otherwise specified, all parts, devices, modules and networks in the present invention are parts, devices, modules and networks known in the prior art.

[0061] Specific embodiments of the present invention are given below. It should be noted that the present invention is not limited to the following specific embodiments, and all equivalent changes made on the basis of the technical solution of this application fall within the protection scope of the present invention.

[0062] Example:

[0063] This embodiment provides an intelligent multi-way valve comprehensive training and testing system. Figure 1 As shown, the system includes a fault simulation module, which includes a pipeline leakage simulation module 1, a valve core sticking simulation module 2, a circuit breaking simulation module 3 and a pipeline damage simulation module 4.

[0064] like Figure 1 As shown, it also includes a comprehensive load simulation module 5 and an oil source module 6.

[0065] The pipeline leakage simulation module 1 is used to simulate pipeline leakage.

[0066] The valve core stuck simulation module 2 is used to simulate the valve core stuck.

[0067] The circuit breaker simulation module 3 is used to simulate an electrical circuit breaker fault.

[0068] The pipeline damage simulation module 4 is used to simulate pipeline damage.

[0069] The comprehensive load simulation module 5 simulates the pressure and flow load of the intelligent multi-way valve 7 to be trained and tested through a two-position two-way proportional valve, thereby training the control accuracy and stability of the intelligent multi-way valve 7 to be tested when the valve core hydraulic power changes under different pressures and different flow load environments.

[0070] The oil source module 6 is used to provide high-pressure hydraulic oil for the intelligent multi-way valve comprehensive training and testing system.

[0071] As a preferred solution of this embodiment, Figure 2As shown, the pipeline leakage simulation module 1 includes a pipeline 101 connected to the oil port of the intelligent multi-way valve 7 to be tested, the end face of the pipeline 101 is fixed by a frame 102, and the end face of the pipeline 101 is sealed and installed with a leakage gland 104 through a seal 103.

[0072] like Figure 2 As shown, the leakage gland 104 provides pressure through the hydraulic cylinder 105, and the hydraulic cylinder 105 is connected to the servo valve 107 and the accumulator 108 respectively through the hydraulic oil pipe with the pressure sensor 106.

[0073] like Figure 2 As shown, an annular leakage oil collecting groove 109 is opened along the circumferential direction on the inner wall of the leakage gland 104, and the leakage oil collecting groove 109 is connected to an oil discharge hole 110 that passes through the side wall of the leakage gland 104, and a measuring cup 111 is provided at the outer end of the oil discharge hole 110.

[0074] In this embodiment, the pipeline 101 is connected to the corresponding oil port of the intelligent multi-way valve 7 to be tested to simulate the oil port leakage fault; the frame 102 is fixed to the test bench frame to provide force support for the leakage gland 104; the seal 103 and the leakage gland 104 are kept sealed by pressure, and the leakage amount generated is different when the pressure is different, so as to achieve different sealing forces; the leakage gland 104 applies thrust through the hydraulic cylinder 105 to press the leakage gland 104 and the end face of the pipeline 101 tightly to achieve different sealing forces; the hydraulic cylinder 105 is a single-acting hydraulic cylinder, and the hydraulic cylinder 105 measures its internal pressure through the pressure sensor 106, and realizes the closed-loop control of the pressure in the hydraulic cylinder 105 through the servo valve 107; the hydraulic cylinder 105 is connected in parallel with the accumulator 108 to stabilize the pressure in the hydraulic cylinder 105. The leaked oil of the pipeline 101 flows into the measuring cup 111 from the oil discharge hole 110 after passing through the leakage oil collection groove 109.

[0075] As a preferred solution of this embodiment, Figure 3a and Figure 3b As shown, the valve core stuck simulation module 2 includes a valve body 201, in which a valve core 202 is installed, one end of the valve core 202 is connected to one end of an extension rod 203, and the other end of the extension rod 203 extends into the electromagnetic brake 204. The movement of the wedge block 205 in the electromagnetic brake 204 can achieve the jamming or blocking of the extension rod 203.

[0076] In this embodiment, specifically, Figure 3a As shown, when the coil in the electromagnetic brake 204 is not energized, the extension rod 203 and the wedge block 205 are in a clearance fit, and the valve core 202 can move freely in the valve body 201. Figure 3bAs shown, when the coil in the electromagnetic brake 204 is energized, the magnet in the electromagnetic brake 204 moves leftward under the action of the electromagnetic force of the coil, thereby pressing the wedge block 205 and locking the valve core 202.

[0077] As a preferred solution of this embodiment, the circuit breaking simulation module 3 is connected to the electrical circuits of the intelligent multi-way valve 7 to be tested through multi-way relays, and the electrical circuit breaking fault of the intelligent multi-way valve 7 to be tested is simulated by turning on and off the relays; the electrical circuit breaking fault includes power supply circuit breaking fault, communication cable circuit breaking fault and sensor signal line circuit breaking fault.

[0078] As a preferred solution of this embodiment, Figures 4 to 6 As shown, the pipeline damage simulation module 4 includes a pipeline 401, a damage opening 402 is opened on the side wall of the pipeline 401, a sealing tube core 403 that can move in the pipeline 401 is installed in the pipeline 401, and there are three working states between the sealing tube core 403 and the damage opening 402, namely, a sealed state, a semi-damaged state and an open state.

[0079] Sealing status: Figure 4 As shown, the sealing tube core 403 and the sealing end surface reserved on the pipeline 401 are tightly combined, and the pipeline 401 is in a completely sealed state.

[0080] Semi-damaged state: Figure 5 As shown, the sealing tube core 403 partially overlaps with the notch on the pipeline 401, simulating a partial damage state of the pipeline 401, and the hydraulic oil will flow out from the notch.

[0081] Open state: Figure 6 As shown, there is no overlapping area between the sealing tube core 403 and the notch on the pipeline 401, simulating a state where the pipeline 401 is completely cut off, and the hydraulic oil will flow out from the notch without pressure loss.

[0082] As a preferred solution of this embodiment, Figure 7 As shown, the comprehensive load simulation module 5 includes a two-position two-way proportional valve 501 connected to the intelligent multi-way valve 7 to be tested, the oil inlet end of the two-position two-way proportional valve 501 is connected to the first flow valve 502, and the oil outlet end of the two-position two-way proportional valve 501 is connected to the second flow valve 503.

[0083] As a preferred solution of this embodiment, the oil source module 6 adopts a known and commonly used oil source module. The specific connection structure of the preferred oil source module 6 is as follows: Figure 7 Specifically in this embodiment, the oil inlet of the two-position two-way proportional valve 501 is connected to the hydraulic pump 602 in the oil source module 6 through the first flow valve 502 for oil supply, and the oil outlet of the two-position two-way proportional valve 501 is connected to the oil tank 601 in the oil source module 6 through the second flow valve 503 for oil return.

[0084] Embodiment 2:

[0085] This embodiment provides a comprehensive training and testing method for an intelligent multi-way valve. Figure 7 As shown, the method installs the intelligent multi-way valve 7 to be tested in the intelligent multi-way valve comprehensive training and testing system given in Example 1, and performs fault simulation test and load simulation test on the intelligent multi-way valve 7 to be tested.

[0086] Step 1: Fault simulation test:

[0087] In the fault simulation test phase, different fault simulation modules are controlled to generate corresponding fault actions. At the same time, the fault mode recognition function is trained through the convolutional neural network embedded in the controller of the intelligent multi-way valve 7 to be tested.

[0088] In this step, in the fault simulation test phase, the corresponding fault simulation modules are controlled to generate fault actions according to the preset programs, such as the valve core stuck simulation module 2 locks the valve core or holds it with a certain force, so that the valve core is stuck or the movement resistance of the valve core is increased; the circuit breaker simulation module 3 disconnects the electrical modules required to simulate the circuit breaker, such as the communication line, power supply line, sensor signal line, etc., to simulate the communication equipment disconnection, sensor disconnection and other faults; the pipeline leakage simulation module 1 and the pipeline damage simulation module 4 respectively control the leakage end cover pressure and the position of the sealing tube core 403 to simulate different leakage amounts and different pipeline damage state faults. At the same time, the fault recognition function is trained through the convolutional neural network embedded in the intelligent multi-way valve 7 to be tested, and the fault mode recognition accuracy and speed of the intelligent multi-way valve 7 to be tested are improved in the comprehensive fault simulation training process.

[0089] In step 1, the training method of the fault mode recognition function includes the following steps:

[0090] Step 101, setting the load state:

[0091] According to the working conditions of the intelligent multi-way valve 7 to be tested, the key load parameter information is set.

[0092] Key load parameter information includes maximum / minimum load force, hydraulic cylinder bore / rod diameter, motor displacement, and load motion target curve.

[0093] Step 102, calculate the hydraulic theoretical flow:

[0094] According to the load state set in step 101, the theoretical hydraulic flow required for load movement is calculated.

[0095] In this step, for the hydraulic cylinder, the target displacement is x(t), the cylinder diameter is D, and the rod diameter is d, then the theoretical flow rate of the rodless chamber hydraulic pressure is x(t)*πD 2 / 4, the theoretical flow rate of the rod cavity is x(t)*π(D 2 -d 2 ) / 4.

[0096] In this step, for the hydraulic motor, the target angle is θ(t), the target speed is n(t), the displacement is r, and the hydraulic theory is n(t)*r.

[0097] Step 103, training the fault mode recognition function:

[0098] Taking the theoretical flow in step 102 and the key information collected by the controller of the intelligent multi-way valve 7 to be tested according to the built-in sensor of the valve plate of the intelligent multi-way valve 7 to be tested as input, and the corresponding probability of the fault mode of the intelligent multi-way valve 7 to be tested as output, the convolutional neural network in the controller is trained and learned based on the BP (Back Propagation) algorithm and the gradient descent algorithm to form a fault judgment rule.

[0099] Key information includes P port pressure, T port pressure, A port pressure, B port pressure, valve core displacement, valve core control voltage, valve supply voltage and communication level.

[0100] The failure modes include no-fault mode, pipeline leakage mode, valve core sticking mode, circuit breaker mode and pipeline damage mode.

[0101] In this step, the P port, the T port, the A port and the B port are all known ports of the intelligent multi-way valve 7 to be tested.

[0102] Specifically in this embodiment, the convolutional neural network adopts a convolutional neural network known in the art. Figure 8 The figure shows the structure of the convolutional neural network, which consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The convolutional layer extracts the key features of different valve pressure, displacement, flow, and other information, and inputs the key features into the pooling layer. After training and identification by the fully connected layer, the probabilities of different faults are output. Figure 8 The convolutional neural network shown in the figure realizes the training of the fault mode recognition function of the intelligent multi-way valve 7 to be tested. The trained convolutional neural network will be embedded in the intelligent multi-way valve 7 to be tested to improve its fault mode recognition performance.

[0103] In this step, the fault mode recognition training is completed through a large amount of training data to ensure that the accuracy of the probability distribution of the neural network output terminal meets the use requirements. By randomly simulating the fault state, the fault mode recognition performance of the intelligent multi-way valve 7 to be tested is verified, and finally the intelligent multi-way valve 7 to be tested can accurately identify different fault modes.

[0104] Step 2: Load simulation test:

[0105] During the load simulation stage, the working port pressure of the intelligent multi-way valve 7 to be tested is controlled by the two-position two-way proportional valve in the comprehensive load simulation module 5, the outlet flow of the overflow valve is measured by the flowmeter, and the flow through the intelligent multi-way valve 7 to be tested is controlled by the intelligent multi-way valve 7 to be tested, thereby realizing flow closed-loop control training and testing under different working pressures.

[0106] At the same time, the controller optimization training is carried out through the neural network embedded in the controller of the intelligent multi-way valve 7 to be tested. According to different specific application scenarios and working conditions, the control parameters are adjusted in real time through the neural network, thereby realizing the optimization of the controller of the intelligent multi-way valve 7 to be tested.

[0107] In step 2, the controller optimization method of the intelligent multi-way valve 7 to be tested includes the following steps:

[0108] Step 201, setting the load state:

[0109] Step 201 is the same as step 101 .

[0110] Step 202, calculating the hydraulic theoretical flow:

[0111] Step 202 is the same as step 102 .

[0112] Step 203, controller optimization:

[0113] The neural network embedded in the controller takes the target pressure (simulated load force), target flow (simulated displacement), current pressure of each working port, valve core displacement and measured flow (simulated displacement sensor data) as input. The neural network takes the control parameters of the controller of the intelligent multi-way valve 7 to be tested as output, and the neural network takes the mean square error of the control error e as the loss function of the neural network. Based on the BP algorithm, the control parameters of the controller of the multi-way valve are adjusted in real time through the neural network, and the controller of the intelligent multi-way valve 7 to be tested is trained and optimized.

[0114] In this step, the control parameters, taking PID as an example, include known parameters Kp, Ki and Kd.

[0115] In this step, in order to ensure the valve core control performance of the intelligent multi-way valve 7 to be tested, another relatively simple neural network is used with the target pressure (simulated load force), flow (simulated displacement), current working port pressure, valve core displacement, measured flow (simulated displacement sensor data), etc. as input, and the control parameters of the controller of the intelligent multi-way valve 7 to be tested (taking PID as an example), that is, Kp, Ki, Kd as the output of the neural network, and the mean square error of the control error e is used as the loss function of the neural network. Based on the BP algorithm, Fig. 9 The control structure shown is used to train and optimize the multi-way valve controller.

[0116] Specifically in this embodiment, the neural network adopts a neural network known in the art. Fig. 9 The neural network structure is shown in the figure. The neural network is connected to the controller, and the neural network outputs the three sets of control parameters of the controller. Fig. 9 The neural network shown realizes the training optimization of the controller of the intelligent multi-way valve 7 to be tested, so as to improve the targeted improvement of the control performance of the intelligent multi-way valve 7 to be tested in the application scenario, and finally realizes the improvement of the comprehensive performance of the intelligent multi-way valve 7 to be tested.

[0117] Furthermore, in this step, according to the pressure-flow curve of the specific working condition, training and control parameter optimization are carried out for all working conditions required for the intelligent multi-way valve 7 to be tested, and the control parameters of the corresponding working conditions are solidified, so as to achieve targeted optimization of the control performance of the intelligent multi-way valve 7 to be tested in the usage scenario before leaving the factory.

Claims

1. An intelligent multi-way valve comprehensive training and testing system, characterized in that: The system comprises a fault simulation module, wherein the fault simulation module comprises a pipeline leakage simulation module (1), a valve core sticking simulation module (2), a circuit breaking simulation module (3) and a pipeline damage simulation module (4); It also includes a comprehensive load simulation module (5).

2. The intelligent multi-way valve comprehensive training and testing system according to claim 1, characterized in that: The pipeline leakage simulation module (1) comprises a pipeline (101) connected to the oil port of the intelligent multi-way valve (7) to be tested, the end face of the pipeline (101) is fixed by a frame (102), and the end face of the pipeline (101) is sealed and installed with a leakage gland (104) through a sealing member (103); The leakage gland (104) provides pressure through a hydraulic cylinder (105), and the hydraulic cylinder (105) is connected to a servo valve (107) and an accumulator (108) respectively through a hydraulic oil pipe with a pressure sensor (106). An annular leakage oil collecting groove (109) is provided on the inner side wall of the leakage gland (104) along the circumferential direction. The leakage oil collecting groove (109) is connected to an oil discharge hole (110) penetrating the side wall of the leakage gland (104). A measuring cup (111) is provided at the outer end of the oil discharge hole (110).

3. The intelligent multi-way valve comprehensive training and testing system according to claim 1, characterized in that: The valve core stuck simulation module (2) comprises a valve body (201), a valve core (202) is installed in the valve body (201), one end of the valve core (202) is connected to one end of an extension rod (203), the other end of the extension rod (203) extends into the electromagnetic brake (204), and the movement of the wedge block (205) in the electromagnetic brake (204) can achieve the stuck or jammed extension rod (203).

4. The intelligent multi-way valve comprehensive training and testing system according to claim 1, characterized in that: The circuit breaking simulation module (3) is connected to the electrical circuits of the intelligent multi-way valve (7) to be tested respectively through multi-way relays, and the simulation of the electrical circuit breaking fault of the intelligent multi-way valve (7) to be tested is realized by turning on and off the relays; the electrical circuit breaking fault includes a power supply circuit breaking fault, a communication cable circuit breaking fault, and a sensor signal line circuit breaking fault.

5. The intelligent multi-way valve comprehensive training and testing system according to claim 1, characterized in that: The pipeline damage simulation module (4) comprises a pipeline (401), a damage opening (402) is provided on the side wall of the pipeline (401), a sealing tube core (403) which can move in the pipeline (401) is installed in the pipeline (401), and there are three working states between the sealing tube core (403) and the damage opening (402), namely a sealed state, a semi-damaged state and an open state.

6. The intelligent multi-way valve comprehensive training and testing system according to claim 1, characterized in that: The comprehensive load simulation module (5) comprises a two-position two-way proportional valve (501) connected to the intelligent multi-way valve (7) to be tested, the oil inlet end of the two-position two-way proportional valve (501) being connected to the first flow valve (502), and the oil outlet end of the two-position two-way proportional valve (501) being connected to the second flow valve (503).

7. A comprehensive training and testing method for intelligent multi-way valves, characterized in that: The method installs the intelligent multi-way valve (7) to be tested in the intelligent multi-way valve comprehensive training and testing system as claimed in any one of claims 1 to 6, and performs a fault simulation test and a load simulation test on the intelligent multi-way valve (7) to be tested respectively; Step 1: Fault simulation test: In the fault simulation test phase, different fault simulation modules are controlled to generate corresponding fault actions, and at the same time, the convolutional neural network embedded in the controller of the intelligent multi-way valve (7) to be tested is used to train the fault mode recognition function; Step 2: Load simulation test: In the load simulation stage, the working port pressure of the intelligent multi-way valve (7) to be tested is controlled by a two-position two-way proportional valve in the comprehensive load simulation module (5), the outlet flow of the overflow valve is measured by a flow meter, and the flow through the intelligent multi-way valve (7) to be tested is controlled by the intelligent multi-way valve (7) to be tested, thereby realizing flow closed-loop control training and testing under different working pressures; At the same time, optimization training of the controller is carried out through the neural network embedded in the controller of the intelligent multi-way valve (7) to be tested, and the control parameters are adjusted in real time through the neural network for different specific application scenarios and working conditions, thereby achieving optimization of the controller of the intelligent multi-way valve (7) to be tested.

8. The intelligent multi-way valve comprehensive training and testing method according to claim 7, characterized in that: In step 1, the training method of the fault mode recognition function comprises the following steps: Step 101, setting the load state: According to the working conditions of the intelligent multi-way valve (7) to be tested, key load parameter information is set; The key load parameter information includes the maximum / minimum load force, hydraulic cylinder bore / rod diameter, motor displacement, and load motion target curve; Step 102, calculate the hydraulic theoretical flow: According to the load state set in step 101, the theoretical hydraulic flow required for load movement is calculated; Step 103, training the fault mode recognition function: Taking the theoretical flow rate in step 102 and key information collected by the controller of the intelligent multi-way valve (7) to be tested according to the built-in sensor of the valve plate of the intelligent multi-way valve (7) to be tested as input, taking the corresponding probability of the fault mode of the intelligent multi-way valve (7) to be tested as output, based on the BP algorithm and the gradient descent algorithm, the convolutional neural network in the controller is trained and learned to form a fault judgment rule; The key information includes P port pressure, T port pressure, A port pressure, B port pressure, valve core displacement, valve core control voltage, valve power supply voltage and communication level; The failure modes include a no-fault mode, a pipeline leakage mode, a valve core stuck mode, a circuit break mode and a pipeline damage mode.

9. The intelligent multi-way valve comprehensive training and testing method according to claim 8, characterized in that: In step 2, the controller optimization method of the intelligent multi-way valve (7) to be tested comprises the following steps: Step 201, setting the load state: Step 201 is the same as step 101; Step 202, calculating the hydraulic theoretical flow: Step 202 is the same as step 102; Step 203, controller optimization: The neural network embedded in the controller takes the target pressure, target flow, current pressure of each working port, valve core displacement and measured flow as input; the neural network takes the control parameters of the controller of the intelligent multi-way valve (7) to be tested as output, and the neural network takes the mean square error of the control error e as the loss function of the neural network. Based on the BP algorithm, the control parameters of the controller of the multi-way valve are adjusted in real time through the neural network, and the controller of the intelligent multi-way valve (7) to be tested is trained and optimized.

Images