Search

CN-122022489-A - Electro-hydraulic proportional valve self-adaptive real-time efficiency optimization and predictive maintenance method

CN122022489ACN 122022489 ACN122022489 ACN 122022489ACN-122022489-A

Abstract

The invention discloses a self-adaptive real-time efficiency optimization and predictive maintenance method of an electro-hydraulic proportional valve, which relates to the technical field of maintenance control, and the invention collects real-time operation state data in the operation process of the electro-hydraulic proportional valve by installing a sensor, processes the collected real-time operation state data in a data processing mode after the collection is completed, inputs the processed real-time operation state data into a long-short-period memory network after the processing is completed, predicts the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time by the long-short-period memory network, meanwhile, based on the nonlinear parameter drift state of the electro-hydraulic proportional valve predicted in real time, the health state is predicted through a deep learning model, the electro-hydraulic proportional valve is subjected to risk assessment through an uncertainty quantization mode, finally, based on the risk assessment result of the electro-hydraulic proportional valve, an optimal control instruction of the electro-hydraulic proportional valve is generated through a multi-objective robust optimization decision, and the electro-hydraulic proportional valve is driven based on the optimal control instruction, so that the control accuracy of the electro-hydraulic proportional valve is improved.

Inventors

  • ZHANG JIABAO

Assignees

  • 北京嘉海鼎盛科技有限公司

Dates

Publication Date
20260512
Application Date
20260227

Claims (9)

  1. 1. An electro-hydraulic proportional valve self-adaptive real-time efficiency optimization and predictive maintenance method is characterized by comprising the following steps of: S1, a sensor is installed to collect real-time operation state data in the operation process of an electro-hydraulic proportional valve; the real-time operating state data comprises current, pressure and response time; s2, processing the collected real-time running state data in a data processing mode to obtain processed real-time running state data; S3, inputting the processed real-time running state data into a long-period memory network, and predicting the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time through the long-period memory network; s4, predicting the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time, predicting the health state through a deep learning model, and performing risk assessment on the electro-hydraulic proportional valve in an uncertainty quantification mode; S5, generating an optimal control instruction of the electro-hydraulic proportional valve through a multi-objective robust optimization decision based on a risk evaluation result of the electro-hydraulic proportional valve, and driving the electro-hydraulic proportional valve based on the optimal control instruction.
  2. 2. The method for optimizing and predictively maintaining the self-adaptive real-time efficiency of an electrohydraulic proportional valve according to claim 1, wherein the processing the collected real-time operation state data by the data processing method to obtain the processed real-time operation state data comprises the following steps: the real-time operating state data for setting a set of criteria must include current data, pressure data, and response time data; The dimension in the real-time running state data is removed through a data conversion mode, and current data, pressure data and response time data in the real-time running state data are converted into numerical data; summarizing the converted numerical data and constructing triples The converted numerical data is stored in the mode; Wherein, the The representation of the constructed triplet is made, Representing current data in the real-time operating state data, Representing pressure data in the real-time operating state data, Response time data in the real-time operational state data; filtering the stored numerical data in a data filtering mode; Traversing the collected numerical data, positioning the data with content deletion in the numerical data, and deleting the numerical data with content deletion; Setting the numerical data after the deletion processing as the processed real-time running state data.
  3. 3. The method for optimizing and predictively maintaining the self-adaptive real-time efficiency of the electro-hydraulic proportional valve according to claim 1, wherein the step of inputting the processed real-time operation state data into the long-short-period memory network and predicting the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time through the long-short-period memory network comprises the following steps: Collecting historical operation state data of the electrohydraulic proportional valve to form a historical operation state data time sequence with the length of m, and setting the historical operation state data time sequence The following is shown: ; Wherein, the The driving current sequence of the proportional electromagnet is shown, Representing the sequence of spool displacements within a time window, Representing a sequence of output traffic within a time window, Representing a valve port differential pressure sequence within a time window; A sequence of hydraulic oil temperatures within a time window is indicated, Indicating the sliding time window length; s22, inputting the obtained time sequence of the historical operation state data into the LSTM network as an input threshold value, and inputting the operation state data at the moment corresponding to the sensor into the LSTM network as a candidate value; The LSTM network comprises an input gate, an output gate and a forget gate; s23, processing the input of the current moment and the historical running state data of the last moment through a forgetting door, and updating; a set of parameters which directly correspond to the EHPV core physical model and can be observed are set, and the corresponding parameters are as follows: ; Wherein, the A EHPV flow gain; is an electromagnetic time constant; is a viscous damping coefficient; Is coulomb friction; is the hysteresis loop coefficient.
  4. 4. A parameter observer based on a long-term and short-term memory network is arranged in the system, and key nonlinear parameter vectors of the electrohydraulic proportional valve are updated ; The observer takes control current, valve core displacement and system pressure of a valve as input, and continuously updates a parameter estimated value through an online identification technology; s24, outputting the historical operation state data processed by the forgetting gate through an output gate to obtain the historical operation state data predicted at the corresponding moment; The LSTM prediction output is as follows: ; Wherein, the Representing predictions The time of day parameter drift is increased by an increment, Representing a trained long-short-term memory network model, outputting dimension and parameter vectors The same; Current parameter estimation ; Wherein, the Is that Estimating a vector of parameters at the moment; is a nominal parameter vector; representing the drift increment accumulation sum from the initial time to the current time; s25, iteratively using the LSTM network, and determining and obtaining the trained LSTM network; Setting an error threshold between the LSTM network predicted corresponding time historical operation state data and the acquired corresponding time historical operation state data, and when the error between the LSTM network predicted historical operation state data and the acquired historical operation state data is within the set error threshold range, stopping iteration to obtain the trained LSTM network, otherwise, dynamically adjusting the LSTM network parameters until the error converges; S26, setting an LSTM network which accords with an error threshold range as a nonlinear parameter drift state prediction model, inputting processed real-time running state data into the nonlinear parameter drift state prediction model, and outputting a nonlinear parameter drift state predicted at the next moment through the nonlinear parameter drift state prediction model; calculating a nonlinear integrated drift index based on processed real-time operational state data : ; Wherein, the Comprehensive drift index, dimensionless, , Represents the weight coefficient, satisfies , As a flow gain drift term, the degradation of the valve port flow regulating capacity is represented; Representing a nominal flow gain factor; a flow gain coefficient which is currently estimated; a drift term of an electromagnetic time constant, which represents the degradation of electromagnetic response speed; Is a nominal electromagnetic time constant; an electromagnetic time constant that is currently estimated; the hysteresis force item is used for representing the friction and hysteresis nonlinearity degree of the valve core movement; The current hysteresis loop area or the maximum hysteresis force; Is the nominal hysteresis force; a step response rise time term, representing a dynamic response speed degradation of the valve; The current step response rise time; Is the nominal rise time.
  5. 5. The method for self-adaptive real-time efficiency optimization and predictive maintenance of an electro-hydraulic proportional valve according to claim 1, wherein the method for predicting the state of health through a deep learning model based on the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time and performing risk assessment on the electro-hydraulic proportional valve through an uncertainty quantization mode comprises the following steps: S31, setting a nonlinear parameter drift state set Wherein Representing the 1 st nonlinear parameter drift state of an electrohydraulic proportional valve, wherein Representing the s-th nonlinear parameter drift state of electrohydraulic proportional valve, and controlling instruction set Wherein the method comprises the steps of Representing the 1 st control instruction executed in the 1 st nonlinear parameter drift state of the electro-hydraulic proportional valve, wherein The jth control instruction under the s nonlinear parameter drift state of the electro-hydraulic proportional valve is represented; S32, calculating initial probability based on a set nonlinear parameter drift state set and a control instruction set of the electrohydraulic proportional valve, and calculating one-step state transition probability; s33, calculating multi-step state transition probability based on the calculated one-step state transition probability; estimating the possible state of each step through a probability estimation algorithm; S34, establishing a real-time prediction digital twin model based on the output estimated electrohydraulic proportional valve state set, the control instruction set corresponding to each state and the execution probability of the corresponding control instruction; s35, performing risk assessment on the electrohydraulic proportional valve in an uncertainty quantification mode based on the established real-time prediction digital twin model; collecting various indexes of nonlinear parameter drift states of the electro-hydraulic proportional valve, and calculating information entropy of risk assessment of the various indexes; ; Wherein, the , The number of each index of the nonlinear parameter drift state of the electro-hydraulic proportional valve is represented, The weight value of the h index in the s-th nonlinear parameter drift state of the electro-hydraulic proportional valve is represented, Information entropy representing the h index; and setting the calculated information entropy values of all indexes as the risk assessment result of the electrohydraulic proportional valve, wherein the risk is higher as the information entropy value is larger.
  6. 6. The method for adaptive real-time efficiency optimization and predictive maintenance of an electro-hydraulic proportional valve according to claim 4, wherein the calculating the initial probability based on the set electro-hydraulic proportional valve nonlinear parameter drift state set and the control instruction set and the calculating the one-step state transition probability comprises the following steps: The calculation of the one-step state transition probability is as follows: One-step state transition probability distribution:
  7. 7. Wherein, the In order to be a one-step state transition probability distribution, Representing the status After being processed by the rewarding function, the arrival state Is a function of the probability of (1), Representing a bonus function that is based on the received data, , Representing the status Executing control instructions A prize value at that time.
  8. 8. The method for adaptive real-time efficiency optimization and predictive maintenance of an electro-hydraulic proportional valve according to claim 1, wherein the generating the optimal control command of the electro-hydraulic proportional valve by a multi-objective robust optimization decision based on the risk assessment result of the electro-hydraulic proportional valve and driving the electro-hydraulic proportional valve based on the optimal control command comprises the following steps: S51, initializing a multi-objective robust optimization decision; the multi-objective robust optimization decision comprises an accuracy objective decision, an energy efficiency objective decision and a robust objective decision; The precision objective decision function is as follows: ; Wherein, the A trajectory representing the desired state is presented, ; Representation of A predicted state vector for the moment; To minimize the deviation of the state from the expected value in the predicted time domain; the energy efficiency objective decision function is as follows: ; Hydraulic power term: ; Wherein, the The flow unit is (L/min); representing the pressure difference (bar); Representing volumetric efficiency (0 to 1), 600 is the conversion coefficient per unit (1 bar L/min=1/600 kW) Control the rate of change penalty term: ; Wherein, the Representing the control variable in A control input of time; representing a control input rate of change; The robust objective decision function is as follows: ; first part, weighted State error ; Wherein the robustness weight matrix Increasing with increasing parameter uncertainty; Is a robustness adjusting factor; as a parameter uncertainty covariance matrix, ; Second part, control deviation penalty ; Wherein, the Indicating a nominal control amount at the current operating condition, ; To control the bias penalty factor, in general ; Comprehensive optimization problem ; Under the constraint of dynamic model, solving the optimal control sequence Weighting of According to Self-adaptive adjustment; S52, generating an electrohydraulic proportional valve optimal control instruction through a genetic algorithm based on the multi-objective robust optimization decision.
  9. 9. The method for adaptive real-time efficiency optimization and predictive maintenance of an electro-hydraulic proportional valve according to claim 6, wherein the generating the optimal control command of the electro-hydraulic proportional valve by a genetic algorithm based on the multi-objective robust optimization decision comprises the following steps: Summarizing index data in each state of the electrohydraulic proportional valve to construct a population, and setting the genetic population scale and iteration times; Randomly selecting y groups of index data from the constructed population to construct an initial population; Taking a risk assessment result of the electrohydraulic proportional valve and a multi-objective robust optimization decision as an fitness function of a genetic algorithm, and calculating fitness of each individual in the population; Setting index data of each individual representing a group of electrohydraulic proportional valves; the fitness calculation formula of each individual is as follows: ; wherein Fit (k) represents the fitness of the kth individual, Representing a fitness function; Selecting good individuals from all individuals based on fitness values of the individuals and a roulette manner; Crossing selected excellent individuals in a sequential crossing mode after the selection is completed, and generating a new population after crossing ; Randomly selecting an individual from the population to mutate with a set probability to generate a mutated population ; Comparing the fitness difference between the initial population and the population subjected to the genetic algorithm cross mutation ; When (when) <0, Indicating that the population fitness after mutation is higher than the initial population, and the population is accepted when Not less than 0, meaning that the population fitness after mutation is lower than that of the initial population, refusing to accept the population; Further, judging whether the maximum iteration number is reached according to the iteration number of the algorithm, outputting an optimal solution when the maximum iteration number is reached, and continuing iteration when the maximum iteration number is not reached; and setting the output optimal solution as an electrohydraulic proportional valve optimal control instruction.

Description

Electro-hydraulic proportional valve self-adaptive real-time efficiency optimization and predictive maintenance method Technical Field The invention relates to the technical field of rice crust cutting devices, in particular to a self-adaptive real-time efficiency optimization and predictive maintenance method for an electrohydraulic proportional valve. Background The electrohydraulic proportional valve (EHPV) system is a core actuator for realizing high-precision automatic steering of an agricultural tractor. However, EHPV systems are susceptible to non-linear parameter drift (NonlinearDrift) due to their performance and response characteristics during long-term operation due to factors such as oil temperature, oil contamination, spool wear, and internal leakage. This drift is directly manifested as increased flow dead zone, enlarged hysteresis loop, reduced steering accuracy, and increased response time. Disclosure of Invention The invention aims to provide a self-adaptive real-time efficiency optimization and predictive maintenance method for an electrohydraulic proportional valve, which solves the problems in the background technology. In order to solve the technical problems, the invention adopts the following technical scheme that the invention provides an electro-hydraulic proportional valve self-adaptive real-time efficiency optimization and predictive maintenance method, which specifically comprises the following steps: S1, a sensor is installed to collect real-time operation state data in the operation process of an electro-hydraulic proportional valve; the real-time operating state data comprises current, pressure and response time; s2, processing the collected real-time running state data in a data processing mode to obtain processed real-time running state data; S3, inputting the processed real-time running state data into a long-period memory network, and predicting the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time through the long-period memory network; s4, predicting the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time, predicting the health state through a deep learning model, and performing risk assessment on the electro-hydraulic proportional valve in an uncertainty quantification mode; S5, generating an optimal control instruction of the electro-hydraulic proportional valve through a multi-objective robust optimization decision based on a risk evaluation result of the electro-hydraulic proportional valve, and driving the electro-hydraulic proportional valve based on the optimal control instruction. Preferably, the processing the collected real-time running state data by a data processing mode to obtain the processed real-time running state data includes the following steps: the real-time operating state data for setting a set of criteria must include current data, pressure data, and response time data; The dimension in the real-time running state data is removed through a data conversion mode, and current data, pressure data and response time data in the real-time running state data are converted into numerical data; summarizing the converted numerical data and constructing triples The converted numerical data is stored in the mode; Wherein, the The representation of the constructed triplet is made,Representing current data in the real-time operating state data,Representing pressure data in the real-time operating state data,Response time data in the real-time operational state data; filtering the stored numerical data in a data filtering mode; Traversing the collected numerical data, positioning the data with content deletion in the numerical data, and deleting the numerical data with content deletion; Setting the numerical data after the deletion processing as the processed real-time running state data. Preferably, the inputting the processed real-time running state data into a long-short-period memory network, and predicting the nonlinear parameter drift state of the electro-hydraulic proportional valve in real time through the long-short-period memory network comprises the following steps: s21, constructing an LSTM network model; Collecting historical operation state data of the electrohydraulic proportional valve to form a historical operation state data time sequence with the length of m, and setting the historical operation state data time sequence The following is shown: ; Wherein, the The driving current sequence of the proportional electromagnet is shown,Representing the sequence of spool displacements within a time window,Representing a sequence of output traffic within a time window,Representing a valve port differential pressure sequence within a time window; A sequence of hydraulic oil temperatures within a time window is indicated, Indicating the sliding time window length; s22, inputting the obtained time sequence of the historical operation state data into the LSTM network as an input th