Control system of three-position four-way intelligent valve terminal

Abstract

The invention discloses a control system of a three-position four-way intelligent valve terminal, and relates to the technical field of remote control. In a reversing prediction stage, an LSTM (Long Short Term Memory) network is adopted to carry out reversing trend analysis, and reversing parameters are adjusted in advance, so that the reversing process is stable; dQN deep reinforcement learning is additionally introduced, and reversing time and back pressure compensation are dynamically adjusted, so that the system can adapt to different load conditions; for execution errors caused by remote control lag, remote signal time delay prediction compensation is added in an AI anomaly detection module, and when instruction transmission delay is detected, a system can adjust reversing time in advance, so that the synchronism of an execution mechanism is improved, and the operation precision is improved; fixed reversing logic is usually adopted in a traditional control mode, complex working conditions are difficult to deal with, while the scheme is combined with fuzzy logic control, the position of a reversing handle, reversing time and a pressure compensation strategy are dynamically corrected according to the current running state, and errors caused by the fixed logic are avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of remote control, and in particular to a control system for a three-position four-way intelligent valve island. Background Art

[0002] In the remote control system of well control equipment, a three-position four-way directional control valve is used to control the hydraulic flow direction to ensure the normal operation of key equipment such as blowout preventers and hydraulic gates.

[0003] Traditional control methods rely on proportional valves or solenoid valves and combine with PLC for logic control. However, in actual applications, pressure fluctuations are likely to occur during the commutation process. In a high-pressure environment, the switching of the directional control valve may cause instantaneous pressure mutations, resulting in impacts on the actuator during operation, accelerating the wear of hydraulic components. In an emergency situation, if the system fails to quickly establish pressure, it may lead to a lag in the response of the actuator, affecting the safety of well control equipment. For this reason, some traditional solutions use accumulators or adjust the oil circuit design to relieve pressure shocks. However, under complex load changes, these methods are difficult to ensure smooth pressure regulation, and remote control further amplifies this problem.

[0004] In remote control, due to a certain delay in signal transmission and the inertia of the hydraulic system itself, it is difficult to synchronize the execution of the directional control valve, which further affects the operation accuracy. When alternating between manual and remote operations, the position of the handle is very crucial. If the handle of the directional control valve is in the middle position, the system needs to re-establish pressure during commutation, which may lead to a response lag. If the handle is maintained in the working position for a long time, the hydraulic pump needs to continuously supply pressure, increasing energy consumption and accelerating the wear of components. Different application scenarios have different requirements for these two modes, and the traditional PLC control method is difficult to balance during the adjustment process. It is difficult to completely eliminate commutation lag and avoid additional energy consumption. Therefore, there is an urgent need for a control system for a three-position four-way intelligent valve island to solve such problems. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a control system for a three-position four-way intelligent valve island to solve the problems of large commutation pressure fluctuations, remote control lag, difficult balance in handle position selection, high energy consumption, and serious component wear.

[0007] To solve the above technical problems, the present invention provides the following technical solutions:

[0008] An embodiment of the present invention provides a control system for a three-position four-way intelligent valve island, which includes,

[0009] The data acquisition module is used to collect the operating parameters of the hydraulic system, monitor the pressure fluctuation state of the reversing valve, and store the operating parameters and pressure fluctuation state data in the control system as operating data;

[0010] The operating parameters include the inlet pressure, outlet pressure, flow rate, and temperature;

[0011] The commutation prediction module analyzes the commutation trend using an AI prediction model based on historical operating data, and adjusts the commutation strategy based on the commutation trend;

[0012] The remote control module determines whether an abnormality occurs during the commutation process based on the real-time monitored operating data;

[0013] In the remote control module, the abnormalities include pressure fluctuation, commutation hysteresis, and overpressure supply;

[0014] The feedback control module dynamically corrects the handle position, commutation time, and backpressure compensation during the commutation process based on the abnormality monitoring results of the remote control module.

[0015] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, wherein: the control method of the system includes:

[0016] Step S1, the data acquisition module collects the operating parameters of the hydraulic system and monitors the pressure fluctuation state of the reversing valve;

[0017] Step S2, based on the historical operating data stored in Step S1, the commutation prediction module analyzes the commutation trend using an AI prediction model, and adjusts the commutation strategy based on the commutation trend;

[0018] Step S3, based on the real-time monitored data in Step S1, the remote control module determines whether an abnormality occurs during the commutation process;

[0019] Step S4, based on the abnormality monitoring results in Step S3, the feedback control module dynamically corrects the handle position, commutation time, and backpressure compensation during the commutation process;

[0020] Step S5, based on the corrected commutation control parameters in Step S4, perform the commutation operation and control the actuator to complete the corresponding action.

[0021] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, wherein: the operating parameters include the inlet pressure, outlet pressure, flow rate, and temperature, and the pressure fluctuation state reflects the pressure change of the reversing valve under different working conditions;

[0022] Store the operating parameters and pressure fluctuation state data in the control system, and input them as operating data to the commutation prediction module and the remote control module.

[0023] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, wherein: the analysis of the commutation trend includes predicting the response characteristics of the directional valve under different load and pressure conditions and optimizing the commutation parameters;

[0024] Generate an optimized commutation strategy and store it in the control system for the remote control module to call.

[0025] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, wherein: the step of analyzing the commutation trend based on the historical operation data stored in step S1 by the commutation prediction module using an AI prediction model and adjusting the commutation strategy based on the commutation trend includes,

[0026] Predict the commutation trend based on the long short-term memory network LSTM:

[0027] Collect the operating parameters of the hydraulic system, including the inlet pressure, outlet pressure, flow rate and temperature, monitor the pressure fluctuation state of the directional valve, form time series data, and perform normalization processing. The normalization formula is:

[0028]

[0029] Wherein, X′ t represents the normalized data, X t represents the original data, X min , X max the minimum and maximum values;

[0030] Train the LSTM commutation trend prediction model, and set the commutation state H t as a time series:

[0031] H t = f(P in,t , P out,t , Q t , T t , S t ),

[0032] Use the LSTM network to calculate the commutation trend at the next moment. The prediction formula is:

[0033] H t+1 = σ(WH t + b),

[0034] Wherein, H t represents the commutation state at the current moment, f(·) represents the commutation state function, P in,t represents the inlet pressure at the current moment, P out,t represents the outlet pressure at the current moment, Q tRepresents the flow rate at the current moment, T t Represents the temperature at the current moment, S t Represents the state of the reversing valve at the current moment, H t+1 Represents the predicted value of the reversing state at the next moment. σ(·) represents a non - linear activation function. The ReLU and Sigmoid functions are selected. W represents the weight matrix of the LSTM network, b represents the bias term, and t represents the current time step.

[0035] As a preferred solution of the control system of the three - position four - way intelligent valve island described in the present invention, wherein: the step of analyzing the reversing trend by the reversing prediction module using the AI prediction model based on the historical operation data stored in step S1 and adjusting the reversing strategy further includes

[0036] Optimizing the reversing strategy based on the deep Q - network DQN. The reversing optimization objective is set as:

[0037]

[0038] Wherein, T represents the total number of time steps, λ 1 , λ 2 , λ 3 Represents the weight parameter of the optimization objective, λ 1 , λ 2 , λ 3 Determined by the Bayesian optimization method. The determination formula is:

[0039]

[0040] Wherein, f(λ) is the loss function, E[·] represents the calculation of the expected value, and the optimal λ is found through multiple rounds of iteration 1 , λ 2 , λ 3 value;

[0041] P diff,t Represents the change in oil pressure during the reversing process at the current moment, T sw,t Represents the reversing time at the current moment, T sw,opt Represents the optimal reversing time;

[0042] Optimize the reversing parameters using DQN reinforcement learning:

[0043] Adjust the reversing time through reinforcement learning. The adjustment formula is:

[0044]

[0045] Wherein, Represents the optimized reversing time, T sw,0 Represents the initial reversing time, α represents the learning rate, ΔP represents the change in load pressure,

[0046] The optimal commutation control strategy is trained using DQN, and the training process is expressed as:

[0047] u t = π(H t ),

[0048] where u t represents the optimized commutation control parameter, and π(H t ) represents the reinforcement learning policy function.

[0049] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, wherein: in step S3, the abnormalities include pressure fluctuations, commutation hysteresis, and overpressure supply;

[0050] When an abnormality is detected, an abnormal state signal is generated and input into the feedback control module.

[0051] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, wherein: in step S4, the dynamic correction includes adjusting the control parameters of the commutation valve based on the abnormal state signal, generating the corrected commutation control parameter, and inputting it into the actuator.

[0052] As a preferred solution of the control system of a three-position four-way intelligent valve island according to the present invention, the steps of dynamically correcting the handle position, commutation time, and backpressure compensation during the commutation process based on the abnormal monitoring result in step S3 include,

[0053] Input the abnormal monitoring result, set as the input variable X of fuzzy control flc :

[0054] X flc =(P e , T e , S e ),

[0055] where X flc represents the input variable vector of fuzzy control, P e represents the pressure error during the commutation process, that is, the deviation between the desired pressure and the actual pressure, T e represents the commutation time error during the commutation process, that is, the deviation between the desired commutation time and the actual commutation time, S e represents the handle displacement error during the commutation process, that is, the deviation between the desired handle position and the actual handle position,

[0056] Set fuzzy rules, adopt fuzzy control based on the rule base, and combine an online adaptive adjustment mechanism:

[0057] Y flc = g(Xflc ),

[0058] where Y flc represents the corrected commutation control parameter, and g(X lfc ) represents the control function based on fuzzy inference.

[0059] Fuzzy control decision table is used for inference to calculate the correction value. The formula is:

[0060] Y flc = w 1 P e + w 2 T e + w 3 S e ,

[0061] where w 1 , w 2 , w 3 represent the fuzzy inference weight coefficients, which are used to balance the influence of each error term on the correction result. w 1 , w 2 , w 3 An online optimization strategy based on gradient descent is adopted for weight adaptive adjustment. The adjustment formula is:

[0062]

[0063] where represents the weight value at the t-th iteration, η represents the learning rate, and J represents the optimization objective function, that is, the sum of squares of the commutation control error.

[0064] As a preferred solution of the control system of the three-position four-way intelligent valve island described in the present invention, where: the step of dynamically correcting the handle position, commutation time, and backpressure compensation during the commutation process by the feedback control module based on the abnormal monitoring result of step S3 further includes

[0065] Real-time adaptively and dynamically adjust the handle position, and use fuzzy control to calculate the optimal handle displacement. The calculation formula is:

[0066]

[0067] where represents the corrected handle position, S h represents the original handle position, k 1 , k 2 represent the adaptive gain coefficients, k 1 , k 2 An adaptive adjustment strategy based on dynamic optimization is adopted to calculate the gain parameters. The formula is:

[0068]

[0069] Among them, represents the gain value at the t-th iteration, λ represents the gain adjustment step size,

[0070] Predict the remote control instruction delay τ d , and adjust the commutation time in advance. The adjustment formula is:

[0071]

[0072] Among them, represents the corrected commutation time, T sw represents the original commutation time, γ represents the compensation factor, τ d represents the transmission delay of the remote control signal,

[0073] Adopt an intelligent pressure regulating valve for dynamic adjustment. The adjustment formula is:

[0074]

[0075] Among them, represents the corrected backpressure compensation, P b represents the original backpressure compensation, β represents the pressure regulation coefficient, P diff represents the instantaneous pressure change during the commutation process,

[0076] Through the flow regulation algorithm, optimize the oil flow control. The optimization formula is:

[0077] Q * = Q + μ(P e + T e ),

[0078] Among them, Q * represents the corrected flow rate, Q represents the original flow rate, and μ represents the flow rate adjustment coefficient;

[0079] Execute the corrected commutation parameters, calculate the correction parameters, and generate the set U of corrected commutation control parameters corr :

[0080] Transmit the corrected parameters to the actuator to complete the dynamic optimization of the commutation process.

[0081] The beneficial effects of the present invention are:

[0082] In the present invention, during the commutation prediction stage, the LSTM long short-term memory network is used to analyze the commutation trend, and the commutation parameters are adjusted in advance to make the commutation process stable; additionally, DQN deep reinforcement learning is introduced to dynamically adjust the commutation time and backpressure compensation, enabling the system to adapt to different load conditions and solving the problem of unstable commutation.

[0083] In the present invention, in view of the execution error caused by remote control lag, remote signal delay prediction and compensation are added to the AI anomaly detection module. When it detects that the instruction transmission is delayed, the system can adjust the commutation time in advance, so as to improve the synchronization of the actuator, thereby improving the operation accuracy. The traditional control method usually adopts a fixed commutation logic, which is difficult to cope with complex working conditions. However, this solution combines fuzzy logic control to dynamically correct the commutation handle position, commutation time and pressure compensation strategy according to the current operating state, making the commutation control more intelligent and avoiding errors caused by fixed logic.

[0084] In addition, long-term high-pressure oil supply is the key factor leading to the wear of hydraulic components and the increase of energy consumption. Therefore, with the cooperation of an intelligent pressure regulating valve and a flow adaptive control algorithm, the present invention realizes intelligent backpressure management, maintains an appropriate backpressure in the standby state of the system, avoids unnecessary high-pressure oil supply, and adjusts the flow according to the working conditions during the commutation execution process, further reducing the system energy consumption and improving the overall operation efficiency.

[0085] In summary, under the combined action of LSTM prediction and reinforcement learning to optimize the commutation strategy and intelligent feedback control, the present invention makes the control of the reversing valve more accurate and stable, improves the reliability of remote operation, reduces the commutation impact and response lag of the system, and effectively reduces the energy consumption and the wear of hydraulic components at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained according to these drawings.

[0087] Figure 1 It is a schematic framework diagram of the control system of a three-position four-way intelligent valve island of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0088] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification.

[0089] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0090] Secondly, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments.

[0091] Embodiment 1, referring to Figure 1 , this embodiment provides a control system for a three-position four-way intelligent valve island, including:

[0092] A data acquisition module, configured to acquire the operating parameters of the hydraulic system, monitor the pressure fluctuation state of the directional valve, and store the operating parameters and pressure fluctuation state data in the control system as operating data;

[0093] The operating parameters include the inlet pressure, outlet pressure, flow rate, and temperature;

[0094] A commutation prediction module, based on historical operating data, analyzes the commutation trend using an AI prediction model, and adjusts the commutation strategy based on the commutation trend;

[0095] A remote control module, based on the real-time monitored operating data, determines whether an abnormality occurs during the commutation process;

[0096] In the remote control module, the abnormalities include pressure fluctuation, commutation hysteresis, and overpressure supply;

[0097] A feedback control module, based on the abnormality monitoring results of the remote control module, dynamically corrects the handle position, commutation time, and backpressure compensation during the commutation process accordingly.

[0098] This embodiment also provides a control method for the control system of the above three-position four-way intelligent valve island, including:

[0099] Step S1, the data acquisition module acquires the operating parameters of the hydraulic system and monitors the pressure fluctuation state of the directional valve;

[0100] The operating parameters include the inlet pressure, outlet pressure, flow rate, and temperature, and the pressure fluctuation state reflects the pressure change of the directional valve under different working conditions;

[0101] The operating parameters and pressure fluctuation state data are stored in the control system and input into the commutation prediction module and the remote control module as operating data;

[0102] Step S2, based on the historical operating data stored in Step S1, the commutation prediction module analyzes the commutation trend using an AI prediction model and adjusts the commutation strategy based on the commutation trend;

[0103] The analysis of the commutation trend includes predicting the response characteristics of the directional valve under different load and pressure conditions and optimizing the commutation parameters;

[0104] Generate an optimized commutation strategy and store it in the control system for the remote control module to call;

[0105] Based on the historical operation data stored in step S1, the steps for the commutation prediction module to analyze the commutation trend using the AI prediction model and adjust the commutation strategy include,

[0106] Predict the commutation trend based on the long short-term memory network LSTM:

[0107] Collect the operating parameters of the hydraulic system, including the inlet pressure, outlet pressure, flow rate, and temperature, monitor the pressure fluctuation state of the commutation valve, form time series data, and perform normalization processing. The normalization formula is:

[0108]

[0109] where X′ t represents the normalized data, X t represents the original data, X min ,X max is the minimum and maximum values;

[0110] Train the LSTM commutation trend prediction model, and set the commutation state H t as the time series:

[0111] H t = f(P in,t ,P out,t ,Q t ,T t ,S t ),

[0112] Use the LSTM network to calculate the commutation trend at the next moment. The prediction formula is:

[0113] H t+1 = σ(WH t + b),

[0114] where H t represents the commutation state at the current moment, f(·) represents the commutation state function, P in,t represents the inlet pressure at the current moment, P out,t represents the outlet pressure at the current moment, Q t represents the flow rate at the current moment, T t represents the temperature at the current moment, S t represents the commutation valve state at the current moment, H t+1It represents the predicted value of the commutation state at the next moment. σ(·) represents the non-linear activation function. The ReLU and Sigmoid functions are selected. W represents the weight matrix of the LSTM network, b represents the bias term, and t represents the current time step.

[0115] Based on the historical operation data stored in step S1, the step of the commutation prediction module using the AI prediction model to analyze the commutation trend and adjusting the commutation strategy based on the commutation trend further includes

[0116] Optimizing the commutation strategy based on the deep Q-network DQN. The commutation optimization goal is set as:

[0117]

[0118] where T represents the total number of time steps, λ 1 , λ 2 , λ 3 represents the weight parameter of the optimization goal, λ 1 , λ 2 , λ 3 is determined by using the Bayesian optimization method. The determination formula is:

[0119]

[0120] where f(λ) is the loss function, E[·] represents the calculation of the expected value, and the optimal λ 1 , λ 2 , λ 3 value is found through multiple rounds of iteration;

[0121] P diff,t represents the change in oil pressure during the commutation process at the current moment, T sw,t represents the commutation time at the current moment, T sw,opt represents the optimal commutation time;

[0122] Optimizing the commutation parameters using DQN reinforcement learning:

[0123] Adjusting the commutation time through reinforcement learning. The adjustment formula is:

[0124]

[0125] where represents the optimized commutation time, T sw,0 represents the initial commutation time, α represents the learning rate, and ΔP represents the change in load pressure,

[0126] Training the optimal commutation control strategy using DQN. The training process is expressed as:

[0127] u t = π(H t ),

[0128] Among them, u t represents the optimized commutation control parameter, and π(H t ) represents the reinforcement learning policy function;

[0129] Specifically, in step S2, a long short-term memory network (LSTM) is used to predict the trend of the historical operation data of the hydraulic system, and combined with a deep Q-network for commutation strategy optimization;

[0130] First, data preprocessing is performed on the inlet pressure, outlet pressure, flow rate, temperature, and commutation valve state, and key features are extracted through normalization and principal component analysis (PCA). Then, based on the LSTM time series prediction model, the change trend of the commutation state is calculated to obtain the commutation response at a future moment;

[0131] In terms of commutation strategy optimization, the DQN aims at the optimal commutation time, minimum pressure fluctuation, and reasonable backpressure compensation, and uses reinforcement learning for parameter optimization so that the model can adapt to different load conditions;

[0132] Here, through AI prediction and reinforcement learning optimization, the response accuracy of the commutation process of the hydraulic system is improved, and the system shock caused by commutation lag or pressure fluctuation is reduced;

[0133] In step S3, based on the real-time monitoring data in step S1, the remote control module determines whether an abnormality occurs during the commutation process;

[0134] In step S3, the abnormalities include pressure fluctuation, commutation hysteresis, and overpressure supply;

[0135] When an abnormality is detected, an abnormal state signal is generated and input into the feedback control module;

[0136] In step S4, based on the abnormal monitoring results in step S3, the feedback control module dynamically corrects the handle position, commutation time, and backpressure compensation during the commutation process;

[0137] In step S4, the dynamic correction includes adjusting the control parameters of the commutation valve based on the abnormal state signal, generating the corrected commutation control parameter, and inputting it into the actuator;

[0138] Based on the abnormal monitoring results in step S3, the steps for the feedback control module to dynamically correct the handle position, commutation time, and backpressure compensation during the commutation process include,

[0139] Input the abnormal monitoring results, set as the input variable X of fuzzy control flc :

[0140] X flc =(P e ,T e ,Se ),

[0141] where X flc represents the input variable vector of fuzzy control, and P e represents the pressure error during the commutation process, that is, the deviation between the desired pressure and the actual pressure. T e represents the commutation time error during the commutation process, that is, the deviation between the desired commutation time and the actual commutation time. S e represents the handle displacement error during the commutation process, that is, the deviation between the desired handle position and the actual handle position.

[0142] Set fuzzy rules, adopt fuzzy control based on the rule base, and combine an online adaptive adjustment mechanism:

[0143] Y flc = g(X flc ),

[0144] where Y flc represents the corrected commutation control parameter, and g(X flc ) represents the control function based on fuzzy inference.

[0145] Use a fuzzy control decision table for inference and calculate the correction value. The formula is:

[0146] Y flc = w 1 P e + w 2 T e + w 3 S e ,

[0147] where w 1 , w 2 , w 3 represent the fuzzy inference weight coefficients, which are used to balance the influence of each error term on the correction result. w 1 , w 2 , w 3 Adopt an online optimization strategy based on gradient descent for weight adaptive adjustment. The adjustment formula is:

[0148]

[0149] where represents the weight value at the t-th iteration, η represents the learning rate, and J represents the optimization objective function, that is, the sum of the squares of the commutation control errors;

[0150] Based on the abnormal monitoring results in step S3, the steps for the feedback control module to dynamically correct the handle position, commutation time, and backpressure compensation during the commutation process also include

[0151] The handle position is adjusted in real-time and adaptively. The optimal handle displacement is calculated using fuzzy control. The calculation formula is:

[0152]

[0153] Among them, represents the corrected handle position, S h represents the original handle position, k 1 , k 2 represents the adaptive gain coefficient, k 1 , k 2 The gain parameter is calculated using an adaptive adjustment strategy based on dynamic optimization. The formula is:

[0154]

[0155] Among them, represents the gain value at the t-th iteration, λ represents the gain adjustment step,

[0156] Predict the remote control instruction delay τ d , and adjust the commutation time in advance. The adjustment formula is:

[0157]

[0158] Among them, represents the corrected commutation time, T sw represents the original commutation time, γ represents the compensation factor, τ d represents the remote control signal transmission delay,

[0159] The dynamic adjustment is performed using an intelligent pressure regulating valve. The adjustment formula is:

[0160]

[0161] Among them, represents the corrected backpressure compensation, P b represents the original backpressure compensation, β represents the pressure regulation coefficient, P diff represents the instantaneous pressure change during the commutation process,

[0162] The oil flow control is optimized through a flow regulation algorithm. The optimization formula is:

[0163] Q * = Q + μ(P e + T e ),

[0164] Among them, Q * represents the corrected flow rate, Q represents the original flow rate, and μ represents the flow rate adjustment coefficient;

[0165] Execute the corrected commutation parameters, calculate the correction parameters, and generate a set of corrected commutation control parameters U corr :

[0166] Transmit the corrected parameters to the actuator to complete the dynamic optimization of the commutation process;

[0167] Specifically, here, the abnormal commutation state is corrected in real time based on fuzzy logic control to ensure the stability of the commutation process. Specifically:

[0168] Extract the pressure error, commutation time error, and handle displacement error as the fuzzy control inputs, and use fuzzy inference based on the rule base to calculate the corrected commutation control parameters. To further optimize the accuracy of fuzzy inference, the gradient descent method is used for adaptive optimization of the weight parameters, so that the error weights are dynamically adjusted according to the system state changes; In terms of dynamic correction, an adaptive gain adjustment strategy is used to calculate the handle position correction gain, and the predicted remote control signal delay is used for commutation advance compensation;

[0169] In addition, for the problem of hydraulic shock, an intelligent pressure regulating valve is used for back pressure compensation, and the flow rate during the commutation process is controlled to change smoothly through flow rate adaptive adjustment, so as to achieve real-time commutation correction based on fuzzy logic control FLC. Combining adaptive optimization and predictive compensation can effectively reduce the lag, shock, and overpressure during the commutation process, and improve the response speed and stability of the system;

[0170] Step S5, based on the corrected commutation control parameters in step S4, perform the commutation operation and control the actuator to complete the corresponding actions.

[0171] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A control system for a three-position four-way intelligent valve island, characterized in that: include, A data acquisition module is used to collect operating parameters of the hydraulic system and monitor the pressure fluctuation state of the reversing valve, and store the operating parameters and pressure fluctuation state data in the control system as operating data; The operating parameters include oil inlet pressure, oil outlet pressure, flow rate and temperature; The commutation prediction module uses an AI prediction model to analyze commutation trends based on historical operation data and adjusts the commutation strategy based on the commutation trends; The remote control module determines whether an abnormality occurs during the commutation process based on real-time monitoring of operating data; In the remote control module, the anomalies include pressure fluctuations, switching hysteresis and excessive pressure supply; The feedback control module dynamically corrects the handle position, switching time and back pressure compensation during the switching process based on the abnormal monitoring results of the remote control module.

2. A three-position four-way intelligent valve island control system as claimed in claim 1, characterized in that: The control method of the system includes: Step S1, the data acquisition module collects the operating parameters of the hydraulic system and monitors the pressure fluctuation state of the reversing valve; Step S2, based on the historical operation data stored in step S1, the commutation prediction module uses an AI prediction model to analyze the commutation trend, and adjusts the commutation strategy based on the commutation trend; Step S3, based on the real-time monitoring data of step S1, the remote control module determines whether an abnormality occurs during the commutation process; Step S4, based on the abnormal monitoring result of step S3, the feedback control module dynamically corrects the handle position, switching time and back pressure compensation during the switching process; Step S5, based on the modified commutation control parameters of step S4, execute the commutation operation and control the actuator to complete the corresponding action.

3. A three-position four-way intelligent valve island control system as claimed in claim 2, characterized in that: The operating parameters include oil inlet pressure, oil outlet pressure, flow rate and temperature, and the pressure fluctuation state reflects the pressure change of the reversing valve under different working conditions; The operating parameters and pressure fluctuation status data are stored in the control system and input into the reversing prediction module and the remote control module as operating data.

4. A control system for a three-position four-way intelligent valve island as claimed in claim 3, characterized in that: The analysis of the switching trend includes predicting the response characteristics of the switching valve under different load and pressure conditions and optimizing the switching parameters; Generate an optimized commutation strategy and store it in the control system for the remote control module to call.

5. A three-position four-way intelligent valve island control system as claimed in claim 4, characterized in that: The step of using the AI ​​prediction model to analyze the commutation trend based on the historical operation data stored in step S1 and adjusting the commutation strategy based on the commutation trend includes: Commutation trend prediction based on long short-term memory network LSTM: Collect the operating parameters of the hydraulic system, including the oil inlet pressure, oil outlet pressure, flow rate and temperature, monitor the pressure fluctuation state of the reversing valve, form time series data, and perform normalization. The normalization formula is: Among them, X′ t represents the normalized data, X t represents the original data, X min ,X max The minimum and maximum values ​​of Train the LSTM commutation trend prediction model and set the commutation state H t As a time series: H t =f(P in,t ,P out,t ,Q t ,T t ,S t ), The LSTM network is used to calculate the reversing trend at the next moment. The prediction formula is: H t+1 =σ(WH t +b), Among them, H t represents the current commutation state, f(·) represents the commutation state function, P in,t Indicates the current oil inlet pressure, P out,t Indicates the current oil outlet pressure, Q t represents the current flow rate, T t Indicates the current temperature, S t Indicates the current state of the reversing valve, H t+1 represents the predicted value of the commutation state at the next moment, σ(·) represents the nonlinear activation function, ReLU and Sigmoid functions are selected, W represents the LSTM network weight matrix, b represents the bias term, and t represents the current time step.

6. A control system for a three-position four-way intelligent valve island as claimed in claim 5, characterized in that: The step of using the AI ​​prediction model to analyze the commutation trend based on the historical operation data stored in step S1 and adjusting the commutation strategy based on the commutation trend also includes: Based on the deep Q network DQN optimization commutation strategy, the commutation optimization goal is set as: Where T represents the total time step, λ1, λ2, λ3 represent the weight parameters of the optimization target, and λ1, λ2, λ3 are determined by the Bayesian optimization method, and the determination formula is: Where f(λ) is the loss function, E[·] represents the expected value calculation, and the optimal λ1, λ2, and λ3 values ​​are found through multiple rounds of iterations; P diff,t Indicates the oil pressure change during the current switching process, T sw,t Indicates the current switching time, T sw,opt represents the optimal commutation time; Using DQN reinforcement learning to optimize commutation parameters: The switching time is adjusted by reinforcement learning, and the adjustment formula is: in, represents the optimized commutation time, T sw,0 represents the initial switching time, α represents the learning rate, ΔP represents the load pressure change, DQN is used to train the optimal commutation control strategy. The training process is expressed as: you t =π(H t ), Among them, u t represents the optimized commutation control parameter, π(H t ) represents the reinforcement learning policy function.

7. A three-position four-way intelligent valve island control system as claimed in claim 6, characterized in that: In step S3, the abnormalities include pressure fluctuation, switching hysteresis and excessive pressure supply; When an abnormality is detected, an abnormal state signal is generated and input to the feedback control module.

8. A three-position four-way intelligent valve island control system as claimed in claim 7, characterized in that: In step S4, the dynamic correction includes adjusting the control parameters of the reversing valve based on the abnormal state signal, generating a corrected reversing control parameter, and inputting the corrected reversing control parameter to the actuator.

9. A control system for a three-position four-way intelligent valve island as claimed in claim 8, characterized in that: The step of dynamically correcting the handle position, switching time and back pressure compensation during the switching process by the feedback control module based on the abnormal monitoring result of step S3 includes: Input abnormal monitoring results, set as the input variable X of fuzzy control flc : X flc =(P e ,T e ,S e ), Among them, X flc represents the input variable vector of fuzzy control, P e It represents the pressure error during the switching process, that is, the deviation between the expected pressure and the actual pressure, T e It represents the commutation time error during the commutation process, that is, the deviation between the expected commutation time and the actual commutation time, S e It indicates the handle displacement error during the commutation process, that is, the deviation between the expected handle position and the actual handle position. Set fuzzy rules, use fuzzy control based on rule base, and combine with online adaptive adjustment mechanism: Y flc =g(X flc ), Among them, Y flc represents the corrected commutation control parameter, g(X flc ) represents the control function based on fuzzy reasoning, The fuzzy control decision table is used for reasoning and calculation of the correction value. The formula is: Y flc =w1P e +w2T e +w3S e , Among them, w1, w2, w3 represent the fuzzy inference weight coefficients, which are used to balance the influence of each error term on the correction result. w1, w2, w3 use the online optimization strategy based on gradient descent to perform weight adaptive adjustment. The adjustment formula is: in, represents the weight value at the tth iteration, η represents the learning rate, and J represents the optimization objective function, that is, the sum of squares of the commutation control errors.

10. A three-position four-way intelligent valve island control system as claimed in claim 9, characterized in that: The step of dynamically correcting the handle position, switching time and back pressure compensation during the switching process by the feedback control module based on the abnormal monitoring result of step S3 also includes: The handle position is adjusted dynamically and adaptively in real time, and the optimal handle displacement is calculated using fuzzy control. The calculation formula is: in, Indicates the corrected handle position, S h represents the original handle position, k1, k2 represent the adaptive gain coefficients, k1, k2 use the adaptive adjustment strategy based on dynamic optimization to calculate the gain parameters, the formula is: in, represents the gain value at the tth iteration, λ represents the gain adjustment step size, Predict remote control command delay τ d , adjust the commutation time in advance, the adjustment formula is: in, Indicates the corrected commutation time, T sw represents the original commutation time, γ represents the compensation factor, τ d Indicates the transmission delay of the remote control signal. Intelligent pressure regulating valve is used for dynamic adjustment, and the adjustment formula is: in, Indicates the corrected back pressure compensation, P b represents the original back pressure compensation, β represents the pressure adjustment coefficient, P diff Indicates the instantaneous pressure change during the switching process. Through the flow regulation algorithm, the oil flow control is optimized, and the optimization formula is: Q * =Q+μ(P e +T e ), Among them, Q * represents the corrected flow rate, Q represents the original flow rate, and μ represents the flow rate adjustment coefficient; Execute the corrected commutation parameters, calculate the corrected parameters, and generate the corrected commutation control parameter set U corr : The corrected parameters are transmitted to the actuator to complete the dynamic optimization of the commutation process.

Images